Finally a downward surprise
(We apologize for the delay but in the last few days we have experienced some IT issues, as the folks in London know..)
Finally, a downward surprise. The September flash core HICP figure came in well below expectations (NSA level 117.07 vs 117.47 expected by us). As usual, we wait for the final reading to estimate the movements of the distribution. Having said so, today’s report is the first significant downward surprise of the year and marks a deviation from the (NSA) 2022 path. While the deceleration is certainly welcome, the puzzling aspect is that the surprise came from core services, while the trajectory of core goods was roughly in line with our expectations. Without the details it is hard to say something; in a sense, the report looks “too good to be true”, but we would be happy to be proven wrong by the details. Nevertheless, we took some signal and revised down our near-term forecast (to be confirmed with the final data). The NSA unchained level of the series by year, including our forecast, can be seen here. The medium-term models also took a bit of signal.
Good news but still too little to change monetary policy. In the week in which the 10y BTP tested 5% everything seems of a secondary order. Today, we got undoubtedly good news. However, we need more because the old issues are still here. Even considering today’s downward surprise, the ECB/NCBs staff forecast for 2024 continues to appear low, and (most importantly) the medium-term models continue to be above 2% at the end of the forecasting horizon. Translated: a step in the right direction but in order to change the message of monetary policy we need more. Until then, we continue to believe that the long-end of the curve will remain under stress.
A PDF containing all relevant charts can be downloaded here.
Details
Finally, some good news. We estimate that in September core HICP prices grew 30bps MoM (sa) or 3.6% at annual rate (chart of MoM is shown below). (note: the seasonal adjustment in September is more problematic than usual and the filters are probably overstating a bit the true pace. Please, get in touch for more details). This brings the 3m/3m (ar) to 4.4% (Figure 1). On a quarterly basis, we estimate that in Q3 core HICP prices have grown at 4.2% at annual rate. The 3m/3m (ar) is below the YoY, signaling that the annual variation should moderate further in the coming months. Acquired core HICP inflation for 2023 is 4.9% (or 5.0% on NSA data).
Figure 1. Core HICP metrics.
Note: the figure shows the metrics of core HICP. All figures are seasonally adjusted. “ar”” stands for “annual rate”. The 3m/3m and the 6m/6m are chained (that is, using the US BEA method).
ECB/NCBs staff forecast for 2023 looks reasonable after today’s downward surprise. Figure 2 shows history of the MoM (sa) of core HICP, together with our own judgmental forecast and the average consistent with the latest ECB/NCBs staff projection. The YoY is expected to average 5.1% in 2023, still a bit below the estimated model (see next section). Conditional on this forecast, the “carry-over effect” for 2024 is 1.7%/1.8%. The NSA (unchained) level of the series by year, including our forecast, can be seen here.
Figure 2. Core HICP MoM (sa, %).
Note: the figure shows the MoM (sa, not annual rate) of core HICP prices. The blue line shows history and our own bottom-up forecast. The red-dashed line shows the forecast consistent with the latest ECB staff macroeconomic projections.
Medium-term model-based forecast is revised down. Today’s data imply a lower starting point of the model (Q3), and a slightly lower persistency of the process. The model forecast using the output gap as a measure of “slack” (Figure 3 right panel below) is at 4.8% (Q4/Q4) in 2023 (average of four quarters at 5.2%), 3.7% in 2024, and 2.6% in 2025. Using the unemployment rate (Figure 3 left panel) produces a similar forecast in 2023 and 2024 but a higher forecast in 2025 (this happens because in the latest ECB/NCBs staff projections the Urate goes sideways in the forecast).
Figure 3. Core HICP: YoY forecast of our “main” Phillips curve model.
Using Urate as a measure of “slack”
Using outputp gap as a measure of “slack”
Note: the figures show the YoY forecast of our “main” Phillips curve model for core HICP price inflation. The confidence intervals are calculated from the estimated parameters distribution. Last quarter in-sample is 2023:Q3.
Implications for the ECB staff and monetary policy
October is on autopilot. “For longer” message unlikely to change in December. As mentioned, today brought good news for the ECB/NCBs staff, as the level of core HICP appears to be in line with the latest macroeconomic projections for 2023. On our side, before concluding about the September data we need to see the details, as the deceleration in services is probably too good to be true and might be offset going forward. In any case, even taking today’s reading at face value, the big picture for monetary policy remains the same, as the forecast at the end of the medium-term continues to be above target, as discussed in London this week. For this reason, in our view and estimate it is unlikely that the message of monetary policy will change any time soon. For now, it is “for longer”.