DISCUSSION PAPER Report No. UDD-56 INCOME TRANSFERS AND URBAN PROJECTS: RESEARCH FINDINGS AND POLICY ISSUES FROM CARTAGENA, COLOMBIA by Daniel Kaufrnann and Michael Bamberger January 1985 Water Supply and Urban Development Operations Policy Staff The World Bank The World Bank does not accept responsibility for the views expressed herein vwhich are those of the author(s) and should not be attributed to the World Banik or to its affiliated oraanizations. The findings, interpretations, and coniclusions are the results of research supported by the Bank. The designations employed, the presenitation of material, and any maps used in this document are solely for the convenience of the reader and do not imply the expresSion of any opinion whatsoever on the part of the World Bank or its affiliates concerning the legal status of any country, territory, city, area, or of its authorities, or concerning the delimitation of its boundaries, or national affiliation. Daniel Kaufmann, an economist, is with the East Africa Industrial Development and Finance Division of the World Bank. Michael Bamberger, a sociologist is a regional coordinator with the Economic Development Institute of the World Bank. At the time when this research project was conducted, both authors worked with the Water Supply and Urban Development Department of the World Bank. Working Paper No. 2 Research Project No. 672-57 Income Formation and Expenditures of Poor Urban Households of Cartagena, Colombia. The authors would like to thank Claudia Franco who worked as a research assistant on the project and participated in the computer analysis. ABSTRACT This paper is one of a series of working papers which report the principal findings of the study "Income Formation and Expenditure Patterns Among Poor Urban Households in Cartagena, Colombia" (RPO 672-57). The present report describes the determinants of the magnitude, composition and uses of interhousehold transfers, and assesses the way in which transfers affect, and in turn are affected by, the squatter upgrading project being financed by the World Bank in Cartagena. An analysis is also presented of the determinants of a household's decision to invest in upgrading and how this is affected by their perceptions of the investment potential of the changes being produced in the area by the project infrastructure. The analysis shows that transfers of money and of free labor appear to be affected differently by the presence of the project. The study was conducted in the Zona Sur Oriental of Cartagena, Colombia, in a low-income, largely squatter area in which the World Bank was financing an integrated upgrading project. The project included land reclamation, upgrading of basic services and credit and technical assistance for housing and community improvement projects. The project was intended to affect, directly or indirectly, most of the estimated 100,000 population of the Zona. The research involved the application of a questionnaire covering demographic and economic characteristics of the households and their inter- household transfers of money and goods, to a representative sample of 507 households; and a series of in-depth anthropological case studies with some of these households. The objectives of the research were to demonstrate the importance of interhousehold transfers as survival and economic development strategies for the urban poor and to examine their operational implications for the design and implementation of future urban development programs. TABLE OF CONTENTS Page No. 1. Introduction ......................................................... 2. Research Findings on the Magnitude, Composition, Sources and Determinants of Interhousehold Transfers ....................... 2 2.1 Importance of transfers ......................................... 2 2.2 Transfer flows between income groups .............................2 2.3 Kinds of transfers ............................ .3 2.4 Determinants of transfers .....................3 2.5 Comparison of transferbehaviour of male- and female-headed households .................................. 6 3. The Relationships Between Upgrading and Transfers ....................7 3.1 Determinants of household decisions to upgrade and the effect of project interventions on the decision ..................................................7 a. Direct project effects on the upgrading decision ............7 b. Non-project related factors affecting the upgrading decision ....................................... 11 3.2 Effect of the Droject on transfer receipts. ................... 15 4. Policy Implications................................................... 15 5. Conclusions ......................................................... 19 Dictionary of variables used in the regression analysis..................23 Tables (following page 24) Table 1: Proportion of Households Giving and Receiving Monetary and In-Kind Transfers by Income ...................27 Table 2: Interhousehold Transfers: Their Importance and Direction According to Sex ......................... 2 Table 3: Proportion of Households Giving and Receiving Transfers and Their Value .................................. 29 Table 4: Types of In-Kind Transfers Given and Received by Sex of Household Head ................................... 30 (Continued) (Table of Contents - Cont.) Table 5: Modelling the Determinant of Interhousehold Transfers: The Conceptualization of the Econometric Framework........................................ 31 Table 6: The Determinants of Transfer Receipts: Results of the Econometric Analysis ................................ 32 Table 7: Summary of the Main Econometric Findings on Determinants and Magnitude of Receipts and Remittances of Monetary Transfers .......................... 33 Table 8: Determinants of Expenditure on Basic Needs Vs Non-Basic Needs ......................................... 34 Table 9: Differences in Transfer Behavior of Male- and Female-Headed Households ............................... 35 Table 10: Determinants of ICT-Induced Shelter Changes................... 36 Table 11: Determinants of Ongoing Construction Activity ................ 37 Table 12: Determinants of Labor Usage in Housing Development........... 38 1. Introduction The conventional wisdom in much of development economics has been that the only significant source of income for the most poor urban households is labor market earnings. It was assumed that most poor urban households had very limited access to formal credit markets and that access to informal sources of credit and transfer income had been reduced by the break-up of the extended family. An important consequence of this in the design of urban projects has been that estimates of project affordability have usually been derived exclusively from labor market earnings (see Keare and Jimenez,1980). Other sources of income, it has been argued, are either too small, too unstable or too difficult to estimate. The purpose of this paper is to demonstrate the importance of interhousehold transfers as a quantitatively important and relatively stable source of household income; and to assess the policy implications of these transfers for the design and implementation of urban shelter programs. By analyzing the determinants of shelter upgrading we will attempt to provide some generalizations regarding household and social network behaviour, which, couple with their own perceptions concerning the urban project will provide the context to the operational implications stemming from the behaviour of the network. It must be emphasized at the outset that this work should not be interpreted as an evaluation of the ongoing urban project. The socioeconomic survey was not intended as a n evaluation survey. Moreover, it was applied in April 1982, about two years after the Institutio de Credito Territorial (ICT) began implmentation and when the project was still at a relatively early stage. This is particularly important in the context of the present program, sicne the landfill component, which was cuasing hardships to many households -2- at the time of the survey, is likely to prove positive once the project is completed. The report complements Working Paper No.1 which discussed the structure and functions of social networks. 2. Research Findings on the Magnitude, Composition, Sources and Determinants of Interhousehold Transfers in Cartagena 2.1 The importance and stability of interhousehold transfers Approximately 35% of all households in the Zona Sur reported having received a monetary transfer during the month prior to the survey. If we include transfers in-kind, almost one half of the households received some type of transfer. It can be seen that for households who receive transfers, the transfers represent a very significant proportion of total income (Table 2). This is particularly true for poorer households: for the lower two income quintiles transfers represent between one quarter and one half of total household income. For female headed households the transfers are particularly significant. Most transfers also represent a relatively stable source of income. 80% of households who received transfers in the month of the survey had received at least two transfers during the previous 4 months; 50% of the households received transfers at least monthly. For in-kind transfers, 63.2% of transfer receivers got them at least once a month. 2.2 Transfer flows between different income groups Another widely held assumption about transfers was that they represented an "altruistic" gift from higher income to lower income households. Based on this assumption it would he expected that most transfers -3 would flow from higher to lower income households and that the proportion of transfer remitters would decrease for lower income groups. Table 2 shows that the pattern of transfers is more complex than the simple altruistic model. Although the proportion of transfer givers increases and the proportion of transfer receivers declines with income, even the lowest income groups include a significant proportion of transfer remitters. For the lower two income quartiles, more than one half of the male headed households remit transfers; and the proportions for female headed households are significant as well: over one quacter for the lowest quintile and over one third for the second lowest. It can also be seen that the lower income groups spent a higher proportion of their total income on transfers, which again shows that transfer remittance is an important activity among all income groups (Table 2). 2.3 Kinds of transfers Although monetary transfers account for the greatest part of the value of transfers (55% of all transfers) Table 3 shows that the remaining 45% of transfers are given in-kind. These may be in the form of food, clothes, other kinds of goods or the imputed value of free rent (Table 4). Gifts of food are approximately equal to the total value of all other types of in-kind transfers. Working Paper No.1 provides a more detailed analysis of the transfer of people, particularly children, through the social network. 2.4 Determinants of transfers a. Conceptual framework and analvtical procedures Table 5 shows the conceptualization of the econometric framework for the analysis of transfers. The main stages used in the analysis were the following: (a) Analysis of the determinants of a household belonging to a "socially interactive network" (i.e potentially having access to tran.sfers). -4- (b) Determinants of the probability of having received a transfer during the past month. (c) Determinants of the magnitude of total transfer receipts (d) Determinants of the magnitude of net transfer receipts. The same stages were followed in the analysis of transfer remittances. Multiple regression analysis was used when the dependent variable was continuous (magnitude of total and net transfecs received and given), and log-linear analysis was used when the dependent variable was dichotomous (probability of receiving or given transfers). In both cases 4 sets of explanatory variables were used: (a) Household demographic characteristics (sex,age and education of household head, and household size. (b) Income and wealth of the household (income, ownership of capital goods, ownership and value of the house, ownership of productive animals) (c) Access to a socially interactive network (number of households in the network, differences in the economic level of network households and the proportion of network households residing in Venezuela). (d) Project effects (upgrading has taken place after the start of the project and households have internal water connnections). Many of the variables were transformed into a dichotomous form _o0 help in the interpretation of the analysis. The detailed analysis is giv.Rn in Table 6 and a summary of the significant variables is presented in Table 7. b. Determinants of the probability of receiving transfers The probability of receiving a transfer is higher where the -5- household head is over 55 and lower where he/she is under 30. The lower the inzome of the household the higher the propensity to receive zransfers. Ownership of consumer durables is negatively associated, and there appears to be some negative association with changes produced by ICT in the sector where the household lives. The reason for this latter may be due to the fact that the ICT project began in the higher income sectors of the Zona, and there is a tendency for higher income households to receive less transfers. c. Magnitude of net transfers received Analysis is presented in Table 6 of both net and total transfer receipts. The discussion here is limited to net transfers i.e. the net impact of transfers on total household income. Net transfers are significantly affected by demographic variables, with more transfers being received when the head is female or over 55, or where there are a large number of household members. On the other hand the volume of transfer receipts is negatively associated with income and wealth. Families with more network contacts in Venezuela or who belong to a network with larger income disparities among members, tend to receive more transfers. Finally there is some relationship with ICT interventions as houses which have upgraded are less likely to receive transfers whereas houses with water are likely to receive more. d. Determinants of the likelihood of giving transfers The factors affecting giving of transfers are almost the mirror image of factors affecting transfer receipts. Consequently female and older household heads and large families are less likely to give transfers, whereas households with high income and wealth variables are more likely to give. e. Determinants of the magnitude of transfer remittances Small households and those with a well educated household head -6- tend to remit more transfers. Similary remittances increase for higher income households and those with more valuable houses. f. Determinants of expenditure on basic needs and non-basic needs Table 8 shows that expenditure on non-basic needs is higher for high income households, and is lower for those with a female head and with a large number of members. Consistent with our research hypothesis, non-basic expenditures are lower for households who receive a high proportion of their income in the form of transfers. g. The general influence of the main types of variables Transfer behaviour is strongly influenced by demographic variables. The sex and age of the household head as well as household size influence all indicators of transfer receipt and remittances. Income and wealth variables also have a consistent impact. Higher income households and those with greater wealth are more likely to remit whereas low income households and those with less wealth are more likely to receive transfers. The information on network structure is less consistent. Members of larger social networks are more likely to receive and remit transfers, which seems reasonable. However, there is some inconsistency between the fact that income differences between network households increase the magnitude of transfers receipts but reduce the probability of remitting transfers. The influence of ICT activities on transfer behaviour is difficult to interpret. (the issue is discussed in more detail later in the report). 2.5 Comparison of transfer behaviour of male and female headed households Table 9 summarizes some of the main differences in transfer behaviour of male and female headed households. Almost twice as many female headed households as male headed receive transfers (45.5% coripared with 26.8%). On average the amount of transfers received by female headed households is almost -7- 3 times as great and is proportionately twice as high. There is however, very little difference in receipt of in-kind transfers. Conversely only 21% of female headed households give transfers as compared to 46.4% of male headed households, and the-transfers represent a smaller proportion of total income 3. The Relationship Between Upgrading and Interhousehold Transfers 3.1 Determinants of household decisions to upgrade and the effect of project interventions on the decisions a. Direct project effects on the upgrading decision Household owners who had upgraded their house (55.5% of owners) sirice the start of the project, which was approved in 1979, were asked what were the factors determining their decision to upgrade. Of those who had upgraded their house, 90.3% replied that the reason was that the house was "buried" by the landfill component of the project; 3.4% said that they were planning to upgrade in any case; 2.9% said that they took advantage of the project intervention to upgrade; and the remaining 3.4% gave "other" reasons. Households were also asked whether they lived in streets which -had been subject to landfill; 76.4% reported affirmatively. Of those who responded affirmatively, 57.9% did upgrade since the beginning of the project; 42.1% did not; whereas the respective percentages for those households whose streets were not subject to landfill were 47.7 and 52.3. Although relatively small, a T-Test reveals that the difference is significant (one tail test at the 95% level). Having been subject to the project landfill component is positively associated with a higher incidence of own-upgrading by the house- hold. Visual and more qualitative evidence from the field work also -8- suggested a significant correlation between the street landfill and the house- hold's perceived need (and desire) to upgrade. However, the regression analysis reported in Table 10 shows that the landfill variable (LNDFILL), although having the expected sign and an estimate implying that households whose streets had been landfilled had a 14% higher probability of upgrading, lacks the sianificance level reported in the simple T-Test. Once a controlled estimation which searches for the independent contribution of each variable has been performed, as in the regression logistic estimation, it becomes apparent that there are factors other than the landfill component -- like "female headedness," tastes, an active community in a sector -- which are more important, both in terms of mnagnitude and significance level. The above result is not overly surprising in light of the caveats of making significant inferences from descriptive and/or simple correlatif.i analysis. Although upgraders stated that the landfill which resulted in their houses being buried was the most significant factor triggering upgrading, a large majority of non-upgraders were also subject to the landfill component (72.3%). Thus, however significant landfill was perceived to be by upgraders, the question still remains as to why so many households subject to landfill did not upgrade. It is clear that landf-ill is only one of several factors influencing the upgrading decision. The above evidence suggests that altiough a "buried house" may be a significant factor triggering upgrading for many households which either did not face constraints (e.g., lack of male laborers, absence of community parti- cipation, etc.), or had a particularly strong preference for shelter upgrad.ing, these same constraints become more important explanatory factors of the upgrading event decision itself than whether landfill has taken place or -9- not. It is also worth emphasi7ing that having own water supply had no effect on ICT-related upgrading, a finding that counters past evidence (see, for instance, Strassman for Cartagena) that strongly indicated the importance of water in explaining housing investments. It is likely that ICT's imple- mentation problems coupled with the expectation of many SEZ dwellers that they would have to leave the area, or be relocated within the zone at some point in the future, 1/ did not make shelter upgrading appear a good investment at this time. The insignificant result of the variable street improvements (STRTIMPR), which is a project componient that could be associated with appreciation in the housing value, is also suggestive in this context. The issue of the motivational structure regarding the household's upgrading decision can be analyzed further by utilizing the evidence on the household's own perception as to their quality of shelter at the time of survey, compared with the time prior to ICT intervention. Thirty-seven and a half percent of houise owners perceived their housing situation to be better, 31.6% perceived it to be the same, and 30.9% worse than previously. A cross-tabulation reveals that upgraders are over-represented in the "better off" and in the "worse off" categories -- particularly in the latter one --, whereas non-up- graders are over-represented in the "same shelter quality" category. Indeed, a logistic regression procedure -- Table 2, regression 1 -- reveals that, controlling for other relevant variables, having implemented changes in the house ("upgrading") is associated very significantly with a "worse off" 1/ When the survey took place ICT still had not gained ownership of the SEZ land which would eventually permit the regularization of tenure for the households in the area. In addition, rumors of a projected highway to be built between the rehabilitation area and the swamp which could result in households' relocation, further increased the sense of insecurity with regard to tenure. - 10 - situation as compared with either the same or a better off situation. ICT- triggered "upgrading" increases the probability of perceiving the current shelter conditions as "worse off" by 24 percent, ceteris paribus. Interestingly, the only other significant variable in the estimation is water ownership, which increases the probability of a "worse off" perception by 22 percent. If a dynamic urban setting were present, where many investment possibilities existed, one would have expected that housing upgrading would *have been associated with a significant improvement in the perceived shelter qujality by the households, and this improvement would have been particularly significant when the family had their own water supply. 1/ Instead, we find evidence to the contrary, which suggests that instead of investment-led shelter upgrading, the "upgrading" process has rather been one of aiming at a "basic level maintenance,' where the initial landfill component made some shelter chanaes necessary in order to avoid ending up significantly worse-off than previously. 2/ In terms of locational variables, the variable that segments the SEZ into four geographically contiguous sectors of intervention is very signi- ficant. Sector I is the most advanced in terms of project implementation, although much of that zone's improvement took place before the ICT project was fully underway. Sector II has been the most active zone in terms of project intervention by ICT, and in terms of community participation, whereas Sectors III and IV were lagging behind on both counts. It is not surprising, there- fore, that the SECTGR2 dummy is very significant and positive, whereas the 1/ See Strassman, op.cit., on the relationship between own water supply and housing investments in Cartagena. 2/ As discussed previously, a significant portion of households facing the need to upgrade might have faced financial constraints which precluded translating that perceived need into actual upgrading. SECTOR4 dummy is negative and almost significant when compared with the base dummy SECTOR 1. A household in the community-active Sector II has an 18% higher chance of having upgraded than a household in Sector I. It is quite telling that the sector with the most active community participation, which has been another component conducive to removal of labor-constraints, turns out to be this significant. b. Non-project related factors affecting the upgrading decision Table 10 presents the results of the regression analysis 1/ in which determinants of upgrading (CHNGICT) are examined. As this is a Yes/No variable, a logistic regression procedure was utilized. 2/ First, we see that income level is not significantly associated with the decision to upgrade 3/ This might be the result of opposing forces. Higher income households have a higher ability to upgrade; but many lower income households, who live closer to the swamp, have generally faced a more extensive project intervention, especially regarding the landfill. Furthermore, lower income households live in houses more likely to be buried by a given amount of landfill around it, 1/ A more analytical empirical procedure like regression analysis is warranted in this case because a descriptive perception qujeqtion like "why did you upgrade?," however telling, fails to: i) shed light on the differences between upgraders and non-upgraders, ii) identify the relevant constraints to upgrading by concentrating solely on an ex-post factor analysis for households that were both willing and able to upgrade, and iii) give an idea on the relative and independent contribution of each factor explaining upgrading. 2/ An ordinary least squares regression procedure would not be adequate when the dependent variable is discrete. For details, see Theil. 3/ A yearly average income measure (rather than monthly current income) was constructed in order to capture the more inter-temporal elements embedded in the upgrading dependent variable. Furthermore, this income variable was divided into two segments, a low income segment (less than col. fi 16,000; US$1 = approximately col i 57) and a high income segment, in order to test whether there are qualitative behvioral differences between both groups. - 12 - since they live in houses of with lower ceilings and worse floor conditions. 1/ Moreover, in a setting like the Cartagena SEZ, it is important to consider the possibility that much of the desired upgrading by "higher" income households was done prior to the project (in the "consolidated" area); whereas the same is clearly not the case in the newer and poorer "rehabilitation" area. In fact, the marginal increase in the probability of engaging in upgrading is higher for a lower income household which has additional disposable income than for its higher income cotunterpart. This finding, coupled with the evidence showing that the amount spent on construction increases more sharply for higher income households is consistent with the "upgrading need" versus "budget constraint" hypothesis. Upgrading is a continuous, incremental process (in the "progressive development" mold) for low income households, whereas higher income households make fewer but much larger investments. This relationship between household income and upgrading is supported by the relationship between current household income and ongoing construction activity. Table 11 shows that whereas there is a willingness to upgrade regardless of income level, as exemplified by the construction activity event taking place at all income levels, the amount invested in upgrading increases significantly with income. 1/ Even if the project intervention affected households of different incomes equally, higher income households would not necessarily be expected to upgrade -- relatively speaking-- more than their lower income counterparts in a situation where both groups are in some equilibrium consumption/investment path. In fact, under some circumstances of reduced access to credit markets by lower income households, coupled with the possible higher shelter elasticities of such a group, one could predict a priori that successful project interventions would induce more upgrading by the poorer group. 13 - Female-headed households upgraded significantly less than their male-headed counterparts, controlling for income levels and other socio- economic variables. This finding is consistent with the ethnographic field work, where the lack of male laborers ("there is no man in the house") was frequently given by women heads of households as a reason for lack of upgrading. The significance of the non-hired labor force in upgrading is also strongly suggested by the other household composition variables. The inore members the household has, the higher the probability of having upgraded. Conversely, the higher the dependency ratio -- defined as the number of economically inactive household members divided by the total number of house- hold members -- the lesser the chances of having upgraded (albeit not very significantly). These findings on the effects of being a female-headed household are also complemented by the evidence on the construction activity regression in Table 11, which suggests that female-headed households have a lower probability of currently upgrading than their male-headed counterparts, controlling for other socio-economic variables. Our hypothesis that lack of male labor availability is more of a constraint to upgrading by female-headed households than lack of financial resources, is supported by the fact that those female-headed households who manage to upgrade are able to invest as much resources as male-headed households, controlling for income. 1/ The importance of the labor factor is implied in two variables related to the social networks, namely the number of households in the social network (HOGRED), and the HISTORY variable, which inquired whether there had been unpaid labor help from relatives and/or friends during the construction 1/ In fact, on the average they devoted slightly more, although the difference is very insignificant. - 14 - process following the initial invasion and squatting in the Southeast Zone. Both variables are correlated positively with ICT-related upgrading, although the level of significance is not very high. (See Table 12 for determinants of labor usage.) However important the labor availability factor might be in explaining the upgrading decision since the project began, and however important the role of the network in providing labor resources might be, it is interesting to note that the same does not apply to interhousehold monetary interactions with other households. Households that engaged in economic interactions with other households in the social networks during the past year, and households that received larger magnitudes of income transfers in that period, had a lesser chance to upgrade, (although not very significantly). Behavioral empirical analysis of the type performed here generally relies on observable socio-economic and demographic variables. The unexplained compoaent of the variance in the analysis is very commonly ascribed to unobserved individual differences. In this case we attempted to capture -- at least partially -- the household "idiosyncrasy component, or the household preference function. This was done by asking households about their priority expenditures if their monthly income increased by 4,000 pesos, and by also asking them the same question if they faced a one-timne windfall of 40,000 pesos. In the creation of the variable PRIORITY those households that had housing expenditures high in their priority were given a dummy equal to one, otherwise a zero. The estimaticn indicates that having a "taste" for shelter increases the probability of having upgraded by 15 percent, and the result is statistically quite significant. 1/ There appear to be households 1/ The confidence level for a one-tailed test is 94.5%. - 15 - that simply have a "taste" for better housing. 1/ 3.2 Effect of the project on transfer receipts Consistent with the previous findings, ICT-induced upgrading did not elicit additional transfer receipts by the upgrading household; in fact there even appears to be an inverse relationship between the project-induced upgrading and transfers received (Table 6 shows that the probability of receiving a transfer and the amount of transfer received bothl have a statistically significant negative relation with ICT induced upgrading). The descriptive information is also concordant with this evidence. To the question what was the expenditure item to which the first monetary transfer received was intended, households replied in 77.3% of the cases that it was for genieral expenditure purposes; only in 2.5% of the cases was it specifically "earmarked" for housing, and the remaining 20.2% for goods, education, medicine and others. All the above evidence is clearly at variance with our Previous findings for El Salvador and the Philippines, where the project-induced effect on income transfers wa! very significant. We shall explore later the question of what might account for this difference. 4. Policy Implications As mentioned previously, at the time when the socio-economic survey in the Southeast zone of Cartagena took place, the zona dwellers did not have 1/ The "taste" for additional investments in housing was recorded after they. had implemented shelter changes. As socio-economic variables like income and household composition are controlled for; the "taste" relationship might indeed be a proxy for a pure preference variable. 16 - a very positive perception regarding the merits and prospects of the project. The implementation of various project components had been delayed; with a few exceptions community participation had not taken off; street landfill had buried the houses in the large rehabilitation area, and land tenure issues not only had not been resolved, but rumors existed that many households would eventually have to be relocated. These problems explain the householdst perception that their own expenditures in shelter, albeit in many cases necessary for maintaining a basic level of services (i.e., raising the house), were not viewed as necessarily good investments. This perception appe,rs to affect household behavior in various policy-relevant ways. One of the more relevant findings, which contrasts with the earlier studies in El Salvador and in the Philippines, is the lack of project-induced monetary transfers. The existence of shelter-related project components does not trigger additional income transfer-s from other households, and the existing transfers are not specifically channeled (earmarked) to shelter expenditures. It should be noted that these findings are contradictory to W.P. Strassman's results in his Cartagena study on housing upgrading. However, methodological and data problems suggest that Strassman's findings might be spurious. I/ It is quite suggestive to compare the findings of the three countries studied so far. Both settings in El Salvador (sites and services projects in Santa Ana and Sonsonate), and the reblocking and upgrading Tondo 1/ Strassmnan infers that transfers play a very important role in housing upgrading on the basis of a test that has two main problems, namely: i) uses relatives in Venezuela as a proxy for transfers from Venezuela (he also does not attempt to capture information on other transfers), and ii) does not control for total income in the multiple regression that explains upgrading, i.e. the estimated relationship between (proxy for) transfers and upgrading might be no more than an implicit general relationship between household income level and upgrading, as suggested by the econometric work presented in this paper. - 17 - foreshore project in Manila, Philippines, were progressing well at the time of the studies. Both the qualitative and quantitative evidence indicated that the respective project populations regarded shelter-related expenditures as sound investments. As a result, the receipt of interhousehold income transfers incresed and a significant proportion of the total transfers were earmarked 1/ for housing expenditures. The Cartagena qualitative/anthropological field experience revealed that the house is generally perceived by the family as an asset to be built/ upgraded with the financial resources of the household itself. As Parris 2/ writes on the basis of this in-depth interviewing of selected households in Cartagena: "As a general finding or observation, I would state that neither peoras (worsened conditions) nor mejoras (improvements) seem to have elicited much more in the form of transfers than if the project were not going on. Housing in the SEZ appears to be an issue for households themselves to work on, without being able to "coax" additional transfers from network relatives outside the zone. For networks with households within the zone, transfers do not seem to have focussed on housing in such a way as to change the exchange relationships ... Building the house was a consolidation of the household itself, as it were, one they were attempting to do financially by themselves alone In fact, since interhousehold transfers are fairly common for the fulfillment of other basic needs, it is not uncommon to find that households 1/ Earmarked in the sense that the same amount of additional income for an average household, if it was not coming from transfers, would flow in lesser quantities to shelter expenditures. 2/ See S. Parris, "Issues in Income Transfers Among Households in the Southeast Zone of Cartagenam, Colombia: Types of Social Network Participation, Kinds of Exchange and Project Effects" (pp. 42-3). Unpublished. - 18 - which are investing their own resources during housing upgrading do channel existing transfers, or in some cases request additional transfers, in order to satisfy their regular consumption level of other goods. The absence of an investment incentive to remit monetary transfers to assist in the process of upgrading, as in the Cartagena case, does not imply an absence of project-related network support. In fact households received considerable assistance in the form of free laxor to help with "digging out" houses which had become subject to the risk of flooding after the level of the roads had been raised. The lack of monetary transfers to assist with upgrading is consistent with the findings of Peattie and others which suggest that shelter upgrading, albeit important, is of lower priority for the low income household. 1/ Thus, we find that in general, the social network will not transfer to a household significant amounts of money to satisfy the household's desired level of shelter consumption. Those monies will he channeled only if the shelter upgrading is perceived as a sound investment. If it is not necessarily seen as a good investment, as in Cartagena, the social network will in some cases substitute monetary help for labor help, in particular for those households that have faced some downgrading in their shelter conditions due to the landfill component. This behavioral response is concordant with the importance attached by low income households to the 1/ A survey of squatters in Nairobi finds the following order of frequency in terms of their priority list: food; housing at the standard of shelter current at the squatter location; school fees; clothes; money to buy land; money to contribute to the extended family; money to extend a business; money to build an additional room, perhaps for rental; and money to rent or buy a better house or room to live in (A. Hake, 1977). Moreover, Peattie (1979) reports on a study of persons sleeping on the pavements in Bombay, which finds that many had resided in the city in this manner for many years, and were not partictularly anxious to change their living arrangements (Ramachandran, 1972). In Peattie's words, "... they had no housing, but not relative to their concerns, a housing problem." - 19 - maintenance of a basic level of shelter. 5. Conclusions Interhousehold income transfers are significant and relatively stable among the urban poor, suggesting the importance of accounting for these incomes in affordability calculations. An income measure which fails to include transfer income will seriously underestimate the total income of many urban households. In particular, a significant underestimation will occur in the case of many low income families and for a large fraction of female-headed houiseholds. These large amounts of resources that can move from one household to another represent an important potential source for the household residing in the urban informal sector. In many cases this potential source of monetary and in-kind resources is actually utilized by a family in dire straits in order to reach a minimum level of income required for basic needs satisfaciton; in other cases this source is tapped as a substitute to capital markets in order to take advantage of investment opportunities when the house- hold faces a liquidity constraint. In the case of shelter, our evidence from El Salvador, the Philippines and Colombia suggests that whereas the maintenance of a certain (current) level of shelter conditions is regarded as a basic need, implying that the household is "entitled" to have access to the network's resources (especially labor) when those conditions are downgraded, the significant upgrading of shelter conditions is generally not regarded as a basic need. The uipgrading household will have access to the network's financial resources only if such upgrading is perceived as a sound investment, from which the - 20 - network may expect a fair return. In other words, transfers which are used to complement the imperfect capital markets appear to be allocated according to a sound economic rationale. In settings where shelter upgrading is not necessarily seen as a promising investment, as in the Cartagena area at the time of the survey, the upgrading process itself will be closely associated with the preferred "'consumption basket" of the household, and, perhaps as importantly, with the clear desire by the household to consolidate itself as such. As a result, households view shelter upgrading as a relatively "individualistic" piocess, to he financed by the household alone. Labor from other households is indeed utilized in many cases, particularly in helping the household to upgrade, but additional transfer monies do not appear to be triggered by the upgrading process. When reviewing household income data during project desian, it is therefore important not only to include transfers in total income, but also to consider the likely scenarios regarding both the household's perceptions vis- a-vis the project interventions and its ultimate objectives in the upgrading process -- whether the motivation is investment, consumption, household consolidation, or a combination of these. The calculation and interpretation of any measure of affordability and of willingness-to-pay for project components could vary significantly depending on these household perceptions and objectivs, since both the potential access to the network's resources and the household's demand for the project components would vary. The issue of a household's demand for shelter upgrading being affected by perceptions and motivations has implications beyond the already emphasized social network interactions. At a more general level, it appears that whether project components help trigger upgrading will depend on the - 21 - household's assessment of the soundness of the shelter-expenditure as an investment. For instance, owmership of water supply, and street improvements had not contributed to upgrading in the Cartagena setting. The contribution of the landfill component in explaining upgrading wa;; also found to be less than originally thought, in part due to the role of the constraints faced by the families whose houses had been buried. A parti- cular issue which is raised by this finding is the need to incorporate into any affordability analysis the availability of labor. In urban informal sector settings where labor markets function less than perfectly and where the household faces serious financial constraints, a family having access to inexpensive or free male labor, either from its own household or from the network, will be at a significant advantage over a similar family that lacks such access to male labor. I I -23- Dictionrary of Variables (in Alphabetical Order) ANIMALS: Number of pigs, horses, and chickens in the house (weighted average) BETTERRSNG: Dummy for the household perceiving better housing conditions in relation to the condition of the house before ICT project started CARTGYRS: Average years hor.sehold head and spouse have lived in Cartagena CHNGICT: Dummy for changes in the house implemented by the household since ICT project started CONSTRUC: Expenditure in construction during survey month DCONSTR: Dummy for ongoing construction activity at time of survey DEPRATIO: Number of non-working household members over total number of household members DIFFSIT: Dummy for households in the network whose economic situation is different from the household interviewed DOWiNH20: Dummy for water connection for houseowners DSTNCSWMP: Distance from swamp DUMMY ISS: If any of the household's workers belong to Instituto de Seguro Social (Social Security) DUMMY LOW: Dummy for Low Income (if low income greater than 0 the Dummy Low = 1) EDUCATHEAD: Hdousehold head education level FEMHEAD: Female household head FILUCTUATION: Best income obtained in the last 12 months minus worst income obtained in the last 12 months HHMBRS: Number of household members HI INCOME: Income net of transfers if household's income > 10,000 HIPRMINC: If perma.ient income greater than an average of Col. $10,000 pesos per month - 24 - HIREDLBR: Whether the household had hired labor during the construc- tion process following the initial invasion and squatting in the Southeast zone (Dummy) HiISTORY: Whether there had been unpaid labor help from relatives and/or friends during the construction process following the initial invasion and squatting in the Southeast zone (Dummy) HOUSEHOLDS: Number of households living in the same house HSEOWNER: House is owned by the interviewed household HSEVALUE: Value of the house for house owners INTERACTION: Dummy for those households who had any economic interaction with another household in the network duiring the last year LBRUTILZ: Percent of average man-working hours for male workers in the household LNDFILL: Dummy for street land filled (Dummy segmenting the rehabilitation area, which was subject to landfill, from the consolidation area, which was not) LOW INCOME: Income net of transfers if household's income < 10,000 LWPRMINC: If permanent income less than or equal to an average of Col. $10,000 per month MALEMBRS: Number of males in the household MCONSTR: Magnitude of construction outlays during survey month for those that are constructing MONETTRANSEC: Monthly average of monetary transfers received MONETTRANSSENT: Monthly average of monetary transfers sent NETWORK: Number of households in the social network NTWRKLBR: Whether the household had been helped by (unpaid) labor from the social network during the construction process following the initial invasion and squatting in the Southeast Zone (Dummy) OLD: Dummy for age of household head greater than or equal to 55 OTRWEALTH: Other goods bought by the household: refrigerator, television, music equipment, radio-cassette, sewing machine, blender, dining room set, living room set, etc. PERMINC: Permanent income proxy (average yearly income measure) - 25 - PRIORITY: Dummy for priority in intended housing-related expenditures if the household were to have a monthly income increase or a one-time windfall RATIO TRANSFERS: Transfers over net income (composition of income) SECTOR2: Dummy for ICT-2 sector in the Southeast zone in Cartagena SECTOR3: Dummy for ICT-3 sect;r- in the Southeast zone in Cartagena SECTOR 4: Dummy for ICT-4 sector in the Southeast zone in Cartagena STRTUPGRD: Dummy for street upgrading VENEZRTIO: Total number of households in the network that live in Venezuela over total number of households in the network WORSEHSNG: Dummy for households whose condition is worse than it was before the ICT project started YOUNG: Dummy for age of household head less than or equal to 30 I \I .N. k Ir I Table I Proportion of llouseholds Giving and Receiving Monetary and In-Kind Transfers by Income Receivers Givers Net Monthly % Households % Hlouseholds % Households % In-Kind . Households % Monetary % llouseholds % In-Kind Income Receive Transfer Over Receive In-Kind Transfer Over Give Monetary Transfer Over Give Tn-Kind Transfer Over (Quintiles) Monetary Total Income Transfers Total Income Transfer Total Income Transfer Total Income Pesos Transfers 0-8,000 43.5 15.8 , 40.0 9.2 26.2 11.9 27.4 6.3 .8,000-11,000 26.7 4.8 28.9 6.1 36.0 4.8 24.4 3.7 11,000-15,500 25.3 3.3 25.3 3.3 42.2 4.7 25.3 2.6 15,500-24,500 35.3 3.3 31.8 3.3 50.6 2.9' 31.0 2.9 24,500 and more 20.0 1.6 22.4 1.0 50.6 5.3 37.6 3.3 Table 2 Interhousehold Transfers: Their Importance and Direction According to Sex Receivers Givers % llouseholds % of Transfer Transfers Income Over % Of Transfer Income Net Monthly Income Receivers Total Income Average Transfer % Households Over Total Income Average Transfer (Quintiles) Sex (for receiver) for a Receiver Transfer Givers (for Givers) for'a Giver F 66.7 51.6 3,164 26.7 16.9 1,038 0-8,000 M 61.8 40.3 1,823 50.9 39.5 1,825 F 45.0 32.2 4,702 35.0 23.4 2,651 8,000-11,000 M 44.3 18.9 2,252 52.2 16.2 1,979 F 52.6 19.5 3,877 36.8 5.2 1,080 11,000-15,500 H 35.9 22.3 4,027 54.7 13.7 2,217 1 N, F 50.0 17.3 4,079 64.7 4.8 1,040 °° 15,500-24,500 M 52.2 13.2 3,154 64.2 10.7 2,468 F 38.5 11.7 4,375 15.4 7.8 5,660 24,500 and more M 31.9 7.8 2,971 71.4 13.7 6,196 Receivers Givers % Of Transfer % Households Income Over % Of Transfer Income Transfers Total Income Average Transfer % Hlouseholds Over Total Income Average Transfer Sex of Head Receivers (For Receivers) For a Receiver Transfer Givers (For Givers) for a Giver Female 56.2 30.3 3,914 35.2 12.5 2,178 Hale 45.7 19.4 2,917 59.0 18.7 3,176 *Net Income: Income net of interhousehold transfers *Honthly average Table 3 Proportion of llouaeholds Giving and Receing Transfers and Their Value Received Average Tranafer Magnitude Average Transfer Average Transfer Received of Monetary % Houselholds Magnitude Received of Magnitude Received Z Households Receive Transfers Only Receive In-Kind In-Kind Transfers Only Z Households of Both Transfers ;ex of Head Monetary Transfers Only (Pesos) Transfers Only (Pesos) Receive Both Transfers (Pesos) Female 38.2 2,937 20.6 1,979 41.2 5,788 Hale 34.0 2,359 40.0 2,459 26.0 4,340 Average 35.2 2,540 34.3 2,375 30.5 4,903 Given Out Average Transfer Magnitude Z Households Average Transfer Average Transfer % Households Glve Given of Monetary Give In-Kind Magnitude Given of Z lHouseholds Magnitude Received Sex of lHead Monetary Transfers Only Tranafer Only Transfers Only In-Kind Transfers Only Give Both Transfers of Both Transfers Female 32.6 948 39.5 2,872 27.9 2,628 Hale 46.1 2,052 22.6 1,929 31.3 5,729 Average 43.9 1,916 25.4 2,172 30.8 5,264 *Monthly Averages 'able 4 Types of In-Kind Transfers Given and Received by Sex of Hlousehold llead Receivers % hlouseholds lhouseholds Hlouseholds X hlouseholds X Houselholds % Households Z louseholds X liouseholds Transfers Receive Monetary Receive Receive Food Receive Clothes Receive Other Receive Imputed Rent Receive People Sex of hlead Keceivers Transfers In-Kind Transfers Tranisfers Transfers Transfers a/ as Transfers bI Transfers c/ Female 56.2 45.5 34.7 21.1 7.3 4.9 7.3 1.6 sale 45.7 27.8 30.2 14.2 5.8 2.4 11.1 2.9 Average 48.3 32.2 31.3 15.8 6.1 3.0 101 2.6 Givers % Households X Hlousehold % llouseholds X louseholds % liouuelholds Z Households X llouseholds % tUouseholds Transfers Give Monetary Give Total Give Food Give Clothes Give Other Give Imputed Rent Give People llead of Sex Givers Transfers In-Kind Transfers Transfers Transfers Transfers a/ as Transfers b/ Transfers c/ Female 35.2 2L.1 23.8 14.3 6.4 1.9 3.3 7.3 Hale 59.0 46.4 31.9 20.6 9.2 4.8 5.0 6.7 Average 53.1 40.1 29.9 19.0 8.4 4.0 4.6 6.9 s/ Other transfers iaclude: school supplies, books, home appliances. bi Free renters. c/ Household members being taken care of by other households. *Figurs in monthly averages. - 31 - Table 5 Modelling the Determinant of Interhousehold Transfers: The Conceptualization of the Econometric Framework Household Belongs to Socially Interactive Network Yes No Net Transfers Received Net Transfers Received Last Month Last Month Zero Positive Zero Negative Received Made Transfers Transfers vi) Magnitude Zero transferred equal zero; or transfer ii) Absence of transfer event - 32 - Tabte 6 Th. Determinants of Transfer PacuLipts: Resultsa of tile ftono..tric Ana1ls1a 123 4 'anudof ?robabtlity flat Transfers of Rea ...ioi ".agnitud. of 7r.b.biltity of 7tobabLIlty, for Socially Leat ¶o-th for Transfers Rec.i-ed CLimgo :..st 4ooth Hagmitude of trrasfer, of .Net.ork Interactiye Int.ractive for Those That for Int*ractIve Zlveo for Those taterattia* Household. Household Rec..ived L..t I.onttb 1..sthold That G.,,. Last .4.ont2 Variables Se.An (DINTRATI) (X(TROTOYIl) (D13TRh) (XISTRTOT) (0t,fTSI) ~lttT INTERCLET -0.021 -931.71 -0.698 332.64 1.731, 66.39 (0.913) .I (1647.99) (0.93Z) (2075.33) (1.019) (1494.79) L.OW ICOIIF, 2292.8 0.027xl103 -0.032 0.0Z.O29iIO. 0.085 3.042z10o- 0.038 (0.087%10- ) (0.145) (0.o1z10-Z) (0.138) (Q.0a3xLO03 (0.146) HI t4COLIE 13164.6 0.O19X10-3 -0.10354' -O.014z10'3 -0.0077 0 13s:10-3 0 .I1L (O.015X103) (0.0167) (0.0099X103). (0.0215) (oO1OL31O'3) (0.014) DUMMY LOW 0.37 -0.39 -867.71 2.270' -1564.A -0.609 458.14 (0.675) (1194.11) (0.926) (1142.1) (0.69 L) (1210.41) CART. YEARS 44.64 0.00044 5.521 0.0018 10.82 -0.0030 3.61 (0.00403) (7.047) (0.0039) (7.76) (0.0043) (6.11) EDUCAT. HEAD 3.3.7 0.0059 -61.27 0.145 135.21 -0.014 365.46' (0.1249) (229.43) (0.131) (24.7.67) (0. 139). (215.85) Y0UNG 0.17 0.57 -392.59 .0.894' 292.42 228.07 (0.413) (666.92) (0.373) (758.93) (0.400) (591.36) 0 LI) 0.24 .0.58' 123.05' 3.728' 9.42 -14'105.6 (0.30) (599.66) (0.366) (616.5) (0.06) (622.9) ?E?0FAD 0.23 -0.016 1693.94' -0.058 1421.06' -0.84' -512.23 (0.326) (600.98) (621.55) (0.34) (008.12) HSEO6flEI 0.33 0.461 -1032.41' -0.297 -738.30 0.178 532.64 (0.3L9) (580.81) (0.324) (645.24) (0.361) (554.01) HSEVALUE 295076.0 -0.014,10-5 -0.00239' -0.0055X10-5 -O.0L7xLO0Z 0.0lZzlO-4' (0.00t57)' (0.033XL10S) (0.00072) (0.043xL105) (0.094.LO02) (0.0066.LO04) (0.00058) OTHRWEALTH 3.37 -0.0705' -115.97 .0.169* -1.036 0.076' -15.40 (0.0405) (71.16) (0.042) (79.641) (0.044) (66.81L) -DUMMY ISO 0.34 -0.191 0.0016 -0.16 (0.297) -(0.2991) -(0.33) KH)OBRS 6.67 -0.019 250..0' -22.04 --0.091' -313.81' (0.052) (93.04) (0.054.) - (99.37) (0.056.) (88.65) DEPRATIO 0.78 1.375 -581.75 1.21 .20L.53 .0.78 3016.17 (0.929) (1638.17) (0.96) (2022.45) (0.98) (1459.05) NETWrO RK 1.17 0.040 20.8 0.006 152.8' 0.143' 57.36 (0.037) (65.6) (0.037) (751)(.045) (57.20) VENEZRTtO 0.06 0.84 4749.56' 1.21 5202.5' 0.57 .1179.4 (1.13) (2079.35)0 (1.28) (2333.7) (1.34) (1858.3) DT.FFTSIT 0.86 0.58' 2372.9' 1.29' 1436.6 -0.97' -614.19 (0.33) (698.2) (0.40) (950.9) (0.48) (593.17) FLUCTUATION 466).6 -0.079a10- 0.033 - .0270O -0.022 (0.Zx10-) (0.030) - (0.0307) -(0.025) 0W54H2O 0.27 0.05 Z210.5' 0.57 1107.01 ~1 0-1309.56' (0.37) (691.2) (0.39) (810.11) (0.44) (641.66) CHP.fOICT 0.41 .0.37 -998.3' -0.779' -1297.15' 0.3i -.01.4 (0.37) (487.7) (0.27$9) (572.79) (0.30) (434.6) A NL4 AL S 1.77 -0.033 -116.15' -0.043 -19 .15 0.035' 53.5 (0.027) (36.16) (0.033) 74.64 (0.343) 4u5.4) HEAN Or DEPENDENT VARLABLE 0.79 -15.83 0.59 3.198 0.68 2.817 4NMBER or CASES 420 331 332 198 332 225 2 FREDICrED CORRECTLY 79.8 - 71.1 - 75.9 LOG LIKELIHOOD RATIO (LLR) AT 0 431.19 -4.48.60 -415.88 AT COHVERGEMCE 404.66 -368.67 - 334.60 R-SQOAR.E 0.339 -0.170 - 0.414 R-SQUARE AD.XUSTED -0.297 0.078 0.357 a/Standard error in parentheses. * Siguificant at 952 le-el, o-atail.d teat. - 33 - Table 7 Summary of the Main Econometric Findings on Determinants and Magnitude of Receipts and Remittances of Monetary Transfers Dependent Variable Magnitude of Probability Magnitude of Probability Independent Net Transfers of Receiving Transfers of Remitting Variables Received Transfers Given Transfer Demographic Female head + Head over 55 + + Head under 30 Household size + Education of Head + Income/Wealth High income - + Low income House value - + + Consumer durables + Owns house - Owns productive animals - + Network Variables Households in network + Network households in Venezuela + Income differences between network households + Project Variables Upgrading since project began - Water connection + R2 Adjusted 0.297 0.35 % Cases correctly predicted 71.1 75.9 Log linear ratio: At zero 448.6 415 At Convergence 368 314 Number of cases 331 332 225 332 Key: "+" positive coefficient is significant at 5% level. to-lt negative coefficient is significant at 5% level. " " coefficient not statistically significant at 5% level. - 34 - Table 8 Determinants of Expenditure on Basic Needs vs Non-Basic Needs Variables Mean Basic Needs a/ Non-Basic Needs b/ INTERCEPT 4128.601* -290.786 (2091.534) (1135,r598) LOW INCOME 2357.828 0.319 0.912 (0.229) (0.125) HI INCOME 16518.946 0Q439* 0.219* (0.026) (0.014) DUMMY LOW 0.323 1069.891 616.111 (1980.819) (1075.486) RATIO TRANSFERS 0.038 1213.980 -3509.925* (3698.800) (2008.264) FEMHEAD 0.233 -1335.506 1215.670* (866.404) (470.614) HNMBRS 6.669 233.571* -400.485* (139.389) (75.681) DEPRATIO 0.778 232.146 3659,117* (2522.414) (1369.545) DUMMY ISS 0.337 273.476 -76.744 (784.722) (426.065) HfSEOWNER 0.742 618.476 341.855 (821.264) (445.905) MEAN OF DEPENDENT VARIABLE 14510.71 3958.54 NUMBER OF CASES 427 427 R-SQUARE 0.503 0.426 ADJUSTED R-SQUARE 0.492 0.414 *Significant at 95% level, one tailed tests. a! Basic Needs comprise: food, water, electricity, clothes, education, medicines, rent, transport, construction material. b/ Non-Basic Needs comprise: lottery, entertainment, help given to family and friends, debt payments, home appliances, gas for cooking. - 35 - Table 9 Differences in Transfer Behavior of Male- and Female-Headed Households Average Average by 5ex of Head .Average M¶onetary F 1496 Transfer Received 795 5 595 " of Monetary Transfer * F 11.9% Received Over Total lacome 5.7% M 3.9% Average tn-Kind F 719 Transfer Received 747 H 757 ' of Tn-Kind Transfer F 4.2Z Received Over Total Income 4.6% H 4.7% Average Transfer F 2200 Received 1547 M 1332 % of Transfer Received F 16.1% Over Total Income 10.4 M 8.6% Average Monetary F 227 Transfer Given 906 H 1130 Z of Monetary Transfer F 1.6% Given Over TotaL Income 5.8 H 7.1% Average In-Kind F 538 Transfer Glven 703 H 756 Z of In-Kind Transfer F 2.8% Given Over Total Income 3.8 H 4.0Z Average Transfer Given F 767 1598 h 1873 Z of Transfers Giveu F 4.4% Over Total Income 9.5% 8 11.0% Labor Income F 73.8% Over Total Income 79.1% M 80.8% Non-Labor Income F 8.7% Over Total lacome 9.9; X 10.2% Income Sent by Temporary F 0.9% Absent Over Total Income 1.5% M 1.7% Average Total Income F 15494 18877 M 19902 - 36 - Table 10 Determian-t of ICT-Iuduced Shelter Cbhuges Variableg _e__ OH1-ICY INTEPCEPT -0.567 (1.118) a/ 1-0.142) b| LWPRMINC 2230.74 . -0.793zlO4 (O.895d10-4) { -O 197xlO- 4i HIPRAIINC 13393.0 -0.13rlo DUMHY LOW 0.375 0.148 (0.661) 10.037] £HTELkCTIO?4 0.763 -0.279 (0.323) [-0.0701 flONETTRANSR&C 1143.26 0.53xlO 4 (O .51xlO4) [0.13xLo04] MONMLRANSSINT 1148.47 60.65xlO 4 (O .59x10:4) O. 16%10 41 SOCSECRTY 0.330 0.219 (0.239) [0.055]1 FEMlHEAD O.Z53 -0.675' (0.315) [-0.169] DOPRATIO 0.786 -1.286 (0.91.8) [-0.322] HH08RS 6.942 0.107w (0.053) 10.027] HETWORIR 7.226 0.037 (0.038) 10.009] HISTORY 0.311 0.244 (0.252) 10.061] PRIORITY 0.872 0.612 (0.385) [0.153] LNDFILL 0.763 0.578 (0.420) [0.145J STRTUPGRD 0.506 0.329 (0.533) 010082 1 SECTOR 2 0.372 0.716 * (0.2Z11) 10.179] SECTOR 3 0.321 0.225 (O.475) to.056] SECTOR 4 0.164 -0.774 (0.555) I-0.19.J POWNH 20 0.329 0.101 (O.312) 10.025 ' MEAN CHNSIC- 0.567 tiUMBEF. OF CASES 312 , PREDICTED CORRECTLY 66.0 LOG lIKELIHOOD RATIO (LLR)- A' C 427.85 AT CONVERGENCE 374.51 Slgnlficant at 95% level, one tailed taste. *J Standard error ic parenthesee. b/ Traoiforuation of the unbracketed logit coefficient to a probability estimate Lo square bracketa. Cdone at the mean of the dependent vatriable p by the formula l (1-p)3 01.0 00 .41 0 . .4- 0 4. 0 0..4 0 0.0 4 00 !1 . ' 4 .0.0.. . . . 044 0 1 0 0 0 0 01 . 0 0 0 .I1 0 0 - I o *1 ' A 0 0 . 0 0 - 38 Table 12 Deteria4ants of Labor Usage In Hiousing Development VARIABLE c/ MEAN HIREDLER NTWRKLBR INTERCEPT - -1.430 -1.371 (0.958) (1.065) [-0.3.58] b/ |-0.2741 PERMINC 15498.6 0.276x10 4 0.012xl0- (O.142x10'4) (O. 149xlO 4) I0.069x10'4] 1-0.002x10-4) INTERACTION 0.769 -0.Z39 1.219' (0.327) (0.411) [-0.0591 10.2461 MONETTRANS REC. 1172.5 0.004x10-4 0.312x104 (0.487xlO04) (0.487xlO4) [0.0O1x10-4 10.062x104] MONETTRANS SENT 1163.9 0.806x10-4 -0.538xlO 4 (0.58 x10-4) (0.644x10:4) I0.201xlO04] 1-0.107xlO 4j FENHEAD O.Z5 0.133 0.389 (0.318) (0.342) [0.0331 10.0781 HHH8RS 6.92 . -0.017 0.132* (0.073) (0.078) 1-0.0041 10.0261 DEPRATIO 0.767 0.533 0.639 (0.951) (1.046) [O.1331 (0.128) DUMMY ISS 0.329 0.351 -0.6768 (0.282) (0.329) 10.0881 (-0.135] NETWORK 7.26 -0.029 -0.025 (0.038) (0.040) [-0.0071 [-0.005] MALEMBRS 3.40 -0.023 -0.205' (0.110) (0.123) [-0.006] [-0.041] LBRUTILZ 0.589 0.608 -1.561' (0.482) (0.529) 10.1521 1-0.3121 OSTNC. 5YWP Q.233 0.755' .0.534 (0.340) (0.414) [0.1891 [-0.107] OWNH20 0.365 0.834* -0.351 (0.292) (0.345) (0.209] (-0.070] PRIORITY 0.875 -0.169 0.104 (0.396) (0.442) [-0.042) 10.021] Mean of Depeodent Variable 0.44 0.27 Number of Cases 304 303 Predicted Correctly 66.8 74.3 Log Likelihood Ratio (LLR) - At 0 417.2 355.8 - At Convergence 365.5 318.2 -Significant at 95: level, one tailed tests. a/ Standard error in parenthesis. bt Transfor-ation of the unbracketed logit coefficient to a probability eatiwate in square brackets. TranSfor=atioa was done at tbe eean of the dependent variable p by the formula-, e7 e dx c/ lMeaning of variables is axplained ins Dictionary of Variables. t , , 0