The Shape of Economic Risks

How a Stagflationary Shock Could Start an Economic Correction

There’s significant talk these days about tech bubbles and how that relates to today. This talk often predicts a market correction. While this is possible, even probable in a long enough term, the timing of such a correction is inconveniently inaccessible.

The bursting of bubbles based on sentiment is not very predictable. For one, when optimism gets into high gear, it’s usually tied to underlying dynamics that carry a lot of uncertainty. Who really knows what work AI will and won’t be able to do, and more importantly, when?

Second, since bubbles are generally based on predictions of future performance, there’s little immediate forcing function. If the current quarter’s numbers come in low, but your target is 5 years away, it’s not that hard to maintain your 5-year optimism. Shrugging off short-term evidence as non-relevant to the long-term outcome has some sensibility.

Third, even someone who knows enough to be certain that optimism is too high will be tempted to hang on and try to time their exit to take best advantage of others’ optimism. This will delay the bursting of a bubble. This dynamic can’t last forever, but it complicates timing significantly.

With this in mind, too much weight may be placed on the idea that a correction would start within the realm of tech and AI investments. This type of correction isn’t the only one we should be worried about, though. A different, though highly complex, scenario is one that shows a smooth connection from point A (now) to point B (the 5-year optimistic view) isn’t possible, because the “main street” economy that supports it begins to buckle.

The Stagflationary Shock

This presents a concerning story: Tariffs create extra costs, and tariff policies create uncertainty that reduces “main street” business investment. Job growth slows as a result. At the same time, crackdowns on immigrants cause business closures and slower business formation in areas that drew workers from immigrant populations, such as farming, restaurants, and residential or informal construction (like renovations and single-family homes).

This scenario creates two powerful, opposing economic forces simultaneously:

An Inflationary Supply Shock: Tariffs (a tax on goods) and labor shortages (raising wage costs) reduce the economy’s productive capacity, pushing prices up.

A Recessionary Demand Shock: Business closures and slowing investment reduce aggregate demand, putting downward pressure on prices.

This aggregate balance is further complicated because these forces may not be uniform. The inflationary pressures from tariffs are broad, while the recessionary effects of workforce loss and business closures are localized to specific industries and regions. It’s possible for the economy to experience both shocks simultaneously without them canceling each other out on a national scale.

Individuals could absorb the pains from those effects by using savings or by taking on more debt, but they’ll also probably stop spending as freely.

When spending slows, it affects retailers, which indirectly slows advertising revenue, or directly slows online sales.

Risk One: The Disinflationary Bust

The first risk is that these forces don’t simply cancel. The inflation from tariffs itself acts as a tax, slashing consumer purchasing power and causing a recession. In this feedback loop, an initial inflationary spike is followed by a recession that becomes so deep that it overwhelms the inflationary pressure from tariffs. Demand is so thoroughly destroyed that businesses cannot pass on their higher costs and instead reduce production and employment. This would not lead to stagflation, but to a deep disinflationary bust—a severe recession where inflation rapidly slows, or even turns negative.

This specific risk—a disinflationary bust originating from “main street”—is one the U.S. economy has arguably been facing. It has, so far, been largely avoided or masked by a powerful counter-force: a massive wave of investment from tech companies and rising stock market optimism fueled by the promises of AI, which has provided a floor for the wider economy. This reliance, however, creates a fragile co-dependence that leads directly to the other major risks.

Risk Two: The Credit Competition Crisis

This fragile balance, propped up by AI, is still dependent on the “main street” consumer. While they have been absorbing the pain from inflation and job slowdowns by using savings or taking on more debt, this cannot last forever.

The second major risk begins when this consumer resilience is exhausted. As “main street” spending contracts significantly, it directly hits the revenue of the tech companies themselves—slowing advertising, e-commerce, and enterprise sales.

This creates a new, dangerous dynamic. Tech companies, which had been funding their massive AI investments from free cash flow, now see those cash flows dry up. They are forced to turn to debt markets to continue investing. Suddenly, these tech giants are in direct competition for a limited pool of credit with the very consumers and “main street” businesses that are also desperate to borrow to survive. This competition for credit would cause interest rates to spike across the board—even without government action—threatening to choke off both the AI boom and any hope of a “main street” recovery.

This is the point where the crisis becomes systemic and demands a government policy response, leading to the next set of risks.

Risk Three: The Crisis Apex and Policy Dilemma

Faced with the systemic credit crisis from Risk Two, the government is forced to intervene. At this apex, the economy is balanced on a knife-edge, with two new, opposing market dynamics coming into play, even as the government weighs its options.

On one hand, there is a slender thread of hope: AI itself might finally start returning tangible, widespread value to “main street,” allowing new business formation and employment to grow again. If AI-driven productivity gains were to suddenly make “main street” businesses profitable and credit-worthy, it could provide a market-based escape. While the dynamic is simple, the reality of this is unclear, as it’s not known what this value would look like or if it could arrive fast enough.

On the other hand, this is the exact moment that AI optimism itself might end. Faced with cratering “main street” demand and a real credit crunch, investors might finally decide the “5-year optimistic view” is no longer plausible. This wouldn’t be a solution, but a different, compounding dynamic: a full-blown tech correction on top of the “main street” recession, leading to an immediate crash (similar to Risk One, but far more severe).

Barring the slender thread of hope, and assuming AI optimism holds just long enough to influence policy, the government must still act. Its options are severely limited by the original stagflationary shock: inflation (from tariffs) is still high, even as all sectors are desperate for cheaper credit.

In this scenario, the government might still try to counteract the recession and credit crunch by lowering base interest rates. This choice, however, creates a new, highly unstable set of risks.

This policy creates a bifurcated economy. It doesn’t mean credit only benefits AI, but rather that in this environment, banks engage in risk-based credit rationing.

This policy is highly unstable. It fails to help “main street” (which is too risky to lend to) while simultaneously fueling an asset bubble and adding inflationary pressure (by pumping cheap money into the “hot” AI sector).

Risk Four: Path A - The Politically-Driven Hard Landing

Faced with the unstable, inflationary, and politically toxic situation created by Risk Three, policymakers arrive at a critical fork in the road. The first, and most economically orthodox path, is the hard landing.

The policy in Risk Three is politically toxic. A government facing an electorate where everyone is moderately affected by inflation—a universal and highly visible political poison—is far less likely to tolerate it than unemployment, which affects a smaller group severely.

The political pressure would be immense for the Federal Reserve to do the opposite of Risk Three: keep interest rates high (or raise them further) to crush inflation, even at the cost of deepening the recession. This “Volcker-style” hard landing, driven by political necessity, would itself trigger the tech correction by making financing prohibitively expensive, leading to a bust in both tech and “main street.”

Risk Five: Path B - The Idiosyncratic Low-Rate Gamble

There is, however, a second, less orthodox path. An idiosyncratic political administration, one more concerned with short-term stock market performance and headline employment figures than long-term price stability, might attempt to force the Federal Reserve to keep interest rates low, despite the stagflationary pressures.

This path involves rejecting the painful (but historically validated) lesson of the 1970s and 80s, and instead gambling that the economy can “grow its way out” of inflation. In this scenario, the government essentially accepts the unstable dynamic of Risk Three. It continues to fuel the AI asset bubble with cheap credit while hoping for a productivity miracle, all while persistent inflation erodes the value of “main “street” savings and wages.

A Choice Between Two Harmful Paths

Neither fork in the road leads to a happy state. Path A (Risk Four) is the orthodox, “Volcker-style” hard landing: a deliberate and deeply painful recession. It is a harmful path that would cause significant unemployment and business failures, but it is a known strategy designed to crush inflation and prevent an uncontrolled spiral.

Path B (Risk Five) is an unorthodox gamble that likely leads to a worse fate. If President Trump chooses this second path—cutting and holding rates during stagflation—the scenario ends grimly. The policy would fail to stop the main street recession while runaway inflation, now unchecked by the central bank, takes hold.

This onset of runaway inflation would eventually force the government’s hand anyway, compelling them to aggressively increase base interest rates after all. This U-turn would force AI investment to finally slow, but only after significant economic damage has been done. In the end, the economy would be a weaker one, with more inflation, more unemployment, fewer productive businesses, and a complicated unwinding of debts.

How likely is this?

The likelihood of this scenario appears uncomfortably high, as the key premises are not hypothetical but are already visible in late 2025 data.

1. The Stagflationary Shock is Underway. The “Stagflationary Shock” is not a future risk; it is the current reality. Job growth has stalled, averaging just 29,000 per month over the last quarter (near-recessionary levels) and showing “little change since April” according to the Bureau of Labor Statistics. This is happening while inflation remains “sticky” at 3.0%, fueled by tariffs. This combination of weak growth and persistent inflation is the textbook definition of stagflation.

2. Tech Vulnerability is Real. “Risk Two” is also proving plausible. The assumption that Big Tech would fund the AI boom from its own cash reserves is incorrect. Tech giants are already turning to debt markets. Recent reports show AI capital expenditures are approaching 94% of operating cash flow, forcing massive bond sales, including Meta’s recent $30 billion offering. This confirms their new vulnerability to a “main street” downturn. That’s not to say that tech itself is at risk of collapse, but more that they aren’t insulated from the rest of the economy.

3. Fed Independence is Under Attack. “Risk Five,” the “idiosyncratic low-rate gamble,” has become a tangible political risk. The Federal Reserve is under “immense pressure” from the administration. The recent appointment of a political adviser, Stephen Miran, to the Fed’s board—who is already dissenting in favor of the larger rate cuts the White House desires—demonstrates that the Fed’s traditional independence can no longer be taken for granted.

Because the three critical links in the chain—the stagflationary shock, the tech debt reliance, and the erosion of Fed independence—are already in place, this alternative path to a correction is not just a theory, but an unfolding reality.

What should we do?

Confronting the question of what would reduce these risks, the most direct path is to remove the initial stagflationary shocks—the tariff policy and the workforce disruptions from restrictive immigration policy—which are the root cause of the entire risk cascade. Tariff policy is easily undone, especially if it comes with a commitment to not make that mistake again. Immigration policy shifts should be more forward-leaning, creating an environment for orderly, legal immigration that meets workforce needs.

Reducing our risk by these means is the most obvious and sound policy. The question though that arises is, what could convince the Trump administration to follow through with measures that contradict the largely irrational policy it’s pursued so far?

Postscript: Quantifying

The scenario here, it’s plausible, but not quantified. Job numbers, cash flows and attacks on Fed independence, do neutralize the strongest counterarguments. That still doesn’t make this a quantified scenario, and without quantification, the timeline is unknown. Anytime a timeline is unknown, you have to also consider the possibility that the timing is “too far away to matter”.

While there’s a solid story that main street weaknesses could overwhelm optimism, financial tricks, and tech’s contributions, it’s also possible these prop up main street and slow main street erosion. In that time, policies could reverse. Maybe it could even be so long to allow new elections and new policies from that.

It would be nice to have numbers. Numbers are not easy though. Quantifying this scenario presents a significant challenge. I have no intention of attempting a full econometric model of the U.S. economy, which is the minimum that would be required to know the outcome of the interaction of these forces. I’d watch for others taking on that work, but do you really need to wait?

I do have the idea that I can take a deeper look at the tech economy, and think through how the specific companies or layers would be impacted by changes in wider economic behaviors. That’s an interesting question all of its own, but I also think it helps understanding how powerful feedbacks would be. This is still a big topic, so it will be a future post.