GDP Estimates: An Update and a Question for the IMFby Arvind Subramanian | January 19th, 2011 | 01:59 pm
My posting last week on China’s new purchasing power parity (PPP)-based estimates has elicited a number of questions. One theme was: What would my revisions imply for the GDP estimates of other countries?
I have looked into this. I don’t intend to provide estimates for other countries, which analysts can do based on the methodology I used for estimating China’s GDP. But I want to flag one point, which is really a question for the bean counters at the International Monetary Fund (IMF) and its World Economic Outlook (WEO) division.
My argument is that there are two potential problems in the IMF’s PPP-based GDP numbers. One relates to the 2005 estimate and the other relates to the assumed cost of living changes (relative to the United States) between 2005 and 2010. The IMF cannot change the 2005 estimate until the World Bank does so, and it is not clear whether or when that will happen. But the second potential problem is within the IMF’s power to review. In the case of China, my view was that the IMF’s estimate of the increase in costs of living was too high, and much higher than warranted by underlying productivity and growth changes.
Technical aside: The way to back out the cost of living changes (what is called the price level of GDP) is to divide the GDP in current dollars by the GDP in PPP dollars. Take the case of India: For 2005, the price level of GDP is 33 (810/2434) and for 2010 it is 36 (1430/4001), representing an increase in the cost of living of 7 percent. (The price level of GDP is always 100 for the United States: its PPP-based and market exchange rate–based GDPs are identical by construction because the US is the reference country for all PPP calculations).
It turns out that the cost of living changes in the IMF’s WEO database for other countries are also problematic. It is not that there is some systematic pattern here—rather the numbers seem arbitrary. For example, the numbers for India (7 percent) and Turkey (18 percent) seem plausible. But for other countries, the cost of living changes seem implausible: Korea (decline in cost of living of 12 percent); Russia (a 48 percent increase in the cost of living); Brazil (a 65 percent increase); and Indonesia (67 percent increase). Over five-year horizons, it is difficult to imagine such changes in the real cost of living (indeed, one of the virtues of using GDP estimates based on PPPs rather than market exchange rates is that they are less volatile).
I would urge the IMF to take a more careful look at its PPP-based numbers.