Do regulated microfinance institutions achieve better sustainability and outreach? Cross-country evidence

Abstract

In spite of increasing pressure on microfinance institutions (MFIs) operating in developing countries to transform into regulated financial intermediaries, to date, no study has investigated whether regulated MFIs actually achieve better financial results and reach more poor clients than nonregulated MFIs. This article explores the impact of regulation on MFI performance using newly released data for 114 MFIs from 62 countries in an empirical model where performance is specified as a function of MFI-specific, regulatory, macroeconomic and institutional variables. Consistent with recent cross-country evidence on the impact of banking regulations on bank performance (Barth et al ., 2004), this article finds that regulatory involvement does not directly affect performance either in terms of operational self-sustainability or outreach. The article also finds that less leveraged MFIs have better sustainability. The policy implication is that MFIs’ transformation into regulated financial institutions is may not lead to improved financial results and outreach. However, the finding that MFIs collecting savings reach more borrowers suggests that there may be indirect benefits from regulation, if regulation is the only way for MFIs to access savings.

Acknowlegements

Valentina Hartarska is an assistant professor at the Department of Agricultural Economics and Rural Sociology, Auburn University; Denis Nadolnyak is a post-graduate research associate at the Department of Agricultural and Applied Economics, University of Georgia.

Notes

¹ The regulated MFIs’ coefficient of depth of outreach measured by the average outstanding loan balance divided by the GNI per capita is 117%, while the coefficient for nonregulated MFIs is 63.4%.

² Financial intermediaries are also subject to externalities. Contagious bank runs occur when the failure of one bank imposes a negative externality on other banks and jeopardizes the safety of the payment system. Some MFIs provide payment facilities and may potentially create a negative externality but the relevant empirical question is to what extent a small MFI operating in a niche market represents a threat to the payment system of a country and is this threat large enough to justify the costs of regulation. Solvency regulations prevent the impact of negative externality (Diamond and Dybvig, 1983). Solvency regulations, however, rely on the quality of collateral to classify loan risk. Most MFIs, instead of collateral, use group loans and the promise of larger loan size to motivate repayment. This suggests substantial regulation design cost as regulation of microfinance activity needs to be tailored to the MFIs’ peculiarities. Wright (2000) argues that microfinance has not yet achieved the market penetration necessary to cause systemic risk in the financial system.

³ It is possible that only the best performing MFIs self-selected to report their data and selection bias can exist, although it is reasonable to assume that the regulated and nonregulated groups are proportionally represented.

⁴ The Microcredit Summit Project focuses on outreach indicators such as the number of MFIs operating in a country and their borrowers’ poverty level and gender distribution. Unfortunately, this project collected no or very limited financial data, such as MFIs’ self-sufficiency, rendering the data unusable for evaluating the impact of regulation on the MFIs’ performance.

⁵ In the sample 10% of the MFIs are nonregulated with for-profit status, 23% are regulated NGOs and 21% are nonregulated NGOs.

⁶ Most empirical models that study bank performance include loans as a measure of bank risk exposure. Unlike banks, however, most MFIs do not engage in income generating activities other than lending. Therefore, LOANS not only controls for risk exposure but also for MFIs’ focus on lending because using funds for other purposes such as new buildings, cars, etc., is likely to affect income generation in the current period.

⁷ Unfortunately the data does not distinguish between obligatory savings and voluntary savings.

⁸ See http://www.worldbank.org/research/projects/bank_regulation.htm

⁹ This result is not shown to save space but it is tested for all model specifications.

¹⁰ Recent studies have found that institutional factors influence financial development. Numerous indicators were tried out to see whether they can serve as better instruments than the index of economic freedom and the index of property rights. These include indexes provided by the Heritage Foundation, the World Bank Governance Indexes, Economic Freedom of the World Indexes (http://www.freetheworld.com) indexes on legal origin, religious affiliation and geographic location. Unfortunately, most of these indexes turned out to be poorly correlated with regulatory status and could not be used as instruments.

¹¹ In microfinance, there is no clear agreement as to whether outreach and sustainability are complements or substitutes. Many authors analyse the performance of a single MFI to conclude that outreach and sustainability are complements but no study has ever analysed a sample of MFIs operating in different institutional environments.

¹² The industry standard for this dimension is ‘depth of outreach’ calculated as the ratio of average outstanding loan size divided by the per capita GNP. Regression on a smaller sample with depth of outreach as the dependent variable was estimated, but not reported here because the best specification produced very low R-squared. Nevertheless, the coefficient of the savings ratio indicates that MFIs with higher proportion of savings actually serve richer borrowers. Better data and a larger sample would be necessary to validate these results.

¹³ This weak result may also be due to poor data quality since the variable COMPET comes from the data collected by Microcredit summit – to date, the most detailed data on the number of MFIs. Alternative specifications with the number of MFIs scaled by GNP per capita and the number of inhabitants did not produce better results.