A SOCIAL, BEHAVIORAL, & EDUCATIONAL RESEARCH (SBER) CASE STUDY OF BEHAVIORAL ECONOMICS RESEARCH A MICROFINANCE STUDY By Ken Carsn, JD with the SBER Subcmmittee f Harvard Catalyst s Regulatry Fundatins, Ethics, and Law Prgram OVERVIEW The scial, behaviral, and educatinal research (SBER) case studies prvide educatin and guidance n hw t identify and mitigate risks assciated with SBER. These studies may be used by bth IRB administratrs and investigatrs when reviewing and designing research studies that invlved SBER cmpnents. Case studies fllw a standard frmat that includes: 1) a fact pattern, 2) regulatry, cultural, and ethical issues, and 3) a risk/benefit analysis and risk management ptins. This frmat was created t allw fr flexibility in applying the case studies. By identifying cmmn themes, linking them directly t federal regulatins and guidance, and utlining risk mitigatin ptins, the case studies can be used in a variety f ways, which include: 1) as an educatin tl fr training individuals in human subjects research, 2) as a basis fr develping reviewer checklists/wrksheets, and 3) as a tl in designing research prjects. We encurage yu t reprduce and use these materials freely. In ding s, we require that yu acknwledge Harvard Catalyst as the publisher and that yu give apprpriate credit t the individual authrs. Fr additinal infrmatin, visit http://catalyst.harvard.edu/abut/citingsupprt.html. CASE STUDY SCENARIO/FACT PATTERN: Micrfinance includes the prvisin f credit t thse wh d nt have access t traditinal banking services (i.e., residents f rural villages r, mre recently, urban residents living in pverty). Micrfinance prvides ecnmic researchers the pprtunity t design real life randmized interventins t test ecnmic behavir hyptheses with large numbers f participants. In this case, a team f ecnmics researchers seeks t test the tenet that the framing f default chices affects decisin-making. The researchers plan t structure micrfinance lending peratins with different default repayment schemes. A micrfinance lender, wrking with the researchers, will mve int a previusly underserved area in India. The lender will ffer ne plan in 200 villages, an alternative plan in 200 ther villages, and n plan at all in a cntrl grup f an additinal 200 villages. The first plan will have an autmatic savings depsit feature as a default, the ther will nt; this ther plan will nly include regular cmmunicatins encuraging saving. The lending prgram will exclusively recruit wmen. Visit: http://catalyst.harvard.edu/prgrams/regulatry/sber.html Email: regulatry@catalyst.harvard.edu 1 Behaviral Ecnmics Research
The researchers will evaluate: Lan uptake Delinquency rates Saving behavir The researchers will als lk at mre indirect measures f the impact f micrfinance, such as assessments f changes in family assets and schl enrllment f children. Since the lending prgram recruits wmen exclusively, impacts n intra-family relatins will als be studied. Researchers will btain lan data directly frm the micrfinance lender, and additinal infrmatin frm brrwers and their families thrugh detailed, in-persn surveys (lasting ver ne hur). When cnducting the surveys, the researchers d nt plan t disclse their cnnectin t the micrfinance lender r the new lan fferings. The micrfinance lender s agreements with brrwers will disclse that accunt infrmatin may be prvided t third parties wh are evaluating lending prgrams. N cmpensatin will be ffered fr participatin in the research. The experiment is carried ut against a backgrund f reprts that micrfinance brrwers repayment rates are in the high 90 percent range. Hwever, there have been disturbing news reprts f sme micrfinance lenders engaging in what appear t be predatry lending, taking harsh measures t cllect frm brrwers wh are in arrears. Visit: http://catalyst.harvard.edu/prgrams/regulatry/sber.html Email: regulatry@catalyst.harvard.edu 2 Behaviral Ecnmics Research
A MICROFINANCE STUDY DISCUSSION Questins/Cmments fr the researcher: What rle have the researchers played in develping the tw plans the lender will ffer? D the researchers have any cnsulting arrangements with the lender? Des the lender expect t have access t the survey respnses f individual brrwers? If s, culd lan fficers use the infrmatin in a way that culd prejudice the respndents? Will survey participants ever be tld that the research was cnducted with the cperatin f the micrfinance lender, and that their survey respnses were cmbined with accunt infrmatin prvided by the lender? Wuld the micrfinance lender have the capacity t rll ut the prgram in all 600 villages at nce? If s, what is the justificatin fr withhlding credit access in the cntrl grup f 200 villages? If nt, was the selectin f villages that wuld get the service equitable? REGULATORY, CULTURAL, & ETHICAL ISSUES: Understanding the lcal cnditins is a majr challenge fr a US-based IRB. There may be a questin as t whether the lcal research partner knws the ethnic and cultural make-up f target village grupings. A cmmn cnsent issue in this type f research invlves ensuring that brrwers understand participatin in the research is nt a predicate f their receiving a lan. What are the repercussins f cnducting husehld surveys that are quite lng, and ask highly sensitive questins ( D yu keep cash at hme? Gld? Hw much? Where is it kept?)? Micrfinance lans such as these are typically made t wmen in husehlds, and a cmmn line f research is t vary the way husbands are infrmed abut a prgram. Such research questins highlight an underlying dynamic f micrfinance, namely, the effect f ecnmic empwerment f wmen n relatinships between wife and husband. Is there a heightened risk f dmestic abuse as a result f participatin in this study? Suppse the lans are nt repaid: is bad credit, pst-interventin wrse than n credit in the cntrl arm, in a village setting where there are n frmal credit tracking systems? Questins fr the IRB: Des the withhlding f the infrmatin abut the relatinship between the researchers and the lender qualify as deceptin? If s, can it be justified? What part f this activity is research, and what wuld take place even if the researchers were nt invlved? Hw is the risk arising frm the research prperly characterized? If the risk f defaulting n a lan exists independent f the research, since the lender is ging ahead regardless f the research prject, is there a heightened risk f default in ne treatment arm r the ther? RESOLUTION & DISCUSSIONS: Risk/Benefit Analysis: Researchers will frequently argue that the risk f their research is minimal because the lenders are mving int this area regardless, and all the research is ding is evaluating which apprach helps Visit: http://catalyst.harvard.edu/prgrams/regulatry/sber.html Email: regulatry@catalyst.harvard.edu 3 Behaviral Ecnmics Research
brrwers make the best decisins. Even s, steps shuld be taken t address and mitigate risks assciated with participatin in this type f research. Mitigatin/Management f Risks: The data cllectin and strage challenges can be met by hiring reputable surveyrs frm utside the area. Research institutins assciated with universities may be gd resurces; grups wh d this type f research n a regular basis have established meticulus prtcl prcedures fr cding paper surveys, and securing them n a daily basis. The IRB shuld press the researchers t demnstrate that all survey sectins are intended t supprt the research questins (surveys may get adpted frm earlier studies, and superfluus sectins retained). Agencies and NGOs that prvide supprt fr victims f dmestic abuse must be lcated, in apprpriate cases, and infrmatin abut them prvided. Interim mnitring may als be required. In rder t carry ut their study prgrams, lenders may agree that they will nt take measures t btain repayment f delinquent lans frm participants (thugh participants can t be tld that). Pilt trials may be feasible, depending n the timetable f the lender and the researchers. An anthrplgist, wh has a deep, qualitative understanding f the culture and cmmunity structure, shuld review the prpsal t prvide an pinin n, and suggest changes t, the prject. Researchers shuld engage smebdy frm the area wh is knwledgeable abut the ppulatin and familiar with what is acceptable, t assist the researchers with recruitment (and increase likelihd f participatin) and cnsent prcesses. Alternate Details: Identity f the micrfinance lender and relatinship t the research Lender: a divisin f a fr prfit, publicly traded majr in-cuntry bank, r Lender: a nn-prfit that received initial seed funding frm a fundatin knwn fr internatinal develpment wrk in lesser develped cuntries Lender: either f the abve, but the research is being spnsred by: Wrld Bank USAID In additin t the tw interventins mentined abve (alternative repayment prgrams), the researcher wants t knw whether access t financing increases crruptin: are brrwers wh use the lan prceeds t prduce gds in their hme mre likely t be met with payff demands by lcal fficials, in rder t be able t market their gds? Sme cncerns, nt all by any means: If this issue is prbed in interviews, will it heighten risk fr brrwer subjects? What cnsent disclsure wuld be required; if wrd abut studying bribery traveled at the utset, the results f the study, r even the ability t carry ut the study, might be cmprmised. Visit: http://catalyst.harvard.edu/prgrams/regulatry/sber.html Email: regulatry@catalyst.harvard.edu 4 Behaviral Ecnmics Research
Other Events: On cntinuing review, the researcher reprts that, surprisingly, sme brrwers are using the lans fr unanticipated purpses (pssibly any f the fllwing) Financing the sending f husehld members frm village t a distant large city, t wrk in factries with dangerus wrking cnditins Spending n lcal cmmunal scial activities, e.g. purchasing firewrks r supprting religius hliday ceremnies, that prmte status but dn t directly generate incme While the micrfinance lan prgram has incentives and rules t blster schl enrllment and attendance f brrwers children, brrwers are turning t their extended families and neighbrs t recruit very yung children t d piece wrk fr extremely lw wages, while their wn children attend schl Lan prceeds are used t finance smuggling and sales f cntraband gds REFERENCE(S) Researchers affiliated with the Jameel Pverty Actin Lab at MIT have carried ut imprtant studies using large scale randmized studies t answer ecnmics-based questins in develping cuntries. http://www.pvertyactinlab.rg/methdlgy/what-randmizatin. Visit: http://catalyst.harvard.edu/prgrams/regulatry/sber.html Email: regulatry@catalyst.harvard.edu 5 Behaviral Ecnmics Research