Real-time US inflation forecasts

 

By E. Foutou

I have been working on inflation forecasting as part of my undergraduate research work with Professor Dimitrios Thomakos. This is a first post, and hope that I will continue with them, on presenting some real-time US inflation forecasts. I use data from the FRED database, on CPI-based inflation (from Consumer Price Index for All Urban Consumers: All Items in U.S. City Average) and a number of explanatory variables to build ARDL-type of models. These explanatory variables are based on economic supply and demand (industrial production, retail sales and services), unemployment, and food and energy prices. All these variables are in the related literature for inflation forecasting. The particular variables I used are below.

1. Advance Real Retail and Food Services Sales
2. Advance Retail Sales: Retail Trade
3. Consumer Price Index for All Urban Consumers: Energy in U.S. City Average
4. Consumer Price Index for All Urban Consumers: Food at Home in U.S. City Average
5. Consumer Price Index for All Urban Consumers: Food in U.S. City Average
6. Consumer Price Index for All Urban Consumers: Services Less Energy Services in U.S. City Average
7. Global price of Corn
8. Global price of Energy index
9. Global price of Food index
10. Global price of Natural gas
11. Global price of Wheat
12. Industrial Production: Consumer Goods
13. Industrial Production: Manufacturing (NAICS)
14. Industrial Production: Manufacturing (SIC)
15. Industrial Production: Total Index
16. Unemployment Rate

These variables are entered in groups that I have created and then these groups are combined, lagged and estimated in ARDL-type models, followed by model reduction and elimination of non-significant terms. What is left after model reduction is used to perform real-time, multi-step ahead forecasting. I generated many model-based forecasts which I then average and provide standard errors as well. My first results are summarized below, from the end of this month (June) till the end of 2023 – note that the last actual inflation value in FRED is at 4.13% for May 2023:

Screenshot 2024-01-29 143944

Table 1. real-time US inflation forecasts

My work is done in Gretl for the forecasts of the models, which are then are (easily) combined in Excel! Looking forward for their first evaluation next month!

image

Figure 1. real-time forecasts of US CPI-based inflation