As our total stock of debt recently surpassed the $40 trillion mark, commentators warned about the dangers of default risk. This is an important question to ask: will the government (taxpayers who fund it) ever be able to pay back such a large amount of debt?
In a new NBER working paper, MIT economist Ricardo Caballero asks a different question: even if U.S. public debt remains safe and the government never defaults on its obligations, can issuing more of it still hurt the economy? His answer is yes, and we are nearly at the point whereby this starts to occur.
U.S. Debt as a Safe Asset
For years economists have highlighted how investors value the liquidity and safety of U.S. treasuries. The argument is that government bonds are not just IOU’s issued to fill budget gaps, but they are also the economies go-to safe asset.
For example, banks use treasuries as collateral and money managers use them as a savings vehicle. Based on these dynamics, proponents of the safe asset theory argue that issuing more debt makes people feel wealthier, which encourages them to spend more, driving demand higher. Under this model, as long as the government doesn’t go bankrupt, more debt equates to more economic stimulus.
Of course, anyone well versed in the Fiscal Theory of the Price Level (FTPL) would counter that the backing of newly issued debt matters just as much. If debt is issued without a credible commitment to future surpluses, then you don’t get stimulus, you get inflation.
If debt outpaces the public’s expectations of future fiscal discipline, the market will revalue those ‘safe assets’ down via inflation until the real value of the debt matches the real value of the government’s expected surpluses.
Safety Isn’t Free to Produce
Aside from the FTPL critique, this new NBER paper makes a more interesting observation. Government debt only stays liquid and tradeable because banks and bond dealers actively make markets in it. They do this by including it on their balance sheets, buying it from sellers, financing it, and re-selling large amounts of debt stock every time bonds mature.
All of this marketability requires bank balance sheet capacity, and balance sheet capacity is not infinite. It is capped by leverage and capital requirement regulations.
We can think of this as a grocery store shelf. The first units of debt are cheap to stock, but as the government continues to issue more debt relative to the size of the shelf (the financial systems ability to absorb it), dealers have to work harder to continue taking on greater risk to keep placing it, and they charge more for that higher risk.
These pressures on the financial system show up as a spread, which Caballero calls the Treasury absorption premium). Importantly, the government effectively pays for this premium spread even though the bond itself doesn’t stop being “safe”.
The Debt Laffer Curve
At low levels of debt, proponents of the safe asset theory argue that issuing more debt boosts the economy. However, as the debt grows larger relative to the financial system’s ability to absorb it, the cost of safety rises and eats into that positive effect. Eventually more debt issuance passes a point whereby more issuance makes demand worse, not better. This is the safe-debt Laffer curve.
Caballero attempts to measure this “cost of safety” for the U.S. treasury market using bond-swap spreads and inflation-protected bond pricing as his data.
He finds that between 2015 and 2026, the marginal cost of producing a safe treasury dollar has more than doubled from about 80 basis points (bps) to about 187 bps. For the positive wealth effect, he finds an extra dollar of safe debt of roughly 330 bps. The safe debt margin is, therefore, about 143 bps.
Based on 2026 Congressional Budget Office projections, Caballero estimates that this safe margin is shrinking by about 18 bps every year, and this rate of shrinkage will accelerate if debt grows faster than forecast. In other words, we are less than 8 years away from being on the wrong side of the debt Laffer curve.
An Important Caveat
There is an important caveat to Caballero’s framework. The peak of the safe-debt Laffer curve should not be interpreted as the point at which government debt first begins to harm economic growth. Traditional crowding-out effects can operate well before the financial system reaches its Treasury absorption limit. As government borrowing rises, it reduces national saving, puts upward pressure on borrowing costs, and diverts capital away from private investment.
In a recent study of the United States, I found that higher debt is associated with slower growth in the private capital stock, and that slower capital accumulation in turn reduces economic growth. In other words, government borrowing can already impose a supply-side cost by displacing productive private investment long before Caballero’s safe-asset margin reaches zero. His Laffer curve therefore identifies an additional constraint on debt issuance, not the first point at which debt becomes economically costly.
No Easy Fix
To remedy the problem of a shrinking safe debt margin, Caballero hints at loosening bank leverage and capital rules. This is a bad idea. The leverage ratio exists specifically to cap how much banks can hold using borrowed funds. Exempting government debt would let banks hold far more government debt without raising new capital—effectively a form of quantitative easing, just run through private bank balance sheets instead of the Federal Reserves.
Capital rules exist to keep banks solvent, not to subsidize the fiscal profligacy of Congress. Loosening these rules also concentrates interest rate risk onto banks at, as we saw in 2023, can be sunk by safe government bonds when rates move the wrong way.
Another problem with Caballero’s easy fix is that it overlooks the incentives problem. Politicians face a diffuse, delayed cost (a future banking shock) against a concentrated, immediate benefit (cheaper borrowing today). This is the classic asymmetry behind fiscal illusion. Importantly, the absorption premium is a price doing real work, signaling genuine scarcity. Suppressing that premium by regulatory fiat doesn’t remove that scarcity, it hides it and invites the “one exemption invites the next” ratchet regulators themselves have warned about.
There is No Free Lunch
The debate over government debt has mostly been “can we ever pay it back?” This paper says that’s not the only question worth asking, and it also quietly forecloses the easiest-looking answer. If loosening the rules that constrain banks just relocates the cost onto the banking system instead of removing it, then there’s no free lever left to pull.
Even a government that never misses a payment can, at some point, be asking the financial system to absorb more debt than it can comfortably supply, and when that happens, more borrowing doesn’t stimulate the economy, it drags on it. We just crossed $40 trillion in debt. On Caballero’s numbers, we only have a few more years to find out whether we’re already asking for more than the shelf can hold.
The administration is going on in on AI producing productivity increases to service the debt and pay down the principal. Well that, and of course inflation. I feel confident that it should work out like that more or less, but you never know what Black Swan could come along or a new combination of unpredictable variables to throw a spanner in the works. Sure it's gonna be fun watching though....