Pirate subdivisions and the market for residential lots in Bogota

Urban and Regional Report No. 79-12 PIRATE SUBDIVISIONS AND THE MARKET FOR RESIDENTIAL LOTS IN BOGOTA Alan Carroll April 1980 This report was prepared under the auspices of the City Study Research Project (RPO 671-47) as City Study Project Paper No. 7. The views expressed here are those of the author and should not be interpreted as reflecting the views of the World Bank or its affiliated organizations. This report is being circulated to stimulate discussion and comment. Urban and Regional Economics Division Development Economics Department Development Policy Staff The World Bank Washington, .D.C. 20433 PREFACE I am grateful to Gregory Ingram, Director of the City -Study, for overall guidance on this paper. I am indebted as well to Ricardo Paredes and Luis Guillermo Martinez, formerly of the Superintendencia Bancaria in Bogota, for making the data available and for their advice on many topics. I must also thank Luis Carlos Jimenez of the Beparta- mente Administrativo de Planeacion Distrital for providing important information on normas minimas subdivisions and Jose Fernando Pineda, local coordinator of the City Study, for furnishing useful contacts,and comments on the analysis. The paper has benefitted greatly from first-draft comments made by Andrew Hamer, Kyu Sik Lee, Johannes Linn, Rodrigo Losada and Rakesh Mohan. Finally, I would like'to acknowledge the computer programming help I received from Yoon Joo Lee, Jeffrey Lewis and Manfred Wagner, and the manuscript editing of Marcia Fernald. This paper is part of a program of research being conducted by The World Bank on Bogota and Cali, Colombia. The goal of the program is to increase our understanding of the workings of five major urban' sectors--housing, transport, employment location, labor markets, and the public sector--in order that the impacts of policies and projects can be assessed more accurately. Another project paper that deals with housing in this series is Rafael Stevenson, "Housing Programs and Policies in Bogota: An Historical/Descriptive Analysis", Project Paper No. 3 (June 1978). TABLE OF CONTENTS Page No. PREFACE SUMMARY (i) Il. INTRODUCTION ............................................... 1 A. Background ......... .................................... 1 B. The Study ........ .............. 5 C. Data Validity ..........idt............................ 7- D. Pirate vs. Normas Minimas Subdivisions ................. 7 E. Affordability of Lots ................................... 11 F. Services and Profits ................................... 16 II. GENERAL CHARACTERISTICS OF SUBDIVISIONS ..................... 18 A. Basic Features at.18 B. Purchase of Tracts ..27 C. Infrastructure and Development Costs ................... 30 D. SalL,;of Lots ........................................... 41 E. Total Costs of Development ............................. 45 III. RATES OF RETURN TO SUBDIVIDERS ............ ............. 51 A. Background ...kgr. 51 B. Rates of Return in the Sample .......................... C. Determinants of Rates of Return ....................... 57 D. Real vs. Nominal Rates of Return ...................... 60 E. Summary and Conclusions ................................ 62 IV. THE DETERMINANTS OF TRACT PRICES ............................ 64 A. Analytical Framework ................................... 64 B. Results for Subdivision Tracts ......................... 66 V. THE DETERMINANTS OF LOT PRICES .............................. 70 A. Background and Model Specifications ................... 70 B. Aggregated Sample Results ............................. 72 C. Disaggregated Sample ................................ 82 VI. POLICY ISSUES: NORMAS MINIMAS AND THE SUPPLY OF SERVICED LOTS .......................................... 89 A. Improving the Quality of Lots ......................... 89 B. Restrictions on Normas Minimas ......................... 92 C. Conclusions ............................................ 96 TABLE OF CONTENTS (Continued) Page No. APPENDIX I - List of Variables in Superintendencia Bancaria Pirate Subdivision Survey, 1977 ................. 98 APPENDIX II - Calculation of Internal Rates of Return for Subdivisions ................................... 100 APPENDIX III - Hedonic Prices in Competitive Markets ........... 103 APPENDIX IV - Hedonic Price Indexes with Infrastructure Expenditures .................................. .. 107 REFERENCES 114 S LIST OF TABLES Page No. 1. Basic Average Data for Pirate and Normas Minimas Subdivisions 9 2. Bogota Household Income Distribution 1977 12 3. Distribution of Monthly Installment Payments for Lots 14 4. Distribution of Subdivisions by Number of Lots 19 5. Number of Subdivisions by Initiation Year 20 6. Total Lots Sold by Year in Sample 25 7. Average Land Uses Within Subdivisions 26 8. Extent of Infrastructure Installation in Subdivisions 32 9. Sources of Financing for Infrastructure in Pirate Subdivisions 34 10. Average Quantities of Infrastructure Installed per Hectare Usable Area 36 11. Average Costs of Infrastructure Installation per Unit - 37 2 12. Average Expenditure on All Infrastructure per M Usable Area 39 13. Average Professional Services, Publicity, and Administrative Expenditures per Lot 4.0 14. Average Rates of Lateness in Payments Among Lot Buyers 42 15. Distribution of Major Cost Components 48 16. Average Total Development Costs Per Lot 49 17. Alternative Average Nominal Rates of Return to Subdividers 55 18. Relation Between Rate of Return for Low-Income Subdivisions and Subdivision Characteristics 58 19. Comparison Between Average Nominal and Real Rates of Return to Subdividers 61 20. Relation Between Tract Price and Tract Characteristics in Low-Income Subdivisions 67 Relation Between Lot Price And Subdivision Characteristics Observations are Subdivisions 21. Linear Hedonic Price Inaex witH inrrastructure Quantity - Average Lot Price Dependent 75 22. Linear Hedonic Price Index with Infrastructure Quantity - Average Lot Price per Square Meter Dependent 76 LIST OF TABLES (Continued) Page No. Relation Between Lot Price and Lot Characteristics - Observations are Lots 23. Linear Hedonic Price Index with Infrastructure Quantity - Total Lot Price Dependent 84 24. Linear Hedonic Price Index with Infrastructure Quantity - Lot Price Per Square Meter Dependent 85 25 Normas Minimas Subdivision Proposals Submitted and Given Final Approval 1973-1977 93 4 LIST OF FIGURES 1. Location of Pirate and Normas Minimas Developments in Sample '22 --2. - Average Tract Size Across Rings and Sectors of. Bogcta 23 2 3. Average Tract Prices per M Across.Rings and Sectors of Bogota 29 4. Average Tract Price per M 2 Over Time 31 5. Average Lot Prices per M2 Over Time 44 6. Average Lot Prices per M2 Across Rings and Sectqrs of Bogota 46 7. Average Lot Size Across Rings and.Sectors of Bogota 47 SUMMARY Despite their lack of services and illegal tenure, pirate subdivisions are a major part of the residential land market in Bogota because they provide the only homeownership possibility for modest-income households. This paper examines the market for pirate lots, with particular attention to comparisons between pirate and legal "minimum norms" (normas minimas) subdivisions. The main issues raised are (1) whether the rates of return to subdividers are excessive and indicative of excess demand, (2),lot buyers' willingness to pay for different lot attributes, and (3) the potential for the private sector's supplying greater numbers of legal, well-servied lots. The analysis covers 135 pirate and 14 normas m1nimas sl.bdivisions developed in the last 10 years. The data were collected by the Housing Division of the Superintendencia Bancaria in Bogota. Pirate .subdivisions have flourished because Bogota has a large supply of relatively inexpensive, privately owned land on the periphery and because land invasions have traditionally been suppressed by the authorities. Although pirate subdividers are often unscrupulous, they fulfill an important function, accounting for about one third of residential land developed in Bogota. While pirate subdivisions are viewed as settlements of the urban poor, the evidence indicates that the median income of home owners in pirate subdivisions matches the median income for Bogota as a whole. Indeed, families in the lowest third of the income distribution cannot afford to buy pirate lots. Most of the subdivisions in the Superintendencia Bancaria sample are small, with the median size about 100 lots. Gross residential densities are around 300 to .600 persons per hectare. The subdivisions are located principally in the far western and southern areas of Bogota. At the time of the survey most subdivisions had water standpipes, electricity, and graded streets. Services,usually are installed after the majority of lots have been sold. In pirate subdivisions the lot buyers themselves often make significant expenditures for infrastructure. In contrast with pirate,subdivisions, normas minimas developments have more services, better locations, more open space,,and a high probability of legal tenure. The average normas minimas lot buyer pays-about one third more than in the pirate market for a lot about 25 percent smaller. While a small number of pirate subdivider,s earn large profits, most obtain modest rates of return±. In the pirate market the profit margin is generally not large enough to be tapped for greater infrastructure investments, as some observers have suggested. Normas minimas subdividers do earn excess profits, however, because the District government has restricted the number of normas minimas developmet permits. Stronger legal action against pirate subdividers would increase only slightly the supply of adequately serviced lots. Making it easier to develop normas minimas subdivisions, on the-other hand, would probably increase the supply substantially, Although the prices of normas minimas lots would drop if the business were more competitive, they would still be higher than,those of pirate lots because buyers are willing to pay for additional amounts of services. All else being equal, buyers,will pay Col$700-800 pesos for each additional meter of sewer lines; moreover, a lot with access (iii) to a water standpipe will fetch several thousand pesos more than one with no water service. Lot buyers are also willing to pay extra for closeness to the central business district, location in more desirable areas of the city, and larger lot size. As a result, while normas minimas could take over a part of the pirate subdivision market, there remains a large proportion of families who cannot afford lots with adequate service levels or other desirable characteristics. These families will continue to buy pirate lots as long as the supply can be maintained at low prices. The benefits- of the sites and services approach -- wider affordability and better planning -- must be better understood before normas minimas can have a significant impact. I. INTRODUCTION A. Background The "pirate subdivisions" of Bogota, Colombia differ from squatter settlements in that land changes hands through purchase. Modest- incomefamilies buy small lots from entrepreneurs who acquire tracts of land and subdivide them without conforming to zoning laws, subdivision regulations, or service provision standards -- hence the term "pirate". The subdivisions spring up where land.is inexpensive, generally on the periphery of the city. The lots usually provide only a bare minimum of services, often nothing more than some bulldozed streets and a few water stan4pipes. Buyers typically make down payments of between a quarter and a third of the lot value and pay monthly installments over one to four years. Families build their houses, often incrementally, according to their capacity to pay. The small amount of research done so far suggests that pirate subdivisions account for a very large proportion,of the residential land supply in Bogota. One study (Vernez, 1973) estimated that, of the total residential area of the city within the urban perimeter in 1970, almost one-third had been developed as pirate subdivisions. Another study found the fraction to be around 38 percent (Borrero and Sanchez, 1973). In addition, Vernez reported that in 1970 about 45 percent of Bogota's 451,000 families were living on land originally developed by pirate subdividers. The proportion of families living in illegal developments is much lower in current terms, however. For years the - city government has been granting legal status to well-established -2- pirate subdivisions, thus allowing residents to obtain better public services and clear tenure., One study (Fuentes and Losada, 1978) estimated that about 13-percent of Bogota's families lived in illegal subdivisions in 1972-73.1/ The extent of the pirate land business in Bogota reflects a high demand for inexpensive residential land and the lack of alter- natives available to lower-income people. Commercial and government housing in the city is generally occupied by families with incomes. at least twice those of pirate subdivision dwellers.(Lopez and Jimenez; Bender 1975). In addition, land invasion, a widespread solution for the,poor in other Latin American cities, has not been practiced to a significant extent in Bogota.- Two basic conditions have allowed pirate subdivisions to flourish in Bogota: the existence of a large supply of inexpensive, privately-owned land close to the city; and a history of lax enforcement of land use and subdivision regulations. While the authorities have 11 The foregoing figures are about the best available on the extent of illegal land development in Bogota. There appear to be no accurate or up-to-date records in any agencies on the total land area occupied by pirate subdivisions. 2/ Bogota is unusual in that invasion accounts,for a very tiny proportion of lower-income land development: somewhere between one or two percent of residential land. The reasons usually given are (1) the scarcity of the type of land preferred for invasion, that is, public land or land whose ownership is unclear; (2) the history of strong police action in Bogota against squatters.on private property; and (3) the need fcr durable, solid shelter due to Bogota's cool, rainy climate and the consequent disincentive for invasion if the risk of eviction is high after substantial housing investments by families (Vernez, 1973). -3- traditionally used strong measures to suppress land invasions, they have exerted little effective control over land sales between private individuals. Several reasons may motivate a land owner to sell to a pirate subdivider. First, a pirate may simply offer the highest price. Land in certain parts of the city may be unattractive to legitimate developers for the middle-class market because the neighborhood is un- desirable., the parcel is too small, drainage is bad, or the land cannot be serviced properly. In addition., pirate subdividers are often willing to buy land whose development value is otherwise restricted by zoning laws. As a result pirate subdivisions have sprung up in areas of the city where there should be no development, such as flood. zones and steep slopes. A third problem that encourages illegal land subdivision is the long delay -- often one or two years -- posed by the legal development approval process. Pirate subdividers are known for the speed with which they transact business, and a parcel on the margin -- one that could be developed legally or illegally -- may be tipped in the latter direction if the owner is in a hurry to liquidate his property. A fourth factor is that not all tracts of land become pirate subdivisions in arm's length transactions. Pirate subdividers have sometimes coerced land owners to sell by hiring squatters or making plysical threats. While the purchase of the original tract is 1/ While land development regulations have been tightened in the last decade (beginning with Law 66 of 1968), officials in Bogota tend to feel that a crackdown on the large pirate subdivision business would invite more land invasions, a prospect deemed far worse than pirate development. usually a legal transaction, the sale of lots is always illegal. Because pirate subdividers ignore zoning laws as well as minimum lot:size and servicing standards, the ownership documents (escrituras) given to lot buyers are invalid and cannot be legalized until the District government "regularizes" the subdivision.-/ This may take many years. It is not clear, however, how much of a burden illegal tenure imposes on lot buyers. There is evidence, for example, that it does not prevent pirate lots, with or without houses, from being resold. Public officials and the press often portray pirate sub- dividers as exploiters and criminals. They are blamed for cheating the poor and creating slums. The findings of this and other studies suggest that, whatever their motives, pirate subdividers fulfill a useful. function by providing tens of thousands offamilies with plots of land that they could not otherwise obtain without resorting to invasion. Buyers rarely lose their lots, despite their illegal tenure. Families build homes on the lots, often upgrading them substantially over the years. Missing services gradually arrive, and over time some barrios begin to take on a middle-class appearance. On the other hand, there is no denying that many pirate subdividers are unscrupulous and some- times cause harm. It is not uncommon for lot buyers to be left without 1/ The principal laws governing lower-income subdivisions are Special District Acuerdo 25 of 1975, Acuerdos 20, 21, and 22 of 1972, Decretos 1259 and 1260 of 1973, and Law 66 of 1968. 2/ According to data from a 1978 survey of 212 proprietor households in pirate subdivisions in Bogota, 43 percent of the lots in the sample had been purchased directly from previous owners. Of these, two-thirds were bought with houses on them. See Andrew Hamer, Lower-Income Home- owner Households in the Developing Metropolis, Part II, October 1979. -5- promised services, for services to fail due to poor construction, or for lots to become uninhabitable because of poor drainage. Indeed, the social cost of uncontrolled development by pirate subdividers is probably large. B. The Study This paper deals with the market for pirate subdivision lots. Examihing the pirate subdivision business is instructive because it is a unique case of a large-scale., competitive land market for modest income families. This study seeks to apply some basic analytical methods to the following questions: 1) what are the rates of return to pirate subdividers, and do they indicate the presence of excess demand for lots? 2) what are the relationships between prices and characteristics of lots, and what do they suggest about the willingness of lot buyers to pay for different lot attributes? and 3) what are the differences 2/ between pirate and legal "minimum norms" (normas minimas) subdivisions,- and is there a potential for expanding the supply of better-serviced, better-planned subdivisions?1i 1/ These include installing services in unsuitable places such as steep slopes or flood zones; building service networks for inefficient lot layouts; providing transportation services to inaccessible areas; revising street and utility construction programs to take account of unauthorized development; and coping with erosion, pollution, and destruction of natural amenities as a result of development in ecologieally sensitive areas. 2/ Normas minimas refers to a legally-set package of minimum standards for services such as water, electricity, and streets (as well as for lot and dwelling size) which -developers of low-income subdivisions must meet. The normas minimas law, first passed in 1972 (Acuerdo 20 of the Bogota Special District), was designed to lower the cost of required infrastructure and thus encourage the creation of legal subdivisions for the lower-income market. 3/ No attempt is made to treat the pirate market's role within the overall housing or land markets of Bogota. These topics are not unimportant, but they could not conveniently be incorporated into the scope of this study. For background information and broader perspectives on pirate subdivisions,see Vernez (1973a), Doebele (1975), Losada and Gomez (1976), and Blaesser (1979). -6- The analysis covers 149 subdivisions developed over the last ten years. In 1977 the Superintendencia Bancaria in Bogota collected data on 135 pirate subdivisions as well as 14 normas minimas developments. The Housing Division sent out questionnaires to 200-odd subdividers whose names were in their files and in records of the District Planning Department (DAPD). The respondents were informed That they were legally bound to supply the information and that their subdivisions could be "regularized" if they cooperated. Data requested for 'each subdivision included the size of the original tract, its price and date of acquisition; the number of lots; the total street and open space areas; the amounts and costs of services .-installed (sewer, water, telephone, electricity, and paving); and, for each lot sold, its size,price, terms, and date of sale (see Appendix I for a complete list of variables included in the survey)'. The 149 pirate and normas minimas subdivisions for which completed questionnaires were returned represent the efforts of 121 individual subdividers../ The paper is divided into six parts. This introductory chapter is followed by a descriptive overview covering the general characteristics of the subdivisions studied, the purchase of tracts, infrastructure and development costs, and the sale of lots. The third chapter deals with rates of return to pirate and normas minimas developers. The fourth and fifth chapters are devoted to an analysis, based on hedonic price 1/ No data on dwelling units or on the characteristics of families buying lots were collected, hence the focus of this study on the supply side of the lot market. The data do permit inferences to be made about demannd, however. -7- indexes, of the determinants of tract and lot prices, respectively. The final chapter presents conclusions and discusses policy issues related to the effectiveness of the normas minimas laws. C. Data Validity The question arises-of whether figures obtained from illegal subdividers by the authorities can be believed. One would expect sub- dividers to exaggerate their costs and understate their 'revenues so as to minimize their apparent profit rates. Checked against the available information, however, the magnitudes of all the important variables -average tract and lot prices, unit costs of infrastructure, lot sizes, and infrastructure quantitites - were found.to be reasonable. Sources of verification are cited in this paper where appropriate. Only one item in the data seems invalid: average overhead costs per lot (subdivider expenditures on professional services, administration, and publicity) are too high for both pirate and normas minimas subdivisions. Owing to the nature of these items and the , design of the questionnaire, overhead costs were clearly the most difficult to verify. Some subdividers apparently took this opportunity to overstate their expenditures. The study attempts to correct for this in the rate of return analysis. D. Pirate vs. Normas Minimas Subdivisions Normas minimas subdivisions are essentially planned and approved sites and services projects. Special District Acuerdo 20 of 1972 permitted not only the lowering of minimum design standards for residential developments but also the participation of the private sector. The example of pirate subdividers undoubtedly influenced -8- the drafting of Acuerdo 20, which makes it possible for lots to be sold as soon as infrastructure plans are approved, thereby allowing sub- dividers to realize immediate returns on investment. Acuerdo 20 was followed in 1973 by Decretos 1259 and 1260, which set out detailed design and engineering specifications for vehicular and pedestrian access, lot sizes, open and communal areas, and public,services. In practice the chief problem with normas minimas subdivisions has been the failure of many developers to install the required.services after lots are sold. A number of reasons have been suggested for this, including underestimation of infrastructure costs by both public agencies and developers, excessive delays for developers caused by government red tape, and sheer dishonesty. Whatever the real causes, the fact is that most normas minimas developments, including the 14 in this study, do not meet legal service standards, although they are generally of significantly higher quality tian pirate subdivisions.- Table 1 presents separate figures for pirate subdivisions inside and outside Bogota's urban perimeter and for the 14 normas minimas subdivisions. Pirate subdivisions outside the perimeter are much larger and cost less, as might be expected, than those inside the perimeter. In general, land development outside the perimeter is subject to different regulations and institutional forces. Where it 1/ As evidenced by the Superintendencia Bancaria data and visits to several normas minimas subdivisions in Bogota. -9- Table 1 BASIC AVERAGE DATA FOR PIRATE AND NORMAS MINIMAS SUBDIVISIONS - .(all prices in 1976 Col $) Pirate Inside Normas/. Pirate Outside 1/I Average: Urban Perimeter Minimas- Urban Perimeter 2/ 2 Tract-price per m 50 64 18 Tract size (m 2) 35,528 98,700 84,758 Number of lots 15.6 585 150i Lot size 125 92 315 2 Lot price per m 253 456 109 Total lot price 31,625 41,952 23,008 Subdivider expenditu e- on infrastructure per m salable (usab-le) 22 64 9 area Subdivider expenditure on infrastructure per lot 3,256 5,955 3,695 Percent subdivision in open space 10 '22 10 2 Open space per lot (m ) 26 39 38 Total number of subdivisions 109 14 24 Total number of lots subdivided 16,994 8,191 3,604 Total number of lots with sale data 11,540 5,916 2,099 1/ All normas minimas subdivisions are inside the urban perimeter. 2/ The term "tracth" refers to the whole parcel of land that is purchased for subdivision. - 10 - seems appropriate in the analysis, subdivisions inside and outside the perimeter are treated separately. Because all normat minimas subdivisions are (by law, and hence by definition) inside the urban perimeter, it is more accurate to compare them only with pirate subdivisions located inside. The first two columns of Table 1 show that normas minimas developers pay an average of about 28 percent more per square meter initially for their tracts of land than pirate subdividers. This reflects the fact that the normas minimas subdivisions are better located, on flat land in the near western zone of the city. It is also due to the leg&l requirement-that normas minimas subdivisions be built on land zoned for residential use and accessible to public utility connections. Normas minimas developers spend a greater proportion of their total costs on infrastructure than do pirates. They also spend almost twice as much on infrastructure per lot in absolute terms, and their subdivisions are better-serviced (see Table 9). Thus, the average normas minimas buyer pays about one- third more than he would in the pirate market for a lot that is about 25 percent smaller; he receives however, more services, a better-than-average location, more open space, and a greater probability of acquiring legal tenure.- I/ The lot prices and sizes calculated from this data set are consistent with earlier observations by Borrero and Sanchez (1973), Losada and Gomez (1976), Doebele (1975), and DAPD (1978). - 11 - According to this data sample, the average pirate lot buyer makes a down-payment of about -30 percent of the lot price and then pays 36 monthly installments. The average down-payment for a pirate lot is about Col$9500 and the average monthly installment is Col$615 (1976 prices). An average normas minimas lot buyer faces a down-payment of about 35 percent and a term of 39 months. The normas minimas down- payment averages Col$14,700 and the installment Col$700. Term length among all subdivisions varies from as few as 12 or 20 months to as many as 50 or 60. While interest rates are rarely explicit in lot purchase terms, buyers within the same subdivision may pay from two to 30 percent more per square meter if they buy on installment than if they pay in one lump sum. On average this "implicit" finance charge is approximately 5.5 percent per year on the balance (total price minus 1/, down-payment)for all subdivisions.- E. Affordability of Lots According to the 1977 Bogota Household Survey (DANE EH15), median monthly household income in that year was .slightly less than Col$5,000, and 26 percent of households had monthl--, incomes under Col$2,500 (1977 prices, see Table 2). This means that for a family at the.median income level, the down-payment for an average lot would cost between two and three monthly incomes. For a family at the first quartile, earning around Col$2,500 1/ If this average* seems low, it is because the base includes some lots sold with no finance charge. For a description of the method used to calculate the average interest rate, see page 73. .. .- 12. - Table 2 BOGOTA HOUSEHOLD INCOME DISTRIBUTION, 1977 Categories of Monthly Cumulative Income (pesos) Percentage Percentage 0- 499 6.2 6.2 500 - 1499 7.7 13.9 1500- 2499 11.9 2.8 2500 - 4999 27.5 53.3 5000 - 7499 15.6 68.9 7500-- 9999 9.9 78.8 10,000 - 14,999 4.4 83.2 15,000 - 24,999 7.5 90.7 25.,O0\+ 9.3 100.0 Total Number of Households (expanded) = 685,424 Source: DANE 1977 Bogota Household Survey, EH15. Figures compiled by N. Hartline and R. Mohan. -13- per month, the down-payment would equal four to six,months' income. Becoming a homeowner in this market thus means having access to a large lump sum, which suggests that families without savings may be largely excluded. The affordability of shelter based on a pirate or normas minimas lot also depends on the cost of building a structure. Since not even the highest-quality lots come with core dwelling units, lot buyers must spend substantial additional sums to provide themselves with minimal shelter. The cost of a basic dwelling ranges from one 1/ to four times the cost of a lot, depending on quality.- A good estimate of the affordability range for housing on pirate and normas minimas lots is provided by the distribution of monthly installment payments made by lot buyers in the Superintendencia Bancaria sample (Table 3). The median monthly installment is roughly Col$625 for all subdivisions, slightly less than Col$600 for pirate, and around Col$660 for normas minimas. About one quarter of the pirate lots require montly payments of less than Col$400, while only about two percent of normas minimas lots can be had at that price. The distiibution overall is compact, with 60 percent of the installment values falling in the Col$400 to 800 range. Assuming the conventional though somewhat arbitrary standard that 25 percent of income is a reasonable amount to spend on housing, 1/ See, for example, W. Paul Strassman, Basic Shelter in an Urban Framework: The Experience of Cartagena, Colombia. World Bank, URBOR Report, Nov. 1978, p. 118. - 14 - Table 3 DISTRIBUTION OF MONTHLY INSTALLMENT PAYMENTS FOR LOTS 1976 Col$ Percentages Installment All Normas Value (pesos) Subdivisions Pirate Minimas 1-199 2.3 3.1 0.7 200-399 15.2 21.6 1.5 400-599 28.5 27.4 31.7 600-799 31.2 24.7 44.9 800-999 11.4 11.9 10.3 1000-1999 8.7 8.9 8.1 2000+ 2.7 2.2 3.8 Lots counted 16,543 11,247 3,296 Zero installment (paid cash) 2,741 2,131 610 No information 271 261 10 Total lots 19,555 13,639 5,916 - 15 - families at the median income level can afford the monthly payments for almost all the lots on the market. If the cost of a dwelling unit i6 taken into account, however, pirate or normas minimas solutions are much less affordable. If the absolute minimum structure-to-land cost ratio is 1:1, the total median housing expenditure would be about 1/ Col$1,250 per montr- This is tithin range of families.at or above the median income level in Bogota. Conversely, a family earning the median income could afford only about half the lots on the market -- including about a third of the normas minimas lots -- if it uses half its monthly housing expenditure for a structure. Assuming the same i:1 structure-to-land split, a fadily at the first quartile in the income distribution, earning Col$2,500 per month, could afford only the cheapest 20 percent of pirate lots offered and would be barred entirely from purchasing a normas minimas lot. Families in the lowest 20 percent of the income distribution would effectively be shut out of the ownership .market. If the assumption of a 1:1 structure-to-land cost ratio is raised to a more realistic level of, say, 3:1,the affordability picture worstens considerably. Under th&se circumstances ownership is difficult for any family below the median income level, unless an unusually large share of incame -- 30-35 percent or higher - is devoted to housing. 1/ This calculation assumes that families would make monthly payments for structures. This is not necessarily realistic, because house construction expenditures are often erratic due to income fluctuations. It does give a valid indication, however, of the long-term financial burden of building a structure. - 16 - In general, home ownership through purchase of a pirate or normas minimas lot seems accessible to families around the median income level, but not to those in the lowest third of the income distribution. This conclusion is supported by a 1978 survey of 212 households in illegal barrios on the periphery of Bogota, which found that the median income of homeowning families matched the median income for the city as a whole.1' F. Servides and Profits Although the figures in Table 1 suggest that.l0t buyers are willing to pay more for services even if it means getting a smaller lot, two caveats are in order. First, in many of the sub- divisions studied, the lot buyers or the government - not the sub- divider - paid directly for a large proportion of the services. This indicates that, in practice, lot buyers may not expect or be willing to pay for more than abare minimum of services at the outset. Second, in most of the 149 subdivisions studied, infrastructure installation, whether by the subdivider or others, occurred after the sale of lots had begun. This casts some doubt on the extent to which the marginal costs of infrastructure may be fully reflected in the observed variations in lot prices. It depends on whether lot buyers are willing to pay only for infrastructure "on the ground" at the time of sale, or additionally, for services "promised" in the future. 1/ Andrew M. Hamer, Lower Income Homeowner Households in a Developing Metropolis, World Bank, City Study Paper, Oct. 1979. - 17 -. The descriptive f4gures in Table 1 also suggest tha; on average, subdividers make good profits, and that the normas minimas subdividers make higher profits than their pirate counterparts. While the normas minimas pay more for raw land and infrastructure, they also create much larger subdivisions, supply smaller lots, and sell them at higher prices. This study tests the hypothesis that pirate subdividers make "exorbitant" profits. Doebele (1975) and others have suggested that if pirate subdividers do earn huge profits (sometimes put at hundreds of percent per year), public policy could be designed eo encourage legitimate private developers to supply lots with higher service levels while still allowing them to obtain good returns. This was, in fact, an argument in favor of the normas minimas laws. - 18 - II. GENERAL CHARACTERISTICS OF SUBDIVISIONS This chapter is a descriptive overview of subdivisions and lots sold, with particular attention to differences between pirate and normas minimas subdivisions. A. Basic Features Size -and Density: The,149 subdivisions in the sample range widely in area,.the smallest being around 0.1 hectare and the largest 43.5,hectares. 2 The average gross subdivision size overall is 49,475 m2, or about 5 hectares or 12.2 acres. The subdivisions in the sample contain as few as five and as many as 2,000 lots (Table 4); the overall average is 196 lots per sub- division. Most of the subdivisions are not large, however; half have less than 100 lots and over one-third have fewer than 50. On the average, pirate subdivisions outside the urban perimeter have the same number of lots as those inside, but their average lot size is about 2.5 times larger (125m2 vs. 315 m2 2 The average lot size of 125 m in pirate subdivisions inside the urban perimeter implies a density of about 56 lots per hectare, taking into account average street and open space area. Normas minimas density works out to an average of about 62 lots per hectare. Some subdivisions in the sample have densities as high as 100 lots per hectare. Assuming six persons per lot, gross residential densities are in the range of 300 to 600 persons per hectare. When Developed: Four-fifths of the subdivisions in the sample were initiated after 1967, and half of these sprang up during the last five years (see Table 5). There appears to be little relationship between size and age among the pirate subdivisions studied. - 19 - Table 4: DISTRIBUTION OF SUBDIVISIONS BY NUMBER OF LOTS Size Ca'tegory Number of Percent of Cumulative (Number of Lots) Subdivisions Subdivisions Percent 1-24 23 15.4 15.4 25- 49 33 22.2 37.6 50-99 20 13.4 51.0 100-199 26 17.4 68.4 200-299 19 12.8 81.2 300-399 8 5.4 86.6 400-499. 5 3.4 90.0 500-749 7 4.7 94.7 750-999 5 3.4 98.1 1000 + 3 2.0 100.0 TOTAL 149 100.00 - 20 - Table 5: NUMBER OF SUBDIVISIONS BY INITIATION YEAR Age Category of Subdivision Number of Percent Based on Initiation Year Subdivisions of Subdivisions Old (pre 1967) 28 18.8 Middle-aged (1967-72) 60 40.3 Young (197,3-77) 61 40.9 TOTAL 149 100.0 -21- Amount of Land Developed in Bogota: The 125 subdivisions that are within the urban perimeter represent a total of-534 hectares or about 1,33,8 acres of residential land, in gross terms. Using Borrero and Sanchez's estimate of Bogota's total residential area of 14,162 hectares (1973), the sample subdivisions comprise approximately 3.8 percent. Assuming, for rough estimation, that pirate land is one-third of total residential area in Bogota, the sample represents about 12 percent of pirate land area in the city. Location: Twenty-four of the pirate subdivisions studied (16%)are located outside the 197a urban perimeter of Bogota. Most of these are within the Special District of Bogota, but a few are located in towns some distance away in the adjacent Department of Cundinamarca'. Eight subdivisions are in Soacha, about four kilometers beyond the city limits to the southwest. Several subdivisions are in the far north-northwest, two of these in Funza, six kilometers from the perimeter; one in Mosquera, seven kilometers away; one in Madrid, ten kilometers away; and one in Bojaca, even further distant. Ten subdivisions in the sample are beyond the urban perimeter to the south, seven of these along the highway to Villavicencio. As Figure 1 shows, pitate subdivisions cluster in the far southern and western areas of Bogota, with a significant but much smaller number located in the north-northwestern zone. The newest subdivisions are in the west, the older ones in the south. Figure 2, which depicts the relationship between location and tract size, indicates that subdivisions outside the urban perimeter are substantially larger overall. The large average size of subdivisions in the near west is due to the clustering there of eight normas minimas developments which average 12.6 hectares in size, twice the average for the zone as a whole. - 22 - Figure, 1 LOCATION OF PIRATE AND NORMAS MINIMAS DEVELOPMENTS IN SAMPLE 0• bladrid7 Funza. Bojaca' Mosquera and outar reng • 961 0 NorNas Minias-subdi- 0 1 2 3 4 ... ... . .. 23- Figure 2 AVERAGE TRACT SIZES ACROSS RINGS AND SECTORS OF BOGOTA 1200 SOUTH 1000 NORTH/NORTHWEST 800 Average 600 Tract Size (100s of M2) WEST 400., (without normas minimas) 200. p II First Second Outside Ring Ring Perimeter 5-10 km 11-20 km over 20 km (Kilometers from center) -24- There is no hard evidence on why normas minimas subdividers creata much larger subdivisions than pirates. One possible explanation is that normas minimas subdividers are likely to have,better access to financing, enabling them to obtain more capital and to invest in larger tracts of land. Another possibility is that normas minimas subdividers are more efficient; that is, they may be more apt to take advantage of potential economies of scale in the subdivision business. Lots Sold: The Superintendencia Bancaria's questionnaire asked subdividers to submit data on all lots sold in 'each subdivision. The data set includes figures on 19,555 lots in 147 of the 149 subdivisions, or about 68 percent of the total number of lots subdivided in the sample. The re- mainder are lots unsold and a small residual of lot sales not reported. Of those lots sold, 13,639 are in pirate subdivisions and 5,916 in normas minimas developments. It has not been possible to estimate what fraction this represents of the universe of pirate lots' sold in Bogota in recent years. Listed in Table 6 are the number of sample lots sold by year: 87 percent were sold since 1970; three-quarters in the last five years; one-third in 1976-77. Land Uses in Subdivisions: Across all subdivisions an average of 70 percent of the area is devoted to salable lots, 11 percent to communal area and green zones, and 19 percent to streets (see Table 7). Because of the legal standards they must meet, normas minimas subdivisions provide. about twice as high a proportion of communal and green space, at the expense of about 20 percent less salable area, than pirate subdivisions. The proportion - 25 - Table 6: TOTAL LOTS SOLD BY YEAR IN SAMPLE Years of Sale Number of Percent o. Cumulative of Lots Lots Total Percent Pre-1960 109 0.6 0.6 1960-64 56. 0.3 0.9 1965-69 2398 12.3 13.2 70 436 2.2 15.4 71 654 3.3 18.7 72 1135 5.8 24.5 73 1610 8.2 32.7 74 3156 16.2 48.9 75 3594 18.4 67.3 76 3526 18.1 85.4 77 2827 14.5 100.0 Sub-total 19506 100.0 No Date 49 TOTAL 19555 - 26 - Table 7 AVERAGE LAND USES WITHIN SUBDIVISIONS (Percent of tract) Average Per nt of All Pirate Inside Pirate Outside Normas Tract Devoted to: Subdivisions Perimeter Perimeter Minimas Usable (salable) Area 70 71 74 56 Communal and Green Area 11 10 10 22 Street Area 19 19 16 22 Average Communal + Green Area per Lot (m2) 29 26 38 39 Average Street Area per Lot (M2) 47 43 70 37 Number of Subdivisions 147 109 24 14 - 27 - of usable area in most of the normas minimas developments is between 50 and 60 percent. Among pirate subdivisions, however, the fraction varies from over 80 percent to 100 percent of the tract surface. Regarding the fraction of land, devoted to streets, the average is 19 percent in pirate subdivisions and 22 percent in normas minimas. The figures on space per lot in Table 7 also show that normas minimas subdividers offer greater amounts of open and communal area. The somewhat lower average street area per lot in normas miimas subdivisions reflects their smaller average lot size. B. Purchase of Tracts In.. 76 of the 149 subdivisions, works were initiated less than 12 months after the tract was acquired. Twenty-nine other sub- divisions-- six of which were normas minimas - were actually begun - before the official acquisition of the tract, although for half of these these the period was less than one year. This is a 1/ reflection of the loose, clandestine nature of the business.- In contrast, 40 subdivisions were not initiated for two or more years following acquisition, and in 28 of these cases -- one-fifth of the sample -- the subdivider waited six years or more before proceeding. These patterns suggest that while most pirate subdividers behave like businessmen with discount rates -- they want to obtain returns as soon as possible -- a substantial number seem to be either 1/ According to the Superintendencia Bancaria, most normas minimas developers are former pirate subdividers. - 28 - inexperienced or in no hurry to develop their parcels of land. Comments on the questionnaires indicate that many subdividers had owned the tracts for long periods and a few had even inherited the land, suggesting that some fraction of pirate subdividers are one-time entrepreneurs. The tendency for some pirate subdividers to wait, perhaps for a rise in land prices, is also reflected in the relationship between the dates of official tract acquisition and the beginning of lot sales. In 64 of the 149 cases, this period is less than one year; in 12 more, it is between one and two years. In 43 subdivisions more than two years passed before any lots were sold, and in 30 of these the delay was over six years. At the same time, there are 28 subdivisions, including six normas minimas in which lots were sold before the tract was officially acquired, in most cases less than one year before. This is another indication that the business often involves unorthodox transactions, including the use of receipts from lot sales to meet payments for the tract. A majority of subdividers in the sample paid for their tracts in a lump sum. In 39 cases, 42 percent of those with information available, the subdivider made a down-payment and paid installments on the tract. The down-payments averaged about 35 percent of the total tract price. In only 23 cases was an interest charge - ranging from 0.3 percent to 7.0 percent annually - reported to be part of the tract acquisition contract. Figure 3 presents average prices of subdivision tracts in Bogota for the northwest, west, and south./ Within the urban 1/ These land price measurements are consistent with those of McCallum (1974) and Villamizar (1979). - 29 - Figure 3 AVERAGE TRACT PRICES PER M2 ACROSS RINGS AND SECTORS OF BOGOTA 144 128 112 96 80 Average Tract Prices per m2 (constant 1976 Col$) 48 32 WEST NORTH/NORTHWEST 16 SOUTH First Second Outside Ring Ring Perimeter 5-10 1cm 11-20 km Over 20 km (Kilometers from center) - 30 - perimeter the average price per square meter of tracts is consistently highest in the north, Bogota's higher income area. land prices are abor,t twice as high in the west as in the south. , In all sectors land prices decline with distance from the center of the city, alth8ugh in the southern zone the gradient is (all. As Figure 4 illustrates, peripheral tract prices per squa*e meter in Bog&a have been rising in real terms, particularly between 1974 and 1977. While price increases have been relatively gradual in the west and south, the northern periphery has recently experienced spectacular land price growth. The combination of lower relative prices and lower real price growth helps explain why pirate subdivisions are concentrated in the west and south. C. Infrastructure and DevepnCt Because the questionnaire failed to ask about planned infrastructure investments, the data on expenditures and quantities of services may be, on an absolute per-subdivision basis, lower than eventual levels. Nonetheless, this does not interfere with either comparisons of relative magnitudes or the hedonic lot price analysis later on. Of the 149 subdivisions studied, 20 reportedly had no infrastructure--wae., sewer, electricity, telephone, streets, side- walks, and curbs-- installed at all; the rest had at least one type installed. As Table 8 indicates, the most .frequent infrastructure types present are streets (three-quarters of the subdivisions), water (two-thirds), and electricity (almost two-thirds). All but - 31 - Figure 4: AVERAGE TRACT PRICES PER M2 OVER TIME (Constant 1976 Col$) 220 North-northwest 180- 140 Average Tract Prices Per M2 100 West South 60 20 Pre- 1970-71 1972-73 1974-75 1976-77 1970 Time Periods - 32 - Table .8: EXTENT OF INFRASTRUCTURE INSTALLATION IN SUBDIVISIONS Percent of Subdivisions Having Services at Time of Survey All Subdivisions Pirate Notmas Minimas Sewer 46.9 46.7 50.0 Water 68.5 65.2 100.0 Electricity 63.8 61.5 85.7 Telephone 20.8 17.0 57.1 Streets 73.2 70.3 100.0 Sidewalks 16.1 16.3 14.3 Curbs 27.5 23.7 64.3 No Services Whatsoever 20 20 0 % 13.4 14.8 0.0 Total Number of Subdivisions 149 135 14 -33- two of the normas minimas subdivisions have water, electricity, and streets, while half of them have sewer service. While in the normas minimas subdivisions all the infra- structure was paid for by the subdividers, in the pirate subdivisions some infrastructure was financed by other sources. Indeed, the data in Table 9 demonstrate that lot buyers themselves often edup paying for the installation of services. Infrastructure investments usually occur well after acquisition of the tract and initiation of the subdivision. In over three-fourths of the cases a delay of one year or more occurred between the sale'df lots' and the installation of any.infrastructure. Of 93 subdivisions for which the installation date of at least bne type of infrastructure is given, only 18 had a service or services installed prior to the sale of the first lot. In 39 cases, infra- structure was not installed until two or more years after the sale of the first lot, and in 17 the delay was more than five years. In 41 of the subdivisions -- 44 percent of those reporting -- infra- structure provision was delayed until after 50 percent.of the lots had been sold; in 28 of these - five of which are normas minimas - the infrastructure was not installed until after 90 percent of lots had been sold. Streets and water standpipes tend to be installed earlier relative to lot sales than water and sewer lines. The lag in infrastructure provision suggests that: (1) subdividers generally cannot afford to make infrastructure expenditures "up front"; hence they wait until after returns from lot sales come - 34 - Table 9: SOURCES OF FINANCING FOR INFRASTRUCTURE IN PIRATE SUBDIVISIONS Percentage of pirate subdivisions in which service was paid by: Both sub- N- Number of Sub- divider and Subdivisions divider Community community Government with service Sewer 43.5 32.2 16.1 8.1 62 Water 62.8 23.2 5.8 8.1 86 -Electricity 48.8 31.7 6.1 13.4 82 Telephone 4.8 61.9 4.8 28.6 21 Streets 81.4 12.8 2.3 3.5 86 Sidewalks 21.0 47.4 5.3 26.3 19 Curbs 53.3 30.0 3.3 13.3 30 Note: In all 14 normas minimas subdivisions 100 percent of all services were financed by the subdividers. The normas minimas subdivisions are excluded from this table.. - 35 - in; (2) some subdividers are unwilling to install infrastructure at all and do so only after lot buyers and/or the municipal authorities force them to at a later time; (3) Lot buyers who are unable to convince or coerce the subdivider into installing infrastructure at his own expense often choose to compromise by sharing costs with him,paying for the services themselves, or obtaining services financed directly by the public sector. Pirate subdividers most frequently finance sewer, standpipes, electric lines, streets, and curbs. Lot buyers pay more often than pirate subdividers for telephone lines and side- walks, although about one-third of the time they pay for sewer and electric lines. The public sector plays a minor role and does not seem to finance certain infrattructure types more often than others. Average quantities of each infrastructure type installed, presented in Table 10, represent only completed - not planned - investments. While the amounts of infrastructure vary widely across subdivisions, normas minimas subdividers clearly provide more than do pirates. Tae average prices of different infrastructure types per unit, shown in Table 11, were obtained by dividing expenditures by quantities of works completed. The prices vary greatly across sub- divisions and should be interpreted cautiously. On the average one would expect normas minimas subdividers to pay higher prices, assuming that they supply higher quality infrastructure. At the same time, pirate subdividers may pay higher prices for some types of infrastructure because they purchase-inefficiently. According to Table 10: AVERAGE QUANTITIES OF INFRASTRUCTURE INSTALLED PER HECTARE USABLE AREA All Sub- Normas Maximum Minimum* Infrastructure Type Units divisions Pirate Minimas Amount Amount Sewer Lineal Meters 186(128) 168(115) 340(13) 1166 89 Electricity Lineal Meters 292(104) 279(98) 496(6) '255 38 Telephone Lineal Meters 11 6 57 (127) (118) (9) 510 15- Streets M4 2041 1988 (121) 1 (115) 3045 (6) 6100 330 Sidewalks M2 47 42 87( 1060 180 (136) (122) (14) Curbs Lineal Meters 97 91 178 1337 66 (132) (123) (9) Note: Figures in parentheses are the number of cases. Specific. quantities of infrastructure were not reported for some subdivisions even though the services were present. * Figures based only on cases where quantity is greater than zero. 01 Table 11: AVERAGE COSTS OF INFRASTRUCTURE INSTALLATION PER UNIT (Constant 1976 Col$) . . Verification Infrastructure Unit of All Normas (Including Labor and Materials) Type Measurement Subdivisions Pirate" Miiiihias Item Price Source* Sewer Lineal meter 407 399 456 8" Pipe 262 (2) Drainage Conduit 2969 (2) Manhole 8568 (2) Water Pipes Lineal meter 398 - 422 253 8" Pipe 237 (1) 10" Pipe 291 (1) Manhole 4610 (1) Pilas Unit 21,560 23,688 13,993 -(not available) Electricity Lineal meter 342 359 203 (not available) Streets m2 83 72 165 Macadam with base 106 (1) Macadam with base 65-86 (2) Asphait 166 (2) Crushed stone, sand, gravel 15 cm. 52 (2) * Sources for price verification: (1) Contracts .between subdivider and contractors for infrastructure .installation, 1977, supplied by Superintendencia Bancaria. (2) Estudio de Precios Unitarios, Ministry of Public Works, 1978. -38- the data, normas minimas developers do in fact pay more than pirates for sewer and street construction, but pay less - by about two-thirds to three quarters --for water pipes, standpipes, and electric lines. Since inefficient purchasing probably does not account for this large a difference, the high figures for pirate subdividers may reflect some instances of erroneous reporting on the questionfiaires. Never- theless, the price figures in Table 11 are reasonable in overall magnitude. Total expenditures by subdividers, lot buyers and govern- ment on infrastructure per square meter of usable area, shown in Table 12, average around Col$38 at 1976 prices for all subdivisions. Infrastructure investments in normas minimas developments are considerably higher than in pirate subdivisions (Col$57 vs. Col$36). Expenditures reported in this sample seem reasonable in comparison with other sources; for example, the amount spent in pirate sub- divisions in Bogota is about one-quarter to one-third the amount budgeted by the World Bank for infrastructure in low-income urban upgrading projects in El Salvador. Table 13 shows average subdivider expenditures for professional services, publicity, and administration, calculated on a per-lot basis. As explained in Chapter I, some of these figures 1/ Documentation on prices for construction of sewer and water lines and streets was made available by the Superintendencia Bancaria. These verification figures, which also appear in Table 11, are close to the averages-given by the data. However, because details on the types of infrastructure installed are missing, it is difficult to say what level of quality lot buyers are receiving at these prices. - 39 - Table 12: AVERAGE EXPENDITURE ON ALL INFRASTRUCTURE PER M2 USABLE AREA (Constant 1976 Col$) 2 Col$ Per m Number of Usable Area Stbdivisions All Subdivisions 3 8.5 116 Pirate 36.0 102 Normas Minimas 56.9 14 Average. for 4 squatter upgrading projects in - - El Salvador 1/ 92.6 Average for World Bank El Salvador Second Urban Development Project 2/ 130.2 "Urbanization Primer" Mi- nimum Level, Layout No. 2 3/ 91.0 1/ EDURES, A Program for the Integrated Development of Critical Metropolitan Areas in El Salvador, Document No. 25, Vol. 1, May 1978. 2/ World Bank, El Salvador Second Urban Development Project, Appraisal Report, No. 1401a-ES, April 1977. 3/ Horacio Caminos and Reinhard Goethert, Urbanization Primer, World Bank Urban Projects Department, October 1976. -40 Table 13: AVERAGE PROFESSIONAL SERVICES, PUBLICITY, AND ADMINISTRATIVE EXPENDITURES PER LOT (Constant 1976 Col$) Expenditures per Lot Categories of over- Pirate Subdivisions Normas Pirate Subdivisions head expenditures Inside Perimeter Minimas Outside Perimeter Professional Services 2,150* 833 940 Publicity 218 498 '257 Administration, 3,819* 3,419* 1,866 Number of Subdivisions (148) (134) (14) *Absolute value probably exaggerated as reported. - 41 - are overstated. For pirate subdivisions the means-for professional services and administration costs are unduly high; for normas minimas the administration cost mean is exaggerated. According to Superintendencia Bancaria officials, many subdividers (including the normas minimas developers, most of whom had not complied with all legal norms at the time of the survey) attempted to overstate their costs to reduce their apparent profit rates. D. Sale of Lots According to the sample data, it takes an average of 47. months to sell 90 percent of the lots inaa pirate subdivision. For normas minimas the period is 17 months. This large,difference may reflect excess demand for normas minimas lots. It also may suggest that normas minimas subdividers are more efficient salesmen. It is widely believed that buyers of pirate lots are unreliable in their payments. This is offered as one reason why "legitimate" developers are reluctant to go into the lower-inoome housing market. The payment rates displayed in Table 14 indicate-that an average pirate subdivider working within the urban perimeter can expect about 30 percent of his lot buyers to fall behind in their payments. Of these, more than half will eventually lag six months or more. A normas minimas subdivider can, on the average, expect a slightly higher rate of late payments as well as a somewhat larger proportion of serious delays. Pirate subdividers operating outside the urban perimeter seem to have much graver problems; on the average- -42- Table 14.. AVERAGE RATES OF LATENESS IN PAYMBNTS AMONG LOT BUYERS Lot Buyers In Pirate Subdivisions Normas Minimas Pirate Subdivisions Inside Urban Perimeter Subdivisions Outside Urban Perimeter Percent late 30.5 33.7 50.9 Percent late more than six months 18.2 22.1 42.8 Late more than six months as .percent of all late 59.6 65.4 84.0 Total number of lot buyers 11,540 5,916 2,099 Total number of sub- divisions 109 - 14 24 Average number of lot buyers per sub- division 106 423 87 - 43 - half their buyers end up late in payments, and 84 percent of 'these fall behind six months or more. Clearly the problem of lateness in payments and long-term default is significant in the lower-income lot business. On the one hand, these results do seem to reflect the riskiness of the business from the subdivider's point cf view. On the other hand, given that many pirate subdividers do not deliver what they promise, it is understandable that lot buyers may have good reasons to stop their payments. In fact, the experience of the Superintendencia Bancaria and comments on the questionnaires indicate that lot buyers deliberately hold up payments in response to a subdivider"s failure to make improvements or install services. Although it is not possible to determine how many cases like this occur, an educated guess might be 25 percent of the late debtors. The prices of both pirate and normas minimas lots have been rising faster than consumer prices overall in Bogota, as shown in Figure 5. The real rate of increase in pirate lot prices has been especially high since 1973. The average inflation rate for the 1973-77 period was about 24 percent, while the average annual nominal growth rate in prices per square meter of pirate lots was about 111 percent. At the same time, the average size of pirate lots has gradually been decreasing. From 1972 to 1977 the average pirate lot had diminished in size from an average of 2 2 about 180 m to about 105 m2. This reflects increasing costs of land development, as well as continuing high demand for pirate lots. Figure 5 AVERAGE LOT PRICES PER M2 OVER TIME 600 Average lot prices (nominal) 500- 4001 verage lot prices (constant Average 1976 Col$) lot price Xirate only (Col$ per m ) 36 3001 200- 100- 52 54 56 58 60 62 64 66 68 70 72 74 76 78 Years Average lot prices per square meter decline with distance from the center of the city, as Figure 6 illustrates. Prices are highest in the northwestern zone within the perimeter, lower by about one-fifth in the west, and lower by about one-third in the south. Average lot sizes within the urban perimeter, shown in Figure 7, vary little with distance from the center, and,even 2 across sectors the range is only between 100 and 140 m . The greatest variations in average lot size occur outside the urban perimeter. E. Total Costs of Development The distribution of major costs in total expenditures for pirate and normas minimas subdivisions, shown in Table 15, indicate that both types of subdividers spend. about the same proportion of total outlays on land, but normas minimas subdividers devote a greater percentage of resources to infrastructure. Pirate sub- dividers report spending a much larger fraction on overhead. Compared with 19 World Bank sites and services projects around the globe, Bogota subdivisions are more land-intensive. Average total expenditures per lot appear in Table 16. While normas minimas subdividers pay about 28 percent more per square meter of land than pirates, their land costs per lot are actually lower on average because they develop smaller lots. There is almost no difference between development costs per lot for pirate subdivisions inside the perimeter and normas minimas sub- divisions. Even if the likely exaggeration of overhead expenditures - 46 - Figure 6 AVERAGE LOT PRICES PER M2 ACROSS RINGS AND SECTORS OF BOGOTA 500 400. Average lot 300 prices per m2 'WEST (constant 1976 Col$) 200- 100 - NORTH/NORTHWEST SOUTH First Second Outside Ring Ring Perimeter 5-10 km 10-20 km 20 km and over (Kilometers from center) - 47 - Mra7 AVERAGE LOT SIZE ACROSS RINGS AND SECTORS OF BOGOTA 40 0 SOUTH NORTH/NORTHWEST 300 Average lot sizes (m2) 200 WEST 100 _ First Second Outside Ring Ring Perimeter 5-10 km 10-20 km 20 km and over (Kilometers from center) -48- Table 15 DISTRIBUTION OF MAJOR COST COMPONENTS (Percentages) Pirate Normas World Bank,/ Component Subdivisions Minimas Projects Land 39 42 21 Infrastructure 22 32 33 Professional Services 15 3 132g/ Publicity 2 2 Administration 14 13 Unspecified/Miscellaneous 8 8 Plot Development- 33 100 100 100 Number of Subdivisions 133 14 19 Source: Praful C. Patel, Sites and Services Projects: Surve and Analysis of Urbanization Standards and On-Site Infrastructure. World Bank, August 1974, Annex A, p. 11. In the World Bank projects this is "site preaCration; topo/survey work . Construction of sanitary cores or dwelling shells. -49- Table 16 AVERAGE TOTAL DEVEIDPMENT COSTS PER LOT 1976 Col$ Pirate Subdivisions Normas Pirate Subdivisions Cost Per Lot Of: Inside Perimeter Minimas, Outside Perimeter Land 11,387 10,798 10,171 Infrastructure 3,256 5,955 3,695 Overhead 6,187 4.,750 3,063 Total 20,830 21,503 16,929 Number of Subdivisions 109 14 24 *Probably exaggerated. - 50 - is taken into account, by reducing pirate overhead by about half and normas minimas overhead by about one third, the total development costs per lot are still practically the same (roughly Col$17,700 for pirate inside the perimeter vs. Col$18,300 for normas minimas). These figures add to the evidence that normas minimas subdividers make higher profits on average than pirates. Normas minimas sub- - dividers spend only marginally more per lot, but they sell at a substantially higher markup and sell four times the number of lots on the average per subdivision. - 51 - III. RATES OF RETURN TO SUBDIVIDERS A. Background Using data from the Superintendencia Bancaria sample, this chapter will examine the magnitude of rates of return to subdividers and discuss the policy implications of the findings. Some observers have suggested that, since pirate subdividers make large profits,'the public sector should attempt to capture the excess profits and divert them to the provision of adequate levels of services (Doebele, 1975). This could be accomplished either by forcing subdividers to install infrastructure or by providing incentives to legitimate developers zo supply legal lots with services. There is some difficulty here in defining what constitutes an exorbitant rate of return. An annual profit rate of 30 percent may seem large-until one takes into account that inflaction in Bogota averaged around 24 percent yearly between 1972 and 1977. Moreover, a 30 percent nominal rate of return is only slightly better than the yields offered by most of the major financial instruments currently available in Colombia (see below). Finally, nominal- Effective Annual Rates of Return, June 1979 UPAC (indexed savings andloan) ordinary deposits 19.30 UPAC savings accounts 24.95 UPAC savings certificates six months 26.14 one year 27.33 Term deposit certificates three months 26.82 six months 28.07 Savings accounts 19.00 Agro-industrial Titulos (bonds) three months 27.50 six months 28.00 Bonos Cafeteros (coffee bonds) 22.00 Titulos de Ahorro Cafetero (coffee savings bonds) @ 65% and 2-year term 51.29 Certificates of exchange 35.60 Source: Estrategia Economica y Financiera, Bogota, June 1979, No. 23, page 10. - 52 - profit.rates in high-standard housing development ventures during the 1977-78.period were reported to be in the range of 50 to 100 percent 1/ annually.- Losada and Gomez (1976) calculated net benefits for two of the subdivisions they studied. In both cases, the subdivider's expenditures consisted of the tract cost, office costs, commissionists' fees, and expenses for grading of streets. Revenues came from,the sale of lots. -The calculations yield net benefits of 229 percent and 227 percent over five years in constant prices, with the cash flow discounted'at 10 percent annually. 2 These figures are equivalent to an average annual real net benefit of about 27 percent (compound), a good but not outrageous profit margin. Doebele (1975) offers rougher profit figures estimated from Superintendencia Bancaria records for six pirate subdivisions of widely varying size and quality. Doebele estimates "presumptive" profits, based on several asumptions, of 35 to 1,300 percent, with an average of 182 percent. - Doebele suggests that lot sales occur over two or three years, which is consistent with the sales periods observed in this study. At three years, the average annual net prcfit for the six subdivisions is 98.5 1/ Based on background information supplied by several middle and upper- middle class development firms in Bogota, February 1979. 2/ Losada and Gomez, pp. 52-55. 3/ The assumptions are: 67 percent of the tract area is salable and 5 percent of the tract value covers overhead costs. - '53 percent (compound). One difficulty with these figures is that they are based on undiscounted cash flows. Discounting over the lot sale period would lower the present value of benefits and-net benefit margins. In addition, it can be argued that Doebele's estimates are maximum possibilities because the expenditures do not include any infrastructure. Finallyi -,the prices used in Doebele's calculations are current, not constant values. Real net benefits would be significantly lower. It is worth noting that Doebele's figures do reflect the great variation in performance among sub- dividers. Some make enormous profits, while others barely scrape by or even lose money. B. Rates of Return in the Smp.le The Superintendencia Bancaria data provide the first opportunity to measure the profitability of a large number of sub- divisions using a consistent criterion. The measure employed here is the internal rate of return (IRR). The IRR is the discount (interest) rate at which the present value of a net cash flow equals zero. I Two attributes make the IRR a better way to measure profitability than discounted net benefits. One is that there is no need to choose a discount rate. The other is that the IRR is 1/ See E.J. Mishan, Cost-Benefit Analysis. New York: Praeger, 1976; pp. 183-195 conceptually analogous to the periodic percent return an entre- preneur may obtain from other investments--such as bonds. The IRR is essentially the compound interest rate at which the initial investment grows over a given period. The IRR does have one sig- nificant disadvantage. Given a cash flow with unusual characteristics, particularly substantial outlays late in the period, the equation for the IRR may yield more than one root. In this case it is common practice to use the first root as the "dorrect" IRR. In the results presented below, between 10 and 19 (depending on the assumptions underlying the calculations) of the 147 subdivisions with lot sale data presented multiple IRRs and were omitted from the analysis. Table 17 summarizes average rates of return across all subdivisions in the sample. The "medium" average IRRs are based on the data !'as is"; i.e., expenditures and revenues reported by subdividers are unmodified and all lots are assumed to be paid for on time. The "payment default" averages take into account a percentage of failure in lot payments, and the "favorable" rates of return assume that subdividers exaggerated some of their overhead 1/ Appendix II provides an explan4WA of the assumptions used in the alternative IRR calculations. An IRR for each sub- division was calculated using a computer program that assembled a net cash flow and then computed the IRR by successive ap- proximations (see Appendix II for details). The program allowed the testing of the IRR's sensitivity to various assumptions, such as the degree of lateness in lot payments, and permitted easy substitution of real for nominal prices. The program could not, however, calculate IRRs for unusual caffiflows. - 55 - Table 17 ALTERNATIVE AVERAGE NOMINAL RATES7OF RETURN TO SUBDIVIDERS-/ (Figures are monthly; yearly figures in parentheses; rates are unadjusted for inflation) Pirate Subdivisions Normas Pirate Subdivisions Inside Urban Perimeter Minimas Outside Urban Perimeter Monthly rates of return: 2/ Payment default assumptions- Median 1.9 2.8 2.4 (25). .(39) (33) Mean 2.7 ( 6.0 3.7 (38) (0)(55) Standard deviation 3.0 5.7 (94) 4.8 ' (42)(94)(75) N 101 14 21 3/ Medium assumptions- Median 2.4 4.929 (33) (78) 2.9 (41) Mean 3.6 7.04. (53) (125) 4.2 (64) Standard deviation 3.6 5.6 )4.9 53) .(92) (8 N 102 14 21 4/ Favorable assumptions- Median 3.2 7.8 (64.0 (46) (146) (60) Mean 4.6 (72) 10.9 (246) 5.5 (90) Standard deviation 5.0 12.6 6.2 (80) (315) (106) N 95 13 20 1/ Rates of return here are averages over internal rates of return calculated for individual subdivisions. Calculations are based on complete revenue and expenditure figures, with dates, for each subdivision. Basic assumptions in all IRR calculations: All lots in the subdivision are sold; if the total number of lots is greater than the number of lots for which sale data is given, the "excess" lots are counted as sold at the average size and price (in 1977 pesos) and on the last date of sale among the lots listed. 2/ Payment default assumptions: A certain proportion of defaulting lot buyers is given per subdivision in the data;'this incidence of default-is assumed to occur randomly over the period of lot sales; defaulting lot buyers are assumed to pay only the down payment. 3/ Medium assumptions: No default; all lots are fully paid for on time. 4/ Favorable assumptions: Overhead costs are limited to one standard deviation above the mean of figures for normas minimas developments. - 56 - costs. These alternative estimates are intended to bound the results within a valid range by accounting for the most serious sources of uncertainty in the data. Table ,17 shows that profits in both pirate and normas minimas subdivisions are good on the average. The median rates of return are substantially lower than the means, indicating that there are a few enormously lucrative subdivisions in the sample. Because of the presence of these high-value "outliers" that skew the mean. upward, the median is a better measure of central tendency here, While median nominal rates of return for pirate sub- divisions are are well below 100 percent annually ( a range of 25 to 60 percent), they are much higher for normas minimas, up to 146 percent under favorable assumptions. This confirms what was suggested by the averages in Table 1. Although normas minimas sub- dividers pay more for land and spend more on the provision of infra- structure, they earn larger profits than their pirate counterparts because they sell lots at a higher price, they sell more of them, and they sell them faster. Two features of the absolute magnitudes of the IRRs are noteworthy. One is that the medians for pirate subdivisions are modest. According to these data, pirates earn a nominal IRR of 30 to 40 percent annually, which is much lower than earlier estimates. Second, there is a very large amount of variance in IRRs among both pirate and normas minimas subdivisions. Standard deviations exceed -57- means in almost all cases. This supports the idea that subdividers in this market are a heterogeneous group. C. Determinants of Rates of Return What accounts for the large variation in subdivider's profits? To help answer this question, a simple linear and additive regression model was constructed with the nominal rate of return as a function of subdivision characteristics, expenditure levels, and lot prices. Although the model, presented in Table 18, accounts for less than a third of the variation in rates of return, some inferences can be drawn. ..Increments in tract prices seem to have strong negative effects on rates of'return, all else held constant, while marginal expenditures on infrastructure and overhead have weak negative effects. The coefficient for the number of lots in the subdivision is weak and negative, contrary to expectations raised by the large and lucrative normas minimas examples. Average lot price is positive and has a fairly strong effect: a Col$1,000 increase in average lot value yields around a 0.5 percent increase in monthly nominal IRR.. Profitability also seems to improve markedly with how recently the subdivisions ws developed, with the extent to which lot buyers are paying an install- ment terms (prices are higher on installment), and with how soon the subdivider begins selling lots after he acquires the tract. Dummy 1/ Background information collected by Bogota officials shows that they are a mix 6f large and small-scale operators, of the educated and the nearly illiterate, and of experienced and naive entre- preneurs. 58- Table 18 RELATION BETWEEN RATE OF RETURN FOR LOW-INCOME SUBDIVISIONS AND SUBDIVISION CHARACTERISTICS Linear Estimation of Form: IRR =A + Ba 2Var ** Variable Definitions: INPLOT Expenditures by subdividers on infrastructure per lot. BUYPLOT Expenditures by lot buyers on infrastructure per lot. WAIT Number of years between tract acquisition and sale of first lot in subdivision. NNW Dummy variable equal to 1 if subdivision located in north- northwest zone of city (north of Avda. Las Americas), O otherwise. WEST Dummy variable equal to 1 if subdivision located in western zone of city (industrial corridot/Bosa), O otherwise. DIST Kilometers between DANE barrio in which subdivision is located and CDB (defined as intersection of Avda. Caracas and Calle 26). LOTS Total number of lots in subdivision. ALVL Average lot price in subdivision in 100s of 1976 Col$. TIME Year of sale of first lot indexed as 1950 = 1, 1951 = 2, etc. 2 ALSZ Average lot size in subdivision (m ). 2 TRPM2 = Tract price per m in 1976 Col $. APDP - Average down payment as a percent of average lot price for the subdivision. NM= Dummy variable equal to 1 if subdivision is normas minimas, 0 otherwise. PROL = Subdivider expenditure on professional services per lot in 1976 Col $. PUBL = Subdivider expenditure on publicity per lot in 1976 Col $. ADML = Subdivider expenditure on administration per lot in 1976 Col $. -59- Table 18 (Continued) Dependent Variable = Internal Rate of Return - in 10ths of Percent Per Month Coefficients for Subsamples: (Prices in 1976 Col $) . .Pirate Tract Inside Inside Profitable All Sub- Acquisition Urban Perimeter Subdivisions Variable divisions 1970 or Later Perimeter (w/o NM) Only INPLOT -.00126 -.0014 -.00106 . -.00106 -.0014 BUYPLOT -.00003 .00107 .00005 .00009. -.00011 WAIT -.0674 -.0599 -.0855 -.0615 -.078, NNW -2.22 1.13 6.12 5.57 .098 WEST 11.19 19.44 16.17 11.59 14.77 DIST .138 .543 1.69 2.029 -.105 LOTST -.00847 -.0193 -.00014 -.026 -.0056 ALVL .0497 .071.4 .0408 .022 .055 TIME 1.54 .353 1.47 1.62 2.88 ALSZ .0207 -.00921 -.044 -.056 .016 TRPM2 -.142 -.231 -.136 -1.21 -.159 APDP -.390 -.598 -.309 -.273 -.375 NM 17.73 11.37 PROL -.00254 -.00713 -.0023 -.0021 -.0030 PUBL -.00367 -.0055 -.0013 .0028 -.0074 ADML -.00106 -.0034 -.0009 -.00067 -.0011 Constant 21.65 72.08 10.67 10.86 -.77 R .295 .351 .311 .312 .313 R2 .206 .201 .213 .198 .227 Std. Error of Est. 37.53 43.64 36.34 32.95 37.06 Number of Observations (135) (81) (114) (100) (127) Note:: Underlined coefficients are significant at the two-tail .05 level. 1/ Internal rate of return used here is "medium assumptions" figure. - 60 - variables for location in the western part of the city and for normas minimas subdivisions yield extremely high but non-significant coefficients. . Among the explanatory variables omitted from the model are the speed of lot sales and the subdividers" experience and organizational ability, which probably account for a good part of the unexplained two thirds variation in rates of return. It is likely that the most lucrative subdivisions are developed by subdividers who are well-organized, experienced, and well-connected to government officials,subcontractors, and sources of credit. D. Real vs. Nominal Rates of Return Real rates of return (calculated with constant prices) are compared with nominal rates in Table 19. For pirate subdivisions real medians are about one-third (median assumptions) to one-half (default assumptions) lower than nominal. For normas minimas, however, median real and nominal rates of return hardly differ. This suggests that normas minimas subdividers are more skilled than their pirate counterparts at manipulating their expenditures, prices and sales in such a way as to keep abreast of inflation. In addition, while both pirate and normas minimas subdividers earn real profits, normas minimas subdividers'real rates of return are much higher. Interestingly -the proportion of subdivisions that fail, e.g. earn negative rates of return, is about the same for pirate and normas minimas subdivisions. In nominal terms roughly one sub- division in 15 fails under medium assumptions,, while one in eight - 61.- Table 19 1/ COMPARISON BETWEEN AVERAGE NOMINAL AND REAL RATES OF RETURN TO SUBDIVIDERS (Figures are monthly; yearly figures in parentheses) Rates of Return Pirate Subdivisions Inside, Normas Minimas Urban Perimeter Subdivisions Nominal Real Nominal Real Payment Default Assumptions Median 1.9(25) 10(13) 2.9(41) 2.9(41) -Mean 2.7 *0(27 6. 0 5.1 (38) 27)(01) (82) Standard Deviation 3.0(42) 2.8(39) 5.7(94) 5.0(80) Number of Subdivisions with Negative IRR 12 25 2 3 Total N 101 102 14 14 Medium Assumptions Median 2.4(33) 1.7(22) 4.9 (78) 4.7(74) Mean 3.6(53) 3.1(44) 7.0(125) 6.0(101) Standard Deviatiot 3.6(53) 4.6 (72) 5.6(92) 5.0(80) Number of Subdivisions with Negative IRR 6 15 1 2 Total N 102 102 14 14 1/ See Table 17 for explanation of calculations and alternative assumptions. Nominal figures are based on current Col$, and real figures are based on constant 1976 Col$. - 62 - fails under default: assumptions. In real terms the failure rate doubles, going as high as one subdivision in four or five under default 1/. assumptions. l/ E. Summary and Conclusions In general, both pirate and normas minimas subdivisions appear to offer high real rates of return to subdividers who opetate efficiently.. Although the rates of return observed here are not "exorbitant," a small number of subdividers do apparently earn very large profits. Average nominal rates of return in the pirate market are below those that have prevailed recently in the middle and upper-middle class development business and about on par with returns in financial markets. The much higher average real rates of return to normas minimas subdivisions indicate that excess demand exists for better-serviced lots in Bogota. The demand has gone unmet because of the scarcity of reasonably-priced land capable of being serviced legally and having residential zoning classification. This and the controversial nature of normas minimas -- involving both technical and political arguments over how low the standards should be set -- have caused the District government to approve relatively few normas minimas subdivision proposals (32 approvals as of early 1979) since the laws were passed in 1973. I1 It is worth noting here that defaulting lot buyers do more damage, on the average, to the rates of return of pirate subdivisions than to normas minimas. -63- The analysis indicates that pirate subdividers do not earn a margin of "excess" profits large enough to be tapped for greater infrastructure investments.. Considering the risks inherent in the business, the poorly-serviced pirate lots seem to be priced competitively. The excess profits earned by Aormas minimas sub- dividers are apparently due to .restrictions on the granting of normas minimas permits, which are necessary for subdividers to connect to utility networks and give legal titles to lot buyers. The permits are, in effect, licenses to market a scarce product and thereby earn high profits.- Without these restrictions, the normas minimas business would become more nearly competitive, lot prices would fall somewhat, and rates of return for normas minimas subdivisions would drop to around the level of those in the pirate market.. 1/ Eleven of the' 14 normas minimas subdivisions in the sample had obtained resoluciones at the time of the survey. The other three had received approval of their final designs. See Chapter VI for a description of the normas minimas approval process. -64 TV. THE DETERMINANTS OF TRACT PRICES A. Analytical Framework In a competitive,land market like Bogota's one would expect to observe certain spatial and behavioral regularitie,s. Market prices are the key to tracing these patterns. This chapter will introduce a technique for analyzing the determinants of land prices - the hedonic price index - and apply it,to the prices of subdivision tracts. In the next chapter the same technique will be used to analyze lot prices. - Traditional theory in urban land economics posits that-the price of an undeveloped parcel of land is determined by various factors, the most important being: topography, nearness to facilities or amenities, distance from the central business district, tbe quality of the neighborhood, and the timing of the purchase. All of these can be thought of as "attributes" of the parcel. Ideally one might construct a model in which the price of land is a function of its attributes: P = f [A1,A2,A3 ... A n where P is the price of the land parcel and A1,A2, etc. are measures of its attributes. This is the structure of a hedonic price index. The technique assumes that each item or product (in this case a tract or lot) has a market price that is associated with a fixed set of quantities of attributes. The standard method is to regress prices on a vector of attributes. The approach is related - 65 - to the idea of product differentiation; that is, thinking in terms of one generic good (e.g. houses) with different component characteristics (e.g. one bathroom or three) rather than in terms of many related generic goods. The hedonic price is an implicit price whose validity depends on the assumption that the market is in short-run equilibrium.1/ The regression coefficient of each explanatory variable is a measure of the contribution of that variable'to the variation in the price of the good, holding constant the values of the other explanatory variables.2/ In this study the hedonic price index is used to reveal the behavior of buyers and sellers. On the demand side, the regression coefficients measure buyers' willingness to pay, on the margin, for quantities of the corresponding attributes. On the supply side,'the coefficients indicate whether attributes make productive or unproductive contributions, on the margin, to the supplier's return. 1/ See Rosen (1974) for an excellent treatment of the theory of hedonic price indexes. An abstract of Rosen's model appears in Appendix III. A basic work on the.subject is Griliches (1971). 2/ The other key assumptions besides short-run equilibrium that underlie 7 the hedonic price index are: (1) buyers and sellers behave in economically rational ways (maximizing utility and profit); (2) consumers have access to information on the variation of attributes and prices of goods; and (3) a sufficiently large number of products with different attributes is available so that the range of choice among combinations is virtually continuous. - 66 - B. Results for Subdivision Tracts - As an introduction to the workings of the hedonic price * index, Table 20 presents'the results of a simple model of the determinants of prices of tracts purchased for pirate and normas minimas development. Coefficients were calculated for seven explanatory variables in a linear and additive specification. The equation was estimated for four groups of subdivisions: the whole sample,.tracts purchased in 1970 or after, pirate subdivisions only, and tracts within Bogota's urban perimeter. This specification explhins no more than a third of the variation in tract prices. This is not surprising, since the independent variables cover only three explanatory concepts - tract size, location, and date of acquisition - out of many possible ones. While certain independent variables, such as the dummies for location in the northwestern or western zones of the city, may pick up the effects of more specific variables, there are a number of explanatory variables completely missing, including the public services avail- able on or near the tract, the tract's natural amenities and topography, the legal status of the tenancy, and the zoning of the tract and surrounding area. The other major limitation of this specification is the linear and additive form. Linearity implies constant returns to scala; that is, price must vary in a constant ratio to the quantity of each explanatory variable throughout the entire range of values of that variable. A more sophisticated model might have included Table 20 67 RELATION BETWEEN TRACT PRICE AND TRACT CHARACTERISTICS IN LOW-INCOM7 SUBDIVISIONS Linear Hedonic Price Index of Form: Price = A + B Varl + B Var 1 1 2 2' (Prices in 1976 Col $) Variable Definitions: NNW Dummy variable equal to 1 if tract is located in north- : northwest zone of the city. (north of Avda. Las Americas.), O if otherwise. WEST Dummy variable equal to 1 if tract is located in western zone of the city (industrial corridor/Bosa), 0 if otherwise. AQYR Year of tract acquisition by subdivider. DIST = Number of -kilometers between DANE barrio in which tract is located and CBD (defined as intersection of Avda. Caracas and Calle 26). DISTSQ Square of DIST. AQYRSQ -Square of AQYR. 2 TRACTSZ Gross size of the tract in 100s of am. 2 Dependent Variable = Tract Price per m Coefficients for Subsamples: All Tract Acquisition Pirate Inside Urban Variable Subdivisions 1970 or Later Subdivisions . Perimeter TRACTSZ -.0088 -.0176 -.0093 -.0066 NNW 29.73 51.91 30.88 30.61 WEST 23.15 28.95 18.06 20.94. AQYR -3.04 3.09 -2.22 -1.89 AQYRSQ .038 .031 .034 DIST -3.93 -5.15 -2.49 -13.79 DISTSQ .048 -.102 .0095 .54 Constant 94.68 -132.79 62.35 83.66 R2 .302 .351 .269 .348 R .266 .302 .228 .308 Std. Error of Est. 34.44 36.29 34.71 34.43 Number of Observations (144) (87) (130) (120) Note: Underlined coefficients are significant at the two-tail .05 level. 1/ Excludes normas minimas. - 68 - non-linear transformations of certain variables (such as tract size). This model attempts to account for the probable non-linear effects of two variables - tract acquisition year and distance to the CBD - by including their squares as additional independent variables. The results show that the zone in which the tract is located has the strongest effect on price. For all tracts, prices in the west average about Col$20 per square meter higher than in the south; prices in the.northwest are around Col$30 higher than in the south-, other variables held constant. The data also indicate that the desirability of the north-northwest and the west seems to have increased over time. Limiting the sample to tracts acquired since 1970 increases the coefficient for the west by roughly Col$10 and the one for the north-northwest by about Col$20. What accounts for the strengthiof these variables? -The zones are proxies for a number of more specific explanatory factors, including access to non-CBD employment'(particularly in the "industrial corridor" to the west between Calle 26 and Avenida Las Americas); the perceived higher social status or quality of the areas, especially of the northwest; and general topography, the south being hilly and erosion- prone, the west being flat but containing certain flood areas, and the northwest being the most desirable.. The other location variables in the model, distance in kilometers to the CBD and the square of this distance, yield coefficients of the expected signs and magnitudes. The high -69- concentration of jobs in Bogota's center makes accessibility valuable.- For all tracts the price per square meter decreases by about Col$4 with each additional kilometer's distance from the center. The distance coefficient rises sharply for tracts inside the urban perimeter only, which corresponds to theoretical expectations that the land price gradient should be steeper with proximity to the CBD. The negative coefficient,for tract size indicates that there are ."quantity discounts" in purchasing tracts. For every 100 square meters added to the size of the tract, sellers tend to lower the unit price by about Col$l per square meter. The discount is higher for subdivisions developed since 1970. -Finally, the year- variable by itself and the combined effect of the year and its square produce the expected results: since 1970 real tract prices have been increasing by about Col$3 per square meter annually. An important point in using hedonic price indexes is that the coefficients of the independent variables express relationships that are relevant to buyers and sellers imultaneously, given competitive short-term equilibrium. The next chapter presents a similar but more elaborate analysis of lot prices with particular attention to the revealed preferences of lot buyers. 1/ According to data of the Instituto Colombiano de Seguro Social, 25 to 30 percent of all employment is concentrated in the CBD, defined as DANE comuna 31, the old downtown south of Calle 25 and east of Avenida Caracas. See K.S. Lee, Spatial Distribution of Establishments and Employment in Bogota, 1978. World Bank, City Study paper, January 1979, p. 5. - 70 - V. THE DETERMINANTS OF LOT PRICES A. Background and Model Specifications Although the data set used in this study contains no inform- ation about the characteristics of lot buyers, it is'possible to draw some inferences about their behavior. One way to do this is to create a hedonic price index for lots sold in pirate and normas minimas subdivisions. The Superintendencia Bancaria survey offers the opportunity to test a number of different specifications and subsamples. The analysis presented here was done first with sub- divisions as observations, using mean values per subdivision for lot price, lot size, and percent down-payment. Afterwards a sample of lots was drawn and a new group of equations was. estimated. This chapter reports results for both the "aggregated" or subdivision sample and for the "disaggregated" or lot sample. The focus is on what the regression coefficients reveal about lot buyers' willingness to pay for particular lot characteristics. Do lot buyers pay higher prices for additional amounts of services, all other things being equal? Or do they prefer to spend their incremental pesos on other attributes, such as greater lot area or better locations in the city? If the analysis shows that.there is a strong willingness-to pay for well-serviced lots, it would suggest that facilitating an increase in the supply of normas mininas lots may be good public policy. If such willingness is lacking, it would indicate that the attractiveness of normas minimas lots is based on other characteristics, perhaps location. -71- Different specifications were possible for several variables in the hedonic price indexes. First, the dependent variable, lot price, could take two forms: total price or price per square meter. Second, the infrastructure variables could be expressed in terms either of quantities (lineal meters of water, sewer, and electric lines, numbers of standpipes, square meters of streets) or of expenditures. The most useful results come from the equations which use quantities of infra- structure,in that they reveal the value of additional units of services installed rather than the effect of additional sums spent on services. Further, the expenditure variables may give biased results to the extent that.subdividers may, due to inefficiency, install similar levels of infrastructure at widely varying costs.1! The linear and additive form of these equations limits the interpretation of the results. While a more complex specification for some terms might be desirable, the intention of this exercise is not to obtain a "best fit" model; rather, it is to demonstrate a sense of the direction and magnitude of the relationships between the various lot attributes and the lot price. 1/ The infrastructure expenditure equations do provide a rough check on the validity of the2data, however. The results they give are reasonable,including R s, which are about the same as in the quantity equations. (See Appendix IV). - 72 - B. Aggregated Sample Results The effects of the independent variables on lot prices, presented in Tables 21 and 22, are summarized below. Quantities..of sewer, water, electricity, and streets: Using infrastructure quantities clarifies the concept of hedonic price index coeff icients as measures of lot buyers' willingness to pay for increments of these attributes. For example, the coefficient of 20.64 for the variable QSEWLT in the first column of Table 21 means that, according to this model and this sample, one additional lineal meter of sewer line per lot raises the average total lot price by Col$2,064, holding all other lot attributes constant. In other words, Col$2,064 is what lot buyers are willing to pay, on the margin, for a unit (linear meter) of this'service. The results indicate that the number of standpipes also has a strong positive effect on lot prices. Installing one standpipe for every 10 lots raises the average lot price by around Col$5,COO. Demand for electric lines, water pipes, and paving, however, seems much lower. It is likely that buyers are willing to pay substantially more for lots with access to standpipes but not with waterpipes because standpipes are a large marginal improve- ment over trucked-in water or none at all. While individual water pipes are better still, buyers are not willing to pay extra for them because, at the time of lot acquisition, the marginal benefits over standpipes probably do not exceed the marginal costs. Moreover, lot buyers have learned to expect that the District government will - 73 - Tables 21 and 22 RELATION BETWEEN LOT PRICE AND SUBDIVISION CHARACTERISTICS Linear Hedonic Price Index of Form: Price = A + BlVarl + B2Var2 (Observations are subdivisions; all prices in constant 1976 Col $) Variable Definitions: ALSZ Average lot size in subdivision (m 2 DIST Kilometers along major streets from midpoint of DANE barrio in which subdivision is located to CBD (defined as intersection of Avda. Caracas with Calle 26). TIME Year of sale of first lot indexed as 1950 = 1, 1951 2, etc. NNW Dummy variable equal to 1 if subdivision located in north-northwest zone of city (north of Avda. Las Americas), 0 if elsewhere. WEST Dummy variable equal to 1 if subdivision located in western zone of city (industrial corridor/Bosa), 0 if elsewhere. OPLOT Square meters of open space per lot in subdivision. STPLOT Square meters of street area per lot-in subdivision. NM = Dummy variable equal to 1 if subdivision is legal normas minimas, 0 otherwise. TRPM2, TRPLOT = Price of original tract of land per m2 and per lot. 2 PROM, PROL = Expenditure on professional services per m of usable (salable) area and per lot. PUBM, PUBL = Expenditure on publicity per m2 of usable (salable) area and per lot. ADMM, ADML = Expenditure on administration per m2 of usable (salable) area and per lot. PERIM = Dummy variable equal to 1 if subdivision is within the Bogota urban perimeter, 0 if not. APDP Average down payment as a percent of average lot price for the subdivision. -74- 2 QSEWM2, QSEWLT Lineal meters of sewer lines installed per m of usable area and per lot. 2 :PILPM2, PILPLT Number of pilas (water taps) installed per m of usable area and per lot. 2 QWATM2, QWATLT Lineal meters of water lines installed per m of usable-area and per lot. QELEM2, QELELT. Lineal meters of electric lines installed per m' ofs usable area and per lot. 2 QSTRM2, QSTRLT = Square meters of street area installed per m of usable area and per lot. - 75 - Table 21 LINEAR HEDONIC PRICE INDEX WITINFRASTRUCTURE QUANTITY Coefficients for Subsamples: (Observations are Subdivisions) Dependent Variable = Average Lot Price per Subdivision in 100s (All prices in 1976 Col $) All Inside Urban Profitable Profitable In- Variable Subdivisions Perimeter Subdivisions side Perimeter ALSZ -183 .662 .134 .776 DIST -7.16 -13.05 -6.71 -15.77 TIME 1.96 3.98 1.10 3.89 NNW 160.10 191.76 157.46 178.81 WEST 117.97 117.34 103.09 100.26 OPLOT .293 .262 .217 .042 STPLOT -2.82 -.278 -.247 -.059 NM 155.61 176.87 168.77 191.73 TRPLOT .0076 .0074 .0084 .0080 PROL .0123 .0092 .0124 .0089 PUBL .0134 .0053 .0085 .0058 ADML .0015 .00094 .0012 .00007 PERIM 25.69 6.56 APDP -1.30 -1.67 -1.28 -1.25 QSEWLT 20.64 18.18 20.11 16.04 PILPLT 618.16 421.82 555.26 490,87 QWATLT -4.98 -10.68 -6.45 -14.61 QELELT 8.13 9.49 9.15 12.22 QSTRLT -.234 .274 -1.77 .378 Constant 172.43 166.68 211.19 167.98 R2 .659 .673 .668 .686 R2 .583 .583 .592 .597 Std. Error of Est. 133.9C 141.65 133.13 140.07 Number of Observations (100) (80) (97) (78) Note: Underlined coefficients are significant at the two-tail .05 level. - 76 - Table 22 LINEAR HEDONIC PRICE INDEX WtfH-REtETRUCTURE QUANTITY Coefficients for Subsamples: (Observations are Subdivisions) 2 Dependent Variable = Average Lot Price per m per Subdivision (All prices in 1976 Col $) All . Inside Urban Profitable Profitable In- Variable Subdivisions Perimeter Subdivisions side Perimeter ALSZ --.245 -.974 -.275 -.891 DIST -6.53 -9.47 -6.60 -11.19 TiMEm 9.15 8.86 8.22 8.71 NNW 96.96 155.01 94.11 146.51 WEST 110.55 112.32 98.54 101.17 - OPLOT -.0027 .327 -.074 .179 STPLOT .134 .022 .177 .142 NM 76.50 95.95 83.96 102.95 TRPM2 .667 .605 .738 .656 PROM .807 .595 .782 .549 PUBM 2.16 .369 2.12 .664 ADMU .632 A .591 .539 .502 PERIM 8.11 -5.72 APDP -.940 -1.39 -.858 -1.09 ,QSEWM2 1683.3 1676.1 1636.5 1521.2 PILPLT 605.20 484.7 660.14 580.61 QWAT.12 -1017.5 -705.5 -1123.6 -1058.1 QELEM2 354.8 136.5 478.6 . 332.5 QSTRM2 6.84 -15.36 11.12 -4.42 Constant 59.29 221.25 95.19 222.05 R2 .684 .677 .687 .678 R2 .614 .589 .614 .586 Std. Error of Est. 108.51' 112.97 107.50 112.29 Number of Observations (100) (80) (97) (78) Note: Underlined coefficients are significant at the two-tail .05 level. - 77 - eventually install water lines at lower costs than pirate subdividers.- At the same time, lot buyers are willing to pay significantly more for sewer lines. Although this service is expensive, the demand exists because (1) sewer pipes are an enormous improvement over the feasible alternat'ive, pit latrines, (2) septic tanks are unworkable on the small . 2/ pirate and nornas minimas lots- ; and (3) there is no alternative supplier as in the case of water conneccions. The low willingness to pay for electric lines exists because legal connections are readily obtained from the public electric company and, if not, illegal taps are easy to install. Willingness to pay for street construction appears to be noneexistent because virtually all subdivisions have graded, unpaved streets, and the cost of paving is very high relative to the added benefits. Tract price: As an explanatory variable, tract price is a proxy for neighborhood quality, including factors like the social class of the residents, access.to shops and services, the attractive- ness of the landscape, and the presence or absence of nuisances such 1/ Since 1976 the Special District of Bogota has offered individual water connections to lot owners in illegal subdivisions on very favorable terms: a 30 percent down-payment on the household's share of the network cost and a connection fee equivalent to 3 percent of the land value, plus installation costs. Both the balance of the network charge and the whole connection fee are financed at 12 percent over ten years. 2/ Even if septic tanks were an alternative, they are no less ex- pensive than sewer pipes. In fact, the cost to the user of septic tanks may be greater because sewer installation is often subsidized. For a comparison of economic costs per unit of various sanitation technologies, see Johannes Linn, Policies for Efficient and Equitable Growth of Cities in Developing Countries, World Bank Staff Working Paper No. 342, July 1979, p. 255. '-78- as noise or pollution. The tract price may also pick up the effect. of how readily the land can be serviced with public utility connections (e.g. the elevation and proximity of sewer and water mains) and the quality of access roads. The coeficcient for tract price is consistently positive and significant across subsamples, though its value is below unity. Location: Dummy variables for subdivision location in the northwest or west turn out to be among the strongest explanatory variables in the lot price analysis. The coefficients are always significant and have high values. Holding all other characteristics constant, a lot in the west costs about Col$10,000 more than a lot in the south, while a lot in the northwest costs Col$16,000 to Col$18,000 more. In other words, controlling for the level of infrastructure, tract price, lot size, and distance to the CBD, lot buyers are willing to pay 10 to 20 thousand pesos for location in these zones alone. As in the tract price analysis of the last .chapter, access to non-CBD employment, social status,. and.topo- graphical desirability make the zone dummy variables strong determinants of land prices. Distance to the center: The coefficient of distance in kilometers from the subdivision to the center of Bogota has the expected negative sign. In most subsamples it is not statistically 1/ Tract price is in this case an explanatory variable together with some of the lame independent variables used in Chapter IV's model. The low R produced by the tract price equation -- around .3 -- permits this maneuver without raising serious collinearity problems. - 79 - significant because it interacts with other accessibility and location variables such as the zone and tract price. The coefficient behaves well, however, having a larger absolute value for subdivisions inside the urban perimeter. Linearly, total lot price declines by about Col$700 to Col$1,500 per kilometer and the lot price per square meter by about Col$6 to Col$ll. Average lot size: Results differ depending on the specification of the dependent variable. Using total lot price, the coefficient is positive across all subsamples;-with lot price per square meter, however, it is consistently negative showing that unit price decreases with lot size. For both specifications, though,the absolute values of the coefficients are much larger for subdivisions inside the urban perimeter, whereman additional square meter of lot area costs an average of about Col$70 to Col$80. Outside the perimeter one finds very large, low- value lots. Year of sale of first lot: The coefficients for this variable has the expected positive sign in all equations and its absolute size is quite stable. The results indicate that average total lot prices have been increasing in real terms by several hundred pesos per year. Average lot prices per square meter have been going up by about 8 to 10 pesos per year, with a higher annual increase of up to 15 pesos in the last several years. 1/ In Appendix Tables A6 and A7 (equations using infrastructure expenditure variables), the slope is substantially lower among recently-developed subdivisions, indicating a flattening of the price-distance gradient over time. - 80 - Average percent down-payment: Using the average percent down-payment as an independent variable measures the implicit finance charge to lot buyers who pay on installment. Pirate subdividers seldom give lot buyers an explicit interest rate as part of the terms of sale. The general practice .is to charge a higher total price (computed as the sum of the installments plus the down-payment)-. The expectation is for the coefficient of this variable to be negative, since it is hypothesized that the smaller the percent down-payment, the higher the total lot price, all else being equal. The sign is in fact consistently negative and the absolute value is stable over the various specifications and subsamples; indeed, the variable appears to be quite strong. For an average lot, the price increase from paying in full (100 percent down) to paying 30 percent down is about Col$9,000. In terms of average lot price per square meter, this increment is around Col$70. Using these results and the average lot price figures in Table 1, the average implicit finance charge on the balance is about 12 percent yearly for all lots in the sample,1/ Overhead expenditures: The subdivider's expenditures for professional services, publicity, and administration were included to test if lots are priced to recover marginal increases in these 1/ As will be shown, however, this figure is an overestimate. The method of calculation is the following: The total average implicit finance charge is computed by multiplying the average percent balance (100 percent minus the average percent down-payment) by the coefficient for the percent down-payment variable from the hedonic price equation. The average total balance is obtained by multiplying the average percent balance by the average lot price. The first figure as a percentage of the second is the total interest rate over the whole repayment period. This is transformed, using the average term length, into a yearly compound rate. -81--. cost categories. The professional services expenditure coefficient oscillates around unity across the various equations, while publicity expenditures have coefficients above unity in most cases Cespecially in the specifications in Appendix IV). One possible explanation for the productivity of publicity expenditures is that they are a proxy for the subdivider's organizational ability or business acumen. Administration expenditures yield weak coefficients and seem t6 be unproductive. Open space and street area: Although the amount of open space and street area were expected to have positive effects on lot prices, the coefficients for these variables are consistently low in value, are never statistically significant, and have unstable signs. Simple correlations of these variables with lot price are also not sigAificant. Normas Minimas: Table 21 -shows that a normas minimas lot costs between Col$15,000 and Col$20,000 more than a pirate lot, controlling for location, services, and other attributes. One possible reason that buyers seem to be willing to pay more for a normas minimas lot is that there is a higher probability of having infrastructure installed; another possibility is that normas minimas buyers are paying for the greater likelihood of obtaining full legal title. Urban perimeter: A dummy variable for the urban perimeter was included in some equations to pick up the effect of laws prohibiting piblic service connections outside the perimeter. While these laws - 82 - are not strictly enforced, their existence probably does delay the provision of infrastructure in the outside areas. The coefficient for this variable is not statistically significant, and its absolute size and sign are unstable. The ambiguity of the results is. due. largely to the interaction of this variable with the distance to CBD, zone, and infrastructure variables. C. Disaggregated Sample The above results, based on subdivisions as observations, tell a useful story about the key features of pirate lots. There is something more to be learned, however, from estimating these equations using a set of observations of individual lots, The use of subdivisions as observations creates potential for "aggregation bias." This can occur when the variables'in regression equations are averages over individual observations that are grouped in some way. A common example is the use of averages across tensus tracts for variables like household income or family size. A key condition for introducing aggregation bieas is that the data be grouped non-randomly according to values of the dependent variable. If this is done, the absolute values of the coefficients of the explanatory variables will be biased upwards.-/ In the preceding analysis the dependent variable fboth forms of it) and two independent variables - lot size and percent down 1/ This was demonstrated by Smith and Campbell, who aggregated the data for a housing expenditure estimation in three ways: randomly, according to one independent variable, and according to the dependent variable. The parameters in the first two estimations were consistent with those based on the original disaggregated data; the third set were higher. This was termed "pure" aggrepation bias. See Barton Smith and J.M. Campbell, Jr. Aggregation Bias and the ,DDemand for Housing, . Paper presented at the Annual Meeting of the Econometric Society, Atlantic City, New Jersey, September 1976. - 83 - payment - are averages across all the lots sold in each subdivision. Since lots within a subdivision are more homogeneous in price than the total universe of lots, using lot price averages as observations implicitly groups the sample into higher-price and lower-price categories; in other words, it stratifies along the dependent variable. In addition, using lots as observations permits a more realistic.approach to the infrastructure variables. In the aggregate equations infrastrubture was counted as being available no matter when it was installed. Since the lot price was an average across all lots, there was no way to differentiate between those lots sold before or after the installation of a given infrastructure type. In the disag- gregated equations, one specification parallels that used in the aggregate sample, with the infrastructure variable interpreted as "installed ever". The other specification contains separate variables for infrastructure installed before and after the sale of the lot. The latter assumes that the presence or absence of infrastructure at the date of sale is reflected in lot price but allows for the possibility that the expectation of receiving infrastructure may raise the lot price. Tables 23 and 24 present the coefficients for hedonic price indexes based on a random sample of lots drawn from the 19,555 in the Superintendencia Bancaria data. The tables present equations containing infrastructure quantity-variables like those set out in Tables 21 and 22. Lot sample coefficihnts for equations with infra- structure expenditures appear in Appendix IV. In general the results - 84 - Table 23 LINEAR HEDONIC PRICE INDEX WITH INFRASTRUCTURE QUANTITY Coefficients for Subsamples: Observations are Lots Dependent Variable - Total Lot Price in 100s (All prices in 1976 Col$) Variable All Lots Profitable Subdivisions Inside Urban Perimeter Pirate Subdivisions Infra- Infra- Infra- Infra- Infra- structure Infra- structure Infra- structure Infra- structure Aggregate structure Before or structure Before or structure Before or structure Before or Model Ever After Sale Ever After Sale Ever After Sale Ever After sale LSZ .183 .113 .103 .115 .103 1.6. 1.61 .104 .095 LYS 1.96 3.53 2.41 .744 -.907 11.04 11.44 .475 .715 rCDP. -1,30 -.545 -.538 -.432 -.407 -.559 -.613 -.689 -.640 NM 155.61 -142.44 -121.18 -126.73 -107.18 -87.68 -99;2 NNW 160.10 158.34 149.76 148.94 141.0? 153.14 127.94 109.44 123.22 WEST 117.97 171.32 156.59 131.45 112.18 153.69 143.31 171.87 161.96 TRPM2 .0076 1.16 1.28 1.33 1.51 1.41 1.40 1.73 1.45 DIST -7.16 -5.71 -4.92 -3.66 -2.32 -2.65 -4.27 -3.81 -4.52 PROL .0123 -.0031 -.0007 -.00036 .0025 .0041 .0082 .0032 .0021 PUBL .0134 .0753 .0723 .0826 .0808 .055 .058 .101 .087 ADML .0015 .0069 .0066 .0065 .0057 .0046 .0025 .0035 .0048 BEFORE QSEWLT 20.64 7.95 5.11 7.42 5.12 7.14 1.49 2.12 2.60 QWATLT -4.98 -4.24 --611. -4.82 -8.36 -23.08 -15.69 4.55 -3.87 QELCLT 8.13 -.135 2.59 2.29 4.45 1.19 1.61 -1.11 6.73 QSTRLT -.234 -.288 -.466 -.309 -.497 -414 -527 -.243 -.463 PILPT 618.16 15'9.72 2328.9 1700.3 2359.23 683.23 1530.7 1183.7 2419.7 AFTER QSEWLT 6.37 4.65 8.45 3.58 QWATLT -5.45 -3.49 -23.08 -3.41 QELCLT -.726 2.27 .0077 -.561 QSTRLT 2.49 2.67 2.32 2.19 PILPLT 357.49 483.52 326.31 -846.31 Constant 172. Y3 -46.91 27.58 114.80 221.64 -782.17 -790.51 155.24 151.57 R2 .659 .609 .642 .615 .654 .665 .697 .504 .539 R .583 .604 .636 .609 .647 .659 .689 .496 .529 Std. Error of Est 133.90 131.95 126.63 131.49 125.08 126.76 120.89 13Q.68 126.09 Number of Observations (100) (1,129) (1,129) (1,033) (1,033) (912) (912) (1,009) (1,009) Note: Underlined coefficients are significant at the two-tail .05 level. - 85 - Table 24 LINEAR HEDONIC PRICE INDEX WITH INFRASTRUCTURZ QUANTITY Coefficients for SubsamDles: Observations are Lots Dependent Variable = Lot Price per M2 (All prices in 1976 Col$) Variable All Lots Profitable Subdivisions Inside Urban Peiimeter Pirate Subdivisions Infra- Infra- Infra- Infra- Infra- structure Infra- structure Infra- structure Infra- structure Aggregate structure Before or structure Before or structure Before or structure Before or Model Ever After Sale Ever After Sale Ever After Sale Ever After Sale LSZ -.245 -.136 -.126 -.142 -.129 -.729 -.635 -.122 -.116 LYS 9.15 8.07 8.63 6.24 6,15 10.46 11.41 7.35 8.33 PCDP -9.40 -.425 -.456 -.334 -.345 -.341 -.402 -.562 -.534 NY 76.80 -119.61., -139.01 -122.58 -149.19 -94.22 -174 NW96.96 131.12 124.90 127.59 119.93 132.10 105.27 93.61 114.44 WEST 110.55 145.56 125.67 125.53 98.56 138.72 120.42 155.46 136.35 TRP142 .667 1.30 1.57 1.39 1.74 1.36 1.39 1.73 1.51 DIST -6.53 -6.63 -r.29 -5.27 -4.40 -1.41 -2.59 -8.14 -8.41 PROM .807 1.37 1.85 1.28 1.72 .718 1.39 1.02 1.41 PUBM 2.16 .340 .599 1.63 2.20 2.83 2.51 5.09 2.77 ADMM .632 1.71 1.46 1.62 1.30 1.26 .844 1.37 1.42 BEFORE QSEW1i12 168.33 271.48 -494.53 288.79 -444.20 198.12 -478.41 92.64 -358.98 QWALM2 -1017.5 -953.03 -823.88 -976.99 -1068.2 -278222 -1252.7 -327.87 -371.62 QELC.M2 358.4 -721.72 -409.35 -506.15 -218.99 -547.04 -500.81 -357.44 -129..63. QSTRX2 6.84 -26.01 -43.62 -23.57 -42.28 -40.64 -59.95 -32.03 -53.00 PILPLT 605.20 2071.6 1927.37 2352.6 2046.6 1111.26 1516.0 -3.14 -516.43 AFTER QSEWM2 670.97 502.58 431.47 668.81 QWALM2 -719.51 -521.91 -3409.9 -917.66 QELCM2 -535.91 -245.23 -481.37 -294.63 QSTR.M2 267.22 313.52 320.28 246.66 PILPLT 1637.35 1990.9 746.65 1477.98 Constant 59.29 -367.1/ -417.83 -259.96 -270.67 -494.82 -557.90 -290.75 -364.95 R2 .684 .709 .752 .679 .737 .693 .742 .718 .753 .614 .705 .748 .674 .731 .688 .736 .714 .748 Std. Error of Est. 108.51 105.57 97.65 106.12 96.37 107.83 99.16 101.68 95.46 Number of Observations (100) (1,129), (1,129) (1,033) (1,033) (912) (912) (1,009) (1,009) Note: Underlined coefficients are sigtifir nt at the two-tail .05 level. -86- of the individual lot sample do not differ dramatically from those of the subdivision (aggregated) sample. R 2s remain around .6 to .7; all but a few coefficeints have the-same signand:similar values. In addition to supporting the results of the subdivision equations, the lot sample coefficients indicate the following: Inkrastructure quantity: Tables 23 and 24 show two consistent differences from the aggregated subdivision sample equations. First, while the coefficient for quantity of sewer lines is still positive and mostly significant, its absolute value is substantially lower. .Instead of costing about Col$2,000 per lot, an additional lineal meter of sewer line is estimated here to cost about Col$700 - 800. Second, the coefficient for standpipes goes up by a factor of two to three. Installing one standpipe for every ten lots adds about Col$15,000 to the price of an average lot, compared to Col$5,000 as estimated with the aggregated sample. The other infrastructure quantity relationships remain more or less un- changed. Regarding infrastructure quantities installed before and after lot sale, the equations give mixed results. The expectation is that the "before" coefficients should be stronger, based on the premise that people are willing to pay more for lots with infra- structure already installed. Standpipes and electric lines, however, are the only services whose coefficients are consistently higher before lot sale than after. Sewer lines installed before lot sale - 87 - give stronger coefficients in Table 24, but in Table 23 the before and after coefficients for sewer are about the same. Water pipes yield small and unstable coefficients whether before or after lot sale. Square meters of streets constructed after lot sale give positive and significant coefficients, while those for streets.built before sale are weak and negative. The reason may be that subdivisions where street construction occurs after the sale of lots are those in which the greatest marginal benefits accrue to lot buyers. In sum, while services installed prior to lot sale do tend to have stronger effects on lot prices, buyers appear to be willing to pay additional suns for promised services. Lot size and percent down-payment-: Although the coefficient for lot size does not change appreciably in the disaggregated equations, the percent down-payment coefficient is lower in absolute value by about half than in the aggregated specifications. The extra cost of paying on installment thus amounts to about Col$4,200 for an average lot rather than the Col$9,000 estimated originally (assuming a 30 percent down-payment). The average implicit finance charge, which was over- estimated using the coefficient from the aggregated equation, is calculated here to be about 5.5 percent per year on the balance for all lots paid by installment. Normas minimas: In the aggregated equation the coefficient for the normas minimas dummy variable is positive but not statistically - 88 significant. In the disaggregated runs, however, the coefficient is consistently negative and significant for the equations with infrastructure quantity. The change of sign is not clearly due to interaction effects or outliers. Distance to the CBD: All coefficients in the lot sample equations have smaller absolute values than in the sub- division equations. The decline in total lot price per kilometer is estimated here (linearly) to ba about Col$300 to Col$500, less than half that in the aggregated. In summary, the disaggregated sample does seem to help correct bias in the estimates of the effects of sewer lines, standpipes, the percent down-payment on the lot, and distance to the CBD. The equations explain between half and two-thirds of the variance in lot prices. Adjusted R s range from around .6 for the equations in which total lot price is the dependent variable to about .7 for those in which lot price per square meter is dependent. These R s are about as large as can be expected and permit a high degree of confidence in the results. -89- VI. POLICY ISSUES: NORMAS MINIMAS AND THE SUPPLY O' SERVICED LOTS A. Improving the Quality of Lots The key policy issue addressed by this ,study is how to encourage the private sector to increase the supply of adequately serviced'lots for the lower and moderdte income iarket. One of .the . first hypotheses mentioned in this paper was that pirate subdividers make "excesive" profits which should somehow be tapped to increase the amount of infrastructure installed without raising lot prices, The.Bogota District.has followed two strategies based on this idea: bringing sanctions against the most exploitive pirate subdividers and encouraging the private sector to enter the legal lots-with-services business by way of the normas minimas regulations.. The first approach has involved freezing the assets of pirate subdividers, a procedure known as intervencion, which is analogous to the U.S. practice of placing a venture in receivership. Law 66 ef 1968 (Article 12) empowers the Housing Division of the Superintendencia Bancaria to seize (though not confiscate) the assets of illegal subdividers and to turn over the administration of their subdivisions to the Instituto de Credito Territorial (ICT), Colombia" s public housing agency. To help implement the intervencion, a lien is normally placed on the land and on the subdivider's bank accounts. The goal is to hold the property until the subdivider upgrades services to a satisfactory level, or, if this fails, to arrange for upgrading by the public sector. - 90 - The problem with this approach is that violations-are too widespread. In effect, any illegal subdivision is eligible to be seized under Law 66. The practical result is that intervencion is used only in extree cases, which one former head of the Housing Division defines as situations where the subdivider and buyers are deadlocked over the continuation of lot payments versus the installation of services. Between 1976 and 1978 only about 30 intervenciones were carried out in Bogota. In addition the ICT has reportedly done little to upgrade the pirate subdivisions it has taken over under Law 66. The agency is paid to lack the administrative capacity to absorb this responsibility.-/ To the question of whether the pirate subdivision market is really so lucrative that excess profits could be diverted to raising the quality of lots without increasing lot prices, the answer is, on the average, no. The evidence presented in Chapter III demonstrates that rates of return to pirate subdividers are not 1excessive'.' Further, the lot price analysis in Chapter V shows that, because buyers are willing to pay for larger quantities of services, better-serviced pirate lots would be more expensive. For these reasons stricter enforcement is not likely to succeed in forcing pirate subdividers to install more services, even if the enforcement 1/ As of the end of 1978, ICT had over 100 seized subdivisions under its authority, about 80 percent of them located in Bogota. - 91 - mechanisms were more effective. For the most part -the pirate glarket successfully and competitively supplies a relatively low-quality good for which there is a high demand at a modest price and for which there is no satisfactory alternative. The qualifier to this conclusion is that a few pirate subdividers do make very large profits. The top 25 percent of pirate subdivisions in the sample earned a mean nominal rate of return of 166 percent annually, compared with .a. mean of 55 percent for all the pirate subdivisions studied.-V Whether these high profits are due to the subdividers' greater efficiency, to favorable timing, or to some other factors, stricter enforcement could in fact improve the quality of services they provide.. The proportion of pirate lots eligible-for such intervention remains small, however: the most profitable 25 percent of pirate subdivisions studied accounts for only 17.3 percent of the total supply of pirate lots in the sample. The other approach, encouraging the private sector to supply higher-quality lots, has produced mixed results. According to the analysis in Chapter III the normas minimas business is lucrative enough to interest legitimate private developers. The very high profit rates -- on average two to three times higher than pirate subdivisions in real terms -- are probably due to institutional restrictions on the 1/ Figures based on "medium" rates of return. See Chapter III. - 92 - supply of normas minimas lots. B. Restrictions on Normas Minimas Between 1973 and 1977 the District Planning Department (DAPD) granted final permits (resoluciones) to 28 normas minimas sub- divisions averaging 10 %hectares in gross area each, as shown in Table 2 25. Assuming 60 percent salable area and an average lot size of 80 m the supply works out to about 4,200 legal lots annually over the five- year period. Only four more normas minimas subdivisions were given final approval between the beginning of 1978 and April 1979,. The approval process consists of two stages, a preliminary review (consulta previa) which results in either rejection or clearance to begin drafting formal plans, and a final design (proyecto general) which results in either final approval (resolucion) or denial of the proposal. Preliminary and final approval are awarded by the Public Services Committee (Comite de Servicios Publicos), a group composed of representatives from DAPD and the public utility companies plus members of the City Council. If a resolucion is granted, the normas minimas subdivider must still obtain permission from the Housing Division of the Superintendencia Bancaria (permiso de venta) before he can begin to sell lots. DAPD records show that, of a total of 253 normas minimas applications received over the 1973-77 period, 28 were given gesolueiones (11 percent), 97 were rejected (38 percent), and 128 were given preliminary clearance only (51 percent). The three major reasons for rejection related to location: (1) the tract was outside the urban perimter; -93 Table..25.. NORY-AS MINIMAS SUBDIVISION PROPOSALS SUBMITTED AD GIVEN' FINAL APPROVAL 1973 -1977 tearNumber of Number of Final Year Prol5sd.ls Permits (eouins 1973 12 1 74 62 9 7,5 63- 6 76 66 5 77 '*54 7 Total 25 7= 28 Proposals receiving resolucidties in a given year were not necessarily submitted in the same.,year. 2There is a discrepancy of four proposals in the DAPD report between this figure and the sum of proposals rejected and accepted with and without resolucion (253). Source: Departamento Administrativo de Planeacion Distrital, Aplicacion de las Normas Myinimas de Urbanizacion y de Servicios, Unidad de Mejoramiento y Coordinacion de Barrios, Bogota, February 1978. -94- (:2) the tract was in a restricted development zone (e.g. a flood- area); and (3) the tract was in an area zoned for industrial use. It is not clear, though, why 128 proposals, about half, received only preliminary approval. It is very probable that some of these sub- divisions -- no one knows how many -- have gone on the market illegally. Interviews with public officials and private developers suggest the following reasons why the supply of legal normas minimas lots has been restricted: Lack of land: According to the regulations, normas minimas subdivisions may be built only on.land with the proper residential zoning classification ("high density"). DAPD reports that over the last two years the supply of this type of land has practically dried up in Bogota. Although the city's Planning Board (Junta de Planeacion) may rezone land, so far it has not increased the supply of high density areas. Apparently a significant number of normas minimas proposals have been submitted for tracts on residential land having the wrong zoning class; these have tended not to be rejected outright, but to remain in limbo. Delay in official approval: The normas minimas approval process takes at least one year and often two, which is viewed as excessive by most developers, especially when coupled with the uncertainty over whether a resolucion will be granted. This gives the normas minimas system the character of a lottery; subdividers lucky enough to win approval are rewared with permission to earn high profits. The odds of winning, are law- however, and it appears that other types of investments:,, mostly -95- speculative but including the pirate subdivision business, currently provide the opportunity for faster turnover of funds by investors. Technical and financial objections: These have been raised against the normas minimas concept itself. The Bogota Water and Sewerage Company (EAAB), for example, reportedly has not been comfortable with the lower-than-standard engineering specifications for water supply and drainage under the normas minimas regulations. The company feared that the infrastructure would be of inferior quality and create excessive maintenance expenditures in the futuze. Under current practice, water company clients pay installation and connection fees, but all subsequent maintenance costs are borne by EAAB. Social opposition: Some argue that the normas minimas concept implies giving official approval to substandard development ("legitimizing slums"), thus undermining the normal subdivision and building codes. It has been difficult to promote the idea that lowering standards through normas minimas produces significant benefits, namely lower unit costs and hence wider affordability for lower-income families, in addition to more efficient use of land and public resources as a result of planned development. The normas minimas label has not eliminated the fears that tend to segregate income groups in Bogota. Opposition from neighboring land owners and residents may have defeated several normas minimas subdivision proposals, based on the belief that the presence of a normas minimas subdivision would lower property values and reduce neighborhood quality. - 96 - C. Conclusions If normas minimas subdividers make high profits.selling better-serviced lots, why don't more pirate subdiviers mimic them by developing higher-quality illegal subdivisions? One possible explanation is that pirate subdividers cannot obtain connections to -the public service networks. Both the Bogota Water and Sewerage Company (EEAB) and the Bogota Electric Power Company (EEEB) require subdividers to have official approval before they allow connections to be made. This is'circumvented only in cases where established illegal subdivisions are "regularized." Another reason pirate subdividers may be unable to supply unauthorized duplicates of the normas minimas product is that they cannot offer legal tenure. While the lack of legal documents does not seem to impede-the sale and resale*of pirate lots, it may be a problem for buyers willing to pay higher prices for normas minimas lots. This study finds that there is a strong demand for residential lots at a variety of quality and price levels. While the biggest supply gap is in-higher quality lots, there remains a large segment of families who cannot afford lots with adequate service levels. These families will continue to buy poorly serviced pirate lots as long as the supply can be maintained at prices well below those in the normas minimas market. While a bigger supply of normas minimas lots may replace a portion of the pirate subdivision market, it will not eliminate it entirely. The extent of the inroads normas minimas can make depends, apart from institutional problems, on how low the standards are set. - 97 - In summary, the principal findings of this study are that: (1) There is a large demand for modestly-priced low-quality residential lots in Bogota which pirate subdividers fill successfully; (2) There is also a large demand for modestly-priced, legal, adequately-serviced lots which is largely unsatisfied; (3) The normas minimas regulations, designed to permit the supply of high-quality lots, are being applied very restrictively, resulting in excess profits to the few normas minimas subdividers, able to obtain permits; (4) The benefits of encouraging normas minimas development are not well understopd or accepted. These benefits are being lost to the extent that potential normas minimas lot buyers are currently forced to settle for inadequately planned and serviced pirate lots. -98- Appendix I List of Variables in Superintendencia Bancaria Pirate Subdivision Survey, 1977 (coded from questionnaires) A. Subdivision Variables 1. Name of.subdivision 2. Name'of DANE (census) barrio of subdivision 3. ID number of DANE barrio (1973 codes) 4. Date.of initiation of physical works 5. Number of lots 6. Date of DAPD approval, if any 7. Date of Superintendencia Bancaria approval, if any 8. Size of tract (m 2 9. Date of acquisition of tract 10. -Date of tract acquisition document 11. Price of tract 12. Value of installments (up to five) for tract 13. Dates of installments 14. Interest rate on tract installments 15. Number of bus routes within ten minutes walk 16. Existence of neighboring subdivision with water net- work 17. Subdivision inside or outside urban perimeter? 18. Subdivision above or below water company elevation limit? 19. If above, source of water supply 20. For the following types of infrastructure: sewer, water pipes, stand pipes, electricity, telephone, streets, sidewalks, and curbs (a) quantity installed (b) cost of installation (c) source of financing (d) date of installatioa -99- 21. Other infrastructure installed: type, cost, fin- ancing, date 2 22. Usable (salable) area of subdivision (m ) 23. Communal area (m2 24. Green zones (m) 2 25. Street area (m ) 26. Expenditure on professional services 27. Expenditure on publicity 28. Expenditure on administration 29. Other expenditures: type, value 30. Date of sale of first lot 31. Months taken to sell 50 percent of lots 32.. Months taken to sell 90 percent of lots 33. Type of publicity used, if any, to sell lots 34. Commissionists' fees, if any, paid per lot 35. Number of buyers to date 36. Number of buyers behind in payments; number behind less and more than 6 months 37. Value of outstanding payments 38. Value of outstanding late payments 39. Number of legal actions initiated by btyers against subdivider 40. Number of legal actions initiated by subdivider against buyers 41. Number of lots sold to date. B. Lot variables (for lots sold) 1. Date of sale 2. Area (m2 ) 3. Type of sale and property documents 4. Price 5. Value of down payment 6. Term of balance (months) 7. Value of monthly installments 8. Value of installments paid to date 9. Does original buyer live on lot? - 100 - Appendix It Calculation of Internal Rates of Return for Subdivisions I. Assumptions in Alternative IRR Calculations A. Basic Assumptions 1) All payments for lots are made on time, and all lots are fully paid for. 2) If no infrastructure,installation date is given (missing ob- servation) it is assumed to occur on the date of sale of the first .lot. 3) Overhead costs are assumed to occur at the initiation date. 4) If the sale date of the lot is missing, it is assumed to be the date of sale of the first lot. 5) All cost and price figures are taken as is. 6) All lots in the subdivision are sold. 7) Lots not having individual sale data (unsold at time of survey) are assumed sold at the average size and price (in 1977 pesos) and on the last date of sale among the lots with data. B. Payment Default Assumptions 1) Items 2 through 7 in A above. 2) A proportion of defaulting lot buyers is given per subdivision in the data; this incidence of default is assumed to occur random- ly over the period of lot sales; subdividers are assumed to receive only down payments for defaulting lots. C. Favorable Assumptions 1) All items in A above except 5. 2) Item 5 is modified in that the three overhead cost types -- pro- fessional services, publicity, and administration -- are limited in value to one standard deviation above the means for overhead in normas minimas subdivisions. II. Method for Calculating Internal Rates of Return A. Using the expenditure and revenue figures and dates, a net cash flow in monthly intervals is assembled for each subdivision. - 101 - B. The cash flow is summed to check whether the present value is pos- itive at a zero discount rate. If it is negative, the value is stored and work begins on the next subdivision. If it is positive, the process ..continues. C. A series of simple iterative linear algebra calculations are performed to estimate the internal rate of return. The routine is based on the inverse relationship between the discount rate and the presenp value in any "well-behaved" cash flow. This is shown in the figure-helow. The steps are: 1) Choose an arbitrary discount rate (R1) as a starting poinit- 2) Compute the present value of the cash flow at this discount rate (PVJ) 3) Calculate the slope of line PVoA given PVo, PV1, and R1. 4) Compute a new discount rate (R2) given PV, and the slope of PVoA. 5) Calculate the new present value (PV2) given R2* PVO b1 PV1 A b2 Present value PV2 B V' C present value function (PV as 3 discount rate varies) R R R R4 IRR Discount rate - 102 - 6) Determine the slope of line AB and use it to compute the new intercept bl. 7) Given the slope of AB and bl, obtain R3. 8) Continue until PVn approaches zero within a given margin (in this case + 500 pesos) D. Since the internal .rate of return is the discount rate at which the present value equals zero, the Rn that causes the PVn to fall within the desired interval around zero is stored. E. A "badly-behaved" cash flow -- one with no unique IRR -- is indicated by the appearance of a negative PVn or a. positive slope in the course of an iteration. If this happens the calculations are stopped and a .new subdivision is started. 103 - Appendix III Hedonic Prices in Competitive Markets A hedonic price index is an equation in which the price of a good or product is regressed on a vector of quantities of its attributes. As Rosen (1974) describes it, Each product has a quoted market price and is also associated with a fixed value of the vector z [of quantities of attributes], so that.products markets implicitly reveal a function p(z)= p(zl,Z, ...z.) relating prices and characteristics. This function is the buyer's (and seller's) equivalent of a hedonic price regression, obtained from shopping around... (p. 37) The hedonic price index describes behavior in.competitive markets in short- run equilibrium. The- k'y assumptions-underlying-the model are: - buyers and sellers behave in economically rational ways, maximizing utility and profit, respectively. - consumers have good information on the available choices among variants of the generic good with different characteristics and different prices. - a sufficiently large number of products with dif- ferent attributes is available so that the range of choice among combinations is virtually continuous. In Figures A-1 to A-5 the relationships between prices and quantities are shown graphically (after Rosen). The diagrams demonstrate how the hedonic price model describes "a space in which both buyers and seller locate" and how the price function is reached through the equalization of quantities supplied and demanded of each attribute. Figure A-1 shows a traditional budget line with the quantity of attribute z plotted against the quantity of all other goods (or income). The budget line is the opportunity set available to consumers as the quantity of z varies. Figure A-2 depicts the same space but with the Y axis flipped over to measure total expenditure on attribute z1. Buyer B purchases a pro- - 104 - HEDONIC PRICES AND COMPETITIVE EQUILIBRIUM Y (all other A goods or income) bid function (indifference curve) budget line p(z1,z2*, . . . zn Fig. A-1 z, (quantity of attribute z1) Flip over Y axis . . .plz . . ..2. . . .DEMAN SIDE B purchases a product (expenditure containing more of on attri- B attribute z than A does. 11 bute z1 demand price line ----amount of income foregone by A in purchasing quantity q1 of al jil. A-2 1 2 z1 I I I 1 D SUPPLY SIDE p z 1 D has a comparative advan- iso-profit curve tage over C in producing (amount of i larger quantities of z ) firm's I C resources spent in producing supply-price line z1) Fig. A-3 z 1 - 105 - HEDONIC PRICES AND COMPETITIVE EQUILIBRIUM - (cont d) p z D MARKET-CLEARING JOINT ENVELOPE implicit price function-- B maximum price obtainable in the market for different C .models with varying amounts of attribute z1. )A Fig. A-4 - .1. C? derivative of iso-profit curve p1 (supply curve) trace of short-run equilibrium points (demand=supply) derivative of indifference curve (compensated demand curve) A' Fig. A-5 z -106- duct containing more of attribute z than buyer A. He may do this because he has a different utility function, a different income, or both. In Figure A-3 the same price line is shown, with iso-profit curves for two producers, C,and D. The price line is a supply-price, and producer D has a comparative advantage, because of his technology and cost conditions, :in producing larger quantities of z than producer C. Figure A-4 shows the buyers' and sellers' indifference curves together and all tangent,to the price line. The latter is the implicit .price function -- the hedonic price -- in equilibrium. The.price line is essentially a joint envelope.created by one group of bid function and another,of offer functions. In Figure A-5 the relationships are depicted in price-quantity rather than expenditure - quantity space. The bid functions are transformed into compensated demand curves (from which the income effect has been removed) and the iso-profit curves into short-ru,n supply or marginal cost curves. - 107 - APPENDIX IV Tables A6 through All HEDONIC PRICE INDEXES WITH INFRASTRUCTURE EXPENDITURES Coefficients for Subsamples - 108 - Table A6 LINEAR HEDONIC PRICE INDEX WITH INFRASTRUCTURE EXPENDITURE Coefficients for Subsamnles: (Observations are Subdivisions) Dependent Variable = Average Lot Price per Subdivision in 100s (All prices in 1976 Col $) All Tract Acquisition Inside Urban Profitable Profitable In- Variable Subdivisions 1970 or After Perimeter Subdivisions side Perimeter ALSZ .156 .056 .834 .114 .849 DIST -7.76 -2.97 -13.18 -6.31 -15.96 ' TIME 4.73 5.5/_ 9.03 6.18 9.45 NNW 212.25 171.81 239.03 198.79 236.16 WEST 100.78 99.58 90.24 111.21 97.59 OPLOT -.173 .148 -.089 -.287 -.140 STPLOT -.053 .236 -.016 .096 .076 NM 43.29 45.95 53.36 63.82 62.02 TRPM2 .613 1.52 .530 263 .514 PROL .0092 .013 .0059 .010 .0054 PUBL .0542 .041 .059 .037 .055 ADML .0024 .0040 .0024 .0023 .0030 PERIM 52.68 63.07 APDP -1.84 -3.18 -2.05 -1.62 -1.55 INPLOT .0091 .0066 .0085 .0099 .0080 BUYPLOT .0041 .0039 .0039 .0036 Constant 144.23 117.74 85.78 94.92 92.04 R 2 .561 .623 .586 .539 .578 R2 .508 .550 .529 .477 .523 Std. Error of Est. 141.30 140.18 143.40 136.37 145.14 Number of Observations (145) (87) (118) (127) (113) Note: Underlined coefficients are significant at the two-tail .05 level. - 109 - Table A7 LINEAR HEDONIC PRICE INDEX WITi-TMASTRUCTURE EXPENDITURE Coefficients for Subsamples: (Obsezvations are Subdivisions) Deendent Variable = Average Lot Price per m per Subdivision (All prices in 1976 Col $) All Tract Acquisition Inside Urban Profitable Profitable In- Variable Subdivisions 1970 or After Perimeter Subdivisions side Perimeter ALSZ -.239 -.194 -.859 -.321 -.954 DIST -8.80 -5.71 .-11.39 -8.21 -12.76 TIME 10.81 16.91 10.29 11.86 11.33 NNW 135.28 122.02 188.14 126.66 179.25 'WEST 88.51 94.45 96.43 89.09 93.69 OPLOT -.159 -.124 .092 -.207 .011 STPLOT .147 .197 .083 .209 .195 NM 8i.99 81.10 51.42 102.77 72.91 TRPM2 .380 .856 .297 .234 .305 PROM .684 2.19 .453 .745 .385 PUBM 3.99 1.38 3.68 2.71 3.12 ADMM .433 .506 -.499 . .434 .567 PERIM 27.96 27.32 APDP -1.60 -2.73 -1.68 -1t28 -1.12 INPM2 .813 .720 .768 .847 .542 BUYPM2 .164 .305 .151 .137 Constant 72.48 -86.49 218.79 50.14 214.13 R .675 .726 .664 .670 .672 R2 .636 .672 .618 .625 .629 Std. Error of Est. 109.80 109.39 112.02 108.41 110.69 Number of Observations (145) (87) (118) (127) (113) Note: Underlined coefficients are significant at the two-tail .05 level. - 110 - Table A8 LINEAR HEDONIC PRICE INDEX WITHZNEAST<UCTURE EXPENDITURE Coefficients for Subsamples: (Observations are Subdivisions) Coefficients for Individual Infrastructure Expenditures: Sewer, Water, Electricity, Streets (These variables replace INPLOT and INPM2 in preceding tables; all other variables remain the same, all prices in 1976 Col $) All Tract Acquisition Inside Urban Profitable Profitable In- Variable Subdivisions 1970 or After Perimeter Subdivisions side Perimeter Specifica- SEWLT .0256 .0257 .0160 .0373 .017 AinAWATLT .0018 -.0045 -.0022 .0002 -.0005 Dependent Variable: ELELT .0209 .0377 .0163 .0152 .012 Average lot STRLT -.0012 -.0280 .0051 .0028 .0059 pr 'ice in---- loos of Col. Constant 17.17 99.13 84.14 108.04 85.97 R2 .569 .665 .582 .561 .576 R R .506 .583 .511 .488 .505 Std. Error of Est. 141.61 135..05 146.33 134.89 147.82 Specifica- SEWM2 1.87 2.45 1.72 2.91 2.44 tion B WATM2 -.579 -1.87 -1.03 -.415 -.409 Dependent Variable: ELEM2 1.74 3.06 1.17 .755 .086 Average lot STRM2 .444 -1.12 .373 .854 .435 p5ice per m in Col $ Constant 88.03 -115.97 219.34 42.58 204.74 R2 .671 .737 .659 .675 .680 R2 .623 .673 .601 .621 .627 Std. Error 109.34 114.46 109.11 111.02 of Est. 111.8 Number of - Observations (145) (87) (118) (127) (113) Note: Underlined coefficients are significant at the two-tail .05 level. Table A9 LINEAR lEDONIC PRICE INDEX WITH INFRASTRUCTURE EXPENDITURE Coefficients for Subsamples: Observations are Lots Dependent Variable Total Lot Price in 100s (All prices in 1976 Col$) Variable All Lots Profitable Subdivisions lnside Urban Perimeter Pirate Subdivisions Infra- Infra- Infra- Infra- Infra- structure Infra- structure Infra- structure Infra- structure Aggregate structure Before or structure Before or structure Before or structure Before or Model Ever After Sale Ever After Sale Ever After Sale Ever After Sale LSZ .156 .111 .111 .108 .112 .931 .95 .094 .089 LYS 4.73 4.51 4.51 4.46 4.34 10.03 9.94 3.95 4.58 PCDP -1.84 -.847 -.847 -.622 -.618 -1.09 "1.09 -.908 -.906 NM 43.29 17.09 17.09 25.89 20.12 27.73 27.46 NNW 212.25 153.32 153.32 147.56 150.35 . 138.54 138.52 153.26 146.32 WEST 100.78 133.88 133.88 121.04 124.17 121.69 121.89. 151.92 158.81 TRPM2 .613 .015, .0152 .039 .032 % .042 .042 -.0014 -.014 DIST -7.76 -7.58 -7.58 -7.79 -7.96 -3.51 -3.54 -7.43 -82 PROL .0092 .0031 .0031 .0009 .0007 .0072 .0074 -.0009 -.0034 PLB .0542 .081 .081 .079 .082 . .067 .068 .113 .115 ADML .0024 .0035 .0035 .0032 .0031 .0009 .0009 .0033 .0035 INPLOT(before) .0091 .0071 .007i .0091 .0081 .0071 .0068 .0068 .0096 INPLOT(after) .0071. .0110 .0074 .0035 Constant 144.23 -56.39 -56.34 -60.76 -52.03 -571.92 -566.25 -15.26 -49.18 R2 .561 .519 .519 .536 .537 .543 .543 .441 .446 _7.508 .516 .516 .533 .534 .540 .540 .436 .441 Std. Error of Est. 141.30 119.01 119.03 117.46 117.37 115.08 115.10 127.59 127.07 Number of Observations (145Y (2,399) (2,399) (2,226) (2,226) (2,172) (2,172) (1,477) (1,477) Note: Underlined coefficients are significant at the two-tail .05 level. - 112 - . Table A10 LINEAR H.-DONIC PRICE INDEM iTH INFRASTRUCTURE EXPENDITURE I Coldfficients for Subsamples: Observations are Lots Dependent Variable - Lot Price per M2 (All prices in 1976 Col$) Variable All Lots Profitable Subdivisions Inside Urban Perimeter Pirate Subdivisions Infra- Infra- Infra- ''Infra- Infra- structure Infra- structure Infra- structure Infra- structure Aggregate structure Before or structure Before or structure Before or structure Before or Model Ever After Sale Ever After Sale Ever After Sale Ever After Sale LSZ -.217 -.212 -.245 e.233 -.703 -.698 . -.159 -.159 LYS 10.81 14.07 13.50 14.16 13.71 13.54 13.42 12.58 12.48 PCD? -1.60 -.846 -.846 -.614 -.604 -.930 -.932 -.745 -.745 N 81.99 80.71 78.23 80.42 63.04 46.81 46.61 NNW 135.28 143.42 144.31 139.32 145.29 144o56 144.40 140.86 141.78 WEST 88.51 14327 143.00 135.63 145.69 148.04 148.36 175.85 174.77 TREM2 .380 .144 .141 .155 .135 .089 .089 .122 .125 DIST -8.80 -9.68 -9.63 -9.65 -10.18 -2.61 -2.67 -12.04 -11.92 PROM .684 1.02 1.08 .729 .693 .488 .507 .114 .159 PUBM 3.99 .300 .761 .303 .968 2.43 2.52 8.43 8.38 ADmo .433 .216 .211 .133 .087 .122 .121 .457 - 456 INPM2(before) . .393 84 .679 .389 .349 .302 .255 .194 INPM2(after) .5651 .386 .298 Constant 72.48 -.693.16 -.655.09 -.706.46 -672.89 -,651.08 -.642.59 -.583.94 -578.48 R2 .675 .672 .673 .680 .686 .650 .650 .649 .649 9 .636 .670 .672 .679 .684 .648 .648 .646 .646 Std. Error of Ent. 109.80 107.91 107.70 106.23 105.39 106.61 106.63 106.57 106.59 Number of Observations (145) (2,399) (2,399) (2,226) (2,226) (2,172) (2,172) (1,477) (1,477) Note: Underlined coefficients are significant at the two-tail .05 level. Table All LINEAR IEDONIC PRICE INDEK HITil INFRASTRUCTURE EXPENDITURE Coefficients for Subeamples: Observations are Lots Coefficients for Individual Infrastructure Types: Sewer, Water, Electricity, Streets (These variables replace INPLOT and INPM2 in preceding tables; all other variables remain the same; all prices in 1976 Col$) Dependent Variable - Total Lot Price in 100s Variable All Lots Profitable Subdivisions Inside Urban Perimeter Pirate Subdivisions Infra- Infrastructure Before Infra- Infrastructure Before infra- Infrastructure Before Infra- infrastructure Before Aggregate, structure or After structure or After -structure or After structure or After Model Ever Sale Ever Sale Ever Sale Ever Sale SWELT .0256 .0123 .0153 .008q .019 .022 .024 .0057 .0077 .0088 .0112 .0126 .0101 WATLT .0018 .0064 .0066 -.0009- .0038 .0003 -.0026 .019 021 .0013 .0095 .0106 .0035 ELECLT .0209 .0035 .0106 .0038 .0090 .0106 .0038 0 .0041 -.0093 .0058 -.0102 STRLT -.0012 106. .0031 .0113 Aph .0038 .013 0055 .0027 .0104 .0092 .0103 Constant 17.17 -68.07 -40.79 -74.30 -64.36 -541.01 -529.07 -85.55 -81.09 R .569 .519 .522 .530 .541 .544 .547 .445 .449 R .506 .517 .519 .535 .537 .541 .543 .440 .442 Std. Error of Est. 141.61 118.97 118.72 117.25 116.95 115.02 114.76 127.17 126.95 Nu=ber of Observations (145) (2,399) (2,399) (2,226) (2,226) (2,172) (2,172) (1,477) (1,477) Specification J: Dependent Variable Lot Price per 112 SEWH2 1.87 -1.24 -1.49 -.823 -.541 -.995 1.07 -.536 -.735 .0042 -.178 -.244 .653 VATY2 -.579 .532 .141 .357 .322 -.489 .270 .421 .805 -.866 .883 1.18 -.348 ELE:2 1.74 1.19 IJ5 1.03 1.76 1.41 1.73 .505 1.04 .485 -1.14 -1.19 -1.34 STM2 .444 .725 .637 1.25 .831 798 1.56 .649 .505 1.05 1.17 .745 1.67 Constant 88.03 -653.78 -630.82 -655.7 -668.81 -661.87 -654.35 -624.35 -608.51 R2 .671 .678 .680 .684 .691 .652 .654 .652 .655 2 .623 .676 .677 .682 .688 .650 .651 .649 .651 Std. Error of Zat. 111.8 107.00 106.75 105.62 104.65 106.40 106.25 106.20 105.89 Nu=ber of Observations (145) (2,399) (2,399) (2,226) (2,226) (2,172) (2,172) (1,477) (1,477) Note: Underlined coefficients are significant at the two-tail .05 level. - 114 - REFERENCES Bender, Stephen A. "Low-Income Housing Development in Bogota," Rice University Studies, Issues in Income Distribution, Houston, Texas, Vol. 61, No. 4, Fall 1975. Blaesser, Brian. The Private Market and the Process of Lower Income Urbanization in Colombia: The Pirate Housing Submarket of Medellin. Master's Thesis, Massachusetts Institute of Technology, Department of Urban Studies and Planning, June 1979. Borrero, Oscar and Sonia Sanchez. Merca(-eo de Tierras en Barrios Clandestinos de Bogota, Bogota: Departamento Administrativo de Planeacion Distrital, April 1973. Departamento Administrativo de Planeacion Distrital. Aplicacion de las Normas Minimas de Urbanizacion y de Servicios. Bogota: Unidad de Mejoramiento y Coordinacion de Barrios, February 1978. Doebele, William A. The Private Market and Low-Income Urbanization in Developing Countries: The "Pirate" Subdivisions of Bogota. Cambridge, Mass.: Harvard University Department pf City and Regional Planning, Discussion Paper No. D75-11, October 1975. Fuentes, Alfredo L. and Rodrigo Losada. "Implicaciones Socio-Economicas de la Ilegalidad en la Tenencia de la Tierra Urbana en Colombia," Coyuntura Economica, Vol. VIII, No. 1, April 1978. Gauhan, Timothy 0. Some Economic and Political Characteristics of Low-Income Housing Market in Bogota, Colombia and Their Implications for Public Policy Alternatives. Rice University, Program of Development Studies, Paper No. 64, Spring 1975. Griliches, Zvi, ed. Price Indexes and Quality Change. Cambridge, Massachusetts: Harvard University Press, 1971. Lopez Trujillo, Anibal and Fernando Jimenez Mantilla. El Estado de la Actividad de Vivienda en Bogota; mimeo, source and date unknown, 16 pages. Losada Lora, Rodrigo and Hernando Gomez Buendia. La Tierra en el Mercado Pirata de Bogota. Bogota: FEDESARROLLO, June 1976. McCallum, David. "Land Values in Bogota, Colombia," Land Economics, Vol. 82, No. 1, August 1974. Nelson, Joan M. Public Housing, Illegal Settlements, and the Growth of Colombia's Cities. Washington, D.C.: U.S. Agency for International Development, December 1973. - 115 - Paredes, Luis Ricardo and Luis Guillermo Martinez. Alternativa para la Solucion del Problema de la Vivienda para Grupos de Bajos Ingresos: El Sector Privado Tradicional y las Normas Minimas. Document prepared for'INTERHABITAT 1977, VI'Inter-American Conference on Housing, Medellin, Colombia, November 1977. Rosen, Sherwin. "Hedonic Prices and Implicit Markets: Product Differentia- tion in Pure Competition," Journal of*Political Economy, Vol. 82, No. 1 January 1974. Smith, Barton, and J.M. Campbell, Jr. Aggregation Bias and the Demand for Housing. Paper presented at the annual meeting of the Econometric Society, Atlantic City, New Jersey, September 1976. Valenzuela, Jaime and George Vernez. La Actividad Constructora Popular: Analisis General y Elementos para una Politica de Apoyo. Documento II, Departamento Nacional de Planeacion, Bogota: April 1972. Vernez, George. Bogota's Pirate Settlements: An Opportunity for Metropolitan Development. Berkeley, California: University of California, Doctoral Dissertation, April 1973 (a). Vernez, George. -Pirate Settlements, Housing Construction by Incremental Development, and Low-Income Housing Policies in Bogota, Colombia. New York: The New York City-Rand Institute, May 1973 (b). Villamizar, Rodrigo. Land Prices in Bogota Between 1955 and 1978: A Descriptive Analysis. Paper prepared for World Bank - CCRP "City Study" Workshop, Bogota, Colombia, July.1979.