How much debt is too much? It is the wrong question, for a state as for a business

21 September 2026, 13:14

The Economist lines up three things — rising bond yields, wide deficits, enormous debts — and asks what could go wrong, arguing that the rich world is flirting with a fiscal disaster.

The instinctive reaction, on reading it, is always the same: there is too much debt. It is the reaction I have heard a thousand times from the other side of the table too, when a business owner opens the annual accounts and sees the total borrowings. And in both cases it is a reaction that starts from the wrong question.

The stock says almost nothing

For a state, the ratio of debt to gross domestic product is what the ratio of net financial position to EBITDA is for a company: it tells you how much debt there is relative to what is produced, that is, how many years of output would in theory be needed to wipe it out.

It is a useful measure and it deserves attention, but it measures size, not sustainability. It is no coincidence that the six-times threshold was born in a supervisory exercise rather than in a textbook: the Asset Quality Review, the examination in which, in 2014, the European Central Bank, before taking over direct supervision of the banks of the euro area, went through the balance sheets of the main banks with a fine-tooth comb to check what the loans on their books were really worth. Looking at the companies that had ended up among the non-performing loans, it found that almost all of them exceeded that ratio.

That is why the threshold was not defined as a breaking point, but as the level beyond which deeper analysis is needed. The alarm bell, however, does not tell you where the fire is: for that you need the other indicator.

The right question is whether you can pay it

The second number sets the resources available against how much goes out every year on the debt: interest plus the repayment of the principal instalments falling due. In companies we call it DSCR, and banks normally ask for at least 1.2.

The names of the items change, not the logic. Two states with the same debt relative to output can be in opposite positions: the one paying low interest on long-dated bonds has a manageable situation; the one paying higher interest on bonds maturing in a few months has a problem, even with the same stock.

It is exactly the difference between the company with three million in ten-year loans and the one with two million, all on its overdraft. The second looks less indebted. It is not.

What a state's short-term debt actually is deserves spelling out, because it is not obvious. It is the securities issued short term — in Italy BOTs, Treasury bills of up to twelve months — and, above all, the share of long-term bonds maturing within the year: a ten-year bond issued nine years ago is, today, debt that has to be rolled over now. It is the same criterion we apply in companies, where the reckoning of what goes out each year includes the principal of loans falling due within twelve months, even if the loan runs for ten years. What matters is not when the debt was taken on, but when it has to be repaid.

The real risk is not paying, it is rolling over

This is where the point comes in that, in our line of work, we have learnt from reading accounts year after year, and it is why, when we calculate how much a company pays each year on its debt, we leave the principal of revolving lines out of the denominator — overdrafts, invoice advances, factoring. Not because they do not exist, but because they are not repaid: they are renewed.

For as long as someone renews them.

The risk of those lines is not being unable to pay an instalment; it is that there is never an instalment at all, and one day the bank decides not to renew. It is called refinancing risk, and it is invisible to anyone who looks only at the ratio of debt to margin.

A state lives almost entirely on this mechanism: every year it places the maturing bonds back on the market. It does not "pay off" its debt; it rolls it over. That is why rising yields are more serious news than the debt itself: they do not increase the cost of the existing debt, they increase the cost of the debt that has to be rolled over, and they do so as each maturity comes due.

The verifiable consequence

If this reading holds, then the signal to watch is not total debt going up. It is three different numbers.

The average residual maturity of the debt. If it shortens, the problem draws closer even with the debt unchanged, because every year a bigger slice has to be placed again.

Interest as a share of revenue. This is the thermometer: as long as it stays stable, higher yields have not yet reached the accounts; when it rises, they have.

Demand at auctions. This is the rollover working or not working, and it is the only figure that says anything about the real risk.

The way to check whether the reasoning is right is simple. If over the coming quarters the debt grows but those three numbers stay put, the alarm was premature. If instead the debt stays stable while maturity shortens and interest eats up a growing share of revenue, the problem will have arrived, and it will have shown up there before it showed up in the total.

Why this concerns you too

Not because you need to worry about the public finances. Because the method is the same, and that says something about the method: when a criterion works on such different scales, it is usually because it gets at the substance, not the form.

The stock tells you how much. The flow tells you whether you can cope. The maturity profile tells you when you will find out.

This article was prompted by a news item published in The Economist.