Presentation on theme: "1 Research in Micro Finance: Big Questions and How to Answer Them."— Presentation transcript:
1 Research in Micro Finance: Big Questions and How to Answer Them
2 Agenda What is the goal of good research? Why is research needed in microfinance? What are some of the important questions? How might we answer these questions? What limits many common methods? How can we measure causal impact, and why does causality matter? How do we get good data? How do we make research matter?
3 The Goal of Research To define reality as accurately as possible To create useful knowledge
4 Why Research in Microfinance is Needed Despite major growth of microfinance, there are more poor people in India now than ever before Still a young industry We still don’t understand basic impacts, or relative value of different models Many misconceptions about microfinance prevail
5 Important Questions in Microfinance What are the basic impacts of credit? Savings? Insurance? Who benefits from microfinance? What are the returns to receiving capital? Why don’t micro-businesses grow? How do different products affect different people in different ways? (heterogeneous effects) How are loans used for consumption, and is this good or bad? How can the poor mitigate risk? How do the poor think of fees and interest rates?
6 Common Research Methods Focus groups and anecdotes of existing clients – good for understanding colorful information for how impacts happen, or information about a community (PRA’s) but not for understanding prevalence! Finding groups for comparison Comparing clients to non-clients Comparing clients “before” and “after” Comparing new clients to old clients
7 Common Research Methods (and their problems) Focus groups and anecdotes of existing clients (but drop-out rates are 20%+ - what about impacts on them? Are anecdotes reliable?) Comparing clients to non-clients (but clients are inherently different from non-clients) Comparing clients “before” and “after” (but other things may have happened in the meantime with greater effect) Comparing new clients to old clients (but early adopters are inherently different from later adopters)
8 Determining Causal Impact Randomized control trial is the “gold standard” methodology Develop a true “control” group: Randomizing access to a product, and compare outcomes among “treatment” to “control” Just like those used in medical trials Selection bias and attrition bias solved, assures internal validity
9 Getting good data Data from organizations reporting on themselves is suspect (e.g. MIX Market) If you want good data for original research, you have to get it yourself! (but it’s expensive!) Primary data collection is very important for field experiments Data collected is only as good as the sample selection, survey design, collection process, data entry, cleaning, and analysis – every step requires good planning and rigorous monitoring
10 Making research matter Research should inform practice: provides evidence on what works, and what does not. Research should inform policy: document and inform policy debates with empirical data and analysis Research should facilitate a process where research questions emerge from the local policy context and policy and programmatic decisions are guided by research outcomes. Good research on microfinance requires good relationships with practitioners!