Clustering analysis in panel data: An application on the ecological footprint of selected OECD countries

Panel data has become very common in recent years with the advantage of analyzing cross-section and time series data together. In panel data, which consists of the data of units over time of Individuals, firms, countries, etc. there may be differences between units. Neglecting heterogeneity in panel data models can lead to inconsistent parameter estimates. The most basic way to account for heterogeneity is to accept that the slope parameters are different. However, when the desired results are not obtained, dividing the panel data into subgroups and making separate estimations for each group will allow to obtain more consistent results. Similar subgroups can be easily created in univariate panel data. However, forming homogeneous groups in multivariate panel data requires statistical method. Therefore, panel data clustering analysis is used to create subgroups in multivariate panel data. Creating homogeneous subgroups with panel data clustering analysis is of great importance for the efficiency and consistency of parameter estimation. In this study, it is aimed to evaluate the environmental quality of OECD countries through the variables of ecological footprint, biological capacity, domestic income per capita, human capital and population density for the period 1971-2017. Considering that the panel data set is heterogeneous according to the units, clustering analysis was applied to the panel data. Parameter estimation was made for all countries and homogeneous subgroups obtained as a result of cluster analysis. Parameter estimations made on subgroups were found to be more effective and consistent than parameter estimations obtained for all OECD countries.