Alternative General Method of Moment Estimators in Dynamic Panel Data Models
| dc.contributor.advisor | Emmanuel Gabreyohannes | |
| dc.contributor.author | Tegodie Hibstu | |
| dc.date.accessioned | 2025-09-05T23:13:14Z | |
| dc.date.available | 2025-09-05T23:13:14Z | |
| dc.date.issued | 2024-09-02 | |
| dc.description.abstract | Dynamic panel data models make it possible to address dynamic economic relationships through the inclusion of a lagged dependent variable among the explanatory variables. In the presence of lagged dependent variable as a regressor, however, the least squares-based estimators may not be consistent. The generalized method of moments (GMM) estimators are the most popular in dynamic panel data estimation. Two crucial issues in GMM estimation of dynamic panel models are the choice of the initial weighting matrix and the problem of instruments proliferation. In this study we propose alternative one-step and two-step system GMM estimators that utilize sub-optimal initial weighting matrices together with reduced instruments set (specifically lag-limited and partially collapsed instruments). Comparison of the performance of the proposed estimators against the Blundell-Bond system GMM estimator; the conventional system GMM estimator; and the sub-optimal system GMM estimator based on all available instruments was undertaken in terms of mean absolute bias, root mean squared error/standard deviation and coverage probabilities through Monte Carlo simulations. The small sample performance of the proposed estimators was also assessed using panel data of employment for manufacturing firms in Ethiopia. The simulation studies show that sub-optimally weighted one-step GMM estimator based on collapsed instruments is the least biased for moderate and large values of the variance ratio of the individual-specific effects to that of random shocks. In terms of precision, sub-optimally weighted GMM estimator using all available instruments was found to be the most efficient, followed by that utilizing collapsed instruments set for moderate and large values of the variance ratio. As the value of T increases, the bias reduction from the latter outweighs the efficiency gain from the former in relative terms. The results also revealed that sub-optimally weighted two-step system GMM estimator adopting partially collapsed instruments outperforms the standard GMM estimator in terms of both bias and RMSE for large T and large variance ratios as the coefficient of the lagged dependent variable gets close to zero. Under these scenarios, the system GMM estimators based on reduced instruments were also found to perform well in terms of coverage probabilities. Moreover, for relatively small time dimensions and large variance ratios, the sub-optimal system GMM estimator based on all available instruments performs best in terms of precision. However, there is no considerable gain from the use of suboptimal initial weights matrix in combination with reduced instruments as the process approaches a random walk for both one-step and two-step system GMM estimation. Our general conclusion is that the performance of sub-optimally weighted system GMM estimators which utilize collapsed as well as untransformed instruments seems promising. | |
| dc.identifier.uri | https://etd.aau.edu.et/handle/123456789/7392 | |
| dc.language.iso | en_US | |
| dc.publisher | Addis Ababa University | |
| dc.subject | Alternative General Method | |
| dc.subject | Moment Estimators | |
| dc.subject | Dynamic Panel | |
| dc.subject | Data Models | |
| dc.title | Alternative General Method of Moment Estimators in Dynamic Panel Data Models | |
| dc.type | Thesis |