Report No. 8893 Colombia The Power Sector and the World Bank, 1970-1987 (In Three Volumes) Volume II: Technical Report June 28, 1990 Operations Evaluation Department FOR OFFICIAL USE ONLY A\ " V",, A entof the World Bank- This document ,has a restricted distribution and may e usqd by recipients only in the performance of their &ficial duties. I cortents may not otherwise . be disclosed without Wold Bank authorization. FOR OFFICIAL USE ONLY ABREVIATIONS AND ACRONYMS AR - Accounts Receivable COL$ - Colombian Peso CONPES - National Economic and Social Policy Council CORELCA * Corporacion Electrica de la Costa Atlantica CPP - Country Program Paper CPS - Central Projects Staff, World Bank CVC - Corporacion Autonoma Regional del Valle del Canca DNP - Departamento Nacional de Planeacion DS - Debt Service Coverage Ratio ECOPETROL - Empresa Colombiana de Petroleos EEEB - Empress de Energia Electrica de Bogota 1PM - Empresas Publicas de Medellin ENCALI - Empresas Municipales de Cali PEN - Financiera Electrica Nacional GDP - Gross Domestic Product gVh - Gigawatt-hour (106 kWh) IC - Long-run average incremental cost ICEL - Instituto Colombiano do Energia Electrica IERR - Internal Economic Rate of Return IFRR - Internal Financial Rate of Return ISA - Interconexion Electrica S. A. JNT - Junta Nacional do Tarifas kW - Kilowatt kVh - Kilowatt-hour LAC - Latin American and Caribbean Region LaMC - Long-run marginal cost *E - Ministerio de Mines y Energia ORD - Operations Evaluation Department, World Bank OMS - Operational Manual Statement, World Bank OPS - Operations Policy Staff, World Bank ROA - Return on Assets ROI - Return on Investment SAR - Staff Appraisal Report SF - Self-Financial Ratio TWh - Teravatt-hour US$ - United States dollar VC - Working Capital This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. CURRENCY EXCHANGE RATES Year Mid-Year (COL$ per US$) 1970 18.4 1971 19.9 1972 21.9 1973 23.6 1974 26.1 1976 34.7 1978 38.1 1980 47.3 1981 54.5 1982 64.1 1983 78.9 1.984 100.8 1985 142.3 1986 194.3 1987 242.6 VOLUME II COLOMBIA THE POWER SECTOR AND THE WORLD BANK. 1970-1987 TABLE OF CONTENTS Page No. 1. TE COLOMBIAN POWER SECTOR (1970-1987).... ................ 1 1.1 Introduction .......................................... 1 1.2 Background ............................................ I 1.3 The Main Entities of the Sector....................... 1 1.4 1970-1974: The New Sector Organization Settles....... 5 1.5 The 1974-76 Crisis and the 'Acuerdo de Sochagota*..... 7 1.6 The Period 1977-79, A New Start of Bank Lending to the Sector......................... 10 1.7 The Period 1980-82, The Turn for the Worse, Power Shortages, and the Guavio Operation................. 16 1.8 The Post-1982 Period of Permanent Crisis.............. 19 11. GENERAL ISSUES*........................**........ 27 2.1 Introduction .......................................... 27 2.2 Vision and Strategy................................... 27 2.3 The Pervasive Optimism in the Bank's Project or Program Presentation................................ 28 2.4 Macroeconomic and Sector Strategy..................... 29 2.5 Least-Cost Solution and Opportunity Cost of Capital............................. 29 2.6 Project and Program Definition, Project Versus Sector Lending ...................................... so 2.7 The Pressure to Lend.................................. 31 2.8 Quality Control ....................................... 31 2.9 Data Base and Documentation........................... 32 2.10 Cooperation with Inter-American Development Bank (IDB) .......................................... 33 2.11 Agenda for the Future.................................. 34 III. SECTOR OBJECTIVES AND INSTITUTIONAL ISSUES................. 35 3.1 Introduction.......................................... 35 3.2 Sector Objectives..................................... 35 3.3 Sector Structure and Regulation....................... 38 3.4 Restructuring the Sector ....................... ..... 41 3.5 ISA's Structure and Role................ . ...... .. .... 42 3.6 The Role of the Bank.................................. 46 3.7 The Agenda for the Future................. ........... 47 Table of Contents (cont'd.) Page No. IV. ELECTRICITY DEMAND AND SUPPLY ..................... . ... . 50 4.1 Introduction .......................................... 50 4.2 Evolution of Electricity Sales, Demand and Generation................................ 50 4.3 Dynamics of the Structure of Electricity Sales........ 52 4.4 Performance of Sales/Generation Forecasts Relative to Outcomes................................ 57 4.5 Growth of System Losses...... .... .... .. ........ ... 63 4.6 Inability to Meet 1981 Power Demand--Cost to the sconomw ...... ......... ........ .............00... 68 4.7 Surplus Generating Capacity--Cost to the Economy...... 70 4.8 Performance of the Bank ......... ... .. ..... ...... 72 4.9 Agenda for the Future................................. 73 V. POWER TARIFFS .....t... .*............................ o.... 74 5.1 Introduction .......................................... 74 5.2 Evolution of Electricity Prices in the Review Period .............................................. 75 5.3 The Performance of Tariff Forecasts Relative to Outcomes ............................................ 79 5.4 Economic Costs, Retail Tariff Structures, and Levels*..... . .............. ...... ... o ...... *..... 80 5.5 Economic Costs and the ISA Bulk Tariff................ 86 5.6 Tariff Estimates and Internal Rates of Return at Appraisal ........................................ 92 5.7 Cost Differences Between Regional Power Markets....... 103 5.8 Income Distribution................................... 108 5.9 Bank Performance...................... 121 5.10 Agenda for the Future........ ..". ....... .......... 124 VI. FINANCES*...... ......o... .............. . .. . . .... ..... 126 6.1 Introduction .......................................... 126 6.2 The Broad Picture................. ...... ...... 126 6.3 Self-Financing ..........o.......................... 129 6.4 Returns on Investment...................... ........ 136 6.5 Debt Service Coverage....... .... ..... .... ...... ... 143 6.6 Debt and Equity............. **.. ..... . . ...... 147 6.7 Accounts Receivable................................... 149 6.8 Bank Lending Terms........................ ........ 153 6.9 The Foreign Exchange Risk and Losses.................. 155 6.10 Appropriateness of Some Financial Indicators........ 158 6.11 Agenda for the Future....................... ....... 160 Table of Contents (cont*d.) Page No. VII. SECTOR INVESTMENT........................................ 162 7.1 Introduction .......................................... 162 7.2 The Investment Decision Process....................... 162 7.3 Planning Generation and Transuission.................. 163 7.4 Project Implementation Time........................... 171 7.5 Project Costs......................................... 173 7.6 Balancing the Investment Program...................... 177 7.7 Bank Involvement.......... .......................... 179 7.8 Agenda for the Future................................. 180 VITI. LINKAGES BETWER THE PO0ER SECTOR AND THE ECONOMYr.......... 183 8.1 Introduction .......................................... 183 8.2 The Two Types of Power Sector-Macroeconomy Linkages .... 0. 0..........I..........0............... 183 8.3 The Power Sector and Economic Growth.................. 184 8.4 Performance of the Bank............. 187 1. THE COLOMBIAN POWER SECTOR (1970 - 1987) 1.1 Introduction 1. The present chapter sets forth the organizational and historical background against which the following chapters discuss the main sector issues and the way the Bank dealt with them in the review period, which covered four presidencies, those of Presidents Pastrana (1970-74), Lopez (1974-78), Turbay (1978-82), Betancur (1982-86), and the start of a fifth under President Barco. 1.2 Background 2. In Colombia, as in most countries, electrification started at the turn of the century, at a very small scale, in the main centers. Private companies provided this new service. Towards the middle of the century, the general trend towards public ownership of public services led, in Colombia, to the acquisition, mostly by municipalities, of the assets of the power companies. This is in contrast to many other countries (e.g. Brazil and Mexico) where Central Government played a much larger role in the transitional period. However, the development is chacteristic for Colombia, where the various regions--for a long time relatively isolated due to the rugged topography--have very strong identities, which they fiercely defend against what they perceive as the levelling and uniforming influence of Central Government. 3. As the electric systems grew, it became increasingly desirable to interconnect the neighbouring ones to permit the construction of larger, more economic generating plants, too large for a single utility. Thus, beyond reaping the benefits of economies of scale, interconnected power companies were able to reduce their reserve capacity with respect to that they would have required if they had remained isolated. Such regional systems developed around the main cities (Bogota, Medellin, Cali, Manizales, Barranquilla, etc.). As these systems grew and as their limits came closer to each other, their interconnection became desirable. The period under review witnessed the incorporation of all the regional grids into one large interconnected system. 1.3 The Main Entities of the Sector 4. Throughout the period 1970-87, seven entities formed the backbone of public electric power supply in the countrys Empress de EnerUia Electrica de Bogota (REEB), Empresas Publicas de Medellin (EPM), Corporacion Autonoma del Valle del Cauca (CVC), Empresas Municipales de Cali (ECALI), Instituto Colombiano de Energia Electrica (ICEL), Corporacion Electrica de la Costa Atlantica (CORELCA), and Interconexion S.A. (ISA). Table 1.1 sets forth some key data illustrating the development of the sector and its utilities over the period under review, whereas Annexes 1.1 and 1.2 list the Bank and IDB operations in the sector. - 2- Table 1.1s Development of Generation and Retail Sales XUB B M CVC* ICEL CORELCA ISA Total Group Group --e~ion 1971s TUh 2.5 2.2 1.0 1.4 1.4 0.0 8.2 1986: TWh 3.3 5.6 3.0 2.8 4.2 8.6 27.5 1971t Share 312 262 122 172 132 01 1001 1986s Share 121 202 112 102 15% 312 1002 Retail Sales 1971t TUh 1.8 1.6 1.1 1.1 0.9 0.0 6.7 19861 TUh 4.8 3.9 3.0 4.6 3.6 0.0 20.1 1971: Share 272 242 162 162 132 01 1002 1986: Share 242 192 152 232 182 02 1002 * includes EHCALI. 5. EEB, originally a private company, since 1951, is owned by the Mmnicipality of the Special District of Bogota. In the course of time, besides supplying electricity to the capital city, it also provided bulk power to subsidiaries of ICEL in the neighborhood, the area of supply changing over time and with the configuration of the interconnected system. KEEB has had a long history of good management and of satisfactory performance both in the operational and the financial field. As discussed later In the present report, this picture has been tarnished by more recent developments. Between 1960 and 1970, the Bank assisted EEE3 in its institutional consolidation through three projects for which it lent US$85.6 million. In the same period, IDB made one loan for US$21.2 million. Since 1970, there have been four operations with lending in the total amount of US$701 million (including US$359 million for the Guavio project in 1981). In this more recent period, IDB also dramatically increased its lending to EE, as it provided US$606.4 million, mostly for the ill-fated Guavio hydroelectric project. In the early 1970s, the company was a net exporter of electricity to the interconnected system, which by then had just started operating. For over the past decade, it has become a large Importer. How and why, especially In the course of the second half of the period under review, EE's perfotmance deteriorated, when, at the same tine, ID and IBID assisted the sector in general and EEEB in particular at an unprecedented level, is one of the major issues that the following chapters discuss. 6. EPH is also a municipally owned entity. It was established in 1955 and has four departments providing power, water, sewerage, and telecommunica- tions services, respectively. Whenever in following sections the report mentions UP, it refers to the electricity department of the utility. Today, beyond generating power for, and distributing it in, the city of Medellin, EPH sells -3- bulk electricity to the ICEL subsidiary in the Antioquia Department. Since the Interconnected system started operating In the early 1970s, 3PM has consistently been a net exporter of energy to the system. In the formative years, before 1970, the Bank assisted the institutional growth of EPM with three loans (US$37.6 million). In the review period, as in the case of BE, Bank support to EPH increased dramatically, totalling US$430.5 million in four loans. In contrast, until now, IDB has made only one loan (in 1980) of US$106.4 million to RPM. During the past thirty years, EPM has remained a well managed strong utility. As the following chapters will discuss, it was able to come through the recent years of crisis relatively unscathed. 7. Government, strongly supported by the Bank, created CVC in 1954 for the overall development of the resources of the Cauca Valley, somewhat along the lines of the Tennessee Valley Authority. In 1958, CVC obtained the majority of the shares of Central Hidroelectrica del Rio Anchicaya Limitada (CHIDRAL), the main bulk electricity supplier in the Cauca Valley, and until then responsible for planning, construction, and operation of power facilities in the area. CVC took over the planning activities and, in the course of time, also the sale of power to the main distribution entity in the region, Empresas Municipales de Cali (EMCALI), as well as to most other municipalities of the department. The complex setup in the Cauca Valley, in particular the often uneasy relations between CVC, CHIDRAL, and E4CALI has led to substantial inefficiencies as well as to conflicts. (1) In 1950, CHIDRAL became the first power sector entity in Colombia with which the Bank had an operation. Until 1970, CVC and CHIDRAL were the beneficiaries of five loans for a total of US$44.7 million. However, in the review period, there were no further Bank operations specifically with CVCICHIDRAL which, until now, received only one loan for US$63 million from IDB. In the following the acronym CVC will designate the group. 8. ENCALI was created in 1931 as a municipally owned public utility in charge of providing the city of Cali and municipalities in its immediate neighborhood with (i) water supply and sewerage, (ii) electric energy, and (iii) telecommunications services formerly offered by municipal departments and private companies. In contrast to EMB and ZPM, ECALI does not own directly generating plant, but rather purchases all the electric energy it sends out from the CVC group, in which it has a stake in form of a minority participation in CHIDRAL (CVC owning the majority of the shares), and with which, as mentioned, relations, at times, have been strained. EMCALI is the only large power utility in Colombia that has never received a loan from IDB or the Bank. Indirectly, of course, it has benefitted from both banks' assistance to its suppliers, CHIDRAL and CVC, and, more recently, to ISA of which CVC is a shareholder and one of the main clients. 9. In 1946, Government created ELECTRAGUAS, ICEL's predecessor, as a dependency of the then Ministry of Development to promote the development of electric power and irrigation in the country and to formulate comprehensive plans in these sectors. At an early stage of development, ELECTRAGUAS con- centrated on its power sector activities. In 1968, it was reorganized, assigned to the Ministry of Public Works, and renamed ICEL. The institute was intended to become the directing agency for electric power development in Colombia. It would therefore have played a major role in (i) preparing national power -4- development programs, (Ii) securing finance and constructing specific components of such programs, outside the area of influence of the other main sector entities, and (iii) coordinating the activities of the utilities under its tutelage. With the emergence of ISA the general planning role of ICEL disappeared. Thus, during most of the review period ICEL essen-tially was controlling 15 departmental and municipal subsidiaries, which supply the consumers outside the service areas of BEEB, EPM, CVC, and CORELCA. For most of the time since the creation of the entity, its development lagged behind that of the other large ut'.1ities, mostly because it serves poorer, dispersed markets, in smaller centers and in rural areas. However, poor organization, lack of incentive for the internal development of the subsidi-aries, and political pressures, contributed to the lagging performance, which forced Government to subsidize heavily the institute and its subsidiaries. Although, in the 1950s and the early 1960s, the Bank made four loans for a total of US$14.6 million to subsidiaries of ELECTRAGUAM. it never lent to ICEL. IDB, in turn, made three loans totalling US$83.1 million to ICEL and one of its main subsidiaries, CREC. Recently, ICEL's share of retail sales has reached the level of EEEB's, making the two utilities, both in poor shape, the two largest distributors in the country. In the review period, the Bank did not directly contribute to finance ICEL's development, but it tried to address some of the issues within ICEL in the context of operations that were essentially oriented towards the power sector as a whole. 10. Created in 1967 as a decentralized public ea 'ty, CORELCA took over ELECTRAGUAS/ICEL's responsibility for the local utiliti., (Electrificadoras) in seven departments along the Atlantic Coast. CORELCA's system is entirely based on thermal generation. The utility suffers essentially from the same weaknessses as ICEL, as exemplified by the fact that, in 1986, six of CORELCA's nine subsidiaris were technically bankrupt. The Bank made only one loan, in 1981, for US$36 million to CORELCA, for its village electrification program, whereas IDB, until now, did not lend directly to the utility. How-ever, as in the case of ICEL, in the context of the projects that addressed essentially sector-wide issues, the Bank tried to help find solutions to the utility's problems. 11. In 1967, when ISA was created, largely at the Bank's insistance, the company's principal objective was to interconnect the four main systems, i.e. those of its shareholders, EEEB, EPM, CVC, and ICEL. In 1978, after subs."n- tial controversy, CORELCA also became a shareholder. ISA was to plan, con- struct, own, and operate new power plants and the associated transmission facilities for the bulk supply to its sponsors' systems. Over the first half of the review period, the Bank heavily supported ISA through four loans in a total amount of US$250.3 million. The first Bank loan to ISA, for the line interconnecting the original sponsors' systems, was made in 1968, i.e. before the start of the review period. In principle, the 1978 US$50 million loan to Government for the link-up of CORELCA should also count as lending to ISA, as the company became the owner of the line. After 1979, the Bank did not extend further loans to ISA directly. However, the operations aiming at the sector as a whole, even when they involved lending to a single entity (e.g. EE in the case of the Guavio operation), were very much concerned with the issues related to ISA. Over the past 18 years, eight loans for a total of US$1004.5 million (including US$360 million to help finance ISA's increased share in the Guavio - 5 - project) were made, though IDB's direct financial involvement was much larger than that of the Bank. ISA's setup and its role In the sector are two of the main sector issues and therefore at the very center of the present review. 1.4 1970-1974s The New Sector Organization Settles 12. In 1970, the government of President Pastrana took office in the midst of an economic boom, as coffee prices, while declining, were still high. However, as the new government did not command a majority in Congress, it was substantial- ly hampered in its economic decisions. This may have contributed to the disappointing growth of GDP at less than 62 in 1970 and 1971. Inflation, which had been at about 14% per year in 1970-72, jumped to about 25Z per year in 1973- 74, which raised issues the Government of President Lopez had to address as soon as it took office in mid 1974. 13. For the power sictor, the start of the 1970s was a transitional period, as it immediately followed the 1967-68 restructuring of the sector. Government, working through ICEL, which it had created in 1968, and the major urban utilities attempted to improve planning and operations by grouping the smaller systems into six zones and delegating the responsibility of supply within each zone. Also in 1968, Government, In the same vein, had establ ahed CORELCA. Again in 1968, it established within DNP, Junta Nacional de Tarifas (JNT), the National Tariff Board, which was to oversee public utilities' tariffs, in general and electricity rates, in particular. Further, ISA, created in 1967, by the turn of the Oecade, was well on its way Implementing the interconnection of the main urban suppliers (SEES, EPM, CVC/Chidral, and CREC), the Bank providing support through Loan 575. It completed the link between the main systems by 1971, as planned. The connection of ICEL's northeastern system followed in 1973. In the early 1970s, despite these early achievements, ISA, as the other newly established or re- designed entities, was still trying to find its proper role. 14. In 1969, as a first step towards an integrated sector development plan, ISA submitted to DNP a generation and transmission plan to meet the demand in its shareholders' systems until the late 1970s. ISA's plan assumed that, throughout the 19709, demand would grow at an average rate slightly in excess of 10%, i.e. somewhat less than the 11 envisaged for the sector as a whole. It set forth the following generation projects: - Alto Anchicaya (340 MV) near Cali in the CYC area, - Chivor (500 M) on the Rio Bata, east of Bogota, and - Quatape II (280 MM) on the Nare/Guatape rivers, an extension of EPM's existing plant. Government's 1968 decision to allow CVC to build Alto Anchicaya immediately, the lack of interest of BPM in proceeding quickly with the implementation of Guatape II, and some advantages of Chivor I over Guatape II in terms of potential output in average and wet years--which the straightforward least cost analysis did not reflect--resulted in the decision to build Chivor I first. -6* 15. The Bank and IDB, which had sponsored ISA's above analysis of the most desirable sequence of plants, after a further comparison of the plant with a large thermal development, supported the project and, in 1970 and 1971, approved loans of US$52.8 (IBRD loan 681) and US$58.3 million respectively. Obviously, though this does not come out of the Bank's appraisal report, there were already in 1971, doubts about the feasibility of completing Chivor by mid-1975. Further, construction of Alto Anchicaya was definitely late. Therefore, the utilities proposed, and Government agreed to install 198 MW of thermal additions to secure supply in 1974. In 1972, in agreement with ISA's plan, EPM started construction of Guatape II. The Bank supported this venture with Loan 874 of US$56 million. Ultimately, none of the projects started in the period 1968-1971 was completed by 1974 as all suffered delays between 11 months (Alto Anchicaya) and 41 months (Barranca III thermal), which resulted in an exceedingly tight capacity situation in 1974 and a deficit in energy during the dry season 1976-77. Serious shortages would have occurred, if demand had -rown at the anticipated 10%. Actually, it developed at about 9.6Z for energy and 8.6Z for capacity. 16. The two Country Program Papers (CPPs) falling into the period 1970- 74 reflected the Bank's declarn4 policy to shift gradually the emphasis of its lending to Colombia away from infrastructure, in particular power projects, for which long term export credits appeared readily available. This was in parallel with a similar change of emphasis in Government's priorities.(2) Accordingly, the CPP foresaw only two power projects, but the Bank planned to act as a catalyst to mobilize resources to replace those it would no longer be able to assign to the power sector. However, the 1971 CPP provided a fallback position by stating that, should agriculture projects slip, the Bank would consider adding some projects in transportation and/or power. 17. The 1973 CPP essentially follows, for the power sector, the lines set in the previous CPP. It merely adds a further ISA project, the link-up of CORELCA's system. However, already in the second half of 1973, there were signs that the Bank was reversing its position, as, in order to contribute further to the consolidation of ISA, it tentatively planned to appraise, already in the following financial year, the North-Central Interconnection together with the San Carlos project. 18. During the first four years of the 1970s, IDB supported subtransmis- sion and distribution in ICEL (US$25.8 million in 1971), transmission and distribution in EEB (US$21.2 million in 1973), and CHEC's Esmeralda/Insula hydro scheme (US$7.3 million in 1973). Most importantly, in December 1973, its Board approved a US$58.6 million loan to help finance the installation of the 500 M0f second stage of the Chivor hydro plant, a project in which IBRD did not get involved. 19. During 1970-73, the Bank's institution building efforts in the power sector concentrated on two main issues, both closely related to the sector's setups the national power development plan and the power sector finances as reflected in those of ISA. Indeed, since the mid-1960s, the Bank had strongly recommended the preparation of a power sector plan covering institutional development as well as a long term investment program for generation, trans- mission, and distribution. Under Loan 575 (Central Interconnection) the -7- Colombians undertook two steps towards the plan by requesting from ISA a study of the least cost development of generation for the 1970s (see para. 14) and by appointing consultants to prepare recommendations on the operation of the interconnected system, an tariffs ISA should apply, and on the corresponding adjustments of ISA's charter. 20. The twnlementation of the long term power sector development study became a condit .of Loan 874 for the Guatape II Project and was to be completed in 1973. How tdr, it had an exceedingly slow start, mainly because of the reluctance of VA's shareholders, which felt that the enforcement of the plan preparation implied an attack on their independence. The situation was further confused by a public controversy related to the publication of an ICEL study and subsequent statements by the Minister of Mines and Energy to the effect that the country needed a single national power company. In this environment, the regional utilities refused to co-operate in the preparation of the plan. The Bank was very active in trying to convince them to abandon their opposition to the plan.(3) In related discussions, for the first time, it slightly relaxed its, uutil then, strictly centralistic position by assuring that, while strongly supporting the implementation of the study, it did not favor any particular institutional scheme.(4) Despite the Bank's efforts, the sector remained skeptical and the plan, in the course of time, was redefined essentially as an inventory of Colombia's hydroelectric potential. Thus, the result fell far short of the original objective. The consultants completed this study in 1977, at long last. 21. The institutional study, which should have been available in 1970, was finished in 1974. However, ISA's shareholders did not endorse the consul- tants' recommendations and therefore ISA's charter remained as originally agreed. The study included two controversial proposals. One was the relaxation of the rule according to which all future major generating plant in the interconnected system should be built and owned by ISA and the other was the integration into ISA of all further utilities whose systems would be connected to the Central System, which made CORELCA the next candidate for becoming a shareholder. 22. In 1970, the Bank did not yet analyze the finances of the sector in a consolidated way. However, as ISA's position was entirely dependant upon that of its shareholders, the appraisal of the Chivor project investigated the position of the individual companies, finding EEEB's and EPM's situation and prospects satisfactory and defining the required Government contributions to CVC and ICEL. The 1972 Guatape II appraisal only evaluated EPM's finances and projections, and judged them to be satisfactory. However, the sharp increase in inflation in 1973-74, which was not associated with commensurate tariff adjustments, led to a quick and dramatic deterioration of all the utilities' financial position. Worst affected was ISA, as it stopped receiving the proper contributions from its shareholders, themselves in dire straits. 1.5 The 1974-76 Crisis and the *Acuerdo de Sochagota' 23. The start of the crisis roughly coincided with the first dramatic increase in petroleum prices in 1973. The difficult position was exacerbated by Government's reluctance to take unpopular measures in its last six mouths of -8- office (first half of 1974). Once President Lopez had taken over, In mid-1974, his Government concentrated on issues that were even more pressing than those in the power sector. However, in the meantime, sector finances were deteriorat- ing, the sponsors of ISA owing the company some US$14 million. BEEB had fallen US$2 million short of its committment to purchase ISA shares and BPM was refusing to comply with its obligations to ISA, unless ISA's bylaws were revised to require, inter alia, that future generating plant be owned and operated by new satellite companies, in which the directly interested utilities would be shareholders. RPM also wished to make its contribution to ISA contingent upon an equal contribution by Government. As the contributions were defined in ISA's charter, and compliance with the charter was a condition of the Bank loan for the Chivor project, the Colombians found themselves in default of conditions agreed with the Bank. In the fall of 1974, when the Bank saw that little action was coming forth to eliminate the default, it decided to suspend informally the disbursements from the two loans to the power sector that were current at the time. By May 1975, monthly tariff increases had been instituted throughout the country and Government had made funds available to CVC, 1EE, and ICEL enabling the utilities to pay up their arrears to ISA. Further, agreement had been reached with EPM on a schedule for paying its dues and on the company dropping its condition for making its future contributions to ISA. On the basis of these actions, the Bank decided to resume disbursements. Despite the progress made, doubtlessly to a good part as a consequence of the Bank's intransigence, within a month of resumption of disbursements, the Bank had to discuss again the proposal, originally presented by EPM, on the creation of new companies for building and operating individual large new power plants.(5) Further, ISA's shareholders quickly fell back into arrears. By the end of 1975, the situation was even worse than a year before and the sector again in default of the Chivor agreements with the Bank. Tariffs, though now increasing at a rate commensurate with inflation, did not permit the utilities to catch up with the losses incurred in previous years. Thus, at that time, the shareholders owed ISA about US$17 million, while ISA owed its contractors over US$25 million, giving the latter nearly a stranglehold on project implementation, one major reason for delays and increased project costs. 24. In the same month the Bank resumed disbursements from its loans, it set forth a new CPP reflecting a reversal of earlier policies and the intention to resume lending to the sector at a large scale. It justified this change arguing that its active involvement in the sector was crucial to achieving the integration objective and to supporting Goverment's efforts to mobilize regional cooperation for rational investment and rate-making policies. The OED review team judges that, indeed, without lending, the Bank could not expect to have the impact necessary for progress in sector integration. 25. In the meantime, construction of the Chivor and Guatape II plants was delayed mainly because of tunnelling difficulties, lack of funding (Chivor) and because of popular opposition to evacuating the communities to be flooded by the enlarged reservoir for the Guatape facility. The plants were commissioned in 1977 and 1978, respectively, both about two years late. Therefore, neither plant could help avoid the supply shortages which occurred in 1976 as a consequence of a prolonged dry-season. Rationing, amounting to 5-10% of rejuirements during some 5 months, affected mainly the CVC and CHEC systems.( ) However, as -9- generating facilities, existing and under construc-tion, were expected to be able to meet demand barely until 1980, construction of additions had to start in 1976, at the latest. Accordingly, ISA's 1975 development program identified the San Carlos hydroelectric development (620 MW), downstream from Guatape in the Medellin region, and the two-stage Mesitas development (600 MW) on the Rio Bogota as the projects with the highest priority. It seems that ISA's shareholders agreed without major controversy, that ISA should build and own San Carlos and that EEEP would do so for Mesitas; it further appears that the utilities had already informally decided on the latter in 1973, but that the Bank did not know about this decision, which, strictly speaking, was in contradiction with ISA's bylaws, but made sense insofar, as the plant was definitely in EEEB's area of generation and was to work in parallel with other EEB plants. Indeed, it was to use additional water to be diverted to the Rio Bogota catchment through the Chingaza conveyor, which, in the aid 1970s, was already under construction to provide additional water for the supply of the Bogota agglomeration; the Bank helped finance this venture through Loan 741 to Empress de Acueducto y Alcantarillado de Bogota (EAAB). ISA planned completion of San Carlos and Mesitas for 1981, which, in view of the state of project preparation and of the experience accumulated with previous plants, should have been recognized as unrealistic. It further considered the interconnection of the Central System with CORELCA's northern grid possible by 1978. In case of delays it envisaged a more intensive use of existing thermal capacity, but found further thermal plants not to be part of a least cost alternative, mostly because of high development costs of coal mining. Later in 1975, the Salvajina plant in the CVC area was added to the program. It is evident that these projects, although labeled part of the least cost development, were based on analysis that fell short of a comprehensive study of alternatives developed to reasonably comparable levels. (7) 26. Faced with the need for the above large investment, Government and the sector urged the main potential financiers, IDB and IBRD, to provide finance as quickly as possible. However, the institutions agreed to go ahead only on the basis of more clear and acceptable prospects as to the future development of the sector, in particulars - a reasonable financial policy to put the sector back on a healthy footing and to support its long term development; - a strong commitment from Government to support ISA in becoming a central force in the power sector; - the proper incorporation of CORELCA as a shareholder of ISA; and - the designation of ISA as the ultimate owner of the interconnection between the central and the northern systems. It seems worth noting that the Bank set these strong conditions for further lending to the power sector at a time when lending was practically at a stop in other sectors, too, i.e., under circumstances when normally the pressure for further operations is high. Indeed, in FY76, the Bank made only one operation in Colombia, IDB none. - 10 - 27. Although in 1976 electricity tariffa, for the first time in the 1970s, increased in real terms, the crisis worsened, in particular as the required Investment now loomed very large. Indeed, in April 1976, DNP submitted to Government an analysis evaluating for the years 1976-78 the gap in the sector finances at some US$260 million (corresponding to about 1986 US$480 million) i.e., about 401 of planned Investment.( ) In September 1976, confronted with this dramatic situation and the Bank's (and IDB's) refusal, under the prevailing circumstances, to go ahead with a lending operation, Government convened a meeting of the main actors of the sector. This resulted in the 'Acuerdo de Sochagota* which spelled out, albeit partly in somewhat vague tems: - an urgent generation program and the studies for the development in the medium term; - a tariff policy that would imply increases in the rates for large residential consumption as well as commercial and industrial uses; - the creation gin due time, of a mechanism to ease the mobilisation a funds to finance the electricity development plane; and - the induction of CORELCA as a shareholder of ISA, in such a way that the Government owned entities (CVC, ICEL, and CORELCA) could not outvote the two municipally owned utilities, EEES and EPM.(9) 28. The Bank judged the Acuerdo de Sochagota to represent a step in the right direction, although the agreement implied that ISA would have a role substantially different from that envisaged in the late 1960s. Indeed, it meant that, in the iamediate generation program, each utility would construct a plant of its own, thus satisfying to a degree the regional interests: ISA would build San Carlos (ultimately 1240 W5), 13E8 Mesitas (600 XV), BPM Guadalupe IV (213 MM), CVC Salvajina (270 MW), and CORELCA Cerrejon (150 W thermal), Barranquilla (66 M0W thermal) and Cartagena (66 M thermal). Consequently, ISA would be one of the main suppliers of bulk electricity in the interconnected system but by no means the absolutely dominant one. This pattern that, with hindsight, ca already be vaguely discerned in the late 1960s with the assigment of Alto Anchicaya to CVC, Guatape II to EPM, and Chivor to ISA, would be dominant at the next turn of sector development, in 1979. It is also Important to note that the Acuerdo de Sochagota promoting a tariff policy implying substantial tariff increases for industry and commerce and more modest adjustments for residential consumers set the stage for the heavy subsidies of residential consumption that are partly responsible for the sector's distressing financial situation in the mid and late 1980s. 1.6 The Period 1977-79, a New Start of Bank Lending to the Sector 29. This period was dominated by the nearly obsessive preoccupation of the Colombian power sector with the threat of shortages in electricity supply in the early 1980s, on the one hand, and by the protracted procedure that ultimately led to the resumption of Bank and IDB lending to the sector on the other. It was, at the start, a period of substantial improvement in the sector finances mostly due, during 1977-78, to substantial tariff adjustments in real - 11 - terms associated with a relatively low level of sector investment. This went parallel to the marked improvement In the economy as a whole, which had started back in 1975, when coffee prices reached a new high, but which was felt in the power sector mostly in 1977. It also made the rate increases envisaged in connection with the 'Acuerdo de Sochagotal (pars. 27) more palatable. Further, the commissioning in 1977 of Chivor provided ISA with income of its own, thus allowing it to receive at least part of the contributions from its shareholders as payments for services rendered rather than as straight equity contributions. However, the improvement was short-lived as, in 1979, the state of the economy worsened. Inflation jumped to 29% per year, obliterating the financial relief expected from that year's tariff increases. Sector investment, at the same time, more than doubled when construction of the large Bank and IDB assisted projects finally started. Worst affected was ISA, which once more did not receive the proper contributions from its shareholders and therefore was unable to meet its obligations towards its contractors, which, in turn, entailed further delays of works in progress, in particular, of Chivor II. 30. In January 1977, right at the start of the period, the Bank set forth a CPP stating with reference to the wAcuerdo de Sochagota* that, in the power sector, *Government has taken leadership* and that 'it has resolved the long standing institutional problem by its active involvement' in ISA, which "has been given a normative role and has been made responsible for development of future generating facilities.'(10) Hindsight tells that this was a gross misreading of the situation. But, it seems that, even at the time, experience and the rather vague formulation of the agreement warranted a more prudent evaluation of the readiness of the Colombians to relent on their commitment to regional interests and to assign ISA the role the Bank thought appropriate. In the same CPP, the Bank re-stated the policy of increased lending to the power sector already set forth in the 1975 CPP but not implemented for the reasons discussed in the previous section. Accordingly, the Bank planned for the period FY78-82 six loanb for a total of US$445 million, i.e. about one third of its total lending program. Whereas the preceding CPP envisaged one operation in distribution, the new one foresaw loans for generation and transmission only. This obviously reflects the main preoccupation of the Colombians and overlooks repeated reminders from Bank officers of the unsatisfactory and deteriorating state of many distribution systems.(11) 31. Although the 'Acuerdo de Sochagotal addressed most of the issues for which the Bank needed at least the perspective of a solution, the various commitments from the Colombian side were not concrete enough for the Bank to proceed with the appraisal of the next projects, San Carlos and the North- Central Interconnection. It took another six months until the Bank could declare itself satisfied. During this time, the relations between the Bank and the Colombians were often strained, as the latter were anxious to start the projects and the Bank adamant about the fulfillment of the set conditions. The ensuing pressure led, at least in one case, to misunderstandings that contributed to postponing the appraisal even further. When, in April 1977, the Bank finally carried it out, the mission's findings were not encouraging. This led to the formulation of rather stringent conditions of negotiations which, to be fulfilled, also required time. These conditions, inter alia, called for: - 12 - * the submission of the sector study agreed in the context of the Guatape II Project and which should have been available in 1973; - further changes in ISA's by-laws to clarify ISA's role; and - tariff measures implying increases of the rates in real terms such as to allow the main utilities to earn a 9Z return on net fixed revalued assets. 32. As could be expected with the number and the severity of the conditions set, the Bank had to adjust them in the course of time. Several became main issues discussed at negotiations. Finally, it took 13 months from appraisal to Board presentation of the projects, a time during which relations between the Colombians and the Bank remained tense but which was also characterised by considerable friction between the responsible Projects Department, which was reluctant to relax the conditions and the Programs Department which was under pressure from the Government and anxious to complete the programmed operation. The main points of dissent weres the sector's institutional and financial performance, for which Projects' staff wanted to have firm commitments to specific far-reaching improvements, whereas Programs' staff considered it politically unrealistic to try to obtain satisfaction on all points at this stage and preferred to endeavor reaching agreement on some of them in the context of planned subsequent operations. Programs' personnel was also unhappy about Projects' staff modifying the conditions, which, however, was often due to the fact that conditions set at an earlier stage had become unfulfillable and had to be either changed or dropped altogether. A further point of contention were the changes in ISA's bylaws in connection with COBELCA's induction into ISA. Projects' staff wanted them as specific as possible, which made them unacceptable to the Colombians and therefore postponed agreement even further. (12) 33. In May 1977, as the threat of shortages in the early 1980s became ever more real, DNP submitted to Government and the latter approved an emergency plan including (i) diversions to the Chivor reservoir to increase the potential energy output of the scheme, (i) 132 MW and 66 MW of thermal plant in the central and northern systems, respectively. These additions, except for one diversion, were planned to start operating in 1981 and thus to help assure continuous supply. In June 1977, ISA submitted to DNP its development plan for the interconnected system through 1988, which took into account delays that had been incurred in the implementation of the Chivor II and the Chingasa schemes. As in the course of 1978 these delays increased further, ISA revised its program and, in early 1979, proposed a second emergency plan implying the installation of 200 MW of gas turbines at Barrancabermeja in the Central System. However--and this is a sign of the haste that prevailed when these decisions were made--the plan proved unfeasible, as ECOPETROL was not in a position to provide the gas. Therefore, it had to be modified, which led to a solution combining steam and turbogas plants in various areas, a solution 4hich ultimately could not help in the 1981 emergency and only to a limited extent in that of 1982. (13) 34. In parallel with the protracted preparation of the operations for the San Carlos plant and the North-Central interconnection line, the Bank started preparing further projects for accelerated processing as soon as the first two * 13 * operations would have been agreed. This occurred in May 1978, when the Bank's Board approved a US$126 million loan to ISA for the power plant and one of US$50 million to Government for the interconnection line. The next two operations followed quicklys in November 1978, the Board approved a loan of US$84 million to fEEE for the Mesitas hydro development and in June 1979 one of US$72 million to ISA for the second stage of the San Carlos plant (i.e. another 620 MW). In parallel, IDB, in the same period made loans of US$96.6 million to ISA for San Carlos I in December 1977, of US$131.7 million also to ISA for the Jaguas hydro plant upstream from San Carlos, and of US$51.5 million to CVC for the Salvajina multi-purpose development in the Cauca Valley. This indicates that IDB was ready with its San Carlos loan six months before the Bank was in a position to proceed with its own operation, a fact that had created some uneasiness. Indeed, Bank staff felt that, as some of IDB's conditions were less stringent than those the Bank hofed to get accepted, this might untermine its position during negotia- tions.( 4) 35. Beyond detailed conditions on the investment programs and the finances of ISA and of its shareholders, as well as on the obligations of the latter towards ISA, the San Carlos operations contained conditions of which the main ones referring to institutional issues are discussed below. They reflect the Bank's approach to institutional development in the power sector as a whole once it had become evident that the straight centralistic approach was inoperative. The reach of these conditions explains to an extent the long time it took to agree on them. The list further exemplifies the Bank's pervasive optimism as to the implementation schedules for the measures envisageds - completion in 1978 and 1979 of the two parts of the long delayed sector development study carried out with German financing on the basis of terms of reference, which, in the meantime, had been substantially adjusted; the institutional part of the study was completed in 1979, the technical part, which had become an inventory of hydroelectric resources, became available in 1980; (15) - preparation in 1979, by ISA, of a power sector development master plan to the year 2000 including distribution in the systems of ISA's shareholders; this study was received in 1981; whereas the generation and transmission master plan remained essentially valid, if the main criterlun is development at least cost (see Chapter VII of this Volume), the distribution plan soon lost its significance, when financial shortages forced utilities to allocate most of their resources to the completion of ongoing projects in generation (16) - submission in 1979 of studies of the cost of service to final consumers and its implication for power pricing in Bogota, Medellin, Cali, and Barranquilla; this was the Bank's first attempt at inducing the Colombians, in the context of an operation, to consider economically efficient pricing; the Bank received drafts for the first two such studies in 1981 and further ones in 1983;(17) although the principle of pricing at economic cost was ultimately accepted, the measures taken after the studies were completed, unfortunately, fell woefully short of the goals set (see Chapter V); - 14 - - until mid 1979, preparation by ISA, assisted by its shareholders, of a study of losses in the respective systems; this was also the first time the Bank addressed the losses problem in the context of an operation in the Colombian power sector; ISA completed the assessment by early 1981, when it started a second stage; () however, the bare fact is that losses continued to grow in SUB, CVC, ICEL and CORLCA and remained constant in EPM (see Chapter IV); - preparation by the end of 1979 of management, planning, operations, and financial studies for ICEL and CORELCA subsidiaries; these studies were substantially modified in the course of time and ultimately resulted inter alia in proposals to regroup the subsidi-aries; (19) but, again, only little concrete progress was achieved in the wake of these efforts (see Chapter III); - submission by the end of 1981, of a study, which consultants would carry out on behalf of ISA, to define the requirements for coordi- nated dispatching in the interconnected system; (20) this led to the installation of a national dispatch center in ISA, in Medellin, and of three regional centers of which those in Bogota and Medellin are completed; and - execution by ISA and its shareholders of a study on the possible creation of a common system of accounts, an area in which improvement was urgentl needed but where, to this day, little progress has been achieved; ( ) 36. In 1977, ISA submitted a plan, which it modified in 1978, for the further expansion of generation in the interconnected system. This plan included, beyond the plants already decided earlier, the Tasajero steam plant (150 NW) in ICEL's sub-system, the extension of the Cerrejon coal-fired steam plant (158 MV) in CORELCA's subsystem, as well as the hydro developments Guadalupe IV (216 MW) and Playas (200 MW) in the EPM area, Betania (500 MV) in the ICEL system, Guavio (1000 MW) in the Bogota region, and Urra I and II (1050 MV) in the area of influence of CORELCA. All these plants were planned for commissioning between 1984 and 1988. In the course of 1978, construction and ownership of Guadalupe IV and Betania had already been assigned to EPM and ICEL, respectively. 37. In October 1979, the urgency to properly plan construction of new plant led ISA's shareholders to call a meeting in Call. Beyond this urgency, two factors dominated the discussions the heavy commitment of ISA in connec- tion with the implementation of Chivor II, San Carlos I and II, Jaguas, Zipaquira IV, and the 500 kV Iuterconnection, on the one hand, and the proven but, at the moment, allegedly under-utilized capability of EPM and EEEB, on the other. In hindsight, it seems that RPM had some free implementation capability at the time, as it had just completed Guadalupe II. However, this is more doubtful for EE that was just starting construction of Mesitas. More in the background, but evidently not less important, was the individual utilities' trend towards achieving a high degree of self sufficiency in terms of generation capability. - 15 - The result was the "Acuerdo de Calie, which for all intents and purposes distributed the responsibility for construction and ownership of the next program of generation plants among ISA's shareholders. Specifically, it defined inter alia thats (22) - ISA (i) plans development and operation of the interconnected system, (ii) constructs and owns further generation and transmission facilities, to maintain *a logical and normal growth in relation to that of the entire national generation capability, ' (111) coordi- nates the energy interchange as well as the setting of interchange tariffs; - EPM constructs and owns the Playas plant; - 5888 constructs and operates the Guavio plant, ISA participating to 40?; - ISA constructs Urra I and II CORELCA participating; ultimately CORELCA may own up to 50% of the facilities once it demonstrates its financial capability to do so; - CVC constructs and owns the next project in the southwestern part of the country and Calima II once this plant is assigned a definite place in the expansion program. 38. It is evident that the "Acuerdo de Cali' implied (i) a potential conflict with a development at least cost, as it would really be a coincidence if a least cost sequence would involve such a perfect regional distribution of projects and (ii) a substantial reduction of ISA's role as a major owner of generation and transmission facilities. In the previous ten years, the Bank had accepted several compromises that had weakened ISA's positiong however, accepting the 'Acuerdo de CaliB meant abandoning a very substantial part of the position won in agreeeants the Bank had reached with Government and the sector when ISA was conceived. Accordingly, the Bank reacted iamediately (October 26, 1979) by submitting to Government a counterproposal, which, while conceding that, at the time, ISA was over-committed, tried to preserve for ISA as much of the role initially envisaged as was possible under the prevailing circumstances. Thus, the Bank suggested inter alia that, in future, ISA would construct and own an average of 33% of the sector's hydro energy and capacity, in any event all plant of more than 500 MW or with a potential output of more than 2,000 GWh per year. Government answered after more than four months, on March 10, 1980, accepting some minor points but staying firmly or taking an evasive position on the main points. At that time, in the Bank, EPM's Guadalupe IV project was at the stage of preparation for Board presentation. As ownership was not an issue, the plant being of modest size (213 MW) and replacing an old facility of RPM, the Bank did not feel justified in making the satisfactory resolution of the wider ownership issue a condition for the further processing of the Guadalupe IV project. However, otherwise the Bank persisted in its efforts and, in late 1980, in connection with the preparation of the Guavio project, it reached an agreement with Government and sector which, by and large, was in line with its October 1979 proposals ISA would own 332 of generating plant overall. This was - 16 - to Imply the ownership of 50Z of Betania and Urra, with ICEL and CORELCA respec- tively, assuming the other 502 of the plants. In the Guavio case, the dis- tribution was to be 602 to 888 and 40 Z to ISA.(23) It is interesting to note that the arrangements envisaged for Betania and Urra very nearly corresponded to those that RPM had set forth in the early 19709 and which, at that time, the Bank had vehemently opposed. 1.7 The Period 1980-82: The Turn for the Worse. Power Shortages, and the Guavio Operation 39. In terms of the Colombian economy, the period was characterized by a sharp slowdown mainly related to dropping coffee prices. declining manufactured goods exports, and increasing petroleum Imports, and this, soon after 1979, when the second sharp increase in oil prices had taken place. Nevertheless, in early 1982, the Bank still projected a quick resumption of economic growth at 5.8% in real terms, which, however, did not materialize.(24) In the power sector the period was especially dramatic, as sector finances deteriorated sharply, electricity supply fell substantially short of demand in 1981 and 1982, and demand growth decreased from some 102 annually before 1981 to about two thirds of that value after the shortages. 40. In early 1980, ISA was again short of funds from its shareholders (i.e., these were in default with respect to the agreements with the Bank), a situation that the Bank recognised as alarming as it threatened to paralyze (and certainly slowed down) work on Chivor II, San Carlos I and II, and the 500 kV interconnection. (25) In March 1980, in the course of a seminar organized by ISA for all the entities involved in the power sector, a series of measures, including rate increases, contributions from Government, and the creation of new tools to mobilize financial resources, in particular, local ones, were proposed. Implementation difficulties, many related to political issues, permitted only partial solutions. Thus, within a short tine the sector finances, especially those of ISA, were again in a desperate state. This pattern dominated 1980. 81, and 82, in each year, the sector turned out a deficit far above 1986 US$200 million. (26) These same years saw the power shortages, which had been anticipated since the mid-70s, and which the sector had tried to avoid by a flurry of investments, become reality. Indeed, from the fall of 1980 to the end of 1983 (but mainly in 1980 and 1981), the sector found itself unable to meet temporarily up to 182 of the peak demand (which in 1981 was equivalent to about 3.5Z of energy sales). Most affected was the Bogota system and that of ICEL. The delays in the construction of the Chivor II, Mesitas, San Carlos, Zipaquira IV and V generating plants, and of the 500 kV interconnection line were the main reasons for the shortages (see also Chapter VII). However, a fire (in early 1980) that shut down the San Francisco plant (135 MW) in the CVC system for nearly two years and the unavailability of the Alto Achicaya plant (340 MN) for several months beginning in October 1981 contributed to the emergency. 41. The period 1978-82, i.e., the duration of the government of ;resident Turbay, was one of dramatic upheavals in the world economy (second petroleum crisis, start of a worldwide recession, and first signs of the debt crisis) the Bank did not prepare a new CPP for Colombia. During 1980-81, it operated by and large along the general lines established in the 1977 CPP. However, as the * 17 - "Acuerdo de Cali' had distributed the responsibility for the various generating facilities among ISA's shareholders, the Bank now lent exclusively to regional utilitiest in March 1980 US$87 million to EER for the Bogota Distribution Project, in June 1980 US$125 million to SPM for the Guadalupe IV Hydro Power Project, in March 1981 US$85 million to RPM for the Playas Hydro Power Project, in May 1981 US$36 million to CORELCA for the Village Electrification Project, and finally, also in May 1981 US$359 million, the largest loan made until then in the LAC Region, to EEEB for the Guavio, Hydro Power Project. During the same period, IDB made three loans to sector entitiess in August 1980 US$50 million to ICEL for its Rural Electrification Project, in December 1980 US$106.4 million to EPM for the Playas Hydro Power Project, and the third in June 1981, a US$173.8 million to BEE for the Guavio Hydro Power Project; thus, two of these operations were in parallel to Bank loans for the same projects. 42. Quite aside from the volume of lending, the Bank, in the period 1980- 82, did a major effort to get away from financing almost exclusively investment for generation. Indeed, with the loans to EEEB for the Bogota Distribution Project and that to CORELCA for the Village Electrification Project, it tried, more directly than possible through operations involving generation, to address the issues related to improved access to supply and to the limitation of technical losses through efficient distribution. In the case of the CORELCA operatign, additional objectives were: institutional strengthening and the replacement of old and unreliable diesel plants in then isolated systems by interconnections to larger systems. With respect to the CORELCA operation, the 1982 CPP already mentions that lending to weaker organizations like CORELCA required *considerably more time and assistance than normal.' The main question is whether, such resources were actually made available. The recorded staff time is certainly not a precise measure but at least indicative. It suggests that the Bank provided such assistance at best at about the same level as in other projects, as only the 'easy* Guadalupe IV and Playas projects show uses of personnel for project preparation lower than the CORELCA project (see Annex 1.3). About the same seems to apply to supervision work. 43. To understand the Bank's approach at the time calls for taking into account at least four pointes for one, it is evident that lending to indivi- dual regional utilities, in particular lending for distribution and village electrification, as in the case of one of the REEB loans and of the CORELCA operation, is not the ideal vehicle for the Bank to address general sectoral issues like plant ownership and ISA's finances, which are closely related to the tariffs in all the power companies. Then, it is understandable that the Bank did not wish to make the operations for EPM, the utility that was, rela- tively speaking, in the best shape and least delinquent as to its obligations towards ISA, dependant on the resolution of such issues. Third, Bank staff obviously was under a certain pressure, as the Colombians, who were experienc- ing sharp power cuts, were suggesting that the Bank with its protracted preparation of the San Carlos project was at least in part responsible for the occurrence of the shortages. To the Bank's defense in this point, it is important to mention that, even Colombian critics of the Bank's firm stand today concede that some of the measures adopted at the time, in particular the periodic tariff adjustments, would have occurred only much later, if it had not been for the multilateral institutions' intransigence. Finally, in the early - 18 - 1980s, it was increasingly evident that without Bank and IDB financing the sector's development was heading towards an Impasse, especially as foreign capital from sources other than the multilateral development banks was becom- ing scarce. 44. The above explains, but not entirely justifies the less stringent position the Bank took towards the sector during the bunching of six operations for US$655 million in less than two years (between June 1979 and May 1981). Indeed, in the context of the preparation of the San Carlos I operation, the Bank had asked for action before making the loan, and the sector responded, albeit slowly, but taking some very hard measures (e.g., tariffs, plant ownership, and induction of CORELCA in ISA). In 1980-81, in contrast, the Bank based its submissions to the Board much more on promises of action after approval, as it now required that such action take place before the Bank made the loan effective or even later. As more often than not, the measures took much longer to implement than foreseen, this resulted in a dramatic lengthening of the time elapsing between Board approval and loan effectiveness. Indeed, whereas the five loans approved between 1970 and 1978 were made effective, in average, six months after Board presentation, it took an average of 18 months to make the six loans approved in 1979-81 effective (see Annex 1.4). This has all the appearances of hasty lending. 45. A further characteristic of the five operations in 1980 and 1981 (the 1979 San Carlos II operation, which has to be seen mainly as a complement that for San Carlos I is left aside) is that their SARs, much more than those of previous operations fail to evaluate in a reasonably realistic way the actual situation of the sector and the prospects of the measures envisaged. Beyond the fact that they cannot state as facts the measures that in earlier operations were established before Board presentation (see para. 44), they project the image of an essentially sound sector that is afflicted with some major problems, which the Government, the sector, and the Bank are successfully addressing. Most importantly, the five 1980-81 SARs examined, and, in particular, that for the Guavio project, did not take the opportunity to convey to the reader the actual difficulties confronting the Bank in helping the sector implement the measures decided upon and the associated degree of uncertainty about the outcome. This is a characteristic of many Bank appraisal reports, but in the cases discussed here, especially Guavio, it seems more pronounced than in other projects. Annex 1.5 sets forth some examples supporting the concept that the image projected was not realistic, and that the facts known at the time support such an assessment. 46. As mentioned, the Bank's Board approved the Guavio operation in May 1981. In December, Bank staff reported that the project could only be completed one year later than envisaged in the SAR,(27) a fact that, even if it did not necessarily affect base costs, increased the total amounts requiring financing (calculated on the basis of assumptions in the Guavio SAR) by some Col$20 billion resulting from a 152 increase in project cost associated with assumed inflation and a 242 increase in interest during construction. It should be noted that such a delay, which by the early 1980s, the Bank had witnessed frequently with hydro projects in Colombia, was not considered in the sensitivitl analysis carried out in the context of the economic appraisal of the project.( ) Bank SARs normally do not present the results of sensitivity tests carried out in the framework of - 19 - the financial analysis, which is regrettable. In the case of Guavio it would have helped measure the very large risks involved. Unfortunately, the document does not even set forth a qualitative statement on the subject. At the same time, as the first delays became apparent, i.e. about six months after loan approval, a Bank supervision mission jdged the financing of Guavio as foreseen at appraisal Ototally unfeasible,( ) mainly because of the lack of local financing. This led to a postponement of loan signing to March 1982 and, ultimately, of loan effectiveness to October 1982, i.e., after the Govenment and the sector, in particular EEBE, had taken some tariff measures and had promised a series of further steps, which, as mentioned earlier, they were able to implement only in part (see e.g the temporary freeze on the tariffs starting in the last quarter of 1982), thus perpetnating, if not exacerbating, the difficulties. (30) It further seems that by mid-1982 DNP was already envisaging a two-year delay in project completion; however, it is not clear whether for technical or financial reasons and whether the Bank knew about it.(31) 1.8 The Post-1982 Period of Permanent Crisis 47. During the preceding period, as a consequence of the collapse of the coffee boom and the starting recession in world economy, Colombia's GDP had grown only at a rate of less than 2% per year. Inflation had again increased to some 25Z per year and the real exchange rate had appreciated by some 12% between 1980 and 1983. This had led to reduced competitiveness of Colombian products in the world market.(32) In the power sector, these years brought supply shortages essentially unrelated to the slowing down of the economy but masking a sharp reduction in demand growth very much related to the economic development. This new demand growth pattern only became evident in 1983-84. 48. In the same period (1983-84), Government relied on import controls to halt the outflow of reserves while maintaining public expenditures to preserve growth. When, in 1984, this proved insufficient, Government designed a program combining stabilization with growth. Most relevant for the power sector were the accelerated devaluation of the Colombian peso and the re-orientation of the public investment policy. Devaluation culminated in 1985 when it reached over 40Z, which dramatically increased the sector debt and its servicing. In terms of public investment policies, the Government emphasized export oriented activities with quick returns as against projects with long gestation periods in the field of non-tradables like power.(33) It seems crucial to note that Colombia achieved all the major objectives of this economic adjustment program. However, the measures created large problems for the power sector, 49. Mobilizing finance for investment in local currency had always been a problem for the power sector. In 1976, the Government established and fully financed the Electricity Development Fund (Fondo de Desarrollo Electrico-FDE) to address, in part, this issue. In the following six years, in the various financial crises the sector went through, FDE played a major role as a channel for Government finance. Also in 1976, in the wake of the *Acuerdo de Sochagota', Government envisaged setting up an institution which would mainly mobilize resources from the local capital market to provide local exchange finance for the power sector. After a first attempt at the creation of what was to become the National Electricity Fund (Fondo Electrico Nacional-FEN), aborted in Congress . 20 - In the late 19709, the Fund was organized In 1982 with strong support from the Bank, which, In the second half of that year appraised an operation with FEO as the Borrower. In the meantime, however, foreign financing had also become a problem because of world recession and the Latin American debt crisis. Indeed, foreign resources became scarce at the same time as the sector needs were especially high because of the large investment program, and because refinanc- ng requirements were growing fast. The latter was accentuated by the trend towards shorter term loans, last but not least, from the Bank (see Chapter VI). 50. In February 1983, ISA prepared an analysis of the sector finances which concluded that most of the measures taken in 1982, in particular mobilization f Internal credit through FED, contracting of additional external debt, and additional budgetary allocation by the Government had not provided the expected results, which, together with the sector deficit accumulated by the enid of 1982, was darkening the outlook, (34) in spite of substantial tariff increases in real terms.- (35) The SAR9 for the FEN and the Rio Grande operation with ZEM, though they project a less bright picture than those for earlier loans, still do not convey the impression that the sector was in nearly constant crisis and that experience had proven that one could only be cautiously optimistic about the prospect for radical Improvement. As a case in point, the SAR for the PEN operation (dated March 5, 1984) calculates with a growth of sales of 13.8Z for 1983. This is not only about 2.5 times the growth that actually occurred, (36) but it also indicates that, by late 1983, i.e. the time when the SAR had to be essentially finalized, the Bank was not aware of the actual development, although, by then, the sector must have noticed that sales were growing at a pace substantially below that agreed with the Bank as the most likely to molterialize. This raises the question, how well informed the Bank was about major develop- ments in the sector. Further, reading the voluminous SAR it is easy to stay unaware of the fact that one major purpose of the FEN operation was to finance the then still uncovered costs, in particular the large cost overruns due to unexpected geological difficulties in the Mesitas project. The SAR for the FEN operation, characterizes ICEL and CORELCA as the weakest utilities of the sector and their outlook as *unpredictable'I and Ouncertain,'0 respectively. (3 ) However, the document (like earlier ones) does not convey the crucical fact that the two companies, share of the market, which was at some 30Z in 1970 reached about 402 in 1986 and was still growing. Therefore, ICEL's and CORSLCA's weakness, especially in financial performance, was weighing ever more on the sector as a whole. 51. in the years following the Bank operation for which it was the Borrower, FEN was not able to play the role it had been assigned. In particular, it was, largely for reasons beyond its control, unable to mobilize a substantial amount of local funds; thus, it became a channel for foreign finance. As to monitoring sector finances, a task it was also assigned, it may have fulfilled it towards the Government, but there is no Indication that the Bank ever received the annual reviews FEN had agreed to prepare. 52. There is an incongruence between the picture projected in the SAR and the fact that about two months after the Bank's Board had approved the FM operation and one month after it had agreed to the Rio Grands loan to EPM, Government had again to call upon the main actors of the sector to discuss with * 21 - them the financial crisis and the ways out of it. This led to the "Acuerdo de Paipag of July 29, 1984. This agreement has to be seen in the context of the general austerity plan Government was Implementing at the time (see para. 48). It called for (i) an optimization of the sector's efficiency, reducing losses, and limiting operating costs and investment to the minimim consistent with a reasonable functioning of the sector, (ii) the fulfillment of ISA's shareholders' obligations towards the company, (iii) the institutional strengthening of ISA and PEN. It is important to note that the agreement explicitly mentions that tariff adjustments cannot represent the sole solution to the sector's woes, as the Government reserves the right to adjust the unavoidable increases in accordance with the equity policy it pursues.(38) The implementation of the "Acuerdo de Paipa* again took more time than foreseen, in particular with respect to the measures to strengthen ICEL and CORELCA and the budgetary contributions to the sector. This has to be seen In the light of the restrictive public investment policy mentioned above (para. 48). The main changes that occurred in 1985 were the increase of the Government's involvement in ISA through the assignment of two seats on ISA's Board of Directors and the revision of the sector investment plan on the basis of an assumed demand growth of 6.5Z per year, which resulted in the elimination of the planned thermal plants and the postponement by two to three years of several large hydro plants in particular Calima, La Miel, and Urra. Guavio, which in the meantime had run into land acquisition, technical, and labor difficulties, the latter triggered, inter alia, by the loss of more than 200 lives in a landslide at the project site.(39) 53. During the processing of the FEN operation two issues related to the power sector increased in prominences (i) the sector approach which, after having taken center stage in the context of the preparation of the San Carlos loans in the second half of the 1970s had been less in the foreground when, through several loans in the early 1980s in the wake of the *Acuerdo de Cali,* the Bank addressed more utility specific issues and (ii) the embedding of the power sector in the energy sector and the latter in the macro-economic context. The increasing awareness of the linkages between the economy and the power sector was associated with the preparatory work for the Sector Adjustment Loan, which started in 1985 and ended with Board approval in November 1987. This awareness is also reflected in the fact that the 1985 Trade Policy and Export Diversifica- tion Loan (Loan 2551) included conditions about the power sector revenues. Incidentally, it is unclear whether, at the time, the Projects Department was fully aware of these conditions in any event, communications channels between the departments interested do not seem to have worked at their best in this period. 54. The integrating effort had already started in the late 1970s, when Colombia carried out, mostly through local consultants, but with assistance from the Federal Republic of Germany, the National Energy Study (ENE), for which the report was published in 1981. The study had, inter alia, set the most likely electricity demand growth in the medium term at 82 p.a., a figure which the sector rejected outright. Later, there was an attempt to make the study of the energy sector a permanent venture. This had no success. As, in the first half of the 1980s, most efforts to improve power sector operations in a wider context had failed, especially because of the lack of an integrating structure, the Bank tried a new approach by inviting current and former ministers, congressmen, * 22 - current and former heads of utilities and of coal and oil companies, representa- tives of IDB, etc., to discuss sector issues in the framework of a three day workshop that took place in June 1985 in Santa Marta. The emphasis was on identifying objectives, issues and their relationships, away from the day to day conflicts the sector was experiencing. The participants then examined possible solutions starting from ideal concepts and subsequently considering con- straints.(40) Although most praised the Bank for taking the initiative and highly valued the learning experience, little follow-up took place. However, in the framework of the joint UNDP/World Bank Energy Sector Management and Assistance Program (ESMAP), Bank specialists and consultants p,epared the basis for the formulation of an energy policy. (41) The study report included valu- able conclusions about the power sector issues. The Power Sector Adjustment Loan tried to address the most important ones. 55. The ambitious program supported by the Power Sector Adjustment Loan (Loan 2889-CO) of US$300 million to Government pursues the following main objectives:(42) - developing a strong planning and coordinating capability within ISA, - progressing towards a broad regulatory framework with well defined rules for investment decisions and operation enhancing the coordina- tion within the sector, with the rest of the energy sector, and with the macro-economy, - providing a policy framework for efficient operation on a sound financial basis, - increasing the awareness of social and environmental concerns, and the capability to deal with them, - ensuring maximum efficiency in all utilities, and - achieving rates reasonably in line with long run average incremental costs. 56. The main measures to achieve the above objectives aret - The creation of an Energy Board formed of the Minister of Energy and Mines, the director of DNP, and representatives of the energy agencies. This body, on the basis of its own staff's investigations, would set policies, regulate the use of energy resources, and, in particular approve investment plans for the various subsectors. A committee reporting to the Ministry of Energy and Mines would monitor on a continuous basis all the power sector utilities. - ISA would have the authority to own, construct, and operate all future major generation and transmission plant, except for the Urra development, long before assigned to CORELCA. * 23 - * The sector would carry out a reduced but still balanced 1987-90 investment program in generation, transmission, and distribution costing about US$2 billion in current terms and submit before the second tranche of the loan is made available a draft plan for the investments required up to the year 2000. - The sector would implement an action program for improving management of NEEB, ICEL, and CORELCA on the one hand and sector finances on the other, the latter through (i) tariff measures both as to level and structure, (ii) equity injections reducing the debt equity ratio, (iii) lengthening average term of debt, and (iv) improved financial discipline. 57. By mid-1988, progress was disappointing, but the Bank was at least in a position to make the loan effective. Only in 1989, progress in loss reductions and the passage of the law creating the Energy Board signaled a possible turn- about. However, in the review team's judgment, the Bank has again been far too optimistic, if not outright unrealistic in evaluating the chances that the various actors in Colombia would be able and willing to implement the program as foreseen. Indeed, most of the main issues that have weighed down on the sector since 1970 and even before, continue to loom large over it. The following chapters will attempt to draw from the experience of the past eighteen years some more specific conclusions that, hopefully, will contribute to the Bank's future approach to helping the sector tacke its persistent problems. ENDNOTES (1) Operations Evaluation Reports Colombia; Volume III - the Power Sector, p.11. (2) CPP of December 27, 1971. (3) Memo Roberto Chadwick, Loan Officer, to Files of August 10, 1973. (4) Memo Roberto Chadwick, Loan Officer, to Files of August 10, 1973, para. 4... (5) Memo Mr. Z. Vessels, Division Chief, Projects, to Mr D. Knox, Director, Projects of June 3,1975. (6) PPAM for Power Interconnection (Loan 575-Co) and Chivor Hydroelectric Projects (Loan 681-CO) of October 29, 1979, para. 30. (7) Memo of December 13, 1977, from Mr. E. Vessels, Division Chief, LAC Projects, to Mr. R. H. Sheehan, Assistant Director, Energy, Water and Telecomunications Department. - 24 - (8) Francisco Javier Ochoas Incidencia do los Problemas Financieros en los Racionamientos do Energia Zlectrica, Lecturas de Bconomia No.19. Medel- lin, enero-abril do 1986. pp. 64-69. (9) Ministerio de Minas y Energias Acuerdo de Sochagota Sabre Politica de Emergencia en el Sector Electrico Nacional, (officially undated but probably of September 26, 1976). (10) C?P of January 7, 1977, para, 10. (11) See, e.g., Memo from Mr. Chadwick, Loan Officer, to Files of August 19, 1973, and memo from Mr. Z. Veasels, Division Chief, Projects, to Mr. D. Knox, Director Projects, of June 3, 1975. (12) See inter alia memo Mr. Wessels, Division Chief, Projects, to Mr. Wyss of December 30, 1977, and memos from Mr. Favilla, Division Chief, Programs, to Mr. Lerdau, Director, Programs, of January 25, 1978, February 3, 1978, and May 3, 1978. (13) Aspectos del Desarrollo del Sector Electrico y Analisis del Racionamien- to en el Suministro de Energia, Enero de 1982, Chapter 2. (14) Memo Mr. Favilla, Division Chief, Programs, to Mr. Krieger Vasena, Vice President, LAC, of November 8, 1977. (15) SAR San Carlos I and 500 kV Interconnection Projects, May 17,1978, pares. 1.32 and 1.33 and SAR Playas Hydropower Project, February 9, 1981, para. 1.10. (16) SAR San Carlos I and 500 kV Interconnection Projects, May 17, 1978, paras. 1.24 and 8.02, and SAR Playas Hydro Power Project, February, 8, 1981, para, 1.10. (17) SAR San Carlos I and 500 kV Interconnection, May 17, 1978, paras 1.30 and 8.02, and Letter Mr. Linder, Division Chief, Projects, to ISA of July 9, 1981. (18) SAR San Carlos I and 500 kV Interconnection Projects, May 17, 1978, paras. 3.02 and 8.02, and SAR Playas Hydro Power Project, February 8, 1981, para. 1.10. (19) SAR San Carlos I and 500 kV Interconnection Projects, May 17, 1978, para. 8.02, and SAR Guavio Hydro Power Project, May 6, 1981, para. 1.12. (20) SAR San Carlos II Hydra Power Project, May 30, 1979, pares. 4.06 and 7.01. (21) SAR San Carlos II Hydra Power Project, pares. 5.28 and 7.01. - 25 - (22) Memoria de la Junta Directiva de Interconexion Electrica S.A., (October 1979, no official date). (23) Telex Mr. Favilla to Government and sector entities, dated October 22, 1980, confiring agreements reached earlier between ISA, its share- holders, Mr. Linder, Division Chief, Projects, IBRD, and Mr. Camp. Division Chief, IDB. (24) Economic Position and Prospects of Colombia (in Two Volumes) Volume Is Main Report. September 17, 1981, para.xxv. (25) Francisco J. Ochoas Incidencia de los problemas Fluancieros en los Racionamientos de Energia Zlectrica; Lecturas de Sconomia. No. 19. Medellin, enero-abril 1986, p. 70. (26) Francisco J. Ochoas Incidencia de los problemas Financieros on los Racionamientos de Energia Electrica; Lecturas de Economia. No. 19. Medellin, enero-abril 1986, p. 74. (27) Letter Mr. Linder, Division Chief, Projects, to Z8EB of December 21, 1981. (28) SAR Guavio Hydro Project, May 6, 1981, para. 6.05. (29) Supervision Report of December 9, 1981, by Mr. Vietti. (30) Francisco J. Ochoas Incidencia de los problemas Financieros on los Racionamientos de Energia Electrica; Lectures de Sconomia. No. 19. Medellin, enero-abril 1986, 75 to 77. (31) Back to Office Report of May 10, 1982, by Mr. Vietti, Financial Analyst. (32) Colombia: Country Economic Memorandum, October 15, 1987, para. 1.06. (33) Colombia: Country Economic Memorandum, October 15, 1987, pares. 1.07- 1.09. (34) Francisco J. Ochoas Incidencia de los problemas Financieros en los Racionamientoo de Energia Electrica; Lecturas de Economia. No. 19. Medellin, enero-abril 1986, p. 78. (35) ISA* Evolucion de las Tarifas de Energia Electrica, Periodo 1971-1986, Figura No. 1. (36) SAR Power Development Finance Project, March 8, 1984, Annexes 1.3 and 1.4. (37) SAR Power Development Finance Project, March 8, 1984, pares. 5.07 and 5.08. (38) Acuerdo do Paipa, July 29, 1984, resulting from a meeting of the Mini- ster of Mines and fnergy and the Minister of Finance with representati- ves of DNP, FBN, ISA, EBMB, PM, ICIL, CVC, CORLCA, and CHEC. (39) Back to Office Report of August 13, 1983, by Mr. Plusquellec. (40) Memo of June 14, 1985, by Me. T. Sato, Senior Loan Officer,,to Mr. M. Schloss. Division Chief, Programs. (41) Joint UNDP/orld Bank Energy Sector Management Assistance Programs Basis for Formulation of a Colombian Energy Policy; a Report by Consul- tants; Bogota, December 1986. (42) President's Report on Power Sector Adjustment Loan, November 10, 1987, paras. 81 to 84. - 27 - II. GENERAL ISSUES 2.1 Introduction 1. On many occasions, OED's analysis of the Bank's association with the Colombian power sector raised, and in some cases shed light on, issues, which have to be seen in a wider context or which are likely to appear, possibly in a similar way, in other sectors and other countries. The present chapter sets forth some of these issues, as they appear in the context of the present study. It does not deal with them in detail. It rather presents them as (i) contribu- tions to further analysis to be done at a different level, in a wider context, and (ii) as additional background for the discussion of the more specific issues in later chapters. These topics are: (i) the need for a broad vision and strategy, (ii) project vs. sector lending and the Bank's resources, (iii) the pervasive optimism of the Bank's presentation of programs and projects, (iv) the pressure to lend, (v) quality control, (vi) multi-objective power sector planning, (vii) documentation in the Colombian power sector and the Bank, and (viii) cooperation of IDB and IBRD in the Colombian power sector. 2.2 Vision and Strategy 2. This evaluation has involved interviews with dozens of Bank staff and decision makers along with their Colombian counterparts, the reviewing of the extensive Bank records related to the subject and an in-depth assessments of 14 of the 15 power projects appraised in this period. An overriding impression the review team is left with at the end of this process is that somehow the Bank's broad vision of what it wished to achieve in the Colombian power sector was clear up to the late 1970s but after this, it appears to be lost. Indeed, in the late 1960s and throughout the 1970s, Bank policy was guided by the idea, shared with Government and, to a large extent, the sector, of interconnection. This was a clear physical and institutional objective with significant benefits. The strategy to reach this goal was of the kind the Bank was well equipped to assist in conceptualizing and implementing. Accordingly, ISA was established and then provided with independent revenue sources through its ownership of the Chivor and San Carlos plants. However, in 1979, came the *Acuerdo de Cali*. From then on Government and Bank policy fails to reflect a broad vision. The Bank appears merely to finance what seems reasonable and allegedly part of a least cost expansion plan, without articulating a strategic context between the Bank, the Government, and the sector entities. This suggests a ma, -, perhaps more generally applicable lesson, i.e. that without first achieving broad agreement on the big strategic picture of where the sector is going, how it shall arrive there and what role the Bank is to play in getting there, lasting success will remain elusive. It seems that the 'Acuerdo de Cali* agreement should have been perceived to convey that, in the short and medium term, ISA could not achieve some of the objectives set for it in the framework of the strategy of the late 1960s. This applies, in particular, to the monopoly for construction and ownership of major additions to the generating system, which should have been seen as beyond reach. Therefore, the broad approach should have been modified at the time, at least in this point. The need for a renewed such vision seems - 28 - paramount to any further Bank involvement in the sector. Chapter III discusses how, such a new vision would then have to be translated into a comprehensive set of objectives and sub-objectibves with the related criteria. 2.3 The Pervasive Optimism In the Bank's Project or Program Presentation 3. During the investigations for the present analysis, the review team was struck many times by the fact that SARe and other documents addressed to Board and Senior Management of the Bank tended to project a pervasively optimistic view of project or program environment, of the difficulties that the proposed or ongoing operation is facing, of the risks involved, and of the time and cost required to implement political, economic, financial, technical, and institution- al measures identified as crucial to the success of the venture. Annex 1.6 sets forth some instances of such over-optimism, most associated with the admittedly extreme case of the unfortunate Guavio operation. Chapters IV, V, and VI set forth further examples. Many individual statements show only a bias that may well be defendable. Nevertheless, they often combine to project a seriously slanted picture. This, of course is easily recognizable in hindsight. But as, hopefully comes out of the analysis of the examples previously referred to, the situation, as it could have been perceived at the time the discussed documents were prepared, did not warrant the optimism displayed. In the case of the Bank operations reviewed here, the phenomenon was already present in the appraisals of the 1970s. However, it has become more prominent in the 1980s, i. e. during a period of clearly deteriorating sector situation and performance. 4. It is evident that, without a good dose of optimism, very little can be achieved in a context as difficult as that of the sector discussed here. Further, if, in an official document, the Bank overemphasizes some of the risks at hand, there is a non-negligible chance that this enhances the materialization of such risks. Z. g., stating that the excavation of tunnels in a certain type of rock is highly risky, provides ex-ante an officially recognized explanation for delays and thus may weaken the commitment to avoid such delays. Similarly, emphasizing that the environment prevailing in a given country is unfavorable for a needed tariff adjustment, is equivalent to adding an obstacle on the way to reach the objective. This stated, there is a flagrant discrepancy between a reasonably supportive and optimistic outlook and the utterly unrealistic presentation of the chances of successful implementation of many of the measures the Bank considered crucial In connection e. g. with the Guavio and the FEN operations. 5. The present review is not the place to discuss to what extent similar distortions appear in project presentations in other 3ectors and other countries, except that it seems justified to affirm that such biased submissions occur and, even, that they are not uncoamon. Therefore, this report presents the mentioned evidence as a contribution to the discussion of one aspect of the Bank's credibility. * 29 - 2.4 Macroeconomic and Sector Strategy 6. Developments in the Colombian power sector in the mid 1980s provide an illustration for a further issue that, in OZD's experience, crops up in other countries, too: i.e., the occasional inconcruency of policies the Bank helps pursue in one sector, e.g. the macroeconomy and those it endorses in others. In the case of Colombia, from 1984 on, the Bank supported Government's major macroeconomic program, combining stabilization with growth, a venture that provied highly successful. However, this program involved, inter alia, accelerated devaluation of the Col$ and re-orientation of public investment to ventures promising early returns (see Chapter 1, para. 48 of this Volume), both policies affecting the power sector in a major way, as the first was bound to increase dramatically the sector's debt expressed in Col$ and the second to limit the sector's access to Government funds. At the time of Board presentation of the FEN and Rio Grande II projects, the macroeconomic measures were either in the last preparation stage or already under implementation, as indicated by the fact that, in July 1984, Government and sector concluded the "Acuerdo de Paipa," which tried to adjust power sector operations to the austerity measures by then already well started. However, the two Bank operations in the sector did neither discuss, much less take into account, the possible implications for the power sector of the macroeconomic program, though such consequences could be foreseen as severe, both in terms of Government's willingness to provide funds for covering deficits and for allowing the sector to increase appropriately electricity rates. 2.5 Least-Cost Solution and Opportunity Cost of Capital 7. Chapter VII, in particular Section 7.2, discusses the principles applied in Colombia to select the most desirable sequence of generation and transmission facilities. The approach taken essentially corresponds to that endorsed by the Bank for many years, i.e. the determination, among alternative programs of power plants and transmission lines, of the sequence with the least present worth of investment and operating costs discounted at a rate reflecting the opprotunity cost of capital. In the 1970s, this cost, and, hence, the discount rate used across the Bank, increased from some 8% to about 11Z. In the early 1980s, in most developing countries--certainly in Colombia--capital became scarce, which, one would expect, would entail an increase in the rate used to discount cash flows. However, the Bank continued to encourage the entities preparing the projects for Bank Group financing to use a rate of 11Z. 8. It is evident that discounting the cash flows at, say, 152 would have led, in the case of the Colombian power sector, to least-cost solutions quite different from those that resulted from the use of the traditional discount rate. Indeed, it can be expected that these solutions would have included a high proportion of thermal generation rather than the sequence essentially composed of hydro plants that invariably came out of the computation using the traditional, low discount rate. 9. Experience like that in the Colombian power sector discussed in the present report seems not to be unique. This suggests that the Bank has to address the issue of the opportunity cost of capital in its full breadth and - 30 - depth and to provide staff, which presently is, in many parts, somewhat disconcerted Jn this point, with the clear guidance required. 2.6 Projiect and Program Definition, Project Versus Sector Lending 10. The period under review started with operations reflecting the Bank's traditional project lending approach, i. e. operations like those for Chivor I and Guatape II, which were centered in their physical component and which included institutional measures essentially related to the borrowing utility. Further measures aimed at broad sectoral improvements in this multiutility sector were few. In the mid-1970s, the preparation of the operation for the San Carlos I Hydroelectric Plant and the 500 kV North-Central Interconnection Line, which dominated the Bank's power sector work program for more than two years was already heavily sector oriented with an array of measures to improve sector performance through actions of Government on the one hand and ISA's shareholders, within ISA and individually, on the other. The protracted preparation of this operation was not only due to the Colombian's reluctance to meet Bank conditions but also to the sheer complexity of the set of measures identified as necessary to optimize the chances of success. 11. Whereas the San Carlos II and Mesitas operations largely built on the ground prepared in connection with San Carlos I, the Bank substantially changed its approach when it came to the five operations it processed in the years 1980 and 1981, all with shareholders of ISA, (two with ZEEB, two with EPM, and one with CORELCA), none with ISA itself. It seems that this departure from previous policy was dictated by two factst (i) the 'Acuerdo de Cali" (Chapter I) and (ii) the Bank's decision to address for the first time in Colombia distribution and rural electrification issues on a large scale. The two loans resulting from this latter approach (Bogota Distribution Project and Village Electrification Program) as well as the two operations with EPM (Guadalupe IV and Playas) were not good vehicles for achieving sector improvements. Therefore, the Bank's efforts towards such progress went into the fifth operation, that for Guavio. This was difficult because, though ISA itself was involved in the project, some of ISA's shareholders, in particular, EPM, were only marginally affected by it. 12. In the mid-1980s, the Bank adjusted its policy again, by trying the project and the sector approach simultaneously, the first in the Rio Grande II project with EPM and the Bogota Distribution II Project with EEEB and the second in the PEN operation and the Power Sector Adjustment Loan. However, in connection with the latter two operations, it is important to remember that they had substantial components oriented towards individual utilities, in particular EEEB, as they included large contributions to the financing of cost overruns in the Mesitas (FEN Operation) and Guavio projects (Sector Adjustment Loan). 13. The above suggests two conclusionst (i) The Bank, especially in the 1980., may have tried to accomplish too much at the same time and (ii) it may not always have used the adequate vehicle to reach the set goals. These conclusions lead back to those presented in Section 2.2 on the need for a strategy within which individual operations are assigned limited but coherent objectives. In the context of the present section, they call for a clear separation of sub-objectives that can be addressed through project lending from - 31 - those that need sector leading as a vehicle. If sector and project lending is considered at the same time, very careful analysis of possible conflicts between operations is called for. 2.7 The Pressure to Lend 14. During the period under review, Bank lending increasingly took place in the framework of a lending program, outlined in a Country Program Paper (CPP), which itself tried to reflect as much as possible, general economic and sector policies by defining overall objectives the planned Bank operations were intended to pursue. This introduced a measure of systematic, as it embedded Bank lending into the political and economic context of the country. However, it also brought the pressure associated with attempting to meet targets, which were usually set high. Indeed, the pervasively optimistic approach discussed earlier for project presentation recurred in project preparation, where the planned operations, even when they were part of the program to be executed in the year of the CPP, were often affected by a high degree of uncertainty as to project or program content, justification, ind preparation. However, in the case of Colombia, as in that of other countries, power projects are often considered to be easier to prepare than operations in other sectors and had to replace such operations that had suffered delays. The 1971 CPP explicitly foresaw this possibility (see Chapter I, para. 16). 15. It is impossible to define precisely when the essentially sound pressure towards predictability implied in the commitment to a program turns into pressure to lend *at any costa with its negative effects on the quality of operations. Nevertheless, the present review concludes that during the 1970s the approach essentially remained sound. This is documented by the extremely careful preparation of the San Carlos I project despite pressure from many sides, in particular Government. In contrast, OED judges that, in the early 1980s, the pressure to lend increased, as suggested by some of the shortcomings of the operations falling into this period. 16. The above might be construed as pure judgement by OED, based on one- sided evidence. Indeed, some Bank staff heavily involved in the operations of the early 1980s consider that there was no pressure to lead in the above sense. However, many others felt that there was. Even if such pressure did not exist, the least that can be stated is, that responsible Bank staff was singularly unsuccessful in dispelling the widely shared perception of pressure to lend. It seems irrelevant to try to go deeper into this issue in connection with past lending to the Colombian power sector. However, it appears rather important that it be squarely faced in the context of the reorganized Bank. 2.8 Quality Control 17. It is evident that the preceding sections raise the question of quality control during the past decade, far less so for the previous years. Indeed, the fact that the sector and individual borrowers, in most recent years were perennially in default with respect to major covenants or targets (in particular those defining the revenue standards and those concerning reduction of energy losses) does not seem to have led to a particular clo,e scrutiny of new projects * 32 - by units of the Bank outside the directly Interested departments, i. e. especially Central Projects Staff (CPS) and later Operations Policy Staff (OPS), the watchdogs of operation quality. Bank files reveal little about major interventions from this side, except for an Important contribution to the definition of the financial setup of PEN. Interviews with persons involved did not provide more detail. Therefore, it may be considered symptomatic that the Guavio operation was processed without a formal review of the Yellow Cover SAR and that this procedure doesn't appear to have generated the major storm that under normal circumstances would have been unavoidable. This is the more striking as there is plenty of evidence that, at the staff level, there was much uneasiness about the project, especially its financial aspects (see Chapter I. Section 1.7). 18. There is again little purpose in further analyzing the shortcomings of quality control in connection with power lending to Colombia, other than to state the obvious, 1. e. that OPS failed to pinpoint some of the major issues not picked up by regional staff, e.g. the significance of the distortions in tariffs, the lack of credibility of lose reduction measures adopted and of revenue increases agreed. Further, it certainly would have been OPS' role at least to identify the weakening of the Bank's insistence on credible Borrower action before Board presentation of an operation and to investigate whether the standards of compliance used in Colombia were the same as those applied elsewhere. 19. The above considerations raise the question to what extent the reorganized Bank is in a better position than .he pre-1987 institution (i) to identify the type of shortcomings of which the latest operations in Colombia have suffered and (ii) to make sure that it deals with its various borrowers in accordance with reasonably uniform standards. Now quality control essentially is the responsibility of the Country Department concerned and the legitimate question is whether quality control this near to production will be sufficiently imune to the pressures exercised by lending objectives, close contact with borrowers, etc. The question is particularly valid as former OPS staff felt that, even in its relative remoteness from individual operations, OPS fell short of being a neutral controller, unaffected by pressures. Whereas this report cannot answer these questions, it can raise them on the basis of the specific Colombian experience, as a contribution to a discussion that has to take place at a different level. 2.9 Data Base and Documentation 20. The review team, though it knew that it would be difficult to find all the information it would have liked to consult, was taken aback by (i) the degree of incompleteness of the Bank's files and (ii) the lack of reliable consolidated information on the power sector in Colombia. 21. In the Bank, the team found that many important documents, e. g. information referred to in the mandatory annex *Content of Project Files' of appraisal reports, are missing. Some could be found with the operational people responsible for the follow-up of still on-going projects; however, many could not be localized within reasonable time. OED staff has noticed this fact before - 33 - in the context of audits of individual projects, when, however, it is less striking than when, as in the case at hand, the review covers some fifteen projects. Further, after the recent reorganization, it is often even more difficult than in the past to track down certain documents. The study team strongly suggests that the responsible departments of the Bank take corrective action. It would seem logical to have at least the project file either complete from the moment the project is approved or. if this is not possible because part of the information is very bulky and should therefore not be copied without urgent need, the file should mention where the various documents omitted can be found. 22. In Colombia, it is evident that ISA and other entities involved in the sector are applying great efforts to compiling information both on the individual companies and the sector as a whole. However, until now, success has been elusive. Indeed, relatively often, even rather elementary data compiled by different organizations prove to be inconsistent. This ranges, just to give two striking examples, from the modest inconsistences in the historical figures of energy the various utilities sold to, or purchased from others, to the large discrepancies in the data on distribution investment supplied by various Colombian sources. The vast differences in accounting systems the companies use also represent a major obstacle to a clear sector view. All these factors have made planning and monitoring the sector diff i-cult. Within the Bank, they have, inter alia, contributed to an unusual number of inconsistences in the documents supporting the presentation to the Board of successive operations in the sector. It seems evident that, besides the large problems Government faces in the near future in the sector, it will also have to address that of building up a data base that is adequate for activities of the importance those in the power sector have in the national economy. 2.10 Cooperation with Inter-American Development Bank (IDB) 23. As set forth in Chapter I, during the review period, IDB's financial involvement in the Colombian power sector was at the same level as the Bank's, i. e. some US$2 billion. ORD has not analyzed in detail the relations between IDB and IBRD in the context of the present review. However, on the basis of the documents studied, it found that (i) as the Bank, IDB refrained from lending to the sector in the mid-1970s (see Chapter I, Section 1.5); (ii) relations were somewhat uneasy during the last phases of project preparation for the San Carlos I project in 1977-1978; and in recent years the flow of information, in particular concerning the Guavio project, has been quite helpful. Even if incomplete, this suggests a modest level of cooperation. 24. In view of (i) the limited success both institutions had since the late 1970s in the sector under discussion here and this with a total lending of about US$4 billion and (ii) the extremely difficult issues that urgently need a solution, which, realistically, can only be reached over time, OED suggests that IBRD and IDB review their collaboration in the power sector in order to create the environment for the most effective cooperation, reducing duplication to a minimum and using each others' strength to the maximum. It seems that a closer cooperation should be possible without jeopardizing each institution's integrity. Just as an example, and not more than that, there is little evidence that the - 34 - Bank made optium use of the fact that for the past several years IDB had an experienced power sector specialist permanently working at its office in Bogota. 2.11 Agenda for the Future 25. In connection with the Bank's future approach to the Colombian power sector, the most important of the general issues touched above is doubtlessly that concerning the needed broad vision and strategy. To OED, it seems high time to take a distance from the day-to-day emergencies in the sector and to outline a broad view of what the Bank, in cooperation with IDB, Government, and the sector wish to accomplish in, say, the next ten years. Then, It should translate it into a stratep7, in which all future operations would have to fit. This strategy, which would spell out principles that are negotiable and others that are not, would have to be agreed with IDB, Government, and the main decision makers in the sector as well as the authorities to which these report. Indeed, experience has shown that to assure that sector-wide measures in Colombia have a reasonable chance of success, requires a commitment from a a setup of organizations substantially wider than central Government. Planned operations, rather than being necessarily self-contained should complement each other by addressing limited but precise sub-objectives in line with the broader goals. This is particularly important for IDB and IBRD operations in the sector, as, in view of the resource constraints, the complementarity of operations supported by the two agencies becomes even more crucial than in the past. 26. Another high priority issue is the need for a reliable and comprehensive data base in the sector. This is the basis for any progress towards transparency of sector operations. - 35 - III. SECTOR OBJECTIVES AND INSTITUTIONAL ISSUES 3.1 Introduction 1. The present chapter first discusses the general sector objectives and the criteria to be applied to pursue such objectives. These objectives have been vague and have lacked the associated criteria. As set forth in Chapter II, these objectives should reflect a broad vision, shared by Government, sector, and the Bank (and IDB) on what the parties want to achieve in the long run and on the general route that they have to take to reach such goals. Setting such general objectives undoubtedly is a task for Government. The second main issue addressed is sector regulation. In this area, the chapter identifies the need for a setup which unambiguously (i) translates the general objectives into sub- objectives and criteria, specific for each sector entity, (ii) defines the interplay between sector entities, and (iii) monitors the sector entities' progress towards achieving the set sub-objectives, and whenever necessary, adjusts such objectives and induces corrective measures, without interfering with the day-to-day management of the entities affected. Finally, it analyzes the issues associated with the complex setup of the sector, in general and of ISA, in particular, and identifies avenues that might lead to an improved organiza- tion. 3.2 Sector Obiectives The General Oblectives 2. Throughout the period under review, the official general sector objectives have been in one form or another (1)s (i) Meeting the demand for electricity comprising existing demand, additional requirement of consumers already connected, and consumption of consumers to be connected; (ii) achieving an optimal use of resources to meet the above demand; and (iii) reaching financial self-sufficiency. These are about the same objectives most governments set for their power sectors. However, formulated in such general terms, they have limited meaning. Indeed, the main issue usually is how precisely the sub- objectives are spelled out and how single-mindedly the sector entities pursue them. In Colombia, a comprehensive set of such sub-objectives and associated criteria does not seem to have been formulated. 3. During the 1970s, neither Government, nor the sector, nor the Bank felt the need for more precise sector objectives and associated criteria, because the prevailing philosophy said that a perceived demand essentially had to be met, that the 'least cost development' was the most desirable vehicle to do so, and that gradual improvements in efficiency and in tariffs were achievable and would ultimately lead to a state that could reasonably be called self-sufficiency. In the 1980s, worldvide recession and specific sector problems brought about a series of critical developments in the sector, e. g. the 'Acuerdo de Cali* (Chapter I, paras. 37-38), the start of the recession and of the debt problem entailing a slowdown in new financing from the capital markets used by the sector (Chapter I, paras. 40-44), the difficulties associated with the implementation of the Guavio project (Chapter I, paras. 45-46) and the massive devaluation of - 36 - the Col$ (especially in 1983). This generated a series of emergencies in the sector, which called for lmediate measures and tended to remove broader long- term issues like the formulation of a comprehensive set of sector objectives reflecting a broad vision from first priority status. Nevertheless, in the context of the Power Sector Adjustment Loan, several steps towards such a basis for rational sector operation were taken. Meeting the Demand 4. The following table illustrates the magnitude of the development the sector went through in the review period in terms of coverage of population and demand: Table 3.1s Population Coverage Unit 1970 1986 Electricity generated per capita per year kWh 390 950 Installed capacity per capita kW 0.10 0.23 Population with access to electricity 2 45 60 Rural population with access to electricity 1 15 45 Although the above figures are not totally reliable, they show that growth was quite substantial. Indeed, taking into account that 1970-1986 was a period of fast growing urban centers and only marginally increasing rural population, the gain in overall coverage of some 15 percentage points represents a vast number of new connections and in excess of 2.5 million households getting access to electricity in areas previously not supplied. Had the investment program been more balanced between generation and transmission, on the one hand, and distribution and rural electrification, on the other, sector performance in terms of population coverage might have been even better, but certainly still woefully short of the utterly unrealistic goal of 952 coverage of the total population by 1980 set forth without qualification in the SAR for the Guatape II Hydro Project in 1972.(2) 5. Meeting the increasing demand of both, new consumers and consumers already connected, and doing this on the basis of criteria that e.g. did not address uneconomically stimulated demand due to a very low residential prices and took little account of affordability for the economy, has contributed to the main problems the sector is presently facing (see Chapters V and VI). Two issues have to be dealt with in this context: (i) determining the demand to be met, and (ii) deciding on how to meet it. 6. Chapter IV discusses in further detail the sector's spotty record in projecting demand, which suggests a familiar pattern: The sector submits projections, which the Bank, in many cases reduces, and the actual levels of consumption result even lower. This holds in particular for the 1980-84 dramatic drop in demand growth. The spread between projected and actual values would have been smaller, had the sector and Bank availed themselves of the projections that had resulted from demand models trying to reflect the influence of macroeconomic - 37 - parameters. Indeed, these studies envisaged for the mid-1980s, a growth substantially reduced (see Chapter IV) with respect to that projected by the sector. However, it could not have foreseen the virtual collapse of demand growth experienced in the 1980s. Only recently did the sector add to the extrapolation of past demand other methods of estimating future requirements. However, it still considers demand as an essentially exogenous factor. 7. Chapter VII looks into how the sector tried to meet the demand using the "least cost approach, which leads to the most desirable solution provided (i) tariffs are at long run marginal cost, (ii) the discount rate used reflects the opportunity cost of capital, and (iii) the most economic solution is considered most desirable. It is not unusual that none of these conditions is fulfilled. However, in the case of Colombia, the discrepancies are especially large and therefore the determined least cost development is not necessarily the most desirable solution, even if the third condition holds. However, as discussed in Chapter VII, there is a need for an approach to investment planning that is more holistic than the analysis used in the past thirty years. Optimizing the Use of Resources 8. Chapter VII discusses also this point in some detail, again in connection with the determination of the *least cost solution.a Here, it should be recalled that, until about 1984, neither Government nor the Bank emphasized the need for an explicit power sector policy consistent with a comprehensive, national energy policy, also explicitly formulated. Since the 1960s, the sector's planning was essentially based on the premise that whenever hydro-power is available, it was the obvious source of energy for producing electricity. This concept was reinforced by the sharp increases in oil prices in 1973 and 1979. However, in the 1980s, Government became aware, that this approach was too simplistic for the increasingly complex environment. Therefore, after the preparation, in the context of the Joint UNDP/World Bank Energy Sector Management Assistance Program (ESMAP), of the study 'Basis for the Formulation of a Colombian Energy Policy' (December 1986), Government has taken steps towards an integrated energy sector policy, which then would form the framework for a coherent power sector policy aiming at an optimal use of national resources.(3) The suggestion of a strict multi-objective planning goes one step further (Chapter VII). Financial Self-Sufficiency 9. The vast discrepancy between declared objectives and reality in the Colombian power sector is nowhere as blatant as in the case of financial self- sufficiency. Indeed, as discussed further in Chapter VI, the sector's internal generation of funds has been poor and has forced Government to provide large amounts in the form of equity and operation subsidies, etc. Such contribu- tions are estimated at exceeding US$1.1 billion between 1978-85. - 38 - Conclusions 10. The time has come to make the mechanisms of the power sector more transparent. One of the pre-requisites for such transparency is the more precise definition of the sector objectives and associated criteria. Assuming that such objectives remain in the broadest terms the three mentioned earlier, it is necessary for Governments - to spell out the principles according to which the sector decides to meet or not to meet demand; - to set the ground rules for using national resources in the power sector, within the broader framework of an energy sector policy, and the national economy; - to clarify the basic rules of commercial operation of the utilities, in particular what elements of costs revenues would have to cover; - to define the operations which Government would subsidize because they are financially not viable in the short run but have to be carried out in fulfillment of specific socio-economic policies (e.g. elements of rural electrification and transitory re-financing measures to put the sector back on a financially sound footing); - to spell out the principles of setting electricity tariffs, their relation to long run marginal costs and the amounts of cross-subsidies between consumer categories that will be tolerated; and Some of the above principles have already been formulated at one time or another. Government would have to reconsider them in the context of developing the comprehensive set of objectives and criteria advocated here. 3.3 Sector Structure and Regulation Changes in Structure 11. During the review period, the only major changes in the sector's organizational structure were (i) the creation, in 1974, of the Ministry of Mines and Energy (NME), which took over from the Ministry of Public Works the responsibility for the directly Government-owned utilities, ICEL and CORELCA; (ii) the replacement of the Ministry of Agriculture by DNP as the branch of Government sponsoring the multipurpose CVC; (iii) the creation of Financiera Electrica Nacional (FEN); and (iv) the induction of CORELCA as a shareholder of ISA. Changes in Sector Regulation 12. The first three structural changes listed above imply a modest strengthening of the regulatory function of Government in the sector. Indeed, MHE, through its Electric Energy Division, formulates national policies for generation, transmission, and distribution of electricityg(4) in the framework - 39 - of such policies, it issues regulations on the provision of electricity (5). However, the Ministry's resources to watch over the Implementation of the policies have remained limited and so did the Division's impact on the sector. The re-assignment of CVC from the Ministry of Agriculture to DNP is an improvement, as DNP plays a major role in sector regulation whereas the Ministry of Agriculture does not. Insofar as FEN is monitoring financial performance in the sector and its entities and advising the Ministry of Finance on the issues, it also contributes to regulate the sector. (6) 13. A much more important strengthening of sector regulation occurred, without structural change, through modification of the role of the Tariff Board (JNT). Created in 1968, JNT was, from the start, conceived as a regulatory commission, a role that it assumed only to a small part in the 1970s, when it essentially limited itself (and partly was limited) to analysing proposals set forth by the utilities, on the one hand, and to approving (or rejecting) them, on the other. Although, in 1974, it was allowed to set tariffs without a previous initiative from one of the utilities, it remained essentially passive until the early 1980s. In 1983, Government, based on an ISA study initiated by the Ministry of Mines and Energy,(7) formally assigned to JUT the formulation of tariff policy. JNT's subsequent work led in 1984 to Resolution 2545 establishing a uniform tariff structure. The next step in consolidating the role of JNT was Resolution 086 of 1986 which implicitly ratified the commission's capacity to set tariffs and prevented some utilities from further interpreting the tariff resolutions as setting only an upper limit for the rates to be charged. Despite the power-play associated with this controversy, JNT managed, with the cooperation of the utilities (even those engaged in the controversy), to establish itself as a central coordi-nating agency at the analytical level, fostering studies that can be used as a basis for defining common ground, and finding mutually acceptable plans for tariff reform. JNT's success in the power sector has led it to taking up tariff regulation in other public services such as water supply, sewerage, waste collection, and communications, a role it was supposed to assume from inception. Eneray Board 14. Based on conclusions of the study carried out in the framework of the ESMAP (Chapter I, Section 1.8), Government envisages that the main instrument to formulate, implement, and monitor concrete sectoral measures at the energy sector level would be the Energy Board which would essentially consist of the Minister of Energy and Mines, the head of DNP, and representatives of the national energy agencies, ECOPETROL, CARBOCOL, ISA, and PEN. It is evident that this body, for which Congress approved in 1989 legislation, will have to play a major role in power sector regulation. Regulation and Regionalism 15. Experience in the past eighteen years has demonstrated, that, the main reason for the weak structure of the sector and its regulation as well as for slow progress towards a stronger setup is regionalism, which plays a dominant role in Colombia. Often considered an utterly negative force, regionalism has its dynamic, creative side, rooted in *the wisdom of the place. I The reason for - 40 - its negative image, especially in the Bank, may reside, at least in part, in the fact that it is associated with diseconomies, which are to a good extent quantifiable, whereas the benefits are largely not, though they may not be less real for that. However, foremost, regionalism in Colombia is a fact, with which both Government and the Bank have to live, and, this being so, it seems especially important to tap the strengths of the phenomenon. Therefore, measures strengthening sector regulation, in order to succeed, will have to be well balanced to assure, on the one hand, a promotion of the national economic targets and, on the other, an adequate protection of regional interests. Conclusions 16. Despite undeniable progress achieved in the review period, the sector sorely needs a strong and coherent organization unifying--or, at the very least, firmly coordinating--the functions presently exercised in the field of power sector regulation by the Electric Power Division of the Ministry of Energy and Mines; DNP; JNT within DNP; FEN; and, to some extent, ISA. When the body supervising the power sector (i.e. the recently created Energy Board, a subdivision of it, or a similar entity) will be in place, it will have to exercise, inter alia, the following functionst - define for the individual utilities areas of autonomy as large as possible in order to encourage the deployment of initiative, entrepre- neurship, and regional dynamism to avoid endangering the pursuit of clearly identified national objectives; - strengthen sector planning and operational coordination, i.e. in the present setup, continue supporting ISA in its main tasks; - establish and oversee the decision making mechanism connecting technical and economic sector planning, financing, and project implementation; - set, in accordance with the general Government objectives for the sector, the operational and financial targets for the individual utilities, as well as associated incentives and penalties; - monitor operations, identify shortcomings, and agree with the utilities on remedial action; - set the tariffs in agreement with the general objectives set forth by Government as suggested in Section 3.2 above and to monitor their implementation; - oversee the creation and operation of joint ventures to construct and own major sector facilities within the limits established in accordance with the first point above; - link up with the proper Government institutions, in particular the Ministry of Finance, to make sure that Government contributions established on the basis of the sector's investment and operational - 41 - planning in accordance with the general Government objectives discussed in Section 3.2 above, come forth at the proper time, in the proper amounts; - define the rules of interchange between utilities and, in particular. of sharing surpluses and deficits within the sector. 3.4 RestructurinA the Sector Attempts at Consolidation 17. Making the power sector more efficient through structural adjustment and the strengthening of ISA has been a long-standing preoccupation of both Government and Bank. For the entire review period and the years iamediately before, one of the dominant ideas has been the consolidation of smaller utilities with larger ones into regional clusters. One first step into this direction was the creation of CORILCA in 1968 as the entity responsible for generation, bulk supply, and the coordination of power development in the Atlantic Coast region. CORELCA has been functioning since 1972 and in 1975 took over ICEL's controlling interest in the seven local utilities serving the region. (8) 18. In the Cauca Valley, after a vehement controversy on tariff issues between CVC and ENCALI was settled In 1975, the Bank, in 1978, seems to have had the Impression that the two utilities had agreed to join in setting up a new company that would be responsible for public electricity supply in the Cauca Valley. (9) However, this consolidation never materialized. 19. The need to strengthen ICEL's management and operations even after the separation of CORELCA, is formulated in the 1978 SAR for ISA,s San Carlos I project, which includes studies to that effect.(10) In 1980, CONPES, on the basis of DNP's proposals, approved a number of measures concerning ICEL. Most prominent among these was the redistribution of markets through merger of two or more utilities. In connection with the Bank's Guavio operation (Loan 2008) Government agreed to prepare an implementation schedule for the program which was to start by mid-1983. (11) However, to this day, the sector achieved little progress towards the re-grouping of markets decided by CONPES. Two reasons seem to dominate resistance against this measure: (i) local governments are unwilling to give up control of the utilities; and (11) the large utilities, which necessarily would have to become the core and driving force of the respective new groups, are reluctant to be forced into an association with institutionally and financially weak partners. 20. The redistribution of markets would create larger utilities with a better chance to operate efficiently and along commercial lines than the smaller present gelectrificadoras'. Ultimately, the remaining small, isolated systems could be grouped in a Government owned company similar to the present ICEL, only much smaller. It is evident that, first, the political will for such a move must be mobilized. But beyond this, as discussed in Section 3.3 on sector regulation, a more clear-cut sectoral framework has to be in place to make sure that the new groups with the large municipal utilities at their core have (i) the freedom to organize themselves properly and (11) a reasonable guaranty for the financial - 42 - viability of the cluster, which, in any event, would integrate some weak utilities. The lack of such a framework might well have contributed to the slow progress towards the more obvious groupings like the integration of the Cundinamarca and Meta gelectrificadorasO with EEB and that of the Antioquis utility with RPM. Conclusions 21. It seems that the re-distribution of markets, long planned but still not started, would be desirable, as on the one hand it would create the basis for increased efficiency overall and on the other, it would preserve a large component of regional identity and therefore have a reasonable chance of being politically acceptable. However, it would require a stricter regulatory framework, e. g. along the lines suggested in Section 3.3 above. 3.5 ISA's Structure and Role ISA and its Statutes 22. ISA's 1967 statutes set forth that the new utility, owned by 3ZEB, EPM, CVC, and ICEL, would plan, construct, own, and operate (i) the interconnection between its sponsors' systems and (ii) major new generating plant in the interconnected system, except for the facilities that were already committed at the time. The by-laws defined the financing of ISA's fecilities as proceeding from its shareholders' contributions to the local curre icy component, and from external loans for the foreign part of expenditures. They further prescribed that ISA would (I) sell the energy it generates at . ites sufficient to cover operating expenses, debt service, and dividends to the sponsors, and (ii) buy energy from supplying sponsors and sell to recipients *at cost." The Bank considered these by-laws generally acceptable. However, in connection with the first ISA operation, i. e. that for the 1968 original interconnection project, ISA agreed to have consultants study improvements, in particular, of the methodology for determining ISA's tariffs for energy transfers through its system and for energy it would generate. (12) The resulting 1974 consultants' report, which aimed at a broad review of ISA's statutes, set forth proposals one of which was particularly controversial. Indeed, it suggested that companies hose systems would be connected to the Central System should be integrated into ISA. This meant that, CORELCA would have to become a shareholder of ISA as soon as implementation of the North-Central Interconnection would have been decided. The main regional utilities resisted the implied weakening of their position in ISA's Board. Therefore, it took two years of negotiations until, in the 1976 2Acuerdo de Sochagota, the sector decided, in principle, on CORELCA's induction in ISA. Another two years passed until the precise changes in the by-laws were agreed among all the interested parties. The Controversy About Plant Assignments 23. It appears that in 1973 a first major departure from the 1968 principles took place when the sector agreed that EB, not ISA, would construct and own the 600 MW Nesitas plant (Chapter I, Section 1.5). The 'Acuerdo de Sochagota' made this assignment official. At the same time. it defined that RPM would - 43 - construct and own the Guadalupe IV plant. This was easier to justify, as the facility was to have an installed capacity of 215 XV, which did not necessarily make it a *major addition'. Moreover, it replaced an older EPH plant. 24. In 1979, the 'Acuerdo de Cali* (Chapter I, Section 1.5) went much further in destroying ISA's planned monopoly on major additions to the system. as it reserved for each of the main utilities (including ISA) at least one large plant of the next tranche of investment in generation. It also defined that, in future. ISA would construct and own new generating plant in such a way that it would 'maintain a logical growth' of its total capacity which would be *normal with respect to the national generating capacity". (13) The Bank, in a last ditch effort, succeeded in having this exceedingly vague clause specified to say that ISA's capacity should be one third of total capacity installed in the country. Most recently, in connection with the Power Sector Adjustment Loan, the Bank changed again its outlook back to its original view by setting as an objective, ISA constructing and owning all major plant additions, after Urra, (14) which the "Acuerdo de Cali" had assigned to the joint venture CORELCA/ISA. 25. In their 1974 study of desirable adjustments of ISA's by-laws, the consultants had proposed to relax ISA's monopoly in a different way. They suggested to permit the formation of consortia including ISA and regional utilities directly interested in specific plants to build and own these plants. At the time, the Bank opposed such a setup. However, in the 1980s, i. e. after the "Acuerdo de Cali", this type of development won general acceptance. It was applied in the case of the Betania plant, executed as an ISA/ICEL joint venture and of the Guavio facility, for which EEB shares responsibility with ISA. 26. The main reason for the above developments away from the original agreements on ISA, is to be found in the fact that ISA's shareholders, while increasingly accepting ISA in its role of sector planner and coordinator, never identified with the policy that assigned to it the role of sole builder and owner of future large power plants. Indeed, the sponsors felt and still feel that they have been forced under duress to accept such a policy. Accordingly, already in the 1970s, they did not miss an opportunity to strengthen their own individual position on this issue to the detriment of ISA's. Beyond considerations of straight regional pride and identity, they (especially REEEB and EPM) were the entities that essentially had the experience and the human resources to implement sizable power plants, while ISA was building up its capability. This issue came to the surface already in the context of the Chivor project, for which ISA, strongly supported by the Bank, ultimately assumed overall responsibility for the implementation but had to delegate important project management tasks to ZEEB. (15) That the regional utilities had untapped capabilities while ISA' s were over-committed was one of the main arguments used in 1979 as a basis for the 'Acuerdo de Cali* and seems to have been valid to a good extent. A further view, often expressed in the sector, has it that, on the one hand, each regional utility should have a high degree of self sufficiency and, on the other, if regional resources (e.g. hydro potential) are used, the region should reap a substantial benefit from their utilization. On the basis of these facts, it is - 44 - evident (4) that the development away from exclusivity for ISA in the field of major generation was to an extent predictable and (11) that the prospects of the attempt. In the context of the Sector Adjustment Loan, to return to this exclusivity, are less than promising. 27. The question is, whether a policy of joint execution and ownership should not be pursued more systematically whenever straight ownership by ISA is excluded. It is clear that such a solution has the disadvantage of compli- cating the sector's structure. However, it does not necessarily imply inefficiencies, if operation can be assigned to one of the owners with experience. The solution has the definite advantage of direct local representa- tion in the project through the participating municipal or regional utility. It further does not preclude ISA from becoming the sole owner of large plants in which all ISA shareholders have a substantial interest. Moreover, even if ISA owns a plant, this Implies, on the basis of ISA's present operating rules, an internal distribution among the shareholders In accordance with individual interest in the ultimate output. Finally, exclusive ISA ownership of future plant contributes little in the short and medium term to achieving the sector's three sector objectives associated with ISA's activities that have remained elusives the definition and execution of the most desirable generation and transmission plan; the economic dispatching within the interconnected system; and having the ISA bulk tariff reflect economic cost-. ISA's Planning Role 28. The high profile, which the issue of generating plant ownership has had during most of the period under review, has tended to screen ISA's consistent and successful development to become the agency responsible for sector planning in generation and transmission and for operating the interconnected system. More recently. ISA has added to its duties a coordinating function in the planning of distribution, in particular, in connection with the need to reduce both technical losses and illegal consumptions the details of such planning, logically, remaining with the individual utilities. In contrast to ISA's development in the field of generation, its progress in the fields of system planning and operating went essentially unchallenged, as its shareholders explicitly confirmed the company in this role in the *Acuerdo de Cali* (1979) and in the OAcuerdo de Paipal (1984). However, there are still substantial problems in this area, especially the fact that ISA had on several occasions to adjust its investment program to take into account political factors. This, by itself, is not unnatural, but, as discussed in Chapter VII, this political input comes in the midst of planning and impedes evaluating properly the cost, in the broadest sense, of accomodating politically motivated preferences. The need to clearly separate the various elements of decision making becomes even more urgent, should the sector adopt, in due time, a systematic multi-objective planning of the type suggested in Chapter VII, which deals with the various aspects covered by planning (economic, financial, environmental, political, etc.) at the same level. * 45 - Sectoral Issues and ISA 29. The present review identifies four broad sector issues closely related to ISA, its role, structure, and operations - the difficulties associated with ISA becoming, as planned at its creation in 1968 and again in the context of the 1987 Sector Adjustment Loan, the sole developer and owner of major additions to the interconnected system (paras. 23-26 above); - the shortcomings of the investment decision making process, In particular, in the field of major generating plant (para. 27 above and Chapter VII); - the failure of the sector to assign ISA's bulk tariff its proper role as an economic signal (Chapter V) and the related fact that ISA dispatches energy essentially on financial principles with little regard to economic considerations; and - ISA's poor resource generation and liquidity record even compared to the low level of performance required by the company's by-laws in the early 1970s and the higher but still modest standard agreed in the second half of the 1970s. 30. The above issues have three main institutional roots (i) ISA is a national organization in an eminently regionalistic environment and (ii) ISA is owned by the the main municipal and regional utilities, which naturally pursue their own interest, which, however, are not necessarily In line with those of the company they own. 31. As discussed above, the plant ownership issue is largely dominated by the first factor, i.a. even if ISA were not the utilities' own company, in the Colombian context, it would be exceedingly difficult to enforce the kind of monopoly envisaged. The second and third issues above are closely related to the fact that ISA is owned by the main other utilities. Indeed, it is only natural that the shareholders try to influence the decisions on investment in the direction in which they perceive their own Iamediate interests, which are those of their constituency, in each case only a limited part of the nation. A stailar conclusion is evident for the bulk tariff issue, as an adjustment of that rate to an economic level would trigger the need for retail tariff increases in the distributing utilities' markets, increases that, though they would improve their individual financial performance, would also result in strengthening ISA's financial independence from its shareholders--a condition which, judging from the past, they are unlikely to welcome. The fourth issue above, i.e. the poor financial performance with respect to the partly modest standards set, is also largely related to the fact that ISA entirely depends on payments by its shareholders for energy delivered and for services rendered and on contributions by these same shareholders. In an adverse economic environment it is always difficult for a company like ISA to receive such payments and contributions, as - 46 - the debtors will tend to meet their immediate obligations first and those towards a dependant company later. Regionalism and the fact that shareholders may, through vithhelding payments, pursue goals of their own, may contribute to the shortfall in performance. Conclusions 32. Whereas the preceding sections suggest that the Government and the Bank have not succeeded in inducing the sector to assign ISA the quasi exclusivity of building and owning the major additions to the interconnected system in generation and transmission plants, the fact still remains that, by 1986, with a share of 312 of the total energy generated in the country's public system, ISA was--albeit temporarily--the largest producer of electri-city in Colombia. Indeed, it generated about as much as the entire system did in 1971, at the beginning of the review perioc. ISA has also become the central planner of the sector and the central dispatcher of the interconnected system. 33. It seems fair to conclude that the compromise leading to the creation of ISA, which Government, the sector, and the Bank reached in the late 1960s and which they subsequently agreed to adjust, has reaped substantial benefits during the review period. However, the persistent shortcomings discussed above, in particular those concerning the bulk tariff and investment decision process, suggest that the stricter sector regulation suggested in Section 3.3 above, while helping may not be enough to solve the problems. Indeed, the time may have come for a re-examination of ISA's role and structure possibly followed by a re- allocation of some functions. In this context, it seems important to look out for solutions which, inter alia, accommodate as far as possible the following partly contradictory points: - There is an advantage in keeping central planning and execution (both in investment and operations) rather close together, as is presently the case in ISA, as this promotes a less theoretical approach than when planning is done in splendid isolation. - There is a clear need for shielding planning from direct political influence, as it is exercised now through ISA's Board, i.e. the political input into the decision process has to come at a different level and at a later stage than at present. - It would be desirable to reduce the prominence of the plant ownership issue by following a pragmatic approach, e. g. as suggested above (para. 27). 3.6 The Role of the Bank 34. Institutional issues have consistently been a major preoccupation of the Bank in its dialogue with Government and the sector. However, as suggested in Sections 3.2 and 3.3 above, during most of the period under review, the Bank felt that defining a detailed and consistent set of sector objectives and the regulatory framework to achieve these did not deserve high priority. While justified as long as interconnection was of greater import, this attitude, in - 47 - OED's view, may have to change. Indeed, such a change is already perceivable in the Sector Adjustment Loan, which, inter alia, promoted the creation of the Energy Board, for which Congress passed legislation in 1989. 35. The Bank has actively supported all the attempts at partial con- solidation of utilities discussed in Section 3.4, In particular the grouping of NEEB and the Cundinamarca and Meta utilities on the one hand and that of EPM with ICEL's Antioquia subsidiary on the other. 36. Finally, the Bank's and IDB's strong support was very much instrumental in ISA's substantial development in the 1970s and 1980s. Through their intransigence in the mid 19709, the two banks helped induce the integration of CORELCA into ISA, and, in 1979-1980, when the 'Acuerdo de Cali' curtailed ISA's role leaving the company with an ill defined part to play as a generating utility, the Bank's intervention was determinant in preserving a definite, albeit more limited, role for ISA. 3.7 The Agenda for the Future 37. In OED's view, the main sector wide institutional issues that Govern- ment, the sector, and the Bank need to address in the immediate future ares (i) sector objectives and regulation, (ii) sector structure, and (iii) ISA's structure and role. It seems crucial that the parties and other institutions playing a major role in the sector (in particular, IDB), should start a new dialogue on the above (and other) basic sector issues. A good basis for such a new approach might be the spirit of the 1985 Santa Marta workshop (Chapter 1, Section 1.8) during which precisely such issues were discussed. It is likely that a protracted iterative process will be required for the parties to reach a consensus on what OED considerv, a pre-requisite for a successful dialogue on specific issues and for collaboration in establishing a reasonably healthy sectors that is, a broad view of what the sector should accomplish in the future (Section 2.2). Candidness about the realities and a positive approach to regionalism, in particular the willingness to use its strengths, would be especially helpful in this context. 38. In terms of sector objectives and regulation, the aim would be explicitness and transparency of: (i) the goals at the level of the economy as a whole, the energy sector, the power sector, and the entities of the power sector (i. e.s Where is the sector going?), (11) the responsibilities of the sector entities for the achievement of the objectives at all levels (i. e.: 'Who does what?), and (iii) the principles directing the mobilization and alloca- tion of resources in the sector (i. e.: Who pays for what?). 39. In parallel with the above questions and in the light of the possible answers to these questions, the parties would have to address in a constructive way the issues related to sector structure, among them the creation of more efficient units within the sector, in general, and the redistribution of markets (Section 3.4), in particular. A candid analysis and discussion among all parties of the lack of success of the efforts to integrate (I) EEEB and ICEL's * 48 - subsidiaries surrounding it, (Ii) EP and ICL's Antioquia utility and (III) CVC and EMCAL1, could be a valuable first step towards a meanIngful dialogue on this partial Issue. 40. Despite the substantial progress achieved during the review period, the inefficiencies associated with ISA's structure and role are such that these points clearly need re-examination, which in some Instances might lead to the re-assignment of certain functions. ENDNOTES (1) Desarrollo del Sector Electrico 1978-1982, DNP, February 1962, Chapter 1. (2) SAR Guatape II Hydroelectric Project, December 19, 1972, para. 2.03 and Annex 7. (3) President's Report on the Power Sector Adjustment Loan, November 10, 1987, para. 38. (4) SAR San Carlos I Hydro Power and 500 k Interconnection Projects, May 17, 1978, para. 1.08. (5) President's Report on the Power Sector Adjustment Loan, November 10, 1987, para. 48. (6) President's Report on the Power Sector Adjustment Loan, November 10, 1987, para. 48. (7) Estudio Sobre la Unificacion de Tarifass ISA, 1983. (8) SAR San Carlos I Hydro Power and 500 kW Interconnection Projects, May 17, 1978, para. 2.16. (9) SAR San Carlos I Hydro Power and 500 k Interconnection Projects, May 17, 1978, para. 2.14. (10) SAR San Carlos I Oydro Power and 500 kV Interconnection Projects, May 17, 1978, para. 2.15. (11) SAR Guavio Hydro Power Project, May 6, 1981, para. 1.12. (12) SAR Interconexion Electrica S.A., Chivor Hydroelectric Project, May 7, 1970, paras. 3.01 - 3.07. (13) Acuerdo de Call (October 1979). decisions, para. 2. (14) President's Report on the Power Sector Adjustment Loan, November 10, 1987, para. 86. (15) SAR Chivor Hydroelectric Project, November 20, 1970, para. 5.09 and PPAN para. 28. - 50 - IV: ELECTRICITY DEMAND AND SUPPLY 4.1 Introduction 1. This chapter focuses on the analysis of a selection of issues related to electricity demand and supply. A brief review of the evolution of electricity sales, demand and generation across regional markets and in the overall sector during the review period provides the setting for discussing the first issue, i. e. the causes for the change in the structure of electricity sales. The second issue addressed is the performance of sales/generation forecasts relative to outcomes and the factors that have driven the divergences between forecasts and outcomes. The chapter then discusses the challenges posed by the growing problem of power system losses with particular attention on the shortcomings in the Bank's approach to this problem. Fourth, it assesses the cost to the economy of the power rationing experienced in 1981 and the degree to which OED ex-ante and ex-post estimates dramatically differ from those made at the time of Board presenta-tion of the Guavio project in 1981. Fifth, it presents an estimate of the cost of the surplus generating capacity the system will possess up to about 1993. It concludes with an overall assessment of the role of the Bank in connection with these issues and outlines possible directions for future action based on the lessons learned. 4.2 Evolution of Electricity Sales, Demand and Generation 2. From the perspective of power supply development, some important achievements have occurred in the review period. First, there was the interconnection in the early 1970s between the central and southern regions. Second, ISA began generating power in 1977 and by 1986 it covered some 31% of the total system's generation needs. Third, interconnection with the Atlantic region came about in 1984 so that all major systems are now interconnected. In this context, there is an ironical situation that has arisen. Due to the excess supply capacity situation that will exist from about 1986-1992/3 (see Section 4.7), there will be water spillage occurring in the system. One would expect, therefore, that there would be large-scale power and energy transfers to the CORELCA system so that a significant amount of CORELCA's thermal generation could be substituted with hydro resulting in further fuel savings. Though this interconnector (which was financed under Bank loan 1583) was originally designed to transfer up to some 960 M of capacity at 500 kV, it appears that the intertie has had to be operated at much lower capacity. This results in one of the only benefits of the excess hydro capacity--replacement of more thermal generation- -not being fully achievedl Indeed, the low level of energy transfer between 1984 and 1987 [e. g. in 1986 only some 850 gWh of energy (18Z of its needs) were imported by CORELCA] appears essentially due to two factors--first, the limited line capacity arising from 230 kV operation (and not the design 500 kV). When an attempt was made to operate the liu., in 1985 at 500 kV capacity severe voltage drops occurred. This raises questions about the adequacy of the line's original design and of the assumptions of the conditions under which the facility was expected to operate. The second, is the persistence of Government subsidies for natural gas used in power generation in the CORELCA system. This reduces the - 51 - of Government subsidies for natural gas used in power generation in the CORELCA system. This reduces the incentive for CORELCA to Import electricity from the central system. Thus, while water is being spilled in the central zone, natural gas is subsidized in the Atlantic system, an avoidable economic waste. The subsidies, which amount to some US$50 million annually in economic terms, are at variance with covenants undertaken by the guarantor under Loan 1583 that financed the Atlantic region interconnector. The Bank appears to have accomodated this non-compliance, apparently on grounds of the interconnector's limited energy transfer capability referred to above. This subsidy is discussed further in Chapter V, Section 5.5. 3. Throughout the review period hydro generation continued to dominate supply, being responsible for 77? of generation both at the beginning and end of the period, at which time total installed generation capacity was some 6,950 MW for a peak demand of 4,840 MW. Annex 4A.1 provides a breakdown of generation, sales, demand, and load factor data for the overall system between 1971 and 1986, with detailed growth rates by markets set out in Annex 4A.2. This is summarized in Table 4.1 below. There are three points of relevance. First, in 1971-76, growth rates of sales were high for all classes of users, as they averaged about 10.42 annually. This was a period during which the average tariff declined about 8.4? per year in real terms and economic growth was strong at 5.4?. The second period, 1976-1980, was also characterized by strong economic growth, 5.5? annually, but rising electricity tariffs, with average annual increases in real terms of about 7.5?. Sales growth was no longer uniform for all classes of users, while growth in generation exceeded that of sales, indicating an increasing level of power losses. The third period, 1980-86, coincides with that of the recession (average GDP growth of 2.6? per year) and continued increases in real terms of electricity tariffs for commercial and industrial users up to 1983 (after which real tariffs declined). Sales growth was highly uneven for the various consumer types, with commercial about 1/3 of residential. Moreover, losses continued escalating. Table 4.1: AVERAGE ANNUAL GROWTH IN SALES AND GENERATION IN TOTAL POWER SYSTEM, 1971-86 (Percent) Residential Commercial Industrial Total Period Sales Sales Sales Sales Generation 1971-76 10.3 10.6 10.4 10.4 10.5 1976-80 10.9 7.3 4.9 8.2 9.5 1980-86 6.3 2.3 5.5 5.2 6.0 1971-86 8.8 6.3 6.9 7.7 8.4 4. What is of significance is that, in the late 1970s, even when the economy was buoyant, there were signs of a sharp slowdown in the growth of sales, - 52 - especially industrial and commercial, which were experiencing rapidly rising tariffs in real terms. This slewdown was, however, somewhat masked in the case of generation by the fast rise of losses. 4.3 Dynamics of the Structure of Electricity Sales Background 5. A factor that has had negative effects on the financial performance of the power sector as a whole and on most of the retail power companies has been the changing structure of electricity sales. This section assesses the factors that have contributed to these changes; and the degree to which the financial consequences of these shifts were recognized and, if so, what actions were taken by the sector authorities and the Bank to address the problem. 6. In terms of the structure of electricity sales, the review period was characterized by: (i) major changes in the share of sales of the different regional power markets; and (ii) significant shifts in the sales to different classes of electricity consumers, both within regional markets and in the interconnected system. Shares of Retailing Power Companies in Total System Sales 7. The development of the shares of power compan.es in total sales is shown in Table 4.2 below. There are two relevant features. First, the major municipal utilities, EPM and EB, together accounted for about 51? of total system sales in 1971 compared to only 302 from the combined sales of ICEL and CORELCA. By 1986 these shares had become 42? and 41?, respectively. In other words, the two regional utilities, ICEL and CORZLCA, characterized tradition-ally as being financially weak and possessing *weak* markets were together as important in terms of sales as EPM and 1EEB combined--the companies popularly considered as creditworthy and as having Istrong' markets--this concept of a 1stronge retail market being defined as one which was spatially limited, possessed high load densities, and with industrialIcommercial sales having a very significant share of total sales. 8. This increased importance of ICZL and CORELCA and the relative decline of EPM and EEB meant that independently of any other developments the sector's financial performance was likely to deteriorate unless specific and concrete measures were taken to address ICZL and CORELCA's financial problems. When was this trend apparent? It is clear from Table 4.2 that by 1980, at the time of the Guavio appraisal, the trend was established and merely became more evident in the 1980-86 period. When did the Bank become aware of this trend? This is difficult to identify since it was not the type of issue the Bank's appraisals, or such sector work as was done, tended to illuminate, as such work tended to take a regional corporate rather than a sector view. * 53 - Table 4.2t Structure of Total System Electricity Sales, 1971-86, by Company Share Company Share of Total System Sales (Percent) Company/Year 1971 1976 1980 1986 BEB 27.3 27.1 25.9 23.5 BPH 23.8 21.5 19.7 18.4 zMCALI 12.2 11.3 10.4 9.8 CORELCA 13.2 14.9 15.3 17.9 CVC 3.4 3.7 4.0 3.9 ICEL (Group) 16.3 17.9 21.2 22.7 Total 1& 100.0 100.0 100.0 100.0 /a These do not add to 1002 because the small municipal companies, e.g. Armenia, etc., have been excluded. 9. ICEL's share of total system sales, driven by a high growth rate of new connections, increased by almost 402, from 162 in 1971 to 232 in 1986, a level virtually on a par with BEB, while concurrently that of EPH fell, from 242 in 1971 to a level in 1986 commensurate with that of CORELCA of some 182 of total sales. Shifts in the Type of Consumer Sales 10. The second perspective of viewing the changes In total system sales. -according to the type of user--is illustrated In Table 4.3, with Annex 4A.3 providing further details. The significant trend shown is the rise in the residential share of sales by 115 from 412 to 482 between 1971 and 1986, accompanied by a decline in the share of all other user types. In other words, an increased Oresidentialisation' of the electricity market when viewed at the aggregate sector level. Table 4.3: Structure of Total System Sales, 1971-86 by User Type (Percent) Sales CateioryYear 1971 1980 1986 Residential 41.4 45.4 48.4 Commercial 12.8 12.4 10.5 Industrial 33.6 29.6 30.2 Other 12.2 12.6 10.9 Total 100.0 100.0 100.0 * 54 - 11. This change in sales structure was especially marked in the regional markets of REB, ICEL, ENCALI, and CVC (propio),(l) as demonstrated In Annex 4A.4 and suouarized in Table 4.4 below. Table 4.41 Share of Residential Sales Within Same Regional Markets, 1971-86 (Percent) MarketiYear 1971 1976 1980 1986 EB Residential Sales Share 35 39 43 48 ICEL (Group) Residential Sales Share 44 47 53 57 EKCALI Residential Sales Share 35 37 40 44 CVC (propio) Residential Sales Share 39 38 47 51 What is evident is that in all of the above regional markets the increment In residential sales share between 1971 and 1986 has ranged from above 113, in the case of EMEB, to about 114 for SCALI. It was only in the RPM and CORELCA markets that sales structures have remained unchanged, but these markets combined only account for about 1/3 of total system sales. 12. The three main factors that have driven the rapid Oresidentialisa-tion" of the structure of electricity sales have been first, the high rate of addition of residential subscribers to the system in keeping with Government's social objectives of extending electricity coverage. Table 4.5 below shows the rates of change of residential subscriber additions and of electricity use per residential subscriber in the period. The subscriber addition rate has been very uniform throughout the period for the sector as a whole, averaging around 7.6? annually, resulting in some 3.5 million subscribers in 1986, or roughly 602 of households in the country. A dramatic Improvement compared to the 342 coverage in the mid-1960s. The second reason for the shift of consumption toward residential users has been the relatively slow growth of industrial value added, combined with the electric intensiveness of industry which has undergone little increase in the review period. The third factor was the massive level of price subsidization of residential electricity. Indeed, electric power has become the lowest priced modern household energy fuel (except for ecocinol* in Bogota), while having the highest economic cost for uses such as cooking. This has encouraged electricity use for cooking (Chapter V, Section 5.8). - 55 - Table 4.-n Residential Subscribers and Residential Electricity Consumption per Subscriber, 1971-86 (Average Annual Growth Rates) (Percent) Residential Electricity consumption/ Company Residential Subscribers Residential Subscriber 1971-76 1976-80 1980-86 1971-86 1971-76 1976-80 1980-86 1971-86 SEES 9.0 5.5 7.2 7.4 3.5 3.3 -1.5 1.4 BPM 6.0 5.5 5.8 5.8 1.4 0.1 -2.2 -0.4 ICEL (Group) 9.8 10.0 8.6 9.4 3.8 5.6 -0.9 2.4 CORELCA 6.7 8.0 7.9 7.6 0.1 5.8 2.1 2.4 CVC 6.4 4.5 6.4 5.7 5.0 10.7 0.4 4.6 ENCALI 3.1 4.4 7.5 5.1 6.8 3.3 -1.6 2.5 Total System 7.6 7.7 7.6 7.6 2.6 2.9 -1.2 1.1 13. Of some relevance is that despite perhaps one of the lowest average residential tariffs in the world (USj*2.4/kVh in 1986), average residential use per subscriber in the EPM market of about 400 kh/month appears to have reached saturation, as shown in Table 4.5. However, this intensity of use is about double that in the rest of the country,(2) indicating that such saturation is certainly nowhere near In the other regional residential markets. The rising Importance of the ICEL and CORELCA markets also means that, with their average specific residential electricity use much lower than in the RPM and EEB systems. more residential consumption was arising from the first and highest subsidized tariff block (0-200 kUh/month). This weakened further the residential revenue base and, in turn, increased distortions in tariff structures by giving rise to greater cross-subsidizes from commercial and Industrial users to residential ones. 14. Recognizing the modest rate of increase of specific electricity use by residences, but the very high rate of subscriber additions, it is estimated that about 80% of the increase in residential electricity sales in the review period have been driven by new subscribers. In contrast with the 7.62 annual rate of residential subscriber increase, that of industrial value added only averaged some 5.62 annually in the review period, which, when combined with virtually no change in the electric intensiveness of Industry in the period, resulted in a rapidly widening differential between the shares of residential and industrial sales in overall system sales. The electric intensiveness was measured as the ratio of electricity use by industry per peso of Industrial value added, which is shown in Figure 4.1 below, based on the estimates given in Annex 4A.5. This suggests that, barring a fundamental change in the electric intensiveness of the industrial sector, the 'residentialization' of the power market is likely to continue as long as the connection rate of new residential users continues to be well In excess of the rate of growth of Industrial value added. Figure 4.1s Industrial Electric Iatensiveness and Ratio of Electricity Generation TO CDP, 1971-86 COLOMBIA.POWER SECTOR Ra~s of Elec~dcity Requ d to GDP (R) and industdul Ekbc~ty Us to Industda Added Vu (R. Vesus Tine 197-46 1977 Power shortags test Power Shortage 730 10 1i97 72 73 74 75 78 77 78 79 80 81 82 83 84 85 3 Yer -m Begusoen unel. ,., u -61668.933.157 twaart unM1si fculMNti. R4. of tedesIna Eletdeny Us to tressat Valus AOiss t011ain9WUU Pes Actoal. 0.o ota mW UImeiy Resuld to Astual W0. t0tdiWMiUSO2 Peo - 57 - 15. It is clear from the above discussions that very significant changes have occurred in the structure of electricity sales, both from the standpoint of user types and by region. The increased level of national electricity coverage has achieved a major social objective; however, the accompanying heightened gresidentialisation' of the power market and rising share In total sales of the weaker regional utilities has, in turn, exacerbated financial difficulties at both the corporate and sector levels. Somehow, the rapidly increasing Importance of the ICEL and CORELCA markets never appears to have come clearly into the focus of either the sector authorities or the Bank. In many ways, this can be understood as prime attention was given to the higher priority of simply trying to meet demand. However, one of the consequences of this inadequate attention to the shifting regional sales was that, institutionally, these regional utilities were not accorded a level of support comensurate with their increased significance in the national power market. In contrast, by the mid-1980s, the Bank and sector authorities increasingly recognized the significance of 'residentialization, for tariff policy. Conclusions 16. Two significant changes have occurred in the structure of electri-city sales in the review period. First, sales from the traditionally weak regional utilities, ICEL and CORELCA, are now on a par with those from EPM and EEEB. Second, an increased 'residentialization" of the power market has emerged. These structural market shifts have come about because of the very high rate of connection of new residential users (7.62 annually in the review period), contrasted with a much slower rate of growth of industrial value added given that industrial electric intensiveness has not changed signifi-cantly between 1971- 86. What has also contributed to this shift in the market has been a tariff policy involving massive subsidies to residential electricity users which, in turn, has tended to encourage uneconomic electricity demand (Chapter V, Section 5.8). 4.4 Performance of Sales/Generation Forecasts Relative to Outcomes Background 17. Fundamental to any power sector investment program is the forecast of demand and sales underpinning the requirements for new supplies. Indeed, given the increased macroeconomic uncertainties that have prevailed since the late 1970s, future levels of demand have become one of the most important uncertain- ties confronting power system planning. This section, therefore, looks at the performance of the Bank's projections of power demand and sales growth; the factors that caused major deviations from these expectations; the consequences of such deviations; the point in time when the Bank and sector authorities became aware of the major demand collapse; the actions undertaken to deal with these new realities; and the lessons to be learned from these developments. 18. Analysis of demand projections Invariably focusses on the short-comings of such exercises while losing sight of the daunting problems faced in developing forecasts that can capture the essence of the unknown reality a decade into the future. Avoiding such a focus becomes especially necessary in the Colombian - 58 - context where energy demand structures were undergoing signifi-cant and different changes by regions where tariff levels and adjustments varied significantly by region; and where the effects of aggregating these regional trends were by no means clear. The forecasting methodologies used by the power sector in Colombia during the 1970s and then into the 19809 are reviewed in Annex 41.1, however, what is first assessed from an empirical ex-post perspective is what have been the most critical factors affecting the behavior of energy demand for the overall power system in the review period. Total Systems Sales and Generation--Outcomes versus Projections and Correlations with GDP 19. A broad summary of the deviations between actual and projected energy requirements and energy sales for the total power system for the period 1976- 1986 is set out in Table 4.6. These data have been derived from Annexes 4B.2 and 4B.3. The deviations call for three observations. First, the effect of the on-set of the unforeseen rncession beginning in 1981 is dramatic for projections made at the time of the San Carlos I and II operations (1978 and 1979). Before 1980 discrepancies between projected and actual energy require- ments and sales were well within forecasting uncertainties and generally less than 10Z. However, with the long recession from 1981-83/84, energy needs did net recover, resulting by 1985 in the deviations between actual and projected quantities at the time of San Carlos II amounting to between 24Z-312. Of course, the major sector investment program for the 1980s was predicated on the demand projections accepted for San Carlos I and I operations, which at that time were high but not unreasonable. Second, of the three key variables forecasted, deviations were consistently the lowest for energy requirements, followed by peak demand (not shown in Table 4.6) and the highest for energy sales. Generally, the peak demand was not independently forecasted but derived from a projection of the system load factor. This was one area (and indeed the only one found by OED) where the Bank's SARs erred on the side of a conservative projection, whereby the forecasted system load factors were worse (i. e. lower) than what actually materialized. In the case of sales, the deviations were between 114 to 113 greater than for energy requirements with this arising solely from the increasing divergence between actual escalating system losses compared with projections of decreasing system losses (see Section 4.5). This contributed to depressing actual sector revenues by 1985 by about 10? below what they would have been had energy losses in the overall system behaved as projected in the San Carlos II and Guavio operations, i.e. remained roughly constant. 20. Finally, for projections made at the time of the FEN operation there are no important deviations in regard to peak demand (not shown) or energy requirements, however, in the case of sales the discrepancies are again - 59 - Table 4.6s Total Power System Deviation Between Forecasted and Actual Energy Requirements and Sales, 1976-1986 (Percent) San Carlos I San Carlos II Appraisal Appraisal Guavio Appraisal PEN Appraisal Year May 1978 may 1979 May 1981 March 1984 ( Deviation) /a (Z Deviation) /a (Deviation) /a (Z Deviation) /a Energy Energy Energy Energy Require- Energy Require- Energy Require- Energy Require- Energy ment Sales ment Sales ment Sales ment Sales 1976 -3.4 - - - - - - 1977 4.2 - - - - - - 1978 2.0 - 3.6 5.8 - - - 1979 3.6 - 4.3 8.2 - - - - 1980 7.0 - 6.5 12.1 0.8 4.0 - - 1981 15.4 - 16.4 20.2 10.5 12.8 - - 1982 15.0 - 16.2 22.5 12.4 15.6 - - 1983 17.6 - 18.1 24.6 15.9 22.1 0.0 8.8 1984 19.2 - 19.8 27.3 18.4 24.9 0.7 11.4 1985 22.8 - 23.8 31.3 22.4 30.1 3.6 14.4 1986 - - - 24.5 31.3 3.8 15.5 /a Deviations are expressed as a percentage and equal to 1 - ratio of actualiforecasted quantity. significant by 1985/86--only one year after Board approvall In this case, the problem appeared to be the starting point of the projection--19831 In some inexplicable way (FEN SAR, Annex 1.4), the PEU operation (appraised in 1983) considered system losses in 1983 to be around 18?, whereas such losses were actually 24? in 1982 compared to 22.5% in 1980--i.e. actual losses were on a rising trajectory between 1980-82, yet with no specific and credible loss reduction program targeted, the SAR reduced losses by 1/3 In 1983 and then projected a declining trend for the rest of the period. This, in turn, implied that even the meager tariff adjustments advocated for the sector in that operation (see Chapter V, Section 5.6) would have had to have been increased by a further 10Z in real terms above the SAR levels (i.e. virtually doubling the proposed SAR sector tariff increases) if the correct energy loss levels were used for 1983. 21. Of the many factors that exert an influence on electricity demand, what are those that appear ex-post to have been the most important during the review period? To answer this, OED looked at several models, discussed in Annex 4B.4. The best explanation of past trends, at the overall sector level, was provided by the simplest model, which assumes electricity demand growth exclusively driven by GDP. In other words, the ratio 1 of total electricity demand to GDP has been - 60 - a linear function of time. Figure 4.1, shown earlier, also illustrates the behavior of the energy ratio, Rl, between 1971 and 1986, as well as the regression line, with values of the ratio, Rl, given in Annex 43.5. As can be seen from this figure, the monotonic Increase of the ratio has been unaffected by recession, energy shortages (other than the declines in the year of the shortages) and increases or decreases in power prices at the aggregate sector level. What is also of significance are comparisons of actual GDP and forecasts made at periodic intervals in the Bank's Economic Reports, given in Annex 43.6. A broad summary of these deviations between actual and projected GDP is given in Annex 4B.7. Of importance was the projection of GDP made in the June 1979 Economic Report, which formed the basis on which power generation requirements were determined for the overall power system expansion plan of the 1980s--in particular, at the time of the San Carlos II and Guavio appraisals. Indeed, by 1985, the deviation between actual GDP and that projected in the 1979 Economic Report was some 221, which was roughly the same level of deviation between actual energy requirements, for the overall power system, and those projected in the San Carlos II and Guavio operations (see Table 4.6 above) for 1985. 22. The poor performance of the projected energy requirements for the overall power system made at the time of the San Carlos II and Guavio apprai-sals is directly attributable, therefore, to the projection of GDP made in June 1979, which could not foresee the recession and the attendant stagnation of GDP growth between 1981 and 1983184. In contrast, the degree to which the GDP projection made in the August 1983 Economic Report, at the height of the recession, has closely tracked the evolution of actual GDP between 1983 and 1986 is shown in Annex 4B.7. This translated into the small deviation between actual energy requirements and its projection made at the time of the FEN appraisal for the 1983-86 period, which was discussed earlier (see para. 20 above). This relationship between GDP growth and increase in energy require-ments allows the establishment of the total revenue lost by the sector because forecasted sales did not materialize--on the one hand, due to recession and, on the other, due to increases in system losses. In fact, of the total deviation of 8.3 TVh in 1985 (see Annex 43.2) between sector sales projected at the Guavio appraisal and actual sales, about 902 (7.5 TWh) arose because of the unforeseen recession and the remaining 10Z (0.8 TWh) from increased losses, which were around 50? above the levels projected. Sales Outcomes Versus Forecasts in Regional Markets 23. An analysis of the performance of energy saleslgeneration relative to forecasts of these quantities within the regional markets, similar to that undertaken above for the overall power sector, was not attempted, in part, because from the late 1970s regional GDP data are no longer available. 24. Annexes 4B.8-4B.13 provide details of appraisal fcrecasts versus outcomes for energy requirements and sales for the regional companies ZEEB, EPM, ICEL (Group), CORELCA, and CVC (Valle), as well as ISA, respectively, during the review period. For ease of reference the results of these comparisons are summarized in Annex 43.14 by company for the case of energy sales. There are three points to be highlighted. First, deviations between forecasted and actual sales outcomes were within the accepted uncertainties of + 10Z for the period - 61 - 1971-78. These were projections associated with the Chivor I (ISA) and Guatape II (EPM) projects. This was a period of good economic growth In Colombia, despite the external shocks from the first oil price rise in 1973/74. Indeed, the reasonably accurate forecasting results achieved in this period, despite the simple methodologies employed, are likely to have laid the basis for the sector to continue the practice of extrapolat-ing past trends. Second, the starting point of sales forecasts for ICEL, CORELCA, BEES, and the consolidated power sector in 1983 associated with the PEN operation (appraised in 1983) are divorced from reality. In the cases of ICEL and CORELCA the situation was worse than that of the consolidated power sector noted earlier (para. 20 above). Here, the sales in 1983 used as the starting point of the projections were 28? and 301, respectively, above actual 1982 sales and, therefore, with this systematic inflation of projected sales, their performance relative to outcomes in these regional markets was very poor by 1986 (only 3 years after appraisal), when projected sales were almost 302 above actuals. A similar problem arose with the sales data for the starting point of the projection for ICEL in the Guavio project in 1980. Here, the sales projection started in 1980 at a level 292 above the actual 1979 level. These systematically inflated starting values for the sales forecasts tended to render projections of demand and financial income of little value. This suggests that the Bank needs to ensure that appraisals set forth 2-3 years of actual historical data shown for important parameters like sales, tariff levels, incomes, and costs, etc., to ensure that these data at least reflect reality. 25. The third point concerns the significant (except for EPM) differ-ences between actual and forecasted sales dependent on the company. For example, whereas by 1986 EPM's actual sales were only about 16X below those forecasted at the time of the Guavio appraisal in 1980, in the cases of EB, ICEL, and CORELCA, the discrepancies were more than twice as large, ranging from 342-39% under-forecasted levels. The differences between BEEB and EPM are also very evident for projects such as BEEB's Bogota Distribution I (1979 appraisal) and EPM' Gaudalupe IV (1980 appraisal). Once more, by 1985, EE sales were some 38? below those forecasted, whereas for EPM, sales were 18 below expected values. 26. The demand forecasting methodologies used in the Colombian power sector since 1971 are discussed in Annex 4B.1. What is clear is that up to the early 1980s, these methodologies were simple, being based on purely 'statistical models, that relied on fits to historical data from which extrapolations were made to arrive at projections, taking into account forecasts of GDP. They, however, withstood the test of time well in the 1960s and 1970s, during periods characterized by high economic growth and rapid electricity demand increases associated with the early phases of electrification. It was not surprising, therefore, that the power sector relied on such methodologies as the basis to project demand at the time that the 1978-84185 sector Investment program was prepared in 1977178. With the Bank's 1977 and 1979 Economic Reports projecting continued strong economic growth into the 1980s it is again not surprising that the forecasted rates of power demand growth for the 1980s were similar to those of the 1970s. It was not until 1980/81 that econometric demand models, capturing income, price, and demographic effects, began to be introduced into Colombian system planning. One model was developed through the ENE project and the other - 62 - for ISA. It was, however, not until 1983184 that the sector *accepted, the results of econometric forecasting methods, which were projecting lower power demand growth rates (around 82 annually) for the 1980s compared to the 10.82 annually the sector had accepted earlier. 27. By about 1983, as the recession's effects began to become better recognized by the sector and the Bank, power demand projections were lowered. However, it took another 3 years before the sector's investment program was slashed. There appear to be four factors that delayed the lowering of demand forecasts for the mid- to late-1980s to levels compatible with those derived from the econometric models. First, the *statistical' models had served well and had good track records. Second, the econometric models were less trans- parent and unproven in the Colombian context. Third, the power rationing in 1981 unfortunately coincided with the beginning of the recession and, hence, masked the effects of the recession for sometime. Additionally, that ration- ing created an atmosphere that shortages had to be avoided virtually at any cost, though when the rationing did occur in 1981 its cost estimated from an ex-post perspective was quite low (see Section 4.6). Fourth, reducing the demand forecast meant that the investment program in generation would have to be slashed, resulting in some of the projects favored by individual companies being postponed or cancelled--a prospect considered to be too *costly,* especially in regional political terms, to entertain. Conclusions 28. Throughout the period 1971-86, the dominant factor influencing aggregate electricity demand growth was the performance of GDP. With the recession between 1981-83/84 and the associated collapse of GDP growth, power sales in the overall sector grew between 1981-86 at about 62 annually, compared to the 10.0-10.52 annual rate that had been projected in the late 1970s for the 1980s. Indeed, at the sector level the 302 deviation by 1986 between projected and actual sales was due about 902 to the collapse of GDP and only 102 to the increase in energy losses. This illustrates the more critical impact of the recession compared to deteriorating energy losses on expected sector revenues. 29. In the Colombian situation economic demand models have demonstrated their utility and should be further used as a complement to the traditional methodologies, while recognizing that incorrect GDP projections will lead to large variations between power demand forecasts and outcomes, irrespective of the demand models employed. OED suggests that the Bank should encourage the sector (see Chapter VII, Section 7.3): - to subject project investment sequences to greater variations in the range of projected power demand than in the past; and - to incorporate investment strategies, implying greater flexibility, e.g. by including some smaller projects with shorter gestation periods, so that the investment program can be adjusted more readily to unforeseen demand changes. - 63 - 30. Finally, the Bank needs to enforce more carefully than in the past ten years that its presentations in SAR9 include at least three years of actual historical data for key parameters, such as sales, income, tariffs, etc., to ensure that the critical starting point of projections are properly embedded in reality. 4.5 Growth of System Losses Background 31. This section addresses the issue of escalating power system losses, contrasting the different performances of the Colombian companies, the prob- lems posed in handling non-technical losses, and the manner in which the Bank has attempted to deal with the issue. 32. Power system losses began to increase dramatically in most regional markets in the late 1970s. Indeed, Table 4.7 shows that in the markets of EEEB, MCALI, and ICEL, losses remained roughly stable from the early to the late 1970s, with EKCALI's levels being the lowest, due, in part, to the distribution character of the company. In this same period, EPM's losses fell to about 16% in 1978. However, in the CORELCA market, losses increased by 50% between 1971172 and 1978179. By 1979, the level of total losses in virtually all markets (except ECALI, which was at 12Z) was between 18-222, with EPM at the low end and CORELCA at the upper end. At this point, averaged over the whole sector, the losses were some 20Z. 33. After 1978/79, losses began to increase rapidly in the EEEB, EMCALI, and ICEL markets, with the rate of increase in CORELCA's case slowing down somewhat. By 1986, the loss levels in the EEEM, ICEL, and CORELCA markets were about 25Z, while those in the EPM and ENCALI markets stayed around 20% and 18?, respectively. For the system as a whole the losses averaged about 24%. Table 4.7: Transmission and Distribution Losses in All Markets (Technical and Non-Technical Losses) 1971-87 /a 1b (Actuals, Percent) System Year Total EEB EPM c CORELCA ICEL (Group) EMCALI CVC (propio) 1971 16.9 14.1 20.5 12.4 16.8 10.6 25.0 1974 17.4 14.4 18.1 17.2 18.9 9.8 31.7 1978 18.8 15.7 16.2 20.6 20.2 12.9 17.5 1979 20.0 19.3 18.2 21.9 19.7 11.9 19.2 1981 20.1 18.2 19.0 23.6 20.0 12.1 18.1 1984 23.9 25.5 20.2 23.9. 22.8 16.8 15.7 1.986 24.4 25.0 19.9 24.2 25.6 17.6 18.4 1987 24.5 26.4 n.a. n.a. n.a. n.a. n.a. ja EEEB actuals refer to sales within its own municipal markets. lb All losses are expressed as a percentage oZ net energy available. Ic EPM's losses refer to those in its own market, excluding its block sales. 64 - Structure of Losses 34. How did this come about? Table 4.8 below attempts to shed some light on this problem. Table 4.8s Structure of Losses (Percent of Net Energy Available) Year Technical (T&D) Losses 'Non-Technicals Losses Total Total EMBB EPM EHCALI Systee EEB 3PM EKCALI Syst 1976 11.6 9.4 9.8 n.a. 4.3 8.3 0.4 n.a. 1978 n.a. n.a. n.a. 12.6 n.a. n.ea. n.a. 6.4 1979 10.7 11.9 9.8 n.a. 7.6 6.3 1.7 n.a. 1983 n.a. 11.6 n.a. n.a. n.a. 8.0 n.a. n.a. 1985 n.a. 10.5 n.a. n.a. n.a. 9.0 n.a. n.a. 1986 11.3 9.6 n.a. n.a. 13.2 10.4 n.a. n.a. Sources: (a) EPM, for KPM data. (b) ISA study, July 1981 - 'Estudio de Perdidas de Energia en el Sector Electico Colombiano.' (c) 3B study, September 1987 - *Programa de Reduccion de Perdidas Periodo 1987-92.9 Here, a breakdown is shown between the so-called technical losses which repre- sent energy that cannot be consumed because of the physical characteristics of the transmission and distribution (T&D) system. Such losses are associated with transmission, transformer, and distribution losses. The *non-technical losses* are defined by the difference between the measurable 'technical' and 'total" losses--with total losses being the difference between bilted sales and net generation. These 'non-technical' losses represent the sum of consumption and billing losses, the former accounting for the fact that not all consumption is accurately recorded and the latter that not all recorded consumption is correctly billed. The 'non-technical' losses include a myriad of possible causes, such ass (i) incorrect meter calibration; (ii) meter tampering, resulting in low readings; (iii) by-passing of meters with illegal circuits; (iv) illegal connections with no meters; (v) legal connections but without meters; and (vi) genuine errors in meter reading and billing processing, etc. What Table 4.8 shows is that for the few markets (EEEB and EPM) for which data are available over a reasonable period, 'technical' losses have not increased. "Non- technical' losses, however, have trebled in the case of EEEM between 1976 and 1986 and grown by 252 for EM in the same period. 35. A recent study of 3EEB's losses(3) attributes the *non-technical' losses of 13.22 to the following causes: not metered service (1.62)s illegal connections (2.12); fraud (92); and others (0.52). The basis for this allocation will always be uncertain, but especially so in a system such as EEEB's in which in 1986 as many as 133,000 subscribers (95Z of which are residential) were - 65 * legally connected but had no meters. This means that about 182 of the residential subscribers In this market--roughly 1 In every 5 residences in the EEEB area--had no meterl It is difficult to reconcile this with the above low estimate of Onon-technical* losses attributed to this factor. What is especially startling is the rapid growth In the number of such privileged subscribers. Indeed, as recently as 1984 there were only some 78,000 SUB residential subscribers in this category. Thus, growth was in excess of 50% in two years. Though less dramatic than the 1986 figure, the 1984 number is already quite high and raises questions about the degree to which this issue was considered in the Bank's Bogota Distribution I Project, which was still under implementation in 1984. 36. How did the Bank view the increasing loss situation in 88B and in the sector as a whole? The record here is not Impressive. First, the issue of the large number of legally connected consumers without meters was identified as a key problem in the 1981 ISA study on losses, but appeared not to have attracted high focus in the Bank's SARo of its two power distribution loans [(Bogota Distribution I (1980) and II (1985)], which addressed loss reduction as well as network expansion. In the case of the Bogota Distribution II operation, some attempt w*s made to deal with this problem through financing for the installation of additional meters over and above those needed to satisfy the requirements of new customers. The success of these measures should become apparent in the medium term. Second, with each Bank operation after Bogota Distribution I in 1980, the Bank made more and more optimistic forecasts about the rate of EEB's reduction in losses. In other words, as SEEB's losses worsened, the Bank's view of improvement became more optimistic. Figure 4.2 and Annex 4C.1 show that at the time of the Distribution I Project (1980) losses were forecasted to be reduced by 115 within 7 years to 132, in 1985, and on the occasion of the Sector Adjustment loan, EEB's losses were projected to fall by 112 within 5 years to 13Z Of course, in the absence of these loss reduction forecasts, the tariff levels called for in the Bank's operations with EEEB would have to have been even higher than set forth in the SARs. For example, in the case of the Distribution II project (1985), the ex-ante internal rate of return was estimated at 12.0, assuming losses were reduced to 13.52 by 1990 and BEEB's retail tariff was sustained at USC7.9/kWh (mid-1984 prices). If the loss level had been assumed to reach 172 by 1990 and to remain there, tariffs would have to have been projected at USC9.0/kWh (mid-1984 prices), i. e. 141 higher, to ensure the same rate of return. The SAR does not set forth, however, the sensitivity of the project's return to a different loss level. 37. At the other end of the spectrum, though BPM's losses have always been high, even in the 1970s, this company has been able to control the situa-tion and prevent them from escalating. Moreover, unlike the situation with EE, RPM and the Bank agreed on realistic loss reduction forecasts (see Annex 4C.2). i! 一6.- ,，呈””。．:：劉讓•必．巒轡卜“越靈到，園聖墮望聖塑1里壇 CO膩爛馴為馴才州觀群酒闖R 亂亂亂鼠勵•齋亂。叫“．奮屆門闢嗡•州k州”．細勿唱馴勻 。「一一一一一一甲一一一不叮兀才一 總崤工、11 藝憎4才、I〕取、跑I 磚4．上I、～〕馭｀一I I功中•－．尸－－甲開－鬥鬥•－f－•，開，•，甲•－f-．•甲－－甲••．甲••．叩•••叩－-－叩鬥～開r-－門－•．f- 胞n颼禮觔為磚叮抽神口鄙縫勾細騰騙奉口曲幼戲個 廈黝柑 一細細．個州州論U論“‘魚•開．綱陣鑰飼“個開州口細目目細－一。一戲，•細網州勵寧關必臨。楓口一魚．馴口．•細勿．．網斷 •■■•．••柄開馴細網•自膠開鰍頃·認個馴禺．……魚闕馴讓馴，魚•馴勵州留細細馴•口目勾闔口哺啊口才 - 67 38. Another major factor that has aggravated am-technical losses, especially in the ZEZB market, has been the distorted tariff structure whereby commercial and industrial users are charged between 501 to 10OX above their long-run supply costs. This has provided significant incentives for large- scale theft through sophisticated meter tampering and alledged fraud. Up to 1987, the Bank's efforts to get these tariff distortions corrected in the ESEB market have been unsuccessful. Therefore, stronger measures seem to be called for in future (see Chapter V). 39. Given the enormous problems facing ZER at the time of the Bogota Distribution 11 appraisal (1984), such as the difficulties encountered in the Guavio project. the Mesita* delay and cost overrun, the company*9 dire finan- cial situation and its known managerial and supervisory problems, it challenges credibility to suggest that concurrently the company could achieve the kind of lose reduction targets set out in either the Bogota Distribution 11 or the Sector Adjustment loans, unless it radically changed its way of doing business, which neither of these two operations foresaw. The lesson to be learned is that what needs to be focussed on is what is the "tolerable* level of losses in the MMB market and in the sector as a whole given that a certain level of non-technical losses may remain inevitable at the current development stage and commensurate management levels in different companies. This tolerable* loss level is considered as being the level, for a given management capability and development environment, at which the company can prevent losses from deteriorating further but is unable to achieve a significant Improvement. In addition, it seems that it should be achievable to have all legal subscribers metered. Conclusions* 40. Somehow, the Bank's approach to ZEEB's loss problem appears to have lost realism from the time of tbe Bogota Distribution 11 operation, approved in 1985. Not only were unrealistic targets set to reduce losses, but the operation also failed to define the management and supervisory adjustments required to go some distance toward these goals--let alone meet them. Moreover, a substantial, but immeasurable. part of the increase in non-technical losses is traceable to the massive number of non-metered but legal connections. This number rose from some 79,000 in 1984 to about 133,000 in 1986, resulting in one in five Bogota residences being legally connected, without a meter. Financing was provided in the Distribution Il loan for installing meters additional to those for new customers. however, the success of these measures should be apparent only in a few years. 41. Based on the experience up to the end of the review period, it is evident that the Bank needs to re-think its approach to loss reduction, especially in the Bogota market. The problem of lose reduction, and especially its theft component, is difficult. Tn the judgment of OED, in designing future lose reduction programs, the Bank needs to assess more critically the degree to which overall losses can be reduced significantly, as long ass (i) utility profitabil- ity is not an objective at the forefront of the sector's strategy; (ii) goverranent subsidies to the companies remain pervasive; and (iii) utility managers face neither penalties nor incentives in connection with failure or success to achieve lose reduction objectives. - 68 - 4.6 Inability To Meet 1981 Power Demand--Cost to the Economy Background 42. The Colombian power sector experienced several periods of rationing over the past 16 years, in particular, in 1977, 1980, and 1981. These supply shortfalls came about because of a number of reasons, including delays in the start-up of new generating, transmission and distribution capacity, combined with a series of droughts. Of course, in most developing countries power rationing is invariably perceived as having a high political cost. Its economic cost is difficult to measure and highly dependent on the nature of the outages, and equally so on the type of power market being served. This section estimates for the 1981 power rationing what the ex-ante and ex-post costs to the economy were. These OED estimates are, in turn, assessed rela-tive to estimates made in 1981 and endorsed by the Bank. 43. There was no specific analysis of rationing costs in Colombia before the 1986 ISA study. (4) This, along with the above-mentioned caveats about estimating such costs, made it surprising that the President's Report on the Guavio project in May 1981, stated that the electricity shortfall expected for 1981 was likely to be around 1200 GUh in energy terms (or about 6Z of require- ments) and that -the cost of this shortage (in terms of productivity, income foregone, and social welfare) has been estimated at US$3 billion for 1981 alone* (Guavio President's Report, p. 11, para. 28). This Implied a cost of some US$2.50 per kWh not supplied. This estimate appears inordinately high for two reasons. First, rationing policy stipulated that load shedding would be focussed on residential and commercial users, not industry. Second, in 1981, about 452 of power sales were residential then and it was well known, even then, that outage costs to residences were low and unlikely to exceed five times the low voltage long-run marginal cost (then estimated at about USO.6/kfh). 44. High estimates of rationing costs could provide the rationale for increasing supply-side investments to Improve system reliability beyond what is justified. It is for this reason that an attempt is made in this section to estimate the likely ex-ante and ex-post costs of the 1981 rationing. It should be noted that rationing cost was not a variable taken into account by ISA as part of its objective function during the planning stage. Therefore, the very high estimate of the 1981 rationing cost in the President's Report did not involve any judgments taken at the planning stage by the Colombian authorities. 45. Power rationing can take many forms and has very different effects on different consumers, depending on the circumstances. Clearly, the cost of rationing is as varied as its effects. These costs depend on whether the outage is sudden and unannounced or planned and announced. The duration of the outage and its time of occurrence also show important cost variations for users and, most critical, the outage costs differ dramatically between differ-ent users. Generally, the lowest cost and 'best form, of power rationing is that which is announced well in advance so that consumers can take appropriate measures to minimize its impact. Fortunately, in the case of Colombia, the most severe cases of power rationing in the review period all arose from events that were well - 69 - known in advance of the need to reduce power supplies. This is because its generation system consists of about 80% hydro, with considerable storage capacity, and, therefore, the problem of system reliability is one of meeting peak power demand and not energy requirements. This is in sharp contrast to most other countries where power generation is dominated by thermal plants. In such situations, the major generation outage problem faced is that of sudden and unannounced breakdowns in generation plants. In these systems such outage costs are higher than in th.se where the rationing can be planned. 46. The 1986 ISA study on power rationing confirmed that for planned generation outages the costs to industrial users are typically about ten times higher than to residential ones, while commercial customers experience more of a nuisance effect than loss of added value and, accordingly, their planned outage costs are, at the most, double those of households. These costs are summarized in Table. 4.9 belows Table 4.9t Estimated Planned Outage Costs by Compan and Type of User (December 1985 Col$IkVh) Type of User! CORELCA ICEL IEEB EPM CVC TOTAL (Valle) Industrial 153 196 252 154 186 188 Commercial 26 39 48 52 26 38 Residential 15 17 17 20 20 17 Sources 'Costs of Rationing Electric Power - Industrial, Residential, Commercial and Service Sectors," ISA, October 1986. The ISA study also noted that planned outage costs are highest for smaller and light industries than for heavy industry, primarily because the former lack self-generation or even stand-by emergency generation equipment. Cost of the 1981 Power Rationing 47. OED estimates of the ex-post and ex-ante costs to the economy of the 1981 power rationing in the central system are developed in Annex 4D.1. The summary of these ex-post estimates of rationed energy levels and allied costs for the three affected markets of IEEB, ICEL. and CVC are presented in Table 4.10 below. This shows that, for example, in the EE market the ex-post range of ratioved energy in 1981 was between 120-340 GWh. Based on the upper level of this range and the unit outage cost estimates by user type and market presented in Table 4.9. OED estimates that the ex-post cost of the 1981 rationing in the EEB market was some 1985 US$70 million. When similarly estimated costs for the ICEL and CVC markets are added, the resulting total ex-post cost was about 1985 US$130 million, which represents only about 4Z of the ex-ante estimate - 70 - of US$3 billion presented in the President's Report of the Guavio project. Since the basis on which this estimate was arrived at was not stated, it is not possible to comment on it further. Table 4.10: Ex-Post 1981 Rationing Levels and Costs User TypefMarket BEES ICEL CVC -----------(gWh)----------- Residential 80-280 30-40 60 Commercial 30- 40 10 15 Industrial 10- 20 10-20 15 Total 120-340 50-70 90 Total Cost (US$ 1985) Based on Upper Levelst $70 million $30 million $26 million 48. To provide some other yardstick of comparison, ORD estimated from an ex-ante perspective the cost of the 1981 outages based on the same assumed 1200 gVh shortage stated in the President's Report. To determine this cost, OED assumed that the unit outage costs were as presented in Table 4.9 above from the ISA study, and that the shortages were pro-rated across all users in proportion to their share of sales. This is an extreme assumption, which would yield a very high upper bound, since the explicit policy by which the shortages were rationed was totally at variance with this assumption--this policy being that industrial users would be least affected and residences allotted the greatest load shedding. However, even with OED's very conserva-tive assumption, its ex-ante estimate of the 1981 outage cost of 1985 US$500 million is still only 171 of the estimate in the above President's Report. The reasons for the still significant differences between OED's ex-ante and ex-post estimates are twofold. First, the on-set of the recession in 1981 reduced the actual level of demand relative to what had been expected in that year. This, in turn, made the actual rationing level lower. Second, when rationing became inevitable, load shedding was directed most to residences and least to industrial users, thus minimizing the cost to the economy. Conclusions 49. The Bank's endorsement of inflated estimates of the 1981 outage costs was unfortunate, especially given that ORD has difficulty justifying ex-ante estimates even 115 of that presented in the President's Report. It repre-sented another incidence of statements potentially affecting the Bank's credibility (Chapter II, Section 2.3). 4.7 Surplus Generating Capacity--Cost to the Economy Background 50. The final issue addressed in this chapter is that of the surplus generating capacity the country will experience up to about 1993. What has been - 71 - the cost of these surpluses to the economy? This is the question this section seeks to answer. The excess generating capacity was brought about by large additions in the 1980s that were decided in 1978179 on the basis of demand forecasts that reflected trends during the 1970s (about 10Iyear demand growth) but did not materialize in the 1980s (when demand growth dropped to between 52- 62 annually). The underlying reason for the inability of the system to cope with variations in demand lies in the rigidities associated with the long and uncertain construction times for the hydro plants that have dominated tradition- ally the expansion plan. Unless the expansion plan includes a number of shorter lead time thermal projects (such as combined-cycle plants), overshooting or undershooting demand will be inevitable in the future since forecasting of demand implies considerable uncertainties and risks which, unless properly accounted for, will result in the power system stumbling periodically from shortages to surpluses and back. Cost of Surplus Capacity 51. By the end of 1985, the power system was entering a period of excess generating capacity, which is expected to continue to 1993/94. OED estimated the level and costs of this excess capacity by analyzing the energy and peak demand balances of the system every year for the period between 1985-1993. 52. Given that the excess capacity is now unavoidable, the consequences associated with it cannot be avoided either, and, therefore, should be classified as Oregret" rather than as a cost. To estimate this *regret' two alternative approaches are analysed. The methodology of the first approach (a completely ex-post one) consists of estimating the excess energy and capacity based on an energy balance reflecting the current programmed dates of commercial operation for the new power plants and actual demand between 1985-88, with projections to 1993. The cost of the excess supply for the 1985-93 period is estimated by assigning a cost to energy and capacity, and results at 1985 US$450 million (see Annex 4D.2). 53. The alternative approach takes a mixed ex-antelex-post perspective (Annex 4D.2). It finds the least-cost expansion plan that would have resulted (ex ante) had the actual evolution of demand been foreseen (ex-post). Hindsight provides, in particular, actual demand and the delays in commission-ing of the plants. From these elements an adjusted investment plan is constructed that brings plants on stream when demand calls for them. The net present value (NPV) of this adjusted plan is then compared with that of the original investment plan, with its high demand growth expectations and early commissioning dates for projects. The difference in these NPVs then reflects the *excess costs,' from which fuel savings arising from having excess hydro capacity are deducted. Based on this alternative approach, the value of *regret" for the excess capacity is estimated at 1985 US$410 million. Conclusions 54. What these very high values of 'regret' highlight is the vital importance of having more flexible investment programs with some shorter gestation projects so that better responses can be had to the vicissitudes of demand uncertainties. - 72 - What is also of merit to note is that the estimates of ex-post rationing costs in 1981 were about 3 times smaller than the *regret* of excess capacity. Indeed, one of the consequences of the highly visible effects of rationing, in contrast with the invisibility and seeming 'costlessness" of excess capacity, is that it can shift the investment planning focus too strongly toward outage avoidance in contrast to surplus avoidance. 4.8 Performance of the Bank 55. Given the intense supply-side focus of both the Colombian authori-ties and the Bank, at least up to the mid-1980s, it is not surprising that even the broader elements of the shifts in demand structures between regions did not emerge as an issue with important financial consequences for the sector. At the time, correctly, there was great concern for trying to meet a demand that seemed always to grow 101 annually. In OED's judgment, the Bank cannot be faulted for this natural bias in its focus at that time. In addition, there is little to be critical of in the demand forecasting methods used In that period. The demand collapse could not have been foreseen by the Borrower or the Bank. What, however, seemed to be missing on the part of the sector authorities was the timely recognition of what really was happening to demand post-1981. although it was not easy to discern. The Colombian and world economies had entered a very deep recession; simultaneously, power tariffs had been rising, in real terms, quite rapidly in the 1980-82 period, giving rise to legitimate questions as to what degree some of the demand slowdown was due to price or income effects or both and, if so, in what proportion? It appears that the Bank expressed some concern in 1982 that demand was likely to be lower than forecast, however, this was not endorsed by the Colombian authorities until sometime later. 56. Where the Bank can be faulted legitimately was the carelessness with which some of its sales forecasts in some of its SARs were put together. This reflected a collapse of quality control of the most basic character in the appraisal review process. Such quality control did not require re-estimation of rates of return or issues of such a nature, but solely asking whether a 30Z increase in power sales in one year looked reasonable, especially since the end point of that increase became the starting point of a 6-7 year sales projection. This, of course, totally divorced such forecasts of sales and income from reality and was potentially damaging because it painted a buoyant, but totally inadequate, picture of growth. 57. A further area of concern about Bank performance involves the treatment of power losses in ZEEEB, a company which received two distribution loans, both targeted, inter alia, at problems of losses. The Bank needs to be more willing to recognize and say that not all problems have solutions in given institutional arrangements. It was willing to accept that BPM's losses were high and controllable, yet unlikely to be significantly reduced in the medium term. However, somehow it failed to take a similar approach in the case of EEEB, where the issue was first control of losses and then an attempt to reduce them. This cannot be achieved without appropriate focus on the management and supervisory eificiency of the company--an issue which the Bank naturally has difficulty addressing, as well as confronting head-on the severe distortions in tariff structure in the EEEB market. - 73 - 58. The Bank's endorsement in 1981 of the inflated estimate of the cost to the economy of the rationing in that year was unwarranted, as evidenced by OED's very conservative ex-ante estimate. The Bank-endorsed estimate tended erroneously to re-inforce the popular notion that the costs of outages are very high and that rationing was something to be avoided almost at any cost. 4.9 Agenda for the Future 59. OlD considers that in the framework of future lending operations by the Bank to the Colombian power sectort - there is a need for a new approach to the problem of power losses, especially the non-technical component. This calls for (I) decisive action on non-metered but legal connections, (ii) effective measures to improve the management and supervisory efficiency of affected utilities, (III) reduction of the massive distortions in tariff structures that provide incentives for power theft, and (iv) a recognition that there can be little success unless the objective of utility profitability becomes paramount, since the achievement of efficient loss levels is questionable for a sector or utility that has easy recourse to government subsidies and bail outs, and in which company managers have no incentives and face no penalties for success or failure to reduce losses; and - Investment strategies should incorporate greater flexibility so that the investment program can be adjusted more readily to unforeseen demand changes. ENDNOTES (1) CVC (propio) - market in which CVC distributes directly, as against MCALI, e.g. where CVC only provides bulk power. (2) Residential electricity use per subscriber averaged for the whole country was only 235 khmonth in 1986. (3) 'Programa de Reduccion de Perdidas Periodo 1987-92,0 BEB study, Septem- ber 1987. (4) 'Costs of Rationing Electric Power - Industrial, Residential, Commer- cial, and Service Sectors,* ISA, October 1986. - 74 - V. POWER TARIFFS 5.1 Introduction 1. During the period under review, the power sector, as a whole, and its individual utilities, have gone through periodic financial crises of which the current one is the most severe. The solutions that traditionally have been proposed to overcome these crises were debt refinancing, restructuring of investments, and tariff readjustments. These tools were used to varying degrees. The sector resorted most often to external debt refinancing. I-adeed, a refinancing operation is politically less painful than tariff readjustments and associated improvements of the sector's internal generation of funds. In addition, it does not lead to visible effects such as stoppage or severe curtailment of investment programs which could follow in the wake of a restructuring of the investment program. Further, the mechanism of debt refinancing 'gives the impression of having solved a problem when, in reality, the only thing that has been done is delay the solution of a crisis.'(1) In this context, the task of the present chapter is to assess the pricing policies pursued in the power sector in Colombia in the review period; the degree to which tariff adjustments have and have not been used as a tool in addressing the sector's and borrowers' financial problems; the adequacy of the Bank's role in urging reform to pricing policies; and the relation between these policies, the sector's financial crisis, and the economic justification of the system expansion programs. 2. Pricing policy has three key objectives--to promote financial strength, economic efficiency, and more desirable distribution of income. Financial strength is achieved if average prices exceed financial costs, including an "appropriate0 contribution toward covering the local costs of on-going and future investments. Of course, the determination of what level of self-financing is 'appropriate* is critical. As far as economic efficiency is concerned, this is achieved if, in addition to fulfilling the *financial test,* the level and structure of prices reflect the social opportunity costs of electricity supply. Finally, the income distribution objective is satisfied if the incidence of price assists low income groups, or at least does not place too great a burden on them. 3. In this chapter, the unfolding of events in the sector in the review period provide the basis against which the achievements of these pricing policy objectives are "tested." Special emphasis is given, on the one hand, to what was known in an ex-ante sense and, therefore, the consequences of which could have been anticipated, and, on the other hand, what can be learned only from hindsight. In terms of presentation, the chapter first reviews how electricity prices have evolved during the period in order to provide a setting relevant to the issues analyzed. Then it analyzes the first issue, the performance of tariff forecasts relative to outcomes. This is followed by an in-depth analysis of the costs of service and tariffs at the retail and bulk levels during the review period, highlighting the degree to which tariff structures have become increasingly distorted while tariffs remained far below either long-run marginal costs (LRMC) or long-run incremental costs (IC). The third issue analyzed is - 75 - the adequacy of tariff levels and internal rates of return envisaged and achieved in the context of the major investments supported by the Bank. The fourth issue involves the cost differences between regional power markets and the tariff implications of the absence of financial transfer mechanisms between these markets. The analysis then turns to the income distribution impacts of pricing policy and to an assessment of tariff reforms made in the review period to improve the desirable distribution effects while not doing too much violence to economic efficiency and the financial strength of the sector. A further section reviews the cost and price structures of urban household cooking fuels and highlights the stimulus given to electricity use by the large subsidy it benefits from in contrast to other household fuels. The chapter then discusses the role the Bank has played in the development of tariff policy and the manner in which it has handled the difficult problems of tariff levels and structures. The analysis concludes with an outline of directions for possible future action based on the lessons learned. 5.2 Evolution of Electricity Prices in the Review Period Background 4. Fundamental to any assessment of how power tariff levels have developed in the review period is a determination of what deflator should be used to measure such movements. If the objective of pricing policy is to ensure that electricity costs to consumers do not absorb more than a certain fixed share of consumers' income, then the appropriate deflator for tariffs would be one solely reflecting local price movements, i.e. either the consumer price index (CPI) or the GDP deflator index. However, if tariffs are to reflect the costs incurred by the producers in the power sector, then they have to take into account the relation between the local and foreign costs of inputs to the sector. Over the past 16 years, the share of the sector's investment costs has been split roughly 60/40 between foreign and local currency, respectively. With this high foreign component of cost, the sector's capacity to service its foreign debt becomes highly dependent on the tariffs' reflecting exchange rate adjustments. This implies that tariffs need to be deflated using a combination of indicators reflecting movements of (i) local and foreign prices and (ii) the exchange rate. These three indices are not independent and, under competitive conditions, the exchange rate should reflect changes in both internal and external prices. As set out in Annex 5A.l, the GDP deflator is used as the local price index and the Unit Value Index of Manufactured (MUV) Exports to LDC markets from the USA, Japan, the UK, Germany, and France, expressed in US dollars, as the foreign price index. Based on these two indices, and the exchange rate index, a composite deflator index was defined as the sum of the GDP deflator index weighted 40? and the foreign currency index weighted 60?. This composite index was the deflator used to assess tariff movements in the review period. It can capture the fact that the power sector, through its capital intensiveness and its high reliance on imports, is more open to external impacts than Colombia's economy overall. * 76 * Retail Tariffs 5. Table 5.1 below provides a broad summary of how retail tariffs evolved from 1971-1986 for the power sector as a whole, broken down by type of user. Year-by-year detailed movements of these tariffs are given in Annex 53.1, using both the composite and GDP deflator indices to illustrate the effects of their use. The tariffs in Table 5.1 are expressed in *real* 1986 peso terms using the composite deflator index. There are three points to note. First, the decline in real terms in average residential tariffs over the period. Second, the significant increases in the differences in tariffs between different users. For example, whereas in 1971 industrial and com-mercial tariffs were 9? and 502, respectively, above residential levels, by 1987 these differences had risen to 1001 and 1902, respectively. Finally, while residential sales, as a fraction of total sales for the whole system, rose from 412 to 482 between 1971 and 1986, the level of the residential tariff relative to the average national unit revenue has decreased significantly, from 922 in 1971 to some 62Z in 1987. This indicates that as the system has become more residential in its sales these users have contributed a decreasing fraction of total revenues. Table 5.1: Averaae Tariffs by User Type, 1971-1987 (1986 Col$/kVh) Year Residential /a Commercial La Industrial fa Total National /a 1971 5.6 8.4 6.1 6.1 1975 3.8 6.5 4.7 4.4 1980 5.1 9.9 7.4 6.4 1983 6.4 14.6 10.7 8.8 1987 4.6 13.7 9.7 7.4 La Deflated to 1986 pesos using the Composite Index (mid-year) in Annex 5A.1. 6. Table 5.2 below shows the evolution of average tariffs in the review period according to company or power market (Annex 53.2 gives the details), and, most notably, the fact that the spread in average tariff levels between companies has increased in the review period. Indeed, in 1971, this spread was some 55? (between CVC and EPM). By 1987, it was 66? (between EEE and ICEL). - 77 - Table 5.2* AveSagE Tariffs by Company. 1971-1987 (1986 Col$JkWh) Total Year CORILCA ICBL EMt EPM ENCALI CVC National 1971 7.2 5.6 5.9 4.9 7.4 7.6 6.1 1975 6.1 4.3 4.0 3.6 4.8 5.1 4.4 1980 8.7 6.2 6.2 4.o 7.6 8.0 6.4 1983 9.4 7.9 10.9 6.4 9.7 9.2 8.8 1987 7.8 6.0 10.0 6.2 6.9 6.5 7.4 a Deflated to 1986 pesos using the Composite Deflator Index (mid- year) in Annex 5A.1. 7. What this seant in terms of rates of increase in average tariffs across regional markets is set out partly in Table 5.3 below, which presents average annual changes in average tariffs according to the type of user in the overall sector. Annex 5B.3 presents more detailed average changes in individual regional markets for specific users. Table 5.3 highlights three distinct periods. Table 5.3t Percent Annual Changes in National Average Retail Tariffs in Real Terms, 1971-87 (Percent) Residential Commercial Industrial Total 1971-75 -9.2 -6.5 -6.4 -8.1 1975-83 6.6 10.8 10.9 9.2 1983-87 -7.8 -2.2 -2.6 -4.2 1971-87 -1.3 2.9 2.9 1.2 8. The first period, 1971-75, witnessed a uniform major decline in tariffs for all classes of consumers, as well as in all regional markets (see Annex 5B.3). This decline was largely related to high international and domestic inflation during 1973-75. At the sector level, the average unit revenue declined in real terms by some 82 annually, residential tariffs showing the most significant erosion (about 9.2Z annually) compared to 6.52 annually for both industrial and commercial users. The second period, 1975-83, is characterized by a reversal in tariff policy, leading to average tariffs increasing for all consumers in all markets in real terms. At the sector level, the average tariff rose by 9.22 annually, with the most rapid increase occurring for industrial and commercial users (about 10.8Z annually), with residential consumers experiencing a rise of some 6.62 annually. The beginning of this second period witnessed the informal suspension of Bank disbursements from power loans (see Chapter 1) and the implementation of significant tariff adjustments to which the resumption of Bank lending was tied. The third period, 1983-87, was again characterized by decreasing tariffs in real terms, though at a slower rate (4.22 annually) than * 78 - in the first period, except for residential consumers whose tariffs declined at an average national rate of 7.8Z annually. This period saw large adjustments of the exchange rates between the peso, the US dollar, and other major currencies. In addition, it coincided partly with of economic recession in Colombia (1981-84) and much slower growth in power demand than historically experienced. Bulk Tariffs 9. Table 5.4 shows that between 1971 and 1986 ISA's bulk tariff has undergone a similar, albeit less pronounced cycle to that exhibited by retail tariffst first, an annual 2.42 decline in real terms between 1972-75, followed by a rapid rise of 10.72 annually between 1975 and 1983, which turned into a decline of 2.S% annually between 1983 and 1986. For the period as a whole, however, the bulk tariff underwent an average annual growth of 4.9Z, which meant that it virtually doubled in real terms between 1972 and 1986, though in economic terms its level in 1986 was still well below the average IC at the interconnec- tion level (see Table 5.9 below). Table 5.4: Evolution of the ISA Bulk Tariff, 1972-86 Year ISA Bulk Tariff ]a Average Annual Growth Rate (1986 PesosJkWh) /b (Percent) 1972 2.00 1972-75 0 - 2.4 1975 1.86 1975-83 - 10.7 1983 4.21 1983-86 = - 2.3 1986 3.93 1972-86 = 4.9 /a Excludes short-term energy sales (i.e. *optimizing* or secondary energy). b Deflated using the Composite Deflator Index (mid-year) in Anne% 5A.l. Conclusions 10. Looked at over the entire review period, 1971-87, retail tariffs have been unsatisfactory by any standard, especially considering the very low level at which they were in 1971. At the overall national level, the average unit revenue increased by a modest 1.22 annually, resulting from increases of 2.92 annually for industrial and comercial uses and a sustained decline of -1.3Z annually for residences, although in the 1970s residential tariffs were estimated to be less than one-third of the cost of service (see section 5.4). In regard to the performance of tariffs within specific regional markets (see Annex 5B.3) some of the fundamental weaknesses in the sector's institutional structure become apparent. For example, during the entire period, 1971-87, overall average retail tariffs in the CORELCA, ICEL, EMCALI and CVC (propio) systems experienced - 79 - virtually no change in real term while accounting for about 602 of overall sales in the power sector in the 1980s. It is only In the BEED and IPM markets, making-up the remaining 40? of power sector sales, that overall average tariff levels have increased in real terms in the review period--in particular, in the case of EEEB by some 3.82 annually, and for BPM by the modest amount of 1.7? annually. This imbalance has been one of the factors contributing to the recurrent financial problems of the power sector--namely, that the group of companies in the sector whose share of sales has grown from 50% to 60% between 1971 and 1986 have experienced concurrently no change in their average tariff levels in real terms even though these were well below costs at the beginning of the period. The somewhat better relative performance of EEEB regarding tariff levels compared to the rest of the sector needs to be viewed in its proper context. Indeed, EMEB's performance has been achieved at the cost of an increasingly distorted tariff structure as commercial and industrial users subsidized more and more heavily residential users (see Section 5.4). This is likely to have been a factor in NEEB's rising non-technical losses, as discussed earlier (see Section 4.5). 5.3 The Performance of Tariff Forecasts Relative to Outcomes Background 11. Tariff projections are difficult to make due to their dependence on the vicissitudes of government policy regarding a controlled price, sucb as that for electricity, and because of the uncertainty svrrounding the macro-economic parameters, such as inflation and devaluation rates. Despite these difficulties, tariff forecasts form an essential element of financial apprai-sal on which an assessment of the earning capacity of the corporation is based. Procedure 12. An assessment of tariff projections compared to actual outcomes is presented in Annexes 5C.1 and 5C.2, where the procedures followed are outlined. In essence, first, tariffs used in the projected income statements for each company in each SAR were expressed in constant pesos using the composite deflator index derived from projections of the key macroeconomic variables in the pertinent SARs--forecasts of these variables are shown in Annex 5C.3. Second, the actual tariffs for each company were then expressed in constant pesos, of the same base year, using the actual composite deflator given in Annex 5A.l. Finally, ratios of these actual to projected real tariffs for each company during relevant years were calculated. They are summarized in Annex 5C.2. Of course, when forecasting uncertainties are taken into account values of the above ratios falling between 90? and 1102 can be considered reasonable. Conclusions 13. The question focussed on here is "how accurate were the tariffs forecasted relative to actual outcomes?" The main findings were that, in general, except for RE, actual values tended to fall short of forecasted tariffs in real terms. This shortfall increased in 1985 and 1986 as the peso devaluation accelerated and was not reflected by a commensurate increase in - 80 - tariffs. However, overall, and taking into account all factors, especially the long-term character of the projections, tariff forecasts were quite accurate. This is in sharp contrast to the forecasts of electricity sales and demand discussed earlier (see Section 4.4). Indeed, out of a total of 122 point tariff values forecasted, 551 were within +101 of the actual outcome, with those lying outside this interval tending to be lower (252) rather than higher (20%). Additionally, tariff forecasts outside the +101 limits were concentrated in SEE and RPM as overestimates associated with early projects that had forecasted significant real declines in tariffs aid-way through project implementation, and in CORELCA as underestimates. 14. Of course, the above discussion is entirely separate from the issue of whether the projected tariffs in financial statements were at appropriate levels- -this is addressed in the following section. 5.4 Economic Costs, Retail Tariff Structures, and Levels Background 15. Any analysis of tariffs is sterile In the absence of an assessment of service costs against which such tariffs are measured. The relevant such costs are the long-run marginal or incremental costs of supplying power at the interconnection level, increased by the incremental costs of distribution within each regional market, leading to estimated supply costs at medium (MV) and low (LV) voltage levels, i.e. costs that are of interest from an economic standpoint. Starting in the aid 1970., mostly at IDB's and the Bank's instigation, the sector produced information on marginal and incremental costs of generation and high voltage transmission. In turn, little was known about incremental costs of network expansion from the interconnection level to the medium and low voltage levels until, in 1986, ISA produced the first comprehensive review. However, as discussed in Annexes 5D.1 and 5D.2, the cost estimates in that review raise as many queries as they attempt to answer. The purpose of this analysis of costs is to examine the historical relationship between tariffs and marginal or incremental costs across the Colombian power system during the review period in order to ascertain the degree to which tariffs have deviated from marginal or incremental costs. This would assist also in illuminating the role tariffs have played in compounding the sector's financial problems. 16. The marginal cost studies conducted for the Colombian power system are outlined in Annexes 5D.1 and SD.2. Since EMP's distribution network was the only system for which estimates of incremental costs were made in 1979, for the period 1979-1984, these data have been taken as a proxy for incremental network costs in all other markets during this period and, indeed, up to 1986. This introduces a systematic underestimation of the incremental supply costs of electricity down to the MV and LV levels, since, compared to all other power systems in Colombia, the RPM market is characterised by its high load density, i.e. very limited spatial coverage combined with high level of service coverage. Indeed, it covers an area 1/4 that of EZE8, and about 22 and 0.62 of that of CORELCA and ICEL, respectively. Therefore, EPM's incremental network costs are likely to be the * 81 - lowest of all markets. Hence, by taking these costs as representative of the rest of the power system, there is little danger that incremental network supply costs are overestimated. LFNC Estimates for 1978-86 17. Based on the 1979 ISAISPM study of marginal supply costs at the generation level and incremental supply costs froa the Interconnection to LV levels, a reconstruction of such costs, as they could be viewed at that time, is possible. Table 5.5 summarizes these results, broken down into capacity costs related to generation, Interconnection and WILY networks, as well as energy costs. Annex 5D.3 shows further details. Tables 5.5s Estimates of Marginal Supply Costs at Market Prices Based on the 1979 ISAhE14 Study (December 1978 Pesos) Capacity Costs Energy Costs (Col$/kW-Year) (Col$/kVh) Generation 1,540 0.74 Interconnection 678 Interconnection Losses 62 0.03 Sub-total at Interconnection 2.280 0.77 MV Network 2,990 /a MV Losses 0.04 Sub-total Costs at MV Level 5j270 0.81 MV and LV Network 4,700 /a Losses 0.09 Sub-total Costs at LV Level 6.980 0.90 /a See Annex SD.3 for basis of the NW and LV network cost estimates. 18. In energy terms (assuming an average capacity utilization of 602, i.e. 5,256 hourslyear), these costs become in market pricess December 1978 PesoskWh MW !eference cost 1.81 LV reference cost 2.23 - 82 - If these costs are re-adjusted in real terms according to the composite deflator index in Annex 5A.1 (December 1978 - 17.36, mid-1978 = 15.82, and mid-1986 - 100), the above reference costs, when expressed in current peso and US dollar terms between 1978 and 1986, become as set out in Table 5.6 below. 19. The reference costs above are given by LV and MV categories, but, within these, costs vary mainly because of the different load factors or use patterns of different consumers. Therefore, the real supply cost to residential users (especially those with cooking loads) Is expected to be higher than the LV reference cost, and the supply cost to commercial users is somewhere between the LV and MV reference cost levels. Since information is not available to determine the amount of sales to industrial and commercial users in the different markets according to the voltage level, this evaluation has taken this breakdown to be 1001 MV, in the case of industrial sales, for commercial sales, 60% LV and 402 MV, and 100% LV for residential sales across all regional markets. Table 5.6s Estimated Long Run Marginal Costs of the Interconnected System 1978-86 Costs in Current Terms (Market Prices) Ja (Col$IkVh) 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 Medium Voltage (MV) 1.65 2.01 2.52 3.08 3.68 4.47 5.51 7.37 10.43 14.10 Low Voltage (LV) 2.03 2.48 3.09 3.79 4.53 5.49 6.78 9.06 12.83 17.35 Ja Inflated by the composite deflator (mid-year) in Annex 5A.1. Costs in Current US CentslkWh Medium Voltage (MV) 4.22 4.73 5.33 5.65 5.74 5.67 5.47 5.18 5.37 5.35 Low Voltage (LV) 5.19 5.84 6.53 6.95 7.07 6.69 6.73 6.37 6.60 6.58 20. The 1979 marginal cost study by ISA and EPM was based on a detailed monthly generation dispatch for the years 1979-84 while the 1983 ISA study looked at incremental costs over the period 1983-93. Now did these two studies' estimates of incremental costs compare for 1983? At the generation and interconnection level the 1979 study gave an estimate of US*3.7/kWh in end-1983 prices when costs are adjusted between end-1978 and end-1983 using the composite deflator. In contrast, the 1983 study provides an estimate of some USC5.2/kVh - 83 - in end-1983 prices (see Annex 5D.4). This indicates that the 1983 ISA study envisaged a 40? increase in incremental costs at the generation and interconnec- tion level based on the 1983-93 expansion plan. This is strange for a hydro system reputed for its low costs, especially when that expansion plan was structured almost exclusively around hydro generation. This should have suggested either that too many investments were concentrated to service demand increments that were not sufficiently large to justify them or that very high cost hydro plants were part of the plan. Since the latter was not the case, at least on the basis of the ex-ante investment costs contained in the plan, it is difficult to avoid the first conclusion. Both the Rio Grande and Bogota Distribution II SARs referred to the incremental costs estimated in the 1983 ISA study in the context of the level and structure of tariffs in the EPM and EE systems. However, neither SAR commented on the high level of the estimated incremental generation costs in the ISA study, presumably because both projects were looking solely at the EPM and EEEB systems and not at the overall interconnected system. 21. Following the acceptance, by 1985, that energy demand was increasing more slowly than previously envisaged, the investment program for the power sector was cut back and in 1986 ISA undertook an update of incremental costs for the interconnected system for the period 1986-94. These estimates are shown in Annex 5D.4. They suggest two relevant points. First, they estimate incremental costs at the generation and interconnection levels to be US*4.04/kWh, expressed in end-1983 prices, i.e. significantly less than the 1983 ISA estimate referred to above, based on the untenable 1983-93 expansion plan. In addition, this new estimate was more in line with estimates derived from the 1979 ISA/EPM study. Second, the cost difference between primary and secondary distribution is only around 132 compared to 232 for the estimates based on the 1979 ISAIEPH study (see Tables 5.5 and 5.6). This is due presumably to the network cost data differences discussed earlier (see para. 16 above). In order to have some representative costs for the interconnected system covering the 1978-86 period, those estimated in Table 5.6 above based on the 1979 ISAIEPM study, have been used to examine the structure and level of tariffs. Retail Tariff Structures, Levels, and Distortions Relative to LRMC 22. Table 5.7 presents a comparison between estimated average marginal costs, as viewed from 1978, for serving residential, commercial, and industrial users across the interconnected system and the average tariffs experienced by these users between 1978 and 1987. In the case of residential users, tariffs in 1978 were only slightly more than 113 of marginal supply costs. They subsequently improved slightly to about 112 that cost by 1983, and deteriorated rapidly after 1983, ending up by 1987 at a level close to where they were in 1978. The fact that in both 1978 and 1987 the average residential tariff across the entire power system was only just above 1/3 of marginal supply costs, does not mean that little had changed. Indeed, this masks the considerable increase of the social costs of these subsidies, since, by 1986, residential sales for the interconnected system were some 80? higher than in 1978. This implies that, in absolute terms, the economic cost of this subsidy had escalated by 80 in only 9 years, purely due to the growth in residential sales. - 84 - Table 5.71 Deviation Between Average Retail Tariffs by User Type for the Power Sector and Marginal Costs of the Interconnected System, 1978-1987 (Current PesosikWh) Year Residential Commerical Industrial Total National Average Distor- Average Distor- Average Distor- Average Distor- Tariff tion a Tariff tion a Tariff tion a Tariff tion ia (2) (2) (2) (2) 1978 0.73 -63 1.39 -28 1.05 -36 0.93 -51 1980 1.22 -60 2.39 -19 1.78 -29 1.55 -47 1983 2.72 -51 6.27 25 4.59 3 3.77 -25 1985 3.92 -57 10.52 26 7.83 6 5.99 -28 1987 6.20 -64 18.10 13 13.06 - 7 10.02 -38 /a Distortion defined as 1-ratio of average tariff to marginal cost. 23. This was not the case for commercial users and the commercial tariff. Here, there has been a systematic improvement in tariffs, resulting in an average level in 1987 of some 132 above LIMC compared to about 282 below in 1978. Like commercial tariffs, industrial ones improved between 1978 and 1986 relative to costs, though some slippage occurred in 1987. The overall national average unit revenue improved from about 112 of LINC in 1978 to about 213 of this cost by 1987. Table 5.8 sets out how, given the growth in sales between 1978 and 1986, the deviations between tariff levels and LRMC translate into estimates of the economic subsidy enjoyed by different classes of users. Table 5.8: Estimated Gap Between Revenues from Total System Sales and Long Run Marginal Supply Cost According to Consumer Type 1978-1986 (1986 US$ millions) Year Residential Commercial Industrial Total National a 1978 - 230 - 30 -100 - 420 1981 - 240 - 3 - 50 - 350 1984 - 300 + 35 - 5 - 310 1986 - 400 + 15 - 35 - 470 La Total National includes the sum of residential, commercial, industrial, and 'other* consumers (not shown). 24. This table illustrates the following. First, the overall economic subsidy to power consumers was, in 1986, at US$470 million (1986 prices), only slightly higher than in 1978, when it was some US$420 million (1986 prices), - 85 - although it had decreased by up to 113 in the early 1980s. Second, between 1978 and 1986, the economic subsidy to residences has increased by about 70%. while that to Industry and commerce was significantly reduced. All in all, the cumulative economic cost of pricing electricity at the retail level well below its LR14C for the 7-year period, 1980-86, is estimated at some US$2.3 billion in 1986 prices. 25. The long history of very low tariffs relative to LRMC begs the qvestion as to how the power sector has survived financially in the review period? The reasons appear to be threefold. First, through a massive level of borrowing. Second, ICEL, CORELCA, and CVC have traditionally received significant capital transfers from the central government's budget. For example, according to unaudited DNP cash flow data for the power sector, between 1978 and 1985 the total central government capital transfers to the sector were about US$1.1 billion (1985$). Third, up to the mid-1970s, EEB and EPM owned low-cost supply systems. Thus, their average costs were lower than LRMC and, as such, they could sustain low tariffs, for a while in their systems, until costly expansion programs began in the late 1970s. 26. When the tariff distortions relative to LRMC are assessed across regional markets, significant variations, presented in Annex 5D.5, become evident that are masked by the more aggregated presentation above. The most important development these figures highlight is that, by the early 1980s, the tariffs in the EEEB market had become very distorted relative to LRMC, with residential, commercial, and industrial users paying 33?, 190%, and 140Z, respectively, of their marginal costs. Conclusions 27. During the review period retail tariffs became increasingly distorted (especially in the BEB market) relative to the structure of costs of serving different users. Residential users were, and continue to be, massively subsidized, with cimmerciallindustrial consumers increasingly charged in excess of their supply costs. Regarding tariff levels, these have remained significant- ly below the long-run average incremental costs, though there was substantial improvement up to 1983, but further deterioration thereafter. Indeed, the cumulative cost of pricing electricity below its average incremental cost between 1980-86 was about US$2.3 billion (1986 prices), with part of this being off- set by central government capital transfers of about US$1.1 billion (1986 prices) during part of this period, as well as a program of large-scale borrowing. 28. What appears to have continued clouding the issue of the need for retail tariffs to move closer to long-run average incremental costs is the fact that the financial consequences of not adjusting tariffs toward this level are modest in the short and even medium term. Indeed, during the long construction periods of hydro projects there do not appear to be financial difficulties as long as financing is available. Problems appear, however, as soon as debt servicing begins, which increasingly is several years before the plant is even commis- sioned. - 86 - 5.5 Economic Costs and the ISA Bulk Tariff Background 29. The issue of whether the bulk tariff level set by ISA is appropriate can be assessed by reference to the long run marginal cost at the intercon- nection level. Table 5.9 shows the average ISA bulk tariff in current pesos (excluding short-term energy sales) and the reference long-rn marginal supply cost (this cost being derived from Table 5.5). It makes clear that marginal costs at the interconnection level were about 3.5 US cents/kWh in the mid-1980s. However, ISA's bulk tariff has been consistently below these costs, varying from about 112 of marginal costs in the late 1970s to about 2/3 in the early 1980s, but declining to just under 60% of costs in 1986. This reinforces the picture that the issue is not that the bulk tariff has been too high, but, indeed, that retail tariffs in the regional markets have been much too low to permit adequate margins to retailing companies importing bulk energy priced at LMC at the Interconnection level. Table 5.9: Deviation Between Bulk ISA Tariff and LRMC at Interconnection Level, 1978-86 Estimated Gap between Average Revenues from ISA ISA Bulk Estimated LMC at Sales and LRMC Year Tariff Interconnection /a Deviation of Bulk Services ----- (Current COL$/kVh)---- (%) lb (1986 US$ millions) 1978 0.52 1.09 - 52 - 53 1980 0.78 1.66 - 53 - 59 1982 1.54 2.43 - 37 - 47 1983 1.80 2.95 - 39 - 55 1985 3.11 4.87 - 36 - 86 1986 3.93 6.89 - 43 -122 La Obtained from Table 5.5 and expressed in current terms using the Composite Deflator Index in Annex 5A.1. Ab Deviation defined as 1- ratio of the bulk tariff to the marginal cost. ISA Bulk Tariff Policy, Energy Transfers and Corporate Structure 30. From a commercial viewpoint, energy interchanges in the inter-connected system have been regulated by distinguishing so-called 'basic" and loptimizingg energy; these concepts remain valid and correspond roughly to (i) energy needed by a system to supply its demand and (ii) energy that is bought or sold in order to reach a minimam cost dispatch. Bulk tariff policy has revolved mainly around "basic, energy prices; optimizing energy gave rise to substantial controversy in the first years of interconnection but its price, - 87 - reflecting short-run marginal costs, together with a split savings approach, has not been seriously disputed. The development of bulk tariff policy between 1971 and 1986 is summarized in Annex 51.1. 31. The process of ISA's bulk tariff evolution has to be seen as one of trial and error, which has sought to reconcile ISA's shareholders' differences. These differences originate in multiple objectives vis-a-vis ISA and the interconnected system. Some of the principal attributes that have influenced bulk tariff design ares * A perceived need for regional energy independence; * The desire for low-priced electricityl an* * The appearance of equity between power pool members. 32. The energy independence problem lies in shareholders' reluctance to believe that an investment in ISA is as good as if they made it themselves. This fear stems from a rooted perception that in a scarcity situation, supplying customers could turn into a free-for-all debacle. The physical reason for this comes from the fact that, if a given company insists on not curtailing its load, there .s no way to impede it unless interconnection lines are opened, thus creating a massive rationing. Consequently, rationing, such as in 1981, (1) cannot be easily measured or monitored and (ii) a noncoopera-tive attitude cannot be punished in any *reasonable* way. Indeed, power systems confronted with a scarcity situation face a *prisoner's dilemmag in which the rational (economic) outcome consists of noncooperation, to the detriment of all. Consequently, the present ISA operations agreement dedicates a substantial number of its articles to specifying how energy and capacity should be apportioned in cases of scarcity. Offsetting this worry is the fact that the system survived the 1981 rationing situation in a reasonable way, which shows that cooperation for the common good is possible and effective. The economic counter-argument to the preceding discussion would be that in the face of scarcity, tariffs should be raised in order to dampen demand, however, low short-run price elasticities, as well as political impediments, would probably foreclose this alternative. 33. Regarding the equity issue, bulk tariffs have gone through (i) the initial years when sellers were allocated tariffs 'at cost,' (ii) the middle years when they were based on the Chivor cost, and (iii) the later years when a common tariff was established based notionally upon ISA's, as distinct from total system, expansion costs. The pertinent question is whether the bulk tariff mechanism apportions, in an equitable way (as perceived by ISA share-holders), the costs of system expansion and operation. 34. An argument that has been advanced by some parties in Colombia against any bulk tariffs consists in observing that ISA members are paying for local investment costs in proportion to their share of energy and capacity in ISA's new power plants. Therefore, additional funds collected from energy sales should be dedicated exclusively to paying debt service and covering O&M expenses. These could be calculated for each power plant, distributed according to each company's share, in exchange for which energy and capacity would be delivered proportional- - 88 - ly to plant generation. The counter-argument to this simplistic position is that ex-ante conditions, when plants are distributed, do not coincide with ex-post operational conditions because of changes in basic parameters (e.g. demand development) or because system optimization calls for short-term changes in the generation strategy. Consequently, the common bulk tariff mechanism appears as a means to reach a short-run economic dispatch and to supply power pool members' demand in a reasonably agile way. 35. Proceeds from bulk sales enter ISA's 'debt service and operating cost melting pot' with no distinction as to whether they should be applied to any one given project. This means that members with substantial participation in costly projects are subsidized indirectly by buyers with greater participa-tion in low cost projects. If equity considerations dominated, bulk tariffs should strictly be set by the buyer, proportionally to each pool member's participation in plants of the expansion program. This would set the struc-ture back by 15 years, with separate purchase tariffs for each utility and the consequent discussions regarding Ofair" vs. 'unfair" prices. 36. Though the interconnected system has been able to agree on a single bulk tariff structure that serves to generate funds for servicing ISA's debt and covering operating and maintenance (O&M) expenses, but not any significant contribution by ISA to new ISA investments, another point of view consists of posing the question in economic terms, i.e. what is the role of the bulk tariff as a price signal within the interconnected system? In a conventional, decentralized market, prices are supposed to convey information about desirable and undesirable decisions to invest and consume. Given the structure of the Colombian power sector over the past two decades, investment decisions regarding new plants did not respond to pricing signals. Indeed, the power companies' role in most investment decisions has been limited to claiming the right (as in the 'Acuerdo de Cali') to develop facilities on their own for the purpose of maintaining their energy independence rather than in response to the market price of bulk energy transfers. 37. Another role of prices for intermediate goods (bulk energy in this case) is to provide signals for an economic allocation of resources at the consumer level. Indeed, the single most important power tariff in Colombia is the bulk tariff--it sets the tone for all retail tariffs, especially now that close to 452 of supply is derived from power interchanges. Yet, after the late 1970s, the sector and the Bank have given little attention to this key price. The reaction of ISA shareholders with respect to the bulk tariff level has been to consider it, in some cases, to be 'too high* in relation to their own sales and consumer level. In this respect, it is worthbwhile analyzing the margin between bulk tariffs and retail prices. Annex 58.2 shows the Bulk Tariff /Retail Tariff ratio by end-user (with reference to the bulk purchase price of each company). A ballpark figure to judge these values would be that this ratio should not be above 0.5, thus allowing for the companies to recoup their network costs. However, from Annex 5E.2, the average national ratio has been above this value for most of the time, except between 1974-76. The residential ratio has been consistently between 0.6 to 0.8, thus leaving too small a margin for network costs. The corresponding ratios for commercial and industrial consumption are b':low 0.5. In the residential subsector the worst (highest) ratios occur in EPM, - 89 - EM88 and ICBL (not shown). However, EPM is not substantially affected as it is able to meet its demand largely from its own generation, whereas ESEB and ICEL, with their large residential market and large imports, are more sensitive to the small margin and have sought to compensate it through cross-subsidies from their industrial and commercial users. 38. Annex 5.2 also shows the ratios of the ISA bulk tariff to average retail tariffs in regional markets between 1974-86. This ratio reached a maximum (ranging from 0.51-0.61) for ICEL and SE between 1976 and 1979, when the Acuerdo de Cali was agreed. This possibly reflects shareholders' expectations that power plant ownership would be a means to assure lower cost energy for themselves (by comparison with EPM) and strengthening their position. The more correct interpretation of the ratios would have been that their retail tariffs, for the residential market in particular, were and are too low. From this point of view it can be said that the bulk tariff has not had the effect of exerting upward pressure on retail tariffs but rather that energy importers have viewed it as being ': oo high.* 39. With the interconnection of the CORE_CA system in 1984 a new element of conflict has emerged regarding the level of the ISA bulk tariff and the Government's policy of continuing to subsidize heavily natural gas used as fuel by CORELCA Empresa. Such subsidies were introduced prior to inter-connection as a means of reducing the differences in retail tariffs between the all-thermal CORELCA system and the rest of the country, in which hydro was dominant. The level of this subsidy was financially about US$0.3911,000 cubic feet, or 60% of the gas price to CORELCA. However, this price was, in turn, about one-half of its economic cost. This has resulted in the unit cash operating and maintenance (O&M) costs of CORELCA Empress being much less than the ISA bulk tariff between 1984 and 1986, as shown below in Table 5.10. For example, with the gas subsidy the unit cash 0&M costs were around 75Z of the ISA tariff in 1984, declining to 60Z in 1986. This impeded economic substitution of thermal generation and provided little incentive for CORELCA to increase imports of bulk electricity. Indeed, even if the financial subsidy on gas were removed, these cash O&M unit costs would still have been slightly below the ISA bulk tariff (see Table 5.10). Of course, if gas had been priced at its economic cost, this would not have arisen, as then the unit cash 0&M costs of CORELCA Empress would have been around 50Z above the ISA bulk tariff in 1986, providing a strong incentive for increased electricity imports and less gas use. * 90 Table 5.10s ISA Bulk Tariff and CORELCA Eupresa's Unit Cash O&M Costs, 1984-86 (Current Col$1kVh) 1986 1985 1984 ISA Bulk Tariff 3.93 3.11 2.37 Cash O&M Cost per Net kWh Generated / 1. With Gas Subsidy 2.40 1.74 1.78 2. Without Gas Subsidy 3.69 2.81 2.88 Ratio ISA Bulk Tariff to Cash O&H Unit Cost (with Gas Subsidy) 1.64 1.79 1.33 Ja Excluding energy purchased by CORELCA from ISA. 40. What all of this indicates is that with the CORELCA interconnection, natural gas pricing for power nse and the ISA bulk tariff level can no longer continue to be viewed in isolation. Under Bank Loan 1583, that financed the Atlantic region interconnector, it was covenanted that the guarantor would end the gas subsidies when interconnection was achieved. Somehow, this had not occurred up to 1987, and the Bank appears to have accepted this non-compliance because of the interconnector's limited energy transfer capability. 41. The conflicts surrounding ISA's bulk tariffs originate in the institu- tional set up of the company: a shareholder is both an investor and a user of ISA services and, therefore, has conflicting objectives. On one hand, it seeks to obtain a return on investments in ISA, while on the other, it wishes to have cheap energy in order to improve its own financial position while changing its own retail tariffs as little as possible. Indeed, the above discussion has illustrated that during the review period the ISA bulk tariff has essentially played the role of a mere accounting notion for the interconnected system rather than that of an economic signalling tool. This shortcoming is rooted in the structure of ISA and the Colombian power sector, whereby ISA's shareholders, as a matter of policy, do not wish the company to generate meaningful financial surpluses after meeting debt service and its O&M expenses. In a bizarre sense, there is a perception that the shareholders consider their own position strengthened if the position of their own company, ISA, is kept weak. The bulk tariff needs to acquire more of its key economic signalling role communicating across the entire interconnected systam the message of what long-run system expansion costs are. Of course, with the bulk tariff set at LRMC key resource mobilization goals for ISA would be rapidly achieved. From the standpoint of resource allocation at the consumer level, the bulk tariff also has not been effective as an economic signal since by not being raised to its LRMC level it has been unable to act as a pressure for the increase in retail tariffs that are required in all markets. - 91 - Conclusions 42. A significant lacuna in the Bank's tariff dialogue with Colombia has been the issue of the bulk tariff between ISA and its shareholders- -the most important power price in Colombia since it directly impacts close to 45% of the energy available. During the review period ISA's bulk tariff has served an exclusively accounting role to generate funds essentially to cover only ISA's debt service and operating costs since ISA's self-financing ratio after debt- service has been negligible between 1977-86. Given ISA's structure, however, the desirable role of the bulk tariff as a fundamental price signal conveying economic information regarding desirable and undesirable generation investments and consumption patterns has been missing. Traditionally, generation investment decisions by companies have been driven more by their perceived need for minimal imports of energy and their 'energy independence' rather than in response to the market price of bulk energy transfers. In other words, the way generation investment decisions are made the price of bulk energy appears of little consequence. In the absence of any 'economic competition* between companies regarding future power supplies for the bulk system, the price of bulk energy supplies does not act as an economic signal for investment decisions. The bulk tariff needs to acquire more of its key economic signalling role across the interconnected system, indicating what long-run system expansion costs are. 43. The severe downward pressure on the bulk tariff has been reinforced by the inadequate margin between retail and bulk tariffs. This downward pressure has resulted in the bulk tariff being almost 50Z of the average IC at the interconnection level between 1978-80, with this improving to about 702 by 1983, but declining to around 60? by 1986. Clearly, had the bulk tariff been closer to the average IC at interconnection since 1980, ISA's financial performance would ha-e been significantly improved during the decade of the 1980s. 44. The main lessons to be learned in regard to ISA's bulk tariff are: (a) Though it is the most important power price in the country, it has not formed the centerpiece of tariff policy in the past decade. Indeed, the issue of the bulk tariff needs to move to the forefront of the Bank's dialogue on the reform of the sector's tariffs, finances, and structure. (b) The low level of the bulk tariff has had a negative impact on ISA's financial performance and, in this context, there is a need to move this tariff up towards the average IC at the interconnection level. In addition, such a policy shift would provide the proper signals regarding economic costs to ensure that retail tariffs in all markets are moved to their appropriate levels. (c) The benefits of interconnection of the CORELCA region with the rest of the country cannot be maximized unless natural gas for power use is priced close to its economic cost, and the current subsidy is eliminated. - 92 - 5.6 Tariff Estimates and Internal Rates of Return at Appraisal Background 45. As poor internal resource mobilisation has been an endemic problem in the Colombian power sector, in general, and in virtually all the power cam- panies, a key issue addressed here is the degree to which the Bank's ex-ante analysis signalled the inadequacy of tariff levels. Of course, the corollary of this is, if such signals were there, how effective was the Bank in pursuing this core problem to a satisfactory conclusion? Fundamental to an assessment of the ex-ante analysis are two questions. First, from a purely financial standpoint, at what level should the tariff have been set? Second, how does this level compare with economic costs, which were estimated earlier? Defining *Appropriate' Levels of Tariffs from a Financial Standpoint 46. As far as the 'appropriate* level of tariffs is concerned, from a purely financial standpoint the basic issue iss what costs are power consumers expected to bear? There are four areas of relevance. First, there is general consensus that power consumers should at least cover the full cash operating and maintenance costs of supplying electricity. Second, power consumers are expected to pay prices that allow the utility/sector to cover fully its debt servicing costs (including tiat arising from foreign exchange risk). This means, in particular, that adjustments to tariffs have to be undertaken following devaluations of the local currency. This would not appear to have been the Government's policy in Colombia, post-1984. Indeed, even up to the end of 1987, the effects of the 1985/86 accelerated devalua-tions had not been passed through fully to power consumers. Third, power tariffs are to be set such as to cover, say, a significant share of local currency costs of the sector*sutility's expansion plans, after meeting debt service; if so, this is likely to require a self-financing ratio (after debt service) of about 35Z-401. It would appear that Colombia's policy has never been to achieve anywhere remotely near to this for the overall sector, though for short periods some utilities have done so. 47. During the 1960s and early 1970s, it was normal for the Bank to call for self-financing ratios (after debt service) in the range of 30Z-40Z in the power sector of several middle-income countries. Somehow, these standards of utility performance appear to have eroded dramatically since then.(2) Why? Several reasons seem to have led to this, such as (i) higher investment levels in the wake of the oil crisis, (ii) low tariffs, and (iii) much increased operational inefficiency, which inevitably accompanies deteriorating business- like attitudes and falling profitability, as long as it is considered that governments always reward poor performance through bail outs and retention of utility managers even though they may have failed to arrest problems within their areas of responsibility. Allied to these factors, the desire of the Bank to accelerate lending, in itself, is likely to have contributed toward a willingness to accept slippage in the standard of financial performance previously considered acceptable. 48. The fourth question of relevance for defining what costs consumers should bear is: what are the repercussions on public sector finances of the - 93 - power sector running sizeable and sustained financial deficits? Clearly, if the power sector is unable to meet the local currency costs of its investment program from its own resources, this local currency gap has to be financed invariably through contributions from government (equity or loans), since the sector's utilities are generally not creditworthy enough to raise directly capital on the local market. A consequence of this often is an increase in the government's fiscal deficit. In Colombia, the cash deficit of the power sector remained at about 202 of the consolidated public sector deficit during 1981- 1984, rising to around 1/3 by 1985. This meant that the power sector's deficit in 1985 amounted to about 1.2Z of GDP, about 113 of which was financed by net internal credit and the remainder by net external credit (see Section 8.3). 49. This highlights the power sector's impact on the macro financial situation, an issue which gradually entered the dialogue between the Bank and the Government in the 1980s. Indeed, because of the sector's increasingly high share of public sector investment, its inadequate internal resource mobilization, and large external borrowing, its proper finucial management became an element in any effort to achieve overall macroeconomic goals. These linkages between the power sector and macroeconomy are discussed further in Chapter 8. What needs to be mentioned here in connection with setting tariff levels from the financial viewpoint is that, in middle income countries the benefits for public finances of having power consumers meet a significant share of the sector's local currency investment costs afte= covering debt service and O&M costs, are too often not accorded sufficient importance. 50. Of course, governments may decide to subsidize some power consumers, either directly through the tariffs or indirectly through financing with grants network expansion in rural areas, etc., in order to achieve social objectives. Such policies tend to lead away from the financial objectives referred to above. However, the key issue is to what degree? The evidence in the case of Colombia is that the power consumer (about 602 of households by 1986) has enjoyed a 'windfall of subsidies* which negatively affected public sector and utility finances. Tariff Estimation and the Bank's Appraisal Process 51. The Bank's appraisal process relies on the determination of the internal econ3mic rate of return (IERR) to assess the adequacy of power tariffs. Since, however, the economic benefits of such projects are not easily assessed, what is calculated is an adiusted internal financial rate of return (IFRR), this being the discount rate that equalizes the present worth of incremental costs and incremental revenues. IEERRs and IFRRs were estimated in the SARs for all projects in the Colombian power sector that the Bank financed in the review period, except for the FEN operation in 1984. An assessment of these ex-ante rates of return provides considerable insight into the levels of prices that were considered necessary to yield an IERR close to the opportunity cost of capital in Colombia, then considered to be in the 112-12% range. Before commenting on them, however, there are some general observations to be made. 52. In the case of opetations before 1979, i.e. Guatape II (1972), San Carlos I and II (1978 and 1979), 500 KV Interconnector (1978), and Mesitas (1978), the * 94 - IERRs were estimated on the basis of investments and benefits directly associated with the projects being financed. However, after 1979 the est'mation of IERRs changed. In particular, in respect of the Bogota Distribation I and II operations (1980 and 1985), EPM's Guadalupe IV and Playas (1980 and 1981) projects, and ZZEB's Guavio (1981) project, IERRa were based on the entire investment program of the company over a peviod, which included the project in question, rather than on the more narrow focus of the project being appraised. Of course, often power projects are interdependent, and it is difficult to segregate rationally benefits from different parts of an investment program. In such cases, a program approach has to be undertaken in estimating the IERR, though, in most of the above operations, what was being financed by the Bank were specific projects and generation projects, in particular, and not time slices of the investment program. However, care must always be exercised in doing this since looking at the broader investment program can overstate or understate the IERR of the project in question. This arises because the rate of return of the program implicitly allows cross-subsidization between more and less economic subprojects. These issues become very relevant in the case of power distribution projects (see Annex 5F.1). Indeed, what is needed is both a determination of the program IERR for the entire investment program of the company, as well as that of the project being financed within the program, and the components of the project, e.g. the generation and distribution elements separately. 53. A further observation concerns the absence of any ex-ante estimates of the IERR or 1AR for the consolidated power sector's investment programs during the period. The only such attempt was made in the recent 1987 Power Sector Adjustment Loan. In both the Guavio (1981) and FEN (1984) operations financial projections were presented for the consolidated power sector, yet in neither SAR were IFRRs or IEERRs presented for the entire sector investment programs. In the case of the Guavio operation, it was the first time that financial projections for the consolidated power sector were prepared and this was an important step forward that the Bank had called for. The SAR noted that this was *to demonstrate the viability of the Acuerdo de Call proposal and to ensure the feasibility of ISA's shareholders' contributions to the financing of Guavio.1(3) The SAR went on to note that these projections were of a preliminary character and "aimed at displaying the financial outlook (of the power sector) for the 1980-1990 period...and have to be deemed as a basis for discussion, requiring further analysis among the electric sector entities and the Government.8(4) Consolidated Power Sector Investment Programs Ex-Ante IFRRs 54. The ex-ante financial and economic rates of return, based on the financial consolidations of the power sector's investment program, are assessed as they were viewed in 1980 (Guavio appraisal), in 1983 (FEN apprai-sal), and in 1986/87 (Sector Adjustment loan appraisal). This chronological view is critical since it illuminates the points in time at which key signals on pricing and investment program size became apparent, and the responses which these signals elicited from the Bank. - 95 - Consolidated Power Sector Bx-Ante IFR anc. IER at the Time of the Guavio Appraisal - 1980 55. Annex 51.2 sets forth an ex-ante estimate of the IFRR for the 1980- 1988 sector investment program, based on the consolidated financial projec- tions in the Guavio SAR. These projections envisaged that the sector's unit revenue would increase between 1980 and 1988 about 602 in real terms (based on the projected composite deflator implied in the Guavio SAR and shown in Annex 5C.3). The ex-ante IFRR turned out to have been 12.8t. Adjusting for border prices would result in a value of the ICUR somewhat higher. Though neither of these indices were presented in the Guavio SAR for the overall sector investment program, what these high values indicate is that at least from an ex-ante standpoint the projected unit revenue levels for the sector in the Guavio SAR (1981) were adequate to ensure the financial and economic viability of the sector investment program. Given this situation, how did the sector retail tariff evolve between 1980 and 19867 Up to 1983, the actual tariff was about 10% above projected levels in real terms, however, this performance reversed after 1984 so that by 1986 the actual sector unit revenue was about 14%, in real terms, lower than what was called for in the financial projec-tions (Annex 5C.1). Compounding this, sector sales in 1986 (as discussed in Section 4.4) were about 30? below the levels envisaged at appraisal, resulting in the sector revenue being about 60Z of that projected at appraisal. Consolidated Power Sector Ex-Ante AIFRR and IERR at the FEN Apraisal-1983 56. In 1983, during the appraisal of the FEN operation, the Bank undertook another assessment of the consolidated power sector's financial prospects, based on the overall sector investment program for the period 1983-1987. By then, however, the situation had radically changed from that perceived three years earlier. First, energy losses in virtually all markets (except EPM) had increased significantly. Second, energy sales' growth had collapsed, compared to what had been foreseen in 1980. Third, the Mesitas and Guavio projects had already experienced significant cost overruns and their commissioning dates had been delayed significantly. Fourth, the Latin American debt crisis had erupted in 1982, with associated high real interest rates, and though Colombia was still viewed as creditworthy, part of the foreign co-financing needed for the Guavio project was not forthcoming. Finally, between 1980-1983 international inflation decreased rapidly while inflation in Colombia continued at an average 25% annually, resulting in increasing over-valuation of the peso as its devaluation rate did not offset the appreciation of the currency. In other words, at the time of the FEN appraisal in 1983, most perspectives relating to the power sector were much worse than in 1980. Yet, this could not be discerned from the FEN SAR, though it was essentially a rescue operation providing supplemental financing for the Mesitas project, while establishing yet another new institution in the power sector. 57. Government originally considered an institution such as FEN desirable in order to facilitate primarily domestic resource mobilization through the local capital market. Of course, a necessary condition for the viability of such a sector financial intermediary is the existence of creditworthy power companies, however, prospects in this regard at the time of appraisal pointed - 96 - towards a sharp deterioration in creditworthiness of the few utilities that had previously been considered creditworthy by the Bank. This was particularly the case of EBB. In this context, specific measures to address the creditworthiness of companies in the sector, and in particular the internal resource mobilization of both the sector overall and the companies in the sector, appeared to be lacking. 58. The SAR presents extensive financial data on all sector companies, as well as projections (except balance sheets) for the consolidated power sector covering the 1983-1987 time slice of the overall investment program. However, it fails to estimate the IFRR or IBRR for the program. As shown in Annex 5F.3, the ex-ante IFRR, based on the sector's average unit revenue projections set out in the SAR (Annex 5.36), is a very low 6.5Z. To achieve an ex-ante IFRR of 10.3Z, the tariff level would have to have been increased to Col$7.0/kWh (USO8.91kh) in constant 1983 prices, which was around 90Z above the actual 1983 national unit revenue level. However, as far as the consolidated power sector was concerned, the FEN SAR envisaged virtually no change in the sector's unit revenue In real terms during 1983-1987, as the tariffs were projected to increase by 1.82 annually in real terms over the four years (SAR, Annex 5.36). 59. The SAR's silence about the above negative ex-ante signals is strange as massive investments were involved (some US$6.4 billion over a 5-year period). In fact, the 1983-87 sector investment program was neither financially nor economically viable at the tariffs set out in the SAR, yet the Bank proceeded, and did so, for the first time to the Colombian power sector, with a B-loan. The SAR appeared to take comfort from the fact that the on-lending of Bank resources by FEN would only benefit projects the Bank had previously appraised and found satisfactory--i.e. projects already under construction. The SAR noted in this context, *a return on investment for the project has not been quantified. However, it has been established that the projects which are prospective sub- loan candidates for Bank financing would have higher internal rates of return than the 13%-15% estimated at the time of the original Bank appraisal. Tariffs have increased more rapidly than projected, while average costs are approximately the same.-(5) It is difficult to understand the latter statement since, by 1983, the Mesitas and Guavio projects (the two largest being financed by the Bank) had already experienced significant cost overruns (hence the EN supplemental financing for Mesitas) and extensive delays. 60. A closer look at the composition of the 1983-87 sector investment program indicates that 43Z of the investment over this period was intended for new projects (CVC's Calima III, ICEL's La Kiel I, and CORELCA's Urra I and II) and it was by no means clear that these were part of a least-cost program. The breakdown of the program between on-going and new projects in this investment program was as set out in Table 5.11, with new works accounting for over 502 of investment in 1985, rising to 842 by 1987. This, together with the inadequacy of the projected tariffs, both factors discernable at the time of the appraisal, should have induced a more thorough discussion of the planned development. - 97 - Table 5.11s Breakdown of 1983-1987 Power Sector Investment Program (Percent) 1983 1984 1985 1986 1987 On-going Works 88 77 44 28 13 New Works 9 21 53 69 84 Source: FEN SAR, Annex 5.38. 61. Other related ex-ante signals which were available as warnings that something was amiss between tariff levels and investment size are the high average incremental generation costs for the 1983-87 sector investment program and the low InRs obtained for the investment programs of the individual companies as set out in the FEN SAR. The average incremental generation cost is estimated in Annex 5F.4, which uses the generation investment and required generation data from the SAR. The incremental generation cost in 1983 prices turns out to be some USc4.1/kVh. These costs become about US*5.lkVh if interest during construction is included. This cost was high and, indeed, represented as much as 802 (when about 502 could have been a more reasonable proportion) of the estimated generation cost for a base-load oil fired steam plant, which would have represented the highest cost power generation source in 1983. Yet, the 1983-87 expansion plan was dominated by mostly low-cost hydro plants. This situation was evidence that too many generation investments were being concentrated to meet demand increments that were not sufficiently great to justify them. The absence of comments concerning this high incremental generation cost (which reflected a sub-optimal plan) is, again, disturbing. 62. The other area of concern with the FEN appraisal was the low rates of return obtained for the individual power company investments. The particular case of EEEB is set out in Annex 57.5, which shows that the ex-ante IFRR for EEB investments over the 5 year period was only 4.42 if ISA's share of Guavio was taken into account. Of special relevance, however, was that, without Guavio, the IFRR for the EEB 1983-87 investment program remained a low 8.9%. This was a clear indication that either the EEE tariffs assumed (USo6.9/kVh in constant 1983 prices) were too low or IEEB's investment plan, even without Guavio, was over-sized relative to projected energy demand. Consolidated Power Sector Ex-Ante IPR and IERR at the Time of the Power Sector Adjustment Loan - 1987 63. The major new works in generation envisaged under the 1983-87 investment plan did not materialize. By 1985/86, continued low electricity demand growth, increasing concern about supply over-capacity, accelerated devaluation of the peso, and a deepening financial crisis in the power sector forced important cutbacks in the investment program. Though some work on the Urra project was started by CORELCA this was eventually brought to a stop, while the La Miel I and Calima III projects were postponed. The Bank played a very important role in achieving this reduction of the investment program, especially in its decision not to then participate in the financing of the Urra project. - 98 - 64. The Bank approved in November 1987, the first adjustment loan in the sector designed to assist Government reforms. Unlike the previous two Bank operations that assessed the consolidated power sector finances, the President's Report for the adjustment loan did also estimate the IERR/IFRR for the overall 1987-1990 sector investment program; this return resulted at 13%, based on two important assumptions: first, between 1987 and 1992, the sector's unit revenue would rise by about 32 annually in real US$ dollar terms; and second, the unit cash operating and maintenance (O&M) costs would decline in the period by about 6.5% annually in real terms. The basis for the latter assumption appears to lie in the decline, between 1984 and 1986, of the cash O&M costs for the overall sector, expressed in US$ (see Annex 6 of President's Report). However, OED does not consider this drop at all representative of normal conditions in the sector for two reasons. First, half of this two-year decline was due to a one-time drop in fuel usage in the system, as more hydro capacity was commissioned and the CORELCA region was interconnected to the rest of the country. Though some further decline in fuel use can be expectedwhen CORELCA imports more energy from interconnected system in the future, these are likely to be limited since thermal plants have to continue to play a role in the system. Second, the other half of the two-year decline stems from the accelerated devaluation of the peso in 1985 and 1986, which directly translates into much lower O&M costs for those years if expressed in US$ terms, as is the case in the President's Report. 65. The above consideration led OED to consider that the President's Report assumption of a steep decline of unit cash O&4 costs was unrealistic and, in particular, that it was not supported by experience over the 1971-86 period (see Annex 5F.6). Annex 5?.? re-estimated the IFRR for the sector investment program on the basis of OED's more conservative assumption that unit cash O&M costs decline by a more modest 1.42 annually. Under this assumption the re-estimated IFRR becomes 9.32. This return increases to 13.32 when the sector's unit revenue is raised to USc5.5/kWh in 1987 UJS$ throughout the 1987-1991 period. This is around 392 above the actual tariff level in 1987 and still about 252 higher in 1991 than the level envisaged in the President's Report. What is again illuminated is that from an ex-ante perspective OED's estimate of the appropriate level of tariffs for the overall qector is at variance with that set forth in the President's Report. 66. The 1987-1990 sector investment program represented a cutback in investment of almost 602 in real terms compared to the 5-year investment program assessed at the time of the FEN appraisal for the period 1983-1987. It is for this reason that the tariff level required to make the 1987-1990 investment time slice financially viable is only about 55% of what would have been needed to justify the 1983-87 sector investment plan. However, it would still require tariffs to increase by about 72 annually in real US$ terms between 1987-91. The issue is that the average rate called for is more than double the 32 annually set out in the President's Report. The almost secondary relevance of tariff levels is stated eloquently in the loan documents-- while on average tariffs are now quite reasonable (the financial problems of the sector are mainly due to heavy debt service and unused capacity, not to rate levels), and the amount of further adjustment required relatively modest, the tariff structure is quite distorted.0(6) * 99 - Conclusions 67. The analysis of ex-ante signals of the sector's unit revenue requirements perceivable ilt 1980, 1983, and 1986187, at the time of the Guavio. FEN, and Sector Adjustment loan appraisals (summarized in Annex 5F.8), yields five conclusions. First, the tariff levels for the overall sector at the time of the Guavio appraisal in 1980 were appropriate ex-ante. Second, the inadequacy of tariffs at the overall sector level was very much in evidence since 1983. Third, the 1983-87 sector investment program set forth in the FEN SAR was ex-ante neither financially nor economically viable and, indeed, appeared divorced from a least-cost program. Fourth, the EE investment program for 1983-87 supported by the FIN operation was either over-sized (evsn without the Guavio pro ject) relative to projected demand, or ZEEB's tariffs needed to be significantly increased. The FEN SAR did not highlight these realities. Fifth, in the Sector Adjustment loan of 1987, the Bank once more failed to acknowledge the scale of the inadequacy of tariff levels, which, according to OED estimates, needed to increase at the sector level at about double the rate set out in the President's Report. Ex-Ante Analysis of Individual Company Projects Appraisals Before 1979 68. Up to 1979, the Bank's appraisals of power projects in Columbia focussed on the benefits and investments directly linked to the project the Bank was financing. Of the five operations financed between 1972 and 1979, this review looked at three in greater depth, to ascertain the perceivable ex-ante signals about tariff levelas ISA's San Carlos I and I projects and MBB's the Mesitas project. What is worth stating at the outset is that in the case of individual project appraisals, whether before or after 1979, OED's findings are in no way similar to those concerning the program appraisals of the consolidated power sector analyzed above. In the latter case, the ex-ante facts that were not highlighted are significant, however, for the individual project appraisals, the issues OED raises below largely concern focus and alternative methodologies. The San Carlos Complex-1977 and 1978: Ex-Ante IFRR and IERR 69. The ISA's San Carlos hydro complex consists of four parts: San Carlos I and II, partly financed by the Bank (and IDB); as well as the Calderas and Jaguas developments, which did not attract Bank financing. In the economic appraisals of the San Carlos I and II projects the SAR focussed on the appropriate level of retail tariffs in the Central and Atlantic regions in order to justify investments dominantly in generation, but also small investments in transmission and distribution (see Annex 5G.1). There was an important limitation in this approach since ISA, the owner of the San Carlos complex, is a wholesaler and not a retailer of power. In this situation, the wholesale price of San Carlos power to ISA's shareholders that made the project viable was of particular interest. A second and independent component of this analysis would, - 200 - therefore, have been to determine the retail tariff in the Central and Atlantic zones that justifies the distribution Investment based on the wholesale power price equal to that which made the San Carlos complex viable. 70. By not illuminating ISA's exclusive wholesale function, the Bank avoided focussing on the vital role the bulk tariff had to play as one of the main instruments for influencing retail tariffs (through the *cost-push' approach) throughout the entire country, irrespective of whether or not the Bank was lending to a specific regional utility. Annex 50.1 re-estimates the ex-ante IERR for the San Carlos I project, treating ISA as a wholesaler of power. Taken in terms of benefits, the average 1976 ISA bulk tariff of US40.721kwh results in an IERR of some 5.7?, whereas to obtain an I=R close to the opportunity cost of capital a bulk tariff rate of some USc1.60/kWh in 1976 prices would have been required. This implied that ISA's bulk tariff needed to increase by about 1202 in real terms between 1976 and 1982, the date when San Carlos I was expected to be commissioned. By that date, the ISA bulk tariff had risen, in reality, to 95Z above its 1976 level in real terms. This was a significant achievement in the context of the San Carlos project. 71. In Annex 5G.2 the ex-ante IERR is re-estimated for the entire San Carlos complex in which it is again treated as a wholesaler of power. In this case, an appropriate rate of return is achieved when the bulk tariff is set equal to the then current ISA bulk tariff in 1978 of USo1.3/kWh (1978 prices). The signal here was that the ISA bulk tariff in 1978 was reasonable to ensure viability of the San Carlos complex, as a whole. Mesitas-1977 Appraisal (ESEB)-Ex-Ante IERR 72. The economic appraisal of the Mesitas generation project estimated the IERR for the combined generation and related sub-transmission and distri-bution components. There is, of course, nothing wrong with this approach. However, as the project proposed for financing included primarily generation, at a period when several other large generation projects were about to be financed, it was important to know how that generation project would rank relative to that of the San Carlos complex. The analysis presented in the Mesitas SAR did not shed light on this type of issue, which is taken up in the present evaluation. 73. The Moaitas SAR estimated the 1RR to be about 10.4%. In deriving this, a total of some US$42 million (mid-1977 US$ market prices) In benefits was attributed in 1982 and 1983 to fuel savings from the commissioning of Mesitas in 1982. This was incorrect (see Annex SG.3). Ironically, the line of reasoning in the SAR (para. 6.03), which explicitly claimed the existence of fuel savings, conflicted with para. 6.02 of the SAR, which implicitly refuted their existence. In fact, there were no fuel savings. Replacing the fuel savings benefit by the benefit of EED sales to ISA for 1982 and 1983 at the bulk tariff (since all of Hesitas's output could not be absorbed initially in EE's system) lowered the IERR by 0.6Z to 9.82. 74. Given that the entire output of Mesitas could, however, be absorbed in 1982 by the interconnected system, what was the ex-ante bulk tariff that would have rendered the project viable? This is estimated in Annex 5G.4. In order - 101 - to arrive at an I8 of 11.82, the bulk tariff would need to be about USol.7/kWh (mid-1977 market prices), compared with the actual ISA bulk tariff in 1977 of US1.1/kMh, i.e. a 551 increase above this level by 1982 when the project was commissioned--in 1982 the ISA bulk tariff had risen to around USel.6/kWIh (in 1977 prices). This level of the bulk price is roughly what was required also for the San Carlos I project, which demonstrates the competitiveness of the Mesitas generation project. Appraisals After 1979 75. After 1979 Bank economic appraisals of power projects in Colombia were based solely on an appraisal of the entire investment program of the relevant company over a period that included the project in question. Once more, this represents an element of the economic justification process but, by no means should it be construed as sufficient to justify the project being financed, especially when the project may only represent a small component of the overall corporate investment program in the period of relevance. Clearly, in the case of large projects (such as Guavio), which sufficiently dominate the investment program of a company, a program appraisal can, in effect, be close to the appraisal of the project, itself. However, when the investment program is not dominated by the appraised project it is necessary to look also at the economic justification of the project proper. This is very relevant when it comes to distribution projects, which often represent a small fraction of total investment over a period, and, above all, are frequently *sacrificed* (in toto or in part) as the company runs into cost overrun problems on its major generation projects and, therefore, cuts back on planned distribution works. Bogota Distribution I - 1979 Appraisal - Ex-Ante IFRR 76. As explained in Annex 50.5, the SAR for this project based the analysis of the internal rate of return on EEEB's entire construction program between 1979 and 1985 and not on EE's distribution program (of which the Bank-financed project was but a part) or on the Bank-financed element of the distribution program. The key issue In appraising distribution projects is how the upstream costs in generation and transmission are captured and then whether they have been captured correctly. In the case of a company that only distributes power, such as EMCALI and many of ICEL's and CORELCA'S electrificadoras, these upstream costs are reflected in the price of the bulk electricity purchased. Of course, to capture fully the economic upstream costs it is essential that the bulk price used as input to the appraisal of the distribution project be the LRMC of bulk supply. In the case of companies such as EEEB, which are generation and distribution companies but also purchase significant amounts of power from ISA, the proper accounting of full economic upstream costs is most easily achieved by treating the company as a distribution entity purchasing bulk power at LRMC. 77. As shown in Annex 50.5, for the Bogota Distribution I project, if this approach had been followed at appraisal, it would have shown that EEEB's retail tariff needed to be around US05.01kfh (1979 market prices), or some 202 above the level recommended In the SAR. if the company's 1979-85 distribution investment program was to have an IFRR of 11.12 based on upstream costs reflected through the bulk tariff at marginal costs. - 102 - EMP Guadaluye IV - 1980 - Zx-Ante IFR 78. The SAR for this project estimated both the IFRR and IERR for the whole of EPM's construction program between 1979 and 1986 at a low 72, based on 1978 EPM retail tariff levels; however, if retail tariffs were raised by about 50Z in real terms above this level, the IER for the company's program increased to 132. Guadalupe IV, as a separate project, is shown in Annex 5G.6 to yield an IFRR of 11.62 when the projected ISA bulk tariff is taken as the benchmark for benefits and an IERR of 182 when using the overall system generation level LRMC as the benefit (around USQ2.6/kWh). This clearly shows Guadalupe IV's inframarginal character and the fact that it was an eminently desirable generation project. RPM Playas - 1981 - Ex-Ante IFRR 79. The Playas SAR of February 1981, was based upon findings of Septem- ber/October 1980. This was less than a year after Guadalupe IV (dated May 1980, and based upon finding3 of November/December 1979). As in the case of Guadalupe IV, the economic evaluation of the SARwas performed for RPM's entire construc- tion program. This indicated an IERR of 15.5%, based on an increase in EPM retail tariffs of about 702 in real terms above their 1980 level of US0l.8/kvh (December 1978 prices). When viewed separa#2ly, however, the Playas project, as shown in Annex 50.6, was a marginal deveils ant. Indeed, using the projected ISA bulk tariff of USo2.05/kVh (December 197v prices) as benefit, the IFRR of Playas was only 82, increasing to 112 if the bulk tariff benefit proxy was elevated by about 27t. Conclusions 80. There are two broad findings which the above review of SAR's for the individual company projects has illuminated. First, the lack of the Bank focus on ISA's bulk tariff is very apparent from the manner in which ISA's San Carlos I and II projects were appraised from the economic standpoint. Indeed, ISA was not treated as a power wholesaler, since the assessment of San Carlos's worth was not based on what the ISA bulk tariff should be, but rather on what level retail tariffs in the Central and Atlantic regions needed to be to provide an adequate return. Second, in the case of distribution projects (Bogota Distribution I and II), it is important that, at least, the distribution program as a whole be appraised and not only the entire invest-ment program of the company. This was not done for the Bogota Distribution I or II projects. If the suggested approach had been followed in the case of Bogota Distribution I, and upstream costs had been captured through the use of the economic supply cost of bulk power, then the need for BEEB's retail tariffs to be 202 above that recommended in the SAR would have been illuminated. - 103 - 5.7 Cost Differences Between Regional Power Markets Background 81. A major characteristic of the different regional power markets haf been, and remains, their widely different costs. These differences have tended to complicate the task of setting tariffs and it Is essential that the dynamic trends underlying the differences be understood since some such discrepancies are likely to persist well into the future, particularly those between EPM and the other companies. In what follows the factors giving rise to the cost differences between the power companies are discussed, followed by a look at the differences between SPM and EB from the ex-ante incremental cost standpoint during the 1980-1988 period in order to judge the minimu differential that needed to en it between their tariffs. The section concludes with an assessment of the implications for tariffs of the lack of a financial transfer mechanism across the sector. Factors Influencing Cost Differences 82. The cost differences between the companies arise from three factorsa (i) Differences in the bulk supply costs of electricity. These bulk supply costs are made up of two elements. One coming ;rom the cost of generating electricity from plants within their own markets and the other from the cost of bulk power Imported into the markets. The differences in the overall bulk supply cost of power are likely to decrease with time, at least between the markets of ESEB, ICEL, CVC, and CORELCA, as all markets are likely to become roughly equally dependent on bulk energy imported into their systems at a cost lower than that at which they can generate power from new plants in their own systems. To becme reality, however, this requires more effective control at the national decision- making level to prevent sub-optimal projects of regional companies from finding their way into the national *least-costa generation investment plan. (ii) Differences in sub-transmission and distribution network costs. This comes from differences in market sizes and structure. In the case of ICEL, the area of coverage is vast (close to 500,000 sq. km.), the pcpulation density low (about 26 persons/sq. km.). and the sales structure dominated by residential sales (57Z of total sales for the consolidated ICEL group). This Implies low load densities and an extended sub-transmission and distribution network resulting in high costs for these parts of the power delivery system. This will not change significantly in the short or medium term. Though the area of coverage is large in the case of CORELCA (some 130,000 sq. km.), and the population density only about double that of ICEL, the sales structure is dominated (47Z of sales) by industrial and commercial consumption (see Annex 4A.4), unlike any market other than EMCALI. Despite this situation, CORELCA's load densities are still relatively low due to the extent of coverage of the system, resulting in sub- transmission and distribution costs on the high side. Markets with - 104 - lower sub-transmission and distribution costs are those of the urban utilities, ENCALI, 3PM, and EBUB, which are characterized by spatially concentrated high density loads. (iii) Weak management. Management quality varies dramatically across the Colombian power sector. For example, whatever standards of measurement are used, the performance of EEB's management in the past several years has been inadequate and in sharp contrast to that of EPM's. This is likely to be due partly to the lack of autonomy of REEB's management from the Bogota municipality, whereas EPMs management has tended to establish greater autonomy from the Medellin municipality. Costs are affected by poor management through, inter alia, incapacity to control, let alone reduce, power losses; virtual loss of control over escalating accounts receivable; and the promotion of investment levels unaffordable to a company. RPM and ZEEB Ex-An&e Incremental Costs in the 1980s 83. EEEB and EPM inherited, in the early 1970s, a low cost generation infrastructure, which, coupled with high load density markets, enabled them to maintain low tariffs since their average costs were lower than their incremen- tal costs and significantly so in the case of SPM. This period was already drawing to a close for EEEB in the late 1970s and is likely to do so also for EPM by the early part of the next century. This becomes apparent just by tracing the developments in the BEEB and ZPM systems. 84. The generation system of Bogota was based on the Salto hydro chain, which required no dam construction and little tunneling (water connection takes place through external penstocks that feed external power houses). This was followed, in the late 1970s, by the Mesitas project, which was justified as a means to absorb the additional water provided by the Chingaza water supply project. Indeed, Mesitas represented the last high head development on t-..e Bogota river complex. Although this project encountered construction delays a' cost overruns, it relied also, in part, on existing infrastructure and coul, I classified, therefore, as a 'medium' cost development. The essential point is that by the early 1970s the 'low cost' hydro developments in the REEB area had been commissioned and by 1976 EEEB became a large importer of energy from the interconnected system, with imports accounting for about 301 of its energy requirements by 1986, the year before Mesitas was commissioned. The development following this, Guavio, is of the *high cost' category. In other words, EEEB has been on a rising incremental cost curve since the mid-1970s, with its average costs moving up rapidly towards its incremental costs. 85. In the case of EPM, this system saw the last of its really 'low cost' developments through the Guatape II project, which was commissioned in 280 and which benefited also extensively from existing infrastructure. This scheme enabled EPM to maintain its position as a competitive not energy exporter into the mid-1980s. The development of Guadalupe IV, which vas commissioned in 1986, represented a 'medium-low cost' project, which benefited, again, from existing - 105 - infrastructure and, in turn, extended RPM's position as a competitive net energy exporter until the late 1980s. The first "high coat* RM project was that of Playas, which should have been commissioned in 1988. 86. In order to illustrate the significant differences in costs between the EEB and SRP systems an assessment of the ex-ante Incremental costs, for each of the systems separately for the period 1981-1988, was undertaken. Data for these *utmates were based on each company's investment plans (in their own plantsO .ontained in the Guavio SAR for 8888 and the Playas SAR for EPM. The purpost -if this analysis is to highlight the much higher ex-ante level of average incremental costs RB faced compared to EPM and, therefore, to provide a basis as to why REED tariffs had to exceed those of BPM by an increesing margin compared to the mid-1970s, when their tariffs differed by 152-202 (see Annex 5B.2). 87. Annex 5H.1 shows a breakdown of EEEB's energy requirements, variable generation expenses, and investments. As estimated in this annex, the ex-ante incremental cost between 1981-88 for the EER expansion program was Col$3.OkWh (1981 prices). This annex also provides a detailed breakdown of RPM's costs based on data taken from the Playas SAR. Given that RPM is not always a net purchaser of energy from ISA, the variable generation expense was calculated as the difference between its revenues from sales in its own market and the gross revenues estimated In the SAR. The estimated ex-ante incremental costs between 1981-88 for the RPM expansion program was 1981 Col$2.2/kVh. Theamain concludan of this analysis is that for the period 1981-1988, from an ex-ante standpoint, REEB's average incremental costs were about 35Z higher than equivalent costs for EPM, indicating that this was the minimum differential by which EE average tariffs needed to exceed those of EPM in the 1980s. Absence of Financial Transfer Mechanism - Tariff Implications 88. Until the early 1980s differences in tariff levels were accepted owing to differences in generation costs and, above all, the absence of interconnec- tion. In particular, tariffs in the CORELCA service area were higher than those in the central region due to thermal generation, which was recognized to be of higher cost than hydro generation in the rest of the country. 89. With the commissioning of the 500KV interconnection line it was assumed that generation-level costs should become approximately the same for all utilities. In consequence, tariff unification became a goal with the purpose of reducing electricity prices in the CORELCA area as well as rationalizing tariff policy throughout the country. The problem was addressed in a 1983 study(7) by ISA which concluded that equalization of tariffs on a national level was not feasible due to the *financial disequilibrial it would cause for most utilities; however, it recommended that two first steps could be taken in order to reach better balanced tariffes (i) a unification of tariff structure; and (ii) a unification of tariff levels on a regional basis. 1 206 90. Despite the difficulties identified in the ISA study, concern about the myriad of disconnected regulations governing tariff policy resulted in a strengthening of the Junta Nacional de Tarifas (JNT) and, specifically, the promulgation of Decree 2545 of 1984, which established a single, country-wide tariff structure. This decree, notwithstanding its very general wording, created the basis fort (M) a conon criterion for residential billing according to social income strata; and (ii) a general voltage-level orientation for non-residential tariffs. It also marked an end to the operation of JNT as a weak regulating agency that had limited itself to rubber-stamping the utilities' rate proposals and made it into an executive Tariff Board with power to mandate rate reforms. 91. The application of Decree 2545 was not immediate; in some cases, such as that of ZEEB, resistance to JNT's guidelines delayed the application of the decree. However, JNT prevailed and ZEEB had to adopt its provisions; to this end, JNT produced Resolution 020 of February 1986, which contained specifi provisions for a program of tariff reform within EBBE following JNT's regula- tions. 92. Having reached its first goal of adopting a common tariff structure, JNT proceeded to seek a unification of levels. A general framework was established through Resolution 086 of November 1986 that set retail tariff goals based on average long-run incremental costs. Implementation of Resolution 086 is an ongeing process in which JNT has taken an active role by eliminating the utilities' ability to charge lower rates: indeed, in the past, JNT limited itself to approving ceilings, vtereas it is now enforcing actual levels. JNT is also steering a pragmatic course on a case-by-case basis in order not to create major difficulties that could eventually lead to the demise of its effort to reach a unified tariff. 93. From the previous discussion, the absence of a mechanism for financial transfers within the sector appears to have been the most signifi-cant obstacle to the goal of tariff unification. Under current institutional arrangements, it is not possible, for example, to transfer any financial surpluses (at such an unknown future time as they may arise) from utilities with surpluses to those in deficit. This absence of an institutional means to transfer resources between utilities has often been claimed also to be a source of the disparity in the realtive subsidies received by consumers in rural and urban areas, though hard evidence to this effect is not easy to identify. What is discussed below (see Section 5.8) is the significant difference in residential tariffs between urban and rural areas. Though this phenomenon has become less pronounced since about 1984, its change cannot be attributed to the institution of financial transfer mechanisms. 94. There are two relevant issues that need to be raised. First, how important is the goal of 'unification, of tariff levels on a regional basis? and second, how essential are financial transfers within the sector, and, if - 107 - so, what objectives should they achieve? Tariff *unification* is an Important goal if it means that tariffs throughout the country are set according to principles which are applied uniformly, as distinct from retail tariffs that are equal across all markets. For example, the principle that tariffs to each consumer class in each market must cover the costs imposed by that consumer could be such a principle. However, In order to off-set inequities and deal with political realities the principle could further be set that low-income residential users enjoy a discount for some of their consumption. Finally, the principle could imply that the cost of these discounts are covered in urban markets through cross-subsidies from higher income residential clients and from industrial and commercial users in that market, but that in rural markets, these costs are met by a direct Government subsidy to the affected utility to the degree that the above cross-subsidy approach cannot cover completely the costs. Such a system would ensure a unified price policy with clear and specific objectives as to who pays for what and who subsidizes whom and by how much. 95. In terms of the second issue, it is by no means evident to OED why significant financial transfers need to occur within the sector between the five major retail companies, EPM, EEEB, CVC/ZMCALI, ICEL, and CORELCA. First, there are no financial surpluses in any of these companies and this is likely to remain so for some time. Second, there is the fundamental problem of undermining the incentive in the efficient and better managed companies, if parts of their surpluses are channeled to weaker companies, some of which may be structurally weak because of weak market characteristics and low load densities, while others may be managerially weak and inefficiently run. Clearly, one does not reward such inefficiencies by financial transfers from more profitable companies. Conclusions 96. Despite interconnection, cost differences remain between the power markets in Colombia. These arise due to differences ins (a) network costs and load densities; (b) supply costs due to differences in generation mix and dependence on bulk imports; and (c) management efficiency. These differences will persist into the future. Similarly, tariff differences will persist between markets in the measure that costs are reflected in tariffs and that a financial transfer mechanism cannot be implemented within the sector. 97. Tariff *unification* has become a desirable goal in terms of uniformly applying principles relating tariffs to costs across the sector. Yet, tariff differences cannot become or remain too large since this exacerbates regional political difficulties. This suggests that lacceptable' tariff differences between regions is one area of policy that now needs agreement. 98. There appears to be little basis for establishing financial transfer mechanisms across the sector between the utility companies serving the five major markets. The political concensus does not exist for the creation of such mechanisms and on economic grounds it is likely to reward inefficient management in some markets. The absence of such mechanisms implies that tariff levels in each market be set so that costs are covered in that market. To the degree that this cannot be fully achieved, either on equity grounds or because of market structure, then the onus would be on Government to transfer directly to the - 108 - is achieved regarding who pays how auch, to whom, and for what. Other avenues could also be explored. For example, utilities continue to be exempted from income taxes and, as such, fiscal policy measures oriented toward creating a mechanism for cost compensation have never been assessed. This is an avenue of analysis that could yield benefits as an alternative to excessive reliance on inter- or intra-sector subsidies. 5.8 Income Distribution Background 99. This section looks at the issue of the income distribution effects of pricing policy, the degree to which improvements have occurred, the Bank's role in this aspect of policy, and the lessons to be learned from these developments in the review period. Subsidies and Retail Tariff Policies 100. Subsidizing residential electricity consumption, in general, and lower income groups, in particular, has been a constant theme in the ta1ff policy discussion in Colombia (e.g. Guatape II SAR, 1972, para. 2.09). This is a natural expression of the political will to provide low income households some access to electricity for uses that may be identified with certain *basic needs, ' such as lighting. In most countries, such subsidized service has tended to be limited to the first 50-100 kVhlmonth or less of consumption. 101. However, in Colombia very large subsidies (up to 801 of the supply cost) are extended for residential consumption as high as the first 200 kWh/month. Given that residential electricity use per subscriber averaged for the country, was some 235 kWhImonth in 1986, it is evident that the massively subsidized first consumption block covers the entire consumption of about one-half of &ll residential users. To achieve the objective of subaidizing low consumption or low income households the utilities have relied on two tools. First, intra- residential sector subsidies involving financial transfers between low and high income residential consumers, within each market, due to the *progressive' nature of residential tariffs. Second, there are inter-sector subsidies, whereby industrial and commercial users are charged in excess of their costs-of-service so as to off-set the financial losses on residential sales. To the extent that financing gaps still exist, a third mechanism is then invoked--direct Government budgetary support. However, these well-intentioned measures depend on whether there are enough subsidizers able to shoulder the cost burden of other consumers. A dramatic illustration of the limits of inter-sector subsidies is provided by EEB. 102. As discussed in Section 4.3, in terms of energy sales, the E2EB market has been transformed from a dominantly commerciallindustrial one to a residential one. From the standpoint of inter-sector cross subsidies this is significant. For example, in 1965 a cross subsidy to residential consumers of US*1/kWh required a surcharge on industrial and commercial consumers of some USC0.551kMh. However, by 1985 the same level of cross subsidy required a surcharge of USel.16/kVh, i.e., an increase of about 1111. The indiscriminate implementation * 109 - of such cross-subsidisation, compounded by the effect of priceconsumption elasticity, can plunge a utility and a sector Into a vicious circle of demand increases responding to very low prices (the subsidized sector) and dwindling demand from the subsidizing sectors. This is what has occurred in several of the regional power markets, and the overall power sector. In the review period. 103. Intra-sector subsidies Imply steeply rising prices with increasing consumption by residential users, as illustrated In Table 5.12, which sets forth the case of ams Table 5.12s NPUs Lowest and Highest Priced Residential Blocks 1968-84 1968 1973 1975 1982 1984 Lowest Priced (current Col$IkWh) 0.12 0.14 0.15 0.76 1.27 Highest Priced (current Col$IkVh) 0.25 0.40 0.65 5.15 14.50 Ratio 2.1 2.9 4.3 6.8 11.4 Here the ratio between the highest and lowest priced residential tariff blocks has increased from just over 2:1 to more than 11st between 1968 and 1984. The rationale behind this type of structure is the existence of a positive cor- relation between consumption and income. While true for the basic consumption unit (single family-single dwelling unit), a number of factors can invalidate its practical application. Foremost among these is the concept of user: for the power utility a user is associated with a meter, whereas, in reality, a meter can cover multiple homes and families. Consequently, boarding houses with high metered consumption in major Colombian cities, where the poorest sectors of the urban population live, ironically could pay the highest tariffs. 104. Increasing dissatisfaction with the tariff system in 1984 led to Important reforms. The main change was to associate tariffs with what are termed 'social-strata. ' Stratification work was initiated in 1980181 by DAUM in Bogota with Government issuing Decree No. 2545 in October 1984. The key provisions regarding residential users were the establishment of a fixed basic tariff having six levels or *strata' according to the classification of the construction quality of the dwelling, and a variable tariff with five consumption blocks. The specifics are given in Annex 5J.1. This represented a very Important advance in Colombian efforts to reform tariffs by standardizing, in the variable tariff portion, the sizes and numbers of consumption blocks, and secondly, for the fixed portion, by providing through the "strata' classification a better discrimination between households of different incomes. This system has only been in force for some 5-6 years. 105. In 1986, with the issuing of JNT Resolution 086, building on the achievements of Decree No. 2545, a further major advance was made in tariff policy. Government assumed full control of power tariff policy by removing the utilities' discretionary facility to request and apply authorized rates. - 110 - Henceforth, the utilities must apply the tariff set by the JUT, which for the first time incorporated explicit criteria based on efficiency and economic rationale by linking tariffs to the average long-run incremental cost (IC) of supplying electricity to different consumers. This set the framework for tariffs to be structured, across the country, on the basis of some rational principles, replacing the virtual state of tariff *anarchy, that prevailed earlier. The specifics of Resolution 086 are set out in Annez 5J.2. Of course, in setting targets based on the structure of costs of service and social considerations, Resolution 0896 sets out for residential consumers levels of discount from the IC which vary depending on the consumption block. For example, for consumption between 0-200 kWh/month the residential tariff could be set between 202-30Z of IC (see Annex 5J. 2). Despite the tremendous advances in tariff policy that Decree No. 2545 of 1984 and JNT Resolution 086 of 1986 represent, the analysis below highlights some important problems which still require resolution and provides a basis for understanding how tariff policies have contributed to the sector's financial woes. Intra-Residential Sector Subsidies 106. Annex 5J.3 shows the variable and fixed components of residential tariffs (as of June 1987) under the new structure in four different regions--Atlantico (within the CORELCA system), Caldas (within the ICEL system), Medellin (EPM) and Cali (EKCALI). In the case of Bogota, UEEB had not yet implemented the JNT tariff structure of Resolution 086 and, therefore, it is not included in this analysis. Also shown in this Annex are the average residential use per stratum and relative distribution of users between strata in the four different regions. 107. The key question to be answered is: Who subsidizes whom and by how much? As far as the intra-residential sector subsidies are concerned, this is best demonstrated by calculating deviations from the average price paid by the residential sector as a whole, independently from its relation to the *right* tariff. Estimates of the subsidies received/provided by each stratum are given in Table 5.13 for the four regions based on the methodology outlined in Annex 5J.4. This table shows that there are three subsidized strata in Atlantico, Medellin and Cali (I to III), with two in Caldas (I and II), and three subsidizing strata in the former regions and four in the case of Caldas. In all situations, at least 314 of residential consumers enjoy an intra-sector subsidy from the remaining 114 of residential users who are in the higher strata. In other words, the intra-sector subsidy structure appears to meet its objectives in that the burden of subsidizing falls largely on the higher income groups. The only anomaly in the subsidy structure is apparent in Atlantico where the lowest stratum (1) receives a smaller subsidy than either Strata II or III. The subsidy per user in Table 5.13 can also be viewed in a different way by breaking it down into the portion corresponding to the fixed charge and that arising from the variable energy charge--this is shown in Annex 5J.5. Table 5.13: Estimated Bill Averages and Subsidies Per User - June 1987 (000's Col$ per month) Stratum Bill/Average Stratum Subsidy/User ATL CAL MDE EKC ATL CAL MDE EMC I 0.8 0.8 1.3 1.9 I 0.4 0.2 0.5 0.8 II 1.4 0.8 1.7 2.3 II 0.7 0.1 0.4 0.6 III 2.1 1.0 2.0 2.4 III 0.4 -0.1 0.2 0.4 IV 2.9 1.4 2.2 2.8 IV -0.1 -0.6 -0.2 -0.03 V 3.6 1.9 3.0 3.8 V -1,4 -1.6 -1.2 -1.3 VI 6.6 2.3 4.7 5.9 VI -5.6 -2.6 -4.0 -3.4 ATL = Atlantico; CAL = Caldas; MDZ = Medellin; and E - Cali. 108. Despite the classification errors which are bound to be encountered in separating 3.5 million residential users into six strata, it would appear that the stratum concept provides a reasonable proxy for income and undoubtedly has represented a major advance in Colombian tariff policy. On the other hand, in regard to the question of whether the increasing residential tariff operates adequately as a supplementary income discriminator, although there are strong correlations between wealth, income and electricity consumption, it is not obvious that the latter can be used as a good discriminator. This is illustrated in Table 5.14 by the following distribution of residential users in Strata I and VI according to consumption block in Cali and Medellint Table 5.14: Percent of Residential Users kWhIMonth Stratum I Stratum VI Cali Medellin Cali Medellin 0 - 200 40.4 43.6 15.9 9.0 201 - 400 44.8 33.6 18.3 8.5 401 - 800 13.6 21.1 33.4 28.3 801 - 1,600 1.0 1.7 23.8 40.8 1,601 + 0.2 0.0 8.6 13.4 Average kWhIMonth 247 266 777 972 109. This table shows that in Stratum I in excess of 314 of users are in lower consumption blocks (below 400 kWh/month), whereas in Stratum VI, 2/3 of the users in Cali and close to 7/8 in Medellin fall in the higher ranges. However, there is a significant number of outliers in both cases. Most disconcerting is the fact that 1/3 of Stratum VI users in Cali and about 115 in Medellin are enjoying the benefits of the very low tariffs in the first blocks - 112 * due to their low consumption. What this indicates is the need to concentrate the income distribution function in the fixed charge of the tariff with less emphasis on the variable part. One of the JUT's objectives is apparently to move tariff policy In this direction, which seems reasonable in light of the preceding. Inter-Sector Subsidies 110. The above analysis has looked at tariff differences between strata without any reference to the global subsidy received by the residential sector from commercial and industrial consumers. When considering the residential sector as a whole, the comparison is no longer against an average tariff, but against the cost of serving this low-voltage market. The benchmark cost used by ISA(S) to reflect the incremental costs at the low-voltage (LV) level, across the entire system, is December 1986 Col$12.80/kWh, which is equivalent to Col$14.8/kVh (June 1987 Pesos) using the composite deflator index (see Annex 5A.1). Of course, as noted earlier (see Section 5.4), ISA's 1986 estimate of the LV incremental cost is about 15Z below that derived from the 1979 ISA/EPM study due to the differences in estimates of network costs. However, purely for the purpose of illustrating the effects of the inter-sector subsidies, the estimate of LV incremental costs from ISA's 1986 study has been used here. When this value of the cost of service to low-voltage users is applied to the same average residential consumption data (see Annex 5J.3) used in the intra- residential sector subsidy analysis, the costs per user-month and economic subsidy per user shown in Table 5.15 result. Table 5.15s Economic Cost of Service to Residences and Implied Economic Subsidies to Users - June 1987 (000's Col$) Stratum Economic Cost/User-Month La Economic Subsidy/User-Month Lb ATL CAL MDE EMC ATL CAL MDE EKC I 1.5 2.8 4.0 3.7 1.2 2.2 3.1 2.6 II 2.7 2.9 5.3 4.6 2.0 2.2 4.0 2.8 III 4.2 3.6 6.3 4.8 2.5 2.5 4.5 2.7 IV 5.7 4.8 7.0 5.4 2.8 2.9 4.5 2.6 V 7.3 6.8 9.2 7.5 2.3 3.3 5.0 2.4 VI 13.2 8.2 14.5 11.6 1.0 3.3 5.8 2.3 Average 2.1 2.3 4.4 2.7 ja This is the average consumption level (Annex 5J.3) multiplied by IC estimate as of June 1987. b This is the difference between economic costluser-month and the total monthly bill given in Annex 5J.4. - 113 - 111. Compared to the estimate of intra-sector subsidies (Table 5.13), the picture is entirely different. Indeed, upper strata users (IV to VI), due to their higher consumption, receive a higher absolute economic subsidy per user than the lower strata (I to III), except in the case of DMCALI where all strata get about the same absolute subsidy. In Medellin and Caldas, Stratum I residences receive about 1/2 and 213, respectively, of the level of absolute economic subsidy of Stratum VI. This Io nothing new. The 1972 OED evaluation raised this as a fundamental problem of tariff policy. It remains so today. 112. When one expresses these economic subsidies in terms of kWh consumed and as a percentage of cost, the distortions are less apparent, but equally real, as shown in Table 5.16 below. This table highlights that, on average, the higher income groups (Strata IV to VI) receive a lower subsidy, expressed as a percentage of supply cost or per kWh consumed, than the lower income groups. However, in all four markets shown in Table 5.16, Stratum IV enjoys a subsidy per user as high as 1/2 to 2/3 of the supply cost, while even for Stratum V this subsidy is still as high as 1/3 to 1/2 of the supply cost. Indeed, in the cases of Caldas and Medellin the subsidy for Stratum VI is still around 402 of the supply cost? Table 5.16: Economic Subsidies to Residences per kWh Consumed and as a Fraction of Supply Costs (Based on June 1987 Tariffs) Subsidy per User as % Stratum SubsidylkWh (Col$/kVh) of Supply Cost ATL CAL MDI EMC ATL CAL MDE ENC I 11.3 11.7 11.8 10.6 76 79 79 72 II 11.1 11.4 11.3 9.2 75 77 76 62 III 8.9 10.2 10.5 8.5 60 69 71 57 IV 7.3 8.9 9.6 7.2 49 60 65 49 V 4.7 7.2 8.1 4.8 31 49 55 32 VI 1.2 5.9 6.0 3.0 8 40 40 20 Resolution 086 113. Current tariff policy calls for an adjustment of tariff levels along the lines set out in JNT Resolution 086 of 1986, which envisages that residential tariffs will vary according to Incremental costs (IC) and to the discounts/surcharges on IC set out in Annex 5J.2. Of course, as discussed earlier, JNT Resolution 086 establishes the principle of explicitly linking tariffs for consumption to incremental costs; however, as shown in Annex 5J.2, substantial discounts for residential consumption levels below 400 kWh/month are still envisaged. Given the picture outlined earlier regarding the scale of economic subsidies based on actual June 1987 tariffs, issues arise in connection with Resolution 086. First, from the economic signalling standpoint, the implications for inter-fuel substitution are negative if tariffs for the second residential block (200-400 kWh/month) are set at the minimum level enunciated * 114 - in the Resolution. At that level electric cooking could displace use of LPG on the basis of 1986 retail prices even though from the standpoint of economic costs cooking with LPG/natural gas should be favored. Second, given the large size of the first subsidized block (0-200 kWh/month) and the large subsidies envisaged for the second tariff block, it is questionable to what degree the Resolution really addresses the key problem of the sector, namely, the overall inadequacy of tarift levels to provide financial viability for the sector. Third. given that there is some flexibility built into the Resolution, the implications on the overall tariff level for the sector between application of the minimum and maximum ranges of the Resolution need clarification. Four, how the application of Resolution 086 impacts income distribution and the levels of subsidies currently enjoyed by residential users in Strata IV to VI is not clear either. One way of analyzing some of these questions is by assuming that the tariff levels corresponding to Resolution 086 are implemented on the basis of the June 1987 low voltage incremental cost (IC) estimate of Col$14.81kWh. This is done in Annex 5J.6. 114. In that analysis it was assumed that the discounts in the variable part of residential tariffs would be set at the mid-point between the maximum and minimum ranges of Resolution 086, as outlined in Table 5.17 below. Table 5.17 Residential Tariff Variable Part- Consumption Block Tariff as Percent of IC (kWh/month) Resolution 086 This Analysis 0-200 20 - 30 25 201-400 50 - 80 65 401-800 90 - 125 110 801+ 100 - 125 120 115. The conclusions from this analysis include, first, that even Strata VI users would still benefit from a small economic subsidy; the still well-to-do Strata IV and V would continue to enjoy subsidies of up to 1/4 to 1/2 of the supply cost, as shown in Table 5.18 below. Second, the discounts on consumption below 400 kWh/month remain so high that, when combined with the low average residential consumption levels, they lead to an average subsidy enjoyed by residences of between 1/2 to 2/3 of the supply cost. This implies that significant cross-subsidies would continue to be required from the commer- ciallindustrial to residential users. - 115 - Table 5.18t Economic Subsidy 2er User as X of Supply Cost (Mid-1987) (Based on Discounts at Mid-Points of Resolution 086) Stratum ATL CAL MDS ElC I 73 74 64 66 II 72 72 56 59 III 57 61 47 55 IV 47 49 38 47 V 29 31 20 27 VI 6 17 3 9 Average 47 65 43 45 116. The above highlights important deficiencies in current tariff policy as reflected in JNT Resolution 086. In this sense. the tariff structure issue is an integral element of the tariff level problem--the two can no longer be conveniently kept in separate compartments. This has come about because the structure has become so distorted that appropriate average levels for residential users can only be reached if the maximum levels foreseen in Resolution 086 were applied. Indeed, given that residential consumers represent some 48? of system sales and commercial and industrial together some 41?, it is evident that the system as a whole can never reach an overall average tariff level close to IC as long as residential tariffs are still massively subsidized as set out by the lower limits of Resolution 086 and an upper limit of 125% of IC is kept on the tariff level for commercial and industrial users. Either residential subisidies should be reduced or industrial/commercial users made to pay more than 125% of IC. 117. Based on the present structure of sales, and assuming that industrial, commercial, and official electricity users are charged 252 above their incremental supply costs, the average unit revenue for the whole system would only approach the IC level if the average residential tariff for the system roughly doubled. In this way, the average residential tariff for the system would be around 802 of the incremental supply cost. 118. Applying the maximum levels of Resolution 086 to the distribution of residential electricity consumption by tariff block in the regional markets and for the consolidated power sector, (Table 5.19), results in the cases of ICEL, EPM, and the overall sector, in levels of average residential tariffs of 55Z, 71%, and 642, respectively, of the incremental cost. This means that, even with the more rational structure promulgated by the JNT, the inadequacy of overall levels will persist unless action is taken to apply close to the maximum limits of the Resolution for residential tariffs and reduce significantly the level of discounts for consumption below 400 kMh/month. * 110 * Table 5.19: Distribution of Residential Electricity Users and Consumtion in Regional Markets by Tariff Blocks (Percent) Consolidated Power Sector Tariff Block RPM La ICEL (1985187) Lc (1982) (kWhimonth) Con- Consump- Con- Consump- Con- Consump- sumers tion Lb samers tion samers tion 0 - 200 21 n.a. 66 42 49 17 201 - 400 34 n.a. 23 28 29 30 401 - 800 38 n.a. 10 22 18 34 801 - 1,600) 1 6 4 12 >1,601 ) 7 n.a. 1 2 1 7 /a Distribution of consumers refers to 1986 data. b Distribution of consumption refers to 1982 data. Lc The data refer to the entire ICEL (Group) market, excluding the electrificadora of Boyaca. Rural Versus Urban Residential Subsidies 119. An area of pricing policy on which little attention was focused by the Bank in its power operations has been the persistent disparity between the large subsidies received by urban household electricity consumers compared to much lower ones received by rural households. This lack of Bank focus came about largely because (other than its one operation with CORELCA for rural electrifica- tion) the Bank did not venture into lending activities that confronted this type of issue, since its operations were mainly directed at the large municipal utilities and ISA. 120. Annex 5J.7 presents a comparison of average residential tariffs in four utilities with Istrongg markets (EPM, INCALI, EIB, and Atlantico--in the CORELCA market) and six utilities with 'weakw markets (Tolima, Huila, Narino, Cordoba, Sucre, and Cesar). This comparison extends over the period 1972 to 1987. This data highlight the following. In the case of EKED and the Cundinamarca utility covering the rural areas surrounding Bogota, average residential tariffs for MB were about 70Z of those in Cundinamarca in 1972, with this differential disappearing after 1983. Similarly, in the CORELCA region in 1972, average residential tariffs in the Atlantico area were 50Z of those in the rural Cesar area. Again, after 1983, these large differentials vanished, so that tot v. this rurallurban aspect of the income distribution issue of tariff policy is n -Iger a problem. This narrowing of the differential came about because in 1983, when tariffs were temporarily afrozen,' the 'freeze* was extended for ICEL and CORELCA companies for a longer period than for other companies. - 117 - Relative Prices and Costs of Modern Household Fols 121. Energy use by households in Colombia has continued shifting from traditional fuels, such as fuelvood, to modern fuels, like LPG, natural gas, kerosene, and electricity. Though, in 1986, traditional fuels supplied some 60Z of total residential energy in Colombia, most of this use occurred in rural areas; however, in the urban areas the modern fuels dominate and electricity provides the largest share of this market, as shown in Table 5.20. The dominance of electricity in the use of modern fuels in the residential sector is due both to its low price and the limited availability in some urban areas of alternative fuels for cooking, such as natural gas or LPG1 for example, natural gas is only available in the Atlantic coastal region. Part of this irony is that, in the case of LPG, the limited availability of this fuel has been a direct consequence 3f pricing policies (see para. 123 below). The importance of looking at energy use for cooking in household energy demand patterns is evident from the data in Annex 5J.8, which shows that cooking accounts for 60Z-8O of gross energy use in low-income urban households and for 40%-55% of such use in higher income residences. In this context, the pricing and availability of household cooking fuels is a key determinant in residential electricity use. Table 5.20s Urban Residential Energy Consumption Modern Fuels Use - 1985 1017BTU (Z) Electricity 9,170 gVh 31.3 52 Natural Gas 3.6 x 106ft3day 1.3 2 Kerosene 3,970 bble/day 7.7 13 LPG 10,150 bbls/day 13.9 23 'Cocinol* 3,130 bble/day 5.5 9 Total 59.7 100 122. There are severe limitations to assessing the degree to which, in the review period, low residential electricity prices in all regional markets have stimulated inefficient energy use in urban households. First, not all fuel supply mixes are available in all major markets. This is discussed in Annex 5J.8. Second, altitude variations lead to important differentials in the efficiencies of thermal cooking devices compared to those using electric resistance. Third, up to 1984 the residential tariff structures differed widely between regions. Despite this, estimates of comparative cooking fuel costs and prices from 1975-86, based on retail fuel prices in this period, are shown in Annex 5J.9. A number of assumptions have been made in drawing up these comparisons, which are set out in Annex 5J.9. 123. What is apparent from the data in Annex 5J.9 is tbat the distortions between prices and economic supply costs are large; that they have been prevalent over the entire period; and that they affect all fuels. At low altitudes, the economic cost of cooking fuel (LPG and kerosene) was between 2/3 to 112 that of electricity before 1980 and since 1986. Even during 1980-85, when oil prices - 118* peaked, the fuel costs were just about equal to that of electricity. Prices. however, sent different signals, as (i) that of electricity was often lower than that of LPG and kerosene, and (ii) that of kerosene was almost twice as high as that of LPG. Taking Into account that kerosene is the fuel preferred by the lower income households, as evidenced by its high level of use in Stratum II households in Barranquilla and Cali, for example, (see Annex 5J.8), it seems that income distribution concerns were not prevalent In the case of kerosene pricing policies. These data also indicate that, although kerosene cooking at low altitudes wa more costly to the consumer than LPG, the poor used the former fuel more, possibly because of the higher fixed entry costs of LPG cooking, in the form of cash deposits for cylinders, etc. This was, however, not the complete picture. 124. The low controlled price of LPG, in turn, resulted, after 1974. in a rapid decline in LPG availability, since by the mid-1970s Colombia became a net oil importer and LPG imports were unprofitable at such retail prices. Figure 5.1 shows the sharp drop in LPG availability between 1973 and 1979 followed by its increase from 1979 to 1987 (roughly doubling the supply level). However, after LPG prices had risen to levels that brought forth more supplies, the prices were still well below economic costs of supply. The collapse in LPG availability between 1974 and 1979 amounted to about 720,000 barrels, which, if it had been replaced solely by electricity, would have increased residential electricity use by around 350 GVh, or 62 of total residential demand in 1979. The signals conveyed by LPG and electricity prices for cooking at low altitudes changed during the period 1975-1986, with the price of LPG lower to about 1978, which was of questionable significance since the availability of LPG was decreasing. Between 1978 and 1982, however, the electricity price was lower, and after 1982, there was little difference in the prices. Figure 5.1 COLOMBIA.POWER SECTOR LPQ Produdon and Consumpdon Iw two 72 n 74o s a amown. - Peedu. ..*...gSen%V * 119 - 125. For low-income households, at low altitudes, and with no access to natural gas, price signals tended to favor, systematically (by more than 502), electricity rather than kerosene for cooking over virtually the entire period. although, before 1980 and after 1986 kerosene provided lower cost cooking and, that even !n 1980-85. when pertoleum prices were at their height, the differences in economic costs were within the + 152 uncertainty of these estimates. This indicates that pricing policies for kerosene and electricity have encouraged low-income users, at low altitudes, to increasingly use electricity during the review period, though on the basis of economic supply costs, this was not justified. 126. Natural gas currently is only available in the Atlantic region and its price to households is very low, rendering it the cheapest energy source for cooking. As shown in Annex 5J.8, stratum 4 residences in Barranquilla used natural gas to meet up to 362 of their energy needs. However, access to this fuel by low-income households appears very limited due, presumably in large part, to the high infrastructure costs for a gas distribution network. This highlights another anomaly in Colombia's household energy pricing policies, namely, that the beneficiaries of low gas prices are not necessarily the low income groups, because the delivery system often does not reach them. 127. In terms of the economic supply costs of natural gas, data is very limited. The estimated supply cost at the well-head for gas from the off-shore Guajira region is about US$1.00/million BTU,(9) based on a 12% discount rate. However, given the sparse data on gas transport costs in Colombia, estimates of the economic supply cost of gas delivered to different parts of the country are not readily available. The OED evaluation team has, therefore, been unable to estimate the level of economic subsidy implied in the delivered gas price to households in Barranquilla and the Atlantic region. However, it is evident that as long as there is a reasonable industrial demand for natural gas in a given region promising a good capacity utilization for a pipeline, the development of a more extended natural gas pipeline infrastructure is likely to be economically competitive with other means of providing household energy (LPG, electricity). The threshold for the gas cost at delivery is about 1987 US$9-10/million BTU (1987 prices). 128. At high altitudes, in particular Bogota, the fuels of relevance in the review period are LPG, 'cocinol," and electricity. From the economic supply cost standpoint there was rough equivalence between these fuels before 1980, with electricity gaining a clear advantage, as petroleum prices peaked between 1980- 85 and the situation reversing itself since 1986. What have been the price signals? First, 9cocinol,* the poor man's fuel, has been heavily subsidized, with the result that about 68% of the energy needs of stratum 2 households in Bogota were met by this fuel. Second, as far as LPG and electricity were concerned, prices signalled throughout the period that the latter be used. This price signal, compounded by the constraints on LPG availability up to 1979, is likely to have resulted in electric cooking becoming increasingly prevalent in Bogota among the Stratum III and above households. * 120 - Conclusions 129. The income distribution objective has been pursued through power pricing policy, providing large subsidies to residential electricity users. The economic cost of these subsidies has increased from some US$230 million (1986 prices) in 1978, to about US$400 million (1986 prices) in 1986. This was achieved mainly through two mechanisms- intra-residential sector transfers and inter-sector cross-subsidies from industrial and commercial users. The intra-sector subsidies achieve desirable transfers from high to low income residential users. The new tariff structure, introduced through Decree 2545, includes a fixed charge that reflects the household's income level; this is one tool for such transfers, which proved quite successful in meeting part of its objective. However, in OED's judgment, a further concentration of the Income distribution function into the fixed charge of the residential tariff is warranted. This would relieve the pressure for meeting income distribution objectives through adjustments of the variable part of the tariff and for the associated further distortion of the tariffs. 130. The inter-sector cross subsidies from commercial and industrial users, in turn, have hiked these users' tariffs above their costs of service and, in turn, led to middle and upper income households enjoying larger economic subsidies per residence than the targeted lower income households. Additionally, as discussed earlier (Section 4.5), this over pricing of electricity to industrial and commercial users is likely to have led to increases in non- technical losses. 131. For some time, there appears to have been inconsistencies in Colombian household energy pricing policies from the standpoint of income distribution, which somehow did not seem to find a place in the Bank's and the Government's dialogue on energy pricing. Thus, kerosene, an important cooking fuel for low income households at lower altitudes, did not have subsidies comparable to those provided for natural gas, LPG, and electricity, although these latter fuels are, almost by definition, less accessible to the poor due to network limitations (gas and electricity) or high front-end costs (LPG). In this context, the inter- linked issues raised in the pricing of household fuels need to be addressed in the setting of residential power tariffs. Indeed, greater focus is called for on the retail price/economic cost signals associated with each of the modern fuels 'competing' for the urban household cooking market. 132. Significant conceptual progress has been made in tariff policy through the promulgation of Resolution 086. However, in OZD's judgment, the discounts allowed on incremental costs of supply for residential consump-tion--as high as 400 kWh/month--remain large. Indeed, it is only if tariff levels to all consumer classes are set close to the maximum permitted in the Resolution that the overall unit sector revenue would approach the level of the average incremental cost. In addition, the Resolution does not provide guidance for setting tariffs in the case of companies whose total financial costs exceed that of the system's incremental costs--this is currently the situation with REEB and needs to be addressed. Finally, the marginal tariff charged for the second consumption block for residences, 200-400 kWh/month, is important in terms of signals sent regarding economic or uneconomic substitution of modern fuels used in urban - 121 - households to meet cooking needs. If the tariff for this block ia set close to the minimum level of Resolution 086 (i.e. 502 of IC) it tends to stimulate use of electricity for cooking that is uneconomic. 5.9 Bank Performance 133. Evaluating Bank performance in the sector during the review period in the field of tariff policy requires answers to the following sets of questions: (i) How did the Bank see tariff policy in its approach to the sector? (ii) Within the context of tariff policy and project justification, what was the quality of the Bank's assessment of the sector's/borrower's problems, the appropriateness of the solutions advocated, and the role of tariffs In these solutions? (iii) How successful was the Bank in achieving the tariff objectives it had set? (iv) What were the signals (implicit and explicit) that the Bank coMuUnicated to borrowers and the guarantor and how did these signals influence their perception of the Bank? 134. During the entire period, the Bank viewed tariff policy largely from a financial perspective, with the issue of pricing according to economic costs only emerging as a yardstick around the time of the Bogota Distribution II operation in 1985. Indeed, one of the problems from a policy standpoint of setting tariffs from a purely financial viewpoint is that unless the *a- ppropriate* level is defined and agreed, it can mean different things to different people. For example, by not taking a sector view, the Bank did not have to take into account in its definition of the "appropriate' level of tariffs at the sector level, the repercussions on public sector finances of financial deficits in the power sector. It is possible that if both the Colombian authorities and the Bank were more focussed on the sise and negative consequences of such impacts that the objective of a financially 'self-sufficient* power sector (i.e. one that largely only borrows to cover its foreign currency costs) would have become a more meaningful goal for both the Bank and the Colombian authorities. 135. Given the pervasiveness of subsidies to electricity consumers, it is surprising that the Bank appears not to have attempted to illuminate better what the costs of these subsidies were--in particular, who subsidizes whom, in what markets, and by how much? This is yet another casualty resulting from the absence of a sector view by both the Colombian authorities and the Bank. 136. When one turns, in the context of tariff policy, to evaluate the overall quality of the Bank's assessment of the sector's/borrower's problems and the appropriateness of the solutions advocated, it is difficult to give an overall judgment. This is due to significant variations in the quality of the Bank's appraisals of individual operations as well as the considerable differences in the complexity of operations. By any standards of measurement the Bank's - 122 - appraisal of the San Carlos project stands out for Its thoroughness and candor. Coming as it did as the first operation after the informal suspension of disbursements in the mid-1970s it set a certain tone. The simpler operations of Mesitas (IEEB), Bogota Distribution I (8M)e, and Guadalupe IV (EPM) were easily justified in the Bank's appraisals. It is when one turns to the more complex operations that judgments become more critical. This applies particular- ly to the FE and Sector Adjustment loans. In the case of the Guavio operation this was the first occasion on which a consolidation of the sector's finances was attempted. This was an important advance, which showed, at least, that based on the demand forecasted and projected tariff increases, that the overall sector investments could be justified, ex-ante. In addition, the Guavio project itself appeared ex-ante to be very competitive with other hydro projects. 137. In the case of the EN operation, there are two issues involved--first, that of institution building and the establishment of FEN, and second, the measures proposed by the Bank in light of the evidence available at appraisal. The case for launching FEN may have been strong at that time. Raising more external resources for the Mesitas and Guavio projects was also valid given the major changes that had occurred in external capital flows after the beginning of the Latin American debt crisis in 1982. However, what was missing from the appraisal was a certain candor in spelling out the problems in the sector and in the companies' problems, which the numbers set forth in the SAR were illustrating. First, these figures provided clear evidence that there was too much investment relative to demand in the 1983-87 program and, therefore, either investments should have been cut drastically or tariffs virtually doubled. The appraisal instead called merely for minor tariff adjustments, while leaving investments essentially untouched, which resulted in a non-viable program. However, the Bank, to its credit, did not finance any of the proposed new investments in the 1983-84 program, and, indeed, adopted a strong position against financing the Urra Project at that time. Additionally, the Bank did obtain a covenant limiting new investments to only projects that were *economi- cally justified and the financing of which would not affect existing financial obligations of the companies.* Second, mobilizing additional domestic resources for the power sector from the local capital market depended first on building up the creditworthiness of companies in the sector, such as EPM, BEEB, ENCALI, and ISA, and less so on the establishment of a financial intermediary. Indeed, grossly inadequate internal resource mobilization has been a characteristic of most Colombian power companies and the sector as a whole since the 1972 OED review--it remained a key problem in 1983, as it does today. By supporting the FEN operation with conditions which did not appear commensurate with the scale of problems already facing the sector at the time, the Bank projected toward the Colombian authorities the image of a very accomodating lender. 138. The Sector Adjustment loan was associated with an important advance in tariff policy in the form of Resolution 086. Additionally, the degree of candor in the President's Report about the sector's problems shows a marked improvement compared to the EN SAR. However, the main criticism concerns the level of tariffs called for in the operation, which again were too low compared to estimates the Bank should have perceived as required at the time of appraisal. Moreover, because of this the role of tariff levels in the sector's problems is played down in the President's Report and this tends to convey the picture that - 123 - internal resource mobilization problems are less severe than they are (and have always been) and that the rate of adjustments in the sector can proceed at the leisurely pace envisaged. 139. As far as the Bank's success in achieving the tariff objectives it had set, the record has been mixed. There is no doubt that Resolution 086 represented the culmination of a long process to link retail tariffs to long run average incremental costs. The Bank played an important role in nurturing this concept over the years, from the time of the San Carlos I negotiations when the first studies on marginal cost pricing were agreed to and financed. Though there are still problems with detailed elements of Resolution 086, it represents a milestone in tariff policy, similar to the effect the decision in the late 1960s to interconnect the regional power systems had on the overall sector. However, the Bank has not succeeded in inducing Government and sector to set tariffs at levels that provide adequate resources to cover both the sector's and companies' costs of debt service (especially after the 1985186 accelerated devaluations) and *appropriate' contributions to investment. Indeed, the situation with EZEB, in particular, leaves much to be desired. For example, the tariff called for in the Bogota Distribution II operation, on the basis of EEEB's total investment program for the 1985-1990 period, was around US*7.9/kWh (mid- 1984 prices). In mid-1987 prices, this was equivalent to about Col$20.0/kvh, yet, at that time, EEEB's unit revenue only amounted to about Col$13.6/kWh, or about 702 of the required rate. 140. In terms of the Bank's signals to its borrowers and their guarantor on matters of tariff policy and how these signals influenced the Colombian authorities' perception of the Bank, the following points are germanet (i) In the review period, the Bank appears to have shared the Colombian view of the ISA bulk tariff merely as an *accounting tool' and not as a prime lever and economic signal for retail tariffs throughout the country. In this manner, the Bank lost the vision of the bulk tariff as the most important power price in Colombia. This was explicit in the manner in which the appraisals of ISA's San Carlos I and II projects determined the appropriate level of retail tariffs in the Central and Atlantic regions as distinct from what the correct wholesale tariff should be. In other words, ISA was treated as a retailer and not a wholesaler of power. A consequence of this approach to the bulk tariff, which is rooted in the structure of ISA's statue, was that ISA has been in continual financial difficulties since the bulk tariff has always been 302-502 below the average long run incremental cost at the interconnec- tion level. This negative impact on ISA appears to have been consistent with its shareholders' objectives. This raises the fundamental issue of the compatibility between ISA's financial soundness, its shareholders' objectives, and the future role of the Bank in continuing to support such unbalanced arrangements. - 124 - (ii) The continuance of large natural gas subsidies to CORELCA after interconnection (at variance with a covenant under Loan 1583) provides little financial incentive for CORELCA to increase energy imports and, indeed, further undermines the economic signalling role of the ISA bulk tariff. (iii) On tariff structure, the Bank seems to have sent only weak signals, even though it became an issue in the Bogota Distribution II operation (1985). Having increased awareness of the costlprice structure problems in the early 1970s with the OED reviews and having insisted on studies within the context of the San Carlos I operation, the Bank raised the issue in all its main operations in the sector, in particular, Bogota Distribution II. However, solutions remained elusive. (iv) Though the Bank maintained an intense dialogue with the Colombian authorities on the issue of tariffs throughout the period, there is no doubt that the occasion on which substantive action was taken by the government to raise tariffs followed the informal suspension of disbursements on loans to power companies in the mid 1970s. Subsequent- ly, as the technical, financial, and tariff performance indices of power companies and the sector deteriorated, from 1983 onward, OED understands that the issue of suspension of on-going disbursements and delay of new loans to the sector was considered again by the Bank's senior management. However, the opinion then of senior management was that the approach to be pursued to achieve higher and better structured tariffs, as well as sounder financial performance, was through persuasion rather than coercion. The one occasion that the Bank took a forceful position, in the mid 1980s, justifiably in the view of OED, was when it indicated in 1985 that it would not then finance the Urra project (see para. 137). (v) The Bank's willingness to take a tougher position on the sector's tariff and financial problems in the mid 1970s, and its reluctance to do so in the aid 1980s, has to be seen in the context of the overall macroeconomic situation in the country at the respective periods. In the mid 1970s, the macro situation was sound generally, with growth prospects strong, whereas by 1984185, Colombia was in the throws of a major macroeconomic adjustment effort, involving significant peso devaluations, trade liberalization, and restrictions on public sector investments. In such a setting, the macroeconomic concerns prevailed over those of a sectoral character--the fight against inflation took priority over raising tariffs (though the two bore little linkage)t and enforcement of sector covenants was relaxed because of the larger macro concerns. 5.10 Agenda for the Future 141. The lessons learned during the review period highlight the need for the Bank to exhibit resolve and tenacity in its handling of tariff issues. Toward this end, OED considers that, in the context of its future lending operations to the Colombian power sector, the Bank: * 125 - - should make the sector approach to tariffs a focal point of its Involvement to Improve tariff levels, structures, and overall resource mobilizationg - needs to call for substantial action toward economic pricing of electricity each time bfre it submits new power operations to the Board; - needs to highlight the Importance of the bulk tariff in power pricing policy in Colombia and should call for action toward pricing such power at Its economic level, even though this can be expected to have repercussions on the relationship between ISA and its shareholderst and - should provide more assistance to the tariff regulatory agency. JVT, in its further development and implementation of tariff policy structured to reflect economic supply costs. ENDNOTES (1) OComments on Self-Financing for the Colombian Electric Sector, * by Luis A. Yepes G., D.N.P. Document DNP-UINF-DE-022. December 1980. (2) *Review of World Bank Leading for Electric Power,$ by M. Munasinghe, J. Gidling, and M. Mason, IE Department, Paper No. 2, 1988. (3) Guavio SAR (1981), p. 38, para. 5.18. (4) See endnote 3. (5) FEN SAR, p. 38, para. 4.14. (6) President's Report, Power Sector Adjustment Loan, Report No. P-4676, p. 19, para. 70. (7) Estudio Sobre la Unificacion de las Tarifas de Energia Zlectrica,l ISA, July 1983, ISA-OPUN 121S. (8) ISA Document ISA-OPUN 20/11/87-168E, *Zstructur- de Costos del Sistema Interconectedo Colombiano.8 (9) *Basis for Formulation of a Colombian Energy Policy, MUDPISSMAP Energy Sector Strategy Study, December 1986. This report estimated gas costs using an 82 discount rate. - 126 - VI. FINANCES 6.1 Introduction 1. During the period under review, the sector's financial performance was never fully satisfactory; indeed, from the mid-1980s it became definitely poor. The present chapter analyses the development of this performance over the period 1970-87, at the level of the sector as a whole and at that of the main utilities, that have received all but one of the loans the Bank made in the sector during the review period, i.e. SEEB, EPM, and ISA. 2. The chapter first sets forth the broad picture on the basis of approximate statements for the sector's sources and applications of funds assembled from various partial consolidations of sector entities' financial statements, actual and projected. Then, it discusses actual, expected, and covenanted financial results, in particular, under five aspectst (i) self- financing, (ii) return on assets, (iii) debt service, (iv) debt and equity, and (v) accounts receivable. Within that context, assessments are made also regarding Bank performance in terms of the financial targets agreed to and the monitoring of financial performance at the utility level. The subsequent three sections deal with specific financial issues that proved central to the development of the sector and of the various utilities during the review period. OED found that the discussion of the following three issues provided the best insight: (i) the changing maturities, grace periods, and repayment terms of Bank loans, (ii) the impact of the exchange risk on the utilities' finances, and (iii) the appropriateness of some financial indicators used to covenant and monitor performance. The Chapter concludes by outlining possible directions for future action based on the lessons learned. 6.2 The Broad Picture Sector Approach 3. One of the more significant deficiencies in the Bank's lending to the Colombian power sector during the review period was the absence of a consis- tent sector approach, which, inter alia, would have provided the financial framework for the operations with individual utilities. Of course, this absence reflected a similar shortcoming on the Colombian side. Indeed, the fragmentation of the sector, itself an outgrowth of fierce Colombian region-alism, prevented Government from properly assuming its policy setting and regulatory role. This, in turn, is likely also to have constrained the Bank from acting in a more decisive manner on the key issues of internal resource mobilization and investment program size at the overall sector level. Even so, especially as the power sector became an increasingly heavy burden on Government's finances, the central authorities should have had at the very least a clear view of the financial performance and prospects of the sector as a whole. However, such information was and continues to be sorely lacking, an issue stressed repeatedly to OED by the central government financial and planning authorities. In the - 127 - context of the present review, OED assembled a limited amount of sector data providing a broad picture of the sector finances and of the impact of various sector entities on them. Consolidated View 4. An in-depth analysis of the power sector finances calls for state-ments consolidating the finances of all sector utilities. Such sector accounts, in reasonable detail, do not exist because: (i) the individual utilities appear reluctant to help assemble such data, as they often see consolidated accounts as a vehicle that ultimately may lead to a limitation of their independence; and (ii) the discrepancies that persist between the various accounting systems and practices used; e.g. EEEB capitalizes part of its operating costs (presumably that it associates in one way or another with ongoing construction or installa- tion) whereas PHM does not. It seems that, for the past two decades, the strongest impulses to correct this state of affairs came from the Bank. Thus, in 1978, in connection with the San Carlos I operation, the Bank agreed with ISA and its shareholders that, by June 30, 1981, the utilities would carry out a study on the possible creation of a common system of accounts; OED found no evidence of substantial progress in this direction. The SAR for the San Carlos operation, for the first time, set forth the statements of the main utilities except ICEL, but did not consolidate them. OED did this summarily, which yielded approximate actual sector figures (excluding ICEL) for 1976 and projections for subsequent years from the perspective of 1977. The SAR for the Guavio project included a consolidation prepared by ISA, with Bank assistance, and qualified as "preliminary*.() From this document OED extracted actual figures for 1980 as well as the projections at that time. Similarly, the SAR for the FEN operation, which also included an income statement and a sources and applications of funds statement but no balance sheet for the sector as a whole, set forth actual results for 1983 and the associated forecasts.(2) Finally, the President's Report for the Sector Adjustment Loan, which consolidates the companies' finances (except for EMCALI), generated actual figures (in US$) for 1984 and 1985, and estimated amounts for 1986. 5. In the context of the present study, OED consolidated the audited accounts of EEEB, EPH, ISA, and ICEL in a funds flow statement that seemed reasonably representative for the sector as a whole (see Annex 6A.1). Resource constraints prevented OED from integrating the accounts of CVC, EMCALI, CORELCA, and Betania. 6. Table 6.1 compares the actual sources and applications of funds statement extracted from the Bank documents as set forth above (see para. 4) with that resulting from OED's partial consolidation of audited figures. In some points, the two sets of data substantially differ, mostly because OED's consolidation does not include the results of CORELCA and its subsidiaries, which, for most of the review period, performed below the average of the main regional and municipal companies. In particular, during the late 1970s and early 1980s, the CORELCA group had to carry a heavy burden of debt service, while incurring substantial further debt, largely because of poor internal cash generation in the face of a sizable investment program. Nevertheless, the two sets of figures in the table convey broadly the same picture. - 128 - Table 6.1s Consolidated Actual Cash Flowe* (Percent) 1976 1980 1983 1984 1985 1986 Dank6 OlB ank OZD Bank-ORD DBank OZD DBank OZD Bank ORD Sources of Fundes - Cash Flows (after Interest) 23 22 22 25 19 20 25 19 23 11 27 14 - Borrowings 61 47 53 33 61 51 68 61 70 71 66 68 - Other Sources 16 31 25 41 20 29 7 20 7 18 7 18 Total Sources 100 100 100 100 100 100 100 100 100 100 100 100 Application of Funds* - Debt Service (Principal) 21 11 23 9 13 16 20 18 26 12 30 16 - Investments 64 80 78 99 86 82 83 87 82 87 66 87 - Working Capital Adjustments 15 11 -1 -1 -1 2 -3 -5 -8 1 4 -3 Total Applications 100 100 100 100 100 100 100 100 100 100 100 100 Self-Financing (bef.W.C.Adj.) 4 14 -1 16 6 4 7 0 -3 -1 -4 -2 Cash Flows/Debt Ser. 113 202 95 278 134 120 128 100 90 94 91 91 Borrowings/Investm. 94 59 67 34 72 63 81 70 85 81 99 78 *Banks extracted fro various documents. 0D: from OED's consolidation of the audited accounts of BEB, EPM, ISA, and ICEL. Conclusions 7. At the end of the review period it is still difficult to generate more than a rather general picture of the sector's financial performance. Indeed, there are no reasonably detailed consolidated statements for the sector, as a whole, covering the period. The partial consolidations of actual financial data available for a series of years during the period under review are those prepared at the Bank's instigation, as well as those the Bank itself prepared (including ODD). The results, based on the Bank's consolidation, suggest that during the period 1976-86: - the sector's revenues did not fully cover its debt service costs as early as 1980 and then again after the mid-1980s. - 129 - - the sector's contribution to investment from its internally generated resources has been marginal thoughout the review period, becoming negative after 1984. 6.3 Self-Financing General 8. Figures 6.1 to 6.5 show the actual and projected self-financing ratios for the sector, MEB, IM, and ISA defined, in accordance with the Bank's Operational Manual Statements (0MS), as the net internal cash generation before working capital adjustments (except for Figure 6.2, which sets forth self- financing after such adjustments) divided by total annual investment in fixed assets. As a background, Annex 6A.2, Section 1.3 discusses some of the definitional obstacles to the reasonably uniform presentation attempted here. This presentation implies that some of the data used do not conform to those appearing in SARs, as OED adjusted the latter to reflect the definition of self- financing set forth above. 9. Figures 6.1 and 6.2, illustrating the developments for the sector as a whole, use the actual data extracted from the consolidation of EEEB, EPM, ISA, and ICEL, which OED prepared in the context of the present study (curve Al) and from the consolidations (also partial), which OED found in several Bank appraisal reports (curve A2). As discussed in para. 4 above, the data used for curve A2 are those for the years 1976, 1980, 1983. 1984, 1985, and estimates for 1986. The values of the ratio for the years in between are interpolated and, therefore, even less reliable than those for the base years. Nevertheless, the order of magnitude is correct. 10. The two curves for the actual sector self-financing ratios (Al and A2) differ substantially in the early years of the period, as, at that time, both, EEEB and EPM contributed a reasonably high proportion of the investment from their revenues. Indeed, this share was never below 252 during these years. However, CORELCA, whose results curve Al does not reflect, did not fare as well, thus contributing to the lower sector ratios in curve A2. In the latter part -of the period, both EB and BPM's performance sagged leading to similar results for the two consolidations considered here, the corresponding self-financing ratios hovering around zero, even reaching negative values for 1985 and 1986, a clear expression of the utterly unsatisfactory financial performance of the sector during the 1980s. . 130 - FIGUR 6.1 8ECTORa SELF-FINANCING RTI0 &EMOU VORING CAPIUL ADJUS'INTS O. - 0.4 - l 0.3 0.2 1976 1977 1978 1979 1980 191 192 1983 1984 1986 196 Y~ 0 Al + P.SCI 9 P.u a PEN A2 Ur~ 6.2a SETRt SEL-FIn=NING RATO A~TE li0REN CAPITAL ADJSTM~NT a.3 - 1978 1977 197m 1979 19 g 9m i 1982g 19m3 19"4 186 198m Yea t] Al + P.wC 0 P.AM db PyfEN x A2 * 131 * 11. The three projections for the sector self-financing ratio extracted respectively from the SARs for the Bank's San Carlos I (curve P.SCI), the Guavio (curve P.Gua), and the PEN (curve P.FEN) operations unmistakably show what the present study (see Chapter 2) identified as the Bank's pervasive optimism and as one of the issues that most likely transcend the Colombian power sector, as it probably can be detected in the Bank's dealings with other sectors, in other countries, too. 12. Unlike the individual utilities' return on revalued assets in operation (revalued RO) for which a threshold of acceptable performance was stipulated in connection with every Bank loan in the review period, the self-financing ratio was not covenanted in all Bank operations. In particular, loan agreements with EPM and ISA did not define a lower limit to this key indicator of what cash flows should achieve beyond covering operating costs and debt service. As discussed in Annex 6A.2 (Section 1.2), in rapidly expanding and moderately sized utility companies where assets in construction can be a very high fraction of total assets employed, the revalued ROI, based on assets in service, is not a fully satisfactory measure of earnings performance. Therefore, as noted in OED's Mesitas audit report (para. 2.7.1), a self-financing ratio covenant becomes an important supplemental tool through which financial performance should be continuously monitored. EEEB 13. The actual figures for EEEB (curve Act. in Figure 6.3) show reasonably good performance in the 1970s and a sharp deterioration in the 1980s. EEEB has a history of increasing cash operating costs per kWh sold and expressed in constant terms. Indeed, this cost has more than doubled during the period 1971- 1986. This is related to an increase of the cost component attributable to the energy purchased and to the lack of efficiency improve-ment, as reflected-- especially in the 1980s--by the increase in energy 'losses.` During the 1970s, investments remained at a modest level, and , from 1976 on, as tariffs increased substantially in real terms, EEB's contribu-tion to investment generally remained in excess of 30% and peaked at more than 50% in the period 1976-1978. Later, after the large investments in the Mesitas hydroplant had begun, it deteriorated to 162 of investment in 1981. From 1982 on, the even larger expenditures in the Guavio facilities compounded the problem, although tariffs were still rising in real terms until 1984. At this stage, government policy, which did not allow the utilities to recoup the effects of the massive devaluation of the Col$, pushed the self-financing ratio down to an utterly unsatisfactory 6?, from where it does not seem to have substantially risen since. 14. Figure 6.3 also sets forth four projections extracted from the SARs for Mesitas (curve P.Mes), Guavio (curve P.Gua), PEN (curve P.FEN), and Distribution II (curve P.DII). These curves show that, in the short term, i.e. the three first years of implementation, the first two projections were quite reasonable. The latter two, which are also the most recent ones, are much further out of line with reality, mostly because, at the appraisal of the corresponding projects, the Bank could not anticipate the decline of tariffs in real terms that was to occur, especially in connection with the 1985 devaluation of the Col$. 一132。 F萬中開口6., 寫口煙，劊口j斗馴口國轎p編訕返XO馴響切這關勵卹口頃O劇牌1劉匹劇闖叮州開圃輪， O‘•，I■■1 1 11■■1 11·l屆 0.,．卜網．．叫卜．口•中••開•卜 自•．•卜•••‘卜••口叫卜口•閑叫 奉．叫－．•叫卜■■■畸•目••中••闕叫卜•■•中■甲‘“卜卜．哈•．■•卜■妒‘呵戶• 0.‘哺勵馴••叫卜．•’叫卜口，•卜••叫吋‘開－吋••．馴中•口 《鳥j•卜，••叫州•馴甲中•吧三j妒 奉自．卜■•『：細•網，中，．•叫卜■•開中．• O•，•卜•開．悶中瞋悶，．中•••• 01 1 1 11華1 1 1 1 11萬11審 唱個曆•令煙，忽勺自曆森，個斤論t煙曆唱憤個個7唱唱個勵唱仰，•唱勻•O個自•唱t•唱自，卸自韋唱個口略個個自•t徊•編唱惜顱才 屆為畸‘‘戲劍細0 .P月細陶。戶j祖吋‘魚O自 奮劉曉口口二‘ 馴瀾•馴吃J‘了U“面以〕嗚馴牌10劉口陶臘膩奮矚匡n為劇j劉U么以馴網熔，個瞭鐺 騙•1 1 1 1 1 1 1 1 1, 1 1 1 1 11 么•斗·一一于一j一領一一口一j－一一一卜一編Jj一入－一各－一j一－一一1一一L一月―一； 么7叫n鬥■•中－．•神卜頗••畸•••．卜■••畸 么申叫卜畸鬥•中••■中■開•畸•目 亂屆•卜•飼馳叫卜．目■叫卜••口中 么4哺口中••哨卜．口•哺••••中•■．哺勵曰••．卜•向•叫卜．戶•■卜 么j•中網曰勿磚口馴j州h煙豐豐由咸j．磚一•叫卜••碼南 么雲叫卜•．•卜•■■中■•爾司h• 鳥唱‘計－一－j一一一－f-－一州卜 01 11爾1 1 1 1 11•1 1 .1; ，煙，•唱個，細唱個曆磚哺口，唱唱自，0唱•，奮唱•，O哺•個•唱••自唱個自唱唱••當唱爾口•t口口鳥唱細口•唱個口自唱個口常 目向畸．‘粈月口一戶滷啊‘戶搏攔織戶尸驪鬨 奮劉曉細匯‘.5 X名么tSI叫卜劉〕觔鴉n紹訕盯功馴牌中壇啊離別（〕喝劇必 rU二劇亂網熔鬍自開憎 蒙暴 ‘向曉．秦戶癩中．戶〞視網 - 133 - 15. in the Bank's Mesitas operation (1978) the company's self-financing ratio was covenanted at a level of 732 over the period 1981-84, whereas construction wag expected then to be completed in 1982. What appears unusual was that the covenant addressed the period after expected project completion and hence when investment levels were likely to be low, at least as foreseen at the time of project appraisal, i.e. 1977/78. Indeed, it would appear that, to underpin the financing plan for the construction of the Mesitas plant. covenanting self-financing during project implementation would have been important, the more so as the plan anticipated a 45Z overall contribution of revenues to Investment. 16. The SAR for the Bogota Distribution I project (1980) vent one step further as it dismissed the self-financing covenant as redundants I ... with the proposed rate of return covenant the contribution to investment covenant is no longer necessary and, therefore. will be eliminated... I (SAR, para. 5.13). Fortunately, this unsatisfactory approach was abandoned at the time (1981) the Bank prepared the Guavio loan, when a self-financing covenant during project implementation was Introduced. This may have been precipitated by the decline of the ratio to less than one-half its 1978 value of 55Z. The covenanted levels of self -financing called for in the Guavio operation were 352 in 1982-84 and 55Z from 1985 on. Since, by 1980, it had become apparent that the Mesitas project would experience delays in completion and significant cost overruns and that EEEB would have two major projects under construction simultaneously, it could have been evident at the time that achieving these targets would be difficult. Indeed, the 552 target after 1985 was lowered to 351 at the time of the Bogota Distribution 11 loan In 1985. The actual ex-post levels of EEEB's self- financing ratio are shown below in Table 6.2, along with the levels covenanted in the Mesitas, Guavio, and Distribution II opera-tions. As can be seen, a considerable margin existed between the targeted and actual levels of this indicator. Part of the reason for the very low actuals is definitional in that in this review the definition of self-financing used is that given in Annex 6A.2 (Section 1.3), which looks at net cash flow with-out adding back any increase in employees' pension fund liab lity reserves. If these liability reserve increases were included then the ex-post self-financing ratios for 1982 and 1983 would increase to between 25Z and 30Z compared to 92 given in the table. This highlights very clearly how vital the resource of pension fund liability reserves increase had become to ex-post assessments of internal resource generation levels, whereas ex-ante it has always been viewed in SARs as unimportant and never representing more than 41-62 of gross internal resot.rce generation. * 134 - Table 6.2s REEB's Self-Financing Covenants Targets Versus Actuals Source/Year 1978 1979 1980 1981 1982 1983 1984 1985 1986 Covenanteds Mesitas - - - 73 73 73 73 - - Guavio - - - - 35 35 35 55 55 Bogota Dist. It - - - - - - - - 35 Actuals: 54 38 21 16 9 9 5 6 6 EPM 17. During the early 1970s, EPM's unit cash operating costs were at about the same level as EEEB's. But in contrast to what happened in EEEB, in EPM, they remained at about that level in real terms. This was possible because, beyond operating with reasonable efficiency throughout the period, EM was able to supply its clients with relatively low cost energy from its own plants. It was even in a position to export surplus energy to the interconnected system. It also kept its energy losses at a constant, albeit fairly high level (182 to 20Z of energy sent out, see Section 4.5). At the beginning of the 1970s, EPM did not carry out major construction work, therefore, its self-financing ratio (Figure 6.4) was high (702 in 1971). The construction of Guatape II, in the mid- 1970s associated with declining tariffs in real terms, led to a deterioration (i.e. levels between 252 and 282 in the period 1973-75). After the tariff improvements that took place in the aftermath of the 'Acuerdo de Sochagota, inter alia, in connection with the preparation of the San Carlos I operation, self-financing peaked at levels in excess of 70Z in 1980 and 1981, when EPM, with Guatape II behind it, was not yet excessively burdened by the new investments in the Guadalupe IV and Playas projects. These projects, together with the start of Rio Grande II in 1984 and the effects of the devaluation mentioned earlier, reduced EPM's contribution to investment to a poor 92 in 1986. Whereas in the case of EEEB (see para.13 above) the exchange losses incurred are specified in the audited income statements and, therefore, reduce the cash flows used to determine self-financing, in the case of EPM, they are not explicitly identified and their impact on self-financing not isolated. Therefore, some of the ex- post ratios set forth in Figure 6.4 for EPM may be somewhat overstated. 18. The four projections extracted from Bank SARs and set forth in Figure 6.4 show that the forecasts were reasonably accurate in the case of the Guatape II and San Carlos I appraisals (curves, P.GII and P.SCI), not too far off in the short run for Guadalupe IV (curve P.GIV), but again, because of the latest developments, in particular the uncompensated devaluation in 1985, substantially off the mark in the most recent operations (curve P.FEN). 19. In contrast to the case of BEEB, in that of RPM no continuously monitored self-financing ratio covenant has been attached to any Bank loan during the review period. This could be justified at the time the Guatape II project (1972) * 135 - was approved since this hydro development was the only major investment by EM in its own plant in the 1970s. However, in the cases of the Guadalupe IV and Playas projects, which RPM constructed simultaneously during the 1980s, the absence of continuously monitored self-financing covenants was, in OND's judgment, more questionable. Indeed, the company's financing plan for its 1980- 86 investment (which was dominated by the Guadalupe IV and Playas projects) envisaged that net internal resource generation would contribute as much as 602 of the finances required (Guadalupe IV, SAR, page 37). Instead, what was imposed, was a condition of loan effectiveness that called for balancing the financing plan, for the 1980-82 period for Guadalupe IV and for 1981-82 for Playas, which is a weak substitute for a covenant. As shown in Figure 6.4, EPM's self-financing ratio declined steadily during the 1980s. However, the Rio Grande loan agreement, approved in 1984, again did not include a self-financing covenant, although the financing plan envisaged that net internal resource generation would provide 63% of the financial resources for EPM's power investments in the period 1984-91. ISA 20. ISA's original by-laws foresaw that the company's shareholders would share ISA's operating costs, its debt service, and the local currency com-ponent of its investment. They also called for a 9Z return on generating assets, a clause never applied in its original form, as, in 1977, when ISA commissioned its first plant, Chivor, a new arrangement was under negotiation in connection with the preparation of the Bank's and IDB's San Carlos opera-tions.(3) The resulting modifications of ISA's statute, supported by the Loan Agreements for the above operations, foresaw that, from 1978 on, ISA would operate as a covmercial entity, earning a reasonable return on its assets in operation. As curve P.SCI in Figure 6.5 shows, this return was expected to allow the utility's revenues to make a 102 to 20% contribution to investment. This is a relatively low contribution, since it assumes that dividends on ISA shares are re-invested. It is explained, however, by ISA's high projected debt service obligations during the construction of Chivor II and San Carlos. 21. Until 1982, ISA, by and large, performed as expected. From 1983 on, however, its contribution to investment became slightly negative, as its bulk tariff, which, in economic terms was low to start with (see Section 5.5), did not allow it to meet its debt service. This shortfall occurred although ISA failed only by a modest margin to achieve the covenanted 9? return on assets in operation (see Section 6.4). In 1984, in connection with the FEN opera-tion, the Bank anticipated a substantial improvement, which did not materialize, as ISA's bulk tariff remained low, when Government did not allow ISA's shareholders to adjust their rates in the wake of the 1985 devaluation of the Col$. 22. As in the case of EPM, the Bank did not attach a self-financing covenant to its two loans to ISA (San Carlos I and II). Given the fact that ISA's cash generation was expected to contribute only marginally to the company's investment, this seems to have been justified at the time. However, this might have to change, if one of the objectives of future action in the sector is to - 136 - have the bulk supply entity actually operate in accordance with economic and commercial principles on the basis of a bulk tariff closer to the economic level than at present. Conclusions 23. During the 1970s, REEB and EPM generated reasonable contributions to their investments, except for a relative low in BPM's self-financing in the middle of the decade, when the peak of the Guatape II investment concurred with the deterioration of tariffs in real terms. The poor self-financing of the sector as a whole during this period essentially is due to the other entities, especially ISA (where it is built into the statutes) and CORELCA. In the 1980s, the downturn in the economy making it difficult for Government to allow the utilities to adjust properly their rates, in particular as the needed increases became larger due to delays and cost increases in the large hydroplants under construction, self-financing deteriorated in Z9B, EPM, and ISA to very low or even negative levels. This also was the case for the sector as a whole, which means that the utilities became increasingly dependant on direct or indirect Government contributions and subsidies at a time when Government itself had to adjust its economic policies, implying that it could no longer assign to the power sector the high priority electricity supply had enjoyed in earlier years. 24. The Bank covenanted self-financing only in connection with operations in EEES, using the covenant essentially as a complement to the rate of return covenant. This led to inconsistencies in the treatment of various entities, which, doubtlessly were justified, in part, by differences in performance. In retrospect, it seems that there would have been good reasons for covenanting self-financing in several further cases, as some investments were very large for the companies undertaking them. This implied large increases of debt and hence debt service, which tended to make, in the medium term, revenues that provide a reasonable return on investment in service insufficient to contribute significantly to investment. The delays in construction and the cost increases, of course, compounded this effect. OED suggests that, the Bank re-examine the issue of covenanting self-financing in its operations in the Colombian power sector as a complement to the rate of return covenant. At the same time it should take a sectoral approach, which does not mean that conditions should be uniform for the sector, but that the effects of conditionality in one utility on the rest of the sector be taken into account. 6.4 Returns on Investment Actual and Projected ROI's 25. Annex 6A.2, Section 1.2 discusses the various measures for return on investment (R01) and the fact that the indicator the Bank generally uses (operating income less operating cost divided by net average revalued assets in operation) calls for a full revaluation which is not reflected in the Colombian utilities' books, as local law only allows the companies a partial revaluation, - 137 - 26. Figures 6.6 to 6.8 show the actual returns according to the above definition (curves Act. In the three figures) as well as those espected to be achieved as set forth (1) for E3B (Figure 6.6) in the SARs for the Mesitas (curve P.Mes), the Guavio (curve P.Gua), the PER (curve P.PEN), and the Bogota Distribution II (curve P.DII) operations; (ii) for EPM (Figure 6.7) in the SARs for the Guatape II (curve P.GIU), the San Carlos I (curve P.SCI), the Guadalupe IV (curve P.GIV), the FEN (curve P. 33) operations, and (iii) for ISA (Figure 6.8) from the SARs for the San Carlos I (curve P.SCI) and the PER (curve P.PEN) operations. The graphs, seen together with the corresponding curves in Figures 6.3 to 6.5, show how misleading the steadily improving return for 333B is. Indeed, a return of 13Z in 1984 (Figure 6.6) permitted financing from revenues of only a meager 52 of investment (Figure 6.3), largely because of the high debt service obligations and the low asset base which does not include work in progress, i.e. a sum exceeding that of assets in operation. Similarly, in ISA an SZ return (Figure 6.8) in 1984-86 does not even allow the company to meet its debt service (Figure 6.5). 一1）總－ ，廈頗開頃‘.& 匐口購t觔森頂口O口匐間總熔”目口煙 為讓1 1 1 1 1 11思’屋I, JO．口1 1 11•1 jil｀面I: ，齋叫卜勿．•口州卜••••開卜．• 斗1 1 1 1 1 1 11購l勵； 唱．才勵唱自響•唱•，奮唱•7口唱，，口唱口自O唱口細t唱•自蠶t申口細唱••曄t煙口唱唱馴口O •A•矓。‘執闢中自．戶磚開目。p戶奮H誠．.0啊 才X中開矚二7 勵澗寫細森口口O開訪閒驕，目細騰 《甲然件馴回呼鳥挪。仰．.t寧細：仰坤併化盒件》． .1 1 1 11奮1 1 1 11•11; 唱自，自唱•，華唱•，必t口曆．唱•嘐婦唱口乍奮，勻奮中，•零•唱口自O唱••唱唱•口自唱個口細，••磚偉寫口騙t自口O •向曉。‘魚中唱．戶禹中。戶膚即誠戶．悶馴 甲馴曉口匯二• XS么。卹乏很復O鑰刃州視臨個口中瞭 《他取睡t勸”隨馱總t馱織。F勵“他細•．t•組O戶劉比顱滬》 唱讒1 1 111」L 1 1 11 t甲廖奮唱中奮中唱個才•惜甲自ot•口唱勺••蠶，口口細t口口禺勺電自勵勺口•自 俘向咸。。戶』闕潤令戶j電鬨 139 - .Covenants for MMB 27. As further discussed in Annex 6A.2 (Section 1.2), one of the problems related to the return. on revalued assets in service in the context of the Colombian power sector has been its over-valuation, especially in the case of SUB, due to an underestimation of the revalued assets. This was possibly the reason for the wide spread in reported values of ex-post revalued ROIG, especially for 1982-83. For example, the SARs for the FEN and Bogota Distribu- tion 11 operations claim that the 14Z revalued ROI covenanted in connection with the Guavio project (see Table 6.3) had been exceeded In 1982 and 1983, with ex- pR.st values given at 16.32 and 14.82, respectively. In reality, the 1987 OED audit report of the Mesitas project, which was based on 123111 19 audited statements sets them forth at 11.12 and 12.8Z respectively. This inaccuracy of the revaluation reduced the effectiveness of the Bank's monitoring of ENEB's performance under the revenue covenant (see Annex 6A.2, Section 1.2). Table 6.3t EEEBs Covenanted Versus Actual Revalued ROIs (Percent) Loan Agreement/ Year 1978 1979 .1980 1981 1982 1983 1984 1985 1986 Covenanted: San Carlos 1 12 12 12 - - - - - - Bogota Distr. I - - - 14 14 14 14 14 - Guavio - - 14 14 14 14 14 Bogota Distr. 11 - - - - - - - 12 12 Actualst 6.5 8.5 8.4 10.4 11.1 12.8 13.2 n.a. 16.0 28. The 12Z return covenanted in connection with the San Carlos I loan in 1978 was superseded in 1981 by a 142 required return specified in the loan A gree- ment for the Bogota Distribution I operation and confirmed in the context of the 1981 Quavio operation. Demand and tariffs increasing only at about half the pace envisaged (see Annexes 4B.8 aud 5C.1) prevented EEEB from earning these returns between 1978-60, while in the 1978-84 period, even though no large assets were added to the rate base, ZZEB failed to achieve the covenanted level, though its return was steadily increasing. As already set forth in Section 6.3, in this same period, ZZEB's self-financing dramatically deteriorated. The shortfall in cash flow associated vith the sub-par performance on the rate of return over the period 1978-84 was of the order of US$90 million (1986 prices) or about half the cost overrun of the Mesitas project, which ultimately was financed, in part, through the FEN loan. - 140 - Covenants for RPM 29. In the case of EPM, the Bank exercised, in the judgment of OED, legitimate flexibility in targeting financial performance, given the overall satisfactory financial performance and prudent management of this company, particularly after the mid-1970s. Here, again, one of the problems encountered in targeting and monitoring the revalued 101 was the considerable variation in reported ex-post values of this indicator. This was also related to difficulties associated with the Bank's pro forea full revaluation of assets in service. For example, the ex-post returv. given in the YNl SAR for 1979 was about one-half of that reported in the Playas and Rio Grande SARs, as well as in the Guatape II audit report. Further, the ex-post values for 1981-82 in the FEN SAR exceeded those given in the Rio Grande (SAR) by one-fifth. 30. EPk's actual performance as measured by the revalued RoI remained below covenanted levels for much of the review period, as shown in Table 6.4 below. However, throughout the period, save for 1985/86, EPM's performance, as measured by its self-financing and debt-service coverage ratios, was reasonably satisfactory, in contrast to ZEEB's. It is in this context that the lowering of the revalued ROI covenant for EPM must be seen at the time of the San Carlos I loan (1978). High values of this indicator were achieved in 1976 and 1977 (12.2% and 13.O%, respectively) because of the delay in comnissioning the Guatape II project; however, with no new investments expected before 1980 (Guadalupe IV), it appeared reasonable to lover the return on asset covenant in the 1978-79 period as the Guatape II project entered the a set base, and to raise it again to 9% in 1980. At the time of the approval c the Playas project (1981), this covenant was lowered for the first year to X and increased to 101 for 1982. This was consistent with what was being achieved in terms of debt service coverage and self-financing in 1980. However, OED has had difficulty understand- Ing the rationale for the lowering of the indicator for 1984 and 1985 in connection with the Rio Grande II loan, since in that period EPM' s financial performance had eroded significantly compared to the early 1980s. Table 6.4s 1PMs Covenanted Versus Actual Revalued R01s (Percent) 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 Covenants Guatape It 9.0 9.0 9.0 9.0 9.0 9.0 - - - - - - - - - San Carlos 1 - - - - - - 7.0 6.0 9.0 - - - - - - Guadalupe IV - - - - - - - - 9.0 10.5 13.0 13.0 10.0 10.0 - Playas - - - - - - - - - 7.0 10.0 14.0 14.0 12.0 - Rio Grande - - - - - - - - - - - 7.0 10.0 12.0 Actuals 8.3 4.4 6.8 8.4 12.2 13.0 7.4 7.4 6.8 6.5 7.6 10.6 a.a. n.a. 13.0 Covenants for ISA 31. The case of ISA is a good illustration of the Bank dealing pragmatically with a difficult situation in nurturing a new institution through the early years of its existence. From the standpoint of financial performance, the major difficulty facing the Bank was ISA's statute, which constrained the company's resource generation to covering its operating costs and debt service, without significant contribution to investment. It was the intent of ISA's statute that local currenty Investment costs would be covered by ISA's shareholders and the foreign component of such costs by borrowing. This implied that ISA's overall financial performance, virtually by definition, would always be problematic. Therefore, the issue for the Bank was to foster the growth of an institution it strongly supported, on the one hand, while dealing with constraints that limited the company's financial performance, on the other. 32. The covenanted revalued 101 was set at 92 in connection with the San Carlos I loan in 1978 and subsequently maintained at that level. It is shown in Table 6.5 below, along with the ex-post values of this indicator. Compli- ance up to 1982 was quite good, with the average value of this indicator being 8.7, declining to 7.9 for the 1983-86 period. This Weakening spilled over into other aspects of financial performance, in particular, an inability of ISA to cover its debt service costs without a rollover of ISA bonds 'held by share- holders. - 142 - Table 6.5t ISAs Covenanted Versus Actual Revalued ROIs Actuals Loan Agreement/Year 1978 1979 1980 1981 1982 1983 1984 1985 1986 Covenanted: San Carlos 1 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 Actualss 8.4 8.0 9.0 8.7 9.3 7.0 8.4 8.1 8.0 33. It was quite clear at the time of the San Carlos I appraisal that setting the revalued ROI covenant at 92 would not provide adequate cash flow for the company to perform financially in a satisfactory manner. The Bank's SAR candidly acknowledged this problem. Seen in the context of the Bank's dilemma, discussed in para. 26 above, proceeding with the known inadequate cash generation target at that time in ISA's history was, in OED's judgment, appropriate. However, as ISA has now passed its *settling-in' period, it appears imperative that the pendulum swing in the opposite direction, toward giving higher priority to the main bulk electricity supplier's financial strength and cash generation, particularly through setting its bulk tariffs closer to economic levels. Conclusions 34. The return on assets in service was the performance indicator covenanted in all Bank operations in the power sector. The Bank tried to compensate for the shortcomings of the covenant by showing flexibility in setting the level of the threshold of acceptable performance as a function of the perceived needs and of the expected commissioning dates of major plants, which would significantly increase the rate base. 35. From 1977 on, UB did not achieve the covenanted return, except in 1986, when a substantially higher return still did not permit the company's revenues to contribute significantly to investment. EPM exceeded its threshold in the late 1970s and in 1986, when, as in the case of EEEB, its self-financing remained inadequate. ISA, in turn, achieved or only slightly fell short of the covenanted return, which, however, was set at a level consistent with the company's by-laws calling for revenues to provide only a token contribution to investment. 36. The experience during the review period suggests that the Bank, as already mentioned in para. 25 above, re-examine the validity of covenanting self-financing in addition to the return on plant in service. It should also review, as suggested in Section 6.10 below, the case for including work in progress in the rate base when investment is as lumpy as the large hydro develop- ments dominating the Colombian power sector's construction program. - 143 - 6.5 Debt Service Coverage General 37. Debt service coverage (DS) is one of the main indicators used by the Bank to judge an entity's capacity for incurring further debt in the power sector. In the context of the present review, it is defined in accordance with the Bank's Operational Manual Statement as net profit after tax but before depreciation and interest (including exchange losses) on debt divided by principal repayments plus interest (including exchange losses) on debt. Annex 6A.2, section 1.5 discusses the differences between the various DS ratios used in connection with Bank operations in Colombia both to monitor performance and to limit the incurrence of additional debt. 38. Figures 6.9, 6.10, and 6.11 set forth, for 111B, EPM, and ISA, respec- tively, actual DS ratios as well as those projected in various Bank SAR9. As in the case of the self-financing ratio, the values used in the graphs, as extracted from Bank SARa. may differ to some extent from those actually appearing in the documents referred to, because, to achieve reasonable uniformity, some had to be adjusted to reflect the above definition of the ratio. - 144 - FI8'QBE б.9 a.s �= D88! BEaVICa t�V�At�E 9l1lIO э.е J.4 �.п 3 п.о д.е 1.4 п.п п 1.А 1.в 1.4 в.п 1 1 t7п / q7a /0. • 11i7s / У7в / У'77 1 G7� / аТ0 / УСр / р�1 1 Обп 1li�i / УОа /YRA / аС� � Iat. + Р.Мвв о РАVд л PJ�EN лг P.OR кпс,агба б.�о в�хи: nввт sввviсв cov� влтzо s в.а а s.s s п.а п 1.О 1 1п"72 tr'7! 1074 tp'f0 tflв 1l77 1У7д 10Я i�6 1�1 19еп /0ei /9ве 14еА 1Yei а ,Wt. + р,вr о р,в0! е pAtiГ х p,�BJ1 FICO�R б.ц •---. iSA: D88r SВбУ2СВ С0�9�АоЕ 8sTi0 �.п п.т п 1.1 1.0 !.1 1.е 1.0 1.4 1.� 1.п 1.1 1 as аа а� ае as аа о.а 1i1f /f'1в 1f77 /1�7� 1!!� 1Од0 /9df 1Я�� 19�i 1У�4 /9�� /01� р �, ♦ Р.l0/ б P.rlN 143 RUB 39. The development of ZZEDs ratio (Curve Act. in Figure 6.9) closely reflects the company's overall financial performance during the period under review. Indeed, the ratio was reasonably satisfactory (between 1.6 and 2.2) in the period 1972-81. Starting In 1980, when major work began at the Mesitas power plant, it fell, in spite of tariff increases In real term to a low 1.1-1.3 in the period 1982-85, largely a consequence of the increased debt service associated with Mesitas. combined with a sub-par operational performance (see e.g. Section 4.5, which discusses the Increases in energy losses). The projections for EM set forth in the SARe for the Mesitas, Quavio, PEN, and Distribution 11 projects (curves P.Mes, P.Gua, P.FZN# and P.DII in Figure 6.9), except that appearing In the Guavio appraisal, show the high expectations that the Bank had with respect to the company's performance, even in the short term. it also illustrates to what extent the company fell short of these expectations. The increased exchange losses in the years after 1984 and the failure to adjust tariffs for the devaluation of the Col$, in particular, compounded the problem. Thus, during the period 1982-86, the DS ratio averaged only 1.24. EPM 40. In SPH (see curve Act. in Figure 6.10), the development of the DS ratio was also in parallel to that of the utility"s overall performance, i.e., satisfactory until the aid-19709, when the debt eervice associated with the Guatape 11 hydro facility began. This. combined with a low internal cash generation related to decreasing tariffs in real terms, depressed the ratio to less than 1.5. The improvement during the late 19709 is a reflection of the improved earnings situation. After 1280, the ratio remained at about or in excess of 2.0, however, In 1986, mainly because of the uncompensated devalua- tion of the Col$ and t ke beginning of debt service for the Guadalupe IV project, it fell to a marginal value of 1.3. The discrepancies between actual and projected figures were, In general, lose important than in the case of MB (Figure 6.10t curves P.GII, P.SCI, P.GIV, P.FW extracted from the SARe for the Guatape 11, San Carlos I, Guadalupe IV, and PEN projects, respec-tively). The Guatape II projection is the exception, as it foresaw DS ratios substantially less favorable than they turned out. OED has no explanation for the exceedingly large differences appearing In the year this loan was made (1972) and in the next. ISA 41. in ISA (see curve Act. in Figure 6.11), the DS ratio peaked during the two fist years after commissioning of the Chivor I hydroplant, i.e. just after ISA. for the first time, bad substantial revenues of its own. There-after, the ratio stayed at about I after 1983. This corresponds to the level required by ISA's original by-laws, i.e. an Internal cash generation permit-ting the company to meet its operating costs and debt service. However, it falls short of the projections set forth in connection with the San Carlos I operation (curve P.SCI),,which called for ISA revenues also contributing, albeit marginally. to its further investments. After 1983, the servicing of the San Carlos debt began. This, combined with weakened earnings from a bulk tariff declining in real terms, - 146 depressed the ratio to about 1, or slightly below 1, where it remained until 1986. If the debt service due on ISA bonds held by shareholders is excluded from the definition of debt service, the level of the DS ratio Improves somewhat. Though fts value post-1983 is still only marginally above 1.0 (see Table 6.6), its average between 1977-86 is about 1.5 compared to an average of 1.2 if debt service on ISA bonds is not excluded. If one further takes into account the increase in accounts receivable, ISA has been unable to service fully its debt since 1983. Table 6.6# ISAs Covenanted Versus Actual Debt-Service Coverage Ratio 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 Covenanted 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 Actuals 0.9 1.3 1.3 1.5 1.8 1.4 1.0 1.0 0.9 0.9 Actuals excl. debt-service on ISA bonds 1.5 2.0 1.4 1.6 2.2 1.4 1.1 1.2 1.1 1.1 Covenants 42. In connection with all but one of the loans the Bank made in the power sector during the period under review, the Borrower and the Bank agreed on a debt service coverage ratio covenant in one of the two forms suggested in the Bank's Operational Manual Statement 2.22. The exception is the PEN loan discussed below. Until 1982 all loan agreements Included the traditional covenant that made incurrence of new debt subject to Bank approval, whenever the net revenues earned during a 12-month period preceding the intended incurrence, fell below a factor (generally between 1.3 and 2.0), multiplied by the maximum debt service over a future 12-month period. In the case of the Guatape II loan the factor was 1.4; in all other operations it was 1.5. In 1982, the Loan Agreement for the CORELCA loan included a covenant of the second type foreseen in the OMS, i.e. one using a test referring only to future annual revenues and debt service payments. It set the factor in the test at 1.1 for 1982 nd 1.3 for 1983 and thereafter. These weakened conditions have to be seen in the context of a first operation with a still rather fragile organization and appeared justifiable. 43. In 1984, the Bank made the FEN loan, and the corresponding agreement made the incurrence of new debt for all the utilities of the sector subject to approval by FEN, when a power company faced a test similar to that usually used in connection with the Bank. The test here, used only projected revenues and debt service payments. As this agreement did not supersede those that the other borrowets had with the Bank, it meant that ZEEB, EPM, ISA, and CORELCA had an obligation toward FEN that was similar, but not identical to that they had toward the Bank with respect to incurrence of new debt. Whereas this must be seen in the context of one of the FEN operations' main objectives, i.e. to support FEN becoming the financial coordinator of the sector, it is evident that it did not help to make sector financial operations more transparent. - 147 - 44. Six months after the PEN operation the Bank made the Rio Grande II loan to EPM. The corresponding loan agreement included the same covenant as the previous operations with SPM. In 1986, in the context of the Bogota Distribution II loan, the Bank agreed with EXEB on a covenant of the second type suggested in the Bank's Operational Manual Statement 2.22. Therefore, in 1986, when EPM presumably failed the tests for both FEN and the Bank, the company could have incurred further debt only after the concurrence of the two institutions. This implies that these would have to reach agreement on the borrowing at issue. Assuming that RPM's performance improved, the FEN test (using projected revenues) being easier to fulfill than the Bank test (using actual revenues), a point would eventually be reached when the utility would need Bank but not PEN approval. Only with further Improvement would EPM not need any consultation with FEN or the BAnk. In the case of EEB, the test is less stringent than for EPM (see also Section 1.5, Annex 6A.2), which seems somewhat in contradiction with the fact that EEEB's performance has, in general, been weaker than EPM's. OED did not investigate the record of compliance, i.e. whether the Borrower, when failing the test, actually asked the Bank for approval of the incurrence of new debt. Conclusions 45. Until 1981, debt service coverage was satisfactory in all the Bank borrowers and with the economic downturn it worsened in all companies. However, here too, the deterioration was least dramatic in EPM. 46. Until 1984, the Bank's handling of the debt service -overage ratio covenant was straigbtforward. The FEN operation in 1984, introducing a similar, but not identical, obligation of the Bank's borrowers toward FEN, could have led to a measure of confusion and the 1986 Bogota Distribution II Loan to ESEB introduced another type of covenant, setting a less stringent test for REEB than for EPM, i.e. under similar performance EPM has to ask for Bank concurrence, whereas 3EB would not. 47. OED suggests that in the context of a sectoral approach to future operations the Bank try to go back to a transparent set of covenants where the differences consistently reflect differences in performance or in objectives, e.g. closer scrutiny over the build up of debt in one entity rather than in another. 6.6 Debt and Equity 48. Low self-financing and high investments imply large borrowings and equity contributions. If, as was generally the case in Colombia, the latter are limited, the debt-equity ratio increases rapidly and. in the medium term, large debt service payments, combined with inadequate resource generation, further constrain both the company's self-financing capacity and its ability to retain earnings, thereby resulting in an increase in the need for further borrowing and/or equity. of course, these factors are compounded much further by the presence of inflation, which averaged 221 annually in Colombia in the review period. Such a high level of inflation rapidly erodes the equity base. * 148 - 49. Table 6.7 illustrates the behavior of the debtlequity (DIR) ratio for EBEB, EPM, and ISA during the review period. The DIB ratio used corresponds to gross long-term debt divided by total not worth. The table sets forth two ratios for each company. The first is that resulting from the utility's audited ac- counts reflecting only the partial revaluations carried out in accordance with Colombian law. The second ratio is extracted from Bank SARa which show a pro forma full revaluat.on in accordance with normal Bank practice. Annex 6A.2, Section 1.4 further discusses the differences between the two ratios. The following discussion focusses on the DIE ratio emerging from the audited statements as these are the official audited figures on which company management and the general public will essentially base their judgments. Table 6.7: Actual Debt/Equity (DIZ) Ratios, REEB. RPM, AND ISA 1975 1978 1981 1983 1986 ERB DIg Audited Statement 1.58 1.06 1.60 2.19 4.30 D/E Bank SARe 0.79 0.52 0.70 0.69 1.38 EPM DIR Audited Statement 1.60 1.76 1.05 1.00 2.20 D/E Bank SARs 0.68 0.68 0.34 0.45 n.a. ISA DIR Audited Statement 4.01 3.83 2.00 2.70 5.93 DIR Bank SARe n.a. 0.70 0.47 n.a. n.a. 50. In the 1970s, as a consequence of moderate investment and despite an often unsatisfactory average tariff level, REEB's debt-equity ratio improved fron 1.7:1 in 1971 to 0.9:1 in 1979. With the first implementation phase of the Mesitas project, it increased to 2:1 in 1982, when the company began constructing the Guavio plant. From then on, the ratio deteriorated to over 4:1 in 1986, reflecting low self-financing and limited availability of public equity funding in the face of large financing required. EPM's ratio was far less volatiles It varied between 1.5:1 in the mid 1970s to 1:1 in the early 1980s. Only in 1985 and 1986, as a consequence of the large borrowings associated with the Playas and Rio Grande II projects and the exceedingly low self-financing following the devaluation of the Col$. did it deteriorate to 2.2:1; In this case, too, public funds for equity contributions were scarce. In ISA, the ratio was never Lelow 2:1. This was due, in the 1970s, to the large IDB and IBRD loans, 'which finnced the investments in Chivor and the interconnec- tion, on the one hand, and to the shortfall in equity contributions from ISA*s shareholders (one of the reasons for the informal suspension of disbursements in the mid-1970s), on the other. After an improvement in the early 1980s, ISA's - 149 - large commitment in the Guavio and the Betania develop-ments, forced the company into further borrowing, again largely from IDS and IBRD. This resulted in an increase of the debt-equity ratio to nearly 6:1 in 1986. Therefore, at the end of the review period, all three of the Bank's major borrowers in the sector were in a situation that all but impeded borrowing from commercial sources. In the case of ICEL, the debt-equity ratio was most of the time lower than in the other companies not because of better performance but due to the fact that the company was not in a position to incur much debt due to its sub-par performance. Indeed, it relied heavily on Government equity contributions to cover its operating deficits as well as its investments. 51. An undesirable consequence of the high leverage that developed in the 1980s in the sector's main Bank and IDB borrowers, is the level of Bank and IDB exposure in terms of the utilities' equity, as shown in their audited accounts. As Table 6.8 sets forth, in 1986, outstanding IDB and IBRD loans by far exceeded EE's and ISA's equity. Even in the case of EPM, the exposure is significant, though not shown in this table. This is clearly a reflection of the fact that. in the 1980s, the two leading agencies have become, for any practical purpose, the only major external sources of finance. Table 6.8s Bank and IDB Exposure (1986) in 2 of Utility's Equity EEEB ISA IDB/IBRD 271 318 IBRD 159 173 Conclusions 52. Although the debt-equity ratios of the main Bank borrowers have deteriorated during the 1980s, largely because of low-self-financing, scarcity of public funds for equity contributions, and high inflation they do not represent a major issue. To OED it seems far more questionable that much of the equity is not remunerated in the form of paid-out dividends, thus, Implying a further not negligible subsidy to the sector. 6.7 Accounts Receivable General 53. In the context of the present review accounts receivable are expressed as the ratio (expressed in months) of outstanding billings for energy sales to annual revenues from energy sales (see also Annex 6A.2, Section 1.6). 54. In a presentation similar to that adopted in Sections 6.3 to 6.5, Figures 6.12 to 6.14 show actual and covenanted levels of accounts receivable for EREB; actual and projected values for ONM; and actual values and levels agreed with the shareholders for ISA. Again, the figures are essentially self-explanatory. * 150 - 8S8B 55. BEEe's accounts receivable (see figure 6.12). which were at some two months' billings in 1972-75, jumped to about three months' dues in 1976 and remained there for the next two years. From 1979 on, performance drastically deteriorated, although in connection with Nesitas, a gradual decrease to some two months of billings was covenanted (see Table 6.9). A second attempt to covenant a drastic reduction was attempted, again with little success, in 1981 in connection with the Guavio operation. OED has difficulties to understand the rationale for covenanting. in the year of loan approval, a reduction of the receivables from nearly half a year's sales to less than three months' billings. Agreeing on such a covenant meant that ES88 would be in default practically from the day the loan was signed. This same observation applies to the 7loan Agreement for the Bogota Distribution II operation, which set an upper limit of 62 days' billings at a time actual accounts receivable were 2.5 times as high. EPM 56. From 197. on, the values of EPM1s accounts receivable indicator were consistently smaller than those for EEB's (Figure 6.13). Though they also substantially increased between 1978 and 1982, they never exceeded 3.3 months, i.e. about half of the maximum they reached in EEEB. After this, EPM's performance dramatically improved, leaving the index at less than 1.6 months in 1986, i.e. about one-third the value achieved by EEEB. In connection with the operations with EPM, the Bank never covenanted the maximum levels of accounts receivable that would be acceptable. This seems well justified in view of the consistently satisfactory performance of EPM, even under the difficult circumstances prevailing in the mid-1980s. � � � � � • � � � . . � ° � g . � =� • а � � л о � . �„ � � �р � �ое � е ,� � � . в� 3 = , � д � � р ~ д в�'I � в � � � � $ � Л � б � А a � р �, а 0 • 0 � Q � �� е � � �� '� � � � � � � � о � � � . и„� � и� r ь3 � д А � д � • 14 R �ii S � S м . t � + � � S о в� . & " � ° •. S • � 6 ~ r! О А �П О в А М1 О д О� � � i д i .f ci : Аа.Ааq �оавАа�иво�оа�w � А и• о в е м е е г а а.�• �ыi�i eid�fciddeicid :�...:.� . � � • 152 Table 6.91 ZEEBt Actuals Versus Covenanted Accounts Receivable (Days of Billings) Source/ Year 1973 1975 1978 1979 1980 1981 1992 1983 1984 1985 1986 Covenanted: Mesitas - - 73 70 66 62 - - - - - Guavio 77 77 62 62 62 62 Distribution 11 - - - - - 62 Actuals 56 65 92 121 155 188 1" 148 147 152 155 ISA 57. As set forth in Figure 6.14. ISA's accounts receivable Indicator was at an unsatisfactory 9 months in 1975, reflecting the defaulting of the company' s shareholders, which contributed to the Bank's suspension of disbursements from the then current loans. The improvement to 5 months by 1978 represented, In part, the results of the Intensive dialogue that took place between Government, the sector, IDB, and the Bank during the preparation of the San Carlos operations. in this contest. the parties reached an understanding an a reduction of the accounts due by the shareholders to two months of billings (curve P.SCI In Figure 6.14), which was not achieved as the indicator remained between some 5 and 7 months from 1978 to 1982. A similar and retanding was reached In connection with the FEN operation. This was eve less heeded than the previous one, as the environment in which the utilities had to operate seriously deteriorated, in particular. when Government did not allow the companies to recoup the effects of the massive devaluation of the Col$ In the mid-1980s. Indeed, by 1986, ISA*9 shareholders owed their company the remimeration for a full yearl's supplies and services, thus using it as a source of short-term finance. The scale of the problem is illustrated by the fact that, in 1986, accounts receivable represented 9Z of ISA99 total assets. Conclusions 58. In collecting their dues,the performance of the two main Bank borrowers in the sector, "''I and EPH, reflects their general institutional performance. i.e., on the one hand, a reasonably satisfactory one by RPM, even under the difficult circumstances of the mid-19809, and, on the other, one that seriously deteriorated between the 19709 and the 1980s by ZMM. It is evident that together with the losses problem, that of the high accounts receivable needs urgently addressing In MMB. The 1987 Sector Adjustment Loan includes steps to this effect. The issue of relevance here is, however, that of setting covenants that are realistic. This applies particularly to covenants /targets for indicators over which utility managers exercise some control--such as energy losses and accounts receivable--but which require some time lag before results can become apparent. This is in contrast to earnings capacity, which is under - 153 - the con.rol of the guarantor and not the utility. In this case, performance derives from edicts that adjust prices and, hence, a willingness to perform prior to loan approval is likely to be more effective in terms of achieving the desired performance. 59. In ISA, high accounts receivable are not a reflection of poor commer- cial practices. They rather show the relatively low priority paying ISA for its supplies and services has in the eyes of its shareholders, another impor- tant point that has to be taken into account, when, as suggested in Section 3.5, ISA's role and structure needs to be re-thought. 6.8 Bank Lending Terms Development of Terms 60. At the beginning of the review period the terms of Bank loans were quite favorable in terms of maturity and grace periods and with debt service in form of equal semi-annual payments for interest and principal. Thus, the 1970 and 1972 Chivor I and Guatape II loans were for 30 and 25 years, respectively, with grace periods of 7 and 5 years, i.e. reasonably close to the construction times anticipated for the corresponding hydro projects. It also meant that the mismatch between principal repayments and depreciation charges (the latter based on a 40-50 year economic life) was modest and did not call for special measures. However, in 1976, in the wake of the first oil price crisis, the Bank reduced its loan terms, even for such long lead projects as hydroplants, to a maximum of 17 years, with a grace period of 4 years, while simultaneously levelling principal repayments. This meant for hydro projects like those in Colombia that: (i) principal repayments jumped to a level some three to four times that of depreciation charges, and (ii) loan amortization, on average, started 15 months before the plant started operation and 23 months before the Bank made the last disbursement from its loans. A composite indicator that captures well the effect of changing repayment schedules is the average loan life. It is a measure of the time during which the loan proceeds stay with the borrower; it represents the time between a single disbursement of the entire loan proceeds and a single balloon repayment of the loan that would lead to the same availability (in US$ years) as the actual loan. It is measured as the sum of annual loan balances divided by the loan amount. Hence, the longer the average loan life, the longer the period the borrower has use of the borrowed resources. Effect of Changes 61. The effect of the change in the repayment modus is illustrated by the case of the Mesitas loan, where the average loan life was 8.5 years assuming constant annual payments of interest and principal and 7.7 years when principal repayments are levellized. However, the impact of the shortening of grace and repayment periods is even more dramatic, as shown in Table 6.10. Indeed, whereas loan lives for the Chivor I and Guatape II loans were 16 years and more at appraisal and, despite delays, resulted in excess of 14 years, four loans made between 1978 and 1981 had, at appraisal, loan lives of 7.7 to 8.0 years. Actually, these average periods of availability of the resources came out at 3.2 to 6.7 years. The development of the maximum loan balance available to the * 154 - Borrower is another illustration of the effects of the changes In loan terms (Table 6.10). Whereas the Chivor and Guatape loans were expected to reach for a short time a 1002 balance, which they nearly achieved, the loan balance of the post-1976 operations referred to before were expected to reach 88 to L.000Z in reality, it ranged from 31 to 802. The lowest figure is for the extreme case of the Playas operation. In this instance, the shortening of the grace and repayment periods, long delays, and the fact that a project change reduced costs and led to disbursements amounting only to about half the loan amount reduced loan life to about 3.2 years and the maximum loan balance to a mere 312 of the loan amount. Fluctuating exchange rates may further worsen the outcome for all the loans discussed here. Table 6.10s Average Loan Lives and Maximum Loan Balance Loan Number 681 874 1628 1668 1953 Borrower ISA EPM ESEB EPM RPM Project Chivor I Guatape II Mesitas Guadalupe IV Playas Year of Loan 1970 1972 1978 1980 1981 Loan Amount 52.3 56.0 84.0 125.0 85.0 Maturity (Yrs) 30 25 17 17 17 Grace Period (Yrs) 7 5 4 4 4 Loan Life (Yrs)s Projected 19.0 16.0 7.7 8.0 7.8 Actual 18.8 14.3 6.7 5.4 3.2 Max. Loan Balance In Terms of Loan Amounts Projected 1002 1002 962 952 88% Actual 992 942 802 582 31% Bank Approach 62. Though Bank staff recognised(4), in the early 1980s, the effects of the new loan terms, especially for hydropower development, the Bank, in its appraisal and other reports, gave little weight to the financial consequences of the 1976 changes in its lending terms, though they were exceedingly significant for power investment programs heavily weighted with long lead hydroelectric projects as those in Colombia. Indeed, the President's Report for the 1987 Power Sector Adjustment Loan seems to have been the first official Bank document that highlighted the changes in Bank loan maturities and grace periods as contributing to the sector's financial problems, albeit it did not stress the significance of the impact. Such Impact, however, takes the form of substantially increased debt service requirements, which, in turn, call either for more borrowing, or better, if a reasonable contribution of revenues to investment is to be achieved, - 155 - for higher tariffs. The Bank, in its dialogue with Government and sector authorities, seems to have foregone this quite potent further argument for adjusting revenues upwards. Conclusions 63. In the context of the Colombian power sector, with its large investments in hydroplants, prone to long delays, the 1976 changes in the Bank's lending terms (shortening of grace and repayment periods and substitution of constant annual amortization payments for constant total amortization and interest payments), substantially affected the Bank borrowers' finances, in particular in the form of reductions in loan life. Though this effect was recognized by Bank staff in the early 1980s, somehow it was not highlighted in any SARs until the President's Report of the Sector Adjustment Loan (1987) and, as such, did not appear to be used as a further basis for much improved internal resource mobilization efforts by the sector. 6.9 The Foreign Exchange Risk and Losses Background 64. The exchange risk that built-up in the early 1980s and the large exchange losses that appeared in the mid-1980s had a large negative impact on the financial performance of the power utilities and the sector. This section assesses the consequences of the exchange risk and losses, the manner in which they have been handled in the utilities' audited accounts as well as in Bank SARs, and the actions by the sector authorities and Government to address the issue. 65. In the late 1970s and early 1980s the real value of the US$ expressed in Col$ increased by about one-third, with the actual exchange rate failing to off-set this development. Therefore, the exchange risk, which the utilities in the power sector have to bear, built up in this period. It materialized in the mid-1980s when a massive devaluation of the Col$ against the US$ took place. The effects were exacerbated by the decline of the US$ against the hard currencies in 1985186. They became manifest in sharp rises in financial expenses and a shortening of average loan lives. Consequences of Exchange Losses and Risk 66. The first impact of the escalating exchange losses was to increase significantly the utilities' debt service obligations and other financial expenses. Table 6.11 below shows that, from 1982 on, the exchange losses represented rising shares of EEEB's and ISA's operating margins (i.e. the difference between operating revenues and costs), peaking in 1985 at 431 and 451, respectively, but subsequently declining to about 241 in 1986 in both utilities. In the case of EPM, the audited income statement (see Annex 68.5) does not show exchange losses as an explicit financial expense; therefore, Table 6.11 does not set forth the data for this company. - 156 - Table 6.11: Impact of Bachange Losses on 3EDB's and ISA's Operatina Margins 1982-86 (Exchange Loss as I of Operating Margin) 1981 1982 1983 1984 1985 1986 EEEB 12 16 22 32 43 24 ISA 0.5 12 9 19 45 24 67. The accelerated devaluation of the 1985186 period was one of the measures the Government undertook to shift the incentives of production from non- tradeables (such as power) that consume large amounts of foreign exchange to tradeables. Government, however, did not take the next step, which would have reinforced the shift and provided a basis for better financial perfor- mance of the utilities and the sector, i.e. to allow power companies to pass through to consumers the costs of the accelerated devaluation. Since it did not, operating margins failed to reflect the new realities, which contributed, therefore, to increasing the level of exchange losses relative to the operating margins shown in Table 6.11. One of the consequences of not allowing the full devaluation costs to be reflected in tariffs was that the central government had to assume, de facto, some of the foreign debt service payments of the sector. 68. The second impact of exchange losses was to reduce the average life of foreign loans, since greater than expected annual repayments reduced the availability of these loans. For example, in the case of EPM, in 1985 actual repayments of principal were 90Z above those scheduled; in 1986 they declined to 152 above that scheduled. This meant that during two years (1985-86), actual repayments of principal exceeded those scheduled by some 40Z, resulting in the average loan life of the outstanding foreign loans to EPM being reduced by about eleven months, further limiting the availability of these resources to the borrower. 69. Of course, the exchange loss and risk situation has been aggravated by cross-currency fluctuations on Bank loans disbursed under the currency-pool system. Though this effect needs also to be judged over the full life of the loan from disbursement to full repayment, the examples below illustrate some of the difficulties that have arisen. In some cases Bank loans have tended to worsen the exchange risk of the utilities, and the currency-pool system does not appear to have reduced this risk, at least not during the period examined. Indeed, at end September 1988 the accrued risk was greater for pooled loans (averaging 35Z for four loans) than for non-pooled loans (averaging 27% for six loans). Table 6.12 below shows the shares of different lenders' contribution to exchange risk compared with their fraction of the total outstanding foreign liabilities of the borrowers. What this table illustrates is the inordinately high share of the exchange risk on outstanding liabilities associated with World Bank loans (as of end-1986), especially for 8388 and EPM, while on the contrary, the Bank's co-financiers in EPM operations (IDB and four commercial banks--Bank of America, Bank of Tokyo, Swiss Bank Corporation, and Chase Manhattan) contributed less to the risk than their share of EPM's corresponding outstanding loans. - 157 - Table 6.12: fchange Risk Contributed by Different Lenders Londer's Share of Lender's Share of Outstanding Recognised Risk tility Foreign on Outstanding Borrower Lender Liabilities Liabilities (2) (2) EM World Bank 3 (1986) 68 (1986) EPM World Bank 63 (1986) 67 (1986) ISA World Bank 31 (1986) 24 (1986) RPM IDB 13 (1986) 13 (1986) FPM Pour Commercial 21 (1986) 12 (1986) Banks Exchange Rate Forecasting 70. In the Bak's power lending operations, forecasting the US dollarl Colombian peso exchange rate was first undertaken at the time of the appraisal of San Carlos I (1977). All subsequent SARs presented forecasts of local and foreign price movements as well as those of the US$JCol$ exchange rates, though they did not state the assumptions underlying these estimates. The various projections of the key macroeconomic variables given in SARs are set out in Annex 5C.3. The *internal consistency' of these data required that the real exchange rate not increase, i.e. that the projected exchange rate and the projected locallforeign inflation ratio not present a widely divergent trend. It also implies that, when the projected exchange rate increases faster than the localiforeign inflation ratio, relative prices will decrease. This was the case for most of the SARs for which data are shown in Annex 5C.3, except those for Guatape II, San Carlos I, and Mesitas, the latter two assuming, over the period 1178-83, that Colombian prices would increase more rapidly than external prices without appropriate exchange rate adjustments. 71. In reality, domestic inflation could not be restrained to the degree foreseen, and foreign inflation, after 1981, fell to levels below one-half of those projected; and, consequently, the peso's devaluation rate had to be accelerated well beyond that envisaged. It is easy to be wise after the event, however, it is difficult to be too crItical of how the Bank handled this ticklish problem of exchange rate forecasting in the context of project lending. This is one area in which what the Bank presents in a reasonably accessible document, such as a project SAR, on a parameter as sensitive as the projected exchange rate, has to be tempered by broader considerations of what such prognostications can lead to in terms of the country's external resource management, especially if what is projected deviates significantly from the country's exchange rate policy at that point in tine. Added to this, it was not until 1986 that the treatment of foreign exchange risk began to acquire increased attention in Bank lending operations. This was precipitated by the exchange rate volatility, - 138 - beginning in 1985, between the US dollar and the hard currencies. This is evidenced by the December 1986 circular(5) which drew attention to the fact that the ftreatment of foreign exchange risk in Bank/IDA lending and on-lending has become more important,* and, therefore, required a 'stricter adherence to Bank policy.* This policy, in particular, stipulated(6) that autonomous entities enjoying IBRD financing should effectively pay standard IBRD leading rates, including foreign exchange and interest rate risks, while if government assumed part or all of the foreign exchange risk a fee should be charged to borrowers. Financing the Exchange Risk 72. As far as financing the exchange losses and risk, there is the concern with the actual servicing of the utilities' foreign debt (including the exchange losses component) as it becomes due. This should, in principle, be financed out of revenues; however, the treatment of exchange losses and risk in some companies' audited accounts indicates this is not the case. For example, EPM's audited income statements, do not explicitly show exchange losses incurred. Data obtained from ICEL for this evaluation do not detail exchange losses or risk in either balance sheets or income statements. Furthermore, while both EEEB's and ISA's income statements explicitly show exchange losses, in EEEB's balance sheets the FEN loan is treated as a local currency liability, implying no exchange risk, whereas EPM's and ISA's statements count it, correctly, as a foreign liability. Conclusions 73. Exchange losses, in particular those associated with the devaluation of the Col$ in the mid-1980s, significantly increased the utilities' financial expenses (in 1985, by threefold in EEEB and by 602 in ISA). They also reduced the average loan life (e.g. in the case of RPM) by nearly a year. The most salient fact, however, is that the various entities of the sector deal in a very different way with exchange risks and losses, thus, impeding an overall assessment of the issue. 6.10 Appropriateness of Some Financial Indicators Background 74. The resource mobilization indicator advocated in the Bank's Guidelines for power sector lending is the rate of return on net fixed assets in service. Under inflationary conditions the Guidelines call for an adjustment of the asset base to reflect changes in price levels through a revaluation. All of the Bank's lending operations to the Colombia power sector covenanted this resource mobilization indicator. The present section addresses, based on the Colombian experience, some issues related to the use of the indicator (see also Section 1.2, Annex 6A.2). Cash Generation Indicator 75. The practice of using only fixed assets in service to determine the asset base for the rate of return on assets is normal in North American and British utility practice. For large, matured utilities, experiencing limited - 159 - growth rates and for which fixed assets in service are likely to represent at least 85Z-90Z of total fixed assets, the use of assets in service as the rate base is well justified. First, because it represents toe actual revenue- producing fixed assets. Second, as such, this asset base is well defined since it is in operation, and third, these fixed assets in service represent a very large fraction of total fixed assets employed. In the case of fast-growing, small- to medium-sized power utilities, as the Bank's borrowers in the Colombian power sector were for most of the period under review, often more than 502 of resources employed in fixed assets represent assets under construction. In these cases, using only assets in service as the rate base leads to very high returns in order to achieve reasonable financial performance. These high returns called for often are resisted by the borrower due to their shear size. The need for such high returns is compounded especially when construction periods exceed loan grace peiods, resulting in debt service becoming due long before the asset under construction becomes part of the asset in service base. It would appear then that, under such circumstances, the Bank should consider, both for covenanting and for monitoring purposes, a return on an asset base including work in progress, or alternatively covenanting and monitoring self-financing in addition to return on assets. 76. A second problem that arose with the return on asset indicator used by the Bank in its power lending operations--the Revalued ROIs--concerns the revaluation process itself. Colombian law and accounting practice only allow partial revaluations of fixed assets corresponding to the revaluation of outstanding foreign liabilities according to the prevailing end-year peso exchange rate. This has meant that a full revaluation of fixed assets was done on a pro forma basis for the determination of the asset in service base of the Revalued ROI indicator used in connection with Bank projects. As discussed in Section 1.2 of Annex 6A.2, one of the problems that arose, particularly regarding EEEB, was the inaccuracy in the revaluation process which became apparent only some years after the event. Indeed, in the case of REEB it was found, in 1984, that the actual fully revalued fixed assets in service of EEEB as of end-1982 were some 402 higher in value than resulted from revaluations undertaken using the Colombian *blue collar* consumer price index. Since this index, which underestimates the impact of the foreign cost component, was used between 1976 and 1984, during this period, the asset base used in the calculation of Revalued ROI indicator had been increasingly undervalued, resulting in rates of return that were higher than those resulting from a more realistic revaluation. Naturally, this seriously compromised the effectiveness with which the Bank could monitor the earnings performance. 77. The above undervaluation of assets in Bank reports is likely to have contributed to sending unduly optimistic signals on EEEB's revenue perfor-mance. It also illustrates the limitations of the Bank using a procedure, in this case the full revaluation of assets, on a mere pro forma basis when this procedure has no legal basis in the country and the sector does not identify with the signals its results send out. - 160 - Conclusions 78. Some of the problems of covenanting and monitoring earnings performance In the Colombian power sector are related to the use of return on net fixed revalued assets in operation as the main indicator. Indeed, this indicator implies a full revaluation, which the Bank has to carry out on a pro forma basis as the procedure is not used in the country. Further, the revaluation procedure turned out to be inaccurate for the years 1976-1984 but was only discovered several years after the event. Finally, using assets in service for the determination of the rate of return in power systems, where work in pr'gress often represents more than 50Z of total fixed assets, is debatable, 6.11 The Azenda for the Future 79. The lessons learned from OED's review of Bank leading to the Colombia power sector between 1970-87 suggest that as a basis for a financially sound sectors (i) revenues and their collection must improve (i.e. higher tariffs and lower accounts receivable) i (ii) costs should be limited (i.e. more efficient opeations, in particular, reducing losses); and (iii) capitalization has to be re-established at the proper level (i.e. higher equity) to enable the sector to become less dependent on IDB and World Bank financing. To this effect, as already discussed in general terms in Chapter 3 in the context of the institu- tional issues, there is a need to establish first, in broad terms, and successively in more detail: - what the consumer will ultimately have to pay for, what Government will subsidize, and through which channels Government will provide the external subsidy; - which cross-subsidies between tariff categories will be implemented; - what utility revenues would have to cover, i.e. cash operating costs, debt service, proper adjustment to working capital, a reasonable contribution to investment, and an equitable remuneration for equity; - why accounts receivable in individual entities are high and how to remedy this shortcoming; and - what are the amounts of equity needed and when have the contributions to become available. 80. Starting from the above basis and taking into account the regulatory framework, which also should be substantially strengthened (see Section 3.3), Government, the sector, IDD, and the Bank should chart out a course to reach the agreed financial targets, setting precise, coherent, and achievable intermediary goals for the sector and all the utilities. An important early objective should be a reasonably uniform presentation of the utilities' accounts, such as to permit a reliable consolidation that would provide a clear view of sector finances. * 161 * 81. In addition, the Bank should review, inter alia, in the light of what has occurred in the Colombian power sector, the indicators it uses to monitor and covenant financial performance of such a sector and determine possible improvements, in particular, in measuring revenue performance. Finally, as part of a broader effort in improving power sector borrowers' monitoring of financial performance, such monitoring under financial covenants should become subject to audit as part of the external auditors' report on the annual financial statements of Bank borrowers. ENDNOTES (1) SAR Guavio Hydroelectric Project, paras. 5.18 - 5.24. (2) SAR FEN, paras. 5.10 - 5.15 and Annexes 3.36 and 3.37. (3) SAR San-Carlos I Hydroelectric Project, May 17, 1978, para.6.02. (4) Memo from J. Gorrio (LCPE2) to J. Fish, Power Adviser (EGY), dated 29 June 1981, OComments to Draft of Sector Support Strategy Paper, Electric Power.* (5) *Foreign Exchange Risk and On-Lending Interest Rate in Bank/IDA Lending in 1985," 5 December 1986, Memo by f. Kopp to ONS holders. (6) Circular No. OP13183, filed with OMS 3.81. - 162 - VII. SECTOR INVESTMENT 7.1 Introduction 1. The present chapter covers the discussion of a7 selection of issues related to sector investment. After briefly touching on the investment decision process, a point that Chapter 3 highlights from the institutional angle, it looks at planning, in particular of generation and transmission investment, an area in which the sector, through ISA, has achieved substantial progress, although much remains to be done. In this context, it makes the case for a multi- objective approach to investment planning. Next, it examines two of the main factors affecting investment selections Implementation costs and times, which are issues that the 1972 OED review already flagged. Finally, it addresses a further such issue, the balance between investment in generation and transmission on the one hand, and in subtransmission and distribution on the other. 7.2 The Investment Decision Process 2. In the power sector, increasing interconnection is usually associated with increasingly centralized planning and decision making. This was also the case in Colombia during the period under review, which, in the 1970s, brought the interconnection of EEEB's, EPM's, CVC's, and ICEL's main systems into the Central System and in the 1980s the link-up of this grid with CORELCA's Northern System. In the same period, ISA became the central planner of additions to the interconnected system. Since the mid 1980s it also plays an increasing role in coordinating distribution investment plans. 3. During the entire 1970-1987 period, with the above development of ISA in the background, the investment decision process for the generating and transmission programs essentially remained the same. ISA's technical staff prepared the investment plan, which, after discussion in ISA's Board was submitted to DNP for review, especially in the light of the country's develop- ment plans and Government's financial policies. After ISA would have taken care of possible adjustments agreed with DNP, the latter would submit the plan to CONPES for final approval. As discussed in Section 3.5, one of the weak-nesses of this seemingly straightforward procedure is that there is no clear separation of the technical and economical planning from the political pro-cess. Indeed, ISA's Board, as it consists of the representatives of the major regional utilities, while providing a forum for discussing and settling major issues affecting the power sector, already injects a large amount of political considerations into the decision making procedure. Therefore, agencies of the Central Government, in particular DNP and CONPES, are never in a position to assess clearly the cost of less than economically optimal decisions. Another weak point of the process is that the sector lacks the regulatory body that could actually enforce CONPES' decisions (see Section 3.3) - 163 - 7.3 Planning Generation and Transmission Chivor I 4. In the late 1960s, the assignment to ISA of the planning function in the interconnected system quickly led to considerable improvements. Chivor I, a hydroelectric project clearly in BUEB's area of influence, was the first generation project to be planned jointly by the major utilities of the sector, through ISA. As, at the time, ISA was still in its formative period, there was strong disagreement as to which agency, ISA or EEEB, should contract for and administer, first, project preparation by engineering consultants and, later, construction of the plant. In the end, ISA, strongly supported by the Bank, assumed full responsibility for the project but agreed to make use of capabili- ties available in EEEB, as this utility was just completing the construction of the 300 MW El Colegio plant. As set forth in Section 1.4, ISA's first plan, prepared in 1969, presented Alto Anchicaya (340 XV), Chivor I (500 MW), and Guatape II (280 MW) as the most promising projects. To identify the three above developments as such, ISA had considered 14 hydro projects ranging from the 60 MW Timba station to the 1540 MW Samana (later renamed San Carlos) plant. In fact, the actual choice was much smaller, as many of the projects examined were at such a preliminary stage of study that it was impossible to envisage starting their implementation within the next two to four years as the expected demand growth and construction times required. Other projects were part of multi- purpose ventures (e.g. Timba and Salvajina) for which the development of the irrigation and/or flood control components was crucial to their economic justification but was premature. Further, already in 1968, i.e. before ISA had carried out its final analysis, the Government had decided to allow CVC to start constructing Alto Anchicaya-IDB providing the financing. Indeed, the utility felt uncertain about the timely completion of Chivor or Guatape II, in case one of these would result in the most desirable first step. CVC was also anxious to limit its future dependency on imports from other parts of the interconnected system. This is a preoccupation of the main utilities that often would recur. ISA's analysis, which also considered an alternative including the installation of gas turbines before Chivor and Guatape II, concluded that, on the basis of production in the critical hydrological year, Guatape 11 should be built before Chivor. However, as the latter plant was able to produce more energy in an average year and as BPM, the owner of Guatape, was not interested in advancing the construction of the second stage of this development, the Government ultimately decided to go ahead with Chivor, which was to be built by ISA.(1) After a comparison, on the basis of a present worth analysis, of Chivor I with a corresponding thermal plant, the Bank concurred. It seems important to note that, what was decisive for the choice of the sequence to be implemented, was not so much the economic analysis, which essentially was inconclusive, but the state of project preparation as well as operational and political convenience. This statement is reinforced by the results of a very rough calculation carried out by OED using ex ante values given in IDB's ex-post evaluation of the Chivor project.(2) The figures suggest that the strict least cost solution for the three plant sequence including Alto Anchicaya would have Guatape II outranking Alto Anchicaya by a small margin and Chivor I by a somewhat wider but still modest margin. - 164 - Guatape II 5. When the Bank appraised this project, the only hydro projects which, on the basis of their state of project preparation, could be considered as alternatives to the 280 MW extension of Guatape (Guatape II) weres (i) Chivor II (500 W) with diversions providing additional water to the plant and hence allowing it to contribute not only a large capacity but also a substantial supplementary energy potential, (ii) Samana (San Carlos), for which the first stage was estimated to provide 572 MW (later increased to 620 MW), and (iii) Mesitas (515 MW, ultimately increased to 600 MW). The comparison of the present worth of the sequences including all the permutations of the above plants as well as thermal developments, yielded for all discount rates between 8 and 12Z Guatape II as the most desirable first plant followed by Chivor II, San Carlos I, and Mesitas. Here, an important point is that, at the time, it was impossible to complete either San Carlos or Mesitas by 1977, when the demand projections called for additional capacity. Therefore, the choice for the next facility to construct was really only between Guatape II and Chivor II. Taking into account that Chivor I was already delayed, it seems that the most economic plant happened to coincide with the only one that was actually ready for implementation. 6. The limited choice of alternatives mentioned above and the lack of uniformity in the level of project preparation at the time of the definition of the immediate program was one of the signs that overall coordination of sector planning was still limited, a drawback which the Bank had pointed out for already quite some time. In particular, it found that the country urgently needed a national power development plan, properly coordinated with Colombia's overall development objectives and with a national energy policy. The Bank's idea was, that such a plan, which would be reviewed annually, would include a long range generation and transmission program on the basis of which ISA would carry out, or at least coordinate, the further investigations leading to the actual investment decision; it would also serve as a guide for investments in urban and rural distribution systems. ISA, under the first Bank loan it received (Loan 575 of 1968 for the original interconnection), had already started in this direction, carrying out a number of river basin and feasibility studies for new hydro plants (e.g. for Samana/San Carlos). In connection with the Guatape II loan, Government agreed to carry out a broad nation,,l plan which would also suggest institutional reforms to increase the sector's planning and operating efficiency and help coordinate financing of the substantial investment required.(3) As set forth in Section 1.4, the preparation of this plan, which was to be completed in 1973, suffered changes in terms of reference which made it fall short of the original objectives. It was completed, with German assistance, in 1979. At that time, its most important component had become the inventory of hydroelectric resources. San Carlos I and II, Mesitas, Guadalupe IV and Playas 7. The Bank's economic justification of the San Carlos project as the next plant in a least cost development was essentially based on ISA's 1975 generation program, which (see para. 6 above) was not yet integrated in a long term plan. In one of the alternative programs it considered, the Bank included a coal fired - 165 - steam plant in place of Mesitas. The analysis shoved that the sequence with this plant had the highest net present value for all discount rates up to 172. (4) The cost of coal considered at the time was 1977 US$251t. Had the computation used a cost of 1977 US$1St, which is about equal to the value of US$251t (1986 prices) inland coal can, at most, be assigned today, the program with the large thermal plant would have become the least cost solution. (5) This is not to suggest that the calculation included in the SAR was wrong, but it demonstrates a point which will recur in the following, i.e. how far from the assumptions actual values can come outg how wide the range of values should be; and which sensitivity tests and risk analysis project appraisal should ideally cover. 8. The demonstration that iesitas and San Carlos II were the next plant in the least cost development of the sector's generation capacity was done in the context of the appraisal of the San Carlos I project and therefore the observation, made in para. 7 above, as to the valuation of the thermal alternative applies again here. 9. ISA successively adjusted its 1978-85 development program of 1975, to include new projects, the studies for which ISA itself and its shareholders were carrying out. The program also covered the two emergency plans Government had agreed to, in an attempt to avoid shortages in the early 1980s (see Section 1.5). At this stage the procedure definitely lacked clarity, as projects at substan- tially different stages of preparation were considered at the same level. Furthermore, the individual shareholders of ISA were strongly promoting the facilities in their own area of influence and were already implementing or considering implementation of these projects on their own rather than through ISA. This development away from the principles originally adopted for ISA had started in the mid-1970s. It culminated in the OAcuerdo de Cali* which confirmed the assignment to each shareholder of at least one large plant (see Section 1.6) s Mesitas (600 MW hydro), already under way, to BEE as well a. Guavio (1000 MW hydro); Guadalupe IV (260 MW hydro) and Playas (170 MW hydro) to RPM; Salvajina (180 MW hydro) to CVC; Betania (500 MW hydro) to ICEL (under participation of ISA), Cerrejon I and II (2x150 MW steam) as well as Urra I and II (1050 1W hydro) to CORELCA. and San Carlos II (620 Of hydro) as well as Jaguas (170 MW hydro) to ISA. In spite of this somewhat chaotic procedure, the Bank ultimately provided financing for Mesitas, Guadalupe IV, San Carlos II, Playas, and Guavio. IDB, in turn, contributed to Salvajina, Jaguas, Playas, and Guavio. Both institutions argued in each specific case that the project proposed for financing was the next step in the least cost solution for the interconnected system, albeit they pointed out that the process for reaching the investment decision was not optimal [see e.g. (6)]. Guavio 10. Mhereas the SAR for the San Carlos project included the Bank's own evaluation of alternative sequences of plants, the Bank's justification of Guavio as the next plant after Playas and before Urra is entirely based on ISA's July 1977 study of the least cost expansion of generation in the interconnected system, updated in 1978 and 1980. The alternative sequences ISA compared included, as usual, oil- and/or coal-fired steam plants. However, the study used for coal the export prices prevalent at the time, and these were much higher than - 166 - those that could have been considered for non-exportable coal. ORD carried out a very much simplified comparison involving on the one hand Guavio and on the other a series of four 250 MV coal-fired steam units. On the basis of the original cost estimate and a coal cost corresponding to US$25/t (1986 prices) (see para. 7); the results show Guavio as the more economic solution for discount rates below 11Z (i.e. approximately the opportunity cost of capital considered at the time). This assumes that the plants had to meet demand growing at a rate of 10.6Z per year, as projected by ISA, which was higher than the average gr.th rate experienced between 1971 and 1980. If demand is assumed to grow only at 8 per year--which is slightly more than the average growth the sector experienced between 1981 and 1986, the equalizing discount rate falls to some 8Z. Increasing coal costs by 30% leaves the rate at 13Z and 11%, respectively, for the two growth patterns. These results are an indication that, using data that were available at the time of project appraisal, the case for Guavio is less than clear-cut. This is not to suggest that the conclusion should have necessarily been different. However, it shows that the sequences involving thermal plants were about as economical as the hydro alternatives and would certainly have deserved more discussion, as they had the advantage of greater adaptability to changes in demand development, of lower risks during construction and of easier financing, as the capital cost was lower than for the exclusively hydro sequences. Determining the Least Cost Plant Sequence 11. ISA, supported by the Bank has, in the course of the period under review, substantially refined its methods of determining the least cost program of plants. For quite some time, ISA has been defining alternative expansion programs, including hydroelectric projects as well as thermal plants (coal, fuel, and gas fired steam plants, and gas turbines) and designed to meet energy and peak capacity requirements through the period under study. These sequences are introduced into a model which simulates the operation of the system using variable hydrological and demand data but allowing that energy requirements would not have to be met under adverse hydrological conditions occurring at most once in 20 years. Adjustments are introduced to ensure that capacity deficits would have a probability of occurrence of less than 52. The most desirable program is then selected as the sequence with the least present worth of investment and operating costs discounted at the interest rate corresponding to the estimated opportunity cost of capital assumed, [in the case of Chivor, assumed at 8 to 102; in connection with later projects, at 112; and, in the case of Salto Grande, appraised in 1983, also at 112; nevertheless, in this case the alternative cash flows were discounted at 122 (7)). There is little doubt about the validty of the assumption of the opportunity cost of capital at 82 to 112 in the 1970s. Whether this value is still justifiable in the 1980s--vhich witnessed a sharp increase in the scarcity of funds for developing countries--which the discount rate used in determining the least-cost solution is supposed to reflect, seems debatable (see Section 2.5). It is evident that the use of a higher discount rate would have led to least-cost solutions of substantially different composition than the ones which ISA determined and which essentially resulted from the approach the Bank endorsed throughout the review period. - 167 - 12. The two graphs in Annex 7.1 show some of the results of such planning in the Colombian power sector. A selection of the successive plans are set against the actual development. The figures clearly show the periods in the mid-1970s and early 1980s when the available capacity and energy potential were insufficient to supply reliably the required energy and/or capacity; the periods of shortages are further discussed in Sections 1.7 and 4.6. It is important to note that the lowest curve in the figures representing the capability under adverse circumstances cumulates possible hydroelectric output in a dry year with a rather low capacity and energy (particularly the latter) generation potential of the thermal plants. Whereas this corresponds to figures that ISA used, (8) it seems unnatural to assume that, in a year when hydroelectricity is at a low, thermal plants, especially modern steam plants, should not be able to run for an equivalent of, say, 6000 to 7000 hours at full capacity. 13. Throughout the 1970s, the group of projects ISA considered in its alternative sequences comprised the facilities it had studied itself, in part on the basis of the studies (i) of individual watersheds it had carried out in the context of the 1968 original interconnection project and (ii) of proposals prepared by its shareholders. The latter projects were not necessarily prepared on the basis of the same technical criteria ISA was using. As set forth in Section 1.4, and in para. 5 above, the Bank tried to remedy to this unsatisfac- tory state of affairs by requiring already in connection with the Guatape II operation in 1972 the preparation by 1973 of a sector development plan, an effort that was only partly successful as it resulted in a mere inventory of hydroelec- tric resources, and this only in 1979. After the submission of the inventory, the situation should have improved somewhat as to the comprehensiveness of the pool of projects from which to choose. However, as the expected fast growth of demand called for a large number of new installations and as project preparation at feasibility level could not catch up quickly with the demand for such, projects, the alternative sequences analyzed continued to comprise the projects that happened to be available at the time. These sequences, however, ususally included the plants favored, not necessarily on technical and economic grounds, by the-individual ISA shareholders. Sensitivity Tests and Risk Analysis 14. Whereas, until the late 1970s, with the relatively reliable development (in spite of the 1973 first oil crisis), there was some justification for the modest variations assumed as possible for the main project parameters, such a justification disappeared with the second oil crisis in 1979, followed by the recession, which introduced a high degree of uncertainty in demand, costs, availability of financing, revenues, etc. However, not only the environment, in which the projects had to be implemented but the projects themselves became more risky, as the easier ones were either already completed or under construc- tion. Therefore, project evaluations, at least those later in the review period, beyond large contingency allowances, should have included sensitivity analyses involving substantial variations of the main parameters. Furthermore, each assumed variation should have been associated with a judgment on the likelihood that the perceived risks would materialize. It would have been desirable to define the basic data used for further analysis in such a way that they would have accounted for some of the risks materializing. The case of the ill-fated - 168 - Guavio project, which ISA determined on the basis of the usual procedure implying limited variations of the main parameters, illustrates the shortcomings of the Colombians' approach. Indeed, in this instance, nearly all possible risks materialized and did so to a degree which, in most cases, was unexpected$ (i) The project started at the very moment the world economy went into recession and Colombia's external sources of finance dried up; (ii) soon after project start a tragic accident claimed more than 200 lives and ignited labor problems, which continued to plague the project off and on; (iii) unexpected difficulties occurred during tunnellingt (iv) points (ii) and (iii) above led to delays and dramatic cost increases; and (v) revenues not sufficiently adjusted to meet the funding requirements induced EMF to fall into arrears with its contractors, which, in turn, was at the root of further delays and cost increases. 15. Using wider ranges of values for the main parameters, and, in particular, varying the demand projections--which ISA's model does to a limited extent-- may ultimately still lead to the same solution as that suggested when assuming smaller possible discrepancies from the base case. Or, it may suggest the desirability of a more flexible approach, e.g. proceeding by stages within the same basic solution, which, though possibly less economic, at least on paper, than the basic solution, may be more adaptable to developments different from those expected. It may also lead to programs entirely different from the ones originally envisaged, which make up in adaptability to unanticipated circumstan- ces what they lack in straight measurable economy. A combination of substantial thermal units and a medium-sized hydro plant in a well-known area instead of a large, long-lead plant in a risky geology may be such a solution, which may also be easier to finance as the capital costs are smaller, a non-negligible advantage in the case of Colombia where the power sector is in financial difficulties to start with and external financial resources are limited since the early 1980s, in part, as a Government policy. Another, related, model could consider the large hydroplants as the basic building blocks that would meet a demand corresponding to the slowest development conceivable. These would be comple- mented by thermal developments (or smaller hydroplants) with shorter lead times to cover the additional demand, when demand growth becomes better perceivable. 16. The above approach to risk analysis entails that the investment decision comes back from the near automatism it acquires in the computerized simulation to the realm of judgement. This, at the same time, means that further sophistication of the planning tools is only valid as far as it provides a better basis for judgement, not a seemingly more precise economic analysis of the programs. Multi-Objective Planning 17. The traditional "least total economic cost, approach to power system planning used by ISA is essentially a single-objective exercise. This means it selects the alternative which minimizes one attribute, the present worth 4of total investment and operating costs (discounted at the opportunity cost of capital) and, in turn, selects the investment sequence possessing the lowest of such costs. This approach explicitly relegates the financial impact of alternative investment sequences to an issue of secondary significance. Indeed, once the least economic cost investment sequence emerges, it becomes virtually - 169 - axiomatic that finances must be found to Implement it. This has essentially been the approach pursued in Colombia (and in other power sectors supported by the Bank). Often, there has been little focus on the financial affordability of the investment sequence to the sector and the economy as a whole. Indeed, it would appear as though a divorce has occurred between the two essentially iterative elements of investment decision making--namely, the economic and the financial. This absence of an economic and financial assessment of competing investment sequences could result in the selection of a sequence that may be of *least economic cost' but not necessarily of Oleast financial cost* given the terms and type of financing available. This becomes important especially when--as it is likely to be the case in Colombia, in the future--two conditions prevail: - the competing investment sequences have very different mixes of hydro and thermal plants; and - significant differences exist between the economic lives of the assets (and hence the economic cost recovery rate) and the maturity period of borrowed funds. 18. The preceding considerations suggest that the traditional approach focusing on project economics and dealing with other objectives in second priority has run its course. A systematic multi-objective, multi-option, multi- uncertainty approach may lead to more satisfactory results, not only in Colombia. In most instances achieving "least economic cost' may still be a main objective, but the analysis would deal with it on the same level as with e.g. 'minimizing revenue requirements* and/or *limiting environmental Impact.' 19. The discussion of risk analysis (paras. 14-16) has set out the increasing Impact of uncertainties, in particular, those associated with construction times (and hence financing needs) for certain types of projects, like hydro and nuclear compared to, say, gas-fired thermal plants. In addition, significant uncertain- ties surround growth rates of power demand and future exchange rates. The multi- objective focus of such planning would, in particular, provide the vital link needed between the economic and financial attributes of different investment sequences and thereby establish a balance to ensure that what is implemented is economic and is also what is affordable. The tool used for a multi-objective analysis (see a.g. (9)] is less Important than the principle of dealing with several objectives at the same level. One of the consequences of moving to the more complex planning method would be that quite often it may not be possible to 'optimise' any single investment sequence in terms of the relevant objectives and uncertainties sinulatenously. Under these conditions the decision making becomes an exercise in reviewing conflicting factors to find the best trade- off between desirable and undesirable effects. This should bring more judgment back into the investment decision-making process, which, with the advent of computer modelling tools, has acquired more and more characteristics of a 'black box,' highly mechanistic exercise. * 170 - Conclusions 20. As discussed further in Section 3.5, the decision making process for the power sector Investment program requires institutional adjustment that make sure that the political input comes forth at the appropriate time (i.e. once all the main characteristics of the alternatives are available) and at the appropriate institutional level (presently this level is too low). 21. The methodology which ISA uses for planning generation and transmission is standard but calls for some improvementst - As discussed in Chapter 4, the sector is adjusting its methods of estimating future demand. However, success will only materialize in some years. - The set of projects earmarked for inclusion in the alternative sequences to be analyzed should be (I) appropriately comprehensive, to include all the plants that could be of interest for contributing to meeting the demand in the medium term and (ii) reasonably coherent as to the state of preparation of the individual projects. Therefore, the Bank, in connection with any future operation in the sector, should strongly support the measures agreed in 1987 between Government and IDB in the context of the Second Additional FInancing for the Quavio Hydroelectric Project. These aim precisely at comprehensiveness and coherence of the set of projects. - The rough computations mentioned in paras. 7 and 10 suggest that it is particularly important that ISA overcome its long predominant bias against thermal generation (except in the CORELCA system) and prepare full feasibility projects for thermal power plants using non-exportable coal and gas. Indeed, the 1986 energy sector study estimates the value of such coal at US$16/t (1986 prices) (10) as compared to the 1977 US$17 (about US$251t(1986 prices)) the Bank used in connection with the justification of the San Carlos I project and the 1982 US$55 (about US$62/t(1986 prices)) used by ISA in its 1982 revision of the 1986-92 expansion plan.(11) Furthermore, in 1988 such coal was available in Bogota at prices as low as US$9/t though in very low volumes compared to those needed for power generation. Such figures suggest that thermal plants should at least be seriously considered. - A bolder approach to risk analysis seems called for. It should provide for (I) wider ranges than those used in the past for the possible values the main parameters may assume, (ii) judgments about the likelihood for the risks to materialize, and (iii) some of the risks materializing. - 171 - 7.4 Project Implementation Time General 22. The 1972 OED review of Bank operations in Colombia noted that most power projects, In particular hydroelectric plants, financed by the Bank between 1950 and 1970 had suffered major delays In Implementation. The reasons given for this weres(12) - major technical difficulties with underground conditions, in particular, tunnelling problems, - landslides at construction sites requiring substantial changes in design, - inadequate contractor performancel and - lack of coordination between civil works and equipment delivery. 23. Annex 7.2 sets forth a sumary analysis of 21 projects completed in the period the present review covers. The figures have to be taken as fairly rough, as there are major obstacles to a precise definition and measurement. Nevertheless, the order of magnitude of the various times is reasonably reliable and telling. The list includes all the operations which the Bank helped finance in the power sector or of which the sector was one of the beneficiaries. Whereas the sample of projects other than those involving major hydroelectric plants is too small to allow definite conclusions, it nevertheless does not contradict the intuitive knowledge that the problem of delays, especially during the construc- tion phase, is less acute in the field of transmission and thermal plant projects than in hydro developments, where it continues to be a major issue. The list does not include the Guavio project, where delays are particularly dramatic. However, the plant is still far from completion. Even so, the average delay with respect to the time of completion assumed is some two years, one incurred between decision to implement the project and actual start of construction and another during construction proper. Pre-Construction Phase 24. The general study of the developments during the past eighteen years, to some extent reflected in Annex 7.2, suggests the following main reasons for the delays in the pre-construction phases - Decision to construct on the basis of an inadequately prepared project (e.g. Guadalupe IV, Playas, Jaguas) and associated longer design time; - Underestimated time required (i) to settle the usually complex institutional issues associated in the Colombian power sector with the definition of ownership and of participation of utilities and other beneficiaries in the project (e.g. Salvajina and Betania), and (ii) to organize financing (e.g. Chivor II, San Carlos I and II, Salvajina); - Unexpectedly protracted procurement (e.g. Chivor II, Jaguas). - 172 - 25. In order to account for the above experience and to assure a more orderly decision making and pre-construction process, ISA is now envisaging to plan hydro developments In such a way that construction proper starts only some five to six years after the decision to proceed with the construction of that plant has taken place on the basis of the proper update of the general sector development plan. This allows for some 3-1/2 years for post feasibility and final design on the one hand and 1-1/2 year for pre-construction procurement on the other. In parallel, there would be as much as some five years for organising financing. This initiative would tend to support the model set out in pars. 19, which would consider the large hydroplants as the basic building blocks which would allow to meet a minimum demand, having plants with shorter lead times forming the complement. Construction Phase 26. Analysing the performance after construction start, also summarized in Annex 7.2, suggests the following reasons for the partly very long delays in hydro plant implementations - geological, especially tunnelling difficulties (e.g. Guatape 1, Chivor I and II, Mesitas, Chingaza, and Jaguas)a - obstacles to resettlement and acquisition of land and of rights-of-way (e.g. Guatape II and Jaguas); - lack of financing, in particular in local currency (e.g. San Carlos II, Jaguas); and - deliberate slowdown of construction because of sharp reduction in demand growth in early 1980. (e.g. San Carlos II, Jaguas and Guavio). 27. If one takes into account that the redesign of the layout in the case of Playas and Jaguas as well as the difficulties in resettlement militate in favor of more preparatory work and that the corresponding delays have to be imputed to the pre-construction phase rather than to the implementation stage, the main reasons for construction delays are clearly those related to geological difficulties and the lack of funds to pay contractors on time inducing the latter to slowdown execution in order to limit their accounts receivable. 28. The lack of funds is, of course, essentially related to the difficult financial situation the sector, in general, and ISA, in particular, has been facing off and on during the past eighteen years, especially, however, since the early 1980s. This array of issues is discussed in Chapter 6. As to the geological problems, it is well worth noting that of the projects which have suffered in a major way of them, the three most affected (the Chivor and Mesitas hydro plants, on the one hand and the Chingaza diversion tunnel, on the other) are all in the Eastern Cordillera, in the larger Bogota area. This is the same region in which EED is Implementing the Quavio hydro development which, inter alia, also has to overcome major obstacles related to deficient underground conditions, which had been anticipated, but not to the extent they materialized. - 173 - Due to the complex, highly fractured formations in this area, large construction work, especially tunnelling, is substantially more risky there than in other parts of the country, In particular in the essentially compact body of rock dominating the geology In the Central Cordillera, where ZPM's plants and ISA's San Carlos and Jaguas facilities are located. ISA's future generating planning should try to account for these and similar differences In risk through (i) increased preliminary subsoil investigations in more risky areas, (ii) appropriate physical contingencies in cost estimates, (iii) thorough risk analysis (as against a mere sensitivity test) and the corresponding contingency planning. Insofar as it has the resources to do so, the Bank should support the sector in these endeavors. Conclusions 29. Most power sector projects, but most prominently hydroelectric plants, continued to suffer major delays. For the fifteen hydro projects of the sample analysed in this review, the average delay incurred between decision to implement and commissioning was about two years, one suffered before the start of construction proper, the other during construction. Pre-construction delays were mainly due to (i) major design chan,ges, (ii) institutional issues (who builds?, who owns?, and who participates?), and (iii) protracted procurement. The construction delays proper were mostly related to (i) geological difficulties especially in tunnels), (ii) resettlement issues, (III) lack of local currency funds, and (iv) deliberate slowdown after the 1982-83 sharp reduction in demand growth. 7.5 Project Costs Previous Findings and Present Analysis 30. The 1972 review of the Bank's operations between 1950 and 1970 found that there were, in general, substantial cost overruns in the major projects carried out with Bank support. The statement not only referred to the actual costs in current terms but also to those expressed in constant terms. Generating plants were the facilities most affected and the local cost component suffered from the largest overruns. OD also discerned a trend towards improvement. (13) The present analysis concentrates on generation and main transmission facilities, which the Bank helped finance and the sector entities completed within the review period. Because of the project's significance to the sector as a whole, the analysis also covers the Guavio hydroelectric project, which should have been completed within the period, but which, unfortunately, will not be commissioned before 1992. The figures for this project reflect actual costs until 1987 and 1987 projections for the remainder of the construction period. Annex 7.3 sets forth the basic figures for the analysis of the estimated and actual costs in Col$ of the project sample which, however does not include a thermal plant, as no such plant was part of a Bank project in the review period and OED did not find information that would have allowed for such a plant an analysis similar to that which was possible for the Bank projects. It is important to note that the analysis uses exclusively costs in Col$, the foreign component being transformed to the local currency at prevalent exchange rates in the years the costs were incurred. The core of the analysis is the calculation of an actual - 174 - base cost by deflating actual expenditures to the year the cost was estimated. This was done using the GDP deflator for the local currency and the import price deflator for the foreign currency component. Where 1987 US$ costs appear, they are purely indicative. They result from (i) an actualization to 1986 Col$ and (ii) a translation into 1986 US$ at the average 1986 exchange rate. The approach over the translation of yearly expenditures into US$ at the prevalent exchange rates and the actualization using US$ deflators would produce somewhat different results as, during most of the past eighteen years the exchange rate failed, to a varying but often very substantial degree, to reflect cost parities. Base Cost and Physical Contingencies 31. The first comparison shown in Annex 7.3 measures actual base costs in Col$ (deflated as explained in para. 30 above to the year of the cost estimate) in terms of the estimated base cost increased by the allowance for physical contingencies. This essentially is a measure of the accuracy of the quantity and unit cost estimates, on the one hand, and of the adequacy of the physical contingency allowance, which in the case of the hydro power plants physical contingencies correspond to 10% (Guatape II) to 162 (Chivor I and Guavio) of base cost, on the other. For the main transmission lines the appraisals assumed the physical contingencies at 13? of base cost for the 1968 Central Interconnection Project and at 51 for the more recent North-Central Interconnection. Annex 7.4 shows that, whereas the actual base costs of Guadalupe IV and Playas were less than those estimated (mostly because of redesign after appraisal), the other projects including the two involving transmission, had base costs reasonably in line with the estimates including physical contingencies, except for Mesitas and Guavio. These two projects covered major tunnelling in the Eastern Cordillera, the risks of which were obviously underestimated (see also para. 28 above). Surprisingly, in seven of the nine cases analyzed in Annex 7.4, the physical contingency allowance in foreign exchange was estimated to be higher than that in local currency. This assumption was proven to be generally inadequate. Indeed, in all cases analyzed here except the two oldest ones (Central Interconnection and Chivor I) the ratio between actual local and foreign base costs is higher--and in the cases with large overruns--much higher, than that originally estimated, which means that, when there is a cost overrun, that in local exchange is usually higher than that in foreign currency, and, conversely, when there is a cost underrun, that in foreign exchange is lower than that in local currency. This suggests that the problem in evaluating local costs, already noted in the previous review, persists, even at the level of the determination of base costs although both the Colombian power sector and the Bank have sophisticated their methods of determining project costs. 32. Overall, it seems clear that the accuracy of base cost evaluation has improved since the 1950s and 1960s. However, there are still areas needing substantial attention. One such field is the proper consideration of risks in underground work, which has to be specific for the individual situation as against the more generic approach which seems to have prevailed during the period under review when this kind of work was assigned a 15% to 20Z physical contingency allowance; the Mesitas project was the exception as in this case the allowance was 25?, which still resulted substantially too low. This means - 175 - that more of the post-1974 tunnel cost estimates, except that for Mesitas, cctplied with the Public Utilities Department's Guidelines for Estimating Costs of Tunel Construction (January 1974), which suggests that for such construction physical contingencies should not be less than 25Z of basic costs. 33. Physical contingencies are only one expression of the construction risk perceived. Indeed, in the case especially relevant for Colombia of tunnel stretches under large rock cover, --ere exploratory borings cannot, at reason- able cost, reach the area of the planned facility, the required construction methods and lining, if they have to result in an economically optimal solution, will always imply a high degree of uncertainty. Therefore, cost estimates, depending upon the designer's judgment, will include an implicit amount of physical contingencies just by increasing, e.g., the length of tunnel assumed to need reinforced lining beyond that strictly required on the basis of the results of the subsoil investigations. To use again the example of tunnelling, the amount of implicit physical contingencies included in a cost estimate may also vary as a function of the school of thought to which the project engineers belong. 'Whereas in the 1960s and 1970s but to a lesser degree in the 1980s, many European engineers considered that a pressure tunnel should be lined with concrete throughout, except in rare cases under exceptionally good rock conditions, for which the Colombian Andes would give practically no example, the Colombian engineers and their largely North American advisers generally followed less couservative principles often applied in the U.S. and Canada, according to which, lining can be avoided in many places and that even if some is required, the full circumference lining with formed concrete would be the exception rather than the rule. It is evident that the first approach implies less risks (smaller explicit but generally larger unspecified physical contingencies) than the second, for which, logically, explicit physical .ovtinger.cy allowances should be higher. Price ContIngencies and Actual Escalation 34. Annex 7.3 shows that, as further discussed in para. 30 above, physical contingencies were at least in the same order of magnitude as the increases in base costs except for two cases. Nowever, ahe same Annex illustrates that this does no4 apply for escalation compared to price contingencies. Indeed, in most ,ases the former is two to nine times the latter. Annex 7.4 sets forth the main results of a rough analysis of the escalation component of actual costs isolated in Annex 7.3. It shows the eicalation attributable to the effect of inflation on estimated base costs and physical contingencies and on that related to increases in base costs and to implementation delays. An attempt to identify the individual effects of the two latter factors was not successful. 35. In the late 1960s, cost estimates for Bank projects did not yet include price contingencies, which was a reflection of the reasonably stable conditions prevailing in the preceding period. From the sample analyzed in Annex 7.3, the Central Interconnection and the Chivor I projects belong to this type. The first of these projects was little affected by inflation as it was completed withi a short period and on time. Construction of the Chivor I hydro plant, however, fell into a period of severe inflation (1972-76) both local and foreign. Further, it suffered from substantial delays which exacerbated the cost - 176 - increases. The same applies to the Guatape II project, for which the SAR, prepared in 1972. already included a modest allocation for price contingencies based on the assumption of Inflation at 41 per year for foreign costs and 101 for local costs. 36. With the San Carlos hydro project the Bank introduced a more sophisti- cated approach by using variable inflation rates for both the local and the foreign cost components. However, the estimates suffered from the fact that they assumed that after the first year with levels of inflation actually experienced earlier in that year or in the year before, inflation would quickly abate, which it did not, thus contributing to the large discrepancies between estimated and actual escalation set forth in Annex 7.3. The fact that the above bias in the assumed future inflation, without disappearing, lessened In connection with the subsequent projects (Annexes 7.4, 7.5, and 7.6) indicates that a learning process was underway. However, comparison of the actual escalation on base costs plus physical contingencies (first column in Annex 7.3) to the price contingencies (last column in Annex 7.3) suggests that the corresponding Improvement fell short of leading to estimates of escalation reasivdbly comparable to the reality, except for the latest Guavio estimates, which reflect the fact that the original inflation assumptions proved too high at least through the mid-1980s, especially because of the impact of the appreciation of the US$ with respect to other main currencies. 37. It seems that after the Bank overcame the reluctance prevailing 'M the first half of the 1970s to incorporate into project cost estimates large price contingencies, the quality of the estimates improved substantially, but it still fell short of estimating the magnitude of the impact of escalation, even when there were no base cost overruns (see analysis of Guadalupe IV and Playas projects In Annex 7.2 columns 1 and 4). There seem to be two reasons for this. On the one hand, after the first energy crisis in 1973, the economic environment had become much less predictable than in earlier years. On the other hand, the Bank is always reluctant to overestimate inflation, as the values it uses have to be consistent with the projections of economic indicators it assumes in connection with country economic discussions, where it will tend to set them at levels requiring a major effort to achieve the targets; therefore, the appearance in a cost estimate for a Bank project of inflation rates higher than those used in the context of negotiations with Government on general economic policy might contain an element of self-fulfilling prophecy. As a consequence, it seems that there is little that the Bank can do to improve on the estimates for price contingencies except to improve on the economic projections in general, which most probably will only be possible in a less volatile environment than that prevailing since the mid-1970s. Interest During Construction 38. Interest during construction (IDC) is a cost element that doesn't seem to get the high profile it deserves, in particular in the environment prevailing in the Colombian power sector with its history of long delays. Of course, the financial projections in the SARs take these interests properly into account. However, except for the Guavio project, in no case reviewed here, are they - 177 - explicitly stated in the appraisal cost estimate. That they would deserve a more thorough discussion is illustrated by a rough calculation for the Guavio project, which leads to the following results* - Assuming base costs and physical contingencies of 1981 Col$35.0 billion, which approximately corresponds to the SAR estimate, IDC amount to some Col$23.5 billion for a construction period of 7 years and for the appraisal assumptions on the development of Inflation and of the exchange rate. - With the same base costs but a 1981-93 construction period, IDC increases to Col$80.8 billion, which corresponds to an increase of about 251 per year of extension in construction time. - Increasing concurrently base cost including physical contingencies and construction time to the values presently estimated, i.e. Col$65.0 billion (1981 prices) and 12 years, respectively, leads to IDC amounting to Col$135.5 billion or nearly US$1 billion. Conclusions 39. In 1972, OED found that substantial cost overruns were pervasive. However, it discerned a trend towards improvement. The present review concludes that the trend continued. Indeed, for the Bank projects completed between 1971 and 1987 (plus Guavio, to be completed in 1993), it finds that the estimated base cost plus the physical contingencies reisonably approximated the actual cost in constant terms of the year of the estimate, except in four cases. In two of these (Guadalupe IV and Playas) the actual base cost is lower because of major project changes after project appraisal. In the two others (Mesitas and Guavio) the actual base cost is much higher than the estimated one because of the difficulties encountered, inter alia those related to unexpectedly severe geological problems. The analysis further suggests that there is an Improvement in the Bank's estimate of price contingencies. This means that a large part of the vast differences between estimated and actual costs in current terms is related to the cuulatl-e effect of delays and inflation, and that, unless the time scheduling improves, substantial cost overruns are likely to persist. 40. The present analysis makes a case for the SARs explicitly setting forth interest during construction corresponding to the specific project proposed, and to submit these to the sensitivity analysis assuming possible delays in implementation. This seems particularly recommendable for the long lead developments which are prone to substantial delays. 7.6 Balancini the Investment Program Conclusions of the 1972 OED Review 41. A major finding of the previous OED review was that, by 1970, in most of Colombia, sub-transmission and distribution systems were inadequate, as the sector dedicated too few of its resources to such facilities. Specifically, the review set ftovh that between 1965 and 1970 the sector allocated *only some 391 of total investment to this area.' It concludes that this contributed to - 178 - the high losses (15 to 25% of energy sent out) experienced. OED also pointed out that the imbalance was partly due to cost overruns in generation plant construction, requiring a re-allocation of part of the resources originally earmarked for distribution(14), which term, from here on, is meant to cover both sub-transmission and distribution. 42. The earlier OED review also concluded that the Bank should encourage the sector to study ways to optimize distribution standards in Colombia.(15) 1970-87 Development 43. Section 2.6 already discussed the lack of reliable sector data in general and of information on distribution in particular. This fact, obviously, is at the source of the large inconsistencies in the information given on the subject in SARs and other Bank documents of the period. It also impedes a deeper analysis in the context of the present review. ISA has recently been assigned the general coordination of sector activities in distribution. However, the data it has available at this time still falls substantially short of what it needs for a general planning of distribution development, which would be a basis for coordinated, detailed investment programs of the various distributing utilities. The poor sector data base, which is closely related to the regional companies' reluctance to coordinate their activities, has, most probably, contributed to the shortcomings of the Distribution Master Plan, which ISA had agreed to prepare in the context of the San Carlos I project (see Section 1.5), and which, after its submission in 1981, appears to have been practically forgotten. The difference in accounting for distribution from company to company also makes consistent analysis difficult. However, even properly interpreting information from the accounts of a single company is often frought with obstacles. 44. In line with the above limitations, Annex 7.5 sets forth rough figures which show that, in the mid 1970s, EEEB's actual 'distribution investment,' which seemed to have included the cost of at least some transmission facilities, varied between 19% and 442 of total annual investment (including investment in ISA, which is exclusively for generation and high voltage transmission). In the late 1970s and early 1980s, while planned at some 302, it reached in average about 232 . In EPM, the level was planned at about 162 for the first half of the 1970s (no convincing actual figures available), and at 222 to 282 for the period 1980- 83. when it actually reached only 112 to 142 (still including some transmission). It is disputable what the ideal level of distribution investment is, as, inter alia, this level is bound to vary from one utility to the other. However, it seems undisputable that the levels achieved and even some of those planned were substantially too low, a fact that several SARs readily acknowledge, when they convey the Bank's preoccupation about the level of investment in generation persistently staying at 752 of total investment [see e.g. (16) and (17)). 45. At first glance, the inadequate level of distribution investment seems to be reflected in the fact that transmission and distribution 'losses, I instead of declining, have increased in all utilities except EPM, and, it seems, are still increasing, despite the lose reduction programs now, for a decade, associated with Bank lending to the sector, in particular to EEEB. However, as discussed in further detail in Section 4.5, the connection is not that simple, - 179 - as there are three types of Olosses"s (i) technical losses, which seem to have remained about constant, and which can be reduced through additional investment in distribution; (ii) illegal consumption, which seems to have substantially increased, at least in the EEB system, and (III) the unmetered but legal consumption, which appears to be very substantial in BEB. 46. During the period under review, the sector has made only limited progress towards standardization and optimization of distribution facilities. 7.7 Bank Involvement Investment Decision Process 47. The Bank has always been aware of the intricate problem created by the political input into decision making coming through ISA's Board at an early stage of planning generation and transmission and therefore impeding a proper evaluation at the level of DNP and CONPES. It chose not to press for major changes, presumably because such changes are impossible without shaking the very basis on which ISA is built. For a long time, there was a strong justification for this cautious approach. However, as set forth in Section 3.5 in the context of the discussion of ISA's structure and role and of sector regulation, the time may have come, to re-examine this issue and to progress towards more transparent decision making. Investment Planning 48. The Bank has been associated with most of the improvements that materialized in planning sector investment, in particular that for generation and transmission. In connection with the San Carlos I operation, it also sponsored the preparation of a distribution master plan (see para. 50). It has consistently supported ISA in building up and fine-tuning its analytical tools. Unfortunately, until the mid-1980s, it had only limited success, when it encouraged ISA to analyze and discuss thermal developments to the same depth as the utility looked at potential hydroplants. Investment Balance 49. Practically every SAR prepared during the period covered by the present review, emphasizes--usually referring to the 1972 OED Review--the need for increasing the attention given to distribution with the triple objective of providing population not served before with access to electricity, improving the quality of supply, and--last but not least--reducing the losses. Ultimately, seven out of the fifteen operations falling under the present review, contributed to financing distribution. The total cost of the distribution components explicitely identified in SARs amounted to some US$670 million, of which about US$600 million concerned distribution in urban areas and the remaining US$70 million distribution in connection with rural electrification. The Bank earmarked some US$300 million and US$25 million of its own resources to help finance these components. Beyond this, the FY1977 Integrated Rural Development Project, included a small rural electrification component (US$5.4 million, of which the Bank was to finance US$1.2 million). The overall figures mentioned before do - 180 - not include eligible distribution subprojects in the framework of the PEN operation. Neither do they account for the Power Sector Adjustment operation aiming inter alia at Improving distribution but without earmarking specific sub- projects. 50. During the 1970s, only the Guatape II project included a distribution component, which, according to the PCR. benefitted 236,000 new customere.(18) However, the same PCR makes clear that the emphasis of the Bank's involvement was on the generation component. Later in the decade, the Bank deliberately and explicitely concentrated on three main issues: sector planning, organiza- tion, and financing, thus, for the time being, setting aside the distribution issue. (19) The Bank may have been helped in reaching this decision by the fact that, in parallel, IDB was addressing distribution and rural electrification through EEB and ICEL at a quite substantial scale (US$ 47 million in the first half of the 1970s) (see Annex 1.2). From 1980 on, however, the Bank, also helped finance large distribution investments in SEEB and CORELCA, which by the middle of the 1980s added up to the figures set forth in para. 40 above. The WERB projects, though they Included ambitious loss reduction programs, failed to identify the issues related to the legal but unmetered consumption, which represents a substantial part of such "losses.I Evidently this type of consumption, remunerated by a monthly lump sum, presumably meeting only a modest part of the cost to the utility, can be reduced through proper metering. However, it is improbable that the 1980 Bogota Distribution Project, which included only the provision of 48,000 meters was intended to address this isssue at all.(20) The Bogota Distribution II Project, foreseeing the installation of some 300,000 meters, would have been a better vehicle for doing so, but this equipment had also to meet the requirements of 240,000 new connections planned. (21) Neither did the Bank emphasize this area in the Sector Adjustment Loan, which is understand-able, in view of the urgency of other issues it wished to address in priority. However, OD suggests that, now that ISA has a coordinating role in the distribution field, the Bank strongly support all steps likely to lead to standardization and economic optimization of distribution investment. Standardization may also enhance the role of local manufacturing of electric equipment, a development that the earlier OED review also encouraged the Bank to support.(22) 7.8 Agenda for the Future 51. OED suggests that in the context of the future dialogue between Government, the sector, IDB, and the Bank and of possible future lending operations in the power sectors - the Bank help the sector address the shortcomings of the planning and decision making process in generation and transmission Investment through (1) institutional measures (see Chapter 3), (11) improving the comprehensivenes and uniformity of the projects to be integrated in alternative developments, (III) a more realistic scheduling of the projects taking into account that some of the risks Implied in their implementation are likely to materialize, and (iv) sensitivity and risk - 181 * analysis that addresses all the major risks involved and reflects the wide ranges of values that the various parameters can assume with a relatively high probabilitys - Government and the sector, the Bank possibly assisting, envisage to add to their planning tools methods allowing m1ti-objective analysis of development alternatives under various conditions of uncertainty; and - the Bank support ISA In fully assuming its new role in distribution (including standardisation and economic optimisation of equipaent and facilities), such as to allow the sector to strike a rational balance between generation and transmission investment on the one hand and distribution investment on the others this would contribute to reducing losses, an issue that has high priority and that, after the past lack of success, requires a new start (see Chapter 4). ENDNOTES (1) IDBs Evaluscion ex-post del Proyecto Hidroelectrico Chivor I y II, Diciembre 1984, paras 3.03 - 3.14. (2) Evaluaion Ex-Post del Proyecto Hidroelectrico Chivor I y II, pp. 20- 21. (3) SAR Guatape II Hydroelectric Project, December 19, 1972, para. 2.08. (4) Working Papers for Annex 7 of SAR San Carlos I Hydro Power and 500 kV Interconnection Projects, May 17, 1978. (5) See calculation in Lotus Program LEASTCST. (6) SAR Guadalupe IV Hydro Power Project, May 16, 1980, para. 1.32. (7) SAR Salto Grande Multipurpose Project, June 1, 1984, paras. 6.05 and 6.08. (8) Interconexion Ilectrica S.A., Revision del Plan de Expansion, Periodo 1986-1992, Generacion y Transmision, Diciembre de 1982, cuadro 2.3.1.1 (9) Fred C. Schweppe and Byde M. Merrills Multiple Objective Trade-Off Analysis in Power System Planning (paper presented at Ninth Power Systems Computation Conference, Cascais, Portugal, August-September 1987). - 182 - (10) Joint UNDP/World Bank Energy Sector Management Assistance Program. Basis for Formulation of a Colombian Energy Policy, a Report by Consultants. Bogota - Colombia, December 1986, para 6.2.3. (11) Interconexion Electrica S.A., Revision del Plan de Expansion, Periodo 1986-1992, Generacion y Tranamision, Medellin, Diciembre de 1982, para. 8.1.3 . (12) Bank Operations in Colombia, an Evaluation, May 25, 1972, p. 85. (13) Bank Operations in-Colombia, an Evaluation, May 25, 1987, p.87., and Operations Evaluation Reports Colombia, Volume III (Draft 10131171), the Power Sector, p. Exxii, pp.113-124, 164-169, 228-233, (14) Operations Evaluation Reports Colombia; Volume III - the Power Sector (October 1971), inter alia p. xiii and p.xxxv. (15) Bank Operations in Colombia, May 25, 1972, p. 99 (16) SAR Playas Hydro Power Project, February 9, 1981, para. 1.26 (17) SAR San Carlos I Hydro Power and 500 kV Interconnection Projects, May 17, 1978, para. 1.21 . (18) Project Performance Audit Reports Colombia - Guatape II Hydroelectric Project (Loan 874-CO); para. 3.3.1. (19) SAR San Carlos I Hydro Power and 500 kV Interconnection Projects, May 17, 1978, para. 1.13. (20) SAR Bogota Distribution Project, February 22, 1980, para. 4.04. (21) Bogota Distribution II Project, May 28, 1985, paras. 4.03 and 4.04. (22) Operations Evaluation Reports Colombia; Volume III - the Power Sector, October 31, 1971, pp.330 and 331. - 183 - VIII. LINKAGES BETWEEN THE POWER SECTOR AND THE ECONOMY 8.1 Introduction 1. Between 1977 and 1981, at the time decisions were being made to proceed with the power sector investment program for the 1980s. there was not much focus on what the consequences of these investments and their financing would be on the macroeconomy, irrespective of the scenario envisaged, whether business-as- usual or shocks (such as demand collapse, accelerated devaluation or unsus- tainable price policies). The only aspect of power sector/macro-economy linkage that was given prominence was the consequence of an investment shortfall, namely. power shortages--and the effects of this on the economy. This was normal for power sector planning in most countries. However, now that the major develop- ments in the sector have not proceeded as originally foreseen (construction times, exchange rates, load forecasts, power prices, etc.), with benefit of hindsight, the linkages between the power sector's financial health and the rest of the economy have become more apparent. 8.2 The Two Types of Power Sector-Macroeconomy Linkages Direct Links 2. These reflect direct relationships between the power sector and the other parts of the economy. The most immediate ares the electricity require- ments of the economy; the power sector's needs for capital (usually very large) and labor (usually very small); quite direct and, in wlombia, signifi-cantly negative correlation between power sector finances and the Government's fiscal position. The large capital requirements of the sector reflect a high dependence on capital goods and services imports, which, in turn, has made the sector account for an increasing share of the public sector's total outstand-ing external debt and annual debt service payments. This, in turn, contributed to worsening the balance-of-payments (B-O-P), given the currently non-tradeable character of electricity in Colombia and the dominantly residential character of consumption. In terms of backward linkages to the economy, the power sector possesses limited scope since capital goods manufacture, such as turbines, is usually unlikely to be competitive except in large LDC markets. Indirect Links 3. Indirect B-O-P impacts arise through the decreased availability of foreign exchange for imports of non-power sector goods and services due to the foreign exchange needs of the power sector. There are also the indirect effects of power sector policies on public sector credit requirements, on money supply, and the general supply of finance to meet the needs of the non-power sector. In other words, the *crowding out* effects on external and domestic credit access by the non-power sector parts of the economy--some of which (e.g. manufacturing and agriculture) are the lead sectors of the economy upon which economic growth largely depend. - 184 - 8.3 The Power Sector and Economic Growth 4. In the past several years. the question of the impact of various types of power sector investment on economic growth has acquired increased relevance in several countries. In shedding some light on this issue it is important to contrast the roles of the power and petroleum sectors. In the first place, since electricity is not a final product but a factor of produc-tion, it may be argued that its share of GDP is not a useful measure of i,.s contribution to the economy. However, there is still merit in looking at this ratio, especially when viewed in conjunction with that of the petroleum sector--the output of which is another intermediate energy product, albeit tradeable. In the case of Colombia, as in most LDCs, the power sector's share of GDP has been, and is, small, contributing very little to value added. The same observation is valid for the petroleum sector in Colombia between 1975 and 1986 during which period the country imported about 1OZ-15Z of its oil needs. Table 8.1 shows these sectors' contributions to GDP over the period. From this it is clear that both sectors have increased their shares of GDP about threefold since 1975. But as these bases were then very low, by 1986 the sectors still only contributed between 2Z (power) and 3 (petroleum) to GDP. Since the forward and backward linkages of both sectors to the economy are limited, the impact of power investment on economic growth is small. Indeed, there is no capital goods manufacturing sector making power equipment. Thus, about 60% of the sector's needs for goods and services have to be imported. This largely eliminates the multiplier effects that would be expected in the economy from investments of the magnitude that characterize those of the power sector. Table 8.1s Power and Petroleum Sector Shares of GDP, 1975-86 (Percent) Power Petroleum Sector Sector Year Share Share 1975 0.7 1.0 1978 1.0 0.9 1980 1.1 1.6 1983 1.6 2.2 1986 2.0 2.9 5. The large import requirements for the sector has led, in the past decade, to the sector accounting for an increasing share of the country's and of its public sector's total outstanding external debt, as shown in Table 8.2. The power sector share of total public and publicly guaranteed outstanding external debt, while remaining about 115 throughout the 1970s, increased very rapidly in the 1980s to about 113 by 1987. This very rapid rise is likely to have limited private sector access to long- and medium-term external credit. Though only a limited part of external resources used by the power sector is fungible and possibly available to other sectors, the *crowding out' is likely to have negatively affected economic growth. - 185 - Table 8.21 Power, Public, and Private Sector Outstanding External (Medium and Long TerW) Debts, 1970-87 (Us$ Millions) Public and Ratio of Powerl Power Private Publicly Public and Publicly Year Sector Sector Guaranteed Guaranteed (non- (M) guaranteed) 1970 245 280 1.330 18.6 1975 460 375 2,470 18.6 1980 860 525 4,180 20.6 1982 1,525 1,215 6,080 25.1 1986 3,515 1,640 11,750 30.0 1987 3,975 n.a. 12,425 32.0 6. The increasing level of the sector's external debt has led to escalating debt service payments, which, when combined with the non-tradeable character of the sector output, results in a direct and increasingly negative impact on the balance-of-payments. Table 8.3 shows that, for example, Interest payments on the power sector's external debt, in terms of total interest payments on the public sector's external debt, has trebled between 1980 and 1987s Table 8.3s Share of Power Sector to Public Sector Interest Payments on Their External Debts (Percent) 1980 1982 1983 1985 1986 1987 12 16 22 28 33 37 7. The fourth area in which the power sector's development since the late 1970s has possibly had an unfavorable impact on economic growth has been the large portion of public sector investment resources absorbed by the power sector (see Table 8.4). Naturally, as part of the essential effort to build Colombia's economic and physical infrastructure, there was a continued increase in the share of public sector investment devoted to power, from about 7% in the mid-1960s to some 242 in the mid-1970s, and on to around 34? by the mid-1980s. What is of relevance, when viewed from an economic growth perspective, is the high share of public investment resources allocated to power since the mid-1970s, given that the latter's capital requirements relative to output are very high compared to the productive sectors (e.g. manufacturing and agriculture) or even to another energy sector--petroleum. Indeed, Table 8.5 shows that the values of the -186- 1acremental capital output ratio (ICOR) for the power sector have been significantly higher than those of the petroleum sector throughout the 1976- 1986 period even when Colombia was a net oil Importer. Table 8.41 Power and Public Sector Investments 1966-86 (Percent) Ratio of Power/Pablic Year Sector Investments 1966 7.3 1968 8.2 1975 23.9 1980 30.0 1983 33.0 1986 34.0 Table 8.5: Estimates of Lagged Incremental Capital Output Ratios (ICOR) 1976-1987 Year Power Sector Petroleum Sector 1976 16.0 1.8 1980 27.3 0.8 1983 14.0 2.1 1986 8.5 n.a. 1987 n.a. 1.5 8. 'What Table 8.5 also highlights is that measured in terms of value added output, in Colombia, investment requirements for power have tradi-tionally been at least 10 times higher than those for petroleum. Further, the two sectors have radically different fiscal and B-0-P impacts. Indeed, the poor financial record of the power sector in Colombia contributed to weakening the broader economy and, in particular, increased the public ector's fiscal deficit. This is the fifth area in which power sector developments since the ;ate 1970s have unfavorably affected growth. The cash deficit of this sector remained at about 202 of the consolidated public sector deficit between 1981 and 1984 and rose to about 113 by 1985. This means that in 1985. the sector's deficit was some 1.2Z of GDP (US$400 million). In other words, the power sector was a drag on the economy from the fiscal standpoint. In contrast, the contribution of the petroleum sector to the government's fiscal position has been large and positive even before oil exports were resumed in 1986. For example, in 1985, the year in which oil production roughly equalled domestic demand, the petroleum settor contributed in excess of US$50d million to central and regional government revenues through - 187 - royalties on production and taxes on consumption. This highlights that the growth impulses of the petro-leum sector are transmitted to the rest of the economy through the sector's surplus revenue, benefiting its equity owners and public sector finances. The power sector does not possess these atributes, only in part due to the absence of significant economic rents. Perhaps of greater relevance has been a power pricing policy that has steered away from either the sector making any direct positive contribution to the government's fiscal position or from even providing appropriate return on investments in the sector to its public sector owners. Conclusions 9. The macroeconomic difficulties that arose in the early 1980s, of course, were driven by factors primarily outside of the power sector (see Sections 1.7 and 1.8). In particular, the collapse of coffee prices, increased petroleum imports at higher prices, declining manufactured exports, an appreciating peso, and -ising real interest rates. In fact, the diffi-culties in the power sector essentially aggravated an already worsening macroeconomic situation. The Bank, however, stepped in with very effec ive support for the Government's macro- economic adjustment program, beginning in 1984, through which all the major stabilization objectives were achieved. However, the consequences of the macro adjustment measures on the power sector were severe. Most relevant was the sizeable peso devaluation in 1985186 (followed by the unforeseen collapse of the US dollar against the strong currencies), which significantly increased the sector's debt servicing costs, while, on the other hand, tariff levels were restrained by Government from off-setting the costs of the devaluation of the sector. 10. In essence, what emerges, in hindsight, is that though the power sector cannot be considered the prime culprit for the macroeconomic diffi-culties, a heightened awareness of the effects of power sector developments on the macroeconomy is called for. This is especially the case when the power sector investment program rises to the levels seen in the 1980s in terms of its share of public sector investment and external indebtedness and, on the other hand, fails to mobilize adequate savings thereby aggravating public sector fiscal deficits. This is perhaps the most important lesson to be learned, namely, that it is essential that power sector investment planning capture better the critical linkages and impacts its investment sequences, and their allied financial and pricing strategies, can have on the macro economy. The multi-objective planning advocated in Section 7.3 could lead a long way along these lines. 8.4 Performance of the Bank 11. From the standpoint of assisting the Government in its successful economic stabilization program in 1984/85, the Bank's performance deserves to be commended. Of course, the effects of this program on the power sector, in turn, led to the need for the sector to itself undergo an adjustment process, which the Bank supported through the Sector Adjustment Loan in 1987. This * 188 - addressed important new initiativess institutionally (the Energy Board), in tariffs (the linkage to average incremental costs) and in a reduced investment program. O1D's main reservation concerns the modest rate of tariff adjustment called for (see Chapter V) in light of the sector's financial problems. NI
Groupe de la Banque mondiale · IEG Evaluation
Colombia - The power sector and the World Bank, 1979-1987 (Vol. 2 of 3) : Technical report
Voir le document original
Le texte intégral est hébergé par l’organisation qui le publie. lawenc.com indexe les métadonnées et renvoie vers la source officielle.
Texte intégral
Informations clés
Organisation
Groupe de la Banque mondiale
Type de document
IEG Evaluation
Pays
Colombie
Source
Banque mondiale