Finance & PSD Impact MARCH 2016 The Lessons from DECFP Impact Evaluations ISSUE 35 Our latest note highlights the results from a pilot exercise conducted in Peru, which shows psychometric credit scoring can increase access to credit for small and medium enterprises who do not have a credit history, without increasing default risk for the lender Psychometrics as a Tool to Improve Access to Credit Irani Arráiz, Miriam Bruhn, and Rodolfo Stucchi Small and medium enterprises (SMEs) often predict loan repayment behavior, including face greater credit constraints than large the applicants’ attitudes, beliefs, integrity, firms because they lack audited financial and performance. All applicants who statements and other information about their achieved a score higher than a threshold set operations, and as a result, financial by the bank were offered a loan. institutions have difficulties assessing the Peru has several private credit risk of lending to them. bureaus that, together, cover 100 percent of Comprehensive credit bureaus where the adult population. Thus, all loan applicants lenders share information have been shown also have a traditional credit score. But for to increase credit to SMEs. However, not all individuals who have not previously taken countries have credit bureaus and where out a loan from a formal financial institution, bureaus exist, the information they provide this score is based primarily on demographic may be limited, for legal and institutional information rather than actual credit history reasons. Also, loan applicants are subject to a and banks may be reluctant to lend to them. chicken-and-egg problem. Bureau For these applicants without a credit information is most useful for making credit history, we ask whether being offered a loan decisions regarding loan applicants with a based on their EFL score increased use of detailed credit history, but applicants can credit. Clearly, individuals with an EFL score only build that history by getting credit, for above the threshold should have been more which they need a good credit history. likely to take out a loan from the The Entrepreneurial Finance Lab implementing bank than individuals with an (EFL) has developed an alternative credit EFL score below this threshold. However, information tool, based on psychometrics, individuals with an EFL score below the which can potentially be used by lenders to threshold could potentially have gotten a loan better screen loan applicants. We examined from another financial institution. the effectiveness of this tool in increasing A second question we ask is whether access to credit by SMEs in the context of a lending to new clients based on the EFL tool pilot exercise conducted in Peru. increases credit risk for the bank. EFL tool and Context Data and Study Design The financial institution participating in the We obtained data on formal credit study, the fifth-largest commercial bank in usage and repayment behavior from the Peru, piloted the EFL tool starting in March Superintendencia de Banca y Seguros (SBS) 2012, with the goal of expanding its lending in Peru, as well as data collected by the to SMEs. Loan applicants filled out a 45- implementing bank and EFL on 207 loan minute questionnaire (the current version applicants that were part of the pilot exercise. takes 25 minutes) and EFL generated a credit To measure the effect of the EFL tool, score based on the applicant’s answers to we use a regression discontinuity design that questions capturing information that can compares loan applicants who scored just Do you have a project you want evaluated? DECRG-FP researchers are always looking for opportunities to work with colleagues in the Bank and IFC. If you would like to ask our experts for advice or to collaborate on an evaluation, contact us care of the Impact editor, David McKenzie (dmckenzie@worldbank.org) above and just below the EFL score threshold decided not to accept the loan offer after set by the bank for offering a loan. We used seeing the loan terms. different parametric specifications and also  We do not find significant differences in local linear regression to check robustness of loan repayment behavior across the results. applicants above and below the EFL score threshold (see Figure 2). Results  We find that those above the threshold are Figure 2. Number of Days in Arrears more likely to have a loan from any around EFL Score Threshold financial institution than those below the 400 threshold (figure 1), suggesting the EFL tool increased use of credit for SME Nubmer of days in arrears 300 owners without a credit history. About 45% of applicants just below the 200 threshold obtained a loan, compared to 100 80% just above the threshold. Figure 1. Increase in Loan Use at EFL 0 360 380 400 420 440 Score Threshold EFL score Sample average within bin 2th order global polynomial 1 Note: The figure shows maximum number of days in .8 arrears. The average of this variable is calculated for 20 bins on each side of the EFL score threshold. .6 Policy Implications .4 Our results illustrate the importance of .2 information for assessing credit risk and expanding use of credit. Increasing the 0 360 380 400 420 440 EFL score quality of the information that credit bureaus Sample average within bin 2th order global polynomial can access—for example, including data Note: The figure shows a dummy = 1 if the applicant from retailers and utility companies in received a loan from any formal financial institution. addition to financial institutions, and The average of this variable is calculated for 20 bins allowing positive information (payment on each side of the EFL score threshold. history on accounts in good standing) in addition to negative information (late  Note that not all applicants with an EFL payments, defaults, and bankruptcies)— score above the cutoff ended up taking could improve credit-scoring models and out a loan, leading to a non-linear increase lenders confidence in these scores, relationship between the EFL score and even for entrepreneurs who have not the likelihood of having a loan. This previously borrowed from formal financial relationship may be driven by institutions. In the meantime, EFL offers a characteristics, such as risk aversion, that practical solution to financial institutions in are captured by the EFL score. That is, countries where well-developed credit more cautious applicants may have bureaus are in the process of consolidation. For further reading see: Arráiz, Irani, Miriam Bruhn, and Rodolfo Mario Stucchi. (2015) “Psychometrics as a tool to improve screening and access to credit” World Bank Policy Research Working Paper no. 7506 Recent impact notes are available on our website: http://econ.worldbank.org/programs/finance/impact