The Monetary Policy Committee's reaction function: an exercise in estimation

Almost all economists know the story about the (drunk) person searching for his lost wallet in the night under the lamp-post, not because that was the most likely place to have dropped his wallet, but because that was where the light was. I shall argue here that this story is fitting in the case of Taylor-type Central Bank reaction functions. These functions indicate how Central Banks might adjust interest rates in response to deviations of current inflation and current output from some desired level, so that, it = a + b(ðt - ð*) + b2yt + b3it-1 (1) where i is the nominal interest rate, ð the current rate of inflation, y is the estimated output gap, and the final term (b3it-1) is usually included to account for the empirical evidence of auto-correlation in the time path of interest rates.