Granger Causality in Macroeconomics Panel Data

The shortage of macroeconomics time series poses a crucial problem which can be avoided by extending the time series including cross-sectional data to panel data. The first panel data models were built for microeconomic applying large number of cross-section units and short time series. With random sampling being an essential prerequisite for microeconomic panel data, the observations are independent. In macroeconomic data, we face a problem with non-random sampling and thus the possibility of dependent variables. This problem can be solved with the factor and spatial models. In this paper, we focus on the factor models and answer to the question how energy supply influences GDP in the long term period. To test stationarity, co-integration and Granger causality of GDP, energy supply and capital formation, we follow methodologies by Pesaran and Ederhardt. It was found out that energy and capital have positive long-term effect on GDP and vice versa the GDP and capital have positive long-term effect on energy consumption.