National Productivity Week 27th January 2025 | Visit Website

A diverse community of
leading experts, policymakers
and practitioners

The Institute’s key research themes
are led by ten academic partners
spread across the UK.

We’re a UK-wide research
organisation exploring what
productivity means for business

Businesses are crucial to solving
the UK’s productivity problems.

– Jun 7th, 2024

Q1, 2024: U.S. productivity growth stagnates

Productivity Measurement Analysis series – United States, Q1 2024 by Martin Fleming. This version, published on 7 June 2024, is an update to the original insights first published on 3 May 2024.

General Summary and Main Figures

U.S. nonfarm business sector labour productivity increased 0.2% in the first quarter (Q1) of 2024, the U.S. Bureau of Labor Statistics (BLS) reported on June 6th. The report provided revised Q1 2024 estimates.

The June 6 release was based on more recent Q1 2024 and Q4 2023 output, hours, and compensation source data than were available for the May 2 preliminary report. The revisions resulted in a 0.1 percentage point decrease in first quarter 2024 nonfarm business sector labour productivity growth.

In the first quarter, nonfarm business sector output increased 0.9% and hours worked increased 0.6%, both quarter-over-quarter (QoQ) at a seasonally adjusted annual rate (SAAR). From the same quarter a year ago, nonfarm business sector labour productivity increased 2.9%.

Unit labour costs in the nonfarm business sector rose at a 4.0% annual rate in the first quarter, reflecting a 4.2% increase in hourly compensation and a 0.2% increase in productivity. Unit labour costs increased 0.9% over the last four quarters.

US Productivity and Costs, Q1 2024, revised

Quarter-on-year ago comparison, SAAR (Q1 2023) Quarter-on-quarter comparison (Q4 2023) Pre-COVID-19 comparison, SAAR (Q4 2019)
Nonfarm Business      
Labour Productivity 2.9% 0.2% 1.5%
Unit Labour Cost 0.9% 4.0% 3.5%
 
Manufacturing
Labour Productivity 1.1% 0.0% 0.3%
Unit Labour Cost 4.1% 3.1% 4.1%
 
Nonfinancial Corporate
Labour Productivity* 3.2% 1.8% 2.0%
Unit Labour Cost* 1.1% 2.6% 3.2%

For the year 2023, nonfarm business sector productivity rose 1.4%, a reversal from a 1.9% 2022 decline. Unit labour cost rose 2.8%, nearly three percentage points slower than the 5.7% 2022 increase.

Manufacturing sector labour productivity was unchanged in the first quarter of 2024, as output fell 0.2% and hours worked also declined 0.2 %, all QoQ at SAAR. Unit labour costs in the manufacturing sector increased 3.1%, reflecting a 3.1% increase in hourly compensation and unchanged productivity. Manufacturing unit labour costs increased 4.1% from the same quarter a year ago.

For the year 2023, manufacturing sector productivity declined 0.7%, a slightly slower pace than the 1.2% 2022 decline. Unit labour cost rose 5.6%, more than one percentage point faster than the 4.8% 2022 increase.

The first-quarter 2024 measures for the nonfinancial corporate sector were also published in the June 6 release. Productivity increased 1.8% in the first quarter as output and hours worked increased 2.5% and 0.7%, respectively, all QoQ at SAAR. Productivity increased 3.2% from a year earlier.

Insights into the Q1 2024 Productivity Release

Weak Q1 2024 U.S. nonfarm business sector productivity growth contrasts with robust second half 2023 improvement. While the data are seasonally adjusted, the Q1 growth slowdown follows productivity declines in Q1 2022 and Q1 2023. Both years had stronger second half growth. The long-term trend in nonfarm business sector productivity growth remains at 1.5%.

Also in the first quarter, manufacturing sector productivity growth continued at a disappointing pace. The manufacturing sector’s long-term productivity decline continues with the level now six percent below its first quarter 2011 peak.

With the June 6 release, the first view of Q1 2024 nonfinancial corporate sector productivity and related measure were published. Productivity in the nonfinancial corporate sector continues at a somewhat stronger pace than in the much broader nonfarm business sector and the smaller manufacturing sector.

Productivity in the nonfinancial corporate sector increased at an average annual rate of 2.0% since the fourth quarter of 2019 – before the COVID-19 pandemic – while productivity in the nonfarm business sector increased at a 1.5% pace and productivity in the manufacturing sector increased 0.3%. The nonfinancial corporate sector produces 65% and the manufacturing sector produces 11% of gross value added in the nonfarm business sector. While the data are not reported, by implication, productivity in the combined financial and noncorporate sectors grew at a rate of approximately 0.6%% over the period.

The 4.0% (QoQ, annual rate) increase in Q1 nonfarm business sector unit labour costs reflected a 4.2% increase in hourly compensation and a 0.3% productivity increase. On a year-over-year basis, unit labour costs rose 0.9% from Q1 2023, the smallest increase since Q1 2021.

The June 6 release also reports a one-half percent cumulative decline in nonfinancial corporate sector unit labour costs from its Q3 2023 peak. The decline reflects a 2.0% increase in the productivity level offsetting a 1.4% hourly compensation increase over two quarters. By contrast, over the most recent two quarters, unit labour costs were unchanged in the nonfarm business sector and rose 1.0% in the manufacturing sector.

Discussion

Weak Q1 2024 U.S. nonfarm business sector productivity growth adds more fuel to the growing fiery debate over long-term productivity growth and the impact of artificial intelligence. Recently, two very different perspectives have emerged, both relying on very similar data.

Recent work by MIT economist Daron Acemoglu argues that the upside to productivity realised from the adoption and deployment of generative AI as define by Large Language Models (LLMs) is limited. Over the next 10 years, Acemoglu expects a , 0.7% boost to the level of total factor productivity and 1.1% boost to the level of GDP. Acemoglu – one of the most cited academic economists characterises his estimates as non-trivial, but notes that they are much lower “than both the revolutionary changes some are predicting and the less hyperbolic but still substantial improvements forecast by Goldman Sachs.”

In a rebuttal of sorts, the Goldman Sachs economics team writes that they share Acemoglu’s view that AI applications offer the opportunity for task automation but automation of many AI-exposed tasks is not cost effective today as well as for the next decade or so. However, by contrast, the Goldman Sachs view is that the significant potential for cost savings will be combined with a tendency for new technology costs to fall rapidly over time which should eventually lead to more widespread adoption and automation.

In contrast to Acemoglu, the Goldman Sachs view is also that the reallocation of labour and creation of new tasks following AI adoption are very relevant when assessing the economic potential from generative AI.  History points to technology-driven reallocation of resources and expansion of the production frontier as main sources of economic growth. AI will raise output both by raising demand in areas where labour has a comparative advantage and by creating new opportunities that were previously technologically or economically infeasible.

Both the Acemoglu paper and the Goldman Sachs discussion rely on the work of Maja Svanberg, Wensu Li, Martin Fleming, Brian Goehring, and Neil Thompson of the FutureTech project at the MIT Computer Science and Artificial Intelligence Lab (CSAIL). Svanberg et. al., using data from computer vision AI applications, find that because of the large upfront costs of AI systems, only 23% of worker compensation “exposed” to AI computer vision would be cost-effective to automate. Suggesting the conclusion that workers are more economically attractive for firms – particularly those without scale – turns out to be widespread.

Overall, the Svanberg et. al. AI-labour cost framework provides an assessment of the anticipated rapid technology cost declines and the expected industry transformation in how AI technology is produced and consumed.

An initial jump in adoption is expected as some tasks that are already attractive to automate are converted to AI, and then a slower roll-out as cost drops and AI-as-a-Service business model innovation makes AI attractive. Only the first of these will be an abrupt shock to the labour market. But, in the second phase job loss would be smaller than the existing economy-wide job turnover rate, suggesting that labour displacement will be more gradual than abrupt.

Also, if the Acemoglu/Goldman Sachs debate is truly limited to generative AI as define by LLMs, it ignores AI deployment aimed at business processes using the vast quantities of structured and unstructured data behind business sector and government firewalls. These efforts are now underway.