2023: The recession that never came

Twelve months ago almost every forecaster expected a US recession in 2023. Surveys in late 2022 put the probability at sixty-five per cent, an unusually high figure for a consensus view. As the year closes, the call has turned out spectacularly wrong. The American economy picked up speed and posted solid GDP growth despite the fastest tightening cycle in generations.

A consensus built on good evidence

The pessimists were not foolish. Between March 2022 and July 2023 the Federal Reserve lifted its policy rate from close to zero to a range of 5.25 to 5.50 per cent, a move of more than five percentage points in sixteen months. The Treasury yield curve had inverted in the middle of 2022, with two-year yields above ten-year yields, a signal that had preceded every American recession of the past half-century. Leading indicators were falling, and consumer confidence was weak.

The spring then seemed to confirm the gloom. Silicon Valley Bank collapsed in March, followed by Signature Bank and, in May, First Republic. A credit crunch among regional lenders looked like the channel through which higher rates would finally bite. Many forecasters pencilled in a downturn starting in the second half of the year.

What happened instead

The economy did the opposite. Output grew at an annualised rate of 4.9 per cent in the third quarter, driven by consumer spending. Unemployment touched 3.4 per cent in January, its lowest since 1969, and has stayed below 4 per cent all year. Inflation fell a long way without the rise in joblessness that the textbook said would be needed to achieve it.

The bank failures were contained quickly. Regulators guaranteed all deposits at the two failed banks, and the Fed opened an emergency lending programme that let banks borrow against their bond holdings at face value. Lending standards tightened, but the flow of credit slowed rather than stopped.

Why the models missed

Forecasters leaned on models calibrated to earlier cycles. They underestimated two cushions in particular: the excess savings households built up during the pandemic, and the ultra-low rates that companies and homeowners had locked in. Heavy federal spending added a third. The private sector was largely shielded from the first round of rate rises. The recession may simply be running late, with the usual lags stretched well beyond what history suggested.

The mortgage market shows the mechanism most plainly. Most American home loans carry a fixed rate for thirty years, and millions of households refinanced when mortgage rates dropped below 3 per cent at points in 2020 and 2021. When new mortgage rates climbed towards 8 per cent in October, those owners felt nothing in their monthly payments. The pain fell on would-be buyers, and home sales slumped, but the typical household budget was untouched.

Companies had done the same with their debt. Large firms issued long-dated bonds at low coupons during the pandemic, pushing their refinancing needs years into the future. For many of them, higher short-term rates raised interest income on cash holdings faster than they raised interest costs. In earlier cycles a larger share of corporate and household debt reset quickly with the policy rate, which is why tightening used to work faster.

The fiscal tailwind

Government spending pushed in the opposite direction to monetary policy. The federal deficit came to about $1.7 trillion in the fiscal year to September, a large figure for an economy at full employment. The infrastructure law, the CHIPS Act and the subsidies in the Inflation Reduction Act channelled money into construction, and spending on new factories rose sharply. When one arm of policy brakes and the other accelerates, models built around the brake alone will overstate the slowdown.

A long wave of immigration and a recovery in labour-force participation also expanded supply. More workers allowed the economy to grow while wage pressures eased, a combination that most recession models do not allow for. That supply side helps explain why inflation could fall while output rose.

Precedent for a soft landing

History offers at least one clean parallel. In 1994 the Fed doubled its policy rate to 6 per cent within a year, bond markets suffered a rout, and many expected a recession that never arrived. The economy slowed, the Fed trimmed rates in 1995, and the expansion ran for the rest of the decade. Soft landings are rare, but they are not unknown, and the 1990s version also followed a period in which balance sheets were in good shape.

The case for a late recession

The strongest counterargument is that the cushions are wearing thin. Savings built up during the pandemic have been run down, and poorer households appear to have spent theirs first. Repayments of federal student loans resumed in October after a pause of more than three years, taking money from younger borrowers. Delinquencies on credit cards and car loans have risen from very low levels.

The low-rate shield also erodes over time. Every year some corporate bonds mature and must be refinanced at today's yields, and commercial property owners with floating-rate loans are already under strain. Office buildings, hit by both higher rates and emptier desks, are the most exposed. If the lags are simply long rather than broken, the economy could still slow sharply in the year ahead.

Against that, the Fed has already stopped raising rates and its December projections show three quarter-point cuts next year. Inflation is close enough to target that the central bank can respond to weakness, which it could not do a year ago. A recession caused by a sudden shock remains possible, but one caused by the lagged effect of past rate rises becomes less likely each month that policy eases.

What would change the call

The labour market is the signal to watch. A rise of half a percentage point in the three-month average unemployment rate from its low of the previous year has, in the past, reliably marked the start of recessions, a rule of thumb known as the Sahm rule. Monthly payroll gains have slowed from their pace in 2022, and a run of weak reports would matter more than any survey. Bank lending surveys and delinquency data would give earlier warning of a credit squeeze.

Forecasters should enter the new year with less faith in their models. Balance sheets protected from rate rises beat the textbook in 2023, and the same protection may keep the economy growing for a while yet. The likeliest outcome is that growth slows to a more modest pace over the first half of next year without tipping into outright contraction, and that the forecasters who called a recession for 2023 shift it forward by twelve months.