The Hutchins Roundup brings the latest thinking in fiscal and monetary policy to your inbox. Have something you'd like us to include in the next Roundup? Email us and we'll take a look.
This edition was written by Chesapeake Dowdy, Jack Spira, Zixun Tan, and Louise Sheiner
The 2025 Survey of Consumer Finances included a new question asking households how much of a hypothetical windfall equal to one month of their normal income they would spend over the next 12 months, a measure known as the marginal propensity to consume (MPC). Aditya Aladangady of the Federal Reserve Board and co-authors find that the average reported MPC is 22%, toward the low end of estimates in the literature, which generally range from around 10% to 50%. Consistent with standard models of consumption and saving, MPCs generally decline with household income, from 28% among households in the bottom decile to 14% among those in the top 1%.MPCs vary between roughly 20% and 25% across most of the wealth distribution, before dropping to 15% among the wealthiest 1% of households. Economic models predict that households with few liquid assets (“hand-to-mouth” households) have higher MPCs than those with more liquidity, but the authors find a relatively small difference: 22.9% compared with 20.3%, after adjusting for income. Financial literacy is also thought to increase saving, yet households with less financial knowledge have MPCs only 1.5 percentage points higher than those with more financial knowledge, after adjusting for income.
Yingyuan Chen, Deepali Gautam, and Jason Wu of the IMF analyze financial data on 74 of the largest firms in the artificial intelligence (AI) supply chain—from construction and chipmaking to development and services—to identify the strengths and weaknesses of this economic trend and the possible consequences for macrofinancial stability. They find that, relative to the dot-com expansion, the tech industry's growth today is driven by incumbent firms with demonstrated earnings, allowing investment to expand without a commensurate increase in leverage. On the other hand, they warn that a downturn in AI could have significant consequences for the broader economy, given the large share of stock market capitalization accounted for by AI-related firms. These effects could be magnified by circular financing arrangements, in which large firms invest in companies that are also their customers, allowing financial distress to spread throughout the supply chain. The authors also caution that rapid growth in capital investment has increased the industry's sensitivity to technological obsolescence, since much of the investment is in chips and other assets whose economically useful lives may be shorter than their financing horizons.
Pension funds are often seen as market stabilizers that provide liquidity by absorbing supply shocks and buying when others are forced to sell, as their long horizons and predictable liabilities shield them from short-term market pressures. This view may no longer hold as pension funds increasingly face negative cash flows as they pay more in benefits than they collect in contributions. Using annual cash-flow and asset-allocation data for 186 U.S. public pension funds, Aleksandar Andonov of the University of Amsterdam and co-authors challenge the conventional view of pension funds as stabilizing investors. First, funds with larger cash shortfalls do not hold more cash or U.S. Treasuries. On average, funds hold only about 10% of their portfolios in these assets. Second, with such thin cushions, funds meet unexpected cash shortfalls primarily by selling equities rather than drawing down safe assets: each dollar of negative cash-flow shock is associated with roughly 67 cents of equity sales. Third, funds continue to sell equities in falling markets, when rebalancing toward their target allocations would call for buying. The authors take this as evidence that the sales are driven by liquidity needs: after the global financial crisis, funds were net sellers in 10 of the 11 quarters with negative equity returns. They suggest that funds accept these thin cushions because cash and bonds earn less than stocks and because accounting rules allow funds with higher expected returns to report smaller liabilities.
"In relation to the labour market, we do not yet see a major change in aggregate employment trends but hiring plans are muted, which especially affects the prospects for new graduates. The potential for AI to replace some types of workers may also weaken the bargaining power of workers, moderating wage growth and reducing the labour share in total income. In our surveys, there is significant variation across industries, with workers in sectors benefiting from the AI infrastructure buildout currently more optimistic than those in knowledge-intensive sectors that face greater AI substitution risk," says Philip R. Lane, Member of the Executive Board of the European Central Bank.
"At the same time, it is also important to appreciate that the quantitative scale of the European AI surge is of a different order to the AI boom in the United States or East Asia. Over the last two years, digital investment in the United States has grown twice as fast as in the euro area, while US workers spend two to three times as much of the working week using AI as workers in the largest euro area economies. In terms of the AI frontier, the Stanford AI Index calculates that the United States and China have produced 59 and 35 notable AI models, respectively, while France and the United Kingdom have produced one each. In terms of AI computing capacity, the United States hosts 75 per cent of the global aggregate, whereas Europe hosts just 5 per cent."
