This paper is under review which means review has begun. You can track the progress of this review on GitHub over here »

\pkg{prodest} provides functions for TFP estimation following the most widely-known methodologies using the control function approach. Focusing on Value Added production functions, it estimates the two--steps models presented by Olley--Pakes (1996) [@olley_etal96] and Levinshon--Petrin (2003) [@levinsohn_etal03], as well as their correction proposed by Ackerberg--Caves--Frazer (2015) [@ackerberg_etal15]. The system GMM framework proposed by Wooldridge (2009) [@wooldridge_09] is also implemented in two slightly different versions.

Dealing with standard Cobb-Douglas technology in a panel framework, all methods assume that the productivity term evolves according to a first-order Markov process and that a proxy variable exists - i.e., a function of state variables and productivity - invertible and monotonically increasing in productivity. Exploiting these features and with different choices of the proxy variables, the methods yield consistent estimates of labor and capital inputs parameters, allowing for an immediate computation of TFP.

The \pkg{prodest} package features also the DGP used by Ackerberg--Caves--Frazer (2015) [@ackerberg_etal15] and allows for the simulation of datasets according to several measurement errors and random shock variances. It can be used by practitioners for both running Monte Carlo simulations and benchmarking estimate results.

Archive DOI: pending