Tweet

April 24, 2025 — Everyone wants to know if a recession is imminent.  But the most popular recession indicators are not necessarily the best places to look for the answer.

The question is front and center in the US now because of concerns over actions by the Trump Administration.  Six factors currently might trigger a downturn:

• the trade war,
• crash in the US stock market,
• chaotic cuts in USG spending,
• a US fiscal crisis arising from government shutdown, debt limit stand-off, or credit downgrading,
• the blocking of net immigration, and
• increased uncertainty and risk (driven especially by the erratic rollout of US tariff policy), as reflected in sharp increases in the VIX and  bond premia.

1. Leading indicators

How can we tell if a recession is near?  Consider as an analogy a sailing ship navigating through heavy fog, watching out for land, fearing to founder.  If the lookout sights certain birds, it is more likely that land is near.  (There are charts, but the captain refuses to use them.)  Analogously, some leading indicators may signal a heightened probability of recession. But only a probability; nothing is certain.

• An inverted rate yield curve – that is, a long-term interest rate that falls to or below the short-term interest rate – seems to have a track record of predicting recessions. The 10-year bond rate fell below the 3-month Treasury bill rate in March, but currently in April there is little difference. In any case, logically, the rule cannot tell us all that much. It just reflects financial market expectations that the Fed might cut short-term interest rates in the future, which in turn reflects market expectations that economic activity might falter.  There are alternative ways to get at market expectations of recession more directly.
• For example, consumer confidence is a useful leading indicator. It can help predict consumption spending. Two long-established measures of consumer confidence, conducted by the University of Michigan and the Conference Board, showed a sharp loss in confidence in March, as Trump’s ever-shifting tariff policies took hold. The Michigan survey’s index of consumer sentiment has been falling since the start of the year.  On April 11, it plummeted another 11 % (“fell off a cliff”), reaching its second lowest level since records began in 1952, a level far below the average in past recessions.   Another source, the New York Federal Reserve Bank’s Survey of Consumer Expectations, reported on April 14 that households’ year-ahead expectations about their financial situations had deteriorated further.
• Business confidence is particularly relevant for firms’ decisions whether to invest in plant and equipment and to hire workers. Moody’s Analytics’ weekly Business Confidence Index globally “collapsed” on April 4, as confusion soared over Trump’s so-called reciprocal tariffs.
• It is natural to look to professionals who make their living forecasting the economy, collected in several surveys. Year-ahead forecasts by approximately 50 of them are aggregated monthly by the Blue Chip Forecasters Consensus.   Two more surveys are by the Survey of Professional Forecasters and  the Wall Street Journal,  The latter, published April 17, showed a mean forecast for Q1 of 0.44% annualized and much higher recession probabilities than at the start of the year.  But the WSJ and SPF surveys are published only once a quarter and the forecasts can become stale rapidly.
• Economists generally hold that where someone puts their money is a more reliable reflection of their true beliefs than what they say. Indeed, there is statistical support for the superior forecasting ability of on-line betting, so-called prediction markets. They currently show the probability of recession as having risen far above an average year’s 15%. One, Polymarket, shows the odds of a recession almost tripling since February, peaking when the “reciprocal tariffs” were instituted April 6, and now near 53 % as of April 24.  Another prediction market, Kalshi, has fully tripled since Inauguration Day and shows 54 % odds as of April 24.

2. The Timeliest measures of economic activity

The question of leading indicators is different from the question of what are the first tangible signs that a recession may have actually begun. In the analogy of the ship navigating in fog, it would be the first encounter with rocks that may (or may not) constitute a landmass.

• Pundits sometimes refer to the “Sahm Recession Indicator” as if it is a leading indicator which might be useful for forecasting. But, in reality, it is designed only to be a snapshot of a recession that has already begun. The Sahm rule says that the economy is in a recession if a three-month moving average of the unemployment rate (U3) rises by at least 0.50 percentage points (relative to its low-point of the previous 12 months). Claudia Sahm was motivated to find a criterion that could be used in real time to trigger, for example, government stimulus payments.  It is more useful in real time than GDP in that it is reported more frequently and with fewer subsequent revisions.  On the other hand, when reacting to a decline in demand, firms may hold off on the decision to lay off workers, especially at times of uncertainty like the present, until after they have accumulated some unwanted inventory and/or cut back the length of their work week and the volume of their production.  (In any case, unemployment remains low so far, by historical standards.)

A number of other timely measures of actual economic activity are useful.

• Purchasing Managers Indices (PMI) survey private sector firms, asking for example whether they have witnessed new orders rise or fall over the preceding month. The US manufacturing PMI from the Institute of Supply Management fell to 49 in March 2025. Readings below 50 indicate contraction.
• The Census Bureau reports retail sales These data are the earliest measure of household consumption to become available, constituting about 2/3 of the economy. The March figures indicated continued growth, albeit driven in large part by elevated motor vehicle sales to consumers trying to get ahead of impending tariffs.
• A period of negative growth in GDP is the single-variable criterion that comes closest to defining what a recession is (two consecutive negative quarters, in the case of European countries). The GDP statistic is only reported quarterly, and with substantial lags and substantial revisions. So-called nowcasts aggregate together relevant indicators that are available up to the minute, as of a particular date, to give the best estimate of GDP in the contemporaneous quarter.   Perhaps the most prominent US nowcast is the Atlanta Fed’s GDPNow, which relies on the PMI, industrial production, retail sales, personal income, home sales, and other announcements. Its estimate of first-quarter GDP growth fell off a cliff at the end of February, from above + 2% to below -2 %, a contraction.  (Seasonally adjusted annual rates.)  The numbers in the first quarter were distorted by unusual foreign exports of gold to the US; but even adjusted for the gold anomaly, GDPNow shows an April fall in growth to slightly below zero.
• The Bureau of Economic Analysis releases the initial official estimate of quarterly GDP about four weeks after the end of the quarter. However, because the statistics are substantially revised subsequently, they cannot be relied upon in real time. The first estimate of 1st-quarter real GDP growth is set to be released tomorrow, April 25.

3. Ex post determination

After the fact, one may wonder, was that really land, or just some shoals or rocks that weren’t big enough to count as land? As noted, in many countries, the criterion for determining a recession is two consecutive quarters of negative growth. But, again, this is reported only after the end of the two quarters and even then is vulnerable to revision.

The official arbiter of US recessions is the NBER Business Cycle Dating Committee.   It waits until all the data are in, typically a year or so after the fact, before declaring a turning point.  The variables that the NBER looks at include gross domestic output (an alternative way of  measuring GDP), real personal income less transfers (PILT), nonfarm payroll employment, real personal consumption expenditures, manufacturing and trade sales adjusted for price changes, employment as measured by the household survey, and industrial production.  But the task of determining the start and end of a recession after the fact is very different from the task of forecasting the recession ahead of time.

4. So, where are we now?
   

This does not help the sailor who is currently navigating through fog. Based on current information, one could put the odds of a US recession as high as 55 % for the coming year – in line with the prediction markets – and even higher for the next four years.  But you never really know until you actually run aground.

[A shorter version was published in Project Syndicate.  Comments can be posted at Econbrowser.]