BSFM: a computer program for Bayesian stochastic frontier models (Q1381519)

The model \(y_i=h(x_i,\beta)+v_i-z_i\) is considered, where \(y_i\in R\), \(x_i\in R^d\), are known (observations), \(\beta\in R^k\) is an unknown vector parameter, \(v_i\) are normally \((0,\sigma^2)\) distributed, \(z_i\) are gamma distributed with the shape parameter \(j\) (\(j=1,2,3\)) and the unknown scale parameter \(\lambda\), \(h\) is a known function (frontier). A C++ program is described which performs Monte-Carlo iterative calculations of the a posterior distribution of \(\beta\). Results of empirical data analysis are presented for a linear model of the frontier.