Heterogeneous Three-Dimensional Panel Data Models

Latif SHA, Youssef AH and Mohamed AM

Published on: 2023-01-28

Abstract

Panel data models exist in a variety of sizes and shapes depending on the properties that are affected by units and time. The model is referred to as a heterogeneous panel data model when all parameters vary across units and/or time. It's possible to get inaccurate findings if slope heterogeneity is ignored. Panel data models with more than two dimensions are referred to as multi-dimensional panel data models and the parameters in these models may be heterogeneous in one or more dimensions. The standard Delta test was used previously to evaluate slope consistency for big two-dimensional panels, also the F-test and Swamy's test were used in the case of four-dimensional panel data. The literature does not have any suitable tests for slope homogeneity in case of three-dimensional panel data, so the Delta, Heteroskedastic and Autocorrelation robust tests were applied in this paper. These tests, in contrast to the F-test, can be used with static or dynamic panel models and with heteroskedastic errors and that is why we will not use the F-test. We used an actual data from the United Nations Industrial Development Organization (UNIDO). We applied the Cross-Sectional Dependence test and the results revealed a significant evidence of cross sectional dependence. It also showed slope heterogeneity, so we recommended mean group, common correlated effect mean group and augmented mean group estimators as proper estimation methods for heterogeneous three-dimensional panel data models in the presence of cross-sectional dependency. The results showed that the common correlated effect mean group estimator is more accurate than others because it allows for cross-sectional dependency.