Abstract
Abstract GMM methods for estimating dynamic panel regression models are heavily used in applied work in many areas of economics and more widely in the social and business sciences. Software packages in STATA and GAUSS are commonly used in these applications. We provide a new R program for difference GMM, system GMM, and within-group estimation for simulation with the model we consider that is based on a standard first-order dynamic panel regression with individual- and time-specific effects. The program lacks the generality of a full package but provides a foundation for further development and is optimized for speed, making it particularly useful for large panels and simulation purposes. The program is illustrated in simulations that include both stationary and nonstationary cases. Particular attention in the simulations is given to analyzing the impact of fixed effect heterogeneity on bias in system GMM estimation compared with the other methods.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
