Une question de confiance

Le 2 avril, les États-Unis devaient publier leurs plans tarifaires détaillés, mais Scott Bessent les a présentés comme le début des négociations : l'incertitude commerciale restera élevée pendant des mois, avec un prix à payer en termes de croissance, tant pour les cibles des nouveaux tarifs que pour les États-Unis eux-mêmes. Selon les estimations pour 2018-2019, l'incertitude pourrait à elle seule coûter 2 à 4 % des dépenses d'investissement des entreprises américaines. Du côté des consommateurs, notre modèle économétrique confirme que les Américains continuent d'être plus déprimés que ce que l'état « objectif » de l'économie ne le suggérerait. C'était vrai sous Biden et cela reste vrai aujourd'hui. Nous soupçonnons que le bombardement constant d'informations sur les droits de douane n'aide pas. Le sentiment compte : même en contrôlant les variables « dures », la confiance des consommateurs a un impact mesurable sur les dépenses. Il est peu probable qu'elle soit suffisamment puissante, à ce stade, pour précipiter les États-Unis dans la récession, mais combinée à la baisse des cours des actions et à l'attentisme en matière de dépenses d'investissement, le ralentissement significatif de la croissance américaine que nous attendions pour 2026 pourrait se matérialiser plus rapidement.

L'incertitude était également très présente dans l'esprit des membres du FOMC la semaine dernière. Alors que le marché s'est concentré sur certains aspects conciliants des déclarations de Jay Powell, un examen attentif des détails des prévisions de la Fed suggère que, bien que la banque centrale ne soit pas encore prête à passer à un scénario de base de stagflation, le risque de ralentissement de la croissance et, plus probablement, les risques de hausse de l'inflation, autour d'un scénario central encore favorable, sont pris au sérieux.

Bien sûr, l'incertitude commerciale est un autre obstacle pour l'Europe. Pourtant, en Allemagne et en France, la source de l'augmentation significative de l'incertitude globale en matière de politique économique est antérieure à la guerre commerciale 2.0 et est plus nationale. La confirmation du succès du nouveau gouvernement allemand dans le changement décisif de la politique budgétaire pour les dix prochaines années, et en France le retour d'un leadership politique plus clair dans un contexte où la défense et les affaires étrangères dominent, pourraient compenser certains des obstacles externes. Cependant, la séquence reste problématique : les tarifs sont pour aujourd'hui, les dépenses de défense sont pour demain.

Does confidence matter?

Scott Bessent has described a potentially lengthy process to get “hard news” on US tariffs. Indeed, on April 2, US supplier countries will be “notified their number”, i.e. a target for tariffs the US would levy on the products they import from a given location, determined by a broad range of hurdles the US administration considers US products face there. But this will only open a negotiation phase with no hard deadline at this stage. To quote him verbatim, “We are going to go to them and say, 'Look, here's where we think the tariff levels are, non-tariff barriers, currency manipulation, unfair funding, labor suppression, and if you will stop this, we will not put up the tariff wall”. We suspect this “reciprocal” approach would not prevent the US administration from implementing on top of it more “special” tariffs, in the same vein as the one of steel and aluminium. Uncertainty is unlikely to recede quickly. To gauge the immediate impact on the US economy, the confidence channel is likely to dominate until a more precise sense of the final bill for importers emerges.

Let us start with the impact on business investment. Matteo Iacoviello from the Federal Reserve developed a news-based index of “trade policy uncertainty” (TPU), tracking the number of references to trade tension in the press. The current uncertainty level is much higher than during the trade war with China in 2018-2019 (see Exhibit 1), probably because the rates would be higher (up to 60% on Chinese products in Donald Trump’s initial communication) and cover many more countries, while the high level of “to-ing and fro-ing” today (for instance on Canada and Mexico) was not observed during trade war 1.0. Iacoviello has produced different models quantifying the impact of TPU on business investment, both at the firm and at the aggregate level (see the link to his paper here).

What is particularly valuable in his approach was his inclusion of related variables, e.g. actual tariff volatility, not just news about them, which allows to isolate the effect of “pure” uncertainty. His conclusion at the time of the first trade war was that trade-related uncertainty “cost” the US between 1 and 2% of business investment, with a very rapid pass-through (the maximum effect materialises between one and two quarters after the uncertainty shock). Given the magnitude of the current rise in TPU, something closer to 2-4% of business investment would now have to be considered. This is no small change. Note that this ignores other sources of uncertainty, for instance doubts as to the capacity of the US administration to deliver the tax cuts pledged during the campaign (e.g. taking corporate tax from 21% to 15%).

Then there is the consumption channel. We have already explored in Macrocast how consumer confidence has already taken a hit since the new administration came in. Still, a common critique of these surveys is that “sentiment” may not have a significant predictive capacity on actual spending decisions. Indeed, consumer confidence could be driven by exactly the same variables which explain private consumption, households merely synthetising information on the labour market, inflation, disposable income, and asset prices. But precisely, what is interesting today is that consumer confidence in the US is significantly lower than it should be given the “objective” state of the US economy. There may be some proper information content in the confidence indices.

To substantiate this, we estimated a model of the Conference Board consumer confidence index, with the change in the unemployment rate, PCE inflation, the 3-month changes in real disposable income and S&P500 index, all lagged by one month, as explanatory variables. They all have the expected sign, and all come with coefficients different from zero at the 5% significance level. We initially tried to include interest rates in the equation, but they failed the significance test. We estimated the relationship over 30 years (1989-2019). The model captures all key inflexion points (see Exhibit 2), except for the pandemic period. The disconnect at the time of Covid is unsurprising: concerns over health and the prospects of lockdown/normalisation exceeded “usual” economic worries, while the level of government support was so high that most consumers could ignore the dramatic rise in unemployment rate at the time.

Still, what we find interesting is that, since the reopening of the economy, consumer sentiment has constantly been lower than the model predictions. To visualise it better we use the prediction error of our model as a “confidence gap”, in Exhibit 3. The massive labour market rebound, and strong real income growth were probably offset by the inflation shock which impacted confidence beyond the elasticity estimated over the three pre-pandemic decades. The price-related confidence shock was persistent: even as inflation started to recede in the first half of last year – prompting the Fed to start cutting – households remained “grumpy”. ​ It is this “grumpiness” – at odds with the objective situation of the US economy - which probably made Donald Trump’s victory possible in November 2024. Yet, what is equally interesting is that consumer confidence remained lower than what the state of the economy would suggest after Donald Trump’s election, by 1.5 standard deviations as of February 2025.

The last step in our exploration consists in checking if consumer confidence retains some explanatory power for actual household spending when injected in a canonical model of private consumption. It does. The coefficient associated to the Conference Board headline confidence index remains significantly different from zero in a model including changes in real income, the unemployment rate and equity prices. We find that a one standard deviation in consumer confidence would immediately reduce consumer spending by 0.6% annualised.

Would this be enough, by itself, to precipitate the US economy into a recession? Probably not. It would likely take a significant deterioration in the labour market to get us there (in our model a rise in the unemployment rate of one standard deviation would have three times the effect on consumption as the same shock on confidence). Still, combined with the effect on business investment consistent with the Iacoviello model, and the adverse effect from the current decline in US equity prices, a significant slowdown in US domestic demand could be seen in the coming quarters if uncertainty and generic “stress” does not fall back. ​

Fed’s navigation without visibility

Uncertainty does not spare the Federal Reserve (Fed). There is now a stagflationary hue to the Fed’s forecasts for 2025 released last week, with a downside revision to GDP (from 2.1% in December to 1.7%) and an upward revision to core inflation (from 2.5% to 2.8%). Yet, the median trajectory was largely unchanged from the December batch for the end of the forecasting horizon, and the Fed still expects to hit its inflation target in 2027, which, combined with some dovish aspects of Jay Powell’s statements, kept the market happy. However, it is clear that the level of concern at the FOMC around a still benign baseline is rising.

The range of views on growth in the coming years has widened (see Exhibit 4). A handful of members took extreme views. One now has a GDP forecast for 2026 in or near recession territory (0.6-0.7%) - against none in December, while at the other end of the spectrum, just like in December, there is still one “uber-optimist” eying 2.5%. But movements around the centre of the distribution were interesting. In December, the most popular view in the FOMC for 2026 growth (2.0-2.1) was held by 12 members. Now, the most popular view is slightly lower, 1.8-1.9% and shared by only 8 members. Views are thus generally a bit more pessimistic, but without strong convergence around one particular pace.

Movements on inflation were clearer. While the most popular view for 2026 PCE is still 2.1%-2.2%, just tantalisingly close the Fed’s target, and is shared by the same number of members as in December (10), a new cluster of 7 members has appeared at 2.3-2.4% (only 2 members were there in December). And for 2027, even if the median forecast has not changed at 2.0%, a “worried cluster” of 5 members now think the Fed will miss its target. So, in a nutshell, while the stagflationary scenario has moved the median only for this year, this is a risk which is increasingly being contemplated at the Fed across the whole horizon.

This is still a subtle message though, and as long as the baseline itself does not move for the end of the forecasting horizon, the Fed cannot easily justify a change of forecasts for its policy rate. The median FOMC projection for policy rates was thus kept unchanged, at two cuts for this year. In truth, it is difficult for any policymaker to make relevant predictions given the unusually range of policy decisions which could come from the White House in the weeks and months ahead. A point on increasing uncertainty around the outlook was added to the second paragraph of the FOMC statement. The central bank does not ignore the noises from the White House (Powell explicitly linked the revisions to the tariffs) but is unwilling to draw definitive conclusions at this stage. The Fed is firmly in observation mode, Powell reiterating they are in “no hurry” and in a “great position to wait until they have more clarity”.

Still, we were surprised that the level of “stagflation concerns” easily discernible in the forecasts did not find a clearer way into Jay Powell’s rhetoric last week. The discussion of tariffs, following many questions from journalists, was very guarded. Jay Powell said that the Fed would not necessarily react to a tariff hike, as it could be seen as one-off level shock, and went as far as to use the word “transitory” – which we thought had been devalued since the post pandemic inflation shock proved more persistent than the Fed had expected. Of course, he also said that the FOMC would react if inflation expectations were to move up. This is the textbook answer. But the tone was probably more conciliatory than expected, and we were also surprised by how Powell dismissed some of the recent signals from consumer surveys on inflation expectations (qualifying the recent surge in the Michigan survey’s price component as “an outlier”), as he preferred to focus on market-based pricing. We would detect there a willingness not to antagonise the White House, even if there is no blank check to Donald Trump from the central bank.

The Fed also slowed down further its pace of Quantitative Tightening, earlier than expected, from USD25bn to USD5bn. This move was presented as being disconnected from the monetary policy stance proper. Powell pointed to some emerging signs of pressure in the money market in the context of the debt ceiling looming as a justification. This was surprising since there is for now no sign whatsoever of money market tensions emerging in interest rates, and bank reserves remain abundant (a point made by Governor Waller who dissented on this decision). The usual gauge may however be inoperative in the current environment. Indeed, as part of its “extraordinary measures” to delay the hard stop on the debt ceiling, the Treasury is curtailing debt issuance and depleting its general account below the usual values to meet its obligations. Mechanically, this feeds into banks’ reserves, thus artificially raising the level of liquidity on the money market. Yet, the Treasury would probably be constrained to resume issuance at a fast clip when the debt ceiling issue is resolved – this will need to happen by mid-summer – which could then trigger some pressure on liquidity. The Fed is probably being pre-emptive there.

Yet, if the purpose of Jay Powell was to avoid a direct confrontation with the US President, it failed, judging by the message Donald Trump sent on Truth Social a few hours later after the press conference. The President explicitly called on the Fed to accommodate as “tariffs start to transition their way into the economy”. This suggests that the White House is intent on taking measures which they know will – at least transitorily – slow down the economy, in line with the messaging from the last few weeks on the need to accept some short-term pain as the administration is reshaping the American (and global) economy. Tactically, we think this is also preparation for the next steps of the interplay with monetary policy: if the economy does indeed decelerate, and the Fed does not immediately respond with a swift rate cut, the blame will be squarely laid on the door of the Federal Reserve by the White House.

Mind the European sequence

While we wait for the “number” and the result of subsequent negotiations, trade uncertainty will of course weigh on spending decisions in Europe as well, before the “real” effect of likely tariff hikes hit. Yet, in some key European economies, mounting trade uncertainty was only one element in a broader rise in policy uncertainty. Indeed, when looking at the broad policy uncertainty index developed by Baker, Bloom and Davis, policy uncertainty in France and Germany stood at two to three standard deviations above the long term average before attention started to focus on trade war 2.0, a bit before Donald Trump’s election (see Exhibit 6).

Trade war 2.0 added to the general angst, but the root is more domestic: doubts around the growth model in Germany and political instability in France. Reassuringly, while it remains historically high, the policy uncertainty index receded in these two countries in February. We could see some further improvement ahead, given Germany’s decisive strategic shift on fiscal policy, while focus on defence and foreign affairs is putting the French President back at the centre of decision-making in Paris, and calls for his resignation, and hence additional political instability, have faded.

In a nutshell, we now need to balance these two antagonistic forces: trade uncertainty on the external side, and a sense of decisiveness in the EU which was so far lacking. We suspect that when the “number” is known, businesses – and possibly consumers too – will focus on the trade issue, since the materialisation of the defence-related fiscal push will take time to materialise. Still, at least, there are no longer only downside risks to the Euro area’s medium term growth trajectory.