As you know, the Taylor rule penalizes deviations from a given target value of inflation and output in a, well, rule based manner. However, most of the data is only available in preliminary versions and when it comes to a decision, the value of any data used at this point in time is quite questionable.

Below you see the actual fed fund rate (blue) and the Taylor rate according to the first estimate of the output gap (green) and the Taylor rate according to the final estimate of the output gap (red). Both gap data are OECD calculations.

Three things become obvious: First, the Taylor rule implied interest rate is closer to the actual fund rate if instead of revised/final data the actual real time data is taken into account. Second, for some periods this effect is more pronounced then in others. Third, during the last years of the data set (2002-2006) we get a quite good fit of the rate using real time or final data.

Obviously, if the weight on the uncertain output gap is reduced (which is a common recommendation under data uncertainty), the Taylor rate using actual data and final data are nearly congruent. See second picture.