Thursday, August 12, 2010

Reinhart and Rogoff: There’s No There There

Here’s the core problem with Reinhart and Rogoff’s claim that public debt levels above 90% of GDP cause reduced growth: it’s all correlation and no mechanism. It epitomizes the worst aspects of empirical economics, searching tirelessly for statistical regularities, but not the mechanisms that might underlie them. Because economic contexts are highly diverse, often singular, it’s the processes at work, not generalizations about outcomes, that economics has the power to elucidate.

Sorry to be so abstract.

The R&R dataset, as the authors proudly explain, encompasses 44 countries over two centuries. We’ve got Finland, Spain, Japan and the US, Thailand, Mexico and Colombia. We’ve got the aftermath of the American Revolution against England and WWII, banking crises under the gold standard, the third world debt crisis of 1982. It’s all there in one hopper, ready to be crunched. I would convert to Rosicrucianism before I would embrace the belief that a single statistical relationship captures all these places and times.

Paul Krugman has highlighted two cases in particular, the US demobilization following WWII and the Japanese lost decades (still lost). Yes, he says, there is a correlation between public debt and slow growth, but in the US episode it’s spurious (war gave us the debt, demobilization the slow growth), and in Japan the causation runs from slow growth to high debt.

Just scanning the R&R list, I see lots of countries that have battled external deficits, a condition that weakens growth and puts governments in the position of running deficits in order to delay adjustment. And what about price shocks that cripple countries with a narrow export base or particular import dependencies? The R&R list is thick with these cases too. Given these interrelationships, it is revealing that, under “Debt and growth causality”, R&R consider only “Growth-to-debt” and “Debt-to-growth”, without the vast third category of “joint causation by other factors”.

Which gets us back to mechanisms. What are the forces, economic, political or otherwise, that cause runups of public debt? Under what circumstances does debt feed back to these other factors? What mechanisms govern the expansion and contraction of fiscal space? These are the kinds of things we need to know.

R&R have it backwards: they are looking for broad generalizations that might be identified over large samples but have uncertain application to any particular case. A better kind of economics would be one that identified processes that, while they generate diverse outcomes with no discernible central tendency over large samples, can be applied precisely to individual cases.

Like, for instance, ours.


STF said...
This comment has been removed by the author.
STF said...

Good post. You might also be interested in the paper by Randy Wray and Yeva Nersisyan, which goes into significant detail to debunk R/R. Here's the link: (forgot the link last time, which is why I deleted that post)

Neil Cameron (One Salient Oversight) said...

I disagree with this. The reason is that no economic crisis has ever pushed debt levels in a country to beyond 90% of GDP without that country first allowing debt to accrue irresponsibly during economic expansions.

Take the US. About one-third of current public debt is due to the financial crisis while the other two-thirds is due to Reagan and Bush 2. Had the US been fiscally responsible since 1981 and had zero public debt going into this crisis, debt levels would only be about 20% of GDP and no one would be concerned 90% debt levels.

Similarly Japan had public debt of around 48% of GDP in 1991 at the beginning of the "lost decade".

The Arthurian said...

But, Salient, we do not allow debt to accrue irresponsibly. It is our policy to remove money from circulation while encouraging the use of credit; and it is our policy to accumulate debt and "manage" it.

The unthinkable level of debt in our economy today is not the result of accident or irresponsibility. It is by design of policy. The debt itself, I think, is an unintended consequence of our efforts to fight inflation while encouraging growth.

Anyway, public debt is the least of our debt problems.

The Arthurian

TheTrucker said...

The proper balanced budget amendment to the Constitution would simply say that the government cannot borrow. So if government overspends then government will have to print the money. As this will cause inflation, then it becomes clear that government must tax appropriately to remove most of its spending from the economy. That is all.

Steve Roth said...

I agree that the paper is useless, but I disagree on the reasons.

I pointed out two problems with the paper some time ago:

1. Sample size. Their main exhibit shows the U.S.:

But as they point out in a footnote, they have only *five* sample points >90% (I find six), samples that represent a profoundly anomalous economic period:

U.S. Federal Debt as a Percentage of GDP
1944 91.45
1945 116.00
1946 121.25
1947 105.81
1948 93.75
1949 94.60

Krugman, looking at the full data set, finds equally problematic sampling shortage worldwide.

2. They compare debt in a given year to inflation in the same year, making no attempt to suss out long-term effects (which are what matter).

3. They don't make the most widely-accepted of necessary corrections–”convergence” or the “catch-up effect”–the tendency of less-prosperous economies to catch up with their cohorts due to transfers of technology, expertise, trade, capital, etc.

Correlational analysis is inherently problematic, but expecting every correlational analysis to provide a truly convincing causation analysis is unrealistic.