Econbrowser Blogger Menzie Chinn: The Neutrality of US AI Investment Retreat on GDP Does Not Equal Consumption Neutrality
Econbrowser blogger Menzie Chinn points out that the retreat in AI capital expenditure is nearly neutral in US GDP accounting—reduced investment is offset by reduced imports—but the drag on consumption transmitted through falling asset prices and shrinking household wealth is real. Citing research by Bigot and Espic (2026), he estimates that about half of US consumption growth in 2025 came from stock market wealth effects, with a marginal propensity to consume out of wealth of approximately 0.027. Even if consumption growth slows as a result, US GDP growth will not turn negative, but disorderly market adjustments would raise uncertainty, further suppressing aggregate demand.
The direct impact of the retreat in AI capital investment on US GDP may be nearly neutral in an accounting sense. Menzie Chinn notes that the surge in capital expenditure was largely offset by imports in US GDP accounting—the contribution of imports of computers, peripherals, and semiconductors to quarterly data was opposite in direction and similar in magnitude to the contribution of investment. Although data center construction is not included in the capital expenditure component, adding it in would not change the overall picture. If the investment multiplier and import multiplier are roughly equivalent, then the retreat in investment represents an approximate zero-sum accounting event at the US GDP level.
But this is not the whole story. Chinn distinguishes between two transmission channels: one is the mechanical offset in GDP accounting, and the other is the behavioral channel—where changes in expectations drive asset price revaluation, declining asset prices erode household net worth, and shrinking net worth depresses consumption growth through wealth effects. He cites research by Bigot and Espic (2026), estimating that of the approximately 2% US consumption growth in 2025, about half came from increases in stock market wealth; this estimate is based on a marginal propensity to consume out of wealth of approximately 0.027, meaning that for every additional dollar of stock market wealth, consumption increases by about 2.7 cents.