This is a technical note that explains why the ECB staff forecast of core HICP in 2023 is implausible. The reason involves the notion of “carry-over effect” and “acquired inflation”. The intuition is simple: the ECB staff is projecting 2023 annual core HICP inflation at 4.2% but, as a matter of facts, it is already 3.0% in January and it can only grow in the rest of the year. The bottom line is that a more plausible forecast is at least 100bps higher than the ECB staff projects. (Stop here if not interested in the technicalities)
How annual inflation is calculated in the EA vs US
Eurostat computes annual inflation not as the BLS/BEA. In Figure 1 we have plotted a hypothetical core HICP price level in two consecutive years, a base year (the blue one) and the current year (the yellow one). The dots indicate the average of the price levels in each year. For a reason that will become clear in the next sections, we have assumed that in the current year the price level grows only in January and then remains constant throughout the year. As known, the BEA/BLS calculate annual inflation as the change between the blue December and the yellow December (or as the Q4/Q4), as indicated by the green line. On the other hand, Eurostat calculates annual inflation as the change between the two dots, as indicated by the red line (which is equivalent to the average of the YoY across months in the yellow year).
Figure 1. How annual inflation is calculated by Eurostat vs the BLS/BEA
The “carry-over effect” and “acquired inflation”
Using Eurostat approach, a portion of annual inflation is inherited from the previous year. Figure 2 illustrates how annual inflation (as computed by Eurostat) can be split into two contributions: the so-called “carry-over effect” and the contribution of the current year. As shown in Figure 2, the carry-over effect is the change between the average level of the base year (the blue dot) and the December level of the base year. Therefore, unless the MoM changes of the yellow months are negative (which is basically impossible for core inflation, seasonally adjusted) or the carry-over effect represents a lower-bound of annual inflation in the current year, as it is the portion of inflation “inherited” from the base year. On the other hand, the contribution of the current year is the change between the level of the series in December of the base year and the average of the current year (the yellow dot).
Figure 2. The “carry-over effect”
What is “acquired inflation” then? “Acquired inflation” refers to what annual inflation of the current year would be assuming that the level of the series remains constant for the remaining part of the year. In other words, “acquired inflation” is defined as the sum of carry-over effect plus the contribution of the current year up to the current month. For instance, after receiving the January (core HICP) report, “acquired inflation” can be calculated by assuming that core HICP does not grow in the remaining 11 months of the year (in which case the average of the level in the current year is simply equal to the level of the series in January, which is nothing but what we have plotted in Figures 1 and 2). By definition, on January 1st of each year “acquired inflation” is equal to the carry-over effect only, and on December 31st “acquired inflation” is annual inflation itself. The intuition remains the same: in each month, “acquired inflation” gives the lower-bound of what annual inflation will be at the end of the year.
(Note: in the US, “acquired inflation” is nothing but the cumulative sum of the MoM bps throughout the year. The issue is more complicated in the Euro area because of the way annual inflation is calculated)
Carry-over and acquired inflation in core HICP right now
Core HICP has a large carry-over from 2022. Figure 3 shows the level of core HICP (NSA as published by Eurostat, and seasonally adjusted in-house). The crucial point, which is incredibly lost in the debate, is that core HICP has a very large carry-over effect from 2022: 2.6%. Not only, but the January reading has pushed up acquired inflation by 0.4pp to 3.0%. In other words, unless the level of the series will now magically stop growing (as in Figures 1 and 2) or the 2023 annual core HICP inflation will be at least 3.0%. This is why we have been so vocal against the ECB staff forecast of core HICP: the 2022 “inheritance” is too big and prices are growing too fast at the margin.
Figure 3. Carry-over and acquired inflation for core HICP in January 2023
The ECB staff forecast is too low
A more reasonable forecast of core HICP is about 100-150bps higher than the ECB staff. Figure 4 shows the simulated path of core HICP in the rest of the year required to meet the ECB staff forecast (the red dot). In order to have an annual core HICP inflation at 4.2% as the ECB staff is projecting, the level of the series should magically deviate from its trend in February and grow at the pace of 20bps per month (which is unrealistic, given that the distribution of price changes is currently centered around a mean/median of 6% (ar)). Indeed, the current pace (that is extrapolating the January MoM to the rest of 2023), would imply an annual inflation about 200bps higher than the ECB projection. A more realistic forecast (shown in yellow) put annual inflation at 5.3%, in line with the model-based forecast we have circulated on HICP day.
Figure 4. Comparing ECB staff 2023 core HICP forecast to other projections
Conclusion
The data will win in the end, as usual. Given that the 2022 carry-over is 2.6% for core HICP and acquired inflation in January is already 3.0%, how can the ECB staff project 4.2% annual inflation at the end of 2023? Most importantly, how many ECB (and Euro system) staff economists are aware of these effects in the data? Are Lagarde, Panetta and Lane fully aware of the notion of carry-over and acquired inflation? To us, the answers to these questions are pretty clear. We have recently claimed that the ECB staff is in “La-La land”. Pretty soon, we expect the data to wake them up.