Viewpoint
September 05, 2025 | 15:04
Shifting Our Gaze to the Unemployment Rate
Shifting Our Gaze to the Unemployment Rate |
| The August employment report cemented our view that labor market resilience can no longer be taken for granted. The sharp slowdown in net job growth in recent months is undeniable. While not yet signaling outright retrenchment and widespread layoffs, demand for new labor appears to be in a deep freeze that won’t likely let up without more intervention from the Fed to lower short-term interest rates. Net new job creation barely has a heartbeat right now, trending at 29k over the past three months. Not only did the economy create a lower-than-expected 22k nonfarm jobs last month, but June was completely revised away to a net 13k job loss (Chart 1). The most important number investors will need to watch going forward is the jobless rate—not the inflation rate. |
| The CPI report previously took that honor; but even if we see more signs of tariff pass-through and faster consumer inflation for August, it won’t likely get much of a reaction from the Fed. The FOMC still leans toward tariffs being a one-time increase in prices, and markets will quickly discount the prospect of sustained inflation in the face of rising unemployment rates and an increasingly discerning consumer. For August, we are forecasting an increase in the headline inflation rate to 2.9% from 2.7%, with core inflation staying at an elevated 3.1%. An uncomfortable inflation rate no doubt, but one that’s likely to push households further into a financial corner, forcing them to cut back even more on discretionary spending. It was a rapid half-point increase in the jobless rate in the first three quarters of 2024, triggering the Sahm (Recession) Rule, that got the Fed to kick off the monetary easing process last year. It started with a bigger-than-expected 50-bp cut last September and culminated in total rate cuts of 100 bps by the end of December before trade and immigration policy risks from the new Administration pushed the Fed back onto the sidelines. The August jobless rate just hit a new cycle high in the post-pandemic era at 4.3%. The increase in the broader U6 measure of unemployment (which includes discouraged and marginally attached workers and those working part-time for economic reasons) has risen even faster this year and hit 8.1% in August (Chart 2). Rising jobless rates are probably the best indicator we have at the moment of increasing labor market slack and the growing imbalance between labor supply and demand. They also suggest that GDP growth has likely downshifted to a sustained period of below-potential rates between 1.0% and 1.5%. The silver lining is that the three-tenths increase in the official U3 unemployment rate so far this year has not been enough to trigger the Sahm Rule (Chart 3). In other words, the U.S. economy is not likely in recession yet. But with the jobless rate arguably already somewhat above full employment levels of 4.2%, the risk of further labor market deterioration is moving to the forefront for the Fed. We also expect a sizable -800k preliminary annual benchmark revision estimate for nonfarm payrolls on September 9, implying job growth has probably been a lot softer than the numbers have been indicating so far this year. The evidence of growing labor market slack coming from the unemployment rate is an important warning and will be enough for the Fed to resume measured rate cuts at the September 17 meeting, with more to follow in the months and quarters ahead. |
Is AI Moving the Economic Needle?Business spending on AI is lifting investment and GDP, but may not yet be moving the dial much on productivity and employment. |
| This November marks the third anniversary of ChatGPT’s launch and the resulting widespread public use of generative artificial intelligence. Surging technology stocks suggest investors anticipate a big financial payoff from faster revenue and productivity growth for corporate America. Yet three years on, is there much evidence of this reward in the economic data? |
|
|
Business Investment: verdict—yes The role of information technology (IT) investment in the economy has been growing for seven decades (Chart 1). Although accounting for a relatively small 4.5% share of nominal GDP (6.2% in volume terms), IT spending punches well above its weight. Expanding seven times faster than the economy in the year to Q2 (15.2%, Chart 2), it accounted for nearly a third of real GDP growth (0.6 ppts). That’s partly due to tariff front-running, but also to businesses developing and adopting AI systems. In fact, a 26% annualized surge in software spending in Q2 (the fastest in 22 years) added a record 0.6 ppts to quarterly growth. Four corporate giants alone (Meta, Amazon, Alphabet and Microsoft) plan to spend about $320 billion this year on AI technologies and data centers, up from $230 billion last year, enough to lift growth by 0.3 ppts. |
| Construction spending on data centers, largely to feed AI systems, surged 30% y/y to July and is on track to exceed outlays on office buildings (Chart 3). However, the dollar increase ($9 billion) is too small to noticeably propel a $30 trillion economy. Still, technology firms plan to spend hundreds of billions of dollars on microchip fabrication plants, which could reverse a slide in spending on IT manufacturing facilities following a torrid run sparked by the CHIPS Act of 2022. U.S. Trade: verdict—yes A record $483 billion annualized deficit in IT goods and services (Chart 4) accounted for more than 60% of the total U.S. trade gap in Q2. At 1.6% of GDP, the ratio has doubled in the past 25 years. Imports of IT goods and services are rising at double-digit rates (Chart 5), much faster than exports, partly due to AI investment and tariff front-running. The growing deficit in IT goods trade has outrun an improving (though much smaller) surplus in IT services trade, carving an estimated 0.4 ppts from GDP growth in the past year and offsetting about half of the push from IT business spending. The White House’s tariffs are aimed at correcting this and other trade imbalances, but finding sufficient resources for domestic production may pose a hurdle. Industry GDP: verdict—no IT industries added 0.4 ppts to average annualized real GDP growth in the two decades to 2025Q1, or almost a fifth of overall growth (Chart 6). Not bad for a sector that accounts for just 6.2% of GDP. However, the impact is unchanged since ChatGPT’s launch in late 2022, suggesting no material additional effect. Of course, that could change in the years ahead as new microchip and data facilities come on stream. Productivity: verdict—not apparent Nor is there yet much evidence of AI lifting labor productivity (Chart 7), unlike after the Internet’s rollout into commerce in the late 1990s. Although productivity growth rebounded in 2024 from earlier pandemic-related weakness, the pace in the past three years remains below normal. A recent MIT study found that 95% of firms that use generative AI have not seen a return on their investment. To paraphrase, time will tell whether AI is both everywhere and in the productivity statistics. As with past major new technologies, businesses may simply need time to adapt processes and workforces to fully realize the benefits. Employment: verdict—not materially AI has not led to material job losses. Challenger layoff announcements in the three years to August were only moderately above normal. Moreover, AI was cited as a reason for just 27,000 pink slips since May 2023 (when responses were first collected), or less than 2% of the total. But AI could be affecting hiring decisions. IT employment has declined in the past three years, unlike the surge in the late 1990s’ tech boom (Chart 8). This is despite the new jobs required to build AI systems and infrastructure. To be sure, several factors may explain the recent slowing, including the previous rapid expansion after the pandemic, modestly restrictive monetary policy, immigration curbs, and the trade war. But anecdotal evidence suggests some managers are hesitant to hire workers for tasks that AI could perform. A recent Stanford University study found a significant decline in hiring of entry-level workers in occupations highly exposed to AI automation, such as software development and customer service. However, it also found healthy demand for experienced workers in the same occupations, as well as for all work levels in which AI augments human skills, such as helping with research and diagnostics. The jury is out on AI’s ultimate impact on the workforce, but it is likely to create both winners and losers. The rundown: Business spending to adopt and build AI systems is lifting economic growth meaningfully, albeit with some offset from imported AI gear. Nonetheless, the effect on productivity and employment is not yet apparent in the economy-wide data. |