Previous research on total factor productivity (TFP) shows that cross‐country differences in income cannot be fully explained by stocks of capital (K), labor (L) and human capital (E). In addition, the omission of major production inputs or the use of proxies to estimate unobservable inputs leads to biased estimation results. This study addresses the above issues by employing a novel econometric approach and provides empirical evidence that a fixed production input, and therefore a country's income, is positively correlated with the existence of British‐style institutions and negatively correlated with cultural heterogeneity and Spanish‐style institutions. Our methodology is twofold. First, using data for 62 countries from 1980 to 2004, we regressed a random‐coefficients stochastic production frontier that allows estimating a fixed unobservable production input without using proxies. Second, the estimated fixed production input is shown to be related to colonial institutions and cultural heterogeneity by means of ordinary least squares and feasible generalized least squares regressions.