Objective research and reviews to aid investing decisions
 Looking for information on debt consolidation loans? Want to know more about reverse mortgages? |
In this entry, we summarize further analysis of our Inflation Forecast, testing its reliability across its 12-month duration. To accomplish this analysis, we apply our inflation forecasting methodology to generate forecasts for each month from January 2000 through June 2005, a total of 66 12-month forecasts. Next we find the differences between forecasted inflation rates and actual inflation rates (forecast errors) for each of the 12 months in each forecast. Then we calculate the standard deviation of these errors for each of the 12 forecasted months. We find that:
Each pink squiggle on the following chart depicts one of the 66 inflation forecasts described above. The black line shows the actual inflation rate, and the dark red squiggle is the current inflation rate forecast. Visual inspection shows that the forecasts lag changes in the actual inflation, as expected from a technical forecasting approach. Forecasts often help identify the locations, but not the exact sizes, of future hills and valleys. The first month of each forecast tends to be the most accurate because it branches from actual historical data, whereas subsequent months of the forecast branch from forecasted data. Forecasting errors therefore generally grow from forecasted month 1 through month 12.
The following table provides the standard deviation of forecasting errors for each of forecasted months 1 through 12. As expected, this measure of inaccuracy grows from the early months of a forecast to the late months.
As a check, we repeat the steps described above for the inflation rate forecasts from each January during 1990-2005. This small sample of 16 forecasts produces similar standard deviations of errors by forecast month.
In summary, our technical inflation rate forecasting methodology generates forecasts continually in catch-up mode, with accuracy generally degrading from early months to late months for each forecast.
We have incorporated these results as error ranges in our Inflation Forecast, which includes a discussion of alternative methods of inflation forecasting. Inflation forecasting uncertainties carry over into the stock market projection of our Real Earnings Yield Model.
