Untold Human Suffering Caused By Excel Spreadsheet Error

The most influential study claiming to show that high debt levels correlate with slow growth – which people have foolishly taken to mean “high debt levels cause slow growth” – can’t be replicated unless you also replicate their Excel spreadsheet error:

Mike Konczal has an important post on 3 major errors in the Reinhardt-Rogoff study. In a decent world, this would change everything about our approach to the downturn, but it won’t because in reality people primarily like the idea of austerity because of the moralistic ring to it. We misbehaved and now we have to be punished. It’s completely insane and is pointlessly ruining millions of people’s lives:

Coding Error. As Herndon-Ash-Pollin puts it: “A coding error in the RR working spreadsheet entirely excludes five countries, Australia, Austria, Belgium, Canada, and Denmark, from the analysis. [Reinhart-Rogoff] averaged cells in lines 30 to 44 instead of lines 30 to 49…This spreadsheet error…is responsible for a -0.3 percentage-point error in RR’s published average real GDP growth in the highest public debt/GDP category.” Belgium, in particular, has 26 years with debt-to-GDP above 90 percent, with an average growth rate of 2.6 percent (though this is only counted as one total point due to the weighting above).

Being a bit of a doubting Thomas on this coding error, I wouldn’t believe unless I touched the digital Excel wound myself. One of the authors was able to show me that, and here it is. You can see the Excel blue-box for formulas missing some data:

 

 

This error is needed to get the results they published, and it would go a long way to explaining why it has been impossible for others to replicate these results. If this error turns out to be an actual mistake Reinhart-Rogoff made, well, all I can hope is that future historians note that one of the core empirical points providing the intellectual foundation for the global move to austerity in the early 2010s was based on someone accidentally not updating a row formula in Excel.

Comments

  1. GDub says:

    I thought this was a fair treatment by the Economist (not a fan of austerity). http://www.economist.com/news/finance-and-economics/21576362-seminal-analysis-relationship-between-debt-and-growth-comes-under

    The graph is of interest. There is a correlation (using either set of data) between debt and growth. The question is over causation, and as many have pointed out, slow growth could cause higher debt of course–and over whether there’s a magical line where things get suddenly worse.

    A few errors in your post. The Greek (and other nations’) crisis began in 1999, so its a lot of nonsense (or at least hyperbole) to say that “untold suffering” was caused by an excel spreadsheet published in 2010. Many of Greece’s problems aren’t caused by public debt, but overall by a society that consumes a lot more than it produces without US-like borrowing ability.

    A great deal of suffering has indeed been caused by unintended effects of changing the degree of “openness” of small economies through a currency union without unity in fiscal policy. The Euro makes the policy challenge a lot harder by removing many of the options that a country like Greece would have used 20 years ago.

    • Jon Geeting says:

      I agree with all of that. The Excel error isn’t the most damning part of this. The most damning part is that they get the causal argument backwards, as you say. It’s true some countries like Greece don’t have the option of borrowing in their own currency, and so they don’t have the same demand-side policy tools that the United States does. The countries that do have these tools should use them.

Trackbacks

  1. [...] School budgets fare well in NJ elections PennPatriot: Rutgers president to answer budget questions Jon Geeting: Untold Human Suffering Caused By Excel Spreadsheet Error Jon Geeting: Bob Donchez Releases [...]

Speak Your Mind

*

* Copy this password:

* Type or paste password here:

87,557 Spam Comments Blocked so far by Spam Free Wordpress