TESTING MONETARY EXCHANGE RATE MODELS WITH PANEL COINTEGRATION TESTS

The monetary exchange rate models explain the long run behaviour of the nominal exchange rate. Their central assertion is that there is a long run equilibrium relationship between the nominal exchange rate and monetary macro-fundamentals. Although these models are essential tools of international macroeconomics, their empirical validity is ambiguous. Previously, time series testing was prevalent in the literature, but it did not bring convincing results. The power of the unit root and the cointegration tests are too low to reject the null hypothesis of no cointegration between the variables. This power can be enhanced by arranging our data in a panel data set, which allows us to analyse several time series simultaneously and enables us to increase the number of observations. We conducted a weak empirical test of the monetary exchange rate models by testing the existence of cointegration between the variables in three panels. We investigated 6, 10 and 15 OECD countries during the following periods: 1976Q1-2011Q4, 1985Q1-2011Q4 and 1996Q1-2011Q4. We tested the reduced form of the monetary exchange rate models in three specifications; we have two restricted models and an unrestricted model. Since cointegration can only be interpreted among non-stationary processes, we investigate the order of the integration of our variables with IPS, Fisher-ADF, Fisher-PP panel unit root tests and the Hadri panel stationary test. All the variables can be unit root processes; therefore we analyze the cointegration with the Pedroni and Kao panel cointegration test. The restricted models performed better than the unrestricted one and we obtained the best results with the 1985Q1-2011Q4 panel. The Kao test rejects the null hypotheses – there is no cointegration between the variables – in all the specifications and all the panels, but the Pedroni test does not show such a positive picture. Hence we found only moderate support for the monetary exchange rate models