It’s now back to its pre “Liberation day” level, but still far below the 2022 level, after the Russian invasion of Ukraine. We will monitor closely the possible spillovers of energy prices. If ever these consequences happened to be lasting and propagating – i.e. affecting underlying inflation and inflation expectations –, we could possibly adapt our monetary policy: to put it short, no automatic monetary reaction, but “monetary agility” on which I will elaborate further.
Over the longer term, other supply shocks are likely to occur more frequently. Extreme climate events are also likely to disrupt supply chains. Information technologies are developing at a fast pace, with AI and quantum computing likely to be the next technological shocks that will possibly increase productivity. More frequent supply shocks will not mean per se higher inflation, but very probably a more volatile inflation, which could weaken the anchoring of inflation expectations.
Because the current equilibrium is unstable, we must above all avoid being complacent. As we have now entered a territory close to the neutral stance, the natural rate R* becomes a less meaningful guidepost. The debate is no longer about navigating by the “stars” to wonder whether we are restrictive or accommodative. It is about how to temper volatility in a world of uncertainty. Here agility and laying out a clear reaction function play a crucial role.
II. Agile but predictable
Fair enough about agility, and nobody would recommend rigidity. But let us acknowledge that it raises two serious questions.
a/ Agility does not mean that we should rush to come back to the 2% target at short notice, in case of inflation volatility. We have a medium-term orientation around our symmetric objective, which implies we do not have to respond to deviations from target at one point in time. Instead, we have to react to dynamics that risk pushing inflation off target: what matters is whether a deviation from target is more likely to increase or to decrease . And medium term means within our projection horizon.
b/ Agility should not imply that we are unpredictable or indecisive. In the past, when we had strong forward guidance, our actions were described as being on “autopilot”. Today, we are sometimes accused of the opposite – “flying blind” or making “ad hoc decisions.” I do not believe we are doomed to wander between these two extremes: it is time to return to a concept that has been less used in recent years – the central bank reaction function. We need to clearly state not what we will do, but how we think and will act, in a “readable” way.
2.1 The central bank reaction function
What is the Eurosystem reaction function? Since 2023, our reaction function has been based on three criteria: the inflation outlook; trends in underlying inflation; and the transmission mechanism.
When inflation was surging and monetary policy was tightening rapidly, it was natural that the second and third criteria were emphasized. With high uncertainty about the inflation outlook, it was important to ensure that inflation expectations did not become deanchored and the underlying inflation criteria was this backstop.
Likewise it was important to assess how quickly changes in policy rates were transmitting to borrowers and ultimately to the demand and supply of credit.
Indeed, the transmission lags have been at least as swift and satisfactory in this latest cycle as in previous ones: each rise in the deposit facility rate (DFR) was almost fully transmitted to bank loans to businesses and households within a year. Loan volumes in the euro area weakened sharply starting from the end of 2022, with a peak response 18 months after a rate hike. However, we must remain vigilant today to the broader transmission of monetary policy actions to financial conditions. In particular, we need to be careful that international spillovers via higher term premia in long term yields and the exchange rate do not counteract the desired monetary stance. I will come back to it.
Inflation projections and the use of scenarios
That said, now that inflation has returned to target, with inflation expectations well anchored, we can put more weight on the first criteria, the inflation outlook.
But inflation projections rely on various assumptions. Developing scenarios and applying informed judgment about possible future outcomes can assist policymakers. But scenario analysis is no panacea. First, we can only analyse in detail a few scenarios at a time, in practice one or two as we did early June with a severe and a mild scenario on trade. We will inevitably overlook certain shocks that do appear. Second, it might be difficult to assign probabilities to scenarios and therefore to use them for policy risk management, especially in periods of ‘Knightian’ uncertainty – not quantifiable.
Theoretically appealing approaches, like model averaging, are difficult to apply in practice. Instead, we could consider methods inspired by robust control. Specifically, robust control aims at avoiding the worst possible outcome, for example easing monetary policy exaggeratedly when an inflationary shock materializes. But this leads to making decisions and ruling out options based on hypothetical scenarios, irrespective of their likelihood. So scenarios will be an important aid to decision-making but should be used judiciously rather than systematically: not in all occurrences of ECB forecasts, and subject to discretion rather than rules in their policy implications.
Three principles of monetary agility
Given all this, let me give you several thoughts on what agility means:
Speed of assessment. In rapidly changing circumstances or complex shocks, there is always the temptation to wait for additional data to obtain a fuller picture. To be agile is to do this analysis faster by exploiting new real-time data sources, incorporate new models from the latest research, survey new staff analyses on risks and contingencies, and avoid the excess of group thinking. For example, our analysis of the effects of tariffs have benefited from using daily shipping data and freight prices.
Symmetric and state-contingent policy paths. We rightly stress that our monetary action should be as “persistent and forceful” as needed . But is there a contradiction between persistence in policy and agility? No, not if signals are understood to be state-contingent and we avoid being boxed in. If the state changes then so does the policy, including possibly its symmetric direction both ways. Persistence is only relevant in certain states of the world.
Acting not stronger but sometimes faster. And what about a forceful response? If required, it is not compulsory to go in incremental steps. Of course, in the presence of new shocks, central banks may prefer a wait-and-see attitude, in line with Brainard’s principle of mitigation : if the effects of policies are uncertain, it is better to react less. But, conversely, waiting too long for events to materialize can result in substantial losses. The Banque de France research paper entitled “The pitfall of cautiousness in monetary policy” has shown that this attitude can lead to a cautionary bias that undermines inflation expectations . This may imply shorter and swifter interest rate cycles, although any adjustments still have to be orderly.
2.2 What should we do now?
High uncertainty surrounding the global economic outlook obviously calls for humility about our ability to predict future inflation and monetary action. Our primary duty today is to assess risks. I mentioned earlier oil prices, while avoiding to draw automatic monetary conclusions. Overreacting, or conversely systematically “looking through”, would be contrary to the necessary agility. We should monitor at least two other sets of data:
Do inflation expectations reflect a risk of lasting spillover? They don’t, so far, as medium term market based measures of inflation expectations have recently remained clearly below 2% in 2026. The probability that inflation falls below 1.5% in 5 years is around 40% according to the pricing of options on inflation.