Are workers finally getting more productive?

Non-farm business productivity grew at an annualized 0.3 per cent in the first quarter, while unit labour costs rose at 4.7 per cent. That is an awkward result for the economists who have been counting on an AI-driven productivity boom to deliver growth without inflation.

After a strong 2023

The weak reading stands out because the previous year looked so different. Output per hour grew briskly through the middle and end of 2023, helping to explain how the economy expanded strongly while inflation fell. Optimists read that as the start of a new cycle, driven by a wave of business formation during the pandemic, workers moving into better-matched jobs and early adoption of generative AI tools after the launch of ChatGPT in November 2022.

The first quarter punctured that story, at least for now. Output grew more slowly while hours worked kept rising, which is the combination that drags the ratio down. The Bureau of Labor Statistics will publish revised figures early next month, and first-quarter numbers are often revised substantially, but the direction of travel matters to a Fed that has been hoping productivity would do some of its work for it.

Hoarding labour

Bringing inflation down without a recession needs real gains in output per hour. The figures suggest American companies are still hoarding labour, keeping staff on the payroll because they fear they could not hire them back later. Without better productivity, strong wage growth feeds straight into operating costs, and services inflation stays put.

The fear is rooted in recent memory. In 2021 and 2022 employers faced the tightest labour market in decades, with job openings far exceeding the number of unemployed workers and quit rates at record highs. Firms that cut staff during the pandemic spent the following two years paying signing bonuses and higher wages to rebuild. Having learned that lesson, many are reluctant to trim headcount even as demand softens, which keeps hours high relative to output.

How the arithmetic works

Unit labour costs measure what a business pays in wages and benefits for each unit of output. They rise when compensation per hour grows faster than output per hour. With productivity nearly flat in the first quarter, almost all of the increase in hourly pay passed straight through to costs, which is how a 0.3 per cent productivity gain sits alongside a 4.7 per cent rise in unit labour costs.

That relationship is why the Fed watches productivity so closely. Wage growth is consistent with 2 per cent inflation when it roughly equals the inflation target plus the trend growth in productivity. If productivity grows at 1.5 per cent a year, wages can rise by about 3.5 per cent without pushing up prices. If productivity stalls, the same wage increases leave companies with a choice between accepting lower margins and raising prices, and in services, where labour is the main cost, they usually raise prices.

The long wait for a boom

History counsels patience about technology. Personal computers spread through offices in the 1980s, and Robert Solow observed in 1987 that their effect was visible almost everywhere except in the productivity statistics. The payoff came a decade later: from the mid-1990s to the mid-2000s productivity grew at roughly twice the pace of the previous two decades, as firms reorganized around the internet and new software. Alan Greenspan's Fed held rates steady in the late 1990s partly on the bet that productivity would contain inflation, and for several years it did.

Generative AI may follow the same path, with gains arriving only after companies redesign workflows, retrain staff and write off old systems. That process takes years, and in its early stages it can even depress measured productivity, as firms spend money and time on adoption before the benefits show up. Expecting the technology to rescue the inflation outlook this year puts the timetable of the 1990s on fast-forward.

The case for looking through it

The best counterargument is that one quarter of productivity data tells very little. The series is volatile, heavily revised and sensitive to the gap between the output and hours estimates, each of which has its own errors. Over the past four quarters productivity has still grown at a respectable pace, and strong immigration has lifted labour supply, which can depress measured output per hour temporarily as new workers are trained. Several quarters of weak readings would be needed before anyone could call the 2023 revival a false dawn.

That caveat is fair, and it cuts in the Fed's direction of caution. If the true trend is somewhere between last year's strength and this quarter's weakness, wage growth still needs to slow further to be consistent with the target. Average hourly earnings are rising at roughly 4 per cent a year, above the pace most estimates of trend productivity would comfortably support.

Who feels the squeeze

Labour-intensive services bear the brunt. Restaurants, hotels, healthcare providers and retailers cannot easily substitute machines for staff in the short run, so rising unit labour costs either compress their margins or show up in their prices. Capital-intensive and software businesses are better placed, since their costs depend less on hours worked. Workers, for now, are holding on to real wage gains, which is good for household spending and bad for the Fed's inflation timetable. Borrowers waiting for lower rates are the indirect losers.

What would change the view

The revised productivity figures, the second-quarter data at the start of August and the employment cost index will show whether this is a blip. A slowdown in hiring alongside steady output would lift productivity quickly, as firms finally stop hoarding. Evidence of AI adoption translating into lower staffing needs in back-office work would also help, though that is more likely to appear in company earnings calls first than in national statistics.

The AI productivity dividend has not shown up in the data yet. Until it does, rising unit labour costs will keep services inflation high, and the Fed will probably be in no hurry to cut, with the autumn now the earliest realistic window.