Univariate Panel Data Models and GMM Estimators: An Exploration Using Real and Simulated Data

This paper explores the time series properties of commonly used variables in firm-level panels: sales (turnover), employment, R\&D, investment, and cash flow. We focus on two questions: 1) whether the behavior of these series is consistent with stationarity, and if so, 2) what order of autoregressive process best describes them. The answer to these two questions is of substantive interest for those who model the dynamic evolution of firms and their behavior. In particular, we are interested in whether firm data is trend stationary (exhibits regression to individual firm means) or difference stationary (evolves as a random walk, possibly with a non-zero drift). We find that estimation of even these very simple processes using fairly large but short panels is fraught with difficulty and we explore the convergence rate of the GMM estimator using simulation methods. We also report the results of using a new class of tests proposed by Im, Pesaran, and Smith for discriminating between stationary and nonstationary processes in medium-sized panels.