[Notes from my System Dynamics class on forecasting]
- We looked at the Livingston Survey of inflation forecasts made by panel of economists since 1946 and compared against actual inflation. Few interesting observations about the forecasts:
* There is attenuation - forecasts are not nearly as extreme as realized inflation.
* Delayed adjustment - missed turning points.
* Bias towards "good news" - adjustment was faster when inflation was going down.
- First order exponential smoothing does a better job than the Livingston panel in forecasting inflation over the last 50 years. As a matter of fact, first order exponential smoothing does better than any other model in the absence of any prior about the underlying data series.
2:42:32 PM
|