An alternative view on the dollar’s strength and trend over time.
Readers of this Econbrowser are familiar with the various measures of the dollar’s value. These include the nominal and real indices calculated by the Federal Reserve. In the figure below, I present the Fed’s real narrow and broad dollar indices. Both are calculated using CPI’s. Alternative measures by the IMF, the BIS, BoE and OECD would look similar.
Figure 1: US Dollar real value indices, against major currencies (blue) and against broad basket (red). Source: Federal Reserve Board.
One of the mysteries [pdf] of the recent behavior of the U.S. macroeconomy has been the outsized trade and current account deficits, relative to history, against a backdrop of a relatively weak dollar. Furthermore, the continued deterioration of the external accounts despite a substantial decline in the dollar’s value since 2002 has induced additional puzzlement.
In a new paper [pdf], Charles Thomas and Jaime Marquez have proposed a new explanation for why we see these seemingly paradoxical conditions. Essentially, their argument is that the standard exchange rate indices mis-measure the value of the dollar (although in principle, this critique applies to all currencies, with varying force). In this view, the standard measures underweight the prices associated with low cost countries (e.g., China). Hence they propose an alternative approach which addresses this deficiency. It is important to note that their alternative measure is NOT an official measure of real exchange rates in any sense of the word. Their measure is solely their own research construct and they are not trying to replace or debunk any of the Fed’s indexes of real exchange rates. Rather, they are attempting to provide another way of looking at relative prices.
In order to be clear, it is important to provide some technical background. Most currency value indices that are now reported — including those of the Fed, the IMF, and the BIS, are Divisia indices (for more on effective exchange rates, see here [pdf]). Such indices are given by the following formula.
where q[it] is the real exchange rate against another currency, and w[it] is the trade weight, which given the notation can vary over time. Hence, the Divisia index is the cumulated chained weighed averages of changes in bilateral real exchange rates. Since the Divisia only looks at changes in the components, their base period is irrelevant. And since the aggregate is constructed as a series of changes, its base value is entirely arbitrary. In contrast, their geometric index is exactly what it sounds like — the geometric weighted average of the levels of bilateral real exchange rates.
The essential feature of the geometric average is that even as a given real bilateral exchange rate remains constant, if its (trade) weight in the index increases, then the overall exchange rate index will be affected. This does not necessarily occur — except indirectly in affecting the exchange rate index growth rate — in the Divisia index.
Now, sometimes this feature of the geometric average is undesirable. However, depending upon the issue that one wants to examine, sometimes one wants a non-changing bilateral exchange rate to have an increasing effect on the overall index. In such instances, one might prefer the geometric index.
Thomas and Marquez construct and present their geometric measure (black) and Divisia (red) index for the U.S. dollar using both overall and non-oil trade weights in Figure 2 (top panel of Figure 1 from Thomas and Marquez [pdf]). [This measure requires relying on a relative price levels for which they use data from the Penn World Tables. To isolate the differences that the aggregation method makes, they calculate both the Geometric measure and the Divisia index using the Penn World Tables data.]
Figure 2: Top panel, Figure 1 from Thomas and Marquez, “Measurement Matters for Modeling U.S. Import Prices,” mimeo [pdf].
In this paper, Thomas and Marquez focus on pass-through and find that using this alternative measure of relative prices, long run exchange rate pass-through has not declined as has been suggested in other studies (see a previous post on pass through here).
What is remarkable to me is that in this reformulated dollar index, the dollar at the begining of 2005, after depreciating from its most recent peak in 2002, is as strong as it was at the dollar peak in the mid-1980′s. To me, this suggests that sustained reduction in the the trade deficit would need to be accompanied not by depreciation to unheard of lows, but back to levels more consistent with the averages of the 1990′s.
Further departing from the Thomas-Marquez paper, consider the implications. The most important one comes in response to the question of why the dollar has remained so strong over the past six years? There are obvious candidates — such as the Rmb peg and the increasing integration of China into the global production system. But I think we should not forget home-grown reasons, including a boom in residential investment and other non-tradables expenditures spurred by expansionary monetary policy, and surging government spending biased toward home goods (after all, we don’t buy our main battle tanks from overseas).
[minor editorial changes at 9:25am 10/17 - mdc]