Analysis of the Quality of Twelve-Month Forward Inflation Forecasts
- Forecasts of inflation 12 months forward carry tremendous importance, and this study attempts to examine the quality of the forecasts obtained from the various sources and to create from them a more precise forecast that is absent any deviation.
- We examined the quality of the forecasts in the sample and outside the sample, and found that the best forecast is generated by the weighting that includes the forecasters’ projections and the expectations derived from the capital market, banks’ internal interest rates, and futures contracts. The weighting that includes just the capital market and futures contracts also has the ability to provide a good forecast.
- We also found that when the examination is based on quarterly data, the Companies Survey improves the precision of the forecasts.
- The Business Tendency Survey explains inflation well within the sample, but its forecasting outside the sample is imprecise.
The forecasts of inflation 12 months forward carries great importance, since they present the public’s expectations regarding inflation, which reflect, inter alia, the public’s confidence in monetary policy. Policy makers take these forecasts into account even though it is more difficult to be precise with them than with forecasts for one month forward, since they relate to a longer horizon and are therefore more exposed to shocks. We found that in recent years, there is a high correlation between 12-month inflation forecasts and actual inflation, but there is a fixed upward deviation.
Forecasts of inflation in the coming 12 months are obtained from a number of sources: the capital market, professional forecasters, the banks’ internal interest rates, Last Price quotes from futures contracts, the Business Tendency Survey, and the Companies Survey. Figure 1 shows the forecasts from the various sources compared with actual inflation.
This paper examines the forecasts in two ways. First, an in-sample examination reflects the extent to which the forecasts “predicted” the data that were obtained. This is used to examine the quality of the forecasts derived from each source separately and the quality of the forecasts derived from their various weights. The second is an out-of-sample examination that looks at the extent to which the forecasts can predict the data that have still not been obtained. This is used to examine the quality of the forecasts derived from the weight of the sources in two ways—when the size of the sample is variable, and when the size is fixed (in a moving window).
The in-sample examination shows that for most of the sources, the forecasts exceed the simple average. The Business Tendency Survey provides the most precise forecast, but there is a fixed upward deviation since the interceptor in the regression is negative and statistically significant. The professional forecasters are also generally precise, and when the examination is conducted without an interceptor, their forecasts are the most precise. We also examined a variety of possible weighting scenarios and found that they improve the forecasts obtained from each of the sources separately, and that most of the interceptors are statistically significant. The best forecast is generated by the weighting that includes the capital market, the forecasters, the banks’ internal interest rates, and futures contracts, and by the weighting that includes only the capital market and the forecasters.
The out-of-sample examination is more important than the in-sample examination since in the end, we are interested in forecasting ability and not in post-facto explanatory ability. We found that there are two weightings that are good for out-of-sample forecasting— (1) the capital market and futures contracts, and (2) the capital market, professional forecasters, the banks’ internal interest rates and futures contracts.