But much remains to be done. Most microfinance institutions (MFIs) are still small and vulnerable to constraints on their resources and to the risks inherent in sin- gle-issue portfolios. Most depend upon donors and governments to remain in opera- tion. There is much waste and duplication, and some mature programs have declining loan recovery rates, even as competition for borrowers rises from conventional banks and finance companies.

Analyzing the failures of credit programs aimed at small farmers and the successes of other programs showed the need for new understanding of the ways that poor house- holds make spending, borrowing, and saving decisions. This area was previously neglected in policymaking on food security issues. The International Food Policy Research Institute (IFPRI) supported household surveys in nine Asian and African countries during the 1990s that analyzed formal and informal financial transactions, and it also evaluated the success of innovative approaches at some MFIs. The overall goal was to clarify the conditions under which state investment in microfinance pro- grams might improve life for poor people more than state investment of the same funds in education, health, nutrition, or infrastructure development.

The research led to the concept of the "critical triangle of microfinance".the need for any MFI to manage simultaneously the problems of outreach (reaching the poor both in terms of numbers and depth of poverty), financial sustainability (meet- ing operating and financial costs over the long term), and impact (having discernible effect upon clients' quality of life). This book elaborates on these objectives and shows that the most successful MFIs expand all sides of that triangle. Tradeoffs are sometimes inevitable, but even so, synergies among the three make the concept valuable.

The Functions of Financial Services
Financial services such as loans and savings instruments can raise household income, consumption levels, and investments in three ways: by allowing expanded production that will generate income; by allowing more risk-efficient "asset portfolios" (animals, land, material, and so on); and by smoothing consumption patterns (tiding a family over a drought, seasonal scarcity, crop failure, illness, and so on). Different levels of poverty are associated with different patterns of demand, different risks and risk per- ceptions, and different means of access to finance.those individuals with slightly more education, technical knowhow, or market access will have different capacities and demand schedules for seeking loans than the very poor. The difficulty in meeting the demand of the poor does not always have to do with providing simple access to services.

In Burkina Faso, for example, strong seasonal uncertainties mean the poor need savings and insurance services, not just loans that will let them increase production. A Pakistan study found that fear of adult male illness led families to invest in relatively liquid assets (cash, food, animals) that could be sold in emergencies, rather than in long-term production assets, thus lowering their lifetime wealth. Studies in Mexico found that wealthier individuals received more credit, in part because they applied for loans more often than the poorest. They did so because they faced lower costs when applying for loans (in terms of time, effort, and likelihood of success) than the poorer clients. A joint-liability group approach was more effective in reaching the poorest in one Bolivia study than individual loans, but in Cameroon individual loans had lower transaction costs. In South Africa, graduated payment schedules for land-acquisition loans were found to work better than short-term uniform payment schedules, because the former allowed farmers to recover from up-front costs and accumulate savings to repay the debt. Different products, in short, have to be applied differently in different places.

MFIs can therefore usefully employ the concept of the triangle of outreach, financial sustainability, and impact as they choose their target clients and create the products they will offer, the loan conditions they will set, and the application procedures they will require. The more innovative and locally adapted the project design, the better the prospects all around.

The triangle indicates, for example, that broader and deeper outreach to the poor may require a tradeoff in financial sustainability. While the poorest are generally considered less credit-worthy, the real cost to MFIs is in obtaining information about an individual's credit-worthiness, no matter how big or small the loan. Reaching large numbers of the poorest may therefore justify the provision of subsidies to MFIs specialized in serving them. The need for subsidies, of course, raises the issue of donors' goals. But even subsidies may not be enough to ensure program success. In a study in Latin America, reliance on subsidies correlated with the breadth of MFIs' outreach.the greater the external support, the larger the numbers of poor people served. But in Malawi, access to credit was found to have no real effect on farming efficiency unless farmers also had access to better farming technology. Additional help to clients in the form of training, technology, marketing, and followup can raise profitability and improve a program's impact but will also increase its costs. However, synergies exist too: if clients perceive the MFI to be sustainable, more of them will seek its services. As the MFI strives for sustainability it will come up with better products and demand-oriented approaches and increase its efficiency with cost-reducing information systems, new lending technologies, and other such measures. These will in turn increase impact.

If all these factors are interdependent, they still must operate within the context of public- and private-sector policies. The state has a direct role in supporting microfinance, for example by helping with startup costs for MFIs and creating favorable regulatory and supervisory systems. In China, administrative interference and distorted pricing systems that did not cover risks resulted in meager outreach and high fragility of rural MFIs. The state also has an indirect role in setting overall policy to stimulate economic growth; in creating legal systems, land registries, and inheritance laws that enforce contracts and protect property rights, especially for women; and in removing the traditional development policy bias against agriculture, rural areas, and small businesses. Education, health care, and infrastructure development are also essential for creating a pro-poor environment for MFIs and their clients.

Future research needs to untangle the impact of financial services like loans and savings instruments from the impact of nonfinancial services like training programs. But MFIs are clearly no panacea against poverty, and their usefulness varies by community and economic context. While MFIs clearly do best in areas with better infrastructure, market access, and complementary services, the most reliable indicator of impact is their continued use by clients and their ability to function without direct subsidies. Marketing studies of MFI customer satisfaction, therefore, are likely to generate more immediately useful information for policymakers than long-term impact assessments.