Debt Overhangs: Past and Present
Debt levels above 90 percent are associated with growth that is 1.2 percent lower than in other periods.
Public indebtedness in advanced economies is historically high and rising after the recent financial crisis. In Debt Overhangs: Past and Present (NBER Working Paper No. 18015), co-authors Carmen Reinhart, Vincent Reinhart, and Kenneth Rogoff address some of the potential economic consequences of high debt loads. Specifically, they use cross-country historical data on public debt levels to examine the long-term growth consequences of prolonged periods of exceptionally high public debt.
The authors define "public debt overhang episodes" as periods when public debt-to-GDP ratios exceeded 90 percent for more than five years. That five-year minimum was chosen to exclude normal business-cycle-frequency slowdowns, although in fact there turn out to be only a handful of such short episodes in this two-century dataset. The 90 percent threshold is based on a growing body of research, including work by the authors, demonstrating that there may be significant nonlinearities in the relationship between debt and growth so that the effect of an increase in the debt-to-GDP ratio from 60 to 70 percent may differ from the effect of an increase from 90 to 100 percent. The three authors find 26 such episodes in 22 advanced economies since 1800. This does not include the unfolding post-crisis cases in Belgium, Iceland, Ireland, Portugal, and the United States, which do not yet meet the five-year minimum criterion, although it seems nearly certain they will ultimately do so. Ongoing debt overhangs in Greece, Italy, and Japan started before the financial crisis, have already cleared the five-year mark, and are included.
The authors find that on average, debt levels above 90 percent are associated with growth that is 1.2 percent lower than in other periods (2.3 percent versus 3.5 percent). Importantly, 20 of the 26 episodes lasted more than a decade, and the average duration of debt overhang episodes in the sample is 23 years. The authors explain that the consequences of this long duration are two-fold. First, it suggests that the association of debt and growth is not just a cyclical phenomenon, which contradicts the view that the correlation is caused mainly by debt buildups during recessions. Second, it implies that the cumulative shortfall in output from debt overhang is potentially large. By the end of the median episode, the level of output is nearly a quarter below that predicted by the trend in lower-debt periods.
The researchers also explore the association between high public debt and real interest rates. A little over half the episodes are associated with high real interest rates and indeed, many pre-World War II debt overhangs ended in restructurings. However, the evidence suggests that the growth-reducing effects of high public debt are not transmitted exclusively through high real interest rates. Indeed, in 11 of the 26 debt overhang cases, real interest rates were either lower or about the same as during the lower debt/GDP years, yet growth was impaired as in the other cases. Therefore, the growth effects are significant, even in the many episodes where debtor countries were able to secure continual access to capital markets at relatively low real interest rates. Why might growth be lower even absent an outright debt crisis? The authors note that even with pro-active fiscal adjustment and high credibility, countries with large debt burdens must ultimately choose some mix of higher distorting taxes, lower government spending, or greater financial repression, all of which reduce growth.
The authors caution that the causal relationship between debt and growth is still an open research question. They note, however, that most studies find relatively little correlation between debt levels and growth rates at debt-to-GDP levels significantly below 90 percent, which is difficult to explain if the link between debt and growth is solely due to slower growth causing high deficits.
--Claire BrunelThe Digest is not copyrighted and may be reproduced freely with appropriate attribution of source.