My last post, has reminded me of a piece I have in the mailbox from Dave who is worried about Stephen Roach's use of data. His latest point only hammers it home even further, in economics there are only lies, damn lies, and statistics. Put more another way we should think very carefully about the limitations of the data we are working with, always ask ourselves why we are presenting something one way rather than another, and always try to bear in mind the weaknesses inherent in any particular data set. Which is why I think it is madness for the ECB to have placed such emphasis on the HICP numbers.
One further point on my original question: I was not using PPP data, just straightforward current and constant currency national income accounts from the Fund's International Financial Statistics. I share your reservations regarding "real" numbers, yet when the discussion concerns output growth I can't see the point of using anything else. With nominal dollar data you have both the distortion of price-level changes (admittedly small in recent years) and, more seriously, the effect of exchange rate movements. By my calculations, world aggregate GDP actually declined in 1997 - 1998 in current US dollar terms. That can't possibly tell us anything meaningful about output growth. And, of course, once we're able to extend the series to 2003 and 2004 the end-point to end-point US share of growth will collapse.
I could re-phrase Steven Roach to use these data as a more elaborate argument for dollar adjustment - if we assume that the dollar has been correctly valued, then we obtain the absurd result that the US has accounted for two-thirds of world output growth in the past seven years, therefore we are driven to conclude that the dollar's rise cannot be justified, or something like that. But I don't believe that was his point.