How much government debt is too much government debt? That question is pertinent in most economies today, but is especially pertinent in the US, where Congress is debating whether to raise the government debt ceiling, and if so on what terms. Unless economists can give sensible advice on the appropriate maximum level for public debt, much of the debate on budgetary policy is based on little more than political bias or, even worse, gut feeling dressed up as expert opinion.
Until last year, economists had rather little to say on this crucial matter. Then, along came Carmen Reinhart and Kenneth Rogoff with their painstaking study of public debt through the ages, This Time Is Different, from which sprung this highly influential paper in the American Economic Review a year ago. The paper examined the link between the level of public debt and real GDP growth in 44 countries over a time span of 200 years. Professor Reinhart has recently summarised the key conclusion as follows:
The main finding of our study is that the relationship between government debt and real GDP growth is weak for debt-to-GDP ratios below 90 percent of GDP. Above the threshold of 90 percent, however, median growth rates fall by 1 percent, and average growth falls considerably more.
The 90 per cent safety limit for gross government debt has rapidly taken on the status of a “stylised fact” which has influenced public policy in a number of countries. However, like most stylised facts, this one has wobbled as it has come under increased scrutiny.
The main problem with the Reinhart-Rogoff (henceforward R-R) work is that it fails to distinguish clearly the direction of causation in the relationship between high levels of debt and GDP growth. If GDP falls during a recession, there will be an automatic rise in the debt ratio, but this will not indicate that the recession was caused by excessive debt. Indeed, this is precisely what happened after 2008.
The evidence on the direction of causation is a bit mixed. An early analysis of the R-R paper by Irons and Bivens at the Economic Policy Institute found that the direction of causation mostly worked from growth to debt, and not the other way around. However, subsequent work at the IMF by Kumar and Woo showed that, for a large cross section of countries, higher initial levels of public debt have a damaging impact on the rate of growth of GDP for the following 5 years, with much of the effect coming through the crowding out of private investment and lower productivity growth.
This methodology rules out reverse causation. And the damaging impact of excess public debt is quite large – for every 10 per cent rise in the debt ratio, subsequent GDP growth falls by 0.2 per cent per annum. Furthermore, the IMF results appear to confirm the R-R result that the relationship worsens noticeably when the debt ratio exceeds 90 per cent, as the graph at the bottom shows.
The main R-R conclusion has therefore survived empirical examination so far. But it has also been been attacked on grounds of economic logic by researchers Nersisyan and Wray in a paper for the Levy Institute. They make a point which has still been insufficiently recognised in the debate about public debt, which is this: if debt is denominated in a domestic currency (like the dollar in the US), then it is not possible for a government to be forced into default. Because a sovereign government or its central bank can always create domestic currency to pay down debt, they need never default on public bonds when they become due for redemption. (Of course, it is very different for debt denominated in foreign currencies.)
The Levy researchers say that they can find no example in the R-R database of an advanced country defaulting on domestic currency debt, unless that country was encumbered by a fixed exchange rate regime or a link to gold. Because that is the case, they argue that the debt ratio should never be seen as presenting an absolute constraint on the behaviour of any government which has an independent, sovereign currency at its disposal. On this reasoning, the researchers derive the super-Keynesian result that governments should run budget deficits whenever the economy is working below its normal level of capacity, and increase their budget deficits whenever necessary to return to full employment.
What, then, is the ultimate constraint on government deficits in this view of the world? It is not public debt, but instead it is the rate of inflation. The Levy researchers conclude that governments should run budget deficits, and finance them via money creation if necessary to hold interest rates down, unless and until they see inflation rates exceeding targets. (Or perhaps until they see inflation expectations rising – see this earlier blog post.)
Since Japan has never seen inflation rising since they fell into their two wasted decades, this argument would have predicted that Japanese government bonds should not suffer from rising yields, even in the face of continuously rising public debt ratios, and so far that has indeed been the case. By extension, the Levy researchers would presumably argue that the US should not worry about a rise in the debt ratio above 90 per cent of GDP, unless inflation persistently exceeds a 2 per cent target.
My own doubts about these super-Keynesian arguments revolve, first, around the appropriate size for government in a free market society; and, second, around the possible impact on inflation expectations, and then on inflation itself, if governments were to pursue a systematic policy of financing budget deficits by money creation. After all, it is possible for economies to suffer devastating surges in inflation, even while unemployment is above normal levels.
That would clearly be the worst of all worlds. Unless the US falls back into a deflationary recession, in which case further fiscal action would become necessary, a 90 per cent long term ceiling for the gross debt ratio seems a sensible enough target to me. Since the US gross debt ratio will be around 100 per cent of GDP this year, there is already cause for concern.
