The U.S. stock market looks frothy. Valuations hover near record highs, propelled by a handful of mega-cap companies that, in many ways, depend on one another. This analysis asks what would happen if the AI hype were to fade and a significant market correction followed. I find that U.S. households are now more exposed to equities than they were after the 2000 dot-com crash, raising the risk that a market downturn could materially impact the economy.
On the surface, the U.S. economy looks strong. Apart from the Trump-induced collapse in the first quarter of the year (link), growth has been robust for many consecutive quarters. Forward-looking indicators such as the GDP Nowcast point to continued strength. Corporate earnings are solid, supporting rapid increases in share prices, which in turn bolster household wealth and consumption. Bank analysts are optimistic, believing stocks will continue to surge, driven by expansionary fiscal and monetary policy (link). In short, everything appears to be going well.
Beneath the surface, however, doubts linger. GDP growth is increasingly driven by massive investment in data centres and related infrastructure. In fact, the U.S. would have been in recession in the first half of 2025 were it not for AI-related investment (link). Moreover, these investments were previously financed out of firms’ earnings; now, companies are turning more heavily to debt markets (link). At the same time, the stock-market surge looks fragile, as argued in my previous post (link). It is driven by a small cluster of extremely large firms within a single sector that all trade a lot with each other, making the market highly concentrated and valuations stretched. Investors are effectively making a concentrated bet on the AI-sector. Expansionary fiscal policy that stokes bullish sentiment at the same time widens an already enormous fiscal deficit, testing the bond market’s patience. Perhaps underlying growth is not as robust as headline figures suggest.
In this analysis, I play devil’s advocate. I outline some thoughts on how the economy might respond if the AI megafirms were to disappoint. While this is not an outright prediction, it is true that I would not be writing this analysis if I believed a stock-market reversal to be a zero-probability event—or if I were not concerned.
A stock-market drop
The logical place to begin is the U.S. equity market. It has been exceptionally strong in recent years, driven by a few mega-cap companies, as documented in my previous analysis (link).
Here, I ask what might happen if some of these firms disappoint and investors lose confidence. Imagine that the earnings of companies like Nvidia, Apple, or Microsoft surprise to the downside. Several developments could trigger this: new competitors entering the market—remember the brief DeepSeek shock at the start of the year?—or simply weaker-than-expected earnings.
Current equity prices are essentially “priced for perfection”, so any disappointment regarding the AI-linked giants could push valuations sharply lower. Given the outsized weight of these few firms, they would drag the broader market down with them. The key question is: by how much?
As noted last time (link), the market is now valued almost as highly as at the peak of the dot-com bubble in the summer of 2000. After that peak, U.S. equities fell by around 45% over the next two and a half years.
After the financial crisis, the market dropped by roughly 50% between the summer of 2007 and its trough in March 2009. During the Great Depression, from summer 1929 to summer 1932, the S&P 500 fell by more than 80%. Following the post-pandemic inflation and interest-rate shocks, equities fell about 20%.
Which of these episodes is the best comparison? The dot-com bubble is probably closest, although there are differences: while today’s rally has not been as extreme, market concentration is much higher, meaning performance relies on an even smaller set of firms. Industry research shows that high stock market concentration is negatively correlated with future returns (link). Whether today’s situation is better or worse than during the late 1990s therefore depends on which aspect one focuses on.
The 2001 recession was mild
The recession that followed the end of the dot-com boom was comparatively mild. According to the NBER Recession Dating Committee, it lasted only eight months, from March to November 2001. The contraction in economic activity was also modest by historical standards. Figure 1 shows the evolution of private consumption and GDP during the downturn.
Figure 1. Quarterly growth in real US consumption and GDP surrounding the dot.com bubble. Source: FRED of St. Louis Fed and J.Rangvid.
Quarterly real GDP fell by less than 0.5% in Q1 2001 and returned to positive territory the very next quarter. The decline in GDP in Q3 2001 was driven by the shock of 9/11. Throughout the 2001 recession, private consumption continued to grow, as the positive consumption growth rates make clear.
Given that the stock market fell by almost 50%, it is striking how resilient the wider economy proved to be. Some observers take comfort from this episode and infer that any recession triggered by a sharp fall in equity prices will be similarly mild today. I am less confident.
Households are more exposed to equities today
One major—and arguably crucial—difference between the dot-com boom of 2000 and today’s AI-driven surge is that ordinary households are far more exposed to equities now.
U.S. households currently hold USD 129 trillion in financial assets. Of this, USD 51 trillion is invested in equities (link). This number represents directly held corporate equity plus mutual fund shares; in addition, US households are exposed to stocks via their pension savings, but I focus on non-pension equity holdings here in order to provide conservative estimates. That means roughly 40% of household financial wealth is allocated to equities—a record high in the forty years for which data are available, as Figure 2 shows.
Figure 2. US households’ holdings of corporate equity as a fraction of households’ total financial wealth. Source: FRED of St. Louis Fed and J.Rangvid.
Compare this with the situation at the peak of the dot-com bubble. In 2000, U.S. households had allocated around 32% of their financial wealth to equities. Today, that share is a quarter higher, at roughly 40%.
Because equity markets have grown faster than household consumption over the past decades, households’ equity exposure relative to their consumption has also increased. Figure 3 shows the value of U.S. households’ equity holdings relative to total household consumption. In 2000, the value of these holdings amounted to about 160% of annual consumption. Today, the ratio is closer to 250%.
Figure 3. US households’ holdings of corporate equity relative to private consumption. Source: FRED of St. Louis Fed and J.Rangvid.
These developments matter because households typically reduce consumption when the value of their equity holdings falls. The larger their stockholdings relative to their consumption, the larger the resulting cut in spending—and the greater the drag on overall economic activity.
How much would consumption fall if share prices dropped?
Let us explore some illustrative numbers to gauge the macroeconomic implications of a stock-market decline.
Suppose the market fell by 20%, as in 2022. With U.S. households holding USD 51 trillion in equities, such a correction would wipe out around USD 10 trillion of their wealth. A 50% fall—similar to the declines following the dot-com bust and the financial crisis—would erase an extraordinary USD 26 trillion.
A shock of this magnitude would have real economic consequences. When households see their wealth decline, they feel (and are) poorer, and they cut their consumption accordingly. You are unlikely to renovate your kitchen, for example, if your stock portfolio has just halved in value. That lost renovation translates into fewer orders for the kitchen manufacturer, less work for the carpenter, and so on down the supply chain.
How large is the consumption response, more precisely? While the exact magnitude is of course uncertain, academic studies offer useful guidance. The best paper I know on this topic uses detailed household-level data from Sweden to analysis the question (link). The authors find that households in the upper third of the wealth distribution—the group holding the bulk of equities—reduce their consumption by about 3% for each dollar fall in the value of their stockholdings. The richer the household, the smaller the proportional reduction, but the average for the upper tier is roughly 3%.
Households in the lower half of the wealth distribution cut consumption by much more—around 23% per dollar lost—but they hold less than 10% of total household equity wealth. Overall, an elasticity of roughly 3% therefore seems reasonable.
Applying this figure gives the following effects on total private consumption:
- USD 300 billion if the market falls by 20% (3% of a USD 10.2 trillion wealth loss).
- USD 780 billion if the market falls by 50% (3% of a USD 26 trillion wealth loss).
Total personal consumption expenditure in the U.S. is around USD 21 trillion. The percentage declines in aggregate consumption induced by a stock market crash thus amount to:
- 1.4% for a 20% stock-market fall.
- 3.8% for a 50% stock-market fall.
U.S. GDP is roughly USD 30.5 trillion, i.e., consumption accounts for about two-thirds of it. This implies that GDP would decline by:
- 1% if the market falls 20%.
- 2.6% if the market falls 50%.
Since 2000, annual real GDP growth has averaged around 2.2%, and consumption growth about 2.5%. If these figures represent “normal” growth, a 20% equity-market decline would not on its own trigger a recession. A 50% decline, by contrast, likely would. I summarise this in Figure 4.
Figure 4. Effect on US consumption and GDP from a drop in the equity market of 20% or 50%. “Normal” growth is the average growth since 2000. Source: J.Rangvid.
Comparing with the 2000 recession
In early 2000, U.S. households held around USD 11 trillion in equities and USD 34 trillion in total financial wealth, giving an equity share of roughly 32%. When the stock market fell by 50% after the dot-com crash, households therefore lost about USD 5.5 trillion. Applying the 3% marginal propensity to consume out of equity implies a fall in consumption of roughly USD 165 billion. Total consumption at the time was about USD 6.8 trillion, so the percentage decline in consumption was around 2.4%.
This was meaningfully smaller than what we would expect today, simply because households’ exposure to equities was much lower, leading to the prediction that a market decline today would have a larger effect on the economy than back then.
Indirect effects
The figures shown in Figure 3 reflect the direct effects of falling share prices on consumption. But there are indirect effects as well.
As noted earlier, when households cut back on spending—fewer kitchen renovations, fewer durable purchases—businesses see demand fall. The kitchen manufacturer sells fewer kitchens; the carpenter has less work. When the kitchen manufacturer has lower revenues, it may lay off workers or reduce hours worked. These newly unemployed (or underemployed) households then reduce their own consumption, amplifying the initial shock.
The overall effect on GDP depends on several factors. One is the multiplier—the degree to which an initial change in spending ripples through the economy in the way just described. If households reduce consumption by USD 300 billion (following a 20% market fall) USD 780 billion (following a 50% fall), or something in between, then aggregate income declines by a similar amount. If the marginal propensity to consume out of income is, say, 70%, the multiplier is 1/(1–0.7) = 3.33. In this case, the final decline in GDP would be more than three times the direct effect, pointing to a substantially deeper contraction.
Counteracting forces, however, would also come into play. The Federal Reserve would almost certainly cut interest rates to cushion the downturn, and fiscal policy might also be deployed. These interventions would offset some of the multiplier effect.
All told, while we can estimate the direct impact of an equity-market fall with some confidence—if we are willing to accept academic estimates of how much households reduce consumption following a stock market crash. The indirect effects are much harder to pin down. Nevertheless, it is likely that the total economic impact would exceed the direct effect alone.
Conclusion
History does not repeat itself, but it often rhymes. The dot-com crash showed that a major stock-market correction need not trigger a deep recession. But the world has changed. Households today are more exposed to equities than they were twenty-five years ago, and the current market boom is more concentrated.
If the AI story falters, the direct hit to consumption would be larger than in 2000, simply because the pool of household equity wealth is bigger. The indirect effects—operating through labour income, confidence, investment, and financial conditions—could amplify the initial shock. None of this means that a severe downturn is inevitable. It does, however, suggest that the economy is more vulnerable to an equity-market correction than many assume.
I am not necessarily predicting such an outcome, because the outlook for stocks is as always uncertain. But when valuations are stretched, growth depends on a narrow set of firms, and households are more exposed than ever, it seems prudent to think through the potential consequences. Preparing for adverse scenarios is not pessimism; it is sensible economics.
