Fed policy, the Sahm rule and the risk of recession

Policy rules point to lower rates as inflation slowly returns to target, where a materially higher unemployment rate would demand faster action given the Fed currently forecasts little change in unemployment. Unemployment of just over 4% has almost triggered the Sahm rule threshold for recession, a rule-of-thumb that captures the post-WW2 experience of every sharp rise in US unemployment being driven by the job losses that are the hallmark of recession. The Fed responds forcefully to recessions, although the rise in unemployment to date has been driven more by employment failing to keep pace with the expanding workforce rather than broad job losses.

Over the past couple of years, the US economy has held up much better than either economists or the Fed had expected, contradicting the long-running signal from the earlier inversion of the yield curve that a recession was highly likely any time from H2 2023 through 2024.

One explanation for this resilience is that the neutral policy rate – or at least the short-run version of neutral – has increased from its pre-pandemic lows, perhaps because of shocks to the economy like continued easy fiscal policy, something that is supported by market pricing and Fed modelling.

Another explanation is that the overhang from COVID was longer than anticipated, drawing out the already long and variable lags of monetary policy. For example, job vacancies have only recently broadly returned to pre-pandemic levels, raising the possibility that a further material decline in vacancies could finally be reflected in a sharp rise in the unemployment rate.

These two explanations are not mutually exclusive. It may be that high interest rates are finally slowing the economy, but that a higher neutral rate means that average interest rates are likely to be higher than pre-pandemic levels for some time.

Recently, the unemployment rate has started to signal an increased risk that the US could finally enter recession, increasing from a multi-decade level of about 3½% last year to just over 4%. 

At this level, the unemployment rate signals that the labour market is broadly back in balance, matching economist and Fed estimates of the NAIRU that range between 4-4¼% (the Congressional Budget Office estimate of the NAIRU is a little higher at almost 4½%).   

The rise in unemployment has almost triggered the 0.5pp threshold signifying the start of a recession in the “Sahm rule”, with the average unemployment rate over the past three months now 0.4pp higher than its low point over the past year (a probit model based on the Sahm rule would equate a 0.4pp reading with a 10% risk of recession, with the rule’s 0.5pp threshold met if the unemployment rate soon rose to 4.3% or more).  

The Sahm rule of thumb is not precise, mainly because using the change in the three-month average of the unemployment rate introduces some inertia into the calculation, and there is no guarantee that it will continue to hold in the future.

Nonetheless, it does better than any other economic indicator in providing a timely signal of recession. 

This reliability reflects the fact that every sharp rise in the US unemployment rate in the post-WW2 period has been driven by the job losses that are the hallmark of a recession.

(“Soft landings” – which are the rare instances when the Fed has tightened policy without triggering a recession – are a different matter, with unemployment falling during the three soft landings of the 1960s, 1980s and 1990s.)

The recent rise in the unemployment rate has reflected weakness in employment. Employment is volatile, but has edged lower in trend terms over recent months and is broadly unchanged from the level of a year ago.

A natural reaction would be to downplay this weakness because it contrasts with the less volatile number of payroll jobs, which have continued to grow, albeit at a slower rate. 

Payrolls are broader in scope and less volatile than surveyed employment, counting, for instance, the additional jobs worked by some people. They are derived from a separate survey of business and have traditionally been the Fed’s preferred measure of the demand for labour.

However, it seems likely that the unusual divergence between employment and payrolls will eventually be resolved by an upward revision to employment and a downward revision to payrolls.

Employment appears understated because the population benchmarks used to construct employment do not fully capture the recent surge in migration (broadly speaking, employment is estimated by multiplying the ratio of employment to population from a sample of households by population benchmarks).

As a recent analysis by the Federal Reserve Bank of Dallas shows, the surge in international migration estimated by the CBO over the past couple of years – where more illegal migrants have been released into the community – is yet to be reflected in census data on the population.

Nevertheless, even if the level of employment is eventually revised higher, recent growth will likely still be relatively weak given the sample estimates of the ratio of employment to the population should be less affected by the jump in migration.

Payrolls seem biased in the other direction, with more comprehensive, but less timely payroll estimates sourced from unemployment insurance data posting slower growth over last year.    

For the Fed, a further material rise in unemployment would pressure policy-makers to cut interest rates.

Most policy rules incorporate both the expected inflation rate and the anticipated gap between the unemployment rate and the NAIRU, pointing to lower interest rates assuming inflation returns to target over the next couple of years.

Meaningfully higher unemployment would point to a faster decline in interest rates given the FOMC currently forecasts this gap will be broadly in balance over the next few years, with the unemployment rate expected to fluctuate between 4 and 4.2%.

The extent of the pressure on rates would also likely depend on what drove unemployment rate higher. If it reflected the usual job losses seen in a recession, history shows the Fed would respond forcefully in cutting rates.

Alternatively, the Fed might be more cautious in cutting rates if higher unemployment reflected employment failing to keep pace with growth in the workforce, which explains most of the increase in unemployment to date. 

Such an outcome would be unprecedented for the US in the post-WW2 period, although it is a common occurrence in other advanced economies, such as the euro area and Australia.