World Development Vol. 29, No. 6, pp. 1057±1069, 2001 Ó 2001 Elsevier Science Ltd. All rights reserved Printed in Great Britain www.elsevier.com/locate/worlddev 0305-750X/01/$ - see front matter PII: S0305-750X01)00016-X Micro®nance Success Amidst Macroeconomic Failure: The Experience of Rakyat Duringthe East Asian Crisis

RICHARD H. PATTEN, JAY K. ROSENGARD and DON E. JOHNSTON, JR. * Harvard University, Cambridge, MA, USA Summary. Ð The -BRI) unit system is recognized as one of the largest and most successful micro®nance institutions in the world. Indonesia has been more drastically a€ected by the East Asian monetary crisis than other countries in the area. It is therefore worthwhile lookingat the BRI experience duringthe crisisÐnot only the experience in microenterprise credit, but also in small, medium and corporate credit and in savings mobilization. The comparative performance of di€erent parts of BRI during the East Asian crisis suggests essential features in the future design of sustainable micro®nance institutions, products, and delivery systems. Ó 2001 Elsevier Science Ltd. All rights reserved.

Key words Ð Asia, Indonesia, banking, ®nancial crisis, micro®nance, microenterprise

1. BACKGROUND ON BANK RAKYAT SBU Microbanking is in charge of BRI's INDONESIA -BRI) 1 3,694 units. A unit is a small bank oce with four to eleven sta€. The units also maintain 319 BRI is one of three state-owned commercial cash posts which are open three to ®ve days a foreign exchange in Indonesia. BRI is week to receive and pay out savings, and to owned entirely by the Government, but oper- receive loan repayments. Each unit is a separate ates under all of the prudential norms and pro®t/loss center with its own balance sheet and regulations that every commercial bank must pro®t and loss statement. The units make loans adhere to in Indonesia. BRI's primary mission of Rp 25,000 to Rp 25 million. There were 2.6 is the provision of rural and urban community million microloans outstandingtotalingRp 6.9 bankingservices by mobilizingfamily savings trillion at the end of July 2000. The units also and deliveringcredit products to medium, small provide savings, giro, and time deposit ac- and microenterprises. counts; there were 25.1 million accounts in the Although BRI was established in 1968 under units at the end of July 2000 totalingRp 18.5 National Law Number 21/1968, it is actually the trillion. 3 successor to a series of banks in Indonesia going SBU Retail Banking is in charge of BRI's back more than 100 years, beginning with De branches, which provide full bankingservices. Poerwokertosche Hulp-en Spaarbank der In- They make small business loans up to Rp 3 landsche Hoofden, founded in 1895. BRI grew billion, loans against salaries and pensions, to become one of Indonesia's largest banks, as and ``program loans'' -channeling of loans for indicated by its position early in the East Asian government schemes, such as those to coop- Crisis: as of December 1997, BRI's total assets eratives or to the government agency that were $16.7 billion, includinga net loan portfolio provides agricultural price support). The of $12.6 billion, and 1997 pre-tax pro®ts totaled branches also provide savings, giro and time $56.7 million. 2 It had 324 branches covering deposit accounts. The branches have recently the entire country, overseas oces in , introduced an on-line facility for savings ac- HongKongand New York, and a workforce of count holders. over 44,000 employees. Duringthe same year the East Asian Crisis hit, BRI re-organized into four strategic busi- ness units -SBUs). * Final revision accepted: 15 November 2000. 1057 1058 WORLD DEVELOPMENT

SBU Corporate makes large corporate loans This resulted in both risingprices and a -above Rp 3 billion), includingdollar-denomi- shortage of demand from farmers for the nated loans. products and services of micro and other lo- SBU Treasury and Investment handles trea- cal enterprises. sury functions, includingthe maintenance of -ii) For many years Indonesia has main- BRI primary reserves in and the tained an open capital account. The rupiah purchase of Bank Indonesia certi®cates -SBIs). has been allowed to depreciate about 5% a It also controls BRI ®nancial subsidiaries. year against a basket of currencies. High ru- piah interest rates compared to interest rates in neighboring countries, when coupled with the seemingcertainty of the rate of deprecia- 2. THE MONETARY CRISIS 4 tion, brought in large amounts of ``hot'' short-term . Once there was a loss of -a) Onset con®dence, this short-term money could and did move out very quickly, triggering The monetary crisis in Indonesia began -or the decision of Bank Indonesia to allow the was signaled) in mid-1997 when Bank Indone- market to set the exchange rate. sia, the , decided that its foreign -iii) The large Indonesia conglomerates bor- reserves were insucient to support the ex- rowed dollars abroad, runningup a private change rate, at that time Rp 2,450 to the US sector short-term foreign exchange debt more dollar. By mid-1998, the exchange rate had than twice the total foreign debt of the Gov- plunged to Rp 14,900 to the US dollar. As of ernment, most of which consists of long- the end of March 2000, the crisis appears to term concessional loans. Bank Indonesia have ended, with Bank Indonesia predictinga was unable to keep track of this private for- stable exchange rate and growth at a rate of 4± eign debt, which was more than twice the 5% for the second quarter of 2000. country's foreign exchange reserves, and There were at least three causes of the mone- was therefore uninformed of possible short- tary crisis which were relevant to BRI's situation: term demands against the country's foreign -i) There was a severe drought in the 1997±98 exchange reserves if these loans were not season, and one entire crop of rice was lost. rolled over.

Figure 1. In¯ation and one-month SBI rates. EXPERIENCE OF BANK RAKYAT 1059

A large amount of this debt was used for real -GOI) had not issued estate investment, most of which was not bonds that were traded in the market, Bank complete, much less sold or rented, at the Indonesia's main instrument time the crisis began. Even had the funds was the Serti®kat Bank Indonesia -SBI). The been used for ®nancingrelatively quick re- rate for a one-month SBI was raised to over turn activities, these borrowers still could 70% in July and August 1998. Apparently, this not have repaid at the new exchange rate; policy was considered a success; the one-month most are now technically insolvent. SBI rate has been lowered rapidly commencing Unlike the Latin American debt crisis in the in October 1998, and was at 10.91% at the end 1980s and the Mexican peso crisis in 1991, of March 2000 -see Figure 1). government overspending was not one of the The exchange rate has moved from Rp 2,450 root causes of the Indonesian crisis. Including to the US dollar before the crisis to over Rp the proceeds of long-term foreign aid loans on 14,900 at the time of the worst panic, and has the income side and loan repayments on the moved between Rp 6,500 and Rp 8,500 to the expenditure side, the budget had been balanced US dollar most of the time since mid-June or in surplus since the late 1960s. 1999. It improved against the dollar following relatively peaceful elections in May 1999; it -b) E€ects on interest rates deteriorates each time there is an outbreak of violence anywhere in the country, the appear- After Indonesia began to receive assistance ance of political uncertainty, or a delay in the from the International Monetary Fund -IMF), release of IMF funds. On April 12, 2000, the monetary policy for several months concen- exchange rate was Rp. 7,623 to the US dollar. trated on raisinginterest rates to support the BRI, alongwith other Indonesian-owned exchange rate and to control in¯ation. Since the banks, raised their time deposit rates in line

Figure 2. SBI, time deposit, and SIMPEDES interest rates. 1060 WORLD DEVELOPMENT with the SBI rate until May 1998, when they loans apply almost immediately, because the peaked at 60% for one-month time deposits. loan agreements provide for adjustable rates. Because of the large in¯ow of deposits, result- The on KUPEDES -``General ingin part from the closingof several banks, Rural Credit,'' the units' sole microcredit in- BRI was able to keep deposit rates well below strument) was raised from 31.72% before the the SBI rate until October 1998. From October, crisis to 45% on September 1, 1998. Since then, BRI has kept rates just below the SBI rate as the rate has been lowered in steps all the way this has moved down. As of March 2000, BRI back to 31.72%, e€ective since September 1, is paying10.5% per annum on a one-month 1999 -see Figure 3). time deposit, while the one-month SBI rate is New interest rates on KUPEDES apply only 11.91%. Rates on SIMPEDES -``Rural Sav- to new loans, while new rates on savings apply ings,'' a passbook account available only at immediately to all existingsavingsaccounts or BRI units) and other passbook savings in BRI to time deposits taken out or renewed after the have been relatively stable compared to time rate change. Thus, there is a lag in the increase deposit rates: they moved from 13% before the in actual credit interest rates and a squeezingof crisis, to a peak of 20%, and have now declined the spread as interest rates move up and an to 9% as of the end of March 2000 -see Figure increase in spread as interest rates move down 2). -see Figure 4). Minimum interest rates on most small, me- dium and large loans doubled from the 17% to -c) Bank liquidity and solvency 19% per annum range in mid-1997 to a high of 36% from September 1998 to May 1999. They At the start of the crisis, Bank Indonesia gave have since fallen steadily, reachinga level of an unlimited guarantee of all deposits in all 19% to 26% in March 2000. Rates on these banks. The result was provision of enormous

Figure 3. E€ective rates on KUPEDES and SME loans. EXPERIENCE OF BANK RAKYAT 1061

Figure 4. KUPEDES productivity and BRI units' cost of funds.

Bank Indonesia ``liquidity credits'' to many -a) SBU corporate banks to pay o€ their depositors. There were runs on several private banks; the two largest -i) Dollar loans were taken over by the Indonesian Bank Re- Because of the drastic devaluation, the ru- structuringAgency-IBRA). Between Novem- piah value of SBU Corporate dollar loans more ber 1, 1997 and March 13, 1999, 61 banks were than quadrupled, ballooningfrom about Rp closed. Another 11 banks were nationalized, to 3.5 trillion to roughly Rp 16 trillion. In the be consolidated before resale to the original balance sheet of BRI, this meant that capital owners or to the market. Four of the seven state adequacy declined sharply and one or two banks are beingmerged.All four survivingstate loans were more than the legal lending limit banks were scheduled to be recapitalized -20% of the bank's equity). As a result, BI's through the issuance of GOI bonds by Sep- ratingof BRI has gonedown. The rupiah value tember 1999, but this deadline was delayed for of these loans has again declined as the rupiah BRI and one other state bank until June 2000. 5 has appreciated. 6 Many SBU corporate loans were for hotels, which were good foreign exchange earners be- 3. CREDIT REPAYMENT EXPERIENCE fore the tourist trade declined. These hotels are OF BRI DURING THE MONETARY still experiencinglow occupancy rates, although CRISIS these rates have improved since June 1999. Even if there had not also been an increase in The credit repayment experience of BRI's interest rates on these loans, it is extremely Micro, Retail, and Corporate SBUs has not unlikely that they could have been repaid at Rp been uniform, as described below. 7,200 to the US dollar, still a near triplingof 1062 WORLD DEVELOPMENT their previous rupiah value, much less at the Rp There are, however, some problems with the 14,900 rate. At present, most of these loans are relatively small number of payroll deduction nonperforming. loans to employees of private businesses where employees have been laid o€. -ii) Rupiah loans Performance of SBU corporate's rupiah -iii) Program loans loans varies accordingto whether the corporate BRI channels loans for various government customer had also borrowed abroad in dollars, programs. Usually these are ®nanced by li- whether the loan proceeds were put into oce quidity credits from Bank Indonesia, and Bank construction or other activities with a delayed Indonesia and/or the Government carry most return, and whether the customer's business of the risk. Interest rates for the liquidity credits sold goods or services for which demand has that ®nance the programs, as well as interest decreased drastically. rates to the ®nal borrowers, are set by Bank Most of the corporate portfolio has been Indonesia or the Government. written o€ and transferred to the Asset Man- The performance of program loans varies agement Unit of IBRA. accordingto the objectives of the programand the quality of the intermediary between BRI -b) SBU retail banking and the ®nal borrower. The drought reduced repayment of credit to farmers passed through SBU retail lendingincludes several di€erent the Government-sponsored cooperatives. On loan categories, and performance has varied the other hand, it increased the size of credit amongthese categories. passed by BRI to BULOG, the government logistics agency, which has had to import much -i) Small and medium business loans more rice than usual to control prices; there has The small and medium business loans of never been any default on loans to BULOG, SBU retail are all rupiah loans. Most are on a which are 100% guaranteed by the Govern- line of credit basis, rather than requiringre- ment. payment in regular installments that include Followingthe drought,the Government principal. The retail loans have performed pushed a large amount of credit through the better than the large corporate loans, but about cooperatives for farmers, all with either a 95% 41% had missed at least one interest payment. or 100% Government guarantee. There was Retail borrowers face a combination of sharply clearly a need for increased farm credit after decliningdemand, doublingof interest rates, the farmers' savings had been drained away by and, in some cases more than a triplingof the the drought. The increase was, however, very rupiah price of imported inputs. As interest large, an increase of over 30 times from the rates have moved back to pre-crisis levels, re- previous year's total. This credit was due back payment performance has improved. from the cooperatives on March 31; data on After transfer to IBRA of loans considered how much was paid back are not yet com- as unlikely to be repaid, and repayment of some plete. loans as economic conditions have improved, the percentage of the remaining retail business portfolio in arrears is down to 20% as of the -c) SBU microbanking end of March 2000. The BRI units have a single credit instru- -ii) Consumer loans ment, KUPEDES -General Rural Credit). All An important business for BRI branches is KUPEDES loans require repayment in install- makinginstallment loans to salaried employees ments. Repayment schedules are available to ®t and pensioners with repayment through payroll the cash ¯ows of borrowers with various kinds deductions or deductions from pensions which of enterprises, whether agriculture or ®sh pro- are paid through BRI. These are classi®ed as duction, tradingof various kinds, handicraft or consumer loans. small industrial production, or services. KU- Most of the payroll deduction loans are to PEDES is also available for employees whose government employees. Repayment has held up repayment is from salary deduction; most such very well, although real wages have dropped. loans have been used for home improvement, The same is true of loans repaid through de- enterprises of family members, or education of duction from pensions paid through BRI. children. EXPERIENCE OF BANK RAKYAT 1063

-i) KUPEDES repayment experience through The long-term loss ratio -LTLR) shows re- March 2000 sults since KUPEDES lendingbeganin Feb- So far, KUPEDES borrowers have contin- ruary 1983. It is of course not sensitive to ued to pay back more than 97% of everything movements within a single one-month period. that has fallen due, throughout both the mon- The LTLR stood at 2.03% at the end of March etary crisis and drought. 2000. Each month, SBU Microbanking's manage- The twelve-month loss ratio -TMLR) shows ment information system produces three ``loss the same results for the past twelve months, ratios'' that measure the total amount overdue, and is most useful in detectingchangesin includingeverythingwhich has been written o€, trends. The TMLR was 1.72% at the end of divided by the total amount which has fallen December 1999, and declined again to 1.54% at due. the end of March 2000. Interestingly, this ratio The short-term loss ratio -STLR) is for the has declined fairly steadily duringthe crisis and particular month. It is strongly a€ected by the has been below 2% from November 1998 to major Muslim holiday of Lebaran, which was March 2000 -see Figure 5). in early January this year and in mid-January in 1999. In March, April, May, and again in -ii) Changes in size of the KUPEDES portfolio September and October 1999, the STLR was Total KUPEDES outstandingreached its below 1%, but increased in November and highest level in January 1998, when Lebaran December 1999 as it usually does just before occurred at the end of the month. A peak in Lebaran. It remained above 2% duringJanuary outstandingin the 30 days or so before Leba- 2000, but then dropped to 0.28% in February ran, followed by a decline thereafter, is normal. 2000 and 1.37% in March 2000. Traders borrow to stock up before Lebaran,

Figure 5. BRI unit system loss ratios. 1064 WORLD DEVELOPMENT when much of the entire year's purchase of credit portfolio increased by 33.9% in rupiahs clothingoccurs; they usually do not borrow just -to Rp 6.1 trillion) and by 4.0% in number of after Lebaran. From this peak in January 1998, borrowers -to 2.5 million). BRI's total KUPEDES outstandingwas rela- tively stable for a year and a half, not sur- -iii) Savings in the units passingthe previous peak until June 1999; prior Duringthe worst of the monetary crisis, there to the crisis, there was normally an increase in was a very rapid increase in savings, both in the loans outstandingfrom year to year. BRI units and in the rest of BRI. In the units, The levelingo€ of loans outstandingis not total savings more than doubled, increasing the result of any policy instruction from BRI from Rp 8.3 trillion at the end of October 1997 policy makers. BRI has not restricted KU- to Rp 17.9 trillion at the end of October 1999. PEDES lendingduringthis crisis. This is in Total savings declined slightly to the end of contrast to their actions in a somewhat similar, December 1999, but have begun rising again in though milder situation, in 1991, during which 2000. The large increase in savings through time BRI placed restrictions on unit lending mid-1999 was partly a ¯ight to safety, even and loan arrears shot up. The di€erent reaction though all deposits at all banks had been duringthe current crisis is partly because of the guaranteed by Bank Indonesia. Many private high in¯ow of savings in 1998 and 1999, as will banks experienced runs and were closed or be discussed below. taken over by IBRA. Even with the central Nor was the small decline and levelingo€ of bank guarantee, savers appeared to want to KUPEDES outstandingduringthe early part avoid the procedures which would impede of 1998 caused by a rise in the KUPEDES in- withdrawal of their savings when a bank was terest rate. BRI was deliberately slow in raising beingrestructured or closed. the KUPEDES rate as the cost of funds in- There was also a change in composition of creased, instead acceptingdecreased interest savings at the units. In response to monetary spreads because of the psychological impor- policy, the rate on time deposits in BRI jumped tance to KUPEDES borrowers of stability. The from 19% for a one-month time deposit in rate for all new KUPEDES loans was raised January 1998 to 57% in September 1998. The from 1.5% to 2.2% per month ¯at -on the top rate paid for SIMPEDES -rural savings), original balance) only from September 1, 1998, the main unit passbook type of savings instru- when the squeeze on the units' interest rate ment, moved up from 16% to 20% duringthe margin, coupled with no foreseeable widening same period. While the rupiah total of pass- of this margin, became insupportable. It was book savings remained relatively stable, time cut back to 1.95% per month ¯at on April 1, deposits grew from 12.6% of savings in the 1999, and then to 1.65% ¯at on July 1, 1999, in units in February 1998 to 40.5 percent in April response to the overall downturn in market 1999. From June 1999 to March 2000, total interest rates. It returned to 1.5% ¯at on Sep- savings remained stable, but there was some tember 1, 1999. shift from time deposits to passbook savings The main reason for the lack of an increase in accounts with the drop of time deposit interest the KUPEDES portfolio duringthis period rates. Time deposits constituted 21.0% of total appears to have been lower demand from bor- rupiah deposits in the units at the end of March rowers who felt uncertain about their business 2000. activities in the face of in¯ation of roughly 80% Interestingly, the rupiah value of passbook a year that prevailed from January to Septem- savings continued to grow to August 1998, fell ber 1998. Such in¯ation had not been experi- slightly through January 1999, and then began enced in Indonesia since the 1960s. Unit loan to rise again, returning to the August 1998 level ocers and managers reported that some of in May 1999. Since then, the value of passbook their good ``repeat'' borrowers' reaction was to savings has continued to increase at a fairly postpone borrowingagain,and instead to run steady rate. Duringthe 12 months to the end of down their stocks and place some of their March 2000, the total amount in passbook capital in savings at the unit. savings increased 37.4%, from Rp 9.9 trillion to Borrowing began to increase again in April Rp 13.6 trillion. 1999, about six months after prices began to The number of passbook savings accounts stabilize, as borrowers began to try to reach grew throughout this period with the exception previous levels of turnover in real terms. In the of one month, December 1998. For the 12 12-month period to March 2000, BRI's micro- months to March 2000, the total number of EXPERIENCE OF BANK RAKYAT 1065

Figure 6. Savings and credit at BRI units 3rupiahs). passbook savings accounts grew by 2.4 million, leverage, thereby making their enterprises from 21.7 million to 24.1 million. -see Figure 6 less vulnerable ®nancially to external distur- and 7). bances such as a sudden increase in interest rates caused by a shift in monetary policy or a sudden decrease in demand from cus- 4. REASONS FOR STRONG tomers who have missed a crop or become MICROENTERPRISE REPAYMENT unemployed. PERFORMANCE 7 -b) The microenterprises are more likely to be engaged in the purchase and sale of domesti- There appear to be at least four factors that cally-produced essentials, so that they face explain the excellent repayment performance of less cyclical demand than do many of the BRI's microenterprise borrowers, as compared small, medium and large enterprises which to small, medium and large enterprise borrow- deal in less essential goods and services and ers: in goods with higher imported content. As -a) The microenterprise loans are all install- their real income has declined duringthe re- ment loans adjusted to the borrower's cash cession, and prices of goods with high im- ¯ow. As borrowers pay back the install- ported content have shot up, consumers ments, they normally reinvest pro®t to main- have moved ``down market'' to meet their tain the level of business that they had daily needs, substitutingthe domestically- reached with the loan proceeds. As a result, produced goods handled by microenterprises most of these microenterprise owners have for the luxury and/or imported goods han- built up their equity and lowered their loan dled by larger enterprises. 1066 WORLD DEVELOPMENT

Figure 7. Savings and credit at BRI units 3number of accounts).

-c) The rural sector may be less a€ected by The result was an increase in the number the monetary crisis than the urban areas, of borrowers who did not pay on time. This but more a€ected by a severe drought that appears to have been a reaction to BRI not struck Indonesia at about the same time. honoringits end of the units' ``contract'' The e€ect of the drought on farmers' de- with microenterprise borrowers to provide mand for other microenterprise services access to credit as longas the borrower meet and products has been partially overcome all of BRI loan terms and conditions. by two reasonably good rice crops after the drought. Areas that raise export crops have experienced somethingof a boom because 5. FUTURE CHALLENGES of the devaluation of the rupiah. -d) The microenterprise borrowers appear to While the BRI units have performed well value their access to credit and savings ser- duringthe past two years of economic turmoil, vices very highly. They are reluctant to break diculties in the corporate and retail banking their bankingrelationship with the BRI operations of BRI could indirectly constrain units even if they have to squeeze consump- e€orts to enhance BRI micro®nance outreach tion to make their loan repayments on time. and coverage. There appears to still be unmet BRI has been careful to keep the microcredit e€ective demand for micro®nance services in window open so that those who pay on time Indonesia, and thus, ample room to improve are able to borrow again if their enterprise BRI micro®nance products and delivery sys- justi®es it. BRI learned from an earlier expe- tems. rience when this window was partially closed Three speci®c challenges in the expansion of for fear that savings were going to decline. BRI micro®nance activities are: EXPERIENCE OF BANK RAKYAT 1067

-a) Sta€ shortagesÐThe BRI units are short but in fact, really works against the primary of sta€, particularly mantris -loan ocers). policy objective of makingBRI more pro®t- Recent estimates place this stangde®cit able. at more than 1,000 mantris, excludingsta€ that would be necessary to open new units. In normal times, BRI would proceed to hire 6. CONCLUSIONS new sta€ for the units at the teller and book- keeper level, and promote some current em- The performance of the BRI units duringthe ployees to the mantri position. But as part of East Asian crisis demonstrates quite dramati- BRI's overall recapitalization program, cally essential features in the design of a sus- 6,000 BRI employees have recently been re- tainable micro®nance institution. tired early, mainly from BRI branches and the BRI head oce. It has so far proven dif- -a) Ability to repay ®cult to hire new unit sta€ while at the same time retiringBRI employees at the branch -i) Loan products that are compatible with a level and above. Based on past experience, microenterprise's cash ¯ow to facilitate loan the transfer of employees to the units from repayment, and that require pay-down of elsewhere in BRI is seldom an option, as loan principal in scheduled installments, the skills and temperament required to work thereby encouraging reinvestment of pro®ts in the units are quite di€erent from those into the business to decrease loan leverage needed elsewhere in BRI. and business vulnerability. -b) Lack of investment fundsÐEven though -ii) Loan delivery systems that are adapted the BRI units are pro®table, BRI's overall to local market conditions and clientele, in losses in 1998 mean that there is no invest- order to identify market opportunities, cope ment budget to replace aging motorcycles with market failures, and distinguish fairly and computers at existingunits, or to pur- cases of customer inability to repay versus chase equipment for new units. Foreign aid unwillingness to repay. is generally not available for this purpose both because the units are generating a prof- it, and due to the general restrictions that -b) Willingness to repay many donors place on the use of foreign aid for commodity procurement. -i) Development of a long-term banking re- -c) Need for permission to open new unitsÐ lationship with borrowers, characterized by Over 200 BRI unit ``cash posts'' have mutual respect and adherence to pre-agreed reached the level of business volume at loan terms and conditions. which they would normally be upgraded to -ii) Unrationed future availability of credit a full unit. Until recently, however, the Min- for present borrowers who repay their loans istry of Finance prohibited the openingof on time, and for new applicants who are new bank oces ``below the branch level'' deemed to be creditworthy. unless the entire bank was rated ``sehat'' -healthy). While the units are clearly healthy, -c) Willingness to save the BRI as a whole is not rated healthy by BI. The responsibility for decidingwhether -i) Savings products that o€er security, ac- new bank oces below the branch level cess, and a fair return, giving customers may be opened has been transferred to BI, both a ®nancial cushion duringhard times, but BI has yet to rule in the case of BRI and an incentive to keep their funds in the units. bank duringperiods of economic uncer- BRI is now evaluatingoptions to deal tainty. with each of these challenges, as it tries to -ii) Products that di€erentiate between ``sav- reconcile the business needs of the units with ings'' and ``investments,'' i.e., passbook ac- the appearance of inconsistency in policy counts versus time deposits. Savings tend implementation. Ironically, application of to be relatively stable and long-term, and generally sound regulations to the special are used primarily for lumpy payments or circumstances of the units is counterproduc- family emergencies; investments are usually tive; strict adherence to these rules might ap- more volatile, speculative, and ``hot,'' and pear to promote equal and equitable treatment, chase the highest short-term return. 1068 WORLD DEVELOPMENT

NOTES

1. See Patten and Rosengard -1991) for a comprehen- 4. See Radelet and Sachs -1998a,b) for an analysis of sive history of microbankingat BRI. Updates on BRI the causes and consequences of the East Asian microbankingactivities can be found in Boomgardand ®nancial crisis. McGuire -1998) examines implications Angell -1994) and Hook -1995). of the East Asian ®nancial crisis for micro®nance. Levinsohn, Berry, and Friedman -1999) look at the 2. These ®gures all use the pre-crisis mid-1997 ex- impact of the crisis on the poor. Djiwandono -1998) change rate of $1.00 ˆ Rp. 2,450. examines linkages between macroeconomic manage- ment and banking systems in general in light of the 3. Usually a paper such as this would state some East Asian ®nancial crisis, while Franks -2000) ex- information such as outstandingloan volume or total plores linkages between macroeconomic stabilization savings in both rupiahs and US dollar terms, so that and micro®nance. readers could compare these with micro®nance opera- tions which they know in other countries. Duringthe 5. Baker -1997) presents a vivid description of the crisis period, however, the exchange rate has been impact of nonperformingloans on bank soundness extremely volatile, and the meaningof changesin the duringthis period. dollar value of these volumes is unclear. For example, the dollar value of the units' KUPEDES loan portfolio 6. See Bank Rakyat Indonesia -1997, 1998) for a more has swungfrom $1.7 billion in June 1997, before the detailed look at BRI's overall ®nancial position. crisis, to $308 million in June 1998, to $724 million at the end of June 1999. BRI unit customers deal almost 7. These ®ndings are consistent with an extensive exclusively in rupiahs and duringthe period, and as bibliography of micro®nance best practices. For exam- explained in the text and shown in Figures 6 and 7, there ple, see Adams, Graham, and Von Pischke -1984), Von was almost no change in the rupiah loan volume or the Pischke -1991), Chaves and Gonzalez-Vega -1996) and number of KUPEDES borrowers. Therefore, this paper Morduch -1999). states all values in rupiahs and not in dollars, except when citing some general background statistics for BRI as a whole.

REFERENCES

Adams, D., Graham, D., & Von Pischke, J. -Eds.) Hook, R. -1995). The Experience of Bank Rakyat -1984). Undermining rural development with cheap Indonesia. In E. Brugger, & S. Rajapatirana -Eds.), credit. Boulder, CO: Westview Press. New perspectives on ®nancing small business in Baker, S. -1997). Loan portfolios in ¯ames: valuing banks developing countries -pp. 111±121). San Francisco, with rising NPLs. , Indonesia: PT Jardine CA: ICS Press for International Center for Eco- FlemingNusantara. nomic Growth and Fundacion Para el Desarrollo Bank Rakyat Indonesia. -1997). 1997 Annual Report. Sostenible. Jakarta, Indonesia: BRI. Levinsohn, J., Berry, S., & Friedman, J. -1999). Impacts Bank Rakyat Indonesia. -1998). 1998 Annual Report. of the Indonesian economic crisis: Price changes and Jakarta, Indonesia: BRI. the poor. NBER WorkingPaper No. 7194. Cam- Boomgard, J., & Angell, K. -1994). Bank Rakyat bridge, MA: National Bureau of Economic Re- Indonesia's Unit Desa system: achievements and search. replicability. In M. Otero, & E. Rhyne -Eds.), The McGuire, P. -1998) The Asian ®nancial crisisÐsome new world of microenterprise ®nance: Building healthy implications for micro®nance. Microbanking Bulle- ®nancial institutions for the poor -pp. 206±228). West tin. Boulder, CO: Economics Institute. Hartford, CT: Kumarian Press. Morduch, J. -1999). The micro®nance promise. Journal Chaves, R., & Gonzalez-Vega, C. -1996). The design of Economic Literature, 37, 1569±1614. of successful rural ®nancial intermediaries: evi- Patten, R., & Rosengard, J. -1991). Progress with pro®ts: dence from Indonesia. World Development, 24-1), the development of rural banking in Indonesia. San 65±98. Francisco, CA: ICS Press for International Center Djiwandono, J. -1998). Systeme Bancaire et gestion for Economic Growth and the Harvard Institute for macroeconomique: le cas de l'Indonesie. Economie International Development. Internationale, 76-4), 191±200. Radelet, S., & Sachs, J. -1998a). The onset of the East Franks, J. -2000). Macroeconomic stabilization and the Asian ®nancial crisis. NBER WorkingPaper No. microentrepreneur. Journal of Micro Finance, 2-1), 6680. Cambridge, MA: National Bureau of Eco- 69±91. nomic Research. EXPERIENCE OF BANK RAKYAT 1069

Radelet, S., & Sachs, J. -1998b). The East Asian ®nancial Von Pischke, J. -1991). Finance at the frontier: debt crisis: diagnosis, remedies, prospects. Brookings Pa- capacity and the role of credit in the private economy. pers on Economic Activity, 1: 1998. Washington, Washington, DC: Economic Development Institute DC: Brookings Institution. of the World Bank.