Few contrasts better capture today’s conflicting narratives about the U.S. economy than the gap between the University of Michigan’s Consumer Sentiment Index and Robert Shiller’s cyclically adjusted price/earnings ratio (CAPE). Historically, from 1980 to 2020, these two measures were highly correlated. That relationship made sense: the same optimism that boosts consumer sentiment also tends to support higher expectations for economic growth and corporate profits. 

Since the pandemic, however, the two series have diverged sharply. Consumer sentiment has fallen to near-record lows even as equity valuations have climbed to their second-highest level on record. Wall Street appears extraordinarily optimistic, while Main Street seems mired in pessimism.

The media has struggled to reconcile these opposing views of the economy. One common explanation paints a Dickensian picture of an AI-driven economy in which billionaires reap enormous profits while workers are displaced by near-sentient machines. But this doesn’t fit the facts very well. The divergence between sentiment and equity valuations began in 2020, long before AI became a commercially viable technology, and there is still little evidence that AI is meaningfully replacing jobs across the broader economy.

More importantly, neither the prevailing pessimism nor the exuberant optimism accurately reflects the underlying U.S. economy, which remains surprisingly stable. Growth has moderated but is still reasonable, and recession risks appear limited in the near term. The economy expanded by 2.1% in 2025, slower than in the prior two years, largely because of cuts in public spending associated with DOGE and the prolonged government shutdown late in the year.

Growth slowed to 1.6% in the first quarter of 2026, but that weakness was largely driven by a sharp widening in the trade deficit following the Supreme Court’s decision to strike down the Trump administration’s tariffs earlier in the year. Final demand growth—which measures spending rather than output and is often a better leading indicator of economic momentum—rose at a 2.8% pace. Based on currently available data, second-quarter growth appears likely to fall between 2% and 3%.

The Great AI Scare Hasn’t Reached Labor Markets

For all the concern about AI in the headlines, labor markets also remain relatively steady. The U.S. economy has been adding jobs at a moderate pace, averaging slightly more than 100,000 new jobs per month in 2026. April’s job openings rate rose to 4.6%, the highest level since the Trump administration returned to the White House, while the unemployment rate held at 4.3% in May. The sharp decline in immigration last year temporarily reduced labor force growth, but that appears to have been a transitory shock rather than a permanent structural shift. Labor market trends are now returning to a slower but more normal pace.

Although AI has become the explanation of choice for many recent layoff announcements, there has been no meaningful increase in the overall layoff rate or in initial unemployment insurance claims. This is true nationally and even in technology-heavy California, where one might expect AI-related displacement to be most visible. In fact, continuing unemployment claims have drifted downward over the past year, suggesting that displaced workers are still finding jobs relatively quickly.

This implies that AI’s current value to many CEOs may be less about labor replacement and more about providing a convenient narrative for restructuring decisions. There has been a modest increase in unemployment among young workers with bachelor’s degrees—much to the concern of many recent graduates and their parents—but unemployment rates for this group remain substantially lower than for young workers with only a high school diploma or less.

U.S. Debt: The Fiscal Time Bomb Hiding In Plain Sight

So, while the bulls and bears debate which economic narrative best reflects reality, the most significant long-term threat to the U.S. economy receives comparatively little attention: the widening federal deficit and accelerating accumulation of U.S. public debt. The Supreme Court ruling that eliminated tariff revenues effectively destroyed any remaining pretense that last year’s OBBB budget would stabilize the deficit as a share of GDP. The Congressional Budget Office has since released new analyses demonstrating how sensitive its projections are to relatively small changes in interest rates, productivity growth, and labor supply assumptions. Under current assumptions, the fiscal outlook appears only modestly sustainable. Under even mildly adverse scenarios, it becomes clearly unsustainable. 

While it’s true that U.S. debt as a share of GDP remains below levels in countries like Japan and Italy, and roughly comparable to France and Canada, among major developed economies, the United States stands out because of the size of its current deficit. Trends matter more than levels, and U.S. fiscal trends are deteriorating rapidly.

Remarkably, however, there are still few signs that bond markets are prepared to resist financing these deficits. The 10-year treasury remains in the same range it has been in over the last few years, from 4% to 4.5%. The only modest sign of concern has been the growing, but still small, gap between the 10-year and 30-year treasury rates—suggesting fear of inflation in the distant future. As long as capital continues to flow into Treasury markets, the U.S. economy can continue expanding. The real question is what event will eventually change investor sentiment and cause lenders to become more concerned about financing U.S. debt.

Oil Markets and Consumer Spending

One continuing source of strength for the U.S. economy has been consumer spending, despite gasoline prices and persistent headlines about inflation. The war with Iran has disrupted global energy markets, and there appears to be little chance of a rapid resolution despite repeated assurances from the administration. Even so, higher oil prices do not currently pose an existential threat to the U.S. economic expansion or even to the pace of inflation. If that’s difficult to believe, consider that the bond markets are also discounting the potential impact, at least according to the expected inflation information contained in Treasury Inflation-Protected Securities (TIPS) bond pricing.

In fact, it is striking that oil prices have not risen more given the disruption to such a historically important global shipping route. This reflects the dramatic transformation of global energy markets over the past decade, particularly the surge in U.S. oil production made possible by advances in horizontal drilling technology. In 2025, the United States produced 37% more oil than Russia and 40% more than Saudi Arabia, making it by far the world’s largest producer. After decades of deficits, the United States exported more petroleum and petroleum products than it imported from 2022 through 2025.

This energy transformation also helps explain the resilience of the U.S. economy. Prior to the conflict, gasoline prices were near historic lows. Consumer spending on gasoline accounted for just 1.7% of total consumer expenditures in January and February—a record low outside the COVID-19 pandemic period, when mobility collapsed altogether. It is therefore unsurprising that light trucks represented 84% of new vehicle sales last year and that the gas-intensive Ford F-150 remained the nation’s bestselling vehicle.

Even after a roughly 40% increase in gasoline prices, fuel spending has only risen to 2.2% of total consumer expenditures—not enough to meaningfully constrain broader consumption. Non-gasoline retail sales increased by a healthy 1% between February and April, while May auto sales came in at a solid 15.8 million-unit pace (SAAR), above the level recorded one year earlier.

In short, the energy shock has, so far, been manageable. If anything, higher fuel prices may accelerate California’s transition toward electric vehicles. Rising energy prices could also stimulate additional investment in U.S. oil production after several years of restrained capital spending caused by low prices, potentially benefiting California oil fields. Nevertheless, Governor Newsom is unlikely to send President Trump a thank-you note anytime soon.

Overall, consumer spending growth slowed somewhat in 2026, but that shouldn’t be surprising given the concurrent slowdown in population growth. U.S. Census estimates suggest population growth has decelerated to just 0.3% this year, the slowest rate ever recorded outside the pandemic. This is a predictable consequence of the Trump administration’s restrictive immigration policies. Slower labor force growth is likely one of the primary explanations for the moderation in overall job growth.

Still, all of this obscures a more troubling reality. The economy continues to grow largely because both public and private stimulus have expanded in recent months. The elimination of tariff revenues, combined with the obligation to refund previously collected tariffs, pushed the federal deficit to an annualized pace of approximately $2.6 trillion in the first quarter of 2026, according to official data. Treasury balances have not fully reflected this deterioration because the administration has delayed some repayments.

So far, rising federal borrowing has not unsettled bond markets. But when investor sentiment eventually shifts, the adjustment could become disorderly very quickly. The only real uncertainty is when that shift will occur—and whether it is ultimately triggered by the bursting of the current stock market bubble.

How can we reconcile record-low consumer sentiment with stable labor markets, record-high real per-capita consumption, low bankruptcy and foreclosure rates, declining debt-to-income ratios, and rising real wages? The truth is that these trends are fundamentally inconsistent with the depth of pessimism reflected in consumer surveys.
This contradiction may be one of the defining paradoxes of the modern era: a profound sense of social anxiety and populist discontent emerging during what is, by many objective measures, the highest standard of living in American history.

The ‘New’ New Economy

This brings us to the second flawed narrative: the extraordinary optimism embedded in financial markets and the continuing surge in equity prices. After a brief decline at the outset of the Iran conflict, stock markets quickly rebounded to record highs. Nasdaq valuations are now roughly triple their pre-pandemic levels.

Part of this increase is justified. Corporate profits in the United States have roughly doubled over this same period. But profit growth alone cannot explain the full rise in equity prices, which is why valuation measures such as the CAPE ratio remain near historic highs.

The current environment increasingly resembles the “New Economy” bubble of the late 1990s. At that time, internet-related companies drove soaring equity valuations. Any company with a “.com” attached to its name could seemingly guarantee a successful IPO. Today, the equivalent label is “AI.” Then, as now, the dominant narrative promised that everything was about to change. Enormous productivity gains were projected, just as they are today.

The parallels extend further. In the late 1990s, investment in computer equipment and software contributed nearly one percentage point annually to GDP growth—similar to current AI-related investment trends. Venture capital played an outsized role in financing innovation and selecting winners then, just as it does now.

In 1999, economist Kevin Hassett famously co-authored Dow 36,000: The New Strategy for Profiting from the Coming Rise in the Stock Market, published only months before the dot-com bubble began collapsing. Today, Hassett serves as head of the President’s Council of Economic Advisers.

The overselling of new technologies is hardly unprecedented. The AI boom today resembles earlier waves of enthusiasm surrounding electricity in the 1920s and railroads in the nineteenth century. All of these technologies ultimately transformed economic activity. The integration of the internet with personal computing did, in fact, produce the first sustained acceleration in labor productivity growth since the 1960s, while permanently reshaping communication and commerce.

But technological change is typically evolutionary rather than revolutionary. The benefits rarely emerge as quickly or dramatically as investors initially expect because integrating new technologies into existing production and consumption systems takes time. Indeed, numerous recent reports suggest that many corporations rushed into AI adoption only to discover that implementation is far more difficult and disruptive than anticipated.[1]

So far, there is little evidence of a major AI-driven productivity boom in the U.S. economy. Five-year average productivity growth remains below levels recorded a quarter century ago.

The AI boom appears likely to follow a familiar historical pattern. There is little doubt that AI-based systems for processing and generating text will ultimately prove transformative, much as the internet transformed communications in the 1990s. At the same time, the technology is clearly being oversold by both the industry itself and by financial markets.

Despite the excitement surrounding each new generation of AI tools, there is still limited evidence of large economic dividends. Although technology stocks have surged, much of the recent increase in U.S. corporate profits has actually come from finance and manufacturing—the sectors funding AI investments and producing the underlying hardware.

The stock market substantially overshot fundamentals during the late 1990s, as did business investment in information technology infrastructure. When the bubble eventually burst and widespread fraud within the ‘New Economy ‘was exposed, the result was the relatively mild recession of 2001.

That downturn remained limited for three primary reasons. First, the booming economy had temporarily eliminated the federal deficit. In 1999 and 2000, the federal government actually ran budget surpluses, and debt as a share of GDP had fallen from 65% to 55%. This gave policymakers ample room to stimulate the economy once growth slowed.

Second, the U.S. economy was benefiting from a long-term decline in interest rates. Real yields on 10-year Treasury bonds fell from roughly 4% at the start of the 1990s to near zero following the Great Recession.

Third, the housing bubble driven by subprime lending offset much of the decline in household wealth caused by falling equity prices.

Today, the U.S. economy lacks all three of those stabilizing forces. The federal deficit is already the largest ever recorded during a full-employment economy, and federal debt has risen to roughly 120% of GDP. Meanwhile, the inflation surge of 2022 and 2023 has pushed real interest rates significantly higher.

As a result, while the AI bubble strongly resembles the ‘New Economy’ bubble in its formation, the eventual consequences may look very different once it finally bursts.

Footnote

[1] See “AI Productivity Boom is Not Here Yet” Economist https://www.economist.com/finance-and-economics/2026/02/22/the-ai-productivity-boom-is-not-here-yet

“Workslop is destroying productivity” (Harvard Business Review) https://hbr.org/2025/09/ai-generated-workslop-is-destroying-productivity

“Amazon scraps AI leaderboard to stop workers chasing usage scores” Financial Times https://www.ft.com/content/b1a62a7f-6df5-4c90-94ce-64ce9c9961b6?syn-25a6b1a6=1