1929, 2008, and where the next one comes from

Ask what caused the crash of 1929 and you will usually be told: speculation. Ask about 2008 and you will hear: greed, or subprime mortgages. Both answers are true in the way that “fire” explains a burned building. They describe the occasion rather than the mechanism, and because they describe the occasion, they are useless for the question people actually want answered, which is where does the next one come from.

The mechanism is more specific than either story, and once you have it, the student loan question becomes answerable — though not, I think, in the direction most people expect.

1929: leverage meeting a margin call

The 1920s produced a genuine boom with genuine foundations — electrification, the automobile, radio, real productivity growth. What turned the eventual downturn into a catastrophe was the financing structure sitting on top of it. Investors could buy stock on margin with a small fraction down, borrowing the rest against the shares themselves. Investment trusts layered leverage on leverage, sometimes several deep, so that a modest decline in underlying assets could erase the equity above them entirely.

That structure has a specific failure mode. When prices fall, lenders demand more collateral; borrowers must sell to raise it; the selling pushes prices lower; the lower prices trigger the next round of calls. The mechanism is self-reinforcing and indifferent to whether the underlying assets were worth anything — it is arithmetic, not sentiment.

What made it a depression rather than a crash was what happened next. Banks held equities and loans to speculators; failures spread; there was no deposit insurance, so ordinary depositors ran; and the Federal Reserve did not act decisively as a lender of last resort while the money supply contracted sharply. The market break was the trigger. The banking collapse and monetary contraction were the disaster.

2008: the same machine, in different clothes

Change the collateral from stock certificates to houses and the structure is recognizable. Households borrowed against homes with little money down. Those mortgages were pooled into securities, tranched, rated, and pooled again, and the resulting paper was held by financial institutions running leverage that left thin equity beneath large balance sheets. Derivative contracts written against that paper concentrated the exposure further into a small number of counterparties.

Then the same loop. Housing prices fell, the securities were marked down, leveraged holders had to sell into a market with no buyers, marks fell again, and because nobody could see precisely who held what, every institution became suspect and short-term funding froze. Lehman failed; the money market seized; the machinery of ordinary credit stopped.

The crucial difference from 1929 was the response rather than the cause. This time the central bank did act as lender of last resort, aggressively and early, and deposit insurance existed. The result was severe — a deep recession and a market decline over half from peak — but not 1932.

The recipe, stated plainly

Systemic crashes need four ingredients together. Leverage, so small price moves wipe out equity. Collateral that can be sold in a hurry, so falling prices force more selling. Opacity, so nobody knows who is exposed and everyone is treated as exposed. Interconnection, so one institution’s failure is another’s missing asset. A problem missing any of the four can be enormously painful without being systemic. That distinction is the whole of what follows.

Now run student debt through it

The concern is understandable and the underlying problem is real. Balances are historically large, the burden falls hardest on people at the start of their working lives, and the return on some of that borrowing has been poor. None of that is in dispute here.

But test it against the recipe. Leverage: the loans are overwhelmingly held or guaranteed by the federal government rather than sitting on leveraged private balance sheets — the government does not face margin calls, and losses show up as fiscal cost rather than institutional insolvency. Collateral: there is none. A degree cannot be repossessed and sold, which is bad for lenders in the ordinary sense but means there is no asset to dump into a falling market, and therefore no fire-sale spiral. Opacity: the exposure is about as visible as a financial exposure gets, sitting on public balance sheets and reported publicly. Interconnection: the private slice that is securitized is small relative to the mortgage complex of 2007, and it is not the collateral underpinning short-term funding markets the way mortgage paper was.

Three of four ingredients are largely absent. That is not a reason for comfort — it is a reason to expect a different kind of damage. What student debt plausibly does is act as a long, quiet drag: delayed household formation, delayed home purchase, delayed retirement saving, fewer new small businesses, and a fiscal cost that lands wherever forgiveness or default eventually places it. That is a serious generational and political problem, and it is a slow one. It grinds rather than detonates. Slow problems get less attention precisely because they never produce a Black Tuesday, which may make them more dangerous to ignore, but it does not make them a crash.

Where the ingredients are actually accumulating

The useful thing about a checklist is that it points somewhere. If you are looking for the next systemic event, look for the four ingredients together rather than for whichever debt number is largest.

That directs attention toward places where lending has moved away from regulated, transparent balance sheets into structures where valuations are not continuously marked and the holders are harder to enumerate — the growth of private credit, commercial real estate exposures concentrated in particular lender types, and, closer to this publication’s beat, the increasingly debt-financed build-out of AI infrastructure. That last one deserves watching for exactly the reasons the earlier pieces here describe: specialized, rapidly depreciating assets, financed with borrowed money, against revenue projections that assume the demand curve keeps going. Whether those ingredients are combining in dangerous proportion is an empirical question I am not answering today, and anyone who answers it confidently in either direction is telling you about their conviction rather than the evidence.

What the historical record does support is narrower and more useful. The largest number is rarely the dangerous one. The dangerous one is the leveraged, opaque, collateralized, interconnected one — and it is usually being described, right up until the week it matters, as well understood and adequately contained.