A Couple of Thoughts About Reinhart & Rogoff

As quick background, R&R had a study that found that higher government debt levels correlated to lower, even negative, economic growth.  More recently, others have found computational errors that exaggerated this result, and have criticized their methodology, particularly their approach to weighting data from different countries and years.

A few thoughts:

  1. A major reasons the errors were found is that R&R actually made their data available for replication.  This is apparently rare - certainly it is rare in the climate world.  I am glad they are getting kudos for this and hope the academic world can find a way to incentivize / force more data sharing
  2. I would not have expected a direct relationship between country debt levels and economic growth.  What I would expect is that growth can still be good at higher debt levels, but the risk of hitting a tipping point starts to rise dangerously with debt levels.  Eventually levels get so high that an interest rate shock or liquidity shock is almost inevitable
  3. More than a relation between GDP growth and absolute debt levels, I would have expected a relationship between GDP growth and changes in debt level.  Absolute government debt levels may represent resources removed from the productive economy years and decades earlier.  Increases in government debt represent recent decreases in capital available for productive use.
  • Colin Suttie

    The major problem with R & R's analysis is that they treat currency users and currency issuers as identical - as Cullen Roche writes, this is an apples to oranges comparison.

  • I think the real issue is that there are different categories of debt. It isn't just the quantity of debt, but what the debt is used for - in particular whether or not there is a positive return of some sort. How those returns are measured may be different for public spending, but there still has to be a positive return. As an example, a road that lasts 20 years has a definite benefit. Even a national park or a campground has a positive social benefit. Some things, especially things that become entitlements, have may have benefit as long as the number of people on welfare is kept low. Big public works projects, usually those with that have some politician's name over the front door, have completely negative returns.

  • Sorry, R&R is not only sound, it is understated.

    Disprove the 1970s vs. the 1980s/1990s.

    In contests betwixt overregulation zealots (“Zealots” http://www.zerohedge.com/news/2013-04-19/guest-post-important-lessons-domestic-terrorism
    ) vs. free marketeers, the free marketeers win every single time, in economic history.

  • skhpcola

    Don't get in the way of neo-progtards such as Tyler Cowen when they are propagandizing for their dystopian New World Order/Open Borders. Anybody that lacks the intellectual acuity to recognize that higher debt levels have a deleterious effect on growth and the creation of wealth should be dismissed as unserious and incurious. The club of self-proclaimed elites is chock-full of morons that offer flaccid refutations of the historical record and flog each others' backs in congratulatory glee whenever they feel that they have sufficiently that historical record to camouflage their ideological failures and biases.

  • Now if someone would only do a study of the counter parties to the R&R study on a point to point basis it would be an interesting read. History has shown that when the party finally stopped it only did so when the counter parties who had been funding the party withdrew, usually in a spectacular fashion. The party always seems to end in a crash, not a whimper.

  • NB: R&R sent their data to Herndon; it was not posted online. Nature 495:p439 reports that 36% of scientists post their data. That number is MUCH lower for economists (and most of the black magic is in the econometrics). On the more interesting issue of debt v GDP, that depends on trends but less debt is certainly better (there's a linear, negative relationship). Oh, and @mesaeconoguy:disqus -- one country is NOT a dataset.

  • Brian