FORECASTING TURNING POINTS IN METROPOLITAN EMPLOYMENT GROWTH RATES USING BAYESIAN TECHNIQUES

In this paper, I adapt to the regional level a Bayesian approach developed by Zellner, Hong, and Min (1990) to analyze forecasts of turning points in a multicountry setting. The techniques applied to a regional setting treat the individual metropolitan areas in the same way that Zellner, Hong, and Min treated countries. The findings in this study indicate that the techniques and models employed by Zellner, Hong, and Min work just as well at the metropolitan-area level as they did in the multicountry setting. The best models, from those studied here, forecast around 70 percent of the downturns and 80 percent of the upturns correctly, which compares favorably to the performance of the same techniques in Zellner, Hong, and Min.