This paper proposes the Pooled Mean Group (PMG) estimator for dynamic heterogeneous panels, which allows short-run coefficients and error variances to differ across groups while constraining long-run coefficients to be identical. The PMG estimator is compared with the Mean Group (MG) and Dynamic Fixed Effects (DFE) estimators in two empirical applications: aggregate consumption functions for 24 OECD economies (1962-93) and energy demand functions for 10 Asian developing economies (1974-90). The PMG estimator is shown to provide consistent estimates of long-run coefficients without assuming identical dynamics across groups, which is particularly useful when the number of groups (N) and time periods (T) are large. The paper derives the asymptotic distribution of the PMG estimator under both stationary and non-stationary regressor assumptions. It also discusses the implications of long-run homogeneity and the effects of heterogeneity on the estimation of short-run parameters. The PMG estimator is found to be more efficient than the MG estimator in cases where the number of groups is large, and it allows for more flexible dynamic specifications across groups. The paper also addresses issues related to the consistency of the PMG estimator and the effects of heterogeneity on the estimation of long-run coefficients. The results show that the PMG estimator provides more accurate estimates of long-run coefficients and is less sensitive to heterogeneity in short-run dynamics. The paper concludes that the PMG estimator is a useful tool for analyzing dynamic heterogeneous panels, particularly when the number of groups is large.This paper proposes the Pooled Mean Group (PMG) estimator for dynamic heterogeneous panels, which allows short-run coefficients and error variances to differ across groups while constraining long-run coefficients to be identical. The PMG estimator is compared with the Mean Group (MG) and Dynamic Fixed Effects (DFE) estimators in two empirical applications: aggregate consumption functions for 24 OECD economies (1962-93) and energy demand functions for 10 Asian developing economies (1974-90). The PMG estimator is shown to provide consistent estimates of long-run coefficients without assuming identical dynamics across groups, which is particularly useful when the number of groups (N) and time periods (T) are large. The paper derives the asymptotic distribution of the PMG estimator under both stationary and non-stationary regressor assumptions. It also discusses the implications of long-run homogeneity and the effects of heterogeneity on the estimation of short-run parameters. The PMG estimator is found to be more efficient than the MG estimator in cases where the number of groups is large, and it allows for more flexible dynamic specifications across groups. The paper also addresses issues related to the consistency of the PMG estimator and the effects of heterogeneity on the estimation of long-run coefficients. The results show that the PMG estimator provides more accurate estimates of long-run coefficients and is less sensitive to heterogeneity in short-run dynamics. The paper concludes that the PMG estimator is a useful tool for analyzing dynamic heterogeneous panels, particularly when the number of groups is large.