AI Summary. The author argues the AI boom's economic significance lies more in stock market gains than in data center construction, and that the inability to monetize AI investments could end the boom. Simulations using the Federal Reserve's FRB/US model suggest that an end to the buildout plus a roughly 30% market decline would slow growth to about 1.5%, short of recession.

Date Posted:
Is Database:
Database
Is Important:
Important

Furman estimates that the data-center buildout, which began in mid-2024, had raised US GDP by only ~0.3% by 2026Q3, as the equipment is largely imported and the US was already near full employment. Counting AI-driven stock market gains, the effect rises to ~0.6%. If the buildout ended and stocks fell ~30%, growth would slow to ~1.5%.

How would an end to the AI boom affect growth?

Core argument: If the AI data center buildout ended and the stock market fell about 30%, the FRB/US model predicts growth would slow to roughly 1½ percent, short of even a shallow recession.

The data center buildout is set to be among the largest investment expansions in peacetime American history. How different is our economy today because of it? The answer, based on a simulation using the large-scale macroeconomic model used by the Federal Reserve (FRB/US), is not much. The buildout by itself has raised GDP by only about 0.3%, because most of the investment is imported, and imports do not add to U.S. GDP or lower the unemployment rate. The stock market boom that came with it matters more—together they have raised GDP by about 0.6% and lowered the unemployment rate by about 0.3 percentage point.This analysis suggests that the economy could survive less data center buildout just fine. Investment would fall, but so would imports and interest rates, and GDP would be little changed. A broader fall in the stock market would be a bigger deal, The bottom line: so far the AI economy is less a story about data centers than about the stock market—and that is also where the risk is.

Takeaways by Macro Roundup® AI

  1. If the AI data center buildout ended and the stock market fell about 30%, the FRB/US model predicts growth would slow to roughly 1½ percent, short of even a shallow recession.
  2. The economy could withstand less data center buildout, making the AI economy so far less a story about data centers than about the stock market, which is also where the risk lies.
  3. Public policy changes or an inability to monetize AI investments could stop the AI boom, and financial-system complexities such as unwinding leverage, absent from the FRB/US model, pose a tail risk resembling 2001 more than 2007–09.

AI Summary. The growth boost from the artificial intelligence boom is projected to peak around late 2026 or early 2027, with AI investment as a share of GDP slowing and the estimated 0.4pp wealth-effect lift to US consumer spending fading. Across US nonmanufacturing industries, AI adoption shows a statistically significant negative correlation with employment growth and positive correlation with unemployment.

Date Posted:

GS’s Hatzius argues that AI’s growth impulse is nearing its peak, with the year-over-year increase in hyperscaler investment as a share of US GDP projected to peak in Q4 2026—even as AI investment continues to rise.

When will the AI boom's growth boost peak?

Core argument: The AI-driven wealth effect is estimated to have added 0.4pp (0.25% of GDP) to US consumer spending growth over the past year, an impact expected to decline gradually, or more sharply if AI-related stocks keep falling.

The growth impulse from the AI boom was set to peak in late 2026 or early 2027 even before concerns around the sustainability of the AI investment boom intensified in recent weeks. Although AI investment is still rising rapidly, we expect the increase as a share of GDP to slow over the next year. Alongside the fading boost from last year’s increase in US depreciation allowances, this will weigh on capital spending growth, especially in the US. AI has also boosted US consumer spending growth via a positive wealth effect worth an estimated 0.4pp (or 0.25% of GDP) over the past year. This impact is likely to decline over the next year, gradually if stocks perform in line with our strategists’ reasonably upbeat expectations but more sharply if AI-related stocks continue to decline. Our own base case remains that AI investment will ultimately pay off in terms of higher productivity and economywide profits. However, one key channel for this payoff is labor market disruption, which has negative near-term macroeconomic effects. And while DM labor markets still look solid in aggregate, we are starting to see clearer signs that the AI impact is spreading beyond the technology sector. Across nonmanufacturing industries, the share of US firms that have adopted AI in regular business functions now shows a statistically significant negative correlation with employment growth and a statistically significant positive correlation with unemployment.

Takeaways by Macro Roundup® AI

  1. The AI-driven wealth effect is estimated to have added 0.4pp (0.25% of GDP) to US consumer spending growth over the past year, an impact expected to decline gradually, or more sharply if AI-related stocks keep falling.
  2. The growth impulse from the AI boom was set to peak in late 2026 or early 2027, with AI investment’s rise as a share of GDP expected to slow over the next year and weigh on capital spending growth, especially in the US.
  3. Across US nonmanufacturing industries, the share of firms adopting AI in regular business functions shows a statistically significant negative correlation with employment growth and a statistically significant positive correlation with unemployment.

AI Summary. Inflation-protected U.S. Treasury yields rose from about 3.5% in early 1997 to about 4.5% by 2000, compared with roughly 2.4% on the 10-year and 3% on the 30-year today. The author argues that this 1990s experience helps calibrate expectations if new software technologies produce a comparable acceleration in productivity growth.

Date Posted:

Klein compares measures of real rates from the 1990s productivity boom to today’s real rates, using TIPS, as well as “synthetic TIPS” derived by subtracting inflation expectations from nominal rates. Current real rates are very low by comparison.

Would faster productivity growth from new software technologies push real yields higher?

Core argument: Real yields on TIPS were roughly 3.5% when the securities were introduced at the start of 1997 and rose gradually to about 4.5% by the peak of the bubble in 2000.

Real yields on actual TIPS were ~3.5% when they first arrived at the beginning of 1997, and then rose gradually to ~4.5% by the peak of the bubble in 2000. Yields on “synthetic” TIPS derived by removing estimates of inflation expectations from nominal yields were also around 3.5% in early 1997 and 4.5% in the first half of 2000, though they were more volatile in between. The yield on the benchmark 10y TIPS is currently 2.4% while the yield on the 30y TIPS is around 3%. Knowing what actually happened in the 1990s is helpful for calibrating expectations about what could happen in the next decade, if new software technologies lead to a comparable acceleration in productivity growth.

Takeaways by Macro Roundup® AI

  1. Real yields on TIPS were roughly 3.5% when the securities were introduced at the start of 1997 and rose gradually to about 4.5% by the peak of the bubble in 2000.
  2. "Synthetic" TIPS yields, derived by stripping estimates of inflation expectations out of nominal yields, also stood near 3.5% in early 1997 and 4.5% in the first half of 2000, with more volatility in between.
  3. The benchmark 10-year TIPS yields 2.4% and the 30-year TIPS around 3%, while the 1990s experience offers a guide for calibrating expectations if new software technologies deliver a comparable acceleration in productivity growth.

AI Summary. Measured total factor productivity growth is estimated to come almost entirely from higher input utilisation rather than underlying efficiency gains, based on a method that infers unobserved capital and labour utilisation from movements in hours per worker. On this evidence, recent productivity gains reflect "working harder" rather than "working smarter.

Date Posted:
Is Database:
Database

US labor productivity has accelerated, with output per hour growing at a mean annual rate of 2.5% btw 2023–2026Q1, ~1pp above its 2005–2019 pace. The gains have largely been driven by higher input utilisation, which is ‘inherently temporary.”

Does higher input utilisation explain recent measured productivity growth?

Core argument: Since the beginning of 2024, utilisation growth accounts for essentially all measured total factor productivity growth, suggesting the measured productivity gains derive from "working harder" rather than "smarter".

Input utilisation is not observed directly for the entire economy;  it must be inferred. Consider a firm that wants to meet strong demand but has a given capital stock, workforce, and technology. It can ask its existing workers to work longer (which we observe). It can also run its capital longer. Each margin is costly, for example, overtime pay, so firms typically use all of them at once. The observed margin serves as a proxy for the unobserved ones. The method infers changes in industry utilisation from observed movements in hours per worker, scaled by an estimated pass-through of those movements into labour productivity. Industry estimates are aggregated using Domar weights, which reflect each industry’s importance. The stacked bars in Figure 2 show measured TFP growth split into [Fernald's] measure of utilisation growth in light grey and utilisation-adjusted TFP growth in dark grey. In 2023, utilisation growth was a drag on measured TFP growth, and utilisation-adjusted TFP growth soared. Since the beginning of 2024, however, utilisation accounts for essentially all TFP growth. For now, the measured productivity gains appear to derive from “working harder” rather than “smarter.”

Takeaways by Macro Roundup® AI

  1. Since the beginning of 2024, utilisation growth accounts for essentially all measured total factor productivity growth, suggesting the measured productivity gains derive from "working harder" rather than "smarter".
  2. In 2023, utilisation growth was a drag on measured TFP growth while utilisation-adjusted TFP growth soared, according to Fernald's decomposition of measured TFP.
  3. Economy-wide input utilisation is not observed directly and is inferred from movements in hours per worker, scaled by an estimated pass-through into labour productivity, with industry estimates aggregated using Domar weights.

AI Summary. In the US, Europe and Japan, pay growth has decoupled from productivity growth, with productivity pulling ahead. US labor productivity, measured as real GDP per hour worked, is more than 5x its 1947 level while real hourly compensation is only about 3x, and median wage growth lags even further behind.

Date Posted:
Is Database:
Database
Is Important:
Important

In advanced economies, increases in labor productivity growth have decoupled from real wage growth, especially post-1980.

Are wages falling further behind productivity growth in advanced economies?

Core argument: US labour productivity, measured as real GDP per hour worked, is more than five times its 1947 level, while real hourly compensation is only about three times greater.

In the US, Europe and Japan, pay growth has decoupled from productivity growth, with the latter having pulled ahead. US Labour productivity — measured as real GDP per hour worked — is now more than five times greater than in 1947, whereas real hourly compensation is only about three times greater. Median wage growth has lagged even further behind productivity growth, with inequality pushing the average wage higher than the median. (The figures above show average rather than median wages; plotting median wages would show an even greater divergence.)

Takeaways by Macro Roundup® AI

  1. US labour productivity, measured as real GDP per hour worked, is more than five times its 1947 level, while real hourly compensation is only about three times greater.
  2. Pay growth has decoupled from productivity growth in the US, Europe and Japan, with productivity pulling ahead in all three.
  3. Median wage growth has lagged productivity growth by even more than average wage growth, as inequality pushes the average wage above the median.

AI Summary. About 73m US households hold retirement accounts, mostly defined-contribution plans whose balances have risen $6.7tn, and 45m households account for nearly $15tn in annual outlays, close to 3x Germany's economy and 70% of China's GDP. This group's spending has sustained faster growth than comparably wealthy economies.

Date Posted:
Is Database:
Database
Is Important:
Important

Citing the spending of the top 45mm US households, Jason Thomas argues that US economic growth could be characterized less as ‘k-shaped’ than “travelling at two speeds, one roughly the same rate as Europe’s and another that’s racing far ahead.”

Does retirement wealth explain why US spending outpaces peer economies?

Core argument: Balances in US defined-contribution retirement plans have risen by $6.7tn since the start of 2019, a 12.7% annually compounded gain, with roughly 73 million US households holding retirement accounts.

Roughly 73 million US households have retirement accounts (primarily defined-contribution plans like 401ks) that consist primarily of claims on businesses whose revenue growth and market values provide another hedge against inflation. Balances in defined contribution plans have increased by $6.7 trillion (+12.7%, annually compounded) since the start of 2019. US personal savings rates have fallen sharply over the past year (from 5.5% in April 2025 to 3% in May 2026) but that seems entirely rational. If your retirement balance is far above where you expected it to be, why not spend a bit more of your current income than previously intended? If we net across these overlapping cohorts and exclude recent first-time home purchases, we’re left with 45 million households that combine to account for nearly $15 trillion in annual outlays. That’s an extraordinary sum, equal to nearly 3x the size of the entire German economy and 70% of the GDP of China. And this third of the population has not only been insulated from the inflation shock but also exhibits the propensity to spend out of wealth and income to an extent that’s sustained a far higher rate of GDP growth than observed in economies with comparable living standards.

Takeaways by Macro Roundup® AI

  1. Balances in US defined-contribution retirement plans have risen by $6.7tn since the start of 2019, a 12.7% annually compounded gain, with roughly 73 million US households holding retirement accounts.
  2. Netting overlapping cohorts and excluding recent first-time home purchases leaves 45 million US households accounting for nearly $15tn in annual outlays, equal to nearly 3x the German economy and 70% of China's GDP.
  3. The US personal savings rate fell from 5.5% in April 2025 to 3% in May 2026, a decline the article argues is rational when retirement balances sit far above expectations.

AI Summary. Using 5.1m product listings from Sears catalogs, the authors construct a quality-adjusted price index implying much lower goods inflation for 1900–1990 than standard deflators, with real goods consumption growing 39x rather than 10.3x. They estimate faster growth before World War II (5.4% a year, 1900–1939) than after (4.2%, 1946–1980).

Date Posted:
Is Database:
Database
Is Important:
Important

Using 5.1mm Sears catalog listings and LLM-based hedonic estimation, this study finds real goods consumption grew 39-fold from 1900 to 1990, not 10.3-fold as official data imply, with the largest mismeasurement concentrated before WWII.

Does quality-adjusted pricing show faster goods consumption growth before the war?

Core argument: Using a quality-adjusted price index built from 5.1 million Sears catalog listings, the authors estimate real goods consumption grew by a factor of 39 between 1900 and 1990, compared with a factor of 10.3 using standard goods deflators.

Measuring real GDP growth requires distinguishing changes in prices from changes in product quality... systematic quality adjustment. price indexes are unavailable for much of the twentieth century. We construct a new quality-adjusted price index using 5.1 million product listings from Sears catalogs, 1900–1990. The resulting cost-of-living index implies substantially lower goods inflation than conventional deflators between 1900 and 1990, real goods consumption grew by a factor of 39 using our index,compared with a factor of 10.3 using standard goods deflators. Figure 5 shows our estimates of the cost-of-living index for consumer goods in the solid red line. The dashed black line shows average cumulated price changes for the same sample of consumer goods. The gap.is largest before World War II, reversing the conventional view that goods consumption growth was slower before 1945 than in the post-war decades.As Gordon puts it, 'the history of price changes from 1914 to 1947 [is] the black hole where little is known.' This paper sheds light on that black hole. We estimate average annual real goods consumption growth of 5.4% for 1900-1939 and 4.2% for 1946-1980... the era of fastest growth was in the prewar, not postwar, contrary to conventional estimates.

Takeaways by Macro Roundup® AI

  1. Using a quality-adjusted price index built from 5.1 million Sears catalog listings, the authors estimate real goods consumption grew by a factor of 39 between 1900 and 1990, compared with a factor of 10.3 using standard goods deflators.
  2. The authors estimate average annual real goods consumption growth of 5.4% for 1900–1939 and 4.2% for 1946–1980, implying the fastest growth came in the prewar rather than postwar era, contrary to conventional estimates.
  3. The new cost-of-living index implies substantially lower goods inflation than conventional deflators, with the gap largest before World War II, a period Gordon describes as "the black hole where little is known.".

AI Summary. As labor force participation falls, a shrinking group of workers generates the income, taxes, and innovation supporting a growing number of non-workers, leaving the economy more fragile because labor shocks are concentrated among fewer participants. Labor scarcity raises the return on automation, robotics, and software, making growth depend on productivity rather than participation.

Date Posted:
Is Database:
Database

Overall LFP is at a 50-year low. Kedrosky warns, “As fewer adults work or seek work, a larger fraction of voters experience the economy mostly as consumers, retirees, or rentiers, not as workers…[which] changes incentives around….taxation and redistribution.”

How does falling labor force participation reshape growth and fragility?

Core argument: Falling labor force participation leaves a shrinking core of workers generating the income, taxes, and innovation that support a growing number of non-workers, and makes the economy more fragile because labor shocks are concentrated among fewer participants.

As fewer adults work or seek work, a larger fraction of voters experience the economy mostly as consumers, retirees, or rentiers, not as workers. That changes political incentives around wages, immigration, AI, taxation, and redistribution. The economy becomes increasingly dependent on a shrinking core. A shrinking group of workers generates the income, taxes, and innovation that support a growing number of non-workers. The economy becomes more fragile because labor shocks are concentrated among fewer participants. Capital must increasingly substitute for labor. Labor scarcity raises the return on automation, AI, robotics, and software. The AI bet becomes as much about compensating for workers' absence as simple replacement. Growth increasingly depends on productivity, on not participation. Any [productivity] miss will be much more consequential than in the past, given that labor no longer picks up the slack.

Takeaways by Macro Roundup® AI

  1. Falling labor force participation leaves a shrinking core of workers generating the income, taxes, and innovation that support a growing number of non-workers, and makes the economy more fragile because labor shocks are concentrated among fewer participants.
  2. As more voters experience the economy as consumers, retirees, or rentiers rather than workers, political incentives shift around wages, immigration, AI, taxation, and redistribution.
  3. Labor scarcity raises the return on automation, AI, robotics, and software, making the AI bet as much about compensating for workers' absence as replacing them, with growth depending on productivity rather than participation and any productivity miss more consequential than in the past.