How much debt is too much? Or more specifically, how much debt can a country take on before it risks hitting a "tipping point": the point where you risk spiralling into a vicious cycle of higher interest rates putting pressure on your budget, which makes the debt riskier and causes interest rates to rise further.
David Greenlaw, James Hamilton, Peter Hooper and Frederick Mishkin have a new paper out on this question. Their answer is "80% of GDP". Neil Irwin has a very good summary, as well as some useful critique:
Considering that the paper was presented at a monetary policy conference, and that its authors include a former Federal Reserve governor (Mishkin), it seems not to grapple enough with the crucial difference between the United States and many of the countries that are among the nations included in the historical analysis the paper is based on. Almost all of the countries that have experienced major financial crises in the recent past have in some way lacked control over their currency. For Greece and Ireland, that is because they use the euro, whose value is determined by the European Central Bank, the value of which is better suited for Germany and France than for those troubled peripheral European economies. In the East Asian crisis of the late 1990s, much of the damage came about because the countries involved had borrowed money in dollars, so when the value of their domestic currencies fell they suddenly had more onerous debt burdens than they had expected.
So for the U.S. fiscal picture to be as dangerous as the paper suggests, one needs a story of how a crisis of confidence in U.S. debt leads to an outflow of capital and no offsetting effort by the Fed to ease policy, simultaneously strengthening growth and making interest rates lower. This paper doesn’t do that.
As Eric Rosengren, the president of the Federal Reserve Bank of Boston, put it in his response to the paper, the model that the authors use is “parsiminous.” Read: It’s awfully simple, even simplistic, omitting a numer of important variables that determine whether a country is at great risk of crisis. Rosengren mentions financial strength of the banking sector in a country and its political system.
“While this paper provides a good discussion of tipping points, the coefficients and simulations should be viewed as indicative of a more general point – that policymakers should consider how a variety of public policies, including fiscal policies, can influence tipping points,” said Rosengren.
And as Fed governor Jerome Powell put it in a second response, the tipping points that the authors found “are driven to a great extent by the experience of smaller eurozone nations that, of course, borrow in euros. The United States borrows in its own currency–the world’s primary reserve currency. That difference is crucial for investors.” He notes that the United Kingdom and Japan, both, like the U.S., large countries with their own currencies, show no detectable rate increases. “These countries present a serious problem for the authors’ case,” Powell said.
There is a real danger in leaning too heavily on data from small countries that don't have the same control over their currency that the US enjoys. Borrowing in a foreign currency (or ceding control over your monetary policy through mechanisms like currency zones or currency pegs) creates a new category of risks that places like the US don't face.
That said, there's also danger in treating "borrowing in your own currency" as completely exogenous to your model--as if it's a natural feature of the United States, like Yellowstone or Niagara Falls. The reason that so many countries borrow in other currencies is that investors don't trust them. They (correctly) suspect that these countries may resort to inflation to erode the real value of their debt burden, and so the bond buyers demand a hefty interest rate premium for debt denominated in local currency.
It's not as if this couldn't happen to us; it just hasn't, because we haven't engaged in as much "financial repression" (Carmen Reinhart's term) as the countries that have to resort to foriegn-currency loans. So some of the monetary weapons that we theoretically have at our disposal only remain in the arsenal, so to speak, because investors trust us not to use them.
So assume that there is some sort of tipping point for the US, even if it is not actually 80% of GDP. That raises a question that I haven't seen well-answered: how close should we be willing to get to that crucial point?
On the one hand, any threshold will have some give in it; it's not as if every single country that hits a debt-to-gdp ratio of 80% suddenly, almost mechanically, tips into the abyss. (Japan still borrows at attractive interest rate with a ratio that is somewhere between 50-150% higher than that, depending on how you calculate.)
On the other hand, getting too close to the threshold dramatically raises your risk and narrows your choices. Countries in deep contractions have grave difficulty balancing their budgets even when they're trying pretty hard. (Imagine, if you will, trying to cut the US budget in half overnight. No, stop cheering, conservatives: going to shut down border patrol and leave all those soldiers and tanks and thinks stranded in Iraq? Or does your mother suddenly stop getting her social security and medicare? Even if you think the budget should be cut in half, "overnight" is not the correct time frame.) So if they run too close to the edge, they may find themselves pushed over the threshhold when things go south--or be forced to do ham-fisted and destructive austerity to avoid going there. So you probably want to leave a pretty big buffer between yourself and the dangerous debt levels, just in case.
Now that Democrats are starting to argue that our goal should be to just stabilize the debt over the next decade at 75-80% of GDP, these question has some fairly timely policy relevance. If there is a tipping point, and we're close to it, we should probably be trying to back off the precipice. Slowly and cautiously, so as to avoid some sort of rock slide--but back away, none the less.