What is the appropriate method for estimating productivity growth? The new growth theories emphasize issues such as learning by doing, externalities and spillovers that must be addressed in a dynamic framework. Traditional econometric works that estimate productivity growth rely almost exclusively on ordinary least squares (OLS) estimation, and concentrate on identifying factors which explain productivity growth such as government spending and openness. An alternative approach is to adopt an error correction model and for a Phillips and Loretan approach which are both designed to estimate the short-run speed of adjustment, lagged effects and feedback effects of the relevant variables. A dynamic approach better explains productivity growth than an OLS framework, since disparate short and long-run effects, lagged variables and feedback/simultaneity effects are significant. An OLS approach is found to produce biased estimates and a systematic underestimation of investment's effect on productivity growth. The structure is as follows. Section I discusses the development of theoretical and empirical models of productivity growth and identifies shortcomings in these approaches. Section II outlines the motivation and methodology for the models which are presented in Section III. Section IV presents the results from the different models. Section V concludes and summarizes the paper.