Iceland: CFR Versus Krugman (and Ryan Avent)by Arvind Subramanian | July 3rd, 2012 | 03:10 pm
What seems like an arcane squabble over relative growth rates between Iceland and Latvia has erupted into an argument between two heavyweight voices on economic policy—the Council on Foreign Relations (CFR) and Paul Krugman. Obscure though it may be, their disagreement is important. It is actually a proxy for a much bigger debate on whether external devaluation (followed by Iceland and supported by Krugman) or harsh internal austerity measures (pursued by some Baltic countries as well as much of Europe right now) is the better strategy for getting out of the slump afflicting many countries in the region. It is therefore important to understand and settle the numbers. In this post, I demonstrate that CFR was wrong in how it presented the numbers and that the medium-term economic performance of the Baltic countries relative to Iceland is neither as good as CFR implied nor as bad as those who sided with Krugman.
The CFR, in its recent post, asserted that Krugman’s claims on the superiority of the Icelandic performance over those of the Baltic Republics (Estonia, Latvia, and Lithuania) are not based on the facts. Measured from the recent cyclical peak, Iceland’s performance looks better, CFR concedes. But measured from the recent trough Iceland is the laggard. And most important, measured over a longer period (QI 2000), Iceland significantly underperforms that of the Baltic countries.
This post does not address whether peaks or troughs are the better starting points for judging cyclical performance. Rather it assesses medium-run performance. Here, CFR makes a critical and a very common mistake: It fails to take account of a simple principle of growth theory, namely convergence. The theory holds that, on average, poorer countries should and will grow faster than richer countries as they catch up to the economic frontier. So, whether Iceland did better than the Baltics cannot be assessed by looking at simple growth rates but at growth rates adjusted for the fact that Iceland was so much richer than the Baltic countries to begin with.
In a separate blog from the Economist, Ryan Avent makes the same point. But neither he nor Krugman (in a subsequent post) helps us quantify matters. The table below provides this quantification based on the famous Solow growth framework. This framework suggests that for every log point difference between two countries’ initial level of per capita GDP, the subsequent medium-run growth rate (of per capita GDP, not total GDP, a distinction that nobody made) for the poorer country should be greater than that of the richer country by the convergence coefficient. The empirical literature on growth suggests that a value of 2 percent is broadly reasonable for this coefficient.
Take the following example. Iceland’s 2000 level of per capita GDP was $30,693, while that of Estonia was $4136. This is a 2 log point difference, which growth theory suggests should entail Estonia growing faster than Iceland by 4 percentage points. Since Iceland grew by an average of 1 percent per capita, Estonia should have grown by close to 5 percent.
The first column in the table presents the actual annual average growth of per capita GDP between 2000 and 2012 from the IMF’s World Economic Outlook database. The second column indicates what growth should have been—taking account of convergence. In this column, initial level of per capita GDP is based on market exchange rates. In the third column, the initial level of per capita GDP is based on purchasing power parity (PPP) exchange rates. In principle, it is preferable to do the convergence calculation based on PPP exchange rates, but PPP-based calculations have margins of error around them. Hence the need to look at both measures.
|Growth of per capita GDP, 2000–2012 (percent)|
|Actual||What growth should have been (convergence based on market exchange rates)||What growth should have been (convergence based on PPP exchange rates)|
Note: Calculations explained in text.
Source: Author’s calculations based on IMF data.
The bottom line from the table is this: On a PPP-based calculation, the Baltic republics have done better than what they should have, given that they started off poorer than Iceland (the numbers in column 1 exceed those in column 3 for all the three Baltic countries). On a market-exchange rate based calculation, however, they have done worse (growth numbers in column 1 are smaller than in column 2 for all three Baltic countries).
My conclusions are:
- CFR was wrong in presenting the numbers for medium-term growth the way they did.
- Ryan Avent and Paul Krugman were right in drawing attention to CFR’s mistake.
- However, Ryan Avent (and Krugman) did not emphasize that even from a medium-term perspective, the relevant comparison should relate to per capita GDP growth rates—not overall GDP growth rates.
- If one takes account of the convergence effect, the performance of the Baltic countries was neither significantly better than Iceland’s (as CFR implied) nor significantly worse (as Ryan Avent implied). Paul Krugman, unlike Ryan Avent, did not take a view on medium-term performance.