Is China Already Number One? New GDP Estimates
When the presidents of China and the United States meet next week in Washington, neither will likely be aware that, measured in terms of purchasing power, it is Hu Jintao not Barack Obama who represents the world's largest economy. Some time in 2010, the Chinese economy overtook that of the United States. My calculations of GDP for 2010—which of course are subject to the uncertainty associated with all such exercises—are based on new estimates of GDP that will soon be published by the Penn World Tables (PWT) under the guidance of Professor Alan Heston at the University of Pennsylvania.
Cross-country comparisons of economic size and standards of living of the average citizen rely on two approaches. The first uses market exchange rates to convert the economic value of goods and services produced around the world into a common currency, usually the dollar. According to the IMF's latest estimates for 2010, the value of total US GDP was $14.6 trillion while that of China was $5.7 trillion.
But it has long been recognized by many economists that using the market exchange rate to value goods and services is misleading about the real costs of living in two countries. Such goods and services as medical services, retail and constructions services, and haircuts—which are not traded across borders—are cheaper in poorer countries because labor is abundant. Using the market exchange rate to compare living standards across countries understates the benefits that citizens in poor countries enjoy from having access to these goods and services.
Purchasing power parity (PPP) estimates—which take account of these differing costs—are an alternative and, in some respects, more revealing way of computing and comparing standards of living and economic size across countries. These estimates have been published periodically in the Penn World Tables since 1970.1 My calculations (explained in greater detail below) based on the most recent version, which is due in early February, show that the size of the Chinese economy in 2010 was about $14.8 trillion dollars—surpassing that of the United States.
How do I obtain this number?
The IMF has produced its own PPP-based estimates for 2010 (published in the World Economic Outlook in October 2010). But these are problematic in two important respects for China: The GDP number for 2005, which is the starting point for all the PPP-based calculations, is understated, and the price increases between 2005 and 2010 are overstated, which further reduces the GDP number for 2010. Consider each.
When the World Bank published its PPP-adjusted estimates for GDP per capita based on disaggregated data collected by the International Comparison of Prices (ICP) project exercise in 2005, a number of commentators expressed doubts about them, especially because the estimates for China ($4,091 per capita) and India ($2,126) were revised downward by 40 percent relative to the pre-ICP estimates. The IMF uses this lower number for 2005.
Surjit Bhalla (2008) argued that the number for China was unrealistic. Suppose, he said, the revised number was extrapolated backward, using China's growth rate between 1952 and 2004 of 5.52 percent. It would imply a level of GDP per capita in 1952 of $153 (in 1985 prices) that was well below the threshold value of $250 that Pritchett (1997) has argued is the minimum required for subsistence. In Bhalla's evocative description, the revisions meant that there were few living Chinese in 1952.
The validity of doing this backward extrapolation is questionable because, as I pointed out in a paper with Simon Johnson and colleagues (2009), PPP growth rates and market exchange rate-based growth rates are not identical and cannot be used interchangeably.
Nevertheless, Bhalla's point is correct, albeit for slightly different reasons as argued by Deaton and Heston (2010). They suggest that China's PPP-GDP was underestimated (by the ICP project and by the World Bank) because of the urban bias of the price sampling in ICP 2005. Data on prices were collected for 11 cities and their surroundings but no rural prices—which are typically substantially below urban prices—were collected (or rather allowed to be collected by the Chinese authorities).
The latest version of the Penn World Tables (version 7 to be released in early February 2011) have corrected these biases, which result in an upward revision for China's PPP-based GDP by about 27 percent and for India by about 13 percent for the year 2005. I use the new PWT corrections as the starting point for computing new estimates for PPP-based GDP and GDP per capita.
A second correction relates to developments between 2005 and 2010. For this period, if the IMF data are taken at face value, they suggest an increase in the real cost of living in China relative to that in the United States (which is equivalent to a real appreciation of the Chinese currency) of about 35 percent.2 This seems implausible because three alternative ways of assessing currency changes point to a much smaller appreciation.
First, most real exchange rate indices computed for China (for example those of JPMorgan and the Bank for International Settlements) point to an appreciation during this period of between 12 and 20 percent. Analysis of productivity differentials between China and its trading partners, and between tradable and nontradable sectors within China, by the IMF in its 2010 Article IV consultation (p. 19) would also imply an appreciation of the yuan of no more than 10 to 15 percent.3 Third, one could ask what currency appreciation would be implied for China if it behaved like the average country: For this country, estimates suggest that currency appreciation responds to the difference between its own growth rate of per capita GDP and that of the United States (in the jargon, this is called the dynamic Balassa-Samuelson effect). This procedure also yields estimates of about 10-15 percent for Chinese currency appreciation.4
If a currency appreciates, the movement is akin to an increase in the average cost of living. So taking 15 percent as the best estimate of the currency appreciation, rather than the 35 percent estimated by the IMF, requires adjusting China's 2010 GDP upward by 20 percent (because the increase in the cost of living has been overstated by 20 percent). To be conservative, I have not adjusted GDP up by the entire 20 percent.
These two adjustments increase China's GDP from the current estimate of $10.1 trillion to $14.8 trillion (an increase of 47 percent, of which 27 percent is due to the revision in the 2005 estimate, and the rest due to smaller-than-assumed increases in the cost of living between 2005 and 2010). This $14.8 trillion figure exceeds US GDP of $14.6 trillion. It must be emphasized, of course, that the difference is small enough to be within the margin of error.5
Applying the same adjustments to GDP per capita increases the estimate for China from $7,518 (the current estimate in the IMF's World Economic Outlook) to $11,047. The GDP per capita (the average standard of living) is now about 4.3 times greater in the US than in China compared with a multiple of 6.3 without my corrections (and compared with a multiple of 11 if GDP is computed using market exchange rates).
One interesting question is why China did not allow more representative prices to be collected as part of the ICP project in 2005. Professor T. N. Srinivasan of Yale University has long argued that China likes to exaggerate its growth rate (to showcase its strength and dynamism) and simultaneously understate its level of GDP (being seen to be poorer may have advantages internationally, such as not being expected to contribute financially to global institutions or global public goods). Overstating prices has the effect of understating GDP, and thus helps achieve this objective.
Perhaps a more important explanation of China's behavior has to do with exchange rate politics. Had all prices been collected, China's average price level (cost of living) would have been substantially lower. And this would have resulted in estimates of undervaluation of the Chinese currency of close to 40 percent against the dollar (see Subramanian 2010 for the connection between China's price level and the implications for estimating whether currencies are under or overvalued). China's trading partners would have had additional technical ammunition to deploy against its highly sensitive but demonstrably beggar-thy-neighbor exchange rate policy.
See follow up post GDP Estimates: An Update and a Question for the IMF.
Bhalla, Surjit. 2008. World Bank—Most Asians Dead in 1950. Business Standard (August 23).
Deaton, Angus, and Alan Heston. 2010. Understanding PPPs and PPP-based national Accounts. American Economic Journal: Macroeconomics 2, no. 4: 1–35.
Johnson, Simon, William Larson, Chris Papageorgiou, and Arvind Subramanian. 2009. Is Newer Better? Penn World Table Revisions and their Impact on Growth Estimates. National Bureau of Economic Research Working Paper 15455. Cambridge, MA: National Bureau of Economic Research.
Pritchett, Lant. 1997. Divergence, Big Time. Journal of Economic Perspectives 11(3): 3–17.
Srinivasan, T. N. 1994. Database for Development Analysis: An Overview. Journal of Development Economics 44(1): 3–27.
Subramanian, Arvind. 2010. New PPP-Based Estimates of Renminbi Undervaluation and Policy Implications. Peterson Institute for International Economics Policy Brief 10-8. Washington: Peterson Institute for International Economics.
1. The PPP-based estimates of the PWT have been an immeasurably valuable input in, and transformed, the study of economic development. The early authors of the PWT—Irving Kravis, Alan Heston, and Robert Summers (father of Larry Summers)—deserve recognition from the Nobel Committee for their unsexy but invaluable efforts in being the quiet bean counters who have provided the essential data for opening up a whole new field of inquiry into arguably the most pressing economic and moral issue of our time: promoting economic development.
2. The World Economic Outlook estimates the price level of China's GDP (also the inverse of the real exchange rate) at 42 in 2005 and nearly 57 in 2010, representing a 35 percent real increase (the price level is always relative to the US base of 100).
3. Recall the basic theory, which says that exchange rates between two countries should move in line with productivity differentials between them or between tradable and nontradable sectors within a country. Total factor productivity growth differentials were about 15 percent for eight years or roughly 10 percent for the five-year period, 2005–10 for which we need to make the adjustment. Similarly, productivity in the tradable sector increased relative to that in the nontradable sector by 26 percent or roughly 15 percent for a five-year period.
4. China's GDP per capita increased cumulatively by 66 percent and that of the United States by 5 percent. Applying this difference to a plausible Balassa-Samuelson coefficient of 0.2, yields an increase in the price level of China's GDP of about 12 percent rather than the 35 percent assumed by the IMF.
5. In 2011, with China's growth likely to substantially exceed that of the United States, the difference in their levels of GDP will likely move beyond error margins.