Wednesday, April 7, 2010

Microfinance Impact 101

Analyzing the effectiveness of microcredit or microfinance to alleviate poverty has been the subject of many blogs, articles and academic research papers.

  • Some have focused on microcredit only – while others have attempted to look at the effectiveness of microfinance more broadly (not only the provision of loans, but also savings and non financial services such as education, training, etc.).
  • Some have focused on reducing poverty while others have focused on making it more “bearable”.
  • Some have attempted to evaluate the impact of receiving a microfinance services while others have attempted to look at the impact of having access to microfinance services.
  • Some have focused only on the very poor, women, or rural areas.
  • Most have focused on one country and/or one microfinance institution, questioning further the replicability of the results.

Therefore, it is not surprising that results have varied and answering the question “Does Microfinance Work?” assertively (in a way that implies that it works across methodologies, institutional types and countries) has been a difficult task.

Does microfinance work? Define “work”…

… the outcomes
When trying to answer the question, a first consideration is what success is going to be measured by. Impact studies in microfinance have focused on many different outcomes. Did the microfinance program:
  • Reach the poor/ vulnerable?
  • Reach women?
  • Promote equity/equal access to services?
  • Promote the millennium development goals (MDGs)?
  • Increase access to health, education, water, etc.?
  • Expand access to financial services and economic opportunities?
  • Empower clients?
  • Reduce uncertainty?
  • Create jobs?
  • Increase consumption and income?
  • Increase expenditures?
  • Show social responsibility to clients, staff and the environment?
  • Etc.

…the methodology
In order to determine whether an outcome is positive, it needs to be compared to another result. Here also, studies have varied greatly. They have compared outcomes of microfinance clients after receiving a loan or other service:
  • To non microfinance clients (screened by the MFI)
  • To non microfinance clients (randomly selected)
  • To new clients
  • To outcomes of microfinance clients before receiving the loan or service

Impact studies can broadly be grouped into 3 groups: non-experimental, quasi-experimental and experimental, with increasing level of dependability.

Non-experimental studies look at the outcomes of clients receiving microfinance services over time, or compared to non clients. The main flaw of these studies is selection bias, as microfinance clients may be prone to being more successful for reasons that have nothing to do with receiving the microfinance services (due to their background, i.e., better educated from the start, more motivated, etc.)

Quasi-experimental studies use techniques to overcome the differences in background characteristics between clients and non clients, which may lead to different outcomes that have nothing to do with receiving microfinance services.

Randomized studies are considered the gold-standard of evaluation techniques as they address the issue of different background characteristics by choosing clients and non clients randomly. With Randomized studies, it is therefore possible to attribute differences in outcomes to microfinance itself.

There are many other methodological issues linked to assessing impact of any program. For more details, see “A practicioner’s guide to evaluating the impacts of labor market programs” (Fitzsimons and Vera-Hern├índez, 2009). Although focused on labor market programs, the issues discussed are equally relevant to microfinance programs.

What DO we know?

Non-experimental studies
There are scores of examples from non-experimental studies on how microfinance has had a positive impact on clients lives, well documented in the following recap studies:
  1. Is microfinance an effective strategy to reach the millennium development goals? (Littlefield, Morduch and Hashemi, 2003)
  2. Measuring the Impact of Microfinance: Taking Stock of What We Know (Goldberg, 2005)

Some conclusions include:
  • Higher average income higher for clients than non clients
  • Higher spending per capita on food and clothing for clients than non clients
  • Higher spending on housing investments for clients than non clients
  • Clients moving out of poverty faster than non clients
  • Clients educating more of their clients than non clients, etc., etc., etc.

There have also been cases (more limited) where microfinance clients have been negatively affected by microfinance (i.e., due to over indebtedness). To address this issue, the microfinance industry is championing a “Smart Campaign” to protect microfinance clients from abuse and ensure that microfinance does no harm.

Overall, the results have been overwhelmingly positive.

Quasi-experimental studies
Four major studies fall under this category:

The first was a major study was funded by the World Bank in 1997 to attempt to address the major methodological flaws of non-experimental studies. It concluded that increase credit to women (at Grameen Bank) would:
  • Increase the probability of girls’ school enrollment
  • Increase the arm circumference of girls
  • Increase the heights-for-age for both girls and boys

The second questioned the methodology used in Pitt and Khandker (1997), revised it, re-ran the data and found the following results little evidence of impact but confirmed that microfinance leads to more stable consumption over time (and across seasons).

In the third, Khandker reviewed the methodology, simplifying it and leading to less controversial results. It became the most reliable impact evaluation for a while, and showed that borrowers:
  • Women increased their food and non food expenditures (no returns for men)
  • Extreme poverty and moderate poverty declined (extreme poverty more than moderate poverty)
  • Spillover effects reduced poverty for non clients as well
  • Microfinance accounted for 40% of the entire reduction of moderate poverty in Bangladesh

Finally, in the fourth study, Roodman and Morduch re-run the data of the 3 studies mentioned above, and although they found nothing contradicts the results, concluded that they were doubtful, and that “the principle difficulties for studying the effects of microfinance have been a lack of clean quasi-experiments and an absence until recently of randomized trials” (p.40).

Randomized studies
The new randomized studies, or RCTs (Random Control Trials) that are emerging are well documented in the following recap studies:
  1. What do we really know about microfinance’s impact? (Roodman, 2009)
  2. Does Microcredit Really Help Poor People? (Rosenberg, 2010)

One such study, “The miracle of microfinance? Evidence from a randomized evaluation” (Banerjee, Duflo, Glennerster and Kinnan, 2009) shows that microcredit has a “significant and not insubstantial impact on both how many new businesses get started and the profitability of pre-existing businesses. We also do see significant impacts on the purchase of durables, and especially business durables” (p.4).

Results are still limited (given the scope and timeframes of the studies) but the prospect of more rigorous studies is very promising for the industry.

So what?

Despite the limited evidence around whether microfinance lifts people out of poverty, evidence shows that microfinance is critical to help poor people cope with poverty, by smoothing consumption and allowing to better deal with emergencies. One major book substantiates these findings “Portfolios of the Poor: How the Poor Live on $2 a Day” (Morduch, Rutherford and Ruthven, 2009). As Rosenberg (2010) notes, microfinance clients vote with their feet, and find the services so valuable that they are willing to pay high interest rates, and show very high repayment rates. He also points out that microfinance has the notable advantage of requiring fewer subsidies than other poverty alleviation strategies (i.e., education). Therefore, even if microfinance were proven NOT TO move people out of poverty, it should still be considered a very powerful tool to help alleviate its consequences.