Thursday, September 26, 2013

Penn World Table 8.0

The new version of the Penn World Table - version 8.0 - has recently been made available and is now hosted at University of Groningen in the Netherlands. An NBER working paper by Feenstra et al. describes what is new in PWT 8.0.

The new edition of the dataset introduces several new measures of GDP and the working paper is mostly devoted to discussing them as well as the relationship between PPP exchange rates (relative to market exchange rates) and the level of income known as the Penn or Balassa-Samuelson Effect.

GDP is now given both in terms of the output side and the expenditure side. The difference between these is that real output side GDP (RGDP(O)) deflates expenditure on final goods (the standard macro-economic C+I+G - consumption, investment, and government expenditure), exports (X), and imports (M) using separate deflators:

The expenditure side real GDP (RGDP(E)) uses only the final output deflator to deflate the GDP. Feenstra et al. argue that the former expresses better the real production level in each country and the latter the standard of living in each country. Previous versions of the Penn World Table used the expenditure side measure only. The difference between the two measures is due to the terms of trade. Countries with relatively expensive exports and relatively cheap imports will have living standards (RGDP(E)) that are higher than their real productive capacity (RGDP(O)).

GDP is also given in "current" and "constant" prices. This terminology is confusing because usually current prices mean prices not adjusted for inflation and constant prices mean adjusted for inflation. Here constant prices mean the reference prices from a given benchmark year -  in the current version 2005 - and current prices mean using the reference prices from each year though these are adjusted for US inflation. These differ because the reference prices change over time. The constant price series RGDP are better for comparisons across time while the current price series CGDP can be used to compare countries at a single point in time.

Finally, there is also an RGDP(NA) series that uses the growth rates in each country's own national accounts to extrapolate GDP in that country in years other than the benchmark year. National accounts growth rates were used exclusively in previous versions of the Penn World Table. This series can differ substantially from the RGDP(E) series as is shown by this graph for India:

According to RGDP(E) living standards in India fell from 1975 to 1985 while according to India's own national accounts they rose. Which is right? Well, it depends what you want to measure. The change in RGDP(E) measures the change in relative living standards across countries while that in RGDP(NA) measures the change in real expenditure weighted according to the budget shares in the country in question. They differ because budget shares differ across countries. RGDP(E) will also grow faster than RGDP(NA) in a country experiencing an improvement in the terms of trade as, for example, Australia did in the years up to 2009 due to the mining boom.

PWT 8.0 also includes capital stock, human capital, and total factor productivity series. The former was included for some countries in some previous versions but not version 7. The latter are both new.

So, all this sounds more complicated than using The Economist's Big Mac Index or previous versions of the PWT. The User Guide gives a less technical guide on how to use the data.

No comments:

Post a Comment