Stephen R. Bond's paper discusses econometric methods for dynamic panel data models, focusing on microeconomic data with a large number of individuals or firms observed over a small number of time periods. The paper emphasizes single equation models with autoregressive dynamics and non-strictly exogenous explanatory variables, using Generalised Method of Moments (GMM) estimators. Two examples are presented: a simple autoregressive model for investment rates and a basic production function. The paper highlights the importance of identifying parameters in dynamic panel data models, especially when dealing with persistent series and potential serial correlation in errors. It discusses various estimation methods, including first-differenced GMM, and evaluates their performance in different contexts. The paper also addresses the limitations of these methods, particularly in cases where the time dimension is small, and emphasizes the need for careful consideration of moment conditions and the validity of assumptions. The paper concludes that GMM estimators, particularly system GMM, are effective in handling dynamic panel data models with persistent series and provide more accurate estimates compared to other methods. The paper also discusses the importance of testing the validity of additional moment conditions and the implications of using weak instruments in such models.Stephen R. Bond's paper discusses econometric methods for dynamic panel data models, focusing on microeconomic data with a large number of individuals or firms observed over a small number of time periods. The paper emphasizes single equation models with autoregressive dynamics and non-strictly exogenous explanatory variables, using Generalised Method of Moments (GMM) estimators. Two examples are presented: a simple autoregressive model for investment rates and a basic production function. The paper highlights the importance of identifying parameters in dynamic panel data models, especially when dealing with persistent series and potential serial correlation in errors. It discusses various estimation methods, including first-differenced GMM, and evaluates their performance in different contexts. The paper also addresses the limitations of these methods, particularly in cases where the time dimension is small, and emphasizes the need for careful consideration of moment conditions and the validity of assumptions. The paper concludes that GMM estimators, particularly system GMM, are effective in handling dynamic panel data models with persistent series and provide more accurate estimates compared to other methods. The paper also discusses the importance of testing the validity of additional moment conditions and the implications of using weak instruments in such models.