Testing the validity of Wagner's law in four income groups: A dynamic panel data analysis

Abstract

This study endeavors to examine the validity of Wagner's Law, which has received considerable attention in recent years. We develop a panel dataset of 20 countries, taking five countries from each income group defined by the World Bank, for the 1991–2018 periods. Five different versions of the law are tested using this dataset. We add further depth to the model by involve the government's revenue and the volume of trade as independent variables. We inquire into the subsistence of cross-sectional dependence. To determine the order of integration, we conduct LLC, IPS, and CADF tests. The results show that the dataset has I (0) and I (1) series, and no series is found to be of I (2). Then we perform the panel ARDL test, and calculate PMG and DFE estimates. We use the Hausman test to choose among the estimates. In each version of the law, the error correction term indicates the presence of both long-term associations within the variables and an economic convergence process. However, we find no evidence to support the law for any version. Additionally, we conduct the panel cointegration tests, such as Westerlund, Pedroni and Kao. These cointegration tests generate results accordant with the ARDL findings.

Keywords

• Wagner's law
• Public spending
• Government deficit
• Panel data
• Panel ARDL test