Tuesday, April 25, 2017

[From May, 2013] Reinhart and Rogoff Trip Over Data While Attacking Krugman

I am re-upping this post, originally at the CEPR blog to amplify a new paper by Michael Ash, Deepankar Basu, and Arindrajit Dube. See also Dube’s contemporaneous piece,“A Note on Debt, Growth, and Causality” for something more sophisticated than mine.

Yesterday, Carmen Reinhart—she of the infamous Excel error—wrote an open letter to Paul Krugman taking issue with his “spectacularly uncivil behavior.” That his “characterization of our work is selective and shallow.” In particular, Reinhart cites Krugman’s views on Italy. She writes:
However, [falling interest rates in “high-debt Italy”] is meant to re-enforce your strongly held view that high debt is not a problem (even for Italy) and that causality runs exclusively from slow growth to debt. You do not mention that in this miracle economy, GDP fell by more than 2 percent in 2012 and is expected to fall by a similar amount this year. Elsewhere you have stated that you are sure that Italy’s long-term secular growth/debt problems, which date back to the 1990s, are purely a case of slow growth causing high debt. This claim is highly debatable.
In fact, Reinhart recently cited Italy as an example of a “more recent public debt overhang episode.” She cites another paper to back up her claim that the evidence shows the direction of causality runs from high debt to slow growth. But even a cursory examination of the data undermines that case.

Figure 1 takes data from Reinhart’s paper in the Journal of Economic Perspectives and shows very clearly that Italy built up its debt after growth slowed significantly— not the other way around. In fact, when growth slowed back in 1974, Italy’s debt-to-GDP was only 41.3 percent. Italy did not reach 90 percent debt-to-GDP until 1988—some 14 years later.

Figure 1: Real GDP Index (Italy Since 1947) (log)
Source: Reinhart, Reinhart, and Rogoff and author’s calculations.
Note: Specified years indicate first year of high-debt episode (see Reinhart, Reinhart, and Rogoff)

Indeed, there is a clear association in Italy’s post-war data between high debt and slow growth, but it clearly tells a story very different than what Reinhart would have us believe.

From 1947-74, real economic growth in Italy averaged 5.8 percent per year. Over the period 1975-88 (when Italy’s debt grew from 41.3 to 90.9 percent of GDP) economic growth averaged only 2.7 percent per year—a fall of 3.2 percentage points. It is clear, based on Reinhart’s data, that high debt could not have caused this slowdown in Italy’s economic growth, even if Italy’s period of low debt is associated with much faster growth.

Nor is Italy the sole example. In all four such recent examples of advanced countries with episodes of high debt, the slowdown precedes the increase in debt.

Figure 2: Real GDP Indices Since 1947 (log)
Source: Reinhart, Reinhart, and Rogoff and author’s calculations.
Note: Specified years indicate first year of high-debt episode (see Reinhart, Reinhart, and Rogoff)

Though less obvious for Belgium, most of the jump in debt-to-GDP came in 1980 and was largely the result of a series break in the data. According to the data on Reinhart and Rogoff’s website, Belgium’s gross general government debt-to-GDP was 62.5 percent in 1970 and falling (debt-to-GDP stood at 57.8 percent in 1974— the year real GDP peaked). Nevertheless, from the peak in real GDP in 1948 to peak in 1974, economic growth in Belgium averaged 4.2 percent per year. When the economy bottomed out in 1975, debt was only 54.4 percent of GDP, and did not reach 90 percent until 1983. Yet from 1975-83, growth averaged only 2.2 percent per year.

For the other countries, it is even more obvious that the economies slowed well before reaching high levels of debt. Clearly, Reinhart should look carefully to her own data before lashing out at Krugman.

No comments:

Post a Comment