Abstract:
Constructing a novel micro-geographic individual-level data set, we study the relevance of shoe-leather costs on cash withdrawals. An unexplored issue in the literature is the consistent estimation of the marginal effect of travel distance on withdrawals when a fraction of unobserved withdrawals have free/low shoe-leather cost, i.e., consumers withdraw upon conveniently encountering a free/low cost withdrawal opportunity. To overcome this challenge, we propose a classification technique to identify respondents who have incurred these free/low cost withdrawals, and subsequently account for such endogenous selection from the exclusion restriction of the adoption of recent online financial innovations. We find that there exist significant threshold effects of distance on typical monthly withdrawal frequency. For respondents living within 1.56 kilometers of their affiliated financial institution, a one-kilometer reduction in distance is associated with an average marginal increase of 0.31 withdrawals per month. In terms of heterogeneous effects, distance plays a larger role in higher-income and older-age cohorts. These results are robust to various econometric specifications.