Reexamining nonlinear effects of intellectual capital on firm efficiency

This paper first gauges the level of firm efficiency using the Stochastic Nonparametric Envelopment of Data (StoNED) approach. Our firm efficiency score closely reflects a firm's actual operating conditions when using the statistical foundations of both Data Envelopment Analysis (DEA) and Stochastic Frontier Analysis. Secondly, we estimate the nonlinear effects of intellectual capital on StoNED-based firm efficiency using the Generalized Additive Model (GAM). This model lets us depict the possible nonlinear relationship between explanatory variables and the explained variables in an additive manner. Our analysis of 1898 firm-year observations for U.S.-listed firms from 1999 to 2019 indicates that (i) our sample firms generally have about 65% of room left for improvement that could transform resources into wealth, and (ii) of the three major components of intellectual capital, human capital exhibits a concave-up curve, while structural capital and relational capital both demonstrate an upward trend, with each having an inflection in the middle of that curve. The GAM results remain qualitatively similar even after we re-estimate firm efficiency using the network slacks-based measure DEA model, and (iii) we discuss these comparisons and the respective implications of the three components.