Given the market turmoil, it appears that the breadth and depth of Donald Trump’s tariffs surprised all but the most pessimistic of observers. Accounting for the surprise 90-day discount on Wednesday to all countries except China — as well as the hike to 125 per cent on China — his proposals are the equivalent of a 23-percentage-point increase in the average US effective tariff rate, to 25 per cent. That’s the highest it has been since 1909.

The announcement does not change much in terms of the sheer magnitude of these tariffs. If they stay in place beyond 90 days, they will cost the average US household $4,400 in purchasing power this year, according to the non-partisan Budget Lab at Yale, raising the spectre of a recession in 2025. It is certainly reasonable to raise one’s recession probability in light of the past week, especially for anyone who had not pencilled in the possibility of high tariffs. But is a recession now more likely than not?

We economists need to be both sober in our analysis and humble in acknowledging the uncertainty in the data. And right now, if you asked me what the impact of tariffs will be, based on available information, I would say this: bad but not recessionary. If I’m wrong, however, it’s likely to be worse than I thought, not better. The signals from the bond market in the past 72 hours are only making me more anxious.

Consider growth in gross domestic product. The Budget Lab’s estimate of a one-percentage-point hit to US real GDP growth in 2025 from the tariffs is a substantial headwind to the US economy, the equivalent of losing $300 billion. But if US growth was going to be 2 per cent this year — in line with most economists’ estimates in January — then a reduction of that size need not lead to a recession. Growth of 1 per cent is still growth, albeit weak and unimpressive growth.

Likewise, conventional macroeconomic rules of thumb say that a one-percentage-point reduction in real GDP growth is roughly the equivalent of a rise in the unemployment rate of 0.5-0.6 of a percentage point. Again, serious — but based on the latest figures, this implies an unemployment rate of about 4.8 per cent at the end of the year. In the past 75 years, the lowest post-recession peak of the unemployment rate was 6.1 per cent.

What about present data? The National Bureau of Economic Research declares US recessions based on the monthly performance of six measures: payroll and household employment, real income, consumer spending, industrial production and real sales. I use a statistical tool called a dynamic factor model to isolate the common-growth component present in all six of these factors, plus the unemployment rate. As you can see, the common factor tends to fall below zero right before a recession starts. Through March of this year, this model has been solidly above zero, suggesting these underlying measures are more strong than they were in past recessions.

What about that uncertainty I mentioned earlier? Here’s where I explain how imperfect economic models are and why I think the effects of the tariffs are more likely to be worse, not better, than expected.

First, macroeconomic models do a decent job of capturing the direct impacts of policies, but they are not good at capturing the additional damage caused by policy uncertainty — the chilling effect on consumer behaviour or business investment from not knowing where a policy will ultimately land. That uncertainty, in addition to the substance of the tariff policy itself, is helping drive existing market turmoil.

Second, the lack of evidence in the data so far is not dispositive. Tariffs had barely bitten by March; according to the US Treasury, gross customs duty revenue was only $900 million higher in March than it was in February. Had the tariffs that the administration announced in February and March been fully in effect in March, revenue should have been several billion dollars higher at least.

Third, some indicators are already flashing red. The Atlanta Federal Reserve’s GDPNow forecast for the first quarter projects GDP at -2.8 per cent, or -0.8 per cent after adjusting for distortions caused by gold. Real personal consumption expenditures fell 7.1 per cent annualised in January and did not fully recover in February, growing only 1.2 per cent. True, January was before the tariff hikes, but the expectation of tariffs may have played a significant role in this behaviour, as consumer durables spending grew in November and December in anticipation of tariffs.

Fourth, while tariffs by themselves may not be recessionary, other factors in tandem with tariffs may be sufficient to push the US over the edge. Less immigration is likely to reduce growth, and federal spending cuts may also. And then, of course, there are whatever unknown additional shocks the world has in store.

None of this, I concede, is reassuring: it’s cold comfort to say that tariffs are likely to be painful but not quite debilitating. The bigger problem is that if economists are wrong — which has been known to happen — then the possibilities become more catastrophic.

• Ernie Tedeschi is the Director of Economics at the Budget Lab at Yale. Until March 2024, he was the Chief Economist at the White House Council of Economic Advisers