Invention's Other Half, Part 3: The Share That Can't Move
Could your household cover a surprise bill this month? In Latvia, more than four in ten adults say no . Norway’s households owe far more than Latvia’s, some of the heaviest debt in the rich world. Yet Norway is one of the fastest societies on earth at taking up a new tool. Latvia is the slow one. Same word, “debt,” pointing two opposite ways.
The first essay argued that capital buys invention and society buys diffusion ; the second, that debt is the tax that freezes the crossing . I set out to put that tradeoff on one chart. I had measured the debt wrong, and that turned out to be the point.
Two questions, multiplied
A tool spreads only as fast as the people who could take it up, and that is never everyone. The slowest, most stretched majority sets the pace: the frontier firm adopts on day one, the exhausted household adopts never. So the thing to measure is not a country’s average. It is its base : the share of people who could actually make the crossing.
That base comes down to two questions you could ask any adult. Can you learn the new thing? That is the numeracy to pick up a genuinely new tool . And can you survive a bad month while you try? A person retraining is a person betting a paycheck on a year of being nobody, and a household one surprise bill can sink can’t make that bet. Multiply the two shares of yeses and you get the base. It is a product, not a sum: you have to clear both bars, so failing either one drops you out.
The second question is where Part 2 lives: debt read not as the monthly payment it named, but as whether one bad surprise can end you.
I asked the wrong question first
I measured debt the way a bank does first: how much do you owe? I wrote that test down before I looked, so I couldn’t fudge it afterward. Then it backfired: the countries that owe the most turned out to be the fastest at taking up new tools, not the slowest. Norway and its Nordic neighbors led it. The link ran backwards, about minus a half . Measured as a pile of debt, the flexibility tax didn’t just fail to show up. It inverted.
The right question flips the sign
That backwards result was the clue. The Nordics are not fragile . They carry big mortgages, but against solid incomes and real safety nets. Measured as an amount, debt was reading prudence backwards. What Part 2 was always about was never the size of the balance; it was whether a shock ends you. So I asked the household’s question instead: can a family cover an unexpected expense? I had named that test in advance, and the sign flipped. Now debt points the right way: societies whose people can absorb a hit take up the new faster, exactly as the second essay claimed. The burden was never the number owed. It was the missing cushion underneath it.
America fails the other question
With both questions answered the same way, the base is one picture. It ranks each country by the share who clear both bars. The Netherlands sits at the top, Latvia (where we started) near the bottom.
Read the tiers, not a league table. Two different surveys tell me how many people clear each bar, but not who clears both. So the true share probably sits nearer the top of each band than the dot.
The old version of this index ranked the United States near the freest, and that ranking was the failure, not the finding. America’s thin spot is real, but it isn’t a fragile balance sheet. Its $400 cushion is mid-pack: about one in eight adults couldn’t cover it by any means . If anything, that flatters America. The US number comes from an easier test than Europe’s, so on a fair one the cushion looks worse, not better. America’s missing yes is to the first question. A third of American adults score below basic numeracy , the lowest skills floor in the panel. The country that makes the tools has the shakiest footing for learning them.
That is the person the first essay put in an RV and the second put at the mortgage window. They don’t read as broke on a balance sheet. They read as a country that can build anything and, budget by budget, stopped paying to make sure its people can follow.
What this gauge can and can’t say
Now the honest ceiling, stated flat. The joint base does not beat skills alone; at twenty countries the two are too close to tell apart. The buffer question, asked by itself, is only a whisper, too faint to separate from chance. And GDP per capita predicts about as well as either.
So what did measuring the buffer actually buy? It stopped subtracting from skills. Measured as the amount owed, debt was destructive: folding it in dragged the skills signal from strong down to almost nothing. Measured as fragility, it no longer fights skills; it stands beside it, on AI uptake and broad digital diffusion alike. That is the real finding: not two gauges agreeing, but the debt gauge finally pointing the same way as the person it describes.
Why watch the base at all, then, if GDP keeps pace? Because you can’t buy GDP; it is the score at the end, not a dial you can turn. A training allowance is a dial. At twenty countries I can’t pull the base apart from income (they move together), so I won’t pretend it out-predicts GDP. Its value is leverage, not fit: it names a mechanism a country can act on. And the one place the base and GDP part ways is the United States, rich and near the bottom of the base all the same.
We’ve answered both before, for some of us
The cheaper of the two floors to build is the buffer. You can’t hand a country numeracy overnight. You certainly can’t hand it GDP. But you can make it so a bad month doesn’t end a person’s crossing . Two limits keep me honest here. This chart can’t prove a buffer causes faster diffusion . That case rests on the mechanism the second essay laid out, and on the fact that we’ve built the thing before. A cushion alone isn’t the whole fix, because the first essay’s programs only worked when a job waited on the far side. A buffer is necessary, and cheap, and not sufficient by itself.
We have answered both questions at once before. The GI Bill paid for the becoming itself: tuition and a living allowance, no loan to carry. A generation could retrain and move without betting the house on it. It answered the survive-it question so people could go answer the learn-it one. And then it did the thing the ledger misses: it chose who got to stand on it. Segregation and local administration meant Black veterans drew far less. That is one reason the wealth it built still concentrates where it does. It is also the precedent’s other half: a country can build the buffer and still ration it. A ledger that only checks whether the buffer exists will score a rationed one as strength. A country, like a person, faces two questions: did we build the floor, and who did we let stand on it. Other rich countries still run the smaller version. Norway’s state loan fund charges no interest while you study, and turns good grades into a grant. China’s national student loan covers the interest while you’re enrolled, and has waived it after graduation for those who couldn’t find work. Ordinary public finance, in places that decided the next move is a public asset, not a private mortgage.
Deep DiveHow much of the country the GI Bill actually reached
Almost five in ten World War II veterans drew an education benefit from the bill in some form. It paid tuition and a living allowance, with no loan attached. A veteran could go to college or trade school without carrying a payment into the training. That is what made the retrain-and-relocate on-ramp survivable. The same act underwrote zero-down home loans, and the postwar generation pushed homeownership from around 44 percent to 62. A country once paid for the becoming itself, at scale, with public money, and the ledger we built to judge that move called it strength.
We measure invention obsessively, and diffusion barely at all. One whole capacity we don’t collect across nations at all: the freedom to move across the map . Building the floor again is the easy half; we know how. The hard half is doing it for everyone who has to cross this time, not just the ones the last program reached. The capable nations are the ones that made sure their people could keep up. I think we still will, and that we will is the part that is up to us.
A note on the data
The essay ends above. What follows is the method: the two floors, the pre-registered tests, the sign flip, and where the instrument is fragile.
The index is a descriptive base, not a verdict. It reads two floors from primary sources (numeracy for skills, the share who can’t cover a shock for fragility), multiplies them into an addressable base, and draws an uncertainty band around every position because two surveys can’t say exactly who clears both bars.
On the sign flip, and the p-hacking charge it invites. The debt floor was tested first as the amount owed, came back wrong-signed, and was re-specified to fragility. The obvious accusation is that I re-cut the data until it flattered the thesis. The order rules out the cheap version (I recorded the miss before I could edit it) but not the deeper one, that I kept trying debt measures until one worked. What answers that is the mechanism: I switched to the one variable Part 2 was always about, the missing cushion, and named it before I joined it to anything. The two debt measures are cousins of Part 2’s debt-service (the one dollar in nine of after-tax income a US household owes before it decides anything); I use the shock version because it is collected the same way across countries and the payment isn’t.
On the numbers. Part 2 plotted fourteen economies; this tests the twenty where the shock data exists, inside a wider portrait of twenty-seven; the net widened, nothing moved underneath it. At twenty countries the test sees only strong signals, and GDP per capita predicts about as well, so the base is a gauge to watch beside GDP, not a lever to pull instead of it. The US fragility figure is an anchor only: it comes from the Fed’s $400-expense question, a smaller shock than the European survey’s (a month nearer the poverty line), so if anything it flatters the US. Every number traces from the chart to a committed dataset to its primary source; the sources, the construction, the three pre-registered tests, and the full limitations register sit in the panel below.
New here? Start with Part 1: Capital Buys Invention, Society Buys Diffusion.
