Menzie Chinn has an intriguing post on Econbrowser that I hope I can do justice. Chinn is a serious analyst of currency price formation and continues to look at the question of the prospects for the dollar. In looking at the value of a currency, there is quite a lot of artwork in deciding how to measure that value. Chinn notes that the most common approach is a trade-weighed measure; others use assets or liabilities. But even in the universe of trade-weighted measures, there are differences in methodology. The Fed, IMF, and BIS use an index-based approach, which means you start with a base year and measure changes (and as anyone who has worked with indexed data knows, conclusions can change considerably with the choice of different base years). Chinn decides to look at a different approach (using levels of exchange rates in “real, common terms”).
This turns out to be important because as we will discuss, the change in methodology leads to a change in conclusion. Thus, the consensus view, that the dollar has depreciated more or less as far as it needs to against the Euro, but probably has further to fall against the yen and yuan, may not be as clear cut as the consensus believes. But curiously, Chinn downplays his findings, going though his analysis and reserving his (perhaps controversial) conclusion to his summary paragraph.
Last week, I wrote a post examining what the measures of central tendency for the dollar’s trajectory were, based upon some standard forecasts. This week, I want to examine more closely whether we should anticipate more depreciation, in real terms, by way of discussing alternative measures of the dollar’s value.
Chinn decides to use the so-called Thomas-Marquez-Fahle [pdf] index, noting that this measure did NOT come from the Federal Reserve). One of the key features of this index is that it allows the rising importance of East Asian trade to have a direct effect on the measure of the value of the dollar.
Chinn goes through why he likes the Thomas-Marquez-Fahle index. The big reason, and this is a compelling one, is that is has a lot of explanatory power. When trying to estimate the US trade balance, he generally has to rely on certain trends as explanatory variables. Use of the TMF index means he can eliminate some of these trends. In regression land, the fewer the number of explanatory variables, the better the model. Chinn uses this chart to illustrate:
Figure 2: Net exports ex-oil to GDP ratio (blue, left axis), log real value of the USD (red, right axis) and Thomas-Marquez-Fahle index (green, right axis), rebased to 1985q1=0. NBER-defined recessions shaded gray. Source: BEA GDP release of November 29, 2007, Federal Reserve Board, personal communication, NBER, and author’s calculations.
Use of the TMF index points to some other conclusions that may seem surprising. One is that the dollar isn’t all that weak:
Figure 1: Log real value of the USD (broad) and Thomas-Marquez-Fahle index, rebased to 1985q1=0. Source: Federal Reserve Board, personal communication, and author’s calculations.
This suggests to me (this is not asserted in the Thomas-Marquez-Fahle paper) that there is still substantial dollar adjustment required in order to get the net exports ex-oil to GDP ratio to something around 1% of GDP. This conclusion, by the way, is not inconsistent with the view forwarded by Brad Setser that the dollar has declined by a sufficient amount to induce some adjustment. It’s also a view consistent with why we haven’t seen more global rebalancing.
This is a pretty bold finding by a highly respected trade/currency analyst. He appears to have soft-pedaled it a bit due to his reliance on an alternative methodology.
The coming year will reveal whether Chinn is on to something.