Utilizing Financial Market Information in Forecasting Real Growth, Inflation and Real Exchange Rate
Jyväskylä University School of Business and Economics; University of Oulu - Department of Economics
University of Turku - Department of Economics
January 22, 2008
We propose a simple approach to macroeconomic forecasting based on a model that uses only financial market data via some fundamental long-run equilibrium conditions both for the macro economy and financial markets. Using an open economy extension of the famous Gordon (1962) growth model for the stock market we derive a system of equations for forecasting the future values of real growth, inflation and real exchange rate. The extension is based on introducing first the Fisher (1930) and Euler equations to the analysis. For an open economy context, where also the purchasing power parity (PPP) and the uncovered interest rate parity (UIP) have to be taken into account, we suggest a forecasting system of three simple equations as the final form, all based on current financial market information in the form of dividend yields and short-term interest rate.
In the empirical part we utilize standard time series analytical tools, i.e., the autoregressive (AR) and vector-autoregressive (VAR) models and rolling regressions for forecasting. The data cover monthly observations from the U.S., U.K., Euro-zone (Germany before 1991) and Japan, all starting approximately from the end of the second oil crisis and ending in January 2007. We control for some major structural changes in the economies, like the U.K. stepping out from the ERM in September 1992, and the observed liquidity trap in Japan after 1999. The main concern in the empirical part is in the out-of-sample forecasting performance of the derived model when compared to the performance of the benchmark models, like AR representations of the individual macro time series.
Our empirical results indicate that the role of simple financial market information both in the forms of dividend yields and nominal interest rate is highly important. Even though it would seem to be somewhat difficult to obtain economically relevant long-run relationships between the analyzed variables in our system of equations, the forecasting power of financial market variables in our model is strong also in the long run. The dividend yield appears to be a relevant forecasting variable both for real growth and inflation in many cases, whereas for forecasting the time-varying long-run trends in the real exchange rate, the dividend yield spread between foreign and home country seems perhaps less important when using linear techniques in the analysis. Based on our results we strongly stress the importance of including some measure of stock market performance in macroeconomic forecasting systems.
Number of Pages in PDF File: 59
Keywords: Stock market, forecasting, macro economy, exchange rates
JEL Classification: G15, G28
Date posted: January 22, 2008
© 2016 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollobot1 in 0.406 seconds