Today, we had a presentation from Prasada Rao of University of Queensland on the theory of constructing international comparisons of income. He also covered some more practical aspects concerning the recent new international price comparison benchmark - ICP 2005. As everyone knows, the prices of identical goods vary across countries as exemplified by the Big Mac Index. But actually constructing more serious indices using baskets of many goods is very hard. The most recent estimates for China reduced estimated GDP per capita in China by around 40% compared to previous projections based on earlier benchmark studies. The reason was that the benchmark study conducted price comparisons only in the regions of 11 major cities, which has been argued to result in an artificially high price level on the basis that the price of many goods may be cheaper in smaller cities and the countryside. Higher estimated prices mean that a given US Dollar value of GDP is actually really worth less than if prices had been estimated to be lower in China. An alternative explanation put forward by Peter Warr was that projections of the inflation in the prices of non-traded goods in China that had been used before the new study were wrong. It's likely that both arguments are correct to some degree.
So what data should applied researchers use at the moment? The latest World Development Indicators is based on the 2005 benchmark survey and with the exception of the issues with the Chinese data is probably superior to previous estimates. But the data (on PPP) only go back to 1980 and they just use the economic growth rate in local currency units to project back from the 2005 benchmark. The Penn World Table Version 6.3 does not yet use the 2005 survey results. They promise to bring out Version 7.0 incorporating those results by the end of 2009. PWT data go as far back as 1950 in some cases. Given this I'm going to be using PWT data where possible in my current work.
No comments:
Post a Comment