The Stimulus Deception | Rob Arnott Explains What Economists Are Getting Wrong

Episode Summary

Executive Summary: Rob Arnott argues that government stimulus and excessive public spending generally reduce long-run growth, with data showing countries that spend more tend to grow slower over rolling five-, ten-, and intertemporal windows. He advocates a light but necessary state, critiques modern Keynesianism, discusses tariffs as a tactical negotiating tool rather than an ideal policy, and says current market pricing favors value, small cap, and non-U.S. assets over expensive U.S. growth stocks.

Main Topics: Stimulus and government spending vs. growth (Priority: 5/5): The central paper, 'Stimulus Does Not Stimulate,' argues that government spending is persistently associated with slower real per-capita GDP growth across countries and over time, challenging standard stimulus doctrine. What Keynes meant vs. modern Keynesianism (Priority: 5/5): Arnott distinguishes Keynes’s original idea of countercyclical smoothing from modern practice, which often treats stimulus as a permanent growth engine and ignores the need to unwind spending after crises. Finding a sensible size for government (Priority: 4/5): The discussion identifies a rough range where government is needed for rule of law, infrastructure, and safety net functions, but becomes harmful when it expands too far beyond that range. Debt, crowding out, and national balance sheets (Priority: 4/5): The conversation frames debt less as an imminent default event and more as a force that crowds out private borrowing and investment, using Japan and the U.S. as contrasts. Tariffs, trade deficits, and negotiation (Priority: 4/5): Arnott argues tariffs should ideally be zero, but views Trump’s tariff threats as a negotiation tactic and a Laffer-curve-style revenue instrument rather than a stable industrial policy. Asset allocation and expected returns (Priority: 5/5): Research Affiliates’ capital markets expectations tool suggests expensive U.S. equities—especially growth stocks—have low forward returns, while value, small cap, and non-U.S. assets look more attractive. Crowded markets and mean reversion (Priority: 4/5): The segment on market valuation emphasizes that current U.S. concentration and growth-value spreads are historically extreme, creating both risk and opportunity depending on mean reversion.

Key Arguments: Government spending is largely a matter of reallocating resources and picking winners and losers, not creating new wealth; more spending tends to slow growth. The negative relationship between spending and growth appears consistently in rolling three-, five-, and ten-year windows and across most countries over time. Keynesian policy is often misapplied today: Keynes supported countercyclical smoothing, not permanent expansion or the idea that spending can permanently raise growth. A minimal state is necessary for rule of law, property rights, policing, fire protection, roads, and borders, but spending beyond a certain level becomes drag rather than support. High national debt is damaging mainly because it crowds out private credit and investment, even if it does not immediately trigger default or collapse. Tariffs are best viewed as distortions to trade that can sometimes be used tactically in negotiations; very high tariffs function as barriers, not modest revenue tools. Current U.S. equity markets are priced richly versus history, so expected returns are modest for large-cap stocks and especially weak for growth stocks. The biggest forward opportunities likely lie in cheaper assets such as value, small cap, and non-U.S. equities if mean reversion occurs. Crisis responses should be light-handed and targeted; governments should admit uncertainty rather than claim certainty or create permanent programs after emergencies. Market participants overfocus on past returns; a better forecast comes from yield, growth in income, and the likelihood of mean reversion.

Data Points: Historical spending dataset: 80 years - Arnott says Research Affiliates assembled 80 years of government spending data to test the stimulus-growth relationship. Rolling windows tested: 3-year, 5-year, and 10-year - The paper evaluated growth vs. spending over multiple rolling horizons. Five-year result consistency: Negative in every five-year span of the last 40 years - Arnott says the spending-growth relationship was negative across all five-year periods in the sample. Extended test horizon: Back to 60 years - He says an additional paper extended the analysis to 60 years with the same result. Government spending vs growth slope: 10% more spending relative to GDP ≈ 0.75% slower growth - Arnott cites the approximate slope from the scatterplot relationship. Country spending vs growth: Countries that spent more grew slower - This is the recurring finding from both cross-sectional and time-series tests. Government spending sweet spot: 15% to 30% of GDP - Arnott argues this range is enough for core state functions but beyond that spending starts to hurt growth. Average developed-economy spending in early 1960s: About 30% of GDP - He notes this as the historical baseline for developed economies. Average developed-economy spending today: About 45% of GDP - Arnott says developed economies have moved materially toward larger government. U.S. national debt: $36 trillion - Used to illustrate the scale of U.S. public indebtedness. U.S. GDP: A little under $30 trillion - Provided for comparison with national debt. Debt-to-GDP ratio: About 120% - Arnott cites this as the common headline measure of U.S. debt burden. Debt-to-revenue illustration: About 700% of annual tax revenues - He estimates debt relative to government revenues as the more relevant burden measure. Japan national debt: Roughly 250% of GDP - Cited as an example of very high debt without immediate catastrophe. Ireland vs UK per-capita GNP change: From about 50% below to about 20% above the UK - Arnott uses Ireland to argue lower government burden can accompany stronger growth over time. Growth-doubling examples: 3% = 24 years, 1.5% = 48 years, 1% = 70 years - He uses compounding to show how small growth differences dramatically affect long-run prosperity. S&P 500 concentration: Top five stocks nearly 30% of the index - He cites this as the highest concentration ever. Dot-com peak concentration: About 18% - Used as a historical comparison to current market concentration. U.S. large-cap expected return: 3.9% over 10 years - Research Affiliates’ estimate for U.S. large-cap stocks. U.S. large-cap return components: 1.4% yield and 5.3% growth - These are the nominal building blocks before mean reversion effects. Shiller P/E historical average: About 18x - Arnott compares this to current U.S. valuation levels. Shiller P/E entering current quarter: About 34x - Used to argue that U.S. equities are expensive versus history. U.S. large-cap growth expected return: About 2.5% - He says growth stocks are priced for especially weak forward returns. Tariff example: 10% as a 'speed bump' - Arnott uses this to distinguish modest tariffs from prohibitive ones. China tariff example: 145% briefly raised to 245% - Used to illustrate tariff escalation as a barrier rather than a small tax.

Pivotal Quotes: "Countries that spent more grew slower." — Rob Arnott: Core conclusion from the paper on stimulus and government spending. "I don't question the notion of government spending going up in a crisis, but can we please have it go back down after the crisis passes, please?" — Rob Arnott: His summary of how countercyclical policy should work, and where modern practice goes wrong. "The correct rate of tariffs is zero." — Rob Arnott: His libertarian baseline on trade policy before discussing tariffs as negotiation tactics.

Implications: For investors, expensive U.S. growth stocks may offer low future returns while value, small cap, and non-U.S. assets appear more attractive. For policy, the episode argues for restraint: limited, temporary crisis intervention; lower long-run spending; and skepticism toward tariffs and industrial policy.