Despite breaking into the arena of development public policy making atleast two decades back and assuming the role of being the premier anti-poverty instrument in countries like India, there have been very few detailed studies on the impact of Self Help Groups (SHGs) and micro-finance as a poverty eradication tool. All the while policy makers have been increasing their exposure to SHGs as the "magic pill" to deliver on the objective of poverty eradication. This does raise very valid questions about whether we are putting all our "development eggs" in the single "basket of SHGs"?
It is in this context that two recent studies raises interesting questions about the micro-finance driven model of poverty eradication that has become the touchstone for all such efforts in India. From a randomized trial in 104 slums of Hyderabad, Esther Duflo, Abhijit Banerjee and Co find mixed results on economic activity and no impact on social dimensions like measures of health, education, or women's decision-making. From a field experiment on credit expansion for microentrepreneurs in Manila, Dean Karlan and Jonathan Zinman find no evidence of improvements in the well-being of those accessing micro-loans, and also results that are diffuse, heterogeneous, and not directly on the targeted group.
Karlan and Zinman find several indirect effects or positive externalities from micro-loans. They find that targeted micro-entrepreneur households shrink by shedding un-productive workers form their businesses; use loan proceeds to invest in human capital of their children, rather than in capital specific to their businesses; and male entrepreneurs appeared to benefit more than female ones from mico-loans.
The study by Duflo, Banerjee and Co about the micro-finance to SHGs in Hyderabad flags off several interesting issues. They find that those households with high entrepreneurship propensity exhibit the most positive response to microloans, as they use the micro-loans to finance the fixed cost and revolving capital requirements of running a business. Micro-loans are found to increase their consumption/purchases of durables (or investments), either for their businesses or for their personal use. It is observed that micro-loans crowd-in investments in starting the business or expanding existing businesses and crowds-out spending on consumer non-durables and temptation goods (alcohol, betel leaves, tobacco, gambling, and food and tea outside home). Further, existing businesses report a large and significant increase in profits after the micro-laons were provided.
The study indicates limited or no effect of the opening of MFI branches on education, health or women's empowerment. This appears to overturn the conventional wisdom that self-help groups and microfinance activity has much greater utility as a tool of social than economic empowerment.
The purpose of this post is not so much to question the utility of SHGs and micro-finance as to examine the possible changes in approaches and methodologies required to make more effective use of these anti-poverty policy instruments. I will first list out some of the general issues relating to the role of SHGs and the present approach and policies surrounding their functioning and development.
1. What should be the most optimal size of groups? Why should it be 10, as is the case with the majority (if not all) SHGs in the country? Does not the otpimal size vary with the backgrounds of the groups - urban-rural, very poor-poor, literate-illiterate etc? What should be the size of federations? Vary the numbers of members in treatment groups to study this.
2. Are smaller groups more effective for economic empowerment through capital investments, while larger ones are better for social empowerment and consumption smoothing? Does this mean that groups should split as they gather capacity to ensure that the varying requirements of the different members of the group are more effectively addressed? In other words, after the second or third linkage, should the SHGs be split into smaller groups so as to enable the members to move forward at raising credit and expanding their businesses at their choice?
3. What should be the most optimal size of initial and subsequent loans? What should be the repayment tenor and schedule? How should loans be packaged? Should loans be use-directed or left to the discretion of the group? Should loans, especially for business investments, be supported with backward linkage support - business literacy, training, helping to source their capital equipments, marketing etc?
4. Women of which age group are more responsive for each category of loans? Does it not make greater sense to target business loans to women of a specific age group? What should be the most optimal age range of group members? Which age group is the most effective target group for SHG activity and bank linkages? Treatment groups to be divided along different ages and studied.
5. What is the impact of financial literacy, group capacity built, general adult literacy, business training etc on how the loans are utilized? How should different categories of loans be packaged, so as to optimize their cost and effectiveness for the group members? Is it more effective to package different categories of loans as loan plus one of these aforementioned linkages or support? In other words, should some categories of loans be disbursed only as a bundled product?
6. Would any other intervention involving increasing financial access, like opening of bank branches and starting of no-frills accounts have also produced the same impact as SHGs and microloans? What would be their respective impacts? Treatment-control studies on these would be instructive.
7. What is the impact of presence of MFIs on moneylenders? Do they lower their rates in view of the reduced demand for their funds? Do MFIs inculcate market discipline into moneylenders? What other positive externalities do the presence of MFIs exert? This assumes important given the fact that moneylenders perform (and will continue to do so for the foreseeable future) an important credit provisioning role in such societies and the challenge is not to eliminate or drive them out, but to discipline them into moderating their lending practices and terms.
8. What is the comparison of the loan-use break up on loans given by money lenders and MFIs? Did MFIs change the loan distribution profile by encouraging new business or business expansion loans?
9. Which economic, social, and religious groups are most and least responsive to loans? It is important to supplement with other anti-poverty measures for those groups of people who take to SHGs with the least effectiveness. In such cases, it would also be important to identify the deficiencies and the reasons for the relatively poor uptake by these people. What is the treatment impacts of loans for consumption, paying off debts, starting business, expanding old business, and purchasing consumer durables, on different categories (economic and social) of SHG members?
10. Should we not explore getting consumer durables retailers/financial institutions that sell their products in Equated Monthly Instalments (EMI) into lending directly to individual SHG members (who would other-wise take loan and purchase consumer goods)? If need be, the government can then directly transfer the interest subvention differential to the respective bank or financial institution. Is it not possible to nudge established financial institutions to provide individual loans for expanding established business by leveraging the credit worthiness provided by the individual's membership of a good SHG?
11. Should we not have a detailed database which tracks the uses to which different groups and its members have used their share of loans for and their impacts, so as to more efficiently tailor and direct the next round of loans? For example, if one member in a group was repeatedly using a large share of the loans for children's education, it may be useful to help the student linked up with a scholarship provider. The information on consumption loans can be used to provide assistance to access alternate smoothing strategies like insurance etc.
And here are a few questions on the aforementioned Hyderabad study involving the MFI Spandana
1. I cannot but help feel that Spandana, being a commercial MFI, showed selection-bias in identifying the areas it proposed to set up its branches. This is evident in the criteria it used to select the slums - "poor, but not the poorest of the poor", smaller habitations/bastis within the slums, home ownership by 80% of the groups members etc. Did these not provide an implicit guarantee and distort the selected group? SHGs in rural areas may not fit into this profile.
2. Did the comfort provided by home ownership, among 80% of group members, provide an implicit guarantee for Spandana? If this were the case, would the individual members have accessed institutional loans by mortgaging their title deeds (assuming they had title deeds, which they should have had in Hyderabad)?
3. Was the decision of new MFI entrants on where (which places) to enter influenced by the presence of Spandana groups? Did Spandana's presence have spill-over effects which may have created an upward bias on the outcomes of the other groups? What effect did these new entrants have on Spandana groups themselves? Since MFI activity was virtually absent, the impact of Sapndana's entry may have been different from if there was already strong MFI activity.
4. Was there no government SHG activity in these sample areas at the baseline of 2005? I am inclined to believe that there was, more so given the fairly strong network of SHG activity in Andhra Pradesh. Has this effect been controlled for?
5. The quality of such household surveys are suspect, especially given the type of information solicited - asset ownership, decision making issues, expenditures, borrowings, savings etc. The households, especially in urban areas, have been socialized by government surveys to under-report their assets and savings, over-report their expenditures etc.
6. Is the increase in business openings in the aftermath of entry of Spandana due to the MFI loans or due to the fact that a few members who initially accessed these loans set off an emulation effect on their neighbours to start businesses? In other words, did Spandana's work exhibit spill-over effects on those who otherwise may not have taken micro-loans and if they did take it, would have used it to for other purposes? Did the new addition, due to this bring in an attrition bias into the sample, by adding newer memebers into the possible group of beneficiaries?
7. However,the selection-bias inherent in the choosing of Hyderabad, with its urban and relatively empowered socio-economic context, may have masked the cause-effect relationship between microfinance activity and social empowerment. One explanation for the limited assessed impact on the social empowerment dimension may be attributed to the fact that social changes take much more time than the two years of the study to have any noticeable impact. I am also inclined to believe that the effects would have been much more marked in the rural areas.
All these aforementioned issues can (and should) be examined extensively using randomized control studies, so as to re-design policies on SHGs and micro-finance to make them more effective and deliver greater bang for the buck.
Update 1 (20/6/2010)
Abhijit Banerjee argues that "there is now recognition that poor people and small firms have very limited access to capital and risk diversification"; they "do start a lot of firms, but these firms seem intended to remain tiny"; "the poor are not particularly well-suited to be entrepreneurs: They neither have the risk bearing capacity nor the human capital"; "nor will anyone give them enough capital to really grow the businesses". He therefore suggests that the main source of dynamism has to be growth of medium to large firms, though "there is evidence showing that these firms are too rare and too small in developing countries".