Why the Reinhart-Rogoff paper was flawed right from the start
By Andrea Terzi
At this point, it seems that everybody with an internet connection knows about Herndon, Ash, and Pollin’s (HAP) rebuttal of Reinhart-Rogoff’s (RR) paper on public debt and growth. Minutes after the news started to go round the web though, I was asked what I thought by friends whose interests are quite removed from economics. They were shocked that Harvard professors could make such embarrassing mistakes in their excel file, and that austerity decisions could be based on such an imprudent error.
But I gave them a different perspective:
- Reinhart-Rogoff’s paper has more serious flaws than mere excel errors.
- Austerity hysteria will not go away after Herndon-Ash-Pollin’s review.
So this is my message to you in this post, particularly if you are a non-economist.
The RR paper became famous because the authors claimed that they had found the “magic number” – the number that tells us all exactly how high public debt can get before economic conditions begin to sour. The European Union had fixed a ceiling for the public debt/GDP ratio at 60% for countries joining the euro. But that was an arbitrary number that had not been economically justified; it was a purely political compromise. As a result, there was no economically-justified number… until RR found the true figure!
In their paper, they claimed that public debt does not hurt growth that much if it remains below 90%. But once debt goes beyond that 90% threshold, economic growth comes to a screeching halt.
Many policy makers have used this 90% number – notably Olli Rehn in Europe, who used it to claim that there is no alternative to austerity, since it is well known that public debt in a number of EU countries is above 90% of GDP.
HAP’s review of RR’s paper completely undid this particular claim. Once the excel file was corrected, there was no magic number!
And yet, as I wrote above, there are even more serious flaws in RR’s paper apart from miscalculations.
1. Empirical analysis: Does debt/GDP correlate with growth?
RR explained that one key contribution of theirs was the incorporation of new historical data from forty-four countries spanning about two centuries. This was fortunate because when doing statistical analysis one normally wants more observations. And so the authors were proud to inform us that they had used “3,700 annual observations covering a wide range of political systems, institutions, exchange rate and monetary arrangements, and historic circumstances”. This would seem to be a valuable feature then; one could explore the same phenomenon under different circumstances to help identify stable relationships that prove robust even under changing conditions.
But using this approach to study the relation between public debt and economic growth makes no sense whatsoever. Public debt management as well as the causes and consequences of public debt differ enormously, depending on institutional setups such as exchange rate arrangements, gold parity, limits to central bank operations, banking regulation. Any calculated average over such a broad time span for numerous countries is simply mixing apples and oranges, and is not significant at all. For this very reason, I’ve never been interested in the numbers RR’s empirical analysis generated. And now that HAP have found the error, I’m glad I did not waste my time.
2. Theory: Does higher debt/GDP really ‘cause’ lower growth?
In their rebuttal, RR stress that they were “very careful … to speak of ‘association’ and not ‘causality’.” This is important in any professional statistical work. However, their statement is not genuine. When describing how growth and debt relate, RR clearly suggest that public debt is a potential threat to growth and that public debt builds up for two reasons: a) war (and this is intentional) and b) in peace time, “unstable underlying political economy dynamics” (and this is also the result of political decisions). The framework they use to interpret their (hardly significant) numbers is one where the debt/GDP ratio is assumed to be clearly and uniquely interpreted as exogenous, i.e., determined by the political process.
They, thus, choose to ignore that the debt/GDP ratio is largely endogenous, i.e., determined by the growth of output and jobs. They do admit that debt has grown following the recession, but they describe this increase as being the result of deliberate fiscal decisions that, as RR suggest, may be inevitable and useful, but, once the worst is over, policy should act to lower that ratio to avoid even worst outcomes. All this means is that when they find that high debt countries have low growth, they do not even suspect that this may be the outcome of too small a public deficit.
Yes, I said too small a public deficit, because the deficit is the fuel of aggregate demand and the source of private savings. Remember, private assets are public sector liabilities. So, if the deficit gets smaller, private savings get smaller. And if private savings are too low for households’ and firms’ desires, then households and firms will attempt to save more. They will cut their spending and send the economy into a recession. And when this happens, the public deficit will automatically get bigger, preventing a free fall in the economy. When the deficit has gotten ‘bigger-enough’, the economy stops falling.
The problem with austerity – something justified because of RR’s view that public debt is dangerous – is that it forestalls the upturn and forces the economy into more recession.
In sum, yes, correcting an excel error hidden in a Harvard paper is a good thing. Nonetheless, even with ‘correct’ numbers, RR’s empirical analysis remains useless – for the reason discussed above under 1. And the analysis is wholly misleading for the reasons discussed above under 2.
Apart from their poor excel file, RR need to get their theory about ‘public debt’ right.
Dr. Terzi is a Professor of Economics and coordinator of the Mecpoc Project at Franklin College Switzerland. He has focused his research interest on macroeconomics, monetary theory, central banking operations and financial market behavior.
This post was originally published at Mecpoc, a forum for alternative views in economics.