It seems like high public debt levels ought to represent a looming economic problem. Why, then, is it so difficult to demonstrate, conclusively, that they are? It could be due to the econometric challenges posed by any macroeconomic issue: sample sizes are small and the possibility of any number of statistical biases throwing things off is large. Or it could be that debt levels simply aren’t, in many cases, as bad as everyone seems to think.
A new paper illustrates the trouble economists have when they try to show that debt is scary. In a paper prepared the US Monetary Policy Forum, economists David Greenlaw, James Hamilton, Peter Hooper, and Frederic Mishkin conclude that “countries with debt above 80% of GDP and persistent current-account deficits are vulnerable to a rapid fiscal deterioration as a result of…tipping-point dynamics”. But their work is remarkably unpersuasive.
They begin by exploring an intuitive framework for debt tipping points. Markets, in deciding what interest rate to charge sovereigns wishing to borrow, weigh the probability that explicit or implicit default will eventually rob them of expected earnings on their loans. Other things equal, a higher debt load translates into higher probability of default. If the outlook for growth deteriorates, or if political changes make sustainable fiscal policy less achievable, markets may again update their outlook. And if the current-account deficit is larger, suggesting that more of the debt is owed to foreigners, then the temptation to default is greater and the odds of default rise.
The authors put together a model showing how a change in the above variables can lead markets to adjust their default expectations and raise the interest rate charged to the sovereign. But a higher rate worsens the fiscal outlook, leading markets to worry more and raise rates again. Under certain circumstances, expected sovereign borrowing can grow explosively, leading to a debt crisis.
They then turn to a statistical exercise, and report their findings:
The regression covers 20 countries for years t = 2000 − 2011 for a total of 240 observations. Interestingly, the coefficients on gross and net debt are both highly statistically significant. The regression suggests that if the country’s primary deficit increases by 1% of GDP (causing both gross and net debt to increase by one percentage point relative to GDP), the borrowing cost would increase by…4.5 basis points.
How to parse this, though? I was immediately concerned by the data sample: 20 advanced economies over 12 years. What’s particularly distressing is that just over half of the sample countries are members of the euro zone. In choosing to study advanced economies, the authors specifically note the problem of “original sin” in studies of emerging markets—that countries which borrow in foreign currencies are subject to different debt dynamics—only to then use a sample in which most of the chosen economies are unable to print their own money.
With that understood, the possibilities of spurious findings are clear. Prior to the crisis deficits and debt across the sample were generally falling, as were interest rates, helping to build a misleadingly tight statistical relationship between changes in borrowing and in sovereign yields. After the crisis, the relationship didn’t vanish entirely because several of the sampled economies did experience soaring debts and skyrocketing borrowing costs. Troublingly for this analysis, all of those economies were on the euro-area periphery. Even more troubling, yields reversed after the ECB effectively broke the link between banking system solvency and sovereign obligations, even though total debt stocks have continued to grow around the periphery.
If one looks at changes in debt-to-GDP ratios since the onset of the crisis in 2012, one observes that the largest increases occurred (in order) in Ireland, Greece, Spain, Japan, Portugal, Britain, and America. Yields over that time frame soared in Ireland, Greece, Spain, and Portugal, of course. But they fell, sharply, in Japan, Britain, and America. We can draw a lot of important macroeconomic lessons from the sample the authors consider, but I’m not sure that the universal danger of a debt stock above 80% of GDP is one of the top ones.
Does that mean that Japan, Britain, and America can relax? Not necessarily. Consider America. As the economy improves sovereign yields rise, reflecting the increased attractiveness of private investments relative to government debt. That will raise the governments interest-rate costs and partially offset the improvement in the deficit that will also accompany recovery. A lot of America’s outstanding debt is held abroad, and over the course of the crisis and recovery the average maturity of American debt has grown substantially. Both factors make a partial default via above-normal inflation more attractive (though of course the Fed would have to play ball to achieve such a thing). Meanwhile, the spectre of rising pension and health care costs remains. By 2020, the government will be dealing with the substantial fiscal demands of the Boomer generation. These factors might all push rates higher than they otherwise would be, thereby worsening the long-run debt outlook and pushing America toward a tipping point.
Given these threats, why aren’t American yields already soaring? On the one hand, markets might (quite reasonably) focus on shorter time horizons than the Congressional Budget Office. At the moment, the budget picture in 2030 looks daunting. But so much can happen before that time. Technological progress may buoy American growth and hold down health costs. America may welcome more immigrants than anticipated, spreading the burden of existing debts over more taxpayers. There are at least as many ways a bet on an American debt crisis could go wrong as there are ways it could go right. That’s not an argument for complete fiscal irresponsibility in Washington. It does suggest