Tunisia - Economic development (Vol. 5 of 6) : A long term projection models

FILE Cop Report No. 274-TUN RETURIto The Economic Development REPORTS DESK of Tunisia WNE wEI Volume IV: Annex A Long Term Projection Model December 27, 1974 EMENA Region Country Programs Department II Not for Public Use U~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Document of the International Bank for Reconstruction and Development Interriational Development Association. This report was prepared for official use only by the Bank Group. It may not be published, quoted or cited without Bank Group authorization. The Bank Group does not accept responsibility for the accuracy or completeness of the report.  CURRENCY EQUIVALENTS Currency Unit = Dinar = 1000 millimes With effect from 1955 US $1.00 - 0.42 Dinar Dinar 1.00 = US $2.381 With effect from September 28, 1964 US $1.00 - 0.52 Dinar Dinar 1.00 - US $1.90 With effect from December 20, 1971 US $1.00 - o.48 Dinar Dinar 1.00 - US $2.08 With effect from February 1973 US $1.00 - 0.44 Dinar Dinar 1.00 - US $2.27 UNITS AND WEIGHTS AND MEASURES: METRIC BritishAJ.S. Equivalents 1 m = 3.28 ft. 1 m ton 0.981 g. ton = 1.1 US sh. ton 1 m2 = 10.76 sq. ft. 1 kg 2.2 lb. 1 in = 0.62 mi. 1 litre = 0.22 gal. 1 m2 = 0.386 sq. mi. = 0.26 US liq.gallon 1 hectare = 2.5 acres 1 m3 = 1.31 cubic yards  ECONOMIC DEVELOPMENT OF TUNISIA TABLE OF CONTENTS ANNEX VOLUME IV A LONG TERM PROJECTION MIODEL Page No. A. Introduction ., .. 1 B. The Model in General ...... .............. 1 C. Sector Output and Investment . ................. 2 D. Employment and Populat:Lon ..... ............. 9 E. Exports ........................................... 12 F. Imports ........................................... 15 G. Consumption and Saving ...... ...................... 18 H. Balance of Payments ...... ......................... 20 I. -Capital Inflow ansd Reserve Changes ................ 21 J. Money and Prices ................... ................ 22 K. Symbols ............................. ......... . .. . 24 L. Model Equations ................................... 27 M. Projection Results ............... .. ............... 33   A LONG TERM PROJECTION MODEL A. Introduction 1. The model described in this Appendix was constructed in order to capture the essential nature of the Tunisian economy in a generalized series of equations, and use this format for making projections under alternative circumstances. The model is a cor.bination of behavioral equations based on past relationslhips and estimated witlh econometric techniques, and other data and parameters gathered by the bank's mission or based on their judgment of likely future events. In addition, the model also incorporates information furnished by the Tunisian Government, or as estimated in their Fourth Plan. 2. The model was initially constructed and calibrated during early 1973, and used at that time to make comparisons with the Fourth Plan. In the short space of one year, thie rapid changes in the international economy have made both the Plan and the model out of date. Consequently, the model now con- tains new assumptions which pertain to the world situation as the Bank saw it at the end of 1974. It is interesting to note, however, that despite the large and fundamental changes in many of the variables between the two sets of projections, the basic results and lessons to be drawn from the model are remarkedly similar under either set of circumstances. It should also be noted that the projections with the new assumptions also update 1973 to make it agree as much as possible with the prelininary actual data now available for that year. B. The Model in General 3. The model is essentially a multi-sector growth model in which over- all growth depends upon individual sector assumptions concerning growth rates and investment decisions. These assumptions are in turn based on either Tunisian Plan or Bank mission estimates of likely production levels for those sectors whose outputs are not dependent on the usual production function or demand limitations. Output for the pre-determined or "exogenous" sectors generally reflects either physical limitations for a sector whose principal product can be exported without distorsions to world prices (olive oil, phosphates, petroleum), or government policy decisions (agriculture), or external demand factors (tourism). Together, they constitute about 25 per- cent of gross domestic product. A second group consists of service sectors whose output responds to total GDP or its final demand components (investment, consumption), and is labelled "endogenous". A third and smaller group com- prises those sectors constrainecl by physical capacity. Output in these sectors will depend upon investments or additions to capacity. These invest- ments are assumed, however, to reflect government policy decisions or private sector decisions. These sectors are consequently, called "quasi-endogenous", since their outputs depend upon decisions outside the model. In the table below the 13 sectors are summarized under the three headings: exogenous, endogenous and quasi-endogenous.  Sectors by Type Endogenous Quasi-Endogenous Exogenous Energy Textiles Agriculture (non-olive) Food Processing Chemicals Agriculture - olives Construction Other Manufacturing Mining Transport and Communications Petroleum Government Tourism Other Services While the number and size of the exogenous sectors is fairly large, it must be remembered that four of the five are export-oriented. Unlike many models in which exports are completely exogenous, this model derives its export pro- jections for the principal exports from the sector outputs adjusted for do- mestic consumption (phosphates, oil, olive oil, manufactured goods). In this way, an increase in one of the major export categories automatically implies an increase in GDP, consumption, and imports, thus producing a more realistic estimate of the impact of faster export growth on the balance of payments. 4. For the endogenous sectors, investment is functionally related to output, while for the quasi-endogenous and exogenous sectors investment is exogenous, but consistent with sector outputs. Imports are disaggregated by end-use classes and related either to the components of final demand (investment, consumption) or to sector outputs. Private consumption is es- timated using a consumption function, while government consumption is based on the growth of the government services sector in GDP. The model differs from the typical two-gap model in that the resource gap between imports and exports is reconciled with the savings-investment gap by allowing the adjust- ment to fall on changes in stocks (i.e., the level of total investment). The remaining elements of the balance of payments (workers remittances, direct foreign investment, foreign assistance) are largely exogenous estimates, and the reconciliation between the capital and current accounts is achieved by changes in reserves. C. Sector Output and Investment 5. For the endogenous sectors, demand-determining variables were iden- tified and regressions run on the 1960-72 data in constant prices to estimate the relationships in terms of value added in constant 1966 dinars. While it would have been more appropriate to use gross output, in the manner of an input-output table, the value added series is assumed to be a suitable proxy for gross output. These endogenous sectors are: energy (excluding petroleum), food processing, construction, transport and communications, government, and "other services", and made up about 65 percent of GDP in 1972. Energy is related to value added in manufacturing, while food processing is related to private consumption and value added in olive production (since a large part of food processing involves olive oil production). Construction is related to fixed investment, transport-communications is related to total GDP at factor cost, and other services to private consumption. Government services are a function of government current revenues, with a one-year lag.  - 3 - YENER = -2.59 + .243 TYAN R = .914 DW = 1.16 (11.34) YFOODP = 3.156 + .031 PC + .436 YAGOL R = .937 DW = 1.61 (3.64) (53.76) (1962-1972) -2 YGOV = 8.60 + .5098 CURREVt_1 R = .965 DW = 1.84 (1962-72) YCONST = .159 I1 + .470 I2 -+ .202 13 R = .985 DW = 1.71 (4.65) (18.28) (3.83) YTRC = -5.20 + .0921 GDPFI2 R = .903 DW = 1.63 (10.61) -2 YSEROT = 28.84 + .3267 PC R = .982 DW = 2.73 (24.54) where Y = value added, constant 1966 dinars, millions ENER = energy, water MAN = manufacturing FOODP = food processing PC = private consumption expenditures YAGOL = agricultural production - olives CONST = construction SEROT = other services TRC = transport and commanications GDPFC = GDP factor cost I1 = fixed investment in manufacturing, petroleum, energy and mining sectors 12 = fixed investment in government, agriculture and other services sectors I3 = fixed investment in tourism, transport and communications GOV = government services CURREV = government current revenues -2 All of the equations show significant coefficients and high R . The equa- tion for food processing is estimated over the 1962-72 period, since the data on value.added in olive production does not go back beyond 1962. 6. Since investment in each sector will have a different composition in terms of capital equipment, construction, or other services, the value added in construction was related to fixed investment disaggregated into three broad groups. The sectoral investments were allocated to these groups on the basis of probable levels of imported capital goods; I1 is assumed to have a high import component (and therefore a low construction component), I2 is assumed to have a low import component, and I3 is assumed to be an in- termediate case. The three categories were devised on a trial and error basis with regards to the equation for the imports of capital goods (described below), but were found to be equally useful in estimating construction value added. Since the constant term in the original equation was extremely low (.8) and not significant, the equation was reestimated without the constant term.  7. Investment in these sectors is based on using the present growth rate of value added as an assumed indicator of future expected growth in the next year. The expected increment to output is the product of this year's output and this year's growth rate. When multiplied by an assumed incre- mental capital-output ratio, this produces the sector investment this year. This implies a rather simple accelerator function based on expected output, a one-year lag, and output expectations of unit elasticity. In the absence of data on capital stock by sector, or existing capacities, this method is the best available. In equation form, the investment function is: i =ICOR .Yi (Y it -yit-)/it-1 Since past investments in Tunisia have often been unproductive, marginally productive, or have had an abnormally long gestation period, fitting invest- ment equations to the past data an unrewarding exercise. The derivation of the sector ICOR's is explained below. 8. This leaves eight sectors in the exogenous or quasi-exogenous category: olive and non-olive agriculture, mining, petroleum, textiles, chemicals, other manufacturing and tourism. 9. The growth rate of non-olive agriculture is assumed not to depend on investment, but reflect a policy parameter. The growth of this sector is projected to be somewhat higher than in the past decade, when adverse weather conditions lowered the overall growth rate to about 1 percent per annum (1961- 70). For the low growth case, which implies a continuation of conditions as they are with no major changes in government policies, and reasonably good weather conditions, the model assumes a growth rate of 4percent for the years 1974-76 and 3.5 percent thereafter. The high growth case assumes a more optimal set of policies which can achieve a higher production level. This is translated into a growth rate of 5 percent for the years after 1974. Investment in agriculture is set equal to the estimates of the Fourth Plan for the years through 1976, and then assumed to grow at a rate of 3 percent thereafter under all assumptions. 10. For olives, we base our projected output for the period 1972-1976 on the projections of the Ministry of Planning. This in turn was based on an analysis of the aging of the existing tree stock. Since a large portion of the existing trees are only now coming into full maturity, a "normal" level of about 500-600 thousand tons of olives a year can be assumed (versus 332 tons average during 1962-71). Starting from an estimate of 350 thousand tons for 1973, output is projected to rise to about either 530 (low) or 580 (high estimate) by 1976. The estimate of the Office d'luile is that the maturation process of the existing trees should increase production at 4.5 percent per annum, for the period after 1976. This assumes, however, no change in culti- vation practices, etc., which could raise the growth rate by as much as another two percentage points. The "high" estimate is therefore set at 6.0 percent per year.  -5- 11. In manufacturing, we have three sectors whose growth is based on exogenously determined investment levels: textiles, chemicals and other manu- facturing. Output in these sectors is assumed to be supply constrained, or limited by capacity rather than demand factors. Incremental output is based on lagged investment and a given ICOR for the sector. The general form of the output equation used for these sectors is as follows (where Yt is sectoral value added, and I is sectoraL gross investment): Y = 1/ICOR (a-1It1 + a2It-2 + a+31t3) t-1 The "a" coefficients represent investment lags which have been taken from the PST 81 Model developed by the I[nstitut d'Economie Quantitative in Tunis. 1/ The actual lags used for the three sectors are as follows: Investment Lags (Percent) a1 a2 3 Textiles 60 30 10 Chemicals 42 42 16 Manufacturing, Other 49 37 14 The investment levels for these three sectors represent a major policy tool for the Government, both direct:ly in government investment in public enter- prises and indirectly in policies aimed at encouragement of private sector production, including private direct investment from abroad. The exogenous estimates, therefore, for manufacturing investment are the Bank's estimates of what is likely to occur under different circumstances. With the advent of increased export earnings and foreign exchange reserves, the investments for the period to 1976 is slightly higher than the Plan estimates. These projections, which are broken into chemicals, textiles and other manufac- turing investments in the model, are summarize in Table 1. Table 1: INVESTMENT IN MANUFACTURING (Mlillions 1966 dinars) Run 1973 1974 1975 1976 1977 1978 1980 1985 1 35.8 41.5 36.8 41.8 49.2 55.5 67.0 89.0 2 35.8 41.5 40.0 48.5 59.0 67.5 83.5 116.9 1/ Used with the permission of the Institute's director M. Hamza.  - 6 - 12. For petroleum, the availability of known deposits makes medium- term forecasts fairly firm. Longer term prospects are somewhat less certain. Two assumptions for oil production, in terms of tons of crude oil, were developed. They are based largely on the revised Plan projections for the period through 1976, with extensions to 1985 based on assessments of possible future discoveries and the rate of depletion of existing fields. The petroleum sector is defined here to include oil refining, and since Tunisia still imports some crude oil for its refineries, the ratio of value added to gross domestic output in 1972 was 1.04. In addition, the sector includes gas production. Imports of crude oil and domestic consumption are relatively small and based on exogenous estimates (see Table 2). The assumption is also made that no new refineries will be constructed after 1976, and that the effect of the new refinery at Gabes is to raise the value added ratio in petroleum production to 1.10. Investment plans are taken from the Plan, extended to 1985 on the basis of an assumed D 7 million per annum expenditure, largely for exploration. 13. Future productions depends on the success of future exploration and the possibility of secondary recovery from existing fields. The figures in Table 2 reflect a high, low estimate for production and exports. In the low estimate, production rises from 3.9 million tons in 1973 to 5.5 million tons in 1980, while in the high estimate it rises to 7.5 million tons. The corresponding level of exports is 4.2 and 6.0 million tons, with the increase in consumption after 1980 coming from imports. Table 2: CRUDE PETROLEUMA AND PRODUCTS Production, Consumption, Imports and Exports (volume in million tons) Run 1973 1974 1975 1976 1977 1978 1980 1985 Production 1 3.9 3.9 4.3 4.5 4.9 5.2 5.5 5.5 2 3.9 4.1 4.8 6.0 6.6 7.1 7.4 7.5 Import 1 1.0 .7 .8 .8 .8 .8 .8 1.6 2 1.0 .7 .8 .8 .8 .8 .8 1.8 Domestic Consumption 1 1.2 1.3 1.4 1.5 1.6 1.8 2.0 2.9 2 1.2 1.3 1.4 1.6 1.7 1.9 2.2 3.2 Export 1 3.8 3.3 3.7 3.7 4.0 4.2 4.2 4.2 2 3.8 3.5 4.1 5.2 5.7 6.0 6.0 6.0  -7- 14. The prospects of the mining sector revolve around the production estimates of the phosphate mines, which constitute about two-thirds of value added in the mining sector. The growth rate of value added in the sector is based on an assumed constant grorth rate for other minerals, and the influence of two different projections for phosphates. These are based on the Bank's analysis of the GAFSA project, which involves a substantial renovation and ex- pansion in production. The production level for phosphates is estimated to reach about 6.0 to 6.5 million tons in 1981, followed by an annual growth rate of 2 percent (see Table 3). Investment in the mining sector is basically the same in all runs, with some mIinor modifications made in certain years to reflect revised estimates of investments from the GAFSA project. Table 3: MINING SECTOR PROJECTIONS Run 1973 1974 1975 1976 1977 1978 1980 1982 1985 Value Added Mining (Growth Rates) 1 2.6 9.3 5.3 12.5 11.1 4.0 9.3 2.0 2.0 2 2.6 18.0 9.8 13.3 9.8 3.6 8.0 2.0 2.0 Phosphate Production (Million Tons) 1 3.5 3.8 4.0 4.5 5.0 5.2 5.9 6.1 6.5 2 3.5 4.1 4.5 5.1 5.6 5.8 6.5 6.8 7.2 Investment, Mining Sector 1 4.5 9.7 15.7 3.0 5.0 3.0 3.0 3.0 3.0 2 4.5 9.7 18.0 6.0 5.0 3.0 3.0 3.0 3.0 15. The tourism sector was initially projected along the lines suggested by the Fourth Plan. This meant that since hotel capacity would rise rapidly from 48,000 beds in 1972 to 85,000 beds in 1976, and assuming that utiliza- tion and tourist expenditures per day remained constant at 1972 rates, there would be a very rapid expansion in the sector over the Plan period. This expansion was continued through 1985 at the rate of 10 percent per annum. The impact of the energy crisis, including the lower growth rates of GNP pres- ently foreseen for Western Europe, has profoundly changed the outlook for this sector. Instead of the 16 percerLt growth rate experienced in 1972, growth in 1973 and 1974 appears to have been negative. After this period of stagnation, an eventual return to the previously expected long-term growth rate of 10 percent per annum is projected; beginning in 1975, tourism earnings in the balance of payments are projected on the same basis. Investment in tourism is estimated based on recent information provided by the Tunisian government.  -8- 16. Total gross domestic product at factor cost in constant prices is simply the sum of the 13 sectors whose value added has been calculated from the above equations. Indirect taxes and subsidies are taken to be 88 percent of total indirect taxes and, when added to GDP at factor cost, produce GDP at market prices. Likewise, total fixed investment is determined by adding up the sectoral investments. 17. The ICOR's for each sector, where ICOR's are used, are based on a comparison of the implicit ICOR's in the Plan with historical experience. It is interesting to note that, with the increased growth in recent years, the sectoral ICOR's in Tunisia have fallen to a point where they are more in line with world-wide levels. The historical ICORs were calculated using five- year moving averages and assuming -a one-year lag between investment and out- put. 1/ The implicit ICORs from the Plan were calculated the same way, based on the terminal year 1976, and therefore include only investments made or to be made for the years 1972 to 1975. Table 4 shows the results by sector for those ICORs calculated for the years 1968 to 1972, the Plan ICORs, and the ICORs used in the model (where appropriate). Table 4: ICORS BY SECTOR 1968 1969 1970 1971 1972 Plan Model Agriculture -2;.9 -116.3 -17.0 5.0 1.7 2.1 * Mining 5.3 180.4 -25.2 86.3 5.5 24.5 * Petroleum 3.0 3.2 2.9 3.3 5.2 40.3 * Energy 5.7 7.1 6.0 6.6 8.2 13.3 8.0 Manufacturing, Total 5.2 5.6 5.8 3.8 2.4 3.4 * Food Processing 5.3 -6.8 -4.2 3.5 .9 3.8 3.0 Chemicals 1.8 2.4 2.7 2.7 4.0 2.9 3.0 Textiles 6.9 3.9 4.0 3.2 2.6 2.0 2.0 Manufacturing, Other 5.6 5.4 4.5 4.7 4.1 3.6 3.5 Transport, Communications 4.3 5.6 3.4 7.6 4.8 11.5 6.0 Tourism 3.8 4.0 4.5 3.4 3.2 4.6 * Other Services 6.0 4.3 3.4 2.5 1.7 2.6 3.0 TOTAL 5.6 5.0 4.3 3.6 2.3 4.1 * * Not relevant to model as inputs. 1/ To be precise, the equation for the ICOR for any sector i in period t is: i i i i i i ICOR I + I +1I +I + I t t-Il t-2 t-3 t-4 t-5 Yt - yt-5 Sectoral investment figures are taken from the publications of the :Iinistry of Planning, in current prices, and deflated to constant 1966 dinars by using the overall implicit deflator for fixed investment.  - 9 - 18. As can be seen in the above table, the sectoral ICORs from the Plan are roughly in line with the historical averages. The ICORs for agri- culture and food processing are not very useful, since outputs of agricultural products depend on weather and past investments (such as in irrigation) have not been successful in overcoming the effects of drought. The ICOR in mining is extremely high, since new gross investments are necessary merely to main- tain present production levels. The high ICOR for petroleum is a result of combining oil production and oil refining in one sector, and of oil produc- tion declining during a period when further investments are being made in refineries. The low projected ICOR for textiles reflects the fact that new investments in this sector are planned to be largely in the manufacture of clothing, which is less capital intensive than the weaving and spinning in- dustries. 19. For the model we have generally accepted the Plan implicit ICOR's, but modified them when they have differed markedly from historical experience. Since many investments to be made during the Plan will not come into full pro- duction until after the Plan period, downward revision in certain cases appears to be realistic. Basically, energy was dropped from 13.3 to 8.0 percent, and transport and communications was dropped from 11.5 to 6.0 percent. There is also some question whether the heavy planned investments in these sectors can be realized. On the other hand, other services was raised from 2.6 to 3.0 percent reflecting a judgment that Plan projections may be slightly optimistic. In order to prevent wild swings and negative investment levels, the investments in services were smoothed with the use of a three-year moving average, and constrained to be zero if the initial estimate proved to be negative. Invest- ment by the central government (for infrastructure) is based on Plan projections for the 1973-76 period. After 1976, government investment is assumed to grow at the same rate as government value added. After some initial experimenta- tion with the ICOR approach for the food processing sector, an exogenous es- timate was substituted in its place. D. Employment and Population 20. Employment is projected on a sectoral basis using fixed incremental labor-output ratios. While this method is somewhat crude, it allows the es- timates of employment in 1972 in the draft Plan to be used as a base. These estimates are, unfortunately, not available on the same basis for preceding years, and those statistics that are available differ in their definitions of employment. Consequently, it wouLd not be useful to estimate equations using the past data, since it would produce estimates not compaiable with those in the Plan. In addition, a substant:ial part of the past series on employment is itself only an estimate derivel from production statistics using fixed coefficients. 21. The incremental labor output ratios (ILOR's) are derived by taking the incremental employment (in thousands) projected in the Plan, and dividing them by the increment of value added in constant prices (in millions of dinars) also found in the Plan (1972-1976). The result is an implicit ILOR,  - 10 - which is based on the work of the various ministries in evaluating the em- ployment component of specific projects. The results of these calculations are shown in the table below, along with the average labor output ratio (L/Y) for the year 1972. Table 5: AVERAGE AND INCREMENTAL LABOR OUTPUT RATIOS /1 1972-1976 L/Y ILOR Sector (1972) (1972-76) Agriculture 5.38 0.0 Mining 1.61 -2.0 Energy (including Petroleum) .15 .09 Food Processing .97 .40 Textiles 3.69 1.54 Chemicals .94 .11 Manufacturing, Other 1.50 .45 Manufacturing, Total 1.59 .91 Construction .91 .64 Transport, Communications .63 .30 Tourism .65 .65 Services, Other .71 .39 Government 1.18 .84 TOTAL 2.09 .62 /1 These ratios exclude artisan employment to the extent that the value added by this employment is not reflected in the national accounts. The total ratios are calculated with respect to GDP at factor cost. The ILOR for food processing is calculated on the basis of 1973, in- stead of 1972, since the latter is a distorted year because of the record olive harvest and olive oil production. In general, the model has used the above ILOR's to project employment, except that the sub-total ILOR for manufacturing, and the total ILOR for the economy are not utilized, since employment for both of these is based on the sum of the disaggregated components. For agriculture, we have not used the implicit Plan ILOR of 0.0, but substituted a value of 2.5. The Plan assumes that no new employment opportunities will be generated in agriculture, which seems rather extreme. It is difficult to know what an appropriate ILOR is, however, since a substantial amount of underemployment is present in the 1972 level of 800,000 employed in agriculture. In other words, a substantial increase in agricultural production would probably be possible without increasing em- ployment, merely by using the present labor force more efficiently.  - 11 - 22. Of the 800,000 supposedly employed in agriculture, we know from the Plan that only 400,000 are fully employed males, while 150,000 are male seasonal workers and 250,000 are women. If we assume that the latter are also seasonal or part time workers, this means there are 400,000 part time and/or underemployed workers. Assuming that these work only one-third the time of the fully employed, then the full-time equivalent of the 800,000 is about 530,000. Relating this figure to the value added in agriculture in 1972, the average labor-output ratio is 3.56. It can be assumed that the incremental ratio will not be as high, since the introduction of new seeds, expanded irrigation, mechanization and other improved agricultural practices will increase the productivity of those already employed in agriculture. There is no sure way of estimating the ILOR for agriculture in the future, but it will certainly be less than 3.56. Somewhat arbitrarily, a rate of 2.5 has been used for projections, with a base of 530,000 employed in 1972. The 530,000 figure for agriculture has the effect of reducing 1972 employment from the Plan's estimate of 1,385,000 to 1;115,000. This recasts the employment estimates into an approximation of full time equivalents, which is more readily related to output. 1/ It also reduces the overall labor-output ratio from 2.09 to 1.48. Since there is a large element of redundant employment in the mining sector, which will be part:ially eliminated during the Plan period, an exogenous estimate of mining employment which is not related to mining output has been used. It is assumed that the level of employment drops from 19 thous- and in 1972 to 16 thousand by 1976, and remains at that level through 1985. 23. The above table shows a rather interesting aspect of the Plan. The average labor-output ratios are substantially higher than the incremental ratios, implying that new investments have a much higher labor productivity than existing projects. Overall, the L/Y is 2.09, while the ILOR is .62. It is also known that the incremental capital output ratios are equal to, or perhaps slightly higher than, the past ratios, so that the implicit in- cremental capital-labor ratios also must be higher than the average capital- labor ratios. Why the incremental cost of generating new jobs in terms of investment and output should be so much higher in the Plan than the average in the past is difficult to explain fully. In general, new investments tend to bring new technologies, which tend to raise capital intensities, while the expansion of production with present technologies may be difficult or im- possible. Nevertheless, the implicit ILOR's in the Plan may not be as high as they could be if greater emphasis was given to maximizing the employment component of new investments. 24. Population growth is assumed to be 2.8 percent per annum through 1976, and then gradually decreasing to 2.4 percent by 1980, and constant at that rate through 1985. Labor force participation rates are assumed to rise 1/ The 80,000 seasonal workers outside of agriculture are ignored and counted as full time workers. While this produces a mild distortion in the fi- gures, it is not a major problem since their numbers are small. An ad- justment would, moreover, be very difficult since the industry alloca- tion is not known.  - 12 - from 28.9 percent in 1972 to 29.3 percent in 1976, due to the changing age structure of the population. After 1976, a gradual decline is assumed to a level of 29.0 percent, which remains constant for the period from 1980 onwards. The estimate of the labor force obtained from the product of the participation rate and the population estimate is further reduced by the amount of projected migration during the projection period. The downturn in expected growth in Western Europe, combined with the prospect of more stringent controls on the immigration of foreign workers has led the Bank to lower its forecast con- siderably. 25. Unemployment is simply the difference between the projected employ- ment and the projected labor force. Since employment figures have been ex- pressed in full time equivalents, the projected unemployment is not the actual number of people without any type of employment, but the unutilized potential of the labor force. Likewise, the calculated unemployment rate is equivalent to a rate of unutilized capacity of the labor force. E. Exports 26. Commodity exports have been disaggregated into their principal com- ponents, and are largely based on projected sector outputs. It is assumed that Tunisian output is generally unaffected by the world supply price for its products, and that world prices are generally unaffected by Tunisian out- put. Since Tunisia does not have a dominant share in any of its export mar- kets, these assumptions are not unrealistic. 27. The exports of olive oil are based on the projections of olive pro- duction (see above) and an estimate of the extraction rate and of oil retained for domestic consumption. The olive oil extraction rate in the past has ranged from 18 to 22 percent (larger crops tend to have less oil and more water). A conservative 20 percent average for the entire period has been assumed, al- though there are indications 1/ that it could eventually go as high as 25 oercent. 28. Not all olive oil is exported, and the amount retained for domes- tic consumption appears to vary directly with the size of the crop. For our projections we have assumed a domestic retention of 25 percent of production, whiclh is equal to the average rate experienced during 1968-72 (for 1962-72 the rate is 26 percent). Multiplying the retention rate by the extraction rate results in a rate of oil exports in tons equal to 15 percent of the tons of olives produced. To estimate constant price exports, the volume estimate for oil is multiplied by the 1966 price received of 329 dinars per ton. During the period 1967-72 the average price has been between 348 and 361 dinars per ton. The stability of the price received is somewhat remarkable since during 1/ Based on conversations with M. Scalabre, FAO expert, Office d'Huile, Tunis.  - 13 - that period production of olives ranged from a low of 92 (1967) to a high of 900 thousand tons (1972). The ability of the consumers of olive oil to switch to other oils in the face of very small changes in relative prices apparently accounts for the past stability :Ln export prices. We realize, of course, that olive production is a very cyclical phenomenon, extremely dependent on weather conditions plus the bearing cycle of the trees. The projections are, therefore, long-run trends and not forecasts of actual annual crops (with the exception of 1973 and 1974). 29. Exports of phosphates are based on the production of phosphates (see above) with an adjustment for the domestic absorption of phosphates in the manufacture of phosphoric acLd. Domestic consumption was estimated in the following way: the total 1972 production of phosphoric products was translated into equivalent tons of raw phosphate using standard conversion factors for each product, 1/ and this resulted in an estimate of .983 million tons of phosphates. Projected domestic phosphate utilization was estimated on the same basis from the draft Plan, and related to projected value added in constant prices. This resulted in a marginal coefficient of .1144, and the equation for domestic phosphate consumption (DPC) is: 2/ DPC = .23 + .1144 YCHEM Subtracting domestic consumption from total production leaves a balance of crude phosphate exports. Exports of petroleum are based on the sector output (see para. 12). 30. For exports of other agricultural products, an exogenous estimate is used for the years up to 1976 based on forecasts made by the Ministry of Planning; thereafter they are projected to increase at 2.5 percent yearly from 1977 to 1980 and 2.0 percent after 1980. This implies increased supply of domestic market in agricultural products and compares with a growth rate in real terms of about 3 percent per annum for the 1965-72 period. These products are chiefly citrus fruits, dates, almonds and other fresh fruits and vegetables, which face a very competitive market due to competition from European and other Mediterranean sources. 31. Exports of manufactured commodities were projected in the Plan to grow from 45.5 M.D. in 1972 to 116.8 M.D. in 1976, or a growth rate of 26.5 percent on a compound basis. If we remove two major subcategories of chemicals (chiefly phosphate products) and textiles (chiefly made up articles, including shoes), the growth rate of the remainder is only 11.9 percent. The rapid growth of exports of chemicals and textiles is, of course, dependent on the realization of planned investments in those sectors. During the Plan period, 1/ Hyperphosphates, Super Simple Phosphates, Super Trile Phosphates, Phosphoric Acid and Fertilizer. 2/ The constant term is adjusted to make estimate agree with 1973 data on phosphate consumption.  - 14 - we have calculated that the ratio between the increase in value added and the increase in exports is 1.76 for chemicals, 1.84 for textiles and .551 for other manufacturing. Since we may want to change the allocation of investment between sectors in the model, and this should therefore change exports, we have used the following equation for projecting manufacturing exports: XMANt XMANt-1 + 1.84 YTEXT + 1.76 YCHEM + .551 YMANOT + XFOODP where XMAN = exports, manufactured commodities YTEXT = value added, textiles, constant prices YCHEM = value added, chemicals, constant prices = first difference operator YMANOT = value added, other manufactures XFOODP = exports, food processing The exports of processed foods are assumed to grow at 8.6 percent per annum, in line with the Plan estimates. These are a relatively small part of manu- factured exports, and consist of wine, sugar and canned goods. 32. The exports of goods is then the sum of petroleum, phosphates, olive oil, other agricultural, and manufactured goods exports. Exports of non-factor services are divided into two broad groups: tourism and other services. The projections of tourism earnings are based on projected growth of value added as described in paragraph 15 above. This growth rate is applied to the base 1972 earnings and converted to current prices with the use of the projected export price index for services. Exports of other services re- flects the recent Tunisian estimate in the years up to 1975, thereafter it is assumed to grow at 6.5 percent per year through 1985, which is the same growth rate used in the Plan for the 1972-76 period (presumably in constant prices). In the period 1966-72 the growth rate in current prices was about 8 percent, or about 6 percent in real terms. Most of the growth in this sector is due to the sale of transportation services, which in turn is linked to tourism. 33. Export prices in the recent years have under-one dramatic changes in the face of rising prices and world-wide inflation. The prices shown in Table 6 are generally based on the commodity analysis of the IBRD, with some adjustments for the Tunisian situation. The Tunisian petroleum price in 1980 is projected to be $12.20 in nominal term based on IBRD forecast for world market prices. This implies that the real price will be lower than the 1974 price when deflated by the international price index. Phosphate prices are projected to be $33.80 by 1980, as compared to the official IBRD forecast of $53.00 (75 percent BPL, f.a.s., Casablanca). The difference can be accounted for the fact that Tunisian phosphate is somewhat poorer in quality. The olive oil price is projected on the basis of the projected price index for coconut oil applied to the olive oil price in 1974.  - 15 - Table 6: EXPORT PRICES 1973 1974 1975 1976 1977 1978 1980 1985 Olive Oil, $/MT 1,149 1,657 1,476 1,362 1,430 1,502 1,656 2,113 Other agricultural products, growth rate (x) 8.0 5.0 5.0 5.0 4.0 4.0 4.0 4.0 Phosphate, $/MT 10.7 40.2 43.1 39.3 29.5 29.5 33.8 43.1 Petroleum ($/BBL /1 3.9 10.4 9.3 9.7 10.3 10.9 12.2 17.2 (Dinar/Ton 14.2 37.5 -33.5 35.0 37.3 39.3 44.2 62.0 Other goods, growth rate (%) 8.0 12.9 5.0 5.0 4.0 4.0 4.0 4.0 Non-factor services, growth rate (x) 10.2 10.0 5.0 5.0 5.0 4.0 4.0 4.0 International Prices (Index) /2 100.0 119.5 134.0 147.1 159.6 172.4 198.3 278.2 /1 Projection assumes a price of $1.625 per barrel in 1966. /2 December 16, 1974 projection. F. Imports 34. Imports are projected on the basis of the rather standard end-use categories, except that imports of foodstuffs are separated from other con- sumer goods imports. The five classes are: intermediate goods, capital goods, fuels, consumer goods, and food. The demand equations for each group were estimated using regressions on the 1960-72 period data. For intermediate products, textile materials (which are related to the textile industry) were separated from other intermediate goods (which are related to non-textile manufacturing output). The equations for these two groups are as follows: MITEXT = -.033 + .783 YTEXT R = .929 DW = 1.53 MIOTH = 3.157 + 1.059 YMANXT R = .724 DW = 1.80 where MITEXT = imports, textile intermediate goods MIOTH = imports, other intermediate goods YTEXT = value added, textiles YMANXT = value added, manufacturing, textiles excluded  - 16 - For capital goods, it was found desirable to isolate the impact on the balance of payments of different patterns of sectoral investments. It is known, a priori, that investment in manufacturing, for instance, will have a higher component of imported capital goods than investment in agriculture. Further- more, a regression of fixed investment on capital goods produces a rather low coefficient (.426). MCAP = -14.75 + .4262 IF R2 = .940 DW = 1.17 (13.17) where MCAP =imports, capital goods IF =gross fixed investment Unfortunately, there is no information on the import component of investment by sector. Instead, investment was divided into three categories; I1, those investments presumed to have a high imported capital component, I2, those with a low component; and I3, those having a medium component. Candidates for each class were chosen on an a priori basis, and some trial and error methods were used to make the final allocation between the three groups. The final equation for imported capital goods in the three groups is shown below (this equation estimated using maximum likelihood correction for auto- correlation): IICAP = -12.05 + .6740 11 + .2149 I2 + .3841 I3 R - .978 DW = 1.99 (10.59) (2.12) (4.70) where I1 = fixed investment; manufacturing, petroleum, energy and mining 12 = fixed investment; agriculture, government and otther services I3 = fixed investment; tourism, transport and communications 35. The final coefficients seem reasonable; the high group has a 67 per- cent import component, the low group has a 22 percent component, and the medium a 38 percent component. While it is possible that multicolinearity has distor- ted the coefficients the results conform to expectations. 36. Imports of food are assumed to be positively related to private consumption expenditures and a relative price index, and negatively related to agricultural production. The equation estimated is as follows: MIFOOD = -29.99 + 49.39 RPIFOOD + .0866 PC - .3073 YAG -2 (3.61) (3.45) (3.91) R = .875 DW = 2.02 (1962-71) The relative price index for food is defined as the wholesale price index divided by the implicit price deflator for imports. The fit of the equation is -,uite good, and the signs are all in the right direction. It is known, nowever, that during the period over which this equation is estimated import licensing has been important in determining food imports, even though it cannot be included in the equation. It is assumed that there is a limit to the reduction of food imports in the future due to the political problems created by not increasing food imports in the face, for instance, of bad harves ts.  - 17 - 37. Import licensing could also influence the other equations, and the feaeibility of using the data to estimate import demand functions may be questioned. In fact, in the past imports of capital goods and intermediate goods have determined investment and industrial production. The independent and dependent variables have merely been switched around in each equation so that the model can determine, the necessary level of imports for any level of industrial production and investment. In other words, while the regression equations may reverse the causality, they nevertheless capture a valid rela- tionship between the variables. 38. No attempt was made to fit a regression for the imports of consumer goods, since these imports are more fungible and not directly related to con- sumption expenditures. The ratio of consumer goods imports to total private consumption expenditures is shown in the table below: Table 7: PRIVATE CONSUMPTION AND CONSUMER GOODS IMPORTS, 1961-1972 (Mill:Lon dinars, 1966 prices) Private Consumer Year Consumption Imports Ratio 1961 324.2 28.7 8.9 1962 314.1 30.0 9.6 1963 356.0 26.1 7.6 1964 353.7 21.9 7.4 1965 381.3 20.9 5.8 1966 375.7 21.9 5.6 1967 384.7 17.9 5.7 1968 384.4 21.8 4.6 1969 420.7 21.5 5.2 1970 438.9 25.7 4.9 1971 491.8 31.5 5.2 1972 559.8 31.5 5.6 The peak comes in 1962, when consumer goods are 9.6 percent of private con- sumption. After this, the imposition of exchange controls brings a gradual fall in the ratio, until it reaches its low of 4.6 percent in 1968. Since then there has been a gradual tendency for the ratio to rise,, particularly in the last few years, when increased foreign exchange earnfngs have permitted a liberalization of import controls. In 1972 the import-expenditure ratio was 5.6 percent, up from 5.2 percent in the previous year. 39. The level of consumer imports is a policy variable, susceptible, within limits, to government decisions relative to import liberalization. For these projections, it has been assumed that there will be a tendency toward liberalization in light of the greater export earnings and reserve accumulation. The ratio of consumer imports to private consumption is there- fore projected to rise from the actual figure of 5.0 percent in 1973 to 6.5 percent in 1976, then remain at 6.5 percent throughout the projection period.  - 18 - Since the rapid growth projected for the economy implies increased production of domestic consumer goods, it is fair to assume that the ratio will not rise above the previous peak of 9.6 percent. 40. Imports of fuels (including crude petroleum) are based on estimates made by the Tunisian government. For the period up to 1980 it is the judgment that fuel imports can be maintained at a fairly constant level, given that the construction of a new refinery will permit additional import substitution and the increased export of petroleum products. Hiowever import volumes beyond 1980 are projected to rise assuming that there will be an increase in domestic consumption, as indicated in Table 2. 41. Imports of non-factor services are projected with an equation which relates them simply to total GDP. The equation is: MSER = -6.23 + .084 GDP where MSER = imports, non-factor services GDP = gross domestic product at market prices Total imports of goods and non-factor services are simply the sum of inter- mediate goods, capital goods, consumer goods, fuels, food, and services. G. Consumption and Saving 42. The estimates of consumption expenditures and saving are based on the private and the government sector. Private consumption expenditures are estimated using an estimated consumption function related to gross national income. Gross national income (GNY) is simply gross national product in constant prices (GNP) adjusted for the terms of trade, and GNP is GDP at market prices plus net factor income from abroad. The terms of trade adjust- ment is necessary in order to recast exports in constant prices into an ad- justed export series which reflects the import purchasing power of exports. This is done by dividing exports by the import price index, and subtracting exports at constant prices. 43. To estimate private consumption expenditures in constant prices, an attempt was made to fit a Friedmanite consumption function, in which estimates of the permanent and transitory elements of income are isolated, under the assumption that the permanent income hypothesis is relevant. 1/ After several trials with different measures of permanent income, however, it was found that income with a one-year lag performed the best. The re- gression equation is: 1/ For a discussion of the relevance of the permanent income approach in developing countries, see R.F. Mikesell and J.E. Zinser, "The Nature of the Savings Function in Developing Countries: A Survey of the Theoretical and Empirical Literature", Journal of Economic Literature, 1973, p. 1111.  - 19 - PC = 68.70 + .581 GNY* + .3857 (GNY -GNY (11.21) t, +6(3.42) t t1 R = .978 DW = 2.28 where PC = private consumption expenditures in constant 1966 dinars GNY = gross national income 44. Government consumption, as well as government current expenditures, grow at the growth rate derived from the government sector in the national accounts. (While government value added, consumption and current expendi- tures are all slightly different concepts, it is assumed that they-grow in line with each other.) Government saving is estimated by subtracting current expenditures from current revenuies. Current revenues are estimated using two broad equations, one for direct and other taxes, and another for indirect taxes. Direct taxes and other revenues are related simply to total GDP, while indirect taxes are related to imports, all in current prices. Three dummy variables were also inserted to capture the effect of tax rate changes in 1965, 1966, 1968, and 1969. The current revenues so estimated are then deflated (using the implicit deflator for total GDP) to arrive at current revenue in constant prices. The regression equations for taxes are: -2 INDTX 4.559 + .4297 M* + .1566 D68 R .929 DW = 1.53 (15.97) (5.80) DTXOTH 4.990 + .0744 GDP* + 6.205 D66 + 7.545 D 65 + 15.05 D69 (8.46) (2.22) (2.56) (5.22) -2 R = .992 DW = 3.09 where INDTX indirect taxes DTXOTH = direct and other taxes M*= imports, goods and non-factor services, current prices GDP* = GDP in current prices D65 = dummy variable having value of 0 up to 1964, 1 thereafter D66, D68, D69 = as above for 1966, 1968 and 1969 Current revenue in constant prices (CURREV) is then defined as: CURREV = (INDTX + DTXOTH)/IPD where IPD = implicit price deflator for GDP, 1966 = 1.00  - 20 - 45. Government gross national saving (GNSPUB) is equal to revenues less expenditures, while total gross domestic saving is equal to GNP less private consumption and government consumption. The difference between gross national and gross domestic saving is equal to net factor income and net current trans- fers. 46. The resource gap is defined as the difference between imports and exports of goods and non-factor services, which by definition must equal the difference between savings and investment. The equality of the two gaps is insured by allowing changes in stocks to absorb any differences, so that stock changes become defined as: DSTK = GDP - PC - GC - IF + M - X or DSTK = M - X - S -IF where DSTK = change in stocks GC = government consumption M = imports, goods and non-factor services X = exports, goods and non-factor services IF = total gross fixed investment S = gross domestic saving In the long run, one would exp .ct DSTK to be small and positive as stocks grow with the general level of demand. If this is not the case, it means that that some other variable must be changed in order to correct the system. H. Balance of Payments 47. The balance of payments is projected in current U.S. dollars. Cur- rent price exports and imports are derived from the series projected in current dinars, using an exchange rate of 2.27 $/D. 1/ Imports are translated from constant price imports into current price imports using two price indices (1966 = 1.0), one for goods and one for services (see Table 8). Net current transfers, direct foreign investment, capital grants and other factor income are derived from Mlinistry of Plan projections through 1976, and extended to 1985. Net interest is composed of interest on the public debt and other interest. Tnterest on the public debt consists of interest on the existing debt which is based on IBRD estimates, plus the interest on new commitments estimated by using the BSIML projection program. Other interest is composed of a small amount received on reserves plus payments on 'private debt (public debt is defined as debt owed by public bodies, or guaranteed by them. Only the unguaranteed private debt is considered to be private). 1/ For 1972, an exchange rate of $2.08 was utilized.  - 21 - Table 8: IMPORT PRICES 1973 1974 .1975 1976 1977 1978 1980 1985 (rate of growth in percent) Goods 17.5 30.0 8.0 7.0 5.5 5.0 5.0 4.0 Non-factor Services 0.0 16.0 6.0 5.5 5.0 5.0 5.0 4.0 48. Workers' remittances are projected at a lower rate than previously, due to the expected decline in the rate of net migration to Europe from Tunisia. This can be traced equally to the economic downturn expected in Europe due to the energy crisis, and the increased amount of regulation and restriction on foreign workers. Remittances up to 1976 represent a recent estimate in the government's economic budget. For the remainig years model projects with a 5% per year inflation factor plus a 5% yearly increase in real wages, assum- ing there will be no net changes in the external labor force. 49. Direct investment income is determined based on two assumptions: (a) that revenue generating from the petroleum sector will amount to about 10 percent of the petroleum export receipts, and (b) that investment income from the non-petroleum sectors is related to the implicit capital (foreign) stock, earning a 8 percent return. In 1972 direct investment income was about $25 million, with $13 million coming from petroleum revenues, the end 1971 capital stock therefore is estimated at $150 million. The end 1972 capital stock is calculated by depreciating the 1971 capital stock by 3 percent, and adding direct investment made in 1971. Since investment income paid in 1973 is gen- erated from the 1972 capital stock, the net effect is a two year lag between investment and the resulting income flow. I. Capital Inflow and Reserve Changes 50. Various assumptions for the commitments of new foreign capital flows of public and publicly guaranteed loans are given exogenously. These assume, roughly, a continuation of existing commitment levels through 1976, and a gradual increase thereafter. To these commitment assumptions, a dis- bursement pattern is applied to generate new disbursements. The disbursement assumptions for bilateral donors are based on historical experience in Tunisia for the period 1966-68, while those for the IBRD and IDA are based on Bank and IDA world-wide historical experience. For financial credits, a 40-40-15-5 rate is used, based on a cross-country study, which seems appro- priate to Tunisia, while suppliers' credits are assumed to be disbursed in the year of commitment. 51. Given the terms of new borrowings, the BSIML computer program was used to calculate the amortization. When combined with existing amortiza- tion payments, based on the IBRD debt statistics, total amortization and net disbursements are calculated. The balance between the current and cap- ital account is maintained by movements in reserves. The ratio of the re- serve level to imports of goods in current prices is calculated, and it is assumed that reserves will not be allowed to fall below 25 percent of imports  - 22 - (three months). In case this should happen, suppliers' credits are increased until the reserve ratio reaches .25. J. Money and Prices 52. The price level for the entire economy is determined from an equa- tion relating the implicit price deflator for GDP (IPD) to the money supply (MS2 - includes quasi money) and real output. The equation is estimated in logarithmic form as follows: log IPD = -.5871 log GDP + .6552 log MS2 + .2797 R = .953 DW = 1.94 (2.63) (5.25) The money supply is divided into two parts: net foreign assets and other do- mestic credit. The monetary authorities are assumed to be able to keep the total money supply growing at a rate 3 percentage points faster than the growth of real GDP. Net foreign assets (reserves) are derived from the balance of payments, so that domestic credit, which has been adjusted for counterpart funds, government deposits, etc., must grow at a rate sufficient to meet the required overall growth of the money supply. Table 9: DISBURSEMENT ASSUMPTIONS (Percent per year) Year Donor 1 2 3 4 5 6 7 8 9 10 World Bank 3 13 20 20 17 12 6 4 3 2 IDA 2 12 19 16 15 13 10 7 4 2 U.S. 50 25 10 5 5 5 France 30 30 10 10 10 10 Germany 5 10 20 15 15 15 10 10 Others 50 25 10 5 5 5 Financial Credits 40 40 15 5 Table 10: TERMS ASSUMPTIONS Interest Maturity Grace Period Source (Y) (Years) (Years) IBRD 7.25 25 5 IDA .75 50 10 Germany (FRG) 2.50 25 8 France 5.50 10 1 Italy 5.50 12 4 Canada .0 40 10 U.S.A. 3.0 35 10 East Bloc 2.5 8 1 Other Bilateral 3.5 20 5 ADB 5.0 20 1 Financial Credits 8.0 10 3 Suppliers' Credits 9.0 7 1  - 23 - 53. Because of the limited data on prices in Tunisia, no attempt has been made to introduce the eff'ects of relative prices in the equations on sector outputs. The use of relative prices did not prove significant in the import equations, except for imported foodstuffs. The general price level has a real effect in the government revenue and expenditure equations. Since gov- ernment revenues are based in part on imports, which are only slightly affected by domestic price changes, government revenues tend to rise more slowly than government expenditures in an inflationary situation, causing a fall in gov- ernment saving, and therefore a fall in total saving. 54. The following pages give a glossary to aid in identification of the variables, and a complete set of equations for the model. In addition, the output from the model for Run 1, which utilizes the "low" assumptions and Run 2, which utilizes the "high" range of assumptions, are shown.  - 24 - K. Symbols Aggregate Variables (constant 1966 dinars) GDP = Gross domestic product market prices GDPFC = Gross domestic product factor cost PC = Private consumption expenditures GC = Government consumption exoenditures IF = Gross fixed investment DSTK = Change in Stocks K = Imports, goods and n.f. services X = Exports, goods and n.f. services GDY = Gross domestic income (GDP + TTADJ) GNP = Gross national product GNY = Gross national income (GNP + TTADJ) NFY = Net factor income NCT = Net current transfers TTADJ = Terms of trade adjustment GDS = Gross domestic savings (GDY - PC - CG) GNS = Gross national savings (GDS + NFY + NCT) Sector Variables Y = Value added I = Fixed investment ICOR = Incremental capital output ratio AG = Agriculture AGNOL = Non-olive agriculture AGOL = Agriculture, olive only IMIN = Yiining M ER= Energy (excluding petroleum) PET = Petroleum M