The distributive effects of local government finances in Colombia : a review of the evidence

The Distributive Effects of Local Government Finances in Colombia: A Review of the Evidence SWP-235 World Bank Staff Working Paper No. 235 March 1976 This paper is prepared for staff use. The views expressed are those of the author and not necessarily those of the World Bank.   This paper is prepared for staff use. The views are those of the author and not necessarily those of the Bank. THE WORLD BANK Staff Working Paper No. 235 March 1976 THE DISTRIBUTIVE EFFECTS OF LOCAL GOVERNMENT FINANCES IN COLOMBIA: A REVIEW OF THE EVIDENCE Local government in Colombia plays an important role in the provision and financing of public services, particularly in the urban areas. The potential scope for redistribution of income through local government activities should therefore not be neglected, as it generally has been in the literature. This paper brings together scattered evidence availalbe mainly in unpublished sources in order to assess the distributive impact of Colombian local government in four areas of fiscal policy: taxation, public expenditure, public service pricing, and access to public services. An earlier version of the paper was presented at the Rice University Workshop on Income Distribution and Policy in Colombia, on May 2-4, 1975. In revising the paper the author has benefited from extensive comments by Hernando Acevedo, Roy W. Bahl, Albert Berry, Richard M. Bird, Orville Grimes, Douglas Keare, Charles E. McLure, Jr., and Jacob Meerman. He also acknowledges gratefully the assistance of many Colombian officials at the national and local level, who helped in assembling the information on whic h this paper is based. Responsibility for any errors of conception or fact remain with the author. Johannes F. Linn Urban and Regional Economics Division Development Economics Department Development Policy Staff   TABLE OF CONTENTS Page No. I. INTRODUCTION.............................................. II. THE INSTITUTIONAL FRAMEWORK OF LOCAL GOVERNMENT...........3 III. THE CONTRIBUTION OF LOCAL GOVERNMIENT TO TAXATION AND PUBLIC EXPENDITURE..................... IV. THE INCIDENCE OF LOCAL TAXES AND EXPENDITURES.............10 A. Tax Incidence .......................................... 10 1. The Property Tax .................................. 10 2. The Valorization Tax .............................. 16 3. Other Local Taxes ................................. 17 4. Summary of Local Tax Incidence .................... 20 B. The Incidence of TLocal Public Expenditure ............. 22 V. THE INCIDENCE OF USER CHARGES FOR URBAN PUBLIC SERVICES ... 24 A. National Pricing Policy for Public Services in Colombia . . 24 B. Public Service Pricing and the Incidence of Subsidies ........................................... 27 C. An Estimate of Subsidies in Urban Public Utility Pricing ..................................... 33 D. Some Further Issues in Public Service Pricing ......... 35 VI. ACCESS TO PUBTIC SEPRVICES ............. .... . .. .... ...o.... . 3d VII. CONCLUSIONS ......oo-ooo ... 43 APPENDIX*..ooo....oooo...o'....o..ooo................-.......o..... 1. Tax Burden by Level of Government 2. Distribution of Public Expenditure by Level of Government 3. Cali: Cross-Subsidies between Users of Water and Sewerage Services 4. A Framework for a Model of Optimal Public Service Pricing and Investment   I, INTRODUCTION Local public finances have in the past received more attertion in Colombia than has generally been the case in developing countries.!/ How- ever, most authors have considered the distributive impact of local govern- ment activities only in passing, if at all.2/ Two main reasons are likely to explain this neglect: First, the importance of local government in the overall public sector activities and thus the distributive impact of local finances, is generally underestimated. One of the objectives of this paper is to demonstrate that, particularly in urban areas, local government accounts for a significant proportion of public sector responsibilities, and that by implication the potential for redistribution through the local public sector should not be neglected. A second reason for the apparent reluctance to deal with the distr'butive effects of local finances may rest with methodological reservations. 1/ See Richard M. Bird (1970), Taxation and Develonment: Lessons from the Colombian Exnerience, (Cambridge, Mass.: Harvard University Press 1970), Ch. 5; Lauchlin Currie, Una Politica Urbana para los Paises en Desarrollo, (Bogota, D. E.: Ediciones Tercer Mundo 1965), Ch. 10; Sally Marie Hey, A Study of Municipal Public Finances in Bogota, D. E., Medellin and Cali, Colombia (Doctoral Dissertation, Johns Hopkins University, 1970); Arvids Kalnins, Estudio del Ipuesto Predial y Complementarios del Distrito Especial de Bogota, 1950-1962 (Bogota, D. E.: Departamento Administrativo de Planeacion Distrital, 1962); R. A. Musgrave and M. Gillis, ed., Fiscal Reform for Colombia (Cambridge, Mass.: International Tax Program, Harvard Law School, 1971), Chs. 14, 16-18; Dick Netzer, Some Aspects of Local Government Finance in Colombia, Economic Development Report No. 51 (Cambridge, Mass.: Development Advisory Service, Harvard University 1967); Republic of Colombia - IBRD-UNDP, Bogota Urban Development Study Phase II, "Fiscal Studies," Bogota, D. E. 1973 (referred to below as Phase II Report); Milton C. Taylor, et. al., Fiscal Survey of Colombia (Baltimore: The Johns Hopkins Press, 1965). Chs. 7, 10. 2/ In the Phase II report, op. cit., an attempt was made to estimate the incidence of local taxes in Bogota, D. E. In the context of a compre- hensive tax incidence study of Colombia by Charles E. McLure, Jr. (1973), "The Incidence of Colombian Taxes, 1970" Paper No. 41, Rice University Program of Development Studies, 1973, the incidence of local taxes was analyzed along with other higher level taxes. This is also the case for the tax incidence analysis in Taylor, op. cit. The incidence of local public expenditures was analyzed jointly with that of all public sector .spending in Miguel Urrutia Montoya and Clara Elsa de Sandoval, "Politica Fiscal y Distribucion del Ingreso en Colombia," Revista del Banco de la Republica, July 1971, p. 1072-1087). / A further constraint, and possibly the binding one, may have been that factual evidence on local government activity has been very sparse, and that only a case study approach and/or surveys conducted on a large scale can provide meaningful insights into the operations of local govern- ments in developing countries.  The conventional theory of fiscal decentralization holds that central government authorities have a comparative advantage in pursuing the goal of income redistribution, and that therefore concern w1th the distributive effect of local public finances is inappropriate.!/ But one may counter this view on various levels: First, the conventional wisdom is based on a number of assumptions, such as perfect information and perfect mobility of the factors of production, of industrial and com- mercial activities, and of residential location,which would ensure a move- ment of factors and activities sufficient to offset local government ini- tiatives in pursuing a particular target of redistribution. While these assumptions may have some justification in highly industrialized countries such as the U. S., this cannot be claimed for most developing countries like Colombia, where a high degree of regionalization, and a continuing lack of communication probably result in significant barriers to mobility and information flows. j Second, national policy aimed at the distributive goal may uniformly direct local fiscal instruments by setting nation-wide standards which guide their use in the appropriate direction. In Colombia, the national pricing policy for local public services is a prime example for this approach. Similarly, any national plan for local fiscal reform must consider its distributive implications, in order to be able to offset any undesirably distributional consequences. Finally, in a country where all policy instruments, including national fiscal policy, remain extremely imperfect and limited tools for reaching a given target of redistribution, the policy makers will want to be able to utilize all available instruments, even those which may be relatively weak, provided their beneficial distri- butional effects are not more than offset by efficiency losses. The purpose of this paper is therefore to focus explicitly on the distributive impacts of local government finances in Colombia by conducting a comprehensive review of the scattered and frequently unpublished evidence on this issue.2/ The remainder of the paper is organized in six sections: Section II highlights the institutional framework of local government, in order to clarify the linkages between local authorities and other decision makers in the public and private sectors. Section III demonstrates the relative importance of local government in Colombia. Section IV reviews the evidence on the incidence of local taxation and expenditure. Section V presents an initial attempt to analyze the distributive effects of pricing policies for user-charge financed public services, which provide a major and rapidlygrowing share of local fiscal capacity. Section VI focuses on the issue of access to local public services by different income groups. Finally, Section VII pulls together the results and conclusions of the paper in assessing / See e.g., R. Musgrave and P.B. Musgrave, Public Finance in Theory and Practice (New York: McGraw Hill, 1973), p. 606. 2/ The fact that fiscal instruments have been found to be notoriously inef- fective in inducing regional decentralization in developing countries supports this conclusion. A substantial amount of evidence is drawn from two case studies of urban finances in Colcmbia by the author: "Urban Public Finances in Developing Countries: A Case Study of Bogota, Colombia," IBRD, forthcoming; and "Urban Public Finances in Developing Countries: A Case Study of Cartagena, Colombia," IBRD, forthcoming. Further evidence is based on research missions to Cali, in the context of an IBRD research project on public service pricing, conducted by the author in collaboration with R. M. Bird, L. K. Hubbell, and C. E. McLure, Jr.  -3- the policy implications of the preceding discussion. Throughout the paper most attention is given to local government in urban areas, since it is in the urban sector that local government contributes most to public sector activities and where as a result its potential for affecting the distribu- tion of income is most significant. II. THE INSTITUTIONAL FRAMEWORK OF LOCAL GOVERNMENT As a first step in the analysis of local government finance 2 in Colombia it is necessary to describe briefly the institutional framework in which the public authorities operate at the local level. Three sets of agents act, and to some degree interact, in municipal government: The municipal administration, which is the local executive branch, is headed by the mayor; the mayor is appointed by the governor of the department (state), who in turn is an appointee of the President of the Republic. The executive branch of local government is thus a creation of, and to a large degree dependent on higher level authorities.,/ Second, the municipal council, which is the legislative branch of local government, is popularly elected. The autonomous (decentralized) local public enterprises, which are the main agents in providing local public services especially in larger cities, are only to a very limited extent under the control of either the mayor or the municipal council, and therefore can, and frequently do, act independently of outside administrative or political control. Attempts by the executive branches in various of the major cities (e.g., Bogota and Cali) to centralize administrative decision making in the hands of the mayor through the creation of planning offices with comprehensive statutory powers have thus far failed to produce a unified local government planning and decision process.5/ 1/ The definition of "local government" is here restricted to the municipal level; the departmental (state) level is treated only in passing. Note that in Colombia municipalities are the only administrative form of local government, and therefore cover urban as well as rural areas of the country. 2/ This does not rule out conflicts between the local and the higher level executive branches, especially since under the constitutional rules of the National Front parity has to be maintained in the appointment of governors between the two major parties, and the mayor of each depart- mental capital cannot be drawn from the party to which the departmental governor belongs. 3/ According to an announcement in the Presidential Address of February 20, 1976, the mayors are now empowered to appoint the managers of the local decentralized agencies. This may strengthen the local executive branch and indirectly increase the influence of higher level government at the local level.  The political pressures operating on the local authorities include those by the local elite, the local electorate, local ad regional industrial and commercial interests, and the national government. The municipal council responds primarily to the first three of these sets of pressures; the mayor in addition has to consider regional and national influences; while the local public enterprises are strongly influenced by central government, in addition to various local pressures. The latter may be made effective through the mayor and the municipal council in their (limited) control function over the activities of the local enterprises. Given these public agents at the local level and the numerous pressures to which they are subjected it is not surprising that on balance the distributional impact of local government activities tends to be very diverse in different localities of Colombia. For instance, while it appears that in Cali local government has responded to local pressures towards socially pro- gressive policies and appears aware of the distributive impacts of its activi- ties, the local public sector in Cartagena in the past seems to have been characterized by an absence of awareness regarding the distributive effects of its activities. Moreover, the fact that numerous agencies operate at the local level with very different objective functions explains the often con- tradictory results of local government decision making, including the distri- butive impact of local public finances. A striking example for such con- tradictory results may be observed in Cartagena, where the rate structures of the two most important municipal taxes show strongly regressive characteristics, while ;he tariffs for water supply and sewerage services are strongly progres- sive . 2. The most important force which tends to level differences in local fiscal management between various localities is the influence of the national government. While it does not exert direct control over the budgeting process at the subr)ational level, it limits the revenue authority of local public agencies.l/ In the field of taxation national legislation has delegated only two major taxes to local government, the property tax and the industry and commerce tax.l/ All other major tax instruments, in particular the income and 1/ There is, at least in the larger cities, little interaction between muni- cipal and departmental governments. However, local authorities may also be under significant pressure from international lending agencies to the degree that they are dependent on their financial contributions. This occurs especially in the case of local autonomous enterprises. 2/ J. F. Linn, "Cartagena," op.cit. 2/ The central government also shares responsibility in the provision of some local public services, generally through its decentralized agencies. This is the case particularly in smaller towns and rural communities. 4/ With the exception of Bogota and Medellin, the tax base for the property tax is appraised by a national agency, the Geographic Institute "Agustin Codazzi." National legislation has also set maximum rates for property taxation.  wealth taxes, the corporation tax, and the sales tax are exclusively reserved for the national fisc.-V The rate structure for local taxes is generally limited by national legislation. Moreover, public service pricing at all levels of government is regulated by the National Tariff Board, a branch of the National Planning Department, which has to approve the tariffs for all public services. Mainly as a result of the influence of the Board, the tariff structure is quite uniform for the major public services throughout the country, although the tariff levels differ.Z/ The National Planning Department further has to approve all foreign and most domestic borrowing by public agencies, including those at the local level.3/ Finally, central government grants, especially those for public utility construction programs and for education and health may be used to exert central government leverage over local government decision making. In summary, the national government limits local government activities through the constraints placed on revenue sources, and acts as a unifying, leveling force particularly in the area of public service provision and pricing. But local government in Colombia remains functionally fragmented and subject to numerous pressures which operate differentially on the various local agents. As a result, cities may be expected to differ widely as regards the distributional impact of their activities and in their concern for it. Furthermore, within each city prima facie contradictory distributive choices are frequently made. III. THE CONTRIBUTION OF LOCAL GOVERNMENT TO TAXATION AND PUBLIC EXPENDITURE Before turning to an analysis of the evidence of distributive impacts of local government fiscal actions, the contribution of the local public sector to the overall fiscal structure needs to be reviewed, in order to assess the importance of local government in financing and providing public services, relative to other levels of government. 1/ A share of the sales tax is distributed to local government. 2/ For a detailed description of the pricing principles entertained by the Board and the tariff structure, see Section V below. 3/ See National Planning Department, "Credito Public Interno," Documento DNP-998-UDRU-DEDM, December 10, 1972; and "Disposiciones Legales sobre la Contratacion de Credito Externo," Documento DAP-126-GPE, August 28, 1968.  - 6 - Beginning with taxation, Table 1 shows the shares of the three levels of government (national, departmental, and municipal) in the total tax burden for the country as a whole, a s well as an estimate of these shares for three major Colombian cities.-/ For the nation as a whole the share of subnational tax burden amounts to some 20 percent, of which municipal governments contribute a little less than half. In the case of Table 1: PERCENTAGE DISTRIBUTION OF REVENUE BURDEN BY LEVEL OF GOVERNMENT, 1970 All Municipalities Total Bogota Cali Cartagena Taxes Current Revenues Taxes Taxes Taxes National 79.4 75.8 78.3 78.3 77.1 Departmental 11.4 9.0 - 10.1 9.3 Municipal 8.9 15.2 21.7 11.6 13.6 Total 100.0 100.0 100.0 100.0 100.0 Source: Taxes: Appendix Table Al Current Revenues: IBRD, Economic Position and Prospects of Colombia, Report No. 696-CO, Vol. I, Table 5.4. Bogota, where municipal and departmental tax authority is combined, almost 22 percent of the tax burden is accounted for by District taxes. For the other two cities the municiDal governments impose a larger burden than the departmental authorities. Since this is the reverse of the national pattern for all municipalities combined, municipal taxes in the urban sector are likely to be more important than the departmental taxes, while the opposite holds for the rural sector. The importance of the municipal revenue structure is further emphasized when considering all current local revenue sources. Approximately 25 percent of all current revenues of the public sector were collected by subnational governments in 1970, and municipal authorities accounted for almost two-thirds of this (Table 1). These data also indicate that non-tax revenues, consisting mostly of user charges, make an important contribution particularly at the municipal level. A careful analysis of the burden and incidence of local taxes and charges therefore appears warranted (see Sections 1/ For a discussion of the assumptions underlying these estimates and their resulting limitations see the Appendix.  - 7 - IV and V below), particularly in the larger urban centers where the distributive impact of the tax structure and public service prices may be considerable.l/ The importance of local government in total public sector activity particularly in urban areas, is further highlighted when looking at public sector expenditures. Despite a relatively slow growth rate in overall municipal spending between 1970 and 1973, municipal governments contributed over 17 percent to total public sector spending in Colombia in 1973 (Table 2). In the major cities the share of local government was certainly considerably higher, and is estimated to have reached almost 50 percent in Bogota and Cali. Table 2: DISTRIBUTION OF EXPENDITURE BY LEVEL OF GOVERNMENT1/ All Municipalities-C m Bogota m Cali/b CartagenZ Percent Distrib. Growth Rat= Percentage Dis ribution 1973 1970-73 1972 19753a 1972 National Gove't. 71.0 21.2 50.1 44.5 68.4 Central Gov't. (32.4) (20.5) (23.1) (20.1) (31.5) National Decen- (38.6) (22.1) (27.0) (2h.h) (36.9) tralized Agencies Departmental Gov't./4 11.6 22.0 - 6.7 8.5 Municipal Gov't./4 17.4 18.6 49.9 48.8 23.0 Total 100.0 20.8 100.0 100.0 100.0 /1 Net of intergovernmental transfers. /2 Average annual compound growth rate, in nominal terms. /3 Budget figures. /4 Including decentralized agencies at the departmental and municipal levels respectively. Source: /- IBRD, op. cit. /b Annex Table A2. 1/ Attention focuses here on the distributive aspects; the same argument could also be made for the importance of the allocative effects of local taxation and user charges, e.g., the effects of the property tax and user chargers on land use patterns.  - 8 - Despite the relatively slow growth rate of overall municipal expenditures shown in Table 2, further breakdown of municipal spending (not shown in Table 2) indicates, first, that for all municipalities the expenditures of decentralized agencies accounted for over 80 percent of total municipal expenditure in 1973 and that they have grown qt the very rapid annual rate of about 30 percent between 1970 and 1973.21 Since these agencies are largely financed by user charges this result further strengthens the previous conclusion that a study of the distributive effects of local government will have to consider explicitly the incidence of public service prices. A further dimension of the issue of local government participation in total public sector activity may be explored by looking at the distri- bution of responsibility for the main public service functions which are carried out in the urban sector. Table 3 shows the degree of involvement of local agencies in the provision of these services in a number of cities. In Bogota city authorities have primary responsibility for virtually all functions, but local government involvement declines for intermediate and smaller cities, especially as regards social services. However, even in the medium and smaller cities local government generally retains responsibility for water and sewerage, and for a number of the minor public services, such as refuse collection, fire protection, markets, cemeteries, etc., and is secondarily involved in a number of the other services. The preceding discussion of local revenues and expenditures in the overall public finance setting of Colombia, suggests that it is desirable to investigate in greater detail the distributive effects of local taxation, of local expenditures, and of the provision and pricing of local public services. The following three sections will discuss each of these issues in turn on the basis of the empirical evidence presently available and will suggest directions for future research. 1/ Dirreccion General del Presupuesto, "Analisis de los Presupuestos Consolidados del Sector Publico, 1970-73," mimeo 1975.  - 9 - Table 3: LOCAL PUBLIC SECTOR RESPONSIBILITY FOR URBAN SERVICES (P: Primary Responsibility; S: Secondary Responsibility; N: Negligible Responsibility) Functions Bogota/a Cali/b Cartagenaic All Municipalities/d /1 /2 /3 Water p p p 79 (137) 57.7 Sewerage p p p 79 (137) 57.7 Electricity p p _ 24 (144) 16.7 Telephones (Local) P P P 55 (62) 88.7 Roads and Bridges P P P s Street Lighting p p - 24 (144) 16.7 Refuse Collection P P P P Parks and Recreation P P P P Fire Protection P P P p Markets and Abattoirs p p p p Cemeteries p p p p Transportation S N N N Law Enforcement S N N N Education Primary p N N N Secondary p N N N Health S S - N Social Welfare P S - N Housing S S - N /1 Number of cities where local government has primary responsibility for the service. /2 Total number of cities covered. /3 Percentage of cities where local government has primary responsibility f6r the service. Source: /a J.F. Linn, "Bogota," op. cit. b Mission data. /c J.F. Linn, "Cartagena," op. cit. /d Arvids Kalnins, Entidades Descentralizadas Territoriales de Colombia, Presidencia de la Republica de Colombia, Bogota, D.E., March 1974.  - 10 - IV. THE INCIDENCE OF LOCAL TAXES AND EXPENDITURES This section reviews some conventional estimates of the incidence of local taxes and expenditures in Colombia. These studies have the common characteristic of neglecting to consider the incidence of user charges or of user-charge financed public expenditures, which may have important distri- butive effects, as Sections V and VI below will attempt to show. However, traditional tax and expenditure incidence studies remain of considerable importance and are therefore subjected to a thorough analysis in the present section. A. Tax Incidence Two recent estimates of the incidence of local taxes in Colombia are available: In his nation-wide study of Colombian tax incidence McLure has considered the two main local taxes, the property tax and the industry and commerce tax..Y Further, in Phase II of the Bogota Urban Development Study an analysis of the incidence of local taxes in Bogta was made..Z/ Incidence studies are notoriously difficult to compare,)/ but for the pur- pose of this paper a presentation of the results and of the likely causes for differences between the two studies will be instructive in conveying a broad impression of the probable incidence of local taxes, and of the critical analytical issues requiring further study. Since the property tax is the most important of local taxes, and the most controversial as regards the distribution of its burden, this section will concentrate on the inci- dence of property taxation, while reviewing the remaining taxes only briefly. 1. The Property Tax McLure found the property taxJ/ to be strongly progressive in middle and upper income brackets, while it is highly regressive according to the Bogota Phase II Study estimates, especially when measured relative to monetary, not total family income (see Table 4). This difference is largely explained by the differing shifting assumptions employed, and is thus illustrative of the difficulties encountered in measuring and comparing tax incidence. McLure assumes that, in the case of residential properties, the tax is borne by capital owners, and in case of commercial and industrial properties by the firms' share- holders, who are assumed to have incomes above Col$120,000 per annum. In 1/ Charles E. McLure, Jr. (1973), op. cit. 2/ Phase II, Study, op. cit. 3/ See Charles E. McLure, Jr., (1975), "Taxation and the Urban Poor in Developing Countries," IBRD Staff Working Paper No. 222, December 1975, for a discussion of these difficulties. Unfortunately, the Phase II Study does not offer enough detail on estimation procedures to assess the differences between its results and those of McLure, except as regards the shifting assumptions discussed further below. 4/ Including the valorization tax. The incidence of this tax is discussed separately below.  Table 4: ESTIIMATES OF THE INCIDENCE OF THE URBAN PROPERTY TAX, 1970 Property Tax/Income Ratio Income Class Percentage of Al, Phase II-b (Col$ p.a.) Urban Households-i Money Income Total Income McLurei- 0- 12,000 14.7 1.01 0.85 0.1 12,000- 24,000 30.2 0.82 0.73 0.1 24,000- 36,000 17.5 0.77 0.70 0. 36,000- 60,000 16.8 0.83 0.77 0.3 1 60,000-174,000 17.7 0.68 0.65 0.5-0 6/i 174,000 and over 3.1 0.61 0.58 1.7-253w Average 0.76 o.66 o.6-o.7L /1 Average of ratios for income sub-classes, weighted by the share of income in each class. Source: /a Charles E. McLure, Jr., (1973), ov cit. /b Phase II Study, op. cit. contrast, the Phase II Study assumes that the tax is shifted fully to occupiers in the case of residential properties, and to consumers in the case of com- mercial and industrial properties. These contrary sets of shifting assumptions reflect extreme versions of the conflicting views of the "new" and the "traditional" approaches to the analysis of property tax incidence 1/ On the assumption that factors are fixed in supply and perfectly mobile within a country, and that markets are perfectly competitive, the new view concludes that a uniform nation-wide property tax will be borne by all owners of land and capital. And since the ownership of these factors tends to be more highly concentrated than income, the property tax is taken to be progressivel./ The new view also recognizes that property tax rates do not apply uniformly for all types of land and capital, and that rates may differ between and within jurisdictions. It allows that these differentials will induce factor movements, and therefore price, or "excise," effects so that part of the property tax may be shifted either to labor or to consumers. 1/ For a brief review of the controversy see Helen F. Ladd, "The Role of the Property Tax: A Reassessment," in Richard E. Musgrave, ed., Broad-Based Taxes: New Options and Sources, the Johns Hopkins University Press, Baltimore, 1973, pp. 39-86. 2/ See Henry Aaron (1974), "A New View of Property Tax Incidence," American Economic Review, Vol. 64(2), May 1974, pp. 212-231 for a concise exposi- tion of this view, and for references to the original contributions by Harberger and Mieszkowski. See also Henry Aaron (1975), Who Pays the Pro- perty Tax? A New View, (The Brookings Institution, Washington, D.C. 1975).  - 12 - However, most analysts have judged it extremely difficult, if not impossible, to deterxine the net incidence of these excise effects on a country-wide basisAl/ and in fact most i3cidence studies consider only the impact of the average property tax rate.2 In the case of Colombia these excise effects are likely to be of some importance, since business inventories and most machinery are excluded from the property tax base; since exemptions of private and public lands are substantial;_/ and since effective tax rates vary between jurisdictions (see Table 5) and within jurisdictions. The regional rate differences are partly due to differences in the legal rates, which are determined by local jurisdic- tions subject to uniform upper limits set by national legislation. But more Table 5: STATUTORY AND EFFECTIVE PROPERTY TAX RATES IN THREE COLOMBIAN CITIES (percentage) /a /b /c Bogota - Cali Cartagena Statutory Rate/1 1.52 (1972)/2 1.h2 (1973) o.84-1.75L (1972) Effective Average R*e o.57 (1972) 0.45 (1973) 1.04 (1972) on Assessed ValueLt Effective Average Rate 0 58 (1969) n.a. 0.82 (1969) on Es mated Market 0.49 (1972) n.a. 0.55 (1972) valueLP.Oh 192 ..0.5(92 /1 Improved urban properties; the base consists of the capital value of land and improvements. /2 Including rates charged by Regional Development Corporatian (CAR in Bogota, CVC in Cali). /3 Graduated so that the statutory rate declines as one moves from low to high property value classes. /4 Computed as the ratio of property tax revenues to assessed property value. 79 Computed as the ratio of property tax revenues to estimated market value of properties. Source: /a J. F. Linn, "Bogota," op. cit. 7 Data collected in Cali. Tc J. F. Linn, "Cartagena," op. cit. 1/ Aaron (1974), op. cit., p. 218, and Dick Netzer, "The Property Tax: Discussion," American Economic Review, Vol. 64 (2), May 1974, p. 231. 2/ E.g., Charles E. McLure, Jr. (1973), op. cit; and Joseph A. Pechman and Benjamin A. Okner, Who Bears the Tax Burden? The Brookings Instituttion, Washington, D.C. 1974. 3/ In 1969, 12.8 percent of all assessed urban property value was exempted from property taxation.  - 13 - importantly, differences in collection performance and assessment ratios account for the differences in effective property tax rates.!/ The excise effects cannot be assessed in the absence of detailed information on relative capital intensities of taxed versus non-taxed activities; on the elasticities of factor substitution and commodity demand; and on the degree and nature of correlation between effective property tax rates and regional per capita inoDmes. One small piece of the puzzle is shown in Table 5, which indicates that Bogota, which certainly has a higher per capita income than Cartagena, enjpys a lower effective property tax rate. j To the degree that this results in lower relative prices for goods and services traded only locally, and/or in higher wages in Bogota relative to those in Cartagena, the differential tax rates will worsen the overall income distribution. However, it is doubtful that there is in general a negative correlation between effective property tax rates and per capita incomes for all Colombian muni- cipalities.._/ Further study would be required to ascertain more precisely the direction and importance of such excise effects, but until convincing evidence can be found to the contrary, it is probably safe to conclude that they roughly off-set each other and leave the measured overall incidence of the property tax roughly unaffected. More damaging to the new view of property tax incidence, parti- cularly in a country like Colombia, is the lack of justification for the assumptions of complete inelasticity of factor supplies and perfect competi- tion in the rental and commodity markets. In a small country with substantial foreign investment and access to foreign capital mark s capital cannot be assumed to be in fixed supply, even in the short run._ Furthermore, in the 1/ The intra-jurisdictional variations result from differential tax rates applying to urban and rural properties respectively; and from wide dif- ferentials in the effective assessment ratios, given the fact that in most larger Colombian cities reassessments proceed sequentially by areas within the city, rather than covering the whole urban area each time. 2/ Estimates put the per capita income in Bogota at approximately Col$7,000 and in Cartagena at Col$5,000 in 1970 (Linn, "Bogota," op. cit., and "Cartagena," op. cit.). 3/ Aaron (1975), op. cit., p. 45 cites estimates which would indicate a positive correlation for the U. S., but they appear inconclusive. Note also that regional differences in effective property tax rates may be offset by differences in local spending on public services; see Aaron(1974), op. cit., p. 213, and Earl R. Rolph "The Property Tax Discussion," American Economic Review, Vol. 64(2), May 1974, p. 233. 4~/ Proponents of the new view have recognized the possibility of long-run flexibility of capital supply through a variable savings ratio, but the importance of international capital mobility has generally been neglected. Two recent exceptions are McLure (1975), op. cit., and Richard M. Bird (1976), "The Incidence of the Property Tax: old Wine in New Bottles," paper presented to Symposium on Property Taxation, sponsored by the Ministry of State for Urban Affairs, Toronto, Feb. 17, 1976, p. 17.  - 14- rapidly expanding cities of Colombia urban land cannot be assumed to b in fixed supply, especially in the longer run, since lhe rate of conversion of low-taxed rural land into highly-taxed urban landa/ is like3,y to be affected at the margin by changes in the property tax differential. As a result, the property tax in Colombia is almost certainly less progressive than McLure's estimate following the new view would indicate. This conclusion is further strengthened by the observation that rental and commo 5 ty markets are far from perfect in Colombia as in most developing countries.- Thus at least a part of the property tax is likely to be passed on to renters and consumers which also would tend to lessen the progressivity of the property tax. Despite these limitations of incidence estimates based on the new view, one would err in the opposite direction in accepting the traditional view and incidence estimates based thereon. Land and capital, while not entirely fixed in supply, are not likely to be available in perfectly elastic supply to the country as a whole, as is assumed under the traditicnal approach. In tLe case of land, the marginal costs of expansion at the urban periphery are certainly rising in the short run, and possibly even in the long run, as cities expand into areas which are more inaccessible and more difficult and costly to service.4/ In the case of foreign capital, the well known pattern of rising costs of freign borrowing which has been observed for other developing countries,- is certain to apply also to Colombia. Furthermore, while market imperfections as a result of monopoly power exist, they are not likely to be 1/ In Bogota, D. E. rural land is taxed at a rate of 0.695 percent while residential urban land is taxed at a rate of 1.520 percent in nominal terms. 2/ For a general statement d this proposition, see P. Mieszkowski, "The Property Tax: an Excise Tax or a Profits Tax," Journal of Public Economy, Vol. 1, April 1972, pp. 73-96. See also Richard M. Bird (1974),Txing Agricultural Land in Developing Countries (Cambridge, Mass.: Ralvara University Press 1974), pp. 167-168 for a discussion of the flexibility of rural land supply. 3/ On the imperfections in the housing markets of less developed countries, see Orville F. Grimes, Jr., Housing for Low Income Urban Families, (The Johns Hopkins University Press, forthcoming), where it is stated that rent payments implying rates of return on capital of 30 to 100 percent per annum have been observed for low income families in main LDCs. The figures are illustrative of the often huge quasi-rents evident in housing markets of LDCs. For a discussion of the effects of market imperfections on property tax incidence, see Richard A. Musgrave, "Is a Property Tax on Housing Progressive?" American Economic Review, Vol. 6h(2). 4/ Casual observation of the geographic conditions of cities like Bogota, Cali, and Cartagena would confirm this view. L/ See for instance Gary Pursell, "Notes on a Shadow Discount Rate for the Ivory Coast," IBRD Development Research Center, mimeo, December 197h.  so pervasive as to, allow the extreme traditional shifting assumption. Finally, the Phase II Study results were based on current income, not permanent income estimates. Consideration of permanent income would tend to make the incidence of the property tax less regressive, since the permanent income elasticity of demand for housing can be assumed to be higher than the current income elasticity, and probably in the neighborhood of unity. / As a result one would have to con- clude that the incidence of property taxation on residential property would tend to be less regressive than estimated in the Phase II Study, and possibly neutral. The regressive impact of the tax on industrial and commercial properties may remain important, but as Bird has pointed out full forward shifting is not likely "owing to the different relative imqrtance of taxed structures in the production of various goods and services."V/ This may be of particular importance in a country like Colombia, where a substantial portion of commercial and service activity in the informal, low-income sector is likely to be carried on outside (taxed) premises or where property in low-income neighborhoods is underassessed (or the tax on it is undercollected).2/ These arguments present further reasons to conclude that the property tax is likely to be less regressive than appears from the Phase II estimates. The most appealing way of resolving the apparant conflict between the traditional and the new view of property tax incidence was suggested by Bird, who observed that "the traditional view deals, in effect, with the incidence of a tax in a particular locality; the new view, on tpe other hand, focuses on the incidence of the tax in the country as a whole ."IA/ This results from the realization that the chief "traditional" assumption (elastic factor supplies) is morenearly met in the case where an individual jurisdiction changes its property tax rate, while the chief "new" assumption (fixed factor supply, is more appropriate for an entire country. One may therefore conclude that while the two estimates of the property tax incidence in Colombia, which were here reported, reflect extreme assumptions, and therefore unlikely sets of circum- stances, the Phase II estimates are more nearly correct in predicting the incidence of a change in the property tax rate which is restricted to a parti- cular jurisdiction, such as Bogota, while McLure's estimate is more appropriate for the analysis of the property tax incidence on a country-wide basis. Given the caveats previously introduced in the discussion of each "view,"i one may conclude that the expected incidence of a property tax change in a particular jurisdictions in Colombia is likely to be somewhat regressive, with neutrality probably representing the upper limit. In contrast, the expected incidence of a property tax change on a nation-wide basis, such as was planned under a recently proposed law for eduction finance, would probably be progressive, with neutrality representing a lower boundary. 1/ Evidence on housing demand elasticities for LDCs is scant; see Grimes, op. cit., for some data which appear to suggest that the current income elasticity is generally in the neighborhood of unity or slightly below. 2/ Richard M. Bird (1975),"Who Pays the Property Tax?" Policy Paper No. 12, Institute for Policy Analysis, University of Toronto, October 1975, p.13. 3/ I am indebted to John C. English for this point. 4/ Bird (1976), op. cit., p. 13.  - 16 - 2. The Valoriza-ion Tax Another municipal tax, which is related to property taxation and which has become of growing importance in Colombia, particularly in the larger cities, is the valorization tax, a special assessment levy designed to recoup the cost of public infrastructure projects../ The distributive impact of this tax is likely to differ from that of the property tax, since it is linked directly to the costs, and indirectly to the benefits, of a public project. Somewhat superficially, Rhoads and Bird have concluded that the valorization program tends to be neutral in its incidence, "for all tax payers would receive benefits equal to or greater than the payment.",2 Doebele has given some greater, attention to the distribution effects of valorization in con- sidering in qualitative terms the improved access of low income groups to employment and housing; the payments systems; the force sale of property; premature subdivision; and other effects. He concludes that, at least for Bogota, "redistribution is small, and mixed bag."3/ Unfortunately, neither of these two studies provides a comprehensive and quant-tative analysis of the incidence of valorization programs. A comprehensive approach to this question would have to start from the realization that the incidence of a valorization program depends on whether it is complementary with, or whether it substitutes for, other revenue sources and expenditure programs. Four cases may be distinguished: First, the valorization program may be complementary both on the revenue, and on the expenditure side, in the sense that it raises additional financial resources over and above what would have been raised without it; and that it generates investment projects in addition to those which would have been carried out without it. The program would then be regressive, neutral, or progressive depending on whether the net benefit of the program is distributed more or less equally than the existing income distribution. Second, the valorization program may generate additiomal resources which are used to carry out projects that would otherwise have been carried out from general resources. The distributive effect of the additional expenditures which are financed from the general resources that are thus freed by the program, may offset on reinforce the distributive impact of the net benefit generated by the valoriza- tion program. Third, if valorization revenues substitute for general resources which would have been raised in the absence of the program, and if on ithe expenditure side the projects that are thus financed by valorization would have been undertaken anyway, then the program is progressive or regressive depending on whether the supplanted revenue source was more or less progres- sive than the valorization tax itself. Finally, if valorization revenues 1/ An early description and analysis of the valorization system is found in W. G. Rhoads, and R. M. Bird, "The Valorization Tax in Colombia: An Example for Other Developing Countries?" in A. P. Becker, ed., Land and Building Taxes that Effect an Economi2 Development, University of Wisconsin Press, Madison, 1969. For nore recent and more detailed reviews of the valorization system, see W. A. Doebele, "Valorization Changes as a Method for Financing Public Works in Bogota, Colombia," World Bank, mimeo, 1975; and J. F. Linn, Bogota, op. cit., and J. F. Linn, "Cartagena," op. cit. 2/ Rhoads and Bird,op. cit., p. 236. 3/ Doebele, op. cit., p. 1h6.  - 17 - again substitute for other revenues, and lead at the same time to an expendi- ture program which is biased more or less heavily in favor of high income groups than would have been the case otherwise, then the incidence of the program depends on the relative incidence effect of the displaced revenue sources and expenditure programs. This brief, and incomplete review of possible incidence effects of a valorization scheme reveals the complexity of analysis required of a comprehensive approach to the problem. Unfortunately there exists virtually no evidence on the crucial issue of the complementary or substitutive effects of the Colombian valorization program. On the basis of rather scant evidence, Rhoads and Bird inferred that on the revenue side the valorization program has been complementary. While this may be true in the short run, it is less clear that in the long run valorization taxes are complementary to other revenue sources. To the extent to which they appropriate part of the better- ment value of public works, they reduce the degree to which land values rise and thus reduce the growth in the property tax base and revenues..Z/ Given the lack of a well defined theoretical framework and the lack of empirical information, a quantitative approximation of the incidence of valorization programs in Colombia is not feasible, but this reains without doubt one of the more challenging areas for further research.' 3. Other Local Taxes Turning then to the industry and commerce tax one finds that its rate structure and base varies even more between cities than is the case for the property tax.4/ The industry and commerce tax is assessed on commercial and industrial assets,>/ on the value of sales, or on other indicators of industrial and commercial capacity (horse power of machinery installed, rental 1/ Rhoads and Bird, op. cit., p. 217. 2/ One may however argue that without valorization programs the infrastructure expenditures which lead to land value increments could never have been generated in the first place, since in the absence of a satisfactory muni- cipal bond market, municipalities would not have been able to raise the necessary resources. 3/ McLure (1975), op. cit., includes the valorization tax with the property taxes. This is only justified if in fact the valorization tax is a sub- stitute for the property tax and the same expenditure program is followed with, as without the tax. And even then it would be more appropriate to apply the analysis of the "old view" on property tax incidence, rather than the "new view," since valorization programs are in use only in relatively few cities. L/ This is largely due to the fact that this tax is entirely under local jurisdiction, in contrast to the property tax. 5/ In this case, real estate and assets from branches in other municipalities are generally exempted.  - 18 - value, etc.)../ To the extent that the tax is levied on assets or rental value, similar arguments apply concerning its incidence as in the case of the property tax. In particular, it is probably not correct to assume full forward shifting as is done in the estimates by McLure and those of the Phase II Study (see Table 6). As a result, the industry and commerce tax is probably less regres- sive than estimated in both studies. Colombian municipalities further impose numerous minor taxes and stamp duties. Their revenue potential and distributional impact is negligible, with the exception of the municipal vehicle tax, which appears to combine salutory efficiency and distributive effects with a considerable revenue potential.2/An estimate of the distributive impact of this tax in Bogota D.E. (Table 6) shows it to be highly progressive especially in the income classes above Col$60,000. So far the tax has not been utilized by Colombian municipalities to any major extent 3/ which may be explained on two grounds. First, and most important, national import tariffs on motor vehicles are very high; as a result local governments may feel themselves politically unable to add further to the domestic prices of automobiles. Second, problems of administration were encountered especially in Bogota, wlhere it is suspected that a substantial proportion of motor vehicles are registered in municipalities surrounding the Special District, since lower rates apply there.4/ 1/ A survey of statutory regulations concerning rates, bases, etc. of this tax is presently under preparation in the Colombian National Planning Department. Note that wiere the tax is levied on rental value, it amounts to an additional tax on real estate in commercial or industrial use, although the valuation of the base is likely to differ from the assessment of properties in the local cadastre. 2/ In terms of efficiency effects a local vehicle tax is probably preferable to a uniform national tax or import tariff, since differentially higher rates in the urban areas could account for externalities there resulting from congestion and pollution, without discouraging the use of motor vehicles in rural areas, where these externalities are negligible. For a discussion of these issues, see Charles E. McLure, Jr. "Automotive Tax Reform", in Malcolm Gilles, ed. Fiscal Reform for Colombia, International Tax Program, Harvard Law School, Cambridge, Mass. 1971. A metropolitan vehicle tax has been utilized very effectively as a main revenue raiser for local government in Jakarta, Indonesia. See J.F. Linn, Roger Smith, and H. Wignjowijoto "Urban Public Finances in Developing Countries: A Case Study of Metropolitan Jakarta," IBRD, forthcoming. 3/ For the departmental capitals in Colombia budgeted revenues under the heading of "Circulacion y Transito" amounted to only 4.5 percent _r local tax revenues in 1973. In Bogota, the vehicle tax contributed 5.1 percent to Pistrict Tax Revenues in 1972; and in Cartagena its contribution amounted to 2.1 percent; from J. Linn. "Bogota". op. cit. and "Cartagena",op. cit. 4/ Improved enforcement practices and alternative means of taxing motor vehicle use in urban areas are conceivable remedies to the second problem. tor instance, a tax on parking facilities and other congestion charges could be implemented. Also, a wider regional administration of vehicle taxation would reduce the administrative problems experienced in Bogota,D.E. See Phase II, op. _ct. Appendix E.  Table 6: ESTIMATED RATES OF MUNICIPAL AND DEPARfTYENTAL TAXES, 1970G 5, Municipal Taxes Departmental and District Taxes Total inc=e 4/ HisSe.hold Property Tax Ind. and Corn.. Tax Vehicle Tax Sub-total Alcoholic Bev. Tobacco Taxes Sub-total (Cci S p.a.) Phase II McLure 2/ Phase II McLure Phase II Phase II McLure 3/Phase II McLure Phase II McLure Phase II McLure Phase II tcLure c- 12,000 0.85 0.01 0.25 0.40 0.02 1.12 0.41 0.12 0.10 0.15 0.64 0.28 1.30 1.40 1.71 l2,000- 24,000 0.731 0.01 0.2) 0.30 0.03 1.05 0.31 2.21 1.30 0.21 0.80 2.30 2.40 3.35 2.71 24. JZ3- 36,C'jO 0.70 0.01 0.27 0.30 0.03 1.00 0.31 1.72 1.20 0.23 0.40 1.92 1.80 2.92 2.11 36, ('J- 60,000 0.77 0.03 0.21 0.30 0.02 1.03 0.33 1.97 1.20 0.16 0.40 2.16 1.80 3.19 2.13 c6,0u0-L74,000 0.65 0.5-0.6 0.2i 0.30 0.05 0.91 0.80-0.90 0.16 1.00 6/ 0.12 0.30 6/ 0.39 1.40 6/ 1.30 2.20-2.20 1?4,000 and over 0.53 1.7-2.3 0.1-- 0.30 0.13 0.86 2.00-2.60 0.15 0.20 o/ 0.07 0.30 6/ 0.40 0.60 6/ 1.26 2.60-3.20 Pverase 0.66 0.6-0.7 0.21 0.30 0.08 0.95 0.9-1.0 0.75 0.80 0.14 0.60 0.93 1.70 1.88 2.60-2.70 -. a_ II iostmEates apply to Bogota, D.E, (total income); McLure estimates apply to the urban sector. 2/ lncl._des valorization. 3. Includes or.ly the property and the industry and commerce taxes. 4. lncludes departmental sales and gasoline taxes. 5/ Sum of -.unicipal ard departrmental sub-totals. 6/ interpolated by approximation since different boundaries for income classes are used. Source: Pnase .II o0. cit. McLure, 1973, 0o:. c