How Much Does Payment Through A Bank Account Help The Poor? An Investigation With Instrumental Variables
Speaker(s) Prof. Sankar De, Shiv Nadar University

Having a bank account is widely regarded as the first step toward financial inclusion of the poor. A common argument underscoring the importance of a bank account for the unbanked is that funds deposited in a bank account lead to higher savings. Recent evidence from field studies conducted in various parts of the globe (Kenya, Malawi, Philippines) points to positive outcomes of improved access to formal bank accounts. However, the existing literature on the savings potential of new bank accounts for the poor usually stops short of investigating whether the savings are productively used. In this paper we attempt to fill in this gap. We also use a different empirical approach from most existing studies which have used field experiments with a limited sample size and an one-time payment to subjects included in the experiments. To ensure generalizability of our findings, we use a large nationally representative sample and repeated wage payments to those included in our sample. Our empirical tests exploit special features of the National Rural Employment Guarantee Act (NREGA) of India. The test results consistently indicate that the treated beneficiary households (recipients of NREGA wage payments through bank accounts) spend significantly less than the beneficiary households in the control group (recipients in other modes of payment, including cash) on education, arguably the most important human capital development investment for the poor. The results are consistent between standard OLS and instrumental variable regressions designed to correct for omitted variable bias in OLS tests. Financial illiteracy as well as transactions costs of frequent withdrawals from bank accounts, which asymmetrically affect discretionary expenses including educational expenses for the poor, explain our results. Our tests for other discretionary and non-discretionary expenses provide corroborating evidence.

Download Paper