Abstract
We extend a recently developed generalized local polynomial estimator into a semiparametric smooth coefficient framework to estimate a generalized cost function. The advantage of the generalized local polynomial approach is that we can simultaneously choose the degree of polynomial for each continuous nonparametric regressor and the bandwidths via data-driven methods. We provide estimates of scope economies from the joint production of microloans and microdeposits for a dataset of Microfinance Institutions from over 50 countries. Our approach allows analysis on all Microfinance Institutions rather than only those offering just microloans. Moreover, the smooth coefficient estimator provides a general interface in which to account for both direct and indirect environmental factors. We find substantial scope economies in general, of about 10 % at the median, as well as evidence that economies of scope vary across the type of services and country in which the MFIs operate, suggesting key insights into policy prescriptions.
