A critique of Reinhart and Rogoff

Rortybomb summarizes it here, Matt Yglesias here, and the original paper is here (pdf), by Thomas Herndon, Michael Ash, and Robert Pollin.  I will read the paper soon.

Comments

What Rortybomb fails to see * is that the post WWII years are an outlier, and deserve to be excluded, exactly like Rogoff et al did (apparently by accident). After WWII all economies were of high-debt but the reconstruction from devastation is what spurred growth, not the high debt itself. In short, to have a phoenix rise from the ashes you need to burn the place down first. Where is WWIII when we need it now? We don't have it, ergo, high-debt-to-GDP is bad, not good. A better but dubious historical example to support the 'high-debt-is-good' crowd would be pre-Industrial Revolution England, which had 200% Debt-to-GDP, but I would argue that the Industrial Revolution saved England. Where is the new Industrial Revolution today? Maybe graphene? :-) In short, if you believe in the Great Stagnation then Rogoff et al are correct.

* Rortybomb: "[UPDATE: People are responding to the excel error, and that is important to document. But from a data point of view, the exclusion of the Post-World War II is particularly troublesome, as that is driving the negative results. This needs to be explained, as does the weighting, which compresses the long periods of average growth and high debt.]"

Funny, the notion that the depression is based on austerity based on a fat-finger about the low growth rates of debt-deflations.

I believe the Rogoff et al argument is that high debt prevents growth. The post WWII years had high debt and high growth, which is precisely an example of a scenario that refutes their argument. What is the justification for excluding them? The very fact that there was a superb bounce-back from the devastation of the war shows that their thesis is faulty.

In other words, if their thesis were correct, then post-war growth would have been low, since it would have been forestalled by high debt. Would they have excluded the data from those years in that case?

People are just grasping at straws to salvage this fiasco. Yes, it all makes sense that the accidental mistakes they made which were all biased in favor of their controversial and notable result were actually appropriate! Of course they were, because ... that time it was different! Directed by M. Night Shyamalan.

Is it that? Or is the party of science grasping at strawmen? Are they pretending that they didn't do what they claimed they wanted because of a paper? Can't punch their way out of a single paper sack? Or are they claiming that the Republicans stopped them (despite not controlling any branch of government outright) because of a story they already believed?

Misdirection isn't going to salvage this either Andrew. The most widely cited pro-austerity empirical study was revealed today to be a joke.

If you're thinking about massive deficit-fueled stimulus to 'get the economy going', you probably don't want to spend too much time poking around in the 1940s.

True. Policy during the 1940s was not designed to fix the business cycle--it was designed to defeat Hitler! The incidental side effect of the huge deficit-fueled spending spree was that the economy recovered and then began to overheat. Hence the price controls. From the perspective of achieving the NAIRU, it was too much. But doesn't that just go to show that deficit spending in can achieve the NAIRU (and even beyond)?

I should have been more specific. I was thinking of the second half of the 1940s, when government spending was massively reduced and the debt was paid down in a matter of years, causing the Great Postwar Depression, which result was obvious to any right-thinking Keynesian.

Excluding post-WWII data for Canada, New Zealand, and Australia sounds reasonable. But then including US data for that period seems inconsistent, particularly since the US supports their theory and the others don't. "Selective exclusion" seems like a reasonable charge at this point without further justification from RR.

I understand your reasoning that the years following WWII should be excluded as outliers, however, the Reinhart and Rogoff study included some but excluded others from this time period, indicating a selected sample. This is just bad, I am shocked that it has taken two years before anyone looked at this, why was it not peer reviewed before being used to influence public policy around the world?

Dumb, dumb error. If Reinhat and Rogoff are in doubt, this demolishes the last pathetic argument against the idea that leveraging yourself to your eyeballs is a fun and riskless activity, doesn't it?

Exactly. And maybe Paul Krugman will stop thumping "This Time Is Different" all the time. I'm sick of it.

I'm confused. Are you saying that this makes you think the RR claims, state and implied, are MORE legitimate?

Nope. My view hasn't really changed one way or the other- viz. anyone who tells me debt equal to 90% of GDP isn't a major problem is living in fantasyland.

This reminds me of old lawyer's quip: when the facts are on your side, pound on the facts. When the law is on your side, pound on the law. When neither are on your side, pound on the table.

Take it easy on the table, Brian.

Heh. Didn't think I was coming across like that. Perhaps an opportunity for reflection.

I'ma think on it, but my gut says I got facts on my side.

Obviously the topic will still be under dispute regardless of this paper since that isn't the only work showing a negative impact of debt on GDP growth, as this Stanford prof pointed out when applying 3 different scenarios showing how they could lead to reduced US GDP growth:
http://www.hoover.org/publications/defining-ideas/article/133521
There is also an IMF paper and the CBO used its own estimates showing debt having a negative impact on growth, and there are other papers on crowding out. The Stanford prof's page doesn't take into account the feedback loop that higher debt levels in the case of the IMF paper leads to slower growth which leads to less tax revenue and more debt. This page:
http://www.politicsdebunked.com/article-list/budget-lottery
updates a GAO forecast long term US debt forecast taking into account that feedback loop, and includes an interactive graph of future US finances and debt where you can alter some assumptions.

The Excel error appears to have made the growth rate for 90%+ Debt-to-GDP to be 0.2%, instead of -0.1%. Well, sure, it's a difference of 0.3%. But a growth rate of 0.2% is STILL AWFUL.

I don't see how the Excel error undermines the Rogoff-Reinhart conclusion at all.

(I take no position on the other two errors cited by Rortybomb.)

The Excel error is fun for people to pick on, but really the main discrepencies are driven by two presumably intentional decisions by RR: exclusions of the post-war years for Australia, Canada and New Zealand (but not the US), and averages based on country instead of country-year. Those two decisions combined together drive the majority of the difference in the estimates.

Good observation TF79. And wow, I got my 15 minutes of internet fame, a mention by TC to his millions of readers! If they Google my name they'll see I mostly troll under this name (but not here, oh no).

Another observation about the R&R paper: how significant is the error anyway? When a -0.1 percent growth rate turns into 0.2 percent growth it seems like noise to me.

It is a bit more than noise, as Dean Baker explains -

'The most important of these errors was excluding four years of growth data from New Zealand in which it was above the 90% debt-to-GDP threshold. When these four years are added in, the average growth rate in New Zealand for its high debt years was 2.6%, compared to the -7.6% that R&R had entered in their calculation.

Since R&R country-weight their data (each country's growth rate has the same weight), and there are only seven countries that cross into the high-debt region, correcting this one mistake alone adds 1.5 percentage points to the average growth rate for the high-debt countries. This eliminates most of the falloff in growth that R&R find from high debt levels. (HAP find several other important errors in the R&R paper, but the missing New Zealand years are the biggest part of the story.)' http://www.guardian.co.uk/commentisfree/2013/apr/16/unemployment-reinhart-rogoff-arithmetic-cause

Comments for this post are closed