Long-Term Statistical Relationships in Political Science: An Analysis Based on Fractional Integration and Cointegration

In this paper we use new statistical methods to examine the connection between economic expectations and support for the government in the U.S.A. For this purpose, we analyse the order of integration for Congressional Approval and Economic Expectations from a fractional point of view. The results show that though both individual series can be specified in terms of fractional processes, the unit root cannot statistically be rejected, which is consistent with other empirical works in these variables. The possibility of cointegration is also examined from a fractional viewpoint, and the results strongly reject the hypothesis of cointegration. Thus, both series are nonstationary, and there is no linkage between them in the long run, according to an error correction equilibrium relationship.