By E. Foutou
I am very pleased to share the new and updated inflation forecasts for the US CPI. Based on my last post, here, the forecast for January of 2024 was between 2.15% to 2.77% with an actual number (available in FRED) of 3.106%, and for February was between 1.68% and 1.84%. Inflation persistence is well-known as is the property of many models to (locally) mean-revert. In this post I have the updated forecasts from February to September of 2024 and I initialize also the use of the adaptive learning algorithm, see here, based on the limited data of my last post. First, in Table 1 below I provide the updated forecasts and then I use a variation of the adaptive learning algorithm to (well!) adapt them based on past performance.
Table 1. actual values and forecasts of US CPI inflation, forecasts computed February 2024
If I use the previous forecasts my previous forecasts from June 2023 to January 2024 and apply the adaptive learning algorithm I find a value of the adaptive parameter of 0.5844 (you can apply the easiest variation of the method with Excel's solver) which changes the February average forecast from 1.85% to 2.35%, with range of 2.10% to 2.60%. Next months we shall see more updates and whether the adaptive learning algorithm proves more useful (it should!)