The labour market is not softening. It is emptying.

A SURPRISING SIGNAL in America's labour market has prompted headlines in recent weeks. The truth is more prosaic. The headline numbers are not telling a story of economic recovery or impending recession. They are telling a story of workers disappearing.

The evidence accumulated over the summer is striking in its consistency, if not in its immediacy. Private-sector employers added just 44,000 jobs in July, well below the 75,000 expected by economists polled by the Wall Street Journal, according to ADP, a payroll processor. In June the Bureau of Labour Statistics recorded a more respectable-looking 57,000 in nonfarm payroll growth and an unemployment rate that ticked down to 4.2%. Both figures are misleading.

The reason is not hard to see. The unemployment rate fell not because people found jobs but because 720,000 of them left the labour force in a single month. Total employment, as measured by the household survey, fell by 507,000. The unemployment rate is a ratio. Remove enough people from the denominator and it can improve while the economy quietly sheds jobs. The labour market was not improving in June. It was shrinking.

The participation rate—the share of the working-age population that is employed or looking for work—dropped to 61.6% in June, its lowest level since 1976 outside the pandemic.

That fragility now shows up in the hiring data. The July ADP report revealed a labour market split in two. Virtually all of the job gains came from healthcare-related sectors, with goods-producing industries seeing a net loss. The concentration of hiring in a single demographic-sensitive sector is not a sign of vitality. It is a sign of constraint: the places that are hiring are those that cannot automate or import their way out of a shortage.

The incentive structure is clear. Healthcare, construction and education face aging workforces and credentialing barriers that prevent cross-industry movement. They must hire because they have no alternative. Manufacturing and trade are losing ground, caught between tariffs that raise input costs and automation that reduces the headcount needed for a given level of output. The net result is a labour market that looks stable in headline numbers but is, in reality, being hollowed out.

This creates a disquieting dilemma for the Federal Reserve. Mr Warsh, the chairman, has signalled that "prices are too high" and that the fight against inflation is not over. Yet the same labour supply squeeze that constrains hiring is also driving wages for job-changers to 7% year-on-year in July, while job-stayers saw 4.4% pay growth, according to ADP. Workers who leave their jobs command a premium because employers are desperate. That is a wage push that looks like the inflationary pressure the Fed is trying to prevent.

To be sure, a 4.2% unemployment rate does not by itself warrant alarm. The Fed holds the federal funds rate in a band of 3.50-3.75%, and markets are debating whether a hike is coming before year-end. Mr Warsh himself said in early July that jobs data had "been moving in a good direction". The standard playbook would hold: if unemployment is low and growth is acceptable, the labour market is tight, wages are bid up, and the Fed should be cautious about cutting.

The deeper problem is that the standard playbook assumes the labour market is moving along a single curve between unemployment and inflation. It is not. Supply is falling faster than demand, which means the usual signals are scrambled. A declining participation rate can push both unemployment down and wage pressures up, even as total employment falls. The Fed's traditional tools—raising rates to cool demand, lowering them to stimulate it—do not address the structural absence of workers.

There are three forces at work, each pointing in the same direction. Demographic aging is the most predictable: the baby-boomer retirement wave will continue to drain the workforce for years. Immigration policy is the most controllable: foreign-born workers tend to have higher participation rates and are disproportionately prime-age. And AI-driven reallocation, while slower to materialise, is beginning to show up in the sectors where new graduates traditionally enter.

None of these forces will reverse unless policy responds. The first task is to stop making the problem worse. Further tightening of immigration, which would remove the one source of young, high-participation labour, would be a mistake. The second is to loosen credentialing bottlenecks in the sectors that are starved for workers. Nursing, construction and teaching require time-consuming qualifications that deter cross-industry movement even when demand is desperate. Reforming those barriers would not solve the problem but would ease its sharpest edges. The third is fiscal realism: as the workforce shrinks, the tax base that supports an aging population will too. That arithmetic does not change because it is unpleasant.

For investors, the structural implication is straightforward. A labour market that is losing workers is one in which labour costs are a floor, not a ceiling. Companies that rely on large, low-margin workforces—retail, hospitality, logistics—will face persistent pressure. Those that can substitute capital for labour, or that operate in sectors with pricing power strong enough to pass costs through, will do better. The July jobs report, due on 7th August, will not change that dynamic. It will only confirm it.

The surprising signal in America's labour market was never in the headline. It was in the empty chairs.