In our business, precision is power. We build predictive models for clients making some of their biggest fiscal decisions. A couple of months ago, one of our best performers, a model that had nailed inventory needs quarter after quarter, started to drift. Its forecasts weren’t wrong, exactly. Just… fuzzier. The prediction intervals widened. The signals got noisier.

When we investigated, the culprit wasn’t the math; it was the map. Our assumptions, built on decades of reliable data from America’s gold-standard statistical agencies, were suddenly out of sorts. Tariff tremors, policy-driven supply distortions, and even now a federal shutdown have all disrupted the data we depended on.

So, the model wasn’t “broken”. It’s that the longstanding rules by which our economy functions have changed.

That’s why my alarm bells go off every time I hear the Fed’s current path compared to Alan Greenspan’s “perfect soft landing” in 1995. Powell may be flying a similar aircraft, but he’s navigating different skies. Inflation isn’t tamed, trade is tangled, and parts of the dashboard are dark. The 2025 economy is not the 1995 economy, and the risks of acting as if it is could be steep.

As Kent and I highlighted two weeks ago, conventional recession indicators suggest a healthy economy, though recent trends in the labor market for college graduates paint a different picture. We pointed to some appalling anecdotes and statistics, including the case of the Computer Science major who submitted 5,762 job applications, only to hear back from none.

This reality begs the question of whether job creation in the age of AI will ever speed up as individuals, companies, governments, and educational institutions adapt. If so, when?

In a podcast interview, LinkedIn’s Chief Economist, Karen Kimbrough, expresses optimism for the future of work in the age of AI. She acknowledges the unusually slow hiring rate for recent grads, but nonetheless offers the reassuring take that current trends are cyclical and that AI-embracing grads will fare the best in the labor market.

If there really is a rainbow after the storm and her forecasts come to fruition, when will we start seeing the evidence of it? What actions must be taken by companies, governments, and job seekers themselves to ensure a future where AI development works in favor of job seekers rather than against them? What does it mean for job seekers to “embrace AI?” What incentives underpin whether or not we, as a society, will work to ensure that AI does not leave today’s college grads behind?

In 2025, if you asked the average economist about the U.S. labor market, the answer might sound reassuring: the unemployment rate is holding steady around 4%, inflation is relatively under control, and job growth continues month after month. But ask a 23-year-old college graduate with a crisp new diploma and a browser full of unanswered job applications, and you’ll hear a different story.

A quiet crisis is unfolding in the United States that eludes headline economic metrics yet is painfully evident in overflowing inboxes, mounting loan statements and waning optimism among young Americans entering the workforce. In mid-2025 the unemployment rate for recent U.S. college graduates reached 5.8%, according to Bureau of Labor Statistics data, its highest level in more than a decade apart from the pandemic spike. More strikingly, this graduate jobless rate now exceeds the national figure, overturning the long-standing pattern in which new degree-holders enjoyed lower unemployment than the wider labor market.

Earlier this year, John, a seasoned professional with a major firm, decided it was time for a leap. The stock market had been climbing steadily, financial headlines were full of optimism, and investor sentiment seemed to signal a revitalised economy. And John was getting increasingly tired of feeling left out. So, convinced that growth had returned, he left his stable job to join a consumer–facing start‑up.

Six weeks later, John was unemployed.

The start‑up’s sales projections were built on an assumed rebound in household demand, but real personal consumption expenditures had stalled. Delinquency rates on credit‑card loans were climbing toward the highest level recorded since the early 2020s, and unemployment‑insurance claims had been trending upward since April. Meanwhile, corporate profits after tax had registered their first significant drop in more than two years, and real exports had flattened.

John’s decision was shaped by a popular narrative: when equities rise, the economy must be healthy. That narrative is both persistent and dangerous.

Among colleagues who follow the U.S. economy closely, shifts in policy direction don’t usually come as a surprise. Yet, in recent weeks, a series of reports has indicated that the administration aims to select the next Federal Reserve Chair chiefly for ideological loyalty, favoring a candidate inclined to reduce interest rates regardless of macro dynamics; the prospect has given both these authors a pause.

As trained monetary and financial economists, we’ve spent years studying the delicate architecture that allows the Federal Reserve to function independently from political pressures. When that independence is threatened, so too is the foundation of macroeconomic stability.

This moment, in our view, requires more than private concern. It calls for public reflection.