New IMF Paper Shows Yet Again that Reinhart and Rogoff Results Are Erroneous
You all remember the Reinhart and Rogoff claim that debt ratios above 90% lead to slow economic growth. As Yeva Nersisyan and I showed in 2010, their results are crap. (see here http://www.levyinstitute.org/publications/?docid=1273)
I’m pretty sure that the Reinhart and Rogoff “study” is the worst empirical research ever undertaken. They simply lumped together data of questionable verity taken across 800 years and 66 countries to get ratios of debt to GDP and growth rates. They summed across debt types, taking almost no notice of the exchange rate regime or the denomination of the debt. In other words, to R&R it makes no difference if a country pegs to gold or dollarizes, or if it issues debt in foreign currency. Further, at least in their public presentations, they habitually assumed that correlation is causation—ignoring the possibility that slow growth raises debt ratios. Their work was ideologically-driven: they wanted to stoke the deficit hysteria used as a justification for austerity.
As anyone familiar with MMT knows, it matters whether a country pegs its exchange rate to gold or a foreign currency. Unless you can ensure a positive inflow of the thing you peg to, you must adopt austere policy to keep domestic wages and prices low. And if you issue debt denominated in a foreign currency (or even in a domestic currency pegged to a foreign currency), then as your commitments rise you normally must squash domestic demand to increase net exports. Hence, for such countries it is probable that there will indeed be a correlation between rising debt ratios and slowing growth rates.
But for a country that adopts its own currency without a peg, and that avoids issuing debt in foreign currency, there is no necessary relation between debt ratios and growth rates. Now, to be sure, rising private debt ratios can increase fragility—for all the Minsky reasons. This can be associated with financial crises, which can have negative impacts on economic growth. However, sovereign debt actually improves domestic finances—makes the financial system less fragile. Adding income and safe financial assets to the domestic economy, sovereign deficits and debts can actually promote growth. By the same token, slow growth can increase debt ratios (as Japan’s experience over the past 20 years shows). Still, the sovereign deficit and debt ratios by themselves tell us nothing about sustainability. We can have those ratios rising for “good reasons” (to promote growth and development) or for “bad reasons” (because slow growth has kicked in the automatic stabilizers). For a sovereign currency issuer, what really matters is unemployment and living standard, not debt ratios.
A new IMF paper, “Debt and Growth: Is There a Magic Threshold?” by Andrea Pescatori, Damiano Sandri, and John Simon (here: http://www.imf.org/external/pubs/ft/wp/2014/wp1434.pdf) does a pretty good job of laying out the issues without the ideological bias of R&R. The authors use a data set that is less questionable, focusing on IMF member nations with data back to 1875. To take account of the possibility of reverse causation (slow growth leads to higher debt ratios), they look at longer periods of correlation. In other words, they see if high debt ratios (say, above 90%) remain associated with slow growth for years into the future. In addition, they distinguish between trajectories (does a country with a high debt ratio have a rising or falling debt ratio) to see if that makes a difference for the correlation. They also do some adjustments for “outliers” that affect averages (note that the R&R results depended strongly on outliers as well as math errors they made in their calculations).
What they find is that there is no “magic threshold” for the public debt ratio beyond which growth suffers. So far as their study goes, I think what they’ve done is a model for honest empirical work. Here is a quick summary of the main findings:
1. “Consistent with R&R (2010), we observe that GDP growth is particularly low in the year after the debt-to-GDP ratio reaches levels above 90 percent. Indeed, this chart shows that GDP growth averages around 2 percent in countries with debt below 90 percent, and tumbles to about -2 percent in countries whose debt ratio increases above that level. At the same time, the inter-quartile range across all episodes reveals that the growth performance for countries with debt rising above 90 percent is quite diverse. It would be unwise, however, to look for a causal relation between debt and growth from Figure 1 because of the possibility of reverse causation mentioned above. While it is possible that when the debt-to-GDP ratio exceeds 90 percent countries enter a state of distress that leads to a substantial reduction in growth, it is equally possible that increases in public debt above 90 percent are driven by an omitted variable that reduces GDP and tax revenues that, in turn, leads to higher debt.” In other words, the simple correlation does not show causation. We need to be careful in interpreting correlations.
2. Those correlations are substantially driven by outliers: “For example, the debt-to-GDP ratio in Japan increases from 133 percent in 1943 to 204 percent in 1944, and the subsequent growth rate in 1945 was -50%. This observation alone leads to a considerable reduction in the average growth for debt thresholds above 135 percent of GDP.” Gee, lesson learned: don’t start a World War unless you can win it; growth suffers if you are on the losing side.
3. The correlations are also driven by reverse causation: rotten economic growth raises debt ratios. Once you extend the time horizon in an attempt to correct for reverse causation, the correlation essentially disappears: “Extending the horizon of analysis allows us to mitigate the bias in the analysis induced by reverse causality and potential omitted variables issues–it also attenuates the effects of outliers such as the growth observation for Japan in 1945. For example, automatic stabilizers mean that low growth will tend to have an effect on the primary balance, and, thus, on debt over a short-term horizon. Similarly, and even more mechanically, a recession will raise the debt to GDP ratio because the denominator decreases. If high debt (that is, debt above some threshold) operates as a drag on growth over anything but the short-run, however, we would expect to observe weak growth not only in the year after the debt ratio exceeds the threshold, but also during the subsequent years… we show the growth performance of the same episodes over longer horizons…the growth performance improves considerably even at a 5-year horizon. The improvement is particularly noticeable for horizons of 10 and 15 years. Importantly, while higher debt is still associated with milder growth, there is no longer any clear debt-to-GDP threshold above which growth deteriorates sharply.”
4. Even quite high debt ratios do not hurt growth, especially if the debt ratios are coming down (as they likely would do if growth is high): “In fact, even countries with debt ratios of 130–140 percent but on a declining path have experienced solid growth. This observation suggests that the high debt itself is not causing the low growth in these episodes but other factors, associated with increasing debt, are more strongly implicated.” Note that it is impossible to completely exclude reverse causation—no matter how hard you try. It is possible that the debt trajectory is determined by economic growth, rather than that debt trajectory impacts growth.
5. In a further test of robustness of their results, the authors claim: “What we find is that, generally speaking, the growth performance of countries with high debt is fairly close to that of their peers—differences are less than ½ percent per year except at the lowest debt levels.” Another robustness test is used to correct for reverse causation, such that “with the removal of the first 5 years the relation between growth and debt becomes even flatter. Furthermore, we observe that even the direction of debt is no longer a clear predictor of growth…” Again, more reason to doubt that debt trajectory is the cause.
6. In conclusion: “Our analysis of historical data has highlighted that there is no simple threshold for debt ratios above which medium-term growth prospects are severely undermined. On the contrary, the association between debt and growth at high levels of debt becomes rather weak when one focuses on any but the shortest-term relationship, especially when controlling for the average growth performance of country peers.” The more you try to control, the weaker the correlations become—and the more erroneous become the R&R claims.
And note that like R&R, the authors of this IMF paper did not distinguish between sovereign currency issuing nations that issue debt in their own floating currency, and those that peg and/or issue debt in foreign currency. By lumping together all IMF members and over a time period that includes gold standards, Bretton Woods, and today’s “free for all”, one would expect a priori that the empirical results would be inconclusive—in other words, nearly random. And that is precisely what the IMF paper shows.
No, Virginia, there is no magic threshold for government debt beyond which growth falls. It depends.
If you’ve got your own sovereign currency, and you do not peg, and you do not issue debt denominated in a foreign currency, then there is no reason to suppose that higher debt ratios cause lower economic growth. Yes, budget deficits can be too high—causing inflation. They can be too low—causing slumps. Debt ratios can be high for “good reasons” and they can be high for “bad reasons”. Focusing on government debt ratios, alone, tells you nothing about the health of the economy in such cases.
If you do not have your own currency, if you peg to foreign currency, and if you issue debt in a foreign currency, then all bets are off. You might well be up that creek without a paddle.
The next step for extending this fairly well-done IMF paper is to distinguish between these two scenarios.
I have a hunch that I know what such a study would find.
3 Responses to “New IMF Paper Shows Yet Again that Reinhart and Rogoff Results Are Erroneous”
Could there ever be a case where you would recommend that the electorate (in the U.S.) tell their government to cut back on sovereign debt? If so, in what situation? Or is there no situation where the electorate should ever do that?
Nerdguy: sure. beyond full employment where “true” inflation occurs. And if you think govt spending is taking too many resources away from the “private purpose” to be used in the “public purpose” (again, that is beyond full emp of resources, altho could also occur before full emp but in that case the best solution is to relieve bottlenecks).
[…] Author: L. Randall Wray · February 14th, 2014 · RGE EconoMonitor […]