Testing for Experience Effects in Banking

This paper tests for empirical evidence of learning by doing in banking with the aim of identifying a micro-founded driver of financial sector development. Learning by doing entails cumulative experience reducing the amount of labor or other inputs required to produce the same amount of output. However identifying this experience effect poses challenges because firms may increase output as input prices decline, introducing the possibility of endogeneity bias in estimating the cost function. Applying a two-step correction procedure to my bank cost function, I correct for endogeneity as well as selection biases arising from sample dependence. The problem of these biases has not been addressed in empirical work on learning by doing, or the banking efficiency literature, nor have experience effects been focused on in the banking context. Using the corrected model, results suggest that experience is associated with reduced costs: the experience effect is decreasing and fades after around 10 years. For example, on average, a 10 percent increase in experience, for a bank of around 1 year of age is associated with a 10.9 percent decline in variable cost; for a 5 year old bank, that becomes a 2 percent decline in variable cost.