Yeah, that would be me. Some time ago, I included in this post a picture purporting to show the evolution of dollar reserves held by central banks over time. The upshot was that the fraction of dollar-denominated assets in foreign central bank portfolios had fallen dramatically, and very gradually, over the 1980s and into the 1990s, before rising significantly in the late 1990s and early part of this decade. My point was that (a) portfolio adjustments by foreigners are likely; and (b) we've seen it before, and it looked like a pretty orderly and prottracted process.
Almost immediately, several readers -- the charge being led Brad Setser and Menzie Chinn -- suggested that my figure was suspicious. They were right. The data I was using was actually for a very narrow measure of central bank reserves that would not be expected to represent the movement of reserve positions more generally.
Since that time I have been trying to shake loose a consistent time series from the International Monetary Fund, to no avail. But here's an "estimate" based on the maximum and minimum shares publicly reported in various IMF publications.
This is certainly a less sexy picture than the one I showed originally, but I think the basic contour of the story remains. From the early 1980s, the share of dollar-denominated assets unwound very gradually, before increasing back to early 1980s levels in the last part of the 1990s. In fact, this data suggests that the adjustment to more diversified central bank portfolios out of the Bretton Woods system was drawn out over a couple of decades.
A more definitive determination awaits a better treatment of the data. I know there are people working away on this, and will gladly post on any work anyone out there might have done on this subject. Until that time -- and having goofed on this once -- here's a spreadsheet with the data that underlies the picture above (thanks to my Faithful Assistants Janet Miller and Shadya Yazback):
Download IMF_table_1.2_share_of_currency_in_forex.xls
Have at it.