SMA vs. ECB inflation forecasts: Which forecasts drive interest rate expectations in the euro area?

Abstract

This Policy Brief uses data from the ECB’s Survey of Monetary Analysts (SMA) over 2021–2025 to examine the formation of euro area inflation and interest rate expectations. The analysis shows that market expectations for inflation, as reflected in the SMA, often diverge significantly from the ECB’s official (B)MPE inflation projections, with discrepancies typically widening at longer horizons. Despite this, both the SMA and the (B)MPE exhibited comparable forecast accuracy during this turbulent period. The core finding for monetary policy is that SMA participants’ interest rate expectations are more responsive to inflation forecast errors relative to the (B)MPE than to those relative to their own SMA inflation forecast. This suggests that market analysts form their interest rate expectations primarily by observing how well inflation outcomes conform to the ECB’s published inflation forecast. When inflation surprises on the upside, interest rate expectations are revised upward by more than one-for-one, confirming the market’s belief in the credibility and inflation-stabilising nature of the ECB’s monetary policy reaction function.

Introduction

Monitoring the evolution of inflation and interest rate expectations is crucial for central banks, as their combined dynamics determine the level of real interest rates and thus the degree of monetary policy restriction in the economy. In the euro area, the ECB’s Survey of Monetary Analysts (SMA) provides an important source of information on the expectations of analysts working at major financial institutions regarding the future evolution of monetary policy and the economy. The SMA is conducted eight times a year prior to each monetary policy meeting of the ECB Governing Council, which ensures its results are a regular and policy-relevant input to the Governing Council’s assessment (Brand and Hutchinson, 2021).

This Policy Brief examines the dynamics of inflation and interest rate expectations in the euro area since the SMA’s launch in June 2021. The analysis addresses three key questions: first, how do SMA inflation expectations compare with the ECB’s official (B)MPE inflation projections; second, how does the forecasting performance of the two sources stack up against realised inflation; and finally, which set of inflation forecast errors – the SMA’s or the (B)MPE’s – is more important for steering SMA interest rate expectations?

Inflation and interest rate expectations in the SMA, 2021–2025

Over the course of 2021–2025, the period for which SMA results are available, inflation and interest rate expectations in the euro area underwent large revisions.

As Figure 1 shows, the SMA respondents repeatedly revised up their short-term inflation expectations as inflation in the euro area increased sharply in 2021 and 2022. These revisions were large and reflected the unanticipated nature of the post-pandemic inflation surge. For example, while inflation was expected to only mildly overshoot 2% in the June 2021 survey round, it was already expected to reach almost 8% in the June 2022 round. Once inflation reached its peak in the fourth quarter of 2022, the revisions to the short-term inflation outlook became noticeably smaller. Thus, while the strength of the increase in euro area inflation surprised the SMA respondents – to a large degree due to the unanticipated nature of the inflationary shocks – the subsequent disinflation process evolved broadly in line with the ex-ante expectations.

In parallel to these inflation developments, SMA respondents’ interest rate expectations also underwent substantial revisions. While the median expectation for ESTR in early 2024 was at only 0.5% in the June 2021 round, its value was already at close to 1.2% in the June 2022 round and at 3.7% in the June 2023 round. The magnitude of these upward revisions to interest rate expectations was closely related to the magnitude of inflation surprises, a relationship that will be discussed later in Section 4 of this Policy Brief. Once inflation surprises moderated, the revisions to the interest rate expectations also became more moderate.

Figure 1. SMA expectations for euro area headline inflation and ESTR by survey vintage

How different are SMA inflation expectations from the ECB’s (B)MPE inflation projections?

The survey design of the SMA allows for a direct comparison with the ECB’s (B)MPE projections, as both are conducted in parallel, which ensures a comparable information set.1 It can therefore be argued that both the SMA and the (B)MPE should capture the same state of the world in a given forecast vintage, and the differences in their inflation forecasts should therefore represent genuine disagreement in the expected path of inflation conditional on the same information set.

Figure 2 shows that considerable differences in forecasted inflation between the SMA and (B)MPE are more the norm than the exception and that the degree of disagreement typically widens with the forecast horizon. While one-quarter-ahead forecasts largely agree, with an average absolute difference of only 0.1 percentage points, the disagreement expands significantly at the four-quarter-ahead horizon, where the average absolute difference reaches 0.4 percentage points.

Furthermore, it is noteworthy that the SMA tended to exhibit, on average, noticeably lower inflation forecasts than the (B)MPE over 2021–2025, in particular at forecast horizons more than two quarters ahead.

Figure 2. Deviation of SMA inflation expectations from (B)MPE inflation projections by survey vintage

Which inflation forecasts have better performance – the SMA’s or the (B)MPE’s?

The notable and systematic differences in inflation projections between the SMA and (B)MPE lead to the question which forecast source has historically proved more accurate. As Figure 3 shows, the 2021–2025 period was exceptionally challenging for inflation forecasters, resulting in pronounced forecast errors for both the SMA and the (B)MPE, especially at longer horizons. Both sources exhibited similar error patterns: they generally underpredicted inflation during the sharp increase of 2021–2022 and subsequently overpredicted it (though to a substantially lesser degree) during the disinflation from 2023 onwards.

Assessment of forecast performance using the Mean Absolute Error (MAE) shows that the historical accuracy of the SMA and (B)MPE inflation forecasts was broadly comparable over the full sample (see Figure 5 in the paper). While the (B)MPE exhibited a slight advantage over the full sample, reflecting its marginally lower forecast errors during the peak of the post-pandemic inflation surge, the overall results indicate that there is no strong reason to favour one set of forecasts over the other for predicting actual inflation.

Figure 3. Forecast errors of SMA and (B)MPE inflation forecasts

Which inflation forecast errors are more important for steering SMA interest rate expectations?

Given the pronounced differences in inflation forecasts between the SMA and (B)MPE, but their broadly similar forecasting performance, this section examines which forecast errors are more important for steering interest rate expectations of SMA respondents. Do monetary analysts at major European and global financial institutions place greater weight on forecast errors relative to their own assessment of future inflation (SMA) or relative to the ECB’s assessment ((B)MPE)?

To address this question, I examine how SMA expectations for the future path of the ECB deposit facility rate (DFR) change between two consecutive quarters (q -> q +1) in relation to the realised one-quarter-ahead inflation forecast error, evaluated against the inflation projections made in the base quarter (q).

Figure 4 shows the estimated sensitivities from a regression-based analysis, which consistently show that the responsiveness of SMA interest rate expectations is both higher and more significant for the (B)MPE inflation forecast errors than for the SMA inflation forecast errors. This finding suggests that analysts participating in the SMA form their interest rate expectations primarily by observing how well inflation outcomes conform to the ECB’s published inflation forecast.

When inflation surprises on the upside, interest rate expectations are revised upwards, a finding consistent with conventional economic theory and the ECB’s policy reaction function. The largest estimated responses are for the Governing Council meetings that are five to eight meetings ahead of the base period. In these cases, the estimated sensitivity is clearly larger than 1, and it can thus be inferred that the interest rate formation process of SMA participants also conforms to the Taylor principle.2 This finding mirrors the conclusions of Bernardini and Lin (2024), who find that the SMA interest rate expectations aligned well with the prescription of a standard Taylor rule in the period 2022–2023.

These results thus suggest that the SMA analysts perceive the ECB’s reaction function as credible and inflation-stabilising. Inflation surprises lead to an immediate upward repricing of interest rate expectations along the forward curve that is stronger than the inflation surprise itself, implying an increase in real interest rates. Such dynamics contribute to a tightening of financing conditions that supports price stability, even in the absence of an immediate increase in the ECB’s (spot) policy rates.

Figure 4. The estimated sensitivity of SMA DFR expectations to inflation forecast errors

Conclusions

The analysis of the SMA data spanning 2021–2025 yields several key insights. First, the comparison of SMA inflation expectations with the ECB’s (B)MPE inflation projections reveals that differences are more the norm than the exception and that the disagreement between the SMA and (B)MPE tends to increase as the forecast horizon widens. Second, despite these differences, the forecasting performance of the SMA and (B)MPE was broadly comparable over the 2021–2025 period, although the (B)MPE performed slightly better during the post-pandemic inflation surge. Third, and most importantly, the analysis shows that SMA analysts’ interest rate expectations are more strongly steered by inflation forecast errors relative to the (B)MPE than by errors relative to their own inflation forecasts. Ultimately, this indicates that the ECB’s inflation forecasts act as an important anchor for market interest rate expectations.