Development Discussion Papers

Transcription

1 Development Discussion Papers Housing Finance and Low-Income Urban Households in Indonesia: The Recent Experience of Two State Banks Blane D. Lewis Development Discussion Paper No. 497 October 1994 Copyright 1994 Blane D. Lewis and President and Fellows of Harvard College Harvard Institute for International Development HARVARD UNIVERSITY

2 HIID Development Discussion Paper no. 497 Housing Finance and Low-Income Urban Households in Indonesia: The Recent Experience of Two State Banks Blane D. Lewis Abstract The supply of housing finance in Indonesia, especially to low-income groups, has been rather insignificant compared to other developing countries in Asia. This paper reviews some empirical evidence concerning the recent efforts of two state banks, the National Mortgage Bank (BTN) and the Citizens Saving Bank (BRI), to augment the supply of housing credit to low-income urban households. The paper shows that while neither bank has achieved much success at targeting low-income groups per se, BRI has made significant strides in channeling finance to households without traditional forms of collateral and has also obtained good results in getting its unsubsidized loans repaid. The paper makes some suggestions for future efforts in shelter-related lending to low-income urban households in Indonesia. Blane D. Lewis worked on the Indonesia Urban Project from 1992 to He is currently employed as a private consultant.

3 HIID Development Discussion Paper no. 497 Housing Finance and Low-Income Urban Households in Indonesia: The Recent Experience of Two State Banks Blane D. Lewis Introduction It is well known that the availability of housing credit is an important factor in realizing housing demand in developing countries (Struyk and Turner, 1986; Renaud, 1989, World Bank, 1993). The supply of such credit, though, varies significantly across countries (Buckley, 1988). The amount of lending for housing in Indonesia, as a proportion of total housing investment, is quite low relative to other developing countries in Asia at similar levels of income (Struyk et. al., 1990). The supply of housing finance to low-income households in Indonesia is particularly low. It has been estimated, for example, that low-income families, which comprise approximately one-quarter of the total number of households in Indonesia, receive only about thirteen percent of the total amount of housing credit from all sources and only approximately six percent of the total lending from formal-sector sources (Struyk et. al., 1990). The major source of formal sector lending for housing in Indonesia is the government's National Mortgage Bank (BTN). BTN has been making home-ownership loans to what it defines as both low- and middleincome households since the mid- 1970s. BTN has, however, been much criticized for its performance in lending to low-income families in urban Indonesia. Among the criticisms levied against BTN, include charges that it has increasingly financed units inappropriate to the needs and financial conditions of low-income residents and that the beneficiaries of its quite highly subsidized loans have largely been relatively high-income households (World Bank, 1986; Struyk et. al., 1990). Many suggestions have been 1

4 made over the years regarding ways to enhance BTN's performance in extending housing credits to low-income households. And indeed the government of Indonesia has implemented some needed changes in such lending. The extent to which improvements vis-a-vis the above criticisms have been realized is an open question, however. Potential improvements in its operations notwithstanding, BTN does not now, nor will it ever, have the capacity by itself to satisfy the large and growing demand of low-income households for housing credit. There exists, therefore, significant opportunity for other institutions to assist in augmenting the supply of housing finance to low-income families in Indonesia. The Citizens' Bank of Indonesia (BRI) has often been suggested as one of the few formal sector institutions that might be able to play a useful role in increasing housing credit to low-income households (Dunham et. al., 1989; Struyk et. al., 1990). The sanguine attitude about BRI's potential derives from its successful track record in other lending sectors. More specifically, it has been argued that BRI effectively extends loans which hold terms and conditions relatively more appropriate to the needs and abilities of low-income households, that it possesses an unmatched capability to reach large groups of low-income families through its extensive branch network, and that it has achieved good results in getting unsubsidized loans repaid (Dunham et. al., 1989). BRI has, in fact, recently begun to extend housing loans and by now has accumulated enough experience for an initial evaluation. The general intent of this paper is to consider some of the recent experience in formal-sector housing lending to low-income households in Indonesia. More particularly, the paper has two objectives. The first goal is to review and evaluate some of the current evidence regarding BTN's housing loans to lowincome households. And the second purpose is to present and examine some empirical evidence concerning BRI's recently started housing loan program. The paper proceeds with the BTN evaluation first and then 2

5 continues with the investigation of BRI's experience. The paper concludes with a summary of the main points and some suggestions for future efforts in low-income household lending for housing in Indonesia. BTN Housing Finance and Low-Income Households As noted above, the effectiveness of BTN's mortgage finance programs dealing with low-income households has often been questioned over the years. Two of the most frequently voiced concerns relate to the kinds of units that BTN has financed and the targeting of the housing finance to aid in the purchase of those units. The World Bank (1986), for example, claimed that BTN's lending during Repelita III and 1 the early years of Repelita IV moved increasingly away from financing the supply of smaller, less expensive houses towards providing for larger, more expensive (and more profitable) homes. This was apparently, at least in part, due to the increasing role that private developers began to play in constructing BTN-financed houses. Initially, BTN provided finance for houses built by the National Housing Development Corporation (Perumnas) only. In the late 1970s, however, BTN began to finance homes built by private developers also. In recent years, the private sector has begun to develop an increasingly 2 larger share of the units financed by BTN. As a result of the shift towards building houses that were larger and more costly, beneficiaries tended to be drawn more and more from higher income households, the Bank noted. In addition, the Bank asserted that civil servants and military personnel benefitted disproportionately, to the obvious detriment of other urban households, from BTN's lending schemes. 1 Repelitas III and IV are the Government of Indonesia's five year development plans for 1979/80 through 1983/84 and 1984/85 through 1988/89, respectively. 2 Struyk et. al., for example, report that from 1979 to 1988 private developers increased their share of BTNfinanced production from 55 percent to 88 percent. 3

6 In the mid-1980s, the Indonesian government and BTN initiated some policy and programmatic changes that were designed, at least in part, to overcome the above mentioned targeting problems (Struyk et. al. 1990). In particular, the government began to allow private developers to construct units smaller than 36 square meters (which previously only Perumnas was allowed to do), thereby hoping to increase the supply of houses affordable to low-income households. In addition, the government newly regulated that at least 70 percent of all houses financed by BTN must be 36 square meters or smaller. Also, in an attempt to concentrate subsidized funds on those really in need, BTN fixed an upper bound on applicant income of Rp. 300,000 per month. Finally, BTN officially discontinued the practice of giving preferential treatment to civil servants and the military in the allocation of housing finance. Struyk and colleagues (1990) report that BTN did, in fact, realize some progress towards reorienting its lending towards smaller houses after the reforms were introduced. In 1987 and 1988, for example, over 80 percent of aggregate BTN lending was allocated to the production of units 36 square meters or smaller. During the previous year, prior to the policy changes, only about 60 percent of the units constructed with BTN finance were within that size range. On the other hand, the above authors note that, as of 1989, a substantial share of BTN's subsidies continued to be focused on households with relatively high incomes and/or on civil servants and military personnel. Consider Table 1 below which provides some characteristics of both BTN borrowers/homeowners and all homeowners in urban Indonesia. As the table shows, BTN housing loan recipients were apparently still, in general, wealthier and much more likely to be employed by the government than the typical urban homeowner in Indonesia. 4

7 Table 1: Household Characteristics: BTN Borrowers vs. All Urban Homeowners, 1988 BTN/ BTN/ All Urban Perumnas Private Homeowners Median HH Income (Rp./Mo.) 235, , ,000 Median Value of Assets (Rp.) 250, , ,000 Average HH Size Percent Government Employees Source: Struyk et. al. (1990) What has been BTN's more recent experience vis-a-vis the supply and demand of houses financed with 3 its money? Currently, most BTN finance goes towards the production of three main types of houses: core, 4 low-cost, and medium-cost. Core houses are the smallest units, with a maximum size of 21 square meters. 5 They are designed for households with incomes less than Rp. 450,000 per month. Core unit loans are set at a maximum size of Rp. 8.5 million and carry interest rates of 12 percent. Low-cost houses are those that are smaller than 70 square meters and are intended for households with income less than Rp. 900,000 per month; their maximum size is Rp. 20 million and they carry a 17 percent annual rate of interest. Mediumcost units have no restrictions vis-a-vis size or household income. The upper bound on the loan size for the latter units is Rp. 300 million and the interest charged is 18 percent. All loans described above are for a 20 year term. 3 The information provided here is from BTN (1993). 4 BTN has also begun to make loans for serviced lots, shop houses, home improvement, and rental housing construction. These efforts are, however, still in the beginning stages of development. 5 Here, "household income" refers to "applicant and spouse income". 5

8 Funding levels for the various types of houses over the past several years are shown in Table 2 below. As can be seen from the table, total BTN funding rose substantially from its initial levels and peaked in 1988/89; it has declined steadily thereafter. The overall drop in recent years is mostly a function of decreases in funding for low-cost housing, which has declined precipitously since 1988/89. Medium-cost unit financing originated in 1990/91 and increased slightly the next year, although not enough to offset decreases in loans for low-cost homes. Real levels of core-house financing have remained essentially constant since declining from their peak in 1988/89. As the table shows, financing for core units has consistently increased in relative terms since 1986/87 (with the exception of the slight drop in 1989/90) and now stands at just above 45 percent of total disbursements. Table 2: BTN Loan Disbursements 1986/87 to 1991/92 (Constant Rupiah, Millions) House Type 1986/ / / / / /92 Core 18,215 64, , , , ,964 Low-Cost 99, , , , ,991 73,499 Medium-Cost ,509 54,310 Total 118, , , , , ,773 % Low-Income Source: BTN Monthly Report, March

9 It appears therefore, that recently an increasing share of BTN housing finance has, indeed, been going to enhance the supply of the smallest units, that is, those houses that are designed for low-income households. A separate and important question concerns whether those units have, in fact, been purchased by low-income households. As shown above, an increasing relative supply of small houses does not necessarily mean that low-income households are increasingly gaining access to those units. Recent and comprehensive data on the household characteristics of BTN core house buyers are unfortunately unavailable. However, some partial information garnered from BTN concerning the recipients of core unit loans is available and may help to shed some light on the question. Consider Table 3 below, which provides the median household income and employment status of BTN core unit borrowers in Jabotabek for October and November of Table 3: Household Characteristics of BTN Core Unit Loan Recipients in Jabotabek, Oct./Nov., 1993 Median Household Income (Rp./Mo.) 336,000 Percent Government Employees 63.2 Source: BTN The table shows that, during the period concerned, the median monthly household income of Jabotabek core-unit buyers was Rp. 336,000. The term "household" here refers to "applicant and spouse", according to official BTN literature. Privately however, BTN officials acknowledge that, much of the time, loan applicants do not admit of any secondary income (either from the household head or from the spouse); the intended effect, of course, is for the household to remain under the income ceiling and thus qualify for the subsidized funds. To the extent that secondary income is not, in fact, included, then the 7

10 figure shown above substantially underestimates the total amount of the applicant's household income. Struyk and colleagues (1989), for example, have estimated that more than 40 percent of total household income may be derived from sources other than the primary earnings of the head of household. If this is true, then the income figure reported above may be closer to Rp. 560,000. Regardless of the exact actual figure, it seems clear that the household income of the BTN borrowers is probably significantly greater than the income of the typical homeowner household in Jabotabek. The table also shows that 63.2 percent of the total number of BTN borrowers were employed by the government. This is clearly higher than the comparable figure for urban Jabotabek as a whole. As noted above, in 1988, just 18.7 percent of all urban homeowners were employed by the government. The current figure for Jabotabek is unlikely to be significantly greater than that estimate. It is important to emphasize that the above data on BTN borrowers were not generated from a random sample of all such borrowers. Caution should be used therefore in drawing inferences about the entire set of BTN core-unit loan recipients and in making comparisons between those borrowers and the general urban Jabotabek homeowner population. The present data do, at least, seem to suggest, however, that the beneficiaries of BTN loans for the smallest units may still be somewhat richer and more likely to be employed by the government than the typical urban homeowner in Jabotabek. That is, BTN financing continues, most likely, to be poorly targeted to low-income households. Recent BTN experience is, therefore, mixed at best. On the one hand, an increasing share of BTN lending does appear to be going to augment the supply of the smallest, most affordable units. On the other hand, and more importantly, the actual beneficiaries of that lending apparently continue, for the most part, to be relatively high-income households and/or those that are employed by the government. Indications are, therefore, that low-income urban household demand for housing credit is not being met. 8

11 And it could not possibly be met by BTN alone, of course. It has been estimated (Struyk et. al., 1990) that by 1997 total demand for housing investment by those purchasing the smallest units will reach a staggering Rp. 4,875 billion (in 1987 Rupiah terms). The difference between the latter figure and current levels of BTN funding reported above is indeed significant. Given the enormity of the excess demand of low-income households for housing finance, it appears as if there is ample opportunity for other lenders to become involved in the market. BRI has often been suggested as the only such lender from the formal financial sector that could possibly be expected to play a useful role in increasing housing credit to low-income households (Dunham et. al., 1989). As noted at the outset, BRI has had a long and fairly successful experience with lending to relatively low-income households throughout Indonesia. And BRI has, in fact, recently begun to extend credits for housing activity. Below, some empirical evidence is presented concerning BRI's early experience with shelterrelated lending. The evidence is derived from a sample survey of the housing loans extended by BRI's units in Jabotabek. 6 BRI Housing Finance: Initial Empirical Evidence The BRI housing lending program examined below began operations early in fiscal year 1990/91. Since its inception, the BRI program has made an estimated 8,300 housing loans (in the Jabotabek area 6 More specifically, the following section is based on an analysis of some 2200 BRI housing loans. These 2200 loans represent the entire set of all such loans extended during the period from June 1990 (when the program started) to July 1993 by a group of 54 BRI unit kotas. The latter were randomly selected from the collection of all 204 such units in the Jabotabek area. Bank personnel from each of the chosen 54 units were asked to provide information about both individual loans and borrowers. The requested data were derived from the actual loan applications and were recorded on a survey form mailed to the BRI units. The information was collected and processed between August and November

12 7 only), amounting to over 16 billion Rupiah. The BRI housing program is rather small, therefore, at least compared to the BTN effort. Over the last three years for which data exist, BTN has extended over 95,000 loans for core houses alone. And these loans amount to nearly 425 billion Rupiah (in current terms). Although BRI housing lending is currently relatively small, if it turns out that its lending is more appropriate to informal sector operations, then its approach could serve as a model for other similar efforts, thus further augmenting supplies of housing capital. Housing operations in the informal sector are, in general, very small in scale and carried out in an incremental fashion without fully registered title to the underlying property. Loans, it is argued, should therefore be made for the purpose of upgrading existing dwellings and should be small in size, short-term, and should not require formal collateral. In addition, it has been argued that the loans should carry an interest rate that at least approaches the market rate so that costs can be covered and sufficient funds can be generated and sustained over the long-term (Dunham et. al., 1989). Below, the nature of housing loans made under the BRI credit program is described. Table 4 below provides some information on the characteristics of BRI housing loans. More specifically, the table shows, in panels A, B, C, and D, the distribution of BRI loans by use, size, term, and collateral requirements, respectively. As can be seen from the table, the vast bulk of BRI loans is used to upgrade existing dwellings. Over 82 percent of the total number of loans are used to upgrade existing units for the owner's personal use and another 14 percent are utilized to upgrade existing housing to rent out. As the table shows, only a very small percentage of the total loans are used to construct new housing, either for own use (1.23 percent) or to rent out (2.14 percent). 7 These estimates are based on the survey results. 10

13 Apparently, BRI housing loans are, in general, quite small. Panel B of the table provides the complete size distribution of loans. As can be seen, almost 40 percent of the loans are for less than 1.0 million Rupiah and just over 60 percent are for less than 1.5 million Rupiah (the median size loan). Approximately 21 percent of the loans are for amounts greater than 2.0 million Rupiah and less than 10 percent are for sums larger than 3.0 million Rupiah. 11

14 Table 4: BRI Housing Loan Characteristics Panel A: Loans by Use Panel B: Loans by Size Use Percentage Size Category Percentage Upgrade for Own Use 82.3 < Upgrade to Rent to < Construct for Own Use to < Construct to Rent to < to < > Panel C: Loans by Term Panel D: Loans by Collateral Term (Months) Percentage Collateral Type Percentage < Certified Land Title Uncertified Land Title Non-Fixed Assets Letter of Employment Other Source: BRI Housing Loan Survey Loan maturities are short for BRI housing loans. Over 12 percent of the loans are for a period of one year or less. Almost 75 percent of the loans are for a term of 24 months (the 12

15 median size) and another one-quarter are for periods of less than 24 months. Only just over three percent of the loans are for periods longer than 24 months. Finally, as can be seen from the table, collateral requirements for BRI housing credits are rather less stringent than those of other banks. Over 40 percent of the loan recipients possessed only a letter from their employers as collateral. An additional 33 percent of the loan recipients used proof of ownership of non-fixed assets (such as a motorcycle) as collateral. Only 2.5 percent of the loan recipients offered certified, legal title to land as security for the loan while 12 percent offered uncertified title as collateral. Based on the above, BRI housing lending certainly appears to be appropriate to the needs of those residents in the informal housing sector. Loans are, in large measure, employed for upgrading existing homes and loan sizes are small and maturities are short. Most importantly perhaps, the collateral requirements of BRI lending appear to be suitable for the land ownership status of most low-income households. In addition, the rate of interest charged, 32 percent, corresponds closely to the market rate and seems adequate to cover bank operating costs and default risks and ensure a sufficient and sustainable pool of funds. The congruous nature of BRI loans does not mean, however, that the credits are, in fact, being extended to low-income urban residents. Table 5 below provides some data on the income, household size, and employment status of BRI borrowers. As the table shows, the median monthly income of BRI loan recipients was Rp. 325,000. Note that this figure represents applicant income only and not household income. As such, it appears that the income level of BRI borrowers is not particularly different from that of those who received core-unit financing from BTN, as estimated above. 13

16 Table 5: Household Characteristics of BRI Borrowers Median Applicant Income (Rp./Mo) 325,000 Average Household Size 4.9 Percent Government Employees 72.7 Source: BRI Housing Loan Survey The table also reports that 72.7 percent of the borrowers were employed by the government. The comparable figure reported above for BTN borrowers was 68.7 percent. In general therefore, the table shows, somewhat surprisingly, that BRI loan recipients do not appear to be very much different from BTN loan recipients, at least in terms of income and occupation of head of household. Moreover, it seems rather certain that both BRI and BTN loan recipients are better off financially and more likely to be employed as civil servants than the typical urban homeowner. BRI loan recipients are, however, significantly different from BTN borrowers in at least one important regard: loan repayment. The BRI survey data showed that only approximately 6.5 percent of active BRI housing loan recipients were in arrears on loan repayment. On the other hand, Struyk and coauthors (1989) report that almost 50% of BTN loans were in arrears, as of Moreover, nearly 30 percent of all outstanding BTN loans, amounting to approximately Rp billion, were over three months in arrears as of that time, according to Struyk and colleagues. Summary, Conclusions, and Recommendations BTN has achieved mixed results, at best, in its recent efforts to extend housing finance to lowincome residents. It is true that an increasing share of the total volume of BTN lending is apparently 14

17 being allocated to augment the supply of the smallest housing units. Unfortunately, it is also apparently true that the bulk of BTN's highly subsidized funds continues to accrue to relatively high-income households and/or to government employees. In addition, BTN's recent track record concerning loan repayment has been less than auspicious, to say the least BRI's initial experience with housing sector lending has also been somewhat uneven. On the one hand, the attributes of BRI's housing loans do appear to be relatively more suitable to informal sector operations. As was shown above, the purpose to which BRI housing loans are put and the size, term, and collateral requirements of the loans appear to correspond well to low-income household needs and abilities. On the other hand, however, BRI also seems to have been less than effective in targeting loans to low-income households: recipients of BRI housing loans also appear to be relatively well-off financially compared to the average urban household and, in addition, are much more likely to be employed by the public sector than the typical urban resident. BRI might perhaps be able to improve its income targeting of housing finance if it placed some restrictions on the maximum size of its loans. Some analysts have suggested, for example, that a maximum loan size of what amounts to approximately Rp. 1.5 million (in current Rupiah terms) might not be inappropriate in the Indonesian context (Dunham et. al., 1989). As was noted above, this is currently the median size of BRI housing loans, as estimated from the sample survey. The sample data indicate that the median monthly income of borrowers receiving loans less than or equal to Rp. 1.5 million was Rp. 250,000. In addition, BRI might be able to better focus its finance on lower income households by improving certain of its loan application procedures. In this regard, it might attempt to estimate total household income rather than just applicant income, as is currently the case, and, in addition, place some upper bound on that income as a condition of application approval. Also, the sample survey indicates that 15

18 applicant income varies significantly by location (it is lowest for those residents from Bogor) and also by age (the younger the applicant, the lower the income); it might be possible therefore for BRI to improve its overall income targeting by channeling more funds to those households who live in relatively poorer areas and/or to relatively younger heads of households. Interestingly, applicant income is not significantly associated with gender; focusing loans on women, therefore (as is often suggested), may not do much to improve income targeting, regardless of its other possible merits. BRI has, however, demonstrated great effectiveness at serving another heretofore neglected and relatively disenfranchised group of urban households: those without access to typical forms of collateral with which to secure loans. As demonstrated above, households without certified title to land constitute, in fact, the vast majority of BRI housing credit recipients. It is important to recognize that this group of households would most likely not otherwise gain access to housing finance, at least from formal sector sources, if not for the BRI program. Other financial institutions should pay close attention to BRI's experience in this regard as they design new housing loan instruments. BRI's success at serving households without clear title to land, however, should not divert attention away from the need for the Indonesian government to improve its land titling operations. Although the National Land Agency (BPN) has recently embarked, with World Bank assistance, on a 25 year program for registering non-forest land parcels across all of Indonesia, more will need to be done in the short-term to help alleviate the special problems associated with unregistered land rights in urban areas. The now defunct PRONA program achieved considerable success in speeding up the large scale registration of land rights in many rural areas of the country in the 1980s (Hoffman and Marbun, 1990); the program could serve as a model for more immediate efforts in urban land title registration. While the land certification process is being implemented, measures might also be taken to require developers to 16

19 compensate holders of all valid land titles (that is, unregistered as well as registered titles) at the same rate as certified titles, should redevelopment occur. Finally, BRI has also achieved significant success in the area of loan repayment. It has clearly shown that it is, in fact, quite possible to provide unsubsidized loans to borrowers and get the loans repaid. This demonstrates, among other things, that housing sector loan subsidy programs like BTN's are, in addition to being expensive for the central government to maintain, probably also unnecessary. Most importantly perhaps, interest rate subsidies discourage the participation of other formal-sector housing lenders. If the government were to eliminate its subsidies of BTN, it is quite possible that other (heretofore crowded-out) lenders might find it profitable to enter the market, thereby expanding the supply of loanable funds for housing development. And an increased supply of housing credit would certainly be beneficial to potential Indonesian homeowners, in general. How much benefit low-income households would derive from the increased supply is an open question, however. The recent experiences of BTN and BRI show the extreme difficulty of reaching low-income urban households with formal housing credit and suggest that an intensification of efforts in this area is required. 17

20 References Bank Tabungan Negara. (1993). "BTN Monthly Report, Update." BTN, Jakarta, Indonesia. Bank Tabungan Negara. (1993). "BTN Housing Loans: Terms and Conditions." BTN, Jakarta, Indonesia. Buckley, Robert. (1988). "Housing Finance in Developing Countries: A Transactions Cost Approach." World Bank Draft Report. World Bank, Washington, D.C. Dunham, Constance, Tantri M.H. Marbun, and Tammy Kang. (1989). "Expanding the Supply of Housing Loans to Low-Income Urban Households in Indonesia." The Urban Institute and P.T. Hasfarm, Washington, D.C. and Jakarta, Indonesia. Hoffman, Michael and Tantri Marbun. (1990). "Unregistered Land Rights: Their Role in Improving Housing Quality in Indonesia." Urban Institute and P.T. Hasfarm, Washington, D.C. and Jakarta, Indonesia. Struyk, Raymond and Margaret Turner. (1986). Finance and Housing Quality in Two Developing Countries, Korea and the Philippines. University Press of America, New York. Struyk, Raymond, Michael Hoffman, and Harold Katsura. (1990). The Market for Shelter in Indonesian Cities. The Urban Institute, Washington, D.C. Renaud, Bertrand. (1989). "The Role of Housing Finance in Development: Issues and Policies." Transportation and Urban Development Department, World Bank, Washington, D.C. World Bank. (1993). Housing: Enabling Markets to Work. World Bank, Washington, D.C. World Bank. (1986). "Staff Appraisal Report: Indonesia Housing Sector Loan." World Bank, Washington, D.C. 18