I Built a Model of How the U.S. Fiscal System Could Break. Here Is What I Found.

So. Hi.

I wrote a working paper on U.S. fiscal sustainability and AI-driven labor displacement, and I am releasing it to the world. It is called the Crisis Ratio, it lives at crisisratio.com , and there is an interactive calculator on the site so you can put in your own numbers and watch the thing either hold steady or come apart.

I realize this is not what most of you expected to see from me. The Venn diagram of "people who know me from Little Jumbo," "people who know me from Acoustic Syndicate," and "people who would have guessed I spent the last several months building a structural model of federal debt service" is a very small overlap, and I think it might just be me and Kathy, who gets a prize for being extraordinarily patient about all of this.

So let me explain what this is, how it happened, and why I think it matters enough to put my name on it.

What I am and what I am not

I am not a credentialed economist. I am a musician and an entrepreneur who has been watching the fiscal trajectory of this country for a long time and trying to make sense of where AI is taking us. I own a cocktail bar, play several instruments, produce my own records, compose symphonies, and spent some twenty-five years traversing this amazing continent in a band. None of that qualifies me to write about federal debt service, and I want to be honest about that rather than pretend to be someone I am not.

What I do have is a particular vantage point. I run a business, which means I watch consumer behavior, labor costs, credit conditions, and the texture of how people actually spend money from the inside rather than from a dataset. I have studied economics at the collegiate level and worked extensively with professional economists throughout my technology career, both of which have taught me to think like an economist. And I have spent the better part of a year using AI tools seriously enough to form a view about where the technology is going. Those things combined produced an idea I could not shake, so I chased it.

How this started

About eight months ago I started to genuinely reckon with the future of the technology profession. This happened to coincide with my first real experiences using Claude and watching its impact on coding productivity, which was not incremental. It was exponential. In my own efforts what once took me a week could be completed in an hour. And honestly, even though Claude often operated with the same level of ADHD I often find in junior coders, if I kept it on track the results far exceeded the quality of work I could complete on my own. It also replaced the entire practice of "Google the solution" with a coding partner that could not only provide the solution but diagnose pesky and hard to find bug conditions.

I started asking the obvious next question. If this is what is happening to software, then what happens to the people whose jobs are made of cognitive work? Not factory work, which we have automated before and survived. The thinking jobs. The analysts, the coders, the writers, the mid-level managers, the entire tier of the economy that has spent the last fifty years absorbing everyone displaced from the tiers below it.

Then, at the end of March 2026, debt held by the public crossed 100% of GDP, a level not sustained since the World War II era, when it peaked at 106% in 1946. Now, I don't know about you, but I find that number fairly terrifying.

The more I thought about the potential consequences of AI displacement on the cognitive tier, the more it started to worry me about the stability of the American economy. But rather than being an alarmist about it, I decided to build a model that would give me a real perspective on the situation. The good news? It's not as bad as I initially thought. The bad news? Under the current debt load and facing the next potentially disruptive economic revolution, it could get really bad really fast. That's why I like to say this is "a mathematical warning, served cold."

The idea, in plain language

There are three numbers that govern whether a government can carry its debt: how many people are working, how fast prices are rising, and what it costs the government to borrow. Unemployment, inflation, and the interest rate on the national debt.

Under normal conditions, these three move in ways that offset each other. When unemployment rises, inflation usually falls. When inflation rises, the Federal Reserve usually raises rates deliberately to bring it back down. The system has built-in shock absorbers. This is why fiscal cascades, the kind where a government's debt costs spiral out of its ability to pay, are rare. The variables fight each other. One going wrong tends to pull the others back toward balance.

The Crisis Ratio, as I call it, is a simple way of measuring the relationship: federal debt service divided by the revenue available to pay it. It's a simple concept, really. You could calculate the same kind of ratio for yourself: add up your annual debt payments, everything you pay toward your mortgage and your credit cards in a year, and divide it by what you earn in that year. Lenders call that a debt-service ratio, and they use it to decide whether you can take on more. You can compute the score for the American economy right now (it sits around 0.21, below the rough 0.30 threshold where debt service starts eating the discretionary budget) and you can compute it for various stress scenarios.

Here is the part that kept me up at night: the shock absorbers I described depend on the three variables continuing to offset each other. What happens when they stop offsetting and start moving together instead? What kind of shock could actually make that happen? Because if all three deteriorate at once, the arithmetic gets bad very fast. If you thought 2008 was bad, when the risky borrowers started to default, imagine what would happen if the people who are supposed to be safe bets can't pay their bills.

My answer, and the argument of the paper, is that large-scale displacement of cognitive workers by AI is the first plausible shock that could do exactly that. Not because it is certain, but because of the specific way it would work. It displaces the high-wage, high-debt cohort that drives consumer demand. It pushes the gains to capital rather than wages, which inflates assets without the usual wage-driven inflation the Fed knows how to fight. And it does this to the exact tier of the economy that normally absorbs displacement from below, with no tier above it to do the absorbing this time.

That is the cascade. That is the crisis. The paper walks through it link by link, names which links are well-supported and which are contested, and is honest about the fact that the whole thing is conditional. The mechanism might not fire. It might fire slowly. It might not fire at all. The paper is a diagnostic, not a prediction. It measures what happens if , and then it argues the if is more plausible than we are treating it.

The honest part about how it was built

I read a lot. I built a lot of spreadsheets. And I had long, strange, genuinely useful conversations with AI tools that turned out to be far better thinking partners than I expected.

I want to be transparent about this because it matters, and because the obvious objection to a non-economist publishing an economics paper is "you just had a chatbot write it." That is not what happened, and the distinction is the whole point. The ideas are mine, the argument is mine, the direction is mine. As a "citizen scholar" I don't have the benefit of research assistants, so I turned to a very focused and discerning method of AI utilization to help me develop my model. Yes, I used AI to help me think through the consequences of wide AI adoption. As one friend aptly pointed out, how meta. Specifically, what the AI gave me was a tireless interlocutor that could check my reasoning, find the literature I did not know existed, point out where I was using a term wrong, and help me turn a sprawling set of intuitions into something structured. Every quantitative claim in the paper is sourced to the CBO, the Treasury, the BLS, the Federal Reserve, or peer-reviewed work. The footnotes run deep specifically because I knew this question was coming and I wanted every number to be checkable.

The paper got better through criticism, including criticism that came in after I first put it out. Someone with a statistics background pointed out I was using the word "independent" in a way that meant something different to economists than what I intended, and I rewrote the framing. They pushed me on the Phillips curve, the canonical relationship between unemployment and inflation, and I added an entire section addressing it head-on. Another friend corrected a tax point. I ran a regression on historical data that showed one of my own coefficients was more aggressive than the historical record supported, and rather than hide that, I put it in the paper and explained why. That process, the willingness to be wrong in public and fix it, is the only thing that separates honest outsider work from crankery. I am trying very hard to stay on the right side of that line.

Why my own name

I thought about publishing this anonymously. I have a pen name I have used for other writing, and the safe move would have been to hide behind it. A bar owner writing about sovereign debt is an easy target. It's easy to mock a rock musician who wants to talk about the problems with the yield curve.

I decided not to, because hiding would have let the gatekeeping win before anyone even tried to stop me. I believe the world is deeply lacking in substantive reasoning right now. Not short on opinions, noise, or division, but short on people willing to ask why the system works the way it does instead of accepting the status quo as the way things simply have to be.

If I believe that, and I do, then I do not get to ask the question from behind a mask.

So here I am. With the cocktail bar, the guitar, the symphony, the bass, and now, apparently, fiscal cascade dynamics.

What I am asking of you

Read it, if the question interests you. Play with the calculator. Put in your own assumptions and see what the ratio does. If you think I am wrong, tell me where, because that is genuinely the point. This is a working paper, which means it is offered as the start of an argument, not the end of one. If parts of it are wrong, I want to know. If parts of it are useful, I hope they travel.

The paper, the calculator, and the full methodology are at crisisratio.com .

And if all of this is a little much, that is fine too. Maybe I will just see you at Little Jumbo one of these Tuesdays.