Two jobs reports, two stories. Which one matters?

Non-farm payrolls rose by a steady 263,000 in September, evidence that American companies are still hiring briskly. Days earlier the Job Openings and Labor Turnover Survey (JOLTS) had shown vacancies falling by more than one million in August, to 10.1 million. Markets gather every first Friday to pick apart the official employment report as if it told a single story. This month it told two. One survey describes a labour market running red hot, and the other shows demand for workers cooling at a remarkable pace.

Working out which one matters is the main question for investors. The Federal Reserve has tied its policy closely to the labour market, and officials keep saying that wage pressure will not ease until the ratio of job openings to unemployed people falls back towards one. Getting the Fed's path right means looking at how the two surveys are built, and at how companies hoarding labour are bending the usual employment cycle.

Vacancies that never existed

The fall in job openings suits the case for a soft landing. If companies can scrap unfilled job ads without firing anyone, the ratio of vacancies to unemployed people can drop sharply without an ugly jump in joblessness. In theory that takes away workers' bargaining power and lets wage growth settle without the damage of a normal recession.

Treating online job postings as hard data takes a leap of faith, though. During the recovery from the pandemic, human resources departments posted speculative and automated listings across job sites, swelling the reported stock of openings. When business confidence turned in late summer, companies deleted the unfilled ones with a keystroke. Deleting an advert costs nothing, while firing someone and paying severance is a real financial decision. The fall in openings may be little more than tidying up of corporate hiring portals.

The survey itself adds to the doubt. JOLTS is drawn from a much smaller sample of employers than the payroll report, it is published with a lag of more than a month, and its response rate has fallen over the years, so large monthly swings are often revised. Openings peaked at a record of close to 12 million in March, and a series that volatile can give back a million in one month without any change in the underlying demand for staff.

An argument about a curve

The dispute has an academic pedigree. The Beveridge curve, named after William Beveridge, plots vacancies against unemployment, and in July two Fed economists, Andrew Figura and Governor Christopher Waller, argued that at today's extreme level of openings the curve is steep enough for vacancies to fall a long way with little rise in joblessness. Olivier Blanchard, Alex Domash and Lawrence Summers replied that history offers no example of vacancies falling that far without a meaningful increase in unemployment. August gave the optimists their first real support, as openings dropped to about 1.7 for every unemployed person from roughly two, while the jobless rate barely moved.

Payrolls are real money

The establishment payroll survey counts actual net additions to company payrolls, checked against state unemployment insurance tax records. The 263,000 jobs added in September, which pushed the unemployment rate down to 3.5 per cent, represent real money committed by employers. Despite higher borrowing costs and steady tightening from the Fed, managers are still adding staff.

The unemployment rate itself comes from a separate survey of households, and in September it fell partly because slightly fewer people were looking for work. That is a reminder of how noisy the monthly figures can be. The payroll count, by contrast, rests on a far larger sample of employers and is later benchmarked to tax filings, which is why the Fed trusts it more.

Much of that is hoarding. Employers went through severe worker shortages in 2021 and are reluctant to let people go, preferring thinner margins to being caught short-handed in the next upturn. That keeps household incomes up, which keeps consumer spending firm and lets companies hold on to their pricing power.

Where the jobs are

The composition of September's gain points the same way. Most of the new jobs came in leisure and hospitality and in health care, labour-intensive services that were still rebuilding staff numbers lost during the pandemic and that are least sensitive to interest rates. The industries the Fed can reach quickly are already shrinking their hiring: mortgage lenders have been cutting staff as applications collapse, and several large technology companies have announced hiring freezes. Average hourly earnings were still up 5.0 per cent on a year earlier, a pace the Fed regards as too fast to square with its 2 per cent inflation goal.

That split has consequences for who bears the adjustment. Workers in finance and technology face the first cuts, while those in restaurants, hotels and care homes retain their bargaining power for longer. Their wage gains feed straight into services prices, which is the part of inflation the Fed has found hardest to shift.

What would change the Fed's mind

The best case for the vacancy reading is that payrolls lag: companies stop hiring before they start firing, so openings turn first and payrolls follow months later. That fits past cycles, and it means September's gain says more about decisions taken in the summer than about intentions now. The Fed cannot act on a sequence it has not yet seen, though, and its next decision, on 2 November, comes two days before the October payroll report. Weekly jobless claims, still low, and the quits rate, which tracks workers' confidence in finding something better, will show the turn sooner.

Markets took the payroll number at face value. Share prices fell and Treasury yields rose on the day of the report, as traders reinforced bets on a fourth consecutive rise of 75 basis points in November. That reaction fits the Fed's own reading: officials have said they want to see a sustained run of slower job growth, and one month of falling vacancies does not meet that standard.

For the Fed, the payroll number is the one that carries weight. The Fed will keep raising rates until the payroll survey turns down and jobless claims rise, and falling online vacancies will not persuade it to stop sooner. Until then officials will treat the labour market as uncomfortably tight, expectations for interest rates will stay biased upwards, and a fourth large increase on 2 November is the likeliest outcome.