iFlow > Short Thoughts
Published on Tuesdays, Short Thoughts offers perspectives on US funding markets, short-term Treasuries, bank reserves and deposits, and the Federal Reserve's policy and facilities.
John Velis
Time to Read: 7 minutes
Among the Fed’s Chairman’s Task Forces for Advancing Monetary Policy, Productivity and Jobs is notable for its leadership. Chair Kevin Warsh surprised some Fed watchers by selecting experts from outside the traditional Beltway-Wall Street nexus that the Fed typically draws on for perspectives on the economy. Doing so signals that they might bring some out-of-the-box thinking (by Fed standards) to the task force’s work and findings.
The leaders – venture capitalist Marc Andreessen, Microsoft executive Asha Sharma, and Charles Jones, a Stanford academic currently on leave at Anthropic – generally hold optimistic views on the effects that AI will have on the economy. Warsh shares this view: prior to his appointment, he wrote that AI as likely to allow for lower rates because of its productivity benefits.
The effects that AI will have on employment and inflation are still very much in the air and hotly debated. Here we’ll touch on the key macroeconomic considerations.
On one side of the employment debate are those who view AI as likely to lead to job losses given AI’s ability to replace certain routinized tasks. The other side believes that AI will bring about a form of Jevons’ paradox: the greater the productivity unlocked by AI automation, the greater the opportunities for human labor. It’s not so clear to us on which side the task force leadership lands. Sharma, who is the CEO of Xbox for Microsoft, has stated that “games are and always will be art, crafted by humans,” suggesting that she doesn’t necessarily see mass job displacement.
There’s also a separate debate about AI’s inflationary impact: between those who see AI’s demand for energy and other inputs, most notably chips, as inflationary, and those who see the potential for increased productivity as likely to lead to disinflation, making goods and services cheaper, all else being equal. The task force’s leaders appear to favor the disinflationary camp. Andreessen wrote in a 2023 Substack post that parts of the economy exposed to technological change could see slower inflation.
EXHIBIT #1: HIGHER AI ADOPTION ISN’T LEADING TO FALLING LABOR DEMAND YET
Source: BNY Markets, Bureau of Labor Statistics, Bureau of the Census
The argument that AI is ultimately disinflationary rests on the expected productivity gains from its adoption across the economy. Exhibit 1, however, shows that the expected productivity gains are not clearly occurring at present. Yet we see that outside of the technology sector, industries that feature the highest AI penetration don’t correspond to those with the largest job-opening declines.
The May 2026 data use the U.S. Census Bureau’s Business Trends and Outlook Survey to determine AI adoption by industry alongside the JOLTS data for labor demand (in the form of job opening rates). The tech sector is an outlier, falling in the lower right-hand quadrant of the scatter plot – higher AI penetration, less job demand. Unsurprisingly, the other industry that has seen high rates of AI adoption is finance. Labor demand was growing, at least through May this year. It’s admittedly a simple study, and we’re limited by readily available and timely data. However, we can’t declare that AI growth is leading to falling employment at this stage of the technology cycle.
As mentioned above, historical evidence of other productivity booms suggests that productivity growth – one of the secret sauces of long-run economic growth – creates a bigger economic pie that increases aggregate demand at least as much as aggregate supply. On this view, employment grows rather than contracts. A pessimist would have to argue that the AI boom is different from other game-changing productivity bursts seen throughout history. That may well prove true, though we have trouble seeing it yet in the data. This is presumably a key question to be examined by the Productivity and Jobs task force.
EXHIBIT #2: PRODUCTIVITY ISN’T SHOWING SIGNS OF A SURGE IN THE SHORT TERM
Source: BNY Markets, Bureau of Labor Statistics, Federal Reserve Bank of San Francisco
Beyond employment effects, it’s also not clear that we’re now in a productivity surge. Exhibit 2 shows labor productivity growth since just before the pandemic as well as total factor productivity (TFP) growth. TFP measures how efficiently the economy converts its productive inputs – labor and capital – into output. How much, per unit of input employed, does the economy grow, in other words? In the chart, labor productivity growth slightly exceeds TFP most of the time. This is because it doesn’t consider changes in quality and quantity of the capital stock, along with how fully physical inputs are being utilized. The bottom line is that neither productivity statistic is flashing “boom” conditions. Much to the contrary. Productivity growth hasn’t exceeded pre-Covid rates; indeed, in Q1 2026, both were slightly negative. We don’t doubt that productivity will accelerate in coming years as industry adoption, labor adaptation to AI, and economies of scale are achieved, but it’s not on the radar yet.
Over time – and we have over two centuries of productivity data for the U.S. economy – productivity is rather constant, punctuated by years-long (but ultimately episodic) surges before eventually returning to trend. More productivity creates more income and profit, generating more investment and capital spending (and capital deepening) and more hiring to meet increased aggregate demand. Ultimately, the arithmetic of productivity brings it back to earth. Productivity growth is defined as the increase in output divided by the growth of inputs (labor and/or capital). As those inputs increase in response to broad aggregate-demand growth, the entire fraction starts to decline, ultimately reaching the long-term sustainable rate, which has averaged around 1.25 to 1.6% per year, according to rolling 50-year estimates.
The Fed’s task force will have to address these issues, as will the community of economists and policymakers, though it’s not clear what the answers will be. Monetary policy may not be suited to the structural changes that will come to labor markets. Supply shocks often have reallocation effects, through which entire sectors and markets become displaced and, in many cases, undergo radical changes in the way they function, hire, train, and interact with the rest of the market. Monetary policy could very well be too blunt a tool to deal with these profound changes, which are structural rather than cyclical.
It’s also not clear, as we’ve stated, that policy rates will necessarily be lower under an AI productivity boom and reallocation shock. Shortages of skilled labor, key inputs and, crucially, capital can emerge. This can be inflationary. We’ve already seen prices of AI-associated inputs – chips, devices, electricity – rise at the wholesale and retail level. Short-term inflation could ironically become elevated, and the policy response hawkish. Furthermore, the competition to raise capital and deploy it through financial markets and physical investment could actually raise the cost of capital, meaning higher equilibrium real interest rates, including the reference rate. Finally, if AI creates sufficient demand and real wage growth (in response to productivity gains), households with higher real incomes could fuel traditional demand-pull inflation and a concomitant policy response.
EXHIBIT #3: INVESTMENT BOOMS AND BUSTS
Source: Reproduced from data found in BIS Annual Economic Report, 2006
Historical analogies raise some concern about financial stability risks from the AI boom. Consider Exhibit 3. For each historical technological cycle, the chart identifies a starting point that corresponds to the previous trough in capex. It then examines the increase in capex for those industries over the subsequent 10 years, indexed to 1 in the 0th year. Two things to note: first, the increase in spending on the AI build-out in its third year dwarfs that of previous technological booms. Second, typically after several years, capex peaks and returns toward the previous trough. After the boom, a “bust” can disrupt financial commitments made to the technology and private investments seeking to ride the wave, ultimately rippling into the real economy. If this were to occur at some point in the present cycle, it would become an issue for the real economy. We trust the Productivity and Jobs task force will take these historical lessons to heart in its deliberations.
Last week was eventful regarding the rates outlook, with both hawkish and dovish developments. Nevertheless, we maintain our call for no changes to rates this year, although we acknowledge upside risk to this view. Ahead of the July FOMC next week, the probability of a hike priced by the market peaked around 40% a week ago and has since fallen to below 20%. By December, cumulative meeting probabilities still imply nearly 1.5. The forward curve has been volatile lately, so we expect more twists and turns in market expectations, especially with increasingly sparse forward guidance coming out of the Committee; increased front volatility is the natural result of this development.
Warsh delivered his first semiannual testimony to each branch of Congress, and while prescriptive remarks were largely absent, he reiterated on several occasions that the Fed would unequivocally hit its 2% inflation target.
June CPI and PPI were released last week and were both lower than expected, indicating that the U.S.–Iran memorandum of understanding had brought down oil prices and, with them, inflation. However, a resumption of hostilities at the end of the week raised the specter of a second energy shock in coming months.
Finally, several Fed speakers delivered hawkish messages, despite the CPI/PPI relief, signaling that inflation remains the main risk to the outlook. Dallas Fed President Lorie Logan even argued that rates should rise.
Taking these developments on balance, especially the impact of a second energy shock, the rates outlook is indeed trickier and will ultimately be determined by the inflation data. We remain watchful for another bout of higher prices and aren’t certain the Fed would view it as sufficient cause to raise rates. We’re actively reviewing our rates outlook, but for now maintain the no-change view. We note with surprise that Brent oil is still trading below $90 per barrel and the first oil shock earlier this year saw limited, though measurable, inflation pass-through. With rates, as with much of the market, geopolitics and the economy’s reaction to them will be the ultimate arbiter of where rates will go.
