DISCUSSION PAPER Report No.: UDD-55 AN ECONOMIC ANALYSIS OF LOW-COST HOUSING OPTIONS IN EL SALVADOR by Marisa Fernandez-Palacios and Michael Bamberger May 1984 Water Supply and Urban Development Department Operations Policy Staff The World Bank The World Bank does not accept responsibility for the views expressed herein which are those of the authors and should not be attributed to the World Bank or to its affiliated organizations. The findings, interpretations, and conclusions are the results of research supported by the Bank; they do not necessarily represent official policy of the Bank. The designations employed, the presentation of material, and any maps used in this document are solely for the convenience of the reader and do not imply the expression of any opinion whatsoever on the part of the World Bank or its affiliates concerning the legal status of any country, territory, city, area, or its authorities, or concerning the delimitation of its boundaries, or national affiliation. Marisa Fernandez-Palacios is an economist with the Latin American Programs Department of the World Bank. Michael Bamberger is a sociologist with the Water Supply and Urban-Development Department of the World Bank. The fieldwork for this report was conducted as part of an IDRC-IBRD supported evaluation of World Bank shelter projects in El Salvador, Zambia, Senegal and the Philippines. The authors are grateful to the staff of the Fundacion Salvadorena de Desarrollo y Vivienda Minima for assistance in data collection; and to the Instituto de Vivienda Urbana, Fondo Social Para la Vivienda and the Financiera Nacional de Vivienda for providing information on the costs of their projects, Umnuay Sae-Hau and Robert Lynn provided invaluable assistance with data analysis. ABSTRACT This report applies the economic and social methodology for project analysis developed by Lyn Squire and Herman Vaii der Tak to the evaluation of low-cost housing options in El Sa:vador. Data was collected in 1977-78 on the 9 main housing options accessible to urban households in the lower half of the income distribution. The options included conventional housing, both single unit and multi-family; formal sites and services and upgrading; and the three main types of informal housing (squatter settlements, rental rooms and extra-legal subdivisions). For each option information was collected on costs (from both the executing agency and the occupants) and on imputed benefits (through imputed rents). Efficiency and social parameters were calculated using the Squire- Van der Tak methodology. The housing options were then compared, using the Squire-Van der Tak approach to cost-benefit analysis, in terms of both efficiency and social benefits. The latter were estimated using weights for consumption by different income groups, and for impacts on consumption and inv7estment. One*of the innovations contained in the report is the estimation of private costs and benefits. These are derived by computing all costs and benefits from the point of view of the occupant of the unit rather than from the point of view of the nation. It was found that the rank ordering or projects (in terms of their economic rate of return and their net present value) was different for public and private estimates. The main reason for the difference was that most public housing programs contain an element of subsidy, either through not covering all costs or through below-market interest rates. These subsidies raise the private rate of return and encourage households to purchase or rent public housing options whose ec'onomic efficiency is lower than that of some of the informal sector options. TABLE OF CONTENTS SUMMARY OF THE REPORT 1 CHAPTER ONE HOUSING POLICIES IN EL SALVADOR AND SHELTER OPTIONS ACCESSIBLE TO LOW-INCOME FAMILIES IN SAN SALVADOR 17 1. Public policies with regard to urban land, services and housing 17 2. The main types of shelter provided through the informal housing market. 19 3. Housing produced by the FSDVH. 26 4. The accessibility of public and private housing programs to low income families in San Salvador. 28 CHAPTER TWO METHODOLOGY 33 1. The estimation of efficiency and social parameters. 33 2. The methodology for efficiency analysis. 39 3. The methodology for social analysis. 45 4. The methodology for the evaluation of private costs and benefits. 52 5. Sample design and the sources of information on the costs and benefits of each housing option. 54 CHAPTER THREE RESULTS OF THE EVALUATION 57 1. Efficiency analysis 57 2. Social analysis 61 3. Effect of different assumptions on the value of land. 68 4. -Analysis of private costs and benefits. 74 5. Conclusions and recommendations. 81 6. Suggestions for future research. 85 ANNEXES 1. Estimation of parameters. 91 2. The estimation of costs and benefits for each of the housing options studied: 131 2.1 FSDVM: Basic Core Unit: El Pepeto 132 2.2 FSDVM: Serviced Lot. San Jose del Pino 148 2.3 IVU: Multi-family apartments. 156 Amatepec 156 2.4 IVU: Single family units. Quetzaltepeque 165 2.5 IVU: Squatter upgrading, La Fosa 173 2.6 FSV: Single family units, San Jose de Soyapango 181 2.7 Squatter settlement on private land. (Tugurio) 190 2.8 Extra-legal subdivision (Colonia ilegal) 197 2.9 Tenement (Meson). 210 REFERENCES CITED IN THE TEXT. 215 SUMMARY OF THE REPORT 1. This study applies the socio-economic approach to project analysis developed by Lyn Squire and Herman Van der Tak, to the evaluation of the First World Bank financed sites and services urban housing project in El Salvador. The evaluation is based on a comparison, made in 1978, of two sites and services options with 7 other housing options provided either by public agencies or through the informal housing market, all of 'which were theoretically accessible to families in the lowest 50-60% of the urban income distribution. The study is intended to contribute to our understanding of the potential role of sites and services and squatter upgrading in urban shelter programs as well as to evaluate the utility of the Squire-Van der Tak methodology in a different context from those in which it has traditionally been employed. 2. The first El Salvador Sites and Services Project, which was executed by the Fundaction Salvadorena de Desarrollo y Vivienda Minima (FSDVM) was one of the first low-income urban housing projects to be supported by the Bank. As this was also one of the first projects to be completed it i-s hoped the results of the evaluation will be helpful in the definition of future Bank policies in the field of urban housing. Although the primary objectives was to evaluate the FSDVM project, the study included a systematic comparison of 9 different housing options in the formal and informal sectors. Consequently the report,provides detailed comparative information on a number of alternative housing options such as squatter upgrading and tenement housing which present 1/ Lyn Squire and Herman G. Van der Tak Economic Analysis of Projects, published for the World Bank by the Johns Hopkins University Press, 1975. -2- alternative approaches to the provision of shelter accessible to the trban poor. . 3. The methodology itself has been widely discussed and criticized within and outside the World Bank and it is hoped this, report will contribute to the discussion by examining the contribution to policy planning when the methodology is applied in a somewhat different context from earlier studies. Some of the methodological issues which are discussed include: the extent to which a technique based on marginal analysis can contribute to the definition of general policy;- the validity of imputed market rents as an estimate of benefits; the utility of distribution weights to evaluate options in terms of their accessibility to low-income groups; and the problem of potentially important benefits which cannot be quantified within this methodology. 4. The urban housing market presents a number of different methodological questions from those found in earlier applications of the Squire-Van Der Tak methodology. The fact that the study is part of a four country longitudinal evaluation of low-income housing 1/ means that the researchers had access to types of information on costs and to some extent on benefits which are not usually available when this methodology is applied. For these reasons, it is hoped that the study will offer an opportunity to evaluate the potential value of the methodology in a new 1/ The five year evaluation program, jointly sponsored by the International Development Research Center of Canada and the World Bank, covered Bank financed projects in El Salvador, Senegal, Zambia and the Philippines. The original project ended in 1980 although a one year extension was approved for Senegal and the Philippines. -3- context and hence will provide new insights into some of the problems of project evaluation. 5. An innovation not included in earlier applications of the Squire-Van der Tak method is the analysis of private, as well as public, costs and benefits. The study complements the normal estimation of public costs and benefits with an estimation of the costs and benefits accruing to the owner-occupier of each housing type. If there are any price distortions, resulting for example from government housing subsidies to certain groups, then the individual household may make a choice among housing options which, although maximizing his or her private utility, is not the best choice from the point of view of the nation. In this way it is hoped the comparison of public and private costs and benefits will help us understand better the dynamics of the ho,ising market. Methodology 6. The study covers the 9 main housing options potentially accessible to the poorest 50 percent of families in San Salvador. The sample includes three types of public low-income housing (one multi-family medium rise apartment and two types of singlo-family units); three options based on upgrading or sites and services and the three main types of housing pro(.,ced through the informal market (squatter settlements, tenement housing and extra-legal subdivisions). 7. For each option a representative project was selected. Cost data was obtained from the producer, and interviews were conducted with a sample of families to obtain information on their housing costs and income. These souces were complemented by the sample survey work already conducted as part of the World Bank-IDRC supported Monitoring and Evaluation Program and by other secondary sources. -4- 8. The estimation of Efficiency and Social paramters follows closely the Squire-Van der Tak methodology. Three indicators were used for the Efficiency analysis: the Economic Rate of Return, Net Present Value and Net Present Value divided by the discounted total costs. For the social analysis two different assumptions were made about the priority given by the government to income redistribution; the first assumed a low ptiority (with n, the weighting factor being equal to 0.5) and the second assumed a higher but still only moderate priority (n=1.0). The analysis of the social impact of projects on income distribution and dn the preference for consumption over investment was made under both of these assumptions. 9. An important innovation was the introduction of a Private Benefit-Cost analysis in which estimations of costs, benefits and net benefits were made in terms of costs and benefits actually incurred by households. This is a potentially useful method for evaluation of the impact of price distortions on the efficiency of resource allocation. 10. The nature of the study introduces a number of interesting methodological features. The first unusual feature is that nine different housing options are studied, rather than the two or three which are normally compared. This does not affect the logic of the analysis but introduces a greater complexity into the interpretation of the results. 11. A second feature which complicates the analysis is the fact that the options are not exclusive. It is quite possible to recommend the simultaneous construction of several of the options, rather than selecting between mutually exclusive alternatives as is usually the case with this type of analysis. This permits a discussion of the extent to which a -5- technique which indicates which alzernative produc,s the highest marginal net benefits can contribute to general planning of an overall housing policy. Structure of the Report 12. The Report contains three chapters and two appendices. The detailed computations have been assigned to the appendices. Chapter One present an analysis of the low-income housing market in El Salvador and shows the ways in which housing policies have restricted the access of low-income households to public housing. This chapter describes each of the types of housing option included in the analysis. 13. Chapter Two presents the methodology and is divided into five sections. Section 1 explains briefly the methodology for the estimation of social and efficiency parameters. The estimating procedures are presented in greater detail in Annex 1. Sections 2 and 3 explain the methodology for efficiency Lnd social cost-benefit analysis. Section 4 explains the procedures for estimating private costs and benefits, and Section 5 explains the sample design and procedures for data collection. 14. Chapter Three presents the results of the analysis and discusses their policy implications. Recommendations are also included for future research. 15. Appendix 1 presents the methodology for the estimation of efficiency and social parameters. Appendix 2 explains the procedures used for the estimation of costs and benefits for each of the 9 housing options studied. All of the main tables are included so that the interested -6- reader can repeat any of the analysis using different assumptions. Summary of the Main Findings 16. Table 1 presents the main findings of the analysis of public costs and benefits. Each of the 5 indicators mentioned in paragraph 7 is included. Table 2 ranks each housing options from 9 (highest) to 1 (lowest) on each of the five indicators. The final column presents the average rank of*each project on all 5 indicators. 1/ 17. The housing projects are classified into three groups. The first group, upgrading and sites and services includes the IVU (Urban Housing Institute) slum upgrading, and two of the FSDVM (Salvadorean Low-Cost Housing Foundation) sites and services projects. The second group includes the three types of housing produced through the informal market and the final group includes three traditional public low-cost housing programs (one medium rise and two single family units). 18. The highest ranking group contains the two FSDVM sites and services projects which had the highest and second highest ranks, and the IVU upgrading program, which was the highest ranked government housing program. 19. The second highest ranking group includes the housing options generated by the informal housing market. The extra-legal subdivision and squatter settlements tied in third highest rank. It should be clarified 1/ The validity of the use of the average rank as a summary figure is indicated by the high degree of agreement between the different ranks. Kendal's co-efficient of Corcordance (W) has a value of 0.8019 indicating a high degree of agreement between the different ranks. -7- that the study deliberately included one of the more stable squatter settlements as the intention was to evaluate-the effectiveness of this option under relatively favorable circumstances. The tenement houses(mesones) have a much lower ranking than the other two types of housing. The reasons for this lower ranking are not entirely clear, but as the meson is located on prime urban land in the city center, it is much more affected by changing land values. This greater sensitivity to the assumptions made about the value of land, means the rates of return are more difficult to estimate precisely. 20. Another point of interest is the different response of the basic core unit and the serviced lot to the social evaluation. When the two projects are compared in terms of economic efficiency the difference is relatively small with the core unit having a slightly higher rate of return (33 percent compared with 28 percent). When, however, a distribution weight of 1 is used, the difference becomes much greater (42. compared with 28%). This supports initial findings in other countries that poorer families prefer a higher level of construction whereas the slightly richer families prefer to receive only a serviced sites so that they have more flexibility to build according to their own preference. 21. Tables 3 and 4 summarize the main findings of the evaluation of private costs and benefits. The analysis was conducted for all 8 owner-occupied housing options, but not for mesones (tenements). Table 3 compares the results of public and private evaluation of Net Present Value, Net Present Value divided by costs and the Internal Rate of Return. In all cases it can be seen that private costs are lower and with one exception the Internal Rate of Return is higher. This suggests that all projects permit occupiers to enjoy some measure of direct or indirect subsidy. In the case of public programs this is a direct subsidy, whilst in the case of illegal subdivisions and squatter settlements this takes the form of families not having to pay the full economic costs for certain services (often water). Although all projects enjoy some measures of subsidy, this is much greater for the public programs. 22. The effect of public program subsidies can be seen in Table 4 which presents the ranking of projects from the public and private standpoint. It can be seen that in the private analysis the ranking of public housing programs increases considerably. This is a reflection of the fact that families in public programs are not paying the full economic cost. As most public programs are only accessible to families in the upper half of the income distribution these subsidy policies have the effect of producing a negative bias in terms of distribution of housing benefits.to the urban poor. 23. An additional analysis was made for the extra-legal subdivision to determine the impact on private benefit-cost ratios of providing loans on the same terms as those received by FSDVM participants. The effect would be quite dramatic as the private IRR would increase from 14% to 35% and both NPV and NPV/Cost would be more than quadruple. Policy Implications 24. Care must be taken in the'interpretation of the findings as the study is based on the analysis of specific projects, some of which may not be the most cost effective examples of the option, and because many of the important variables are subject to considerable measurement error. At the same time the consistency of the findings over a number of independent indicators suggests the relative magnitudes are indicative of important -9 differences between the projects studied. Within this context a number of interesting policy implications are suggested. 25. Both upgrading and sites and services are potentially effective ways of making housing accessible to the urban poor. On all of the efficiency and social indicators these two projects are consistently ranked higher than traditional housing programs. The fact that the higher rankings hold for both efficiency and social analysis suggest that these are cost-effective ways to provide shelter for the poorest 50 per cent of 1/ the urban population. - 26. Both extra-legal subdivisions and squatter settlements have relatively high rankings, suggesting that the informal housing market is able to provide, with no financial or technical assistance, housing for the urban poor which ranks higher than most traditional programs. The analysis also shows that simply by offering financial terms similar to those provided to FSDVH the private benefits would double or on some indicators quadruple. This suggests this type of project might be 1/ A similar conclusion was reached in a later study of another FSDVM project in Santa Ana, in which it was estimated that by moving from the informal sector to the FSDVM project, participants increased their utility by about 20%. See John Quigley "The distributional consequences of stylized housing programs". Urban and Regional Report No. 80-18. Urban and Regional Economics Division. The World Bank. 1980. - 10 - interesting to consider for support by national or international organizations. The project has the additional advantage that a large number of households could be reached relatively easily and quickly and without the need to establish a very complicated organizational structure. 27. The relative rankings for tenement houses are in most cases much lower, due mainly to the high cost of land and the income transfers to the high income tenement owners. The proximity to places of work and the fact that most of the buildings are structurally sound, makes this another attractive option if a satisfactory approach to the cost of land can be found. The FSDVM has already produced promising results with a pilot upgrading project for one meson which was converted into a .. 1/ condominium.- 28. Some interesting policy issues arise from the different social evaluations of the FSDVM basic core unit and serviced lot. Although the two models have similar efficiency levels, the higher income level of families choosing the serviced site, lowers considerably the social rate of return for this option. This suggests that in an overall shelter strategy, the serviced site should perhaps be targetted to a slightly higher income group than the core unit. 29. A more general policy issue arises from the comparison of public and private benefits. Public housing programs contain considerable subsidy elements thus making them more attractive to consumers in terms of private net benefits, than is merited by the efficiency net benefits. The 1/ This project is described in Bamberger, Gonzalez Polio and Sae-Hau "Evaluation of Sites and Services Projects: The Evidence from El Salvador." World Bank Staff Working Paper. No. 549 1982. - 11 - distortions are made worse by the fact that many of the most highly subsidized programs are mainly occupied by higher income groups so that the subsidies contribute to a more regressive distribution of benefits. This suggests that current housing subsidy policies should be reviewed so as to either reduce the subsidy element or to distribute it more equitably among different income groups. 30. Finally the findings of this study suggest that this method of Economic Analysis can be a useful policy tool, particularly if private analysis is included together with the more traditional approaches. It is believed that similar studies would be useful to urban policy makers in many countries who are seeking ways to make housing accessible to the urban poor. Table 1: COMPARISON OF LOW-INCOME HOUSING OPTIONS IN SAN SALVADOR USING DIFFERENT ECONOMIC ANALYSIS INDICATORS RATE OF RETURN NET PRESENT VALUE Economic Social R/R R/R N=0.5 n=1.0 NPV(Colones) NPV/DTC 1/ Programs based on Upgrading or Progressive Development. FSDVM Basic Unit 33 26 42 4065 1.2016 FSDVM Serviced Plot 28 22 28 2329 0.7269 IVU Upgrading 18 15 15 1078 0.2640 Informal Housing Extra-legal Subdivision 22 24 22 1788 0.3509 Squatter Settlement 20 29 53 373 0.2972 Tenement 12 12 32 30 0.0141 Traditional Government Housing IVU - Multifamily Apartments 1 6.3 13 -1828 -.1304 IVU - Single Family Units 9 7 18 -606 -.0720 Fondo Social - Single Family Unit 13 12 20 452 0.0641 1/ Net Present Value divided by discounted total cost. This adjusts for differences in total investment. At Table 2: RANK ORDERING OF LOW-INCOME HOUSING OPTIONS IN SAN SALVADOR. COMPARISON OF DIFFERENT ECONOMIC ANALYSIS INDICATORS (9 = Highest value; 1 Lowest value) RATE OF RETURN NET PRESENT VALUE AVERAGE RANK Economic Social R/R ON ALL .R/R n=0.5 n-1.0 NPV NPV/DTC 1/ INDICATORS Programs based on Upgrading or Progressive Development FSDVM Basic Unit 9 7 8 9 9 9 FSDVM Serviced Unit 8 6 6 8 8 8 IVU Upgrading 5 5 2 5 6 5 Informal Housing Extra-legal Subdivision 7 8 5 7 7 7 Squatter Settlement 6 9 9 6 4 7 Tenement 3 4 7 3 3 3 Traditional Government Housing IVU-Multifamily Apartment 2 1 1 1 1 1 IVU-Single Family Units 1 2 3 2 2 2 Fondo Social-Single Family Unit 4 4 4 4 5 4 1/ Net Present Value divided by discounted total cost. This adjusts for differences in total investment. Table 3: A COMPARISON OF HOUSING OPTIONS IN TERMS OF PUBLIC-AND PRIVATE COSTS NPV AND IRR PUBLIC COSTS AND BENEFITS PRIVATE COSTS AND BENEFITS PROJECT Cost* NPV** NPV/COST IRR COST* NPV** NPV/COST IRR Colones Colones % Coloneb Colones % FSDVM Basic core 3383 4065 1.2016 33 2976 5368 (5157) 1.8038 (1.73) 129 (94) FSDVM-Serviced lot 3204 2329 .7269 28 2617 3635 1.389 45 IVU-Single family 8414 -606 -.0720 11 7085 4061 .5731 35 IVU-Appartments 14021 -1828 -.1304 9 8537 5254 .6154 54 FSV-Single family 7046 452 .0641 13 7007 4655 .6643 90 IVU-Squatter upgrading 4083 1078 .2640 18 3760 1857 .4939 25 Colonia ilegal 5096 1788 .3509 22 5223 713 .1369 14 Tugurio 1255 373 .2972 20 1227 639 .5208 26 Notes: Meson (tenements) not included in the comparison between public and private costs and benefits * Costs discounted at 12% ** Net present value (discounted at 12%) of Net Benefits Two different estimates are made for private benefit/cost ratios in the FSDVM Basic Core Unit project. The first estimate (a) make maximum assumptions about imputed rent during the first two years, whilst the second (b) makes minimum assumptions about imputed rent. See Section 1 of Annex 2 for a more detailed explanation. r * Table 4: COMPARATIVE INDICATORS OF PUBLIC AND PRIVATE COSTS AND BENEFITS FOR EACH HOUSING OPTION (PB=Public: PT=Private) COST NPV IRR COMPARISON OF RANK ORDERING (8=highest) (PB/PT) (PT/PB) (PT-PB) NPV/COST IRR Public Private Public Private Programs based on up- grading or progressive development. FSDVM Basic Unit 1.14 1.32 +61 8 8 8 8 FSDVM Serviced Plot 1.22 1.12 +17 7 7 7 15 IVU Upgrading 1.09 1.72 +7 4 2 4 2 Informal housing Extra-legal subdivision .98 .40 -8 6 1 6 1 L Squatter settlement 1.02 1.71 +6 5 3 5 3 Traditional government housing IVU - Multifamily- 1.64 2.88 +45 1 5 1 6 IVU - Single family 1.19 6.70 +24 2 4 2 4 FSV Single, family 1.08 10.3 +77 3 6 3 7 4i - 17 - Chapter l Housing Policies in El Salvador and the Shelter Options Accessible to Low-Income Families in San Salvador. 1. Public policies with regard to urban land, services and housing 1/ 1.01 Public housing policies have traditionally focussed on the provision of a relatively small number of high cost units whose access was largely restricted to families in the upper 40 per cent of the income distribution or who were employed in the formal sector of the economy. From 1960 to 1970 an average of 2,600 new housing units were produced annually in urban areas whilst an estimated 10,000 new households were formed every year. *It was estimated that in 1972, 55 per cent of the urban housing 6tock needed improvement or replacement, and in 1975 that 69 per cent of urban households lacked individual water supply and 62 per cent lacked sewerage connections. 1.02 Since 1950 three major public hoziing programs have been established. The first of these agencies was the Instituto de Vivienda Urbana (IVU) established in 1950. During the 1960's it launched a large-scale program of conventional mass-produced single family units and multi-family apartment buildings mainly directed to the middle income groups. By 1978 IVU had constructed more than 23,000 units with direct financing from the central government. Over the past few years IVU has faced a series of organizational problems which have slowed its rate of 1/ Much of the information for this section was taken from Bamberger, Sae-Hua and Gonzalez-Polio "Evaluation of the First El Salvador Sites and Services Project." Chapter 1. -18 - growth. At various times it has experimented with low-cost self-help housing projects, either through sites and services or squatter upgrading but, many of them have been short-lived and have not been integrated into its normal operating activities. It has proved difficult to integrate these projects into normal operations and they have not yet represented a significant proportion of total house production. One of these squatter upgrading projects was included in the present study. 1.03 The second government agency, the Financiera Nacional de Vivienda (FNV) was established in 1965 under the auspices of the Alliance for Progress as a means for channeling private savings into housing construction, through the establishment of savings and loan associations. By 1978 the FNV had financed 26,600 units at an average unit cost in 1978 of $20,000. The houses constructed have been mainly located in the Metropolitan area and are accessible only to families in the top thirty per cent of the income distribution. Most of the cosatruction has been of single-family units although recently a number of condominium apartments have been financed. 1.04 The third agency, established in 1973 is the Fondo Social para la Vivienda (Social Housing Fund). The FSV is financed mainly through a payroll tax and was established as an extension of the Social Security System. The program only covers household heads employed in the formal sector. By 1978 over 5,000 units had been constructed, but although many of the families paying the payroll tax had incomes below $280 per month, most of the units being financed cost between $4000 and $6000 and were only accessible to families in the top 30 or 40 per cent of the income distribution. Although the FSV has begun to experiment with variable - 1R - interest rates, which potentially provides a way to make housing accessible to lower-income families, in fact there is a tendency for the average cost of units financed to increase and for the program to become less accessible to the poorer sectors who are theoretically covered by the program. 1.05 The policy of only constructing public housing for middle and upper income groups is aggravated by a lack of a clearly defined land-use policy. Land is very scarce and hence expensive in El Salvador, and as there is no policy of price control or reserving land for low-income housing, the effect has been to make it extremely difficult to obtain land at a price affordable to poorer families. 2. The main types of shelter provided through the informal housing market 1.06 The previous section showed that access to public housing programs has essentially been restricted to families in the top 40 per cent of the urban income distribution. An inevitable consequence of this has been, as in most other developing countries, that most low-income families live in shelter provided through "the informal housing market." By "informal" is meant types of housing which either contravene one or more laws or regulations or which at best are tolerated by public authorities but not actively assisted. In this section we will describe the main types of informal housing in San Salvador. It is important to remember that by no means all families living in "informal" or even low-income housing are poor. Given the general housing shortage and lack of access to urban land, many middle-income families also occupy these same types of informal housing. The point is often made that some of the wealthiest residential areas of the city are also "illegal" in that they 20 also infringe certain planning regulations. From the point of view of housing research and policy the economic heterogeneity of many supposedly low-income areas proves to be a complex and intriguing issue as on the one hand it makes it more difficult to ensure the benefits of a project reach the low-income families for which it is intended, whilst on the other the interest of higher income families in these areas means that poorer families can potentially generate extra income from subletting or selling their properties. 1.07 It was estimated that the population of the metropolitan area of San Salvador in 1977 was 788,320 and that 54 percent lived in housing which was defined as inadequate. Table 5 shows that five types of inadequate housing were identified of which the most important was the rented rooms (mesones) followed by illegal subdivisions. 1.08 In comparison with other Latin American countries, an extremely low-proportion of families live free on land which has been invaded. The squatter settlements (tugurios) house only 8.6% of families with inadequate housing and less than 5 per cent of all families in the metropolitan area. In fact, even in the tugurios, a significant proportion of families are paying rent to the owner of the land which has been invaded so that the proportion who are living free is even lower. This contrasts very dramatically with the situation in many Latin American cities (for example Caracas, Lima, Rio, Recife, Mexico City) where a quarter or more of the total urban population may be living on invaded land and paying no rent. Table 5: ESTIMATED POPULATION AND NUMBER OF FAMILIES LIVING IN RENTED ROOMS, SQUATTER SETTLEMENTS, ILLEGAL SUBDIVISIONS, TEMPORARY CAMPS AND DETERIORATED HOUSES, SAN SALVADOR, 1977 Type of Housing Population Families Family Size Source No. % No. % Inadequate Housing 425716 54 92797 55 4.8 Rented rooms (mesones) 184910 23.4 48660 28.9 3.8 (1) Ministerio de Planificacion Illegal subdivision 161606 20.5 27672 16.4 5.8 (2) FSDVM Squatter settlements 38000 4.8 8064 4.8 4.7 (3) EDURES Camps 20300 2.6 4229 2.5 4.8 (4) EDURES Deteriorated housing 20900 2.6 4354 2.6 4.8 (5) EDURES Adequate Housing 362604 46 75542 44.8 4.8 (1) Ministerio de Planificacion TOTAL 788320 100 168521 100 Note: All figures were projected to 1977 by-the authors. Source: (1)"La SLtuacion de Vivienda. 1971-75," *Seccion de Investigaciones Muestrales. Ministerio de Planificacion, Cuadro No. 2, 1975. (2) "La Vivienda Popular en El Salvador," FSDVM, 1976. (3) "Rehabilitacion de Asentamentos Marginales," Informe Inicial, Fase 1, Pag. 3, EDURES, Noviembre 1976. (4) EDURES op. cit., Based on Informe de OMCOM, 1976. (5) EDURES op. cit., Based on "Encuesta de Mano de Obra," PREALC, 1974. 22 - 1.09 The reason for the marked difference in the case of San Salvador (and most other Salvadorean cities) is that land is very scarce and hence very little can be considered as "marginal". This has meant that other than small areas which are in ravines or are otherwise topographically difficult to develop, even the poorest families have been forced to pay for their accomodation. 1.10 The five types of informal settlements can be described as follows: Mesones (rented rooms) - the meson normally consists of between 5 and 50 rooms built round one or more central patios. Water supply and sanitary services are communal. Traditionally the mesones were converted middle-class houses but now new mesones are starting to be built. The majority of older mesones are located close to the center of the city and it would appear that people are prepared to pay a higher rent because of the location. In many cases the meson is the place of employment and it is typical for there to be more expensive rooms facing onto the street, thus permitting a small store or service to be installed. With the expansion of the city new mesones are now being built on the edges of the city. 1/ 1.11 In the majority of cases a family will rent a single room and there is an average of 2.8 persons per room. In 1977, the average monthly rent was estimated to be US$12, but there is a wide variation and approximately 20 per cent pay less than US$8 whilst at the other extreme 1/ In theory it is illegal to construct new mesones but the law has not proved effective. - 23 - the top 20 per cent pay maore than US$16. In about 80 per cent of the cases* the cost of water and light is included in the rent. In the case qf water, most tenants consider this to be a disadvantage rather than a benefit because the owner restricts the water supply often to two hours a day and families would prefer to have the opportunity to pay more for a better service. 1.12 Mesones vary greatly in quality and price, ranging from well built brick rooms with neatly kept patios and sanitary areas, to rooms constructed of mud and straw, where water enters for only two hours a day and where 10 or more families share one badly maintained pit-latrine. The quality of mesones tends to decline because the law theoretically forbids reconstruction or repair of mesones, and because many of the owners are seeking a maximum short-run profit and see no reason to invest in improvements. 1/ 1.13 Most of the mesones are located in the central areas of the city and in recent years programs of urban renewal have resulted in a large number being demolished, and many others have been destroyed in a large number of mysterious fires in which mesones and small stores have vanished. As a consequence, rents have been rising vertiginously and cases were found in the study where rents had doubled over a 3-month period. The study also revealed cases in which owners were deliberately locking the toilets or reducing the water supply to force families to leave so that the rent could be raised to the newly entering tenant. 1/ Many owners plan to sell the land when prices rise or to convert into middle-class apartments- 2/ The law prohibits increasing the rent to an existing tenant. - 24 - 1.14 The lack of government control over prices or quality combined with lack of alternative housing options has meant that the meson represents an important economic market producing an estimated gross annual rental income of approximately US$7.5 million in the metropolitan area. 1.15 Illegal Subdivision (colonia ilegal) - The illegal subdivision is a situation in which the owner of a piece of land subdivides it for sale without having obtained planning permission. Almost invariably the land is sold without any services or preparation of the site (other than a minimal access road). As planning permission has not been obtained, it is usually difficult and often impossible for the families to obtain the basic services, particularly water. The normal method of sale of the land is through "renting with promise of sale." Under this system, the family pays monthly installments, usually over a period of 5 to 7 years, and on the completion of the final installment the land is purchased. If the family defaults on a payment, the seller can consider the payments to that point as a simple rental agreement and the tenant may lose all purchase rights. This method of sale is illegal but it tends to be the purchaser rather than the seller who suffers from the illegality and there are many cases in which the purchaser never obtains ownership. The illegality of the sales procedure also means that there is no regulation or control of prices or interest charges and in most cases the purchaser has no idea how much interest he is paying. 1.16 It is important to appreciate that the illegal subdivision is not just a form of low-cost housing and that many middle-class and even upper-class housing developments can be classified as illegal. - 25 - 1.17 The studies in colonias ilegales revealed that in most of the new colonias on the outskirts of the-city. (for example in the Apopa region where the FSDVM is constructing its new projects), a significant number of mesones are being built and that many of the houses are also being built for rent. This is important because it has traditionally been thought that in the colonia ilegal familes constructed almost exclusively for their own use but in fact the colonia may prove to be one of the major forms of rental market in the future. 1.18 Squatter Settlements (tugurios) - Compared with other countries San Salvador has no large tugurios (squatter settlements) and probably none has more than 1,000 families and the great majority house less than 100 or even 50 families. Due to topographical conditions and the high demand for land, almost none of the tugurios are on flat land, and many are in ravines, along former railway routes or on relatively steep hills. In most cases, the life of the tugurio is relatively short although there are several cases of well established communities where families have been living there for 10 or more years and where there is relatively secure tenure. Approximately 15 to 20 per cent of the tugurios are on private land and in many cases occupants are paying rent. 1.19 In the insecure tugurios there is usually no water supply or sanitary services. In the more established tugurios, water has usually been introduced either by the government or by private initiative. In one of the communities studied, the water had in fact been installed by one of the tenants who then resold to the other families at a considerable profit. In the larger tugurios government programs have introduced latrines, often on a communal basis. -26 Currently less than five_ percent of" the popuation of San Salvador live in tugurios and it is likely that with the programs of urban renewal this proportion will fall. In most cases when a tugurios is removed, the residents are offered some type of access to a government housing program, often based on self-help, but at the present time we do not have any information on the proportion of families who are in fact relocated to one of these programs. 1.20 Temporary Camps - These camps were provided by the government for the victims of flooding and other emergencies. They are gradually being removed. 1.21 Deteriorated Housing - The estimates here are inevitably not very precise and as in any other city, the figures are constantly changing as older areas of the city are pulled down for renwal. 1.22 Neither of these latter two types of housing were included in the present study. 3. Housing types produced by the FSDVM 1/ 1.23 The objective of the FSDVM is to provide housing accessible to families too poor to be able to obtain housing through existing public or private programs. As a central policy is to require complete cost-recovery to ensure the replicability of the projects, it has been accepted that in practice the projects will not be accessible to about the poorest 15 per cent of the urban population. 1/ A more detailed description of the FSDVM programs is given in Bamberger, Sae-Hua and Gonzalez-Polio "Evaluation of the first El Salvador Sites and Services Project" Chapter 2 op. cit. - 27 - 1.24 Although some projects have been as small as 100 units, the typical FSDVM project provides between 500 and 2000 housing units. Usually all units are single family and single floor, but recently there has been some experimentation with 2 floor units and with conversion of tenements into condominium ownership. 1.25 FSDVM's normal method of operation is to divide the project into lots of 75 to 100 square meters, all serviced with individual water and sewage connections. Costs are minimized by reducing the area of vehicular roads and by exploiting natural topographical gradients to reduce the cost of earthworks and drainage. Participating families are required to work together in mutual help groups at weekends to construct walls, roof etc. to a level where the house, although not complete, is considered to be habitable. Families can then complete the construction at their own pace and using materials and designs of their own choosing. Loans are provided for the purchase of additional building materials but families will normally have to provide additional financing to complete the house. 1.26 The use of mutual help construction is important as it eliminates the necessity for a 10 per cent down-payment and thus makes the houses more accessible to poorer families. 1.27 Although the majority of plots in a project will be sold through the mutual help system, there may also be some plots sold directly to families who can complete the construction on their own. The present study includes both "basic units" where the family obtains, at the end of the mutual help a habitable unit, and "serviced plots" where the family receives the plot with water and sewage connections but no contruction. - 28 - 1.28 Houses are purchased through fixed monthly payments, over a 20 year period at 61. interest. Houses cannot be resold during the first five years. This provision is included to protect families from selling out before they have had time to understand the worth of their property. 4. The accessibility of public and private housing to the low income families in San Salvador 1.29 Table 6 estimates the accessibility of each of the main types of formal and informal housing to low-income families. The estimates are based on the simplifying assumption that a household will not spend more than 25% of total household income on housing. - Average monthly costs were calculated for'each type of housing and an income distribution curve was computed (Figs. 1 & 2). It was assumed on the basis of the above assumptions that a family could afford a particular type of housing as long as their total family income was at least 4 times greater than the required housing payment. Table 6 indicates the lowest income percentile to which each type of housing is affordable. It can be seen that none of the public housing programs which existed in 1977 (the year in which the estimates were made) were affordable to families below the 48th income percentile. In fact most public housing was only accessible to the top 40 per cent of families. 1/ Although this assumption seems to be valid for renters (see Douglas Keare and Enmanual Jimenez "Affordability, Income and Housing Consumption", lowest income owners appear willing to spend up to 40% of their incomes on housing (see Bamberger, Sae-Hua and Gonzalez Polio Chapter 6). Table 6: ACCESSIBILITY OF FORMAL.AND INFORMAL HOUSING TO THE URBAN POOR, SAN SALVADOR 1977 Institution Type of housing . Lowest percentilL who can afford this type of housing Invasion of public or Squatter settle- Free (accessible to private land ment (tugurios) all income groups) Private landlord Tenements (mesones) poorest quality 6th income percentile Private developer Extra legal sub- division (poorest quality) 10th FSDVM Basic Core Unit 24th "o Private landlord Tenements - adequate quality 24th " Public-IVU Marginal housing in squatter areas (discontinued) 27th Private developer Extra legal sub- divisions (adequate quality) 42nd Public - FSV Normal program 1975-78 48th Public - IVU 2 bedrooms 52nd Public - IVU 4 bedrooms Above 60th Public - IVU Appartments Above 60th Public - FSV Normal program 1978-82 Above 60th Public - FNV Single family houses Above 70th Yote: FSDVM = Salvadorean low-cost housing foundation IVU = Institute for Urban Housing FSV = Social Housing Fund FMV = National Housing Finance Fund. Source: Jim Richard and Nichael Bamberger "Economic evaluation of sites and services programs and their accessibility to low-income groups in El Salvador" Table 2.15. FSDVM Report Series on the Evaluation Program, No. 16 July 1977. - 30 - 1.30 The very poorest families are restricted to poor quality rented rooms, squatter settlements and extra-legal subdivisions. The FSDVM projects are accessible down to about the 25th income percentile (although a few families are found in the second income decile) which is about the same level as for better quality tenement houses. These figures on project accessibility will become important in later sections when we consider the distribution of project benefits by the income level of beneficiaries. Figure 1; DISTRIBUTION OF MONTHLY FAMILY INCOMES IN SAN SALVADOR, 1976 30 25 20 P0 4 500 1000 1500 2000 2500 3000 3500 4000 Monthly Income (Colones) Source: Projections for 1976 by FSDVM on the basis of data from PREALC, Situacion y Perspectivas del Empleo en El Salvador 1975. Io . 32 - 400 3500 3000 r4 2500- oso 5000 65% of the families earn > ~less than-Ø500_peýr month 2000 - - ---- 45% of the families earn co00 lessthan 300 p_er month 300 - - - - - - · 200 00 Oý 20 0 50 O7,08 0 0 Percent of the Families Source: PREALC, Situacion y Perspectivas del Empleo en El Salvador 1975. Graph constructed by FSdM. en El alvado*ven~ - 33 - CHAPTER 2 - METHODOLOGY Section 1 - The Estimation of Efficiency and Social Parameters 2.01 The methodology for the estimation of Efficiency and Social parameters follows closely the Squire-Van der Tak methodology. A more detailed description of the estimating procedures is given in Annex 1 and the purpose of this section is simply to provide a brief overview. Efficiency parameters 2.02 The estimation of efficiency parameters is considered as a first stage of the analysis in which no consideration is given either to the characteristics of the groups receiving the benefits, nor to the use which is made of national resources. The main objective of the efficiency analysis is to compare the net benefits of each project to the nation by converting all costs and benefits to border prices. This involves the estimation and adjustment for all price distortions produced through subsidies, indirect of direct taxes and other contraints on the operation of a free market so as to be able to estimate the true economic, as opposed to financial cost, of each project to the nation. The main elements of the calculation are summarized below. Conversion factors for consumption 2.03 The purpose of the conversion factors is to estimate the real value to the economy of foregone consumption. This is done by converting market prices to border prices by adjusting for import and export taxes. A number of simplifying assumptions were made, such as the assumption that all export goods are exported, to permit the use of the Squire-Van der Tak simple conversion factor. 1/ Separate conversion factors were used for consumption of imported goods by high and low income groups as the import 1/ Squire-Van der Tak, Equation 3 page 59 34 - tax rates are different. The cut-off point between high and low income groups was taken at C700 per month ($280) household income so as to be consistent with other similar studies in El Salvador. This also approximates the upper income limit for participation in FSDVM projects. 2.04 For non-tradables such as land, a simplifed Standard Conversion Factor was used. Conversion factor for capital goods 2.05 Separate estimates were made for construction (labor + materials + other) and for other traded capital goods. The conversion-factors for construction and other capital goods were then weighted by their respective shares of total investment to obtain the Conversion Factor for Capital Goods. Shadow wage rate (SWR) 2.06. In its simplest form the Shadow Wage Rate reflects the opportunity cost to the economy (foregone production) of using labor in the construction of housing. The higher the unemployment rate, the lower the real cost qf withdrawing labor to work on house construction. Separate SWR were estimated for skilled and unskilled labor as quite different assumptions are involved in each case. 2.07. For skilled labor an unemployment rate of 10 was assumed so that the opportunity cost for skilled labor is 0.9 of the average skilled wage rate. It was further assumed that skilled labor is withdrawn equally from all sectors of the economy so that the Standard Conversion Factor is used to estimate the true cost to tho economy. Thus the SWR for skilled labor = 0.87. - 35 - 2.08. For unskilled labor an unemployment rate of 35% is assumed. It was further assumed that all unskilled labor is withdrawn from rural areas so that the opportunity cost is represented by the foregone agricultural output. The SWR for unskilled labor is estimated to be 0.66 2.09 The proportions of skilled and unskilled labor used in house construction were estimated to be in the ratio 2:1 so that the SWR for labor was a weighted average of the SWR for skilled and unskilled labor. Marginal productivity of capital 2.10. This is defined as the net return on a marginal unit of public investment at border prices. This had to be estimated using the macro-economic approach suggested by Squire-Van der Tak as information did not exist to use the more detailed micro-economic approach. 1/ The method used may over-estimate the marginal productivity of capital as only an approximate method could be used to take out the part of the increase in GDP which is due to increased labor productivity. 2. Social parameters 2.11. The purpose of the social parameters is to permit an evaluation of the contribution of the project to achieving the distributional objectives of the government. Every government has an explicit or implicit welfare function which guides its policy planning and two of the main determinants of this are policies relating to the trade-off between present consumption and future growth and the distribution of economic benefits between different income groups. The social analysis begins with a definition of the distributional objectives of the public sector (based on a review of policy statements and inferences about implicit 1/ Squire-Van der Tak p. 110 --36- objectives). Weights are then developed to reflect these objectives and are applied to project benefits. Estimating the values of government distribution preferences 2.12. Although the government states income redistribution as a policy goal there is no indication that this is a very high priority, and as a consequence values of n= .5 and n= 1. both relatively low, were used. 2.13. The Consumption Rate of Interest (CRI) reflects .the discount rate for future benefits. Where government policy is growth oriented the discount rate should be low, but it will be correspondingly higher where policy is oriented towards increasing present consumption. Moderate CRI's of 67. with low priority on income distribution ( n= 0.5) and 7% with higher priority on distribution (n=l) were used. This reflects the fact that the government indicates its interest in growth but at the same time has as an objective a steady increase in consumption. Distribution Weights (d) 2.14. The distribution weight (d) indicates the value which should be attached to benefits accruing to different income groups when the two values of n are used. An average consumption level was estimated as the reference point for the calculation (C 820 per capita in 1976). Five income strata were then defined to correspond approximately to the target group of the main types of housing projects. Weights for each income group were then.calculated for values of n of 0.5 and 1.0. For households below the average consumption level the value of d will be greater than 1.0 whereas for families above this level it will be less than 1.0. The range of weights is of course greater with n = 1 indicating that a higher priority is placed on ensuring redistribution of benefits to lower income groups. - 37 - Critical Consumption Level (CCL) and the value of public income (v) 2.15. The Critical Consumption Level (CCL) is the level of consumption at which the government is indifferent between public and private uses of an additional unit of income. Below this level the government would put higher priority on private consumption. The CCL can be estimated directly (see paragraphs 32 to 35 of Annex 1) but a number of independent consistency checks (minimum daily wage, minimum income for tax exemption etc.) were also used to ensure that the estimated values were realistic. 2.16. The values of v, the value of public income were estimated by both the simple and the complex formulae proposed by Squire-Van der Tat. 1 It was.:decided to use the results of the simpler estimating procedure as the assumptions on which they were based seemed more plausible. This procedure gave a value of v=2 when n=0.5 and v=1.8 when n=O. 2.17. Using these values of v the CCL was estimated to be C 235 per capita per annum with n=0.5 and C 488 per annum with n = 1.0. It can be seen that the assumption about n has a significant impact on what is assumed to be the minimum acceptable level of consumption. Distribution weights (w) 2.18. Distribution weights were computed for each of the 5 income groups and were applied to the net increases in consumption arising from the different housing projects. These weights combine the different values placed on consumption accruing to different income groups and the different values placed on income accruing to government and private individuals and different income levels. - 38 - Social Shadow Wage Rates (SSWR) 2.19. Social Shadow Wage Rates .(SSWR) take into account the effects of increased employment on consumption as well as the opportunity cost of the foregone production. The distribution weights also take into account which income groups receive the extra consumption. 2.20. For skilled labor, where only a 101. unemployment rate is assumed, the additional employment has a relatively small effect on increased consumption. The SSWR is 0.914 for n=0.5 and 0.915 for n=1 indicating that due to the small change in consumption the SSWR is not responsive to changes in n. 2.21. For unskilled labor the impact on consumption is greater due to the fact that an unemployment rate of 35% was assumed. For n=0.5 the SSWR is 0.78 which is considerably higher than the efficiency SWR of 0.66 reflecting the social cost of the extra consumption. However, with n=1 the SSWR drops to 0.615 indicating the higher value placed on the increased consumption of the lowest income groups. 2.22. A global SSWR ias estimated as the weighted average of the skilled and unskilled rates, giving values of 0.868 for n=0.5 and 0.807 for n=1. 2.23. SSWR's were also estimated for the mutual help and self-help construction used by the FSDVM and some other programs. For skilled labor the same SSWR was used. It was assumed that given the high unemployment levels among unskilled workers the shadow value of the beneficiaries own labor was zero, which meant that any reduction in output would be compensated for by hiring other unemployed labor. The Accounting Rate of Interest (ARI) 2.24. This is the discount rate for evaluating the fall in the value of the numeraire over time. The estimated values,.ranging from a low of 7% with n=0.5 to a high of 8.3% with n=1, were not very different from the Consumption Rate of Interest. Section 2: Methodology for Efficiency Analysis 2.25 Costs and benefits were calculated in two separate steps 1 where the first one included the efficiency costs and benefits reflecting the real economic costs of the housing option considered and the second one reflected the social costs and benefits related to income transfers (both within the private sector and between the public and private sectors) and to'changes in consumption and savings resulting from the "project". 2.26 The efficiency analysis takes into account the costs and benefits to the economy as a whole, as opposed to the costs and benefits to consumers and producers. Since we use the numeraire chosen by Squire-van der Tak, (uncommitted public income at border prices expressed in domestic currency), conversion factors have to be calculated which reflect any difference between financial and economic costs and benefits in terms of that numeraire. 2.27 The first step of the efficiency analysis is the use of the conversion factors and parameters, calculated as explained in paragraphs 2.03 to 2.05. For most cost items the elimination of interest payments and indirect taxes, which do not constitute a use of resources and therefore have no border vaue, is the only adjustment made. However, for land and labor there is a clear divergence between financial and economic 1/ In some cases a third step was used when estimates of costs and benefits were made from the point of view of the individual family. cost which has to be accounted for. Shadow prices for labor were calculated as explained in paragraphs 2.06 to 2.09 above. 2.28. Land was shadow-priced assuming that the eventual effect of the housing projects considered would take place in the periphery of the urban area, thus displacing land from agricultural uses. The opportunity cost of land would then be the annual loss in crop production, border-priced using the conversion factor for agricultural output. Although the economic value of land is a function of location, the asumption made here is justifiable in that virtually all new low-income housing projects are located in the periphery and it made the comparison of the results easier regarding policy recommendations. If the optimal location for a project was to be decided, the market value of land would play an important role as a proxy for the opportunity cost, but here we are more concerned with other aspects of the projects reflecting the comparative advantage of housing options in characteristics other than location. In the case of the mesones, however, given that their central location makes them compete with other (mainly commercial and higher income housing) uses, the assumption is more questionable even on empirical grounds, and for this reason the estimated market value of the land is used for the evaluation of the mesones. For sensitivity purposes, however, the analysis is also performed using market values of land for all the options, and the results are presented in Tables 14 through 17. 2.29 The value of the lost agricultural production was calculated as the loss of average net yield per hectare for the main crops in the vicinity of the AMSS. The cost of land calculated in this way is an annual cost, whereas when the market value of land is taken it is a one time cost that occurs in the first year of the project life. - 41 - 2.30 Two different elements are included as benefits in the efficiency analysis. The first one includes the market rent of the project houses inputed by surveys and by comparison with similar dwellings in the market. The rent rather than the imputed market value of the house was used to estimate benefits because the latter is likely to reflect the present value of the future benefits of the dwelling discounted at the owners' discount rate, which does not necessarily coincide with the public discount rate. 2.31 In a later study in two interior cities (Santa Ana and Sonsonate) an evaluation was made of the validity of owners estimates of sales value and imputed market rent. Rent and market value were compared with cost of construction and other indicators and the resulting ratios appeared consistent with findings from other countries. 1/ 2.32 A second component of the benefits is the increase in the consumer's surplus of the beneficiaries. This results from the fact that the price that beneficiaries of all formal sector projects are charged for their new houses is below market rents, as a result of lower interest rates and other subsidies (explicit or implicit). 2.33 This increase in consumer's surplus would be the shaded area in Fig 3, where DD represents the individual household housing demand curve, R1 is the market clearing rent, R2 is the rent actually charged, Q1 is the amount of housing consumed before the project and Q2 the amount consumed after the project. 2/ 1/ Bamberger, Sae-Hua and Gonzalez-Polio "Evaluation of the First El Salvador Sites and Services Project". Chapter 4. 2/ Since project housing is "rationed", it is possible that the amount of housing supplied to any individual is below his desired consumption at that price, which is why in the graph A does not fall on the demand curve. -42 - Fig 3 The Estimation of Consumer Surplus Rh1 - 43 - 2.34 It is difficult to quantify this additional benefit, particularly because housing is not a homogeneous good but a bundle of different characteristics whose separate values are hard to assess. However, an attempt was made here using a triangle approximation to estimate the consumer's surplus (which in fact underestimates the benefits). The part of consumer's surplus generated by the project not reflected in the benefit valued at market rent would be: ACS = 1/2 (r - r 2 1 (1) where r is the market rent imputed on the houses of the project and 1 r is the rent paid by the beneficiaries. Since the houses of the 2 projects studied are not rented but sold to the beneficiaries, r2 is calculated as an equivalent rent using the formula: H1 i-1 H2 i2 r2= -(-m x -m (2) 1+il 1-(+il)m 1+12 1-(l+i2) where H1 = Amount charged by the institutions when selling the house. H2 = Additional cost to the family (materials and/or labor for the mutual self-help stage. 1= Interest rate charged by the institution. 2= Discount rate of family consumption (taken as equivalent to the opportunity cost of capital in border prices). m = life of the project. -44 - Q is taken as I while Q (the amount of housing consumed before the 21 project) is calculated as: 01 = ro rl (3) where r is the market rent paid for the previous dwelling and r1 is the market rent inputed on the project houses. By doing this we are defining the amount of housing consumed as proportional to the market value of the housing unit. 2.35 The net efficiency benefits can be summarized in the following equation: E = B- rl + 1/2 (rli-ra)(1-ro)] - Bj . (4) Where E = net efficiency benefit - imputed market rent of project dwelling_ r2 equivalent rent paid by beneficiaries (Equation 2) o= market rent of the beneficiaries' previous dwelling (taken as proxy for their housing consumption). Bi= conversion factor for the consumption of beneficiary i B= conversion factor for the jth cost item. = cost of the jth component of the housing, - 45 - Section 3: Methodology for the Social Evaluation 2.36 The efficiency evaluation does not take into account the benefits or costs arising from the project as a result of its effect on income distribution. Squire-Van der Tak evaluate the distributional impact of the project by assessing the changes in consumption resulting from it and weighting those changes with a distribution weight (w. = d./v) 1 1 calculated according to the income level of the people affected. The formula used by Squire-Van der Tak for the social evaluation is NSB = E - C (B. - d./v), where E is the net efficiency benefits, C is the 1 1 increase in consumption accruing to the private sector, 13. is the 1 conversion factor (s) for the consumption of the beneficiaries, d. is 1 the distribution weight for income group I and v is the value of public income (and in our case, of private savings also). 2.37 Our approach Is the same, although the data are presented in a slightly different way, because the changes in consumption arising from the use of labor have been added to the efficiency costs, thus yielding social conversion factors different from the efficiency conversion factors which were used to convert the cost elements to give the social costs. Labor is the only cost item for which a social conversion factor (the social shadow wage rate) was calculated, following the method explained in paragraphs 2.06 through 2.09. 2.38 The following equation was used to calculate the net effect of changes in consumption arising from the project: SNB = (B - C ) -,%I .(B. - W.) (5) E Si 1 i where B is the efficiency benefits; s C Sis the social costs-, -46 I. is the change in consumption accruing to income group i; 1 B. is the conversion factor for the consumption of income group 1 i; and w. is the distribution weight for the consumption of income group i (or di/v, where d. is the income distribution weight 1 and v is the value of public income). 2.39 The social costs, Cs are presented in a separate table, while the social benefits are presented in a second table including the following elements: a) The efficiency benefits, B . E b) I,v the increase in consumption accruing to the beneficiaries of the project, calculated as: I. = IR + CS - (AP + SH + M), (6) 1 where IR is the imputed market rent of the project housing; CS is the additional consumer's surplus generated by the project; -AP is the annual payment to the institution; SH is the additional costs to the family resulting from the self-help stage; and M is the maintenance cost paid by the beneficiaries. 2.40 The conversion factor for consumption B. is either B (0.94) 1 e or B (0.89), according to the income level of the beneficiaries. 2.41 The distribution weight, W.. calculated as di/v, is computed for a discrete change from the original per capita consumption of the beneficiaries, C1, to the consumption after the project: C2 = C I I I/m. m is the average size of the household; v is the value of public income, 2 for n = 0.5 and 1.8 when n = 1. - 47 - 2.42 In the calculations it is assumed that all additional benefits (either increased income or reduced.expenditures) are consumed if the beneficiaries fall within he low-income consumption group (those with monthly income below C600 in 1976 colones), which should be close to actual fact. When beneficiaries fall within the high-income group a certain proportion of the additional income is assumed to be saved. Savings are valued as much as public income, and the fact that private savings might lead to additional consumption in the future, is disregarded; thus underestimating the social cost of the increase in the income of the rich. However, none of the projects are likely to have a big impact on savings and the effect of savings on future consumption can be neglected without seriously affecting the accuracy of the evaluation. 2.43 -12t is the social value of the change in low-income housing market rent as a result of the additional supply created by the project. This effect would only take place if the assumption is made that the low-income housing market rent adjusts significantly to changes in the housing stock. The assumption seems to be close to reality given the market conditions prevailing in San Salvador. For instance, rents have been particularly sensitive to the decrease in low-income housing stock caused by the recent destruction by fire of a number of mesones in the city center. 2.44 If a decrease in market rent is assumed to occur there will be a decrease in the rents paid by low-income tenants to relatively well-off landlords. In the efficiency analysis there is no effect, since the tenants' gain is considered as valuable as the landlords' loss and changes in consumer's and producer's surplus cancel each other out. In the social - 48 - analysis the case is different and the consumption of low income tenants is assigned a higher weight than-that of landlords, and therefore the effect has to be taken into account. 2.45 There is very little empirical evidence on which an estimation of this effect can be based. This is further complicated by the fact that there might not be a single homogeneous market for housing but different sub-markets and.in this case the effects of changes in the stock of one market would have little impact on the others. However, an attempt was made to estimate the social effect because it is considered important and the evaluation would otherwise be incomplete. A single low-income housing market is postulated, including only the informal housing options. The formal market low income housing projects are assumed to have an impact because their beneficiaries vacate their previous dwellings in the informal market: 2.46 I2 is calculated as: 12 = X (P2 - P1) (7) N where P1 = market rent before the project. P2 = market rent after the project; N = number of housing units built by the project; Q = stock of low income dwellings; x = proportion of low-income dwelling rented. In order to gauge the net social effect, 12 has to be multiplied by [(B -W T)-(l-S).(B L-W ) where s is the landlords' marginal propensity to save, BT and BL the conversion factors for the consumption of tenants and landlors and w and wL the corresponding distribution weights, calculated for marginal changes in consumption of - 49 - the income groups affected. The social value of the change in rents per new unit is: - X (P2'-P1) Q (BT - WT) - X (1-S)(P2-P1)Q (BL-WL) (8) N N 2.47 To calculate the change in rent (P 2-P 1) it is necessary to know the elasticity of rents to changes in the housing stock. Population is growing at about 6% per year in the AMSS, of which at least 2.5% is accounted for by migration; since it is reasonable to expected that most immigrants will belong to low-income groups the average annual increase in potential demand for low-Income housing can be set at 7.5 per cent. Low-income rents have been increasing at about 4% in real terms in recent years. 2.48 Information in the study "La Vivienda Popular en El Salvador" and in the FSDVM's Apopa Demand Study indicates that the number of families living in low-income houses in the AMSS was about 110,000 in 1977 and the deficit was about 15,000. The housing stock could be estimated at about 95,300 units (see below). Using the population and housing growth rates mentioned above, there was a relative decrease in the low-income housing stock of about 2.6%. The value of the elasticity would then be the ratio of the percentage increase in the rent to the percentage decrease in stock or 1.5. 2.49 In order to estimate the housing stock it was assumed that two different markets exist, the informal and the formal low-income housing markets, and that the effect of the formal institutions' housing projects are passed on to the informal market via the number of vacancies left behind by the beneficiaries. The housing stock, Q, is calculated by - 50 - indexing the three main options in the informal market (rooms in mesones, colonies ilegales and tugurios) according to the average rents paid per unit. V According to a study by the FSDVK average rent per unit in 177 was C34.4 per month in mesones, C29.3 in colonias and C9 in tugurios. If the meson is assigned a value of 1, the colonias would have a value 29.3/34.6 = 0.85 and the tugurio of 9/34.6 = 0.26. The total housing stock expressed in numbers of meson units would be Q = R. x n. where :1 1 R. is the value of the index for housing option i, and n. is the 1 :1 number of units of option i. The value of Q thus estimated is 95,287 in 1977, using data of the Apopa Study by the FSDVM. 2.50 The new market rent after the project can be estimated as P = 2 P (1 - E. Q/Q) where Q is the without-project stock of low-income housing in the market. The total social effect of the income transfer resulting from the changes in market rent brought about by the project can now be estimated as: -(P2-PI1;