Wages are rising — by how much depends on who you ask

Average hourly earnings rose 0.6 per cent in November, a strong monthly gain that sent bond-trading algorithms onto the defensive. Payrolls grew by 263,000 in the same report and the unemployment rate held at 3.7 per cent, so the labour market gave the bears little else to work with. The argument over whether wages are accelerating is getting harder to settle, because the data are noisy. The headline figure from the establishment survey suffers from composition bias, where shifts in the mix of low-paid and high-paid workers distort the average.

Depending on which gauge a committee member picks, the labour market is either in a runaway wage-price spiral or easing into an orderly late-cycle slowdown. When shops, bars and hotels hire lots of lower-paid staff, average hourly earnings can look weak simply because more low-wage jobs are being counted. When that hiring pauses, the average rises automatically. Separating real pay pressure from changes in the mix means looking past the headline earnings number.

A lesson from April 2020

The distortion is not theoretical. When payrolls fell by more than 20 million in April 2020, average hourly earnings jumped, because the workers who lost their jobs were disproportionately low-paid. Nobody took that as evidence of a boom in pay. The same arithmetic runs in reverse during a hiring recovery in hospitality and retail, and it can run either way from month to month as sectors take turns to hire.

That recovery is not finished. Employment in leisure and hospitality is still below its level of February 2020, so the sector keeps adding low-paid jobs in some months and pausing in others. Each pause flatters the average; each burst of hiring drags it down.

Hours matter too. Average hourly earnings divide pay by hours worked, so a month in which part-time shifts are trimmed can lift the hourly figure without any employer raising a single wage. Employers who want to save money without firing anyone usually trim shifts first, which is precisely the pattern that flatters the hourly figure late in a cycle. Monthly moves of this size are therefore weak evidence on their own, whichever direction they point.

Pick a wage gauge

The Employment Cost Index (ECI) tries to strip out these shifts by holding occupational and industry weights constant, which makes it a cleaner measure of what employers are actually paying. The ECI has shown signs of levelling off. The Atlanta Fed's Wage Growth Tracker, which follows the same individual workers over a rolling twelve-month window, is still near record highs, pushed up by the big raises offered to people who change jobs.

Each gauge has its own blind spot. The ECI is quarterly and arrives late, so it cannot settle an argument about the past month. The Atlanta Fed tracker leaves out people entering and leaving work, and this year it has run well above the other gauges.

That leaves investors trading in a fog. Bond desks that set their peak-rate expectations off the establishment survey risk mistaking seasonal swings in hours worked for lasting wage inflation. Equity strategists cheering softer ECI readings, for their part, are overlooking the steady job-hopping that keeps corporate margins from recovering.

Why the Fed is listening

The dispute matters more than usual because Jerome Powell has told markets where to look. In a speech at the Brookings Institution on 30 November, the Fed chairman singled out core services excluding housing as the part of inflation most likely to persist, and tied it directly to the labour market. Wages are the largest cost in most service businesses. A central bank that has made that link will treat stubborn pay growth as a reason to keep rates higher for longer, even as it slows the pace of increases.

The same speech signalled that smaller rate rises could begin as soon as the December meeting. Markets read that as dovish. The wage data suggest the more important message was about how long the Fed intends to hold rates at their peak.

Powell also argued that the shortage of workers owes much to early retirements and slower immigration, neither of which higher interest rates can reverse quickly. If labour supply stays tight for structural reasons, the Fed can only rebalance the market by cooling demand for workers, and job openings above ten million show how much cooling remains.

Real pay is still falling

The case against a spiral rests on one comparison. Average hourly earnings are up 5.1 per cent on a year earlier, while consumer prices rose 7.7 per cent in the year to October. Workers are losing ground in real terms, which is the opposite of the 1970s pattern in which pay chased prices and then outran them through indexed contracts.

Union coverage in the private sector is also far lower than it was then, and few American wage deals carry automatic cost-of-living clauses. Pay is rising because employers are competing for scarce workers, and that pressure can ease as soon as hiring slows. A spiral needs workers with the power to lock in gains, and most do not have it.

Margins keep shrinking

For corporate treasurers the dispute between wage indices makes no difference to profits. Unit labour costs have outpaced productivity growth for the past four quarters, so higher pay has gone straight into thinner margins.

The strongest counterargument is that companies entered this period with unusually fat margins after two years of pricing power, and can absorb a squeeze without cutting jobs. That is true of large firms with brands and market share. It is much less true of restaurants, hotels and care providers, where labour is most of the cost base and customers have started to resist higher prices.

The data to watch are the fourth-quarter ECI at the end of January, the quits rate in the monthly job openings survey, which tracks how many workers feel confident enough to switch, and the Fed's new projections on 14 December. A falling quits rate would be the earliest sign that the job-switching premium is fading.

In competitive service industries, companies are finding it harder to pass on 5 per cent pay rises without losing customers. Until the premium paid to job switchers falls back towards its historical norm, unit labour costs will probably keep eating into service companies' margins well into next year, whichever wage index policymakers prefer to quote.