Nominal interest rates increased significantly in the US and euro area following the post-pandemic inflation surge, yet economic activity and labour markets remained resilient. This column finds that accounting for policy expectations implies a larger increase in the natural rate of interest, r*, than standard approaches suggest. In the US, r* increased by around two percentage points and around 1.5 to 2 percentage points in the euro area. These increases are largely explained by changes in convenience yields. Thus, monetary policy stances were considerably less restrictive than commonly believed and this helps explain the soft landing in economic activity.
Judging by several metrics, the Federal Reserve and the ECB delivered one of the largest policy tightening cycles in modern history during the post-pandemic inflation surge, yet both economies avoided a sharp contraction in economic activity. The dominant explanations emphasise unusually strong aggregate demand supported by expansionary fiscal policies (Bernanke and Blanchard 2025, Hodge et al. 2025, and Giannone and Primiceri 2024), and that adverse supply-side shocks stemming from global supply-chain disruptions and higher energy prices were not so disruptive to economic activity as central banks thought these headwinds would dissipate quickly and chose not to hike rates for some time (Erceg et al. 2024).
Yet these explanations leave an important puzzle unresolved: if monetary policy after the 2022-24 tightening became as restrictive as standard estimates suggest, why did economic activity and labour markets remain so resilient? A growing literature argues that part of the answer may lie in a reassessment of the natural rate of interest, r*. Recent evidence points to a material increase in natural real interest rates following the pandemic (Baker et al. 2023, IMF 2023, Brand et al. 2024, Del Negro et al. 2026). If r* rose alongside policy rates, then monetary policy was tightening relative to a moving benchmark rather than a fixed one. Understanding whether, why, and by how much r* increased is thus central to interpreting the post-pandemic inflation surge and assessing the true stance of monetary policy. Our recent paper (Farkas et al. 2026) contributes to this debate by showing that estimating the natural rate of interest when accounting for policy expectations implies a substantially larger rise in r* in the euro area and the US than standard approaches suggest. Our findings imply that the post-pandemic monetary policy stances were considerably less restrictive than commonly believed and help explain the soft landing achieved on both sides of the Atlantic.
Our paper (Farkas et al. 2026) studies how expectations about future monetary policy affect estimates of the natural rate of interest, r*, in the US and the euro area. The natural rate is the real short-term interest rate consistent with output under flexible prices in the long run. Because it is unobservable, it must be inferred through an economic model. This makes estimates sensitive to the model’s structure and, crucially, to the information used to identify anticipated monetary policy. A misguided assessment of r* can produce an equally misguided assessment of the monetary policy stance, since it depends on the gap between the actual real policy rate and the natural rate.
Our paper’s main contribution is to incorporate expectations of future policy rates directly into a medium-scaled dynamic, stochastic, general equilibrium (DSGE) model for both the US and the euro area to sharpen the identification of r^*. The model includes nominal rigidities, financial frictions, anticipated monetary policy shocks, and convenience yields associated with the safety and liquidity of debt securities. The convenience yield is the non-pecuniary value investors attach to safe and liquid assets. When demand for these assets increases, investors accept a lower return in exchange for safety and liquidity. A higher convenience yield therefore depresses the safe real interest rate and lowers r*.
Besides conventional macroeconomic data we augmented the observable data set with overnight index swap rates, long-term government bond yields, corporate bond spreads, and long-term inflation expectations. These observables helped distinguish anticipated policy interventions from convenience yields and structural forces that affect r*. We found that even though the natural rate is driven primarily by persistent productivity movements and convenience-yield shocks, the inclusion of anticipated policy shocks materially changes the assessment of r*. At first glance, it seemed counterintuitive, as anticipated monetary policy shocks (by model design) do not directly alter r*. Nevertheless, policy expectations change the estimated natural rate indirectly by changing how the model allocates observed movements among competing structural shocks.
The major empirical result, shown in Figure 1, is a significant post-pandemic increase in r* on both sides of the Atlantic. With forward-guidance information, the estimated natural rate rises from about -0.5% to 1.5% in the US and from about -2% to nearly 0% in the euro area. Thus, r* increases by around two percentage points in the US and around 1.5 to 2 percentage points in the euro area. Models that omit expected policy rates imply smaller increases, while the workhorse Holston et al. (2017, 2023) model suggests that r* remained broadly unchanged.
Figure 1 Natural rate estimates across model specifications
a) United States
b) Euro area
Source: IMF staff calculations.
Looking at the driving forces of the natural rate, from 2020 onwards, the benchmark model estimates an increase in r* of 1.9 percentage points in the US, of which 1.5 points are attributed to convenience-yield innovations. In the euro area, the estimated increase is 1.7 percentage points, with convenience-yield innovations contributing 1.8 percentage points. The recent increase is therefore not primarily a conventional trend-growth story; it largely reflects a reversal of forces that compressed safe real rates after the Global Crisis.
The main policy implication is that the post-pandemic rise in nominal interest rates overstates the intensity of effective monetary tightening if the natural rate was rising simultaneously. If r* had remained unchanged, as suggested by the workhorse Holston-Laubach-Williams model, the increases in policy rates would have produced a much larger positive real rate gap. This helps partially explain why economic activity remained resilient despite dramatic policy-rate increases beginning in 2022. The rise in the natural rate worked against the tightening impulse. The stance became less restrictive than would be inferred under a near-constant r* assumption.
This is demonstrated with our estimated models with a counterfactual exercise. Instead of allowing r* to follow the rising path, we assume it followed the flatter Holston-Laubach-Williams path in a counterfactual scenario. The actual and counterfactual real rate paths are shown in Figure 2. These counterfactual scenarios with flatter r* paths produce a real monetary policy stance that eventually becomes about two percentage points tighter in the US and slightly more than 1.5 percentage points tighter in the euro area, respectively.
Figure 2 Actual and counterfactual real rate gaps
a) United States
b) Euro area
Source: IMF staff calculations.
The macroeconomic effects of the alternative tighter real interest rate paths are substantial, as shown in Figure 3. Had the natural rate not risen, the tighter real rate gap would have lowered output (as deviation from trend) by nearly 4% and 2% in the US and the euro area, respectively. Under the unchanged-r* path, US growth would have been approximately zero from 2022 onwards, while euro-area growth would have remained near -2% for a sustained period.
Figure 3 Quarterly output and output growth paths under the unchanged-r* counterfactual
Source: IMF staff calculations.
Overall, our work highlights the importance of accounting for changes in r* when assessing the monetary policy stance. In particular, we find that expectations of future policy rates have an important impact on r* estimates via information on their anticipation effects.
Source : VOXeu
