A Quick Caution in the Productivity Data
- Recent productivity data has been fairly strong, continuing the improved trend from 2018-on as the economy moved past post-GFC credit crunch and labor market slack dynamics.
- The SF Fed’s total factor productivity data shows that capital deepening (a good thing) and more intensive, but ultimately unsustainable, procyclical utilization of capital and labor are driving productivity growth over the past few years. TFP growth’s post-covid whipsaws, combined with the current boost from increasing utilization, seems consistent with typical post-recessionary patterns and yet another indicator of the ongoing underlying cyclical recovery in the US.
- Much like the internet or electrification booms, AI may take some time to show up in the broader data as it slowly shifts business practices across the economy.
- Other shocks are playing a role too. Immigration’s decline may be temporarily boosting productivity growth. Deglobalization and tariffs are likely a drag as they allocate capital towards less efficient, perhaps more robust or societally optimal, business practices.
- On net, we should be cautious in assuming further acceleration in productivity growth from here and a mild near-term deceleration cannot be ruled out. This is another reason for some caution around the inflation outlook as well.
Overall labor productivity growth (output per hour) has shown a moderate reacceleration from its post-GFC lows when weak investment, elevated commodity prices, and very slack labor markets. This process has been ongoing since roughly 2016-18. In recent years there has a potential further reacceleration but it seems far too early to count on that sustaining, particularly given that so many features of the US economy have exhibited recessionary and rebound like dynamics since 2022.
Economists at the SF Fed produce a quarterly decomposition of overall labor productivity (output per hour) growth into investment, improvements in labor quality (education, etc), and total factor productivity (technology and efficiency gains).
Total factor productivity is the ‘unexplained residual’ after removing the impacts of the labor and capital factors we can most easily measure. TFP growth has shown a more modest recent reacceleration. However, these researchers make the case we also need to adjust for more intensive intra-cyclical shifts in the utilization of capital and labor to get a clearer read of underlying technological progress. Unfortunately, this adjustment casts a more sobering view of the data. TFP growth is currently being driven by a surge in the utilization of capital and labor, as often happens coming out of cyclical downturns. Utilization-adjusted TFP growth has shown much less sign of reacceleration.
We should not overly extrapolate what the current data means about the medium- and longer-term future of productivity growth. AI may well provide a substantial lift to TFP, it may be mostly a capital-based shock (where the surge in investment largely accounts for the output/hr gains), or it may simply be a revolution of work tasks but ultimately shift macro-level productivity less than the optimists currently expect.
As economist Robert Solow famously (at least among economist circles) quipped in 1987 that you could see the PC “everywhere except the productivity statistics.” The canonical discussion is the very slow diffusion of electrical power innovations across factories over the span of decades, The Dynamo and the Computer. It took some time but eventually the IT revolution did show up. This took place as both a mild increase in capital deepening (an increase in capital / hours worked) and the surge in total factor productivity growth showing a true shift in underlying technology and more productive business practices that lasted for almost a decade.
This data suggests that there may be some early signs of the AI boom but they are very early signs. Utilization-adjusted TFP growth in the consumption sector (what the authors refer to as the rest of the business economy outside equipment and consumer durables) seems a bit better than its post-GFC doldrums, but the acceleration there is modest overall. This possible further acceleration will be building off of the OK but far from booming recent underlying (capital deepening, labor quality, and TFP) productivity data.
On a cautionary note, the rapid slowdown in immigration and surge in deportations are likely temporarily flattering the productivity growth data at least a little bit. This because these workers tend to fill relatively lower wage and productivity roles whose absence reduces aggregate hours worked more than GDP. The impact of tariffs and broader forces of deglobalization will be a net drag on productivity growth as well, as historic most efficient practices have to be recreated. This is likely at least partly behind the current drop in utilization adjusted TFP growth in the equipment and consumer durables industries. Demand whipsaws may also be playing a role. These are shorter-lived forces but so many shocks hitting the economy at once make getting a sense of underlying activity, let alone productivity, trends a challenge.
