Purchasing power parity, balanced growth, and volatility forecasting: An application of recent developments in time series analysis.

摘要

In this thesis, I use recent advances in statistics and econometrics in an effort to re-test some well-known theoretical propositions, examine whether these new techniques support the theory, provide models that are better fitted to describe and forecast economic time-series. The Purchasing Power Parity theory is tested using the Fisher and Seater (1993) and King and Watson (1997) methodologies and strong evidence in support of PPP is found. I use the general class of ARCH/GARCH processes to model financial times series in an ARIMA framework and the best fitted models outperform traditional ARIMA models in terms of the forecast variance. Finally, I test the balanced growth theory and try to estimate a money demand function using the Johansen and Juselius (1993) methodology. I do not find evidence in support of the balanced growth theory and a stable money demand function, and these results are not sensitive to different monetary aggregates that are constructed according to recent index number theory.