What should be the effect of productivity increases on the real exchange rate?

At a recent Society of Government Economists conference, entitled Public Policy and U.S. Competitiveness that took place last month, a paper written by Giancarlo Corsetti, Luca Dedola, and Sylvain Leduc, was presented that reported the following:

“For each [G7] country, using long-run restrictions, we identify shocks that increase permanently domestic labor productivity in manufacturing (our measure of tradables) relative to an aggregate of other industrial countries including the rest of the G7. We find that, consistent with standard theory, these shocks raise relative consumption, deteriorate net exports, and raise the relative price of nontradables — in full accord with the Harrod-Balassa-Samuelson hypothesis. Moreover, the deterioration of the external account is fairly persistent, especially for the US. The response of the real exchange rate and (our proxy for) the terms of trade differs across countries: while both relative prices depreciate in Italy and the UK (smaller and more open economies), they appreciate in the US and Japan (the largest and least open economies in our sample); results are however inconclusive for Germany. These findings question a common view in the literature, that a country’s terms of trade fall when its output grows, thus providing a mechanism to contain differences in national wealth when productivity levels do not converge. They enhance our understanding of important episodes such as the strong real appreciation of the dollar as the US productivity growth accelerated in the second half of the 1990s. They also provide an empirical contribution to the current debate on the adjustment of the US current account position. Contrary to widespread presumptions, productivity growth in the US tradable sector does not necessarily improve the US trade deficit, nor deteriorate the US terms of trade, at least in the short and medium run.”

The surprising result, at least to me, was that productivity growth relative to the rest of the world appreciated the real exchange rate (this is what is meant by no deterioration of “the US terms of trade”).
Now this may seem like an obvious result to most people, but in theory, this is not an obvious result.

It is true in a world where there are non-tradable goods, then increased productivity levels in tradables relative to that in nontradables, as compared to the rest-of-the-world, leads to a rise in the relative price of nontraded goods, and — with purchasing power parity holding for traded goods — leads in turn to an appreciation of the real exchange rate. This intuitive result is why many economists rely upon this channel — called the Harrod-Balassa-Samuelson approach in their theoretical models to motivate the empirical specifications tested. In my 2000 RIE paper investigating the behavior of East Asian currencies, I relied upon this mechanism.

But there is, in my work at least, an empirical problem with this explanation, at least insofar as it applies to developed economies like those of the G-7. In an (unpublished) paper written with Louis Johnston, we found that the relative price of ostensibly traded goods alone was also appreciated by productivity gains. This outcome means that one cannot appeal to the Harrod-Balassa-Samuelson effect (since only traded goods prices are being examined).

So now one has a conundrum. Increased productivity means increased supply of goods; assuming similar preferences and no wealth effects across countries, the terms of trade should worsen, as there are now more units of goods produced at home.

However, as Alan Stockman pointed out a long time ago (and explained in a simplified version here), when productivity increases in a two-good world with out nontradables, it might have (at least) two effects, depending upon the assumptions. The first is the intuitive, terms-of-trade-worsening effect as more home goods are produced, driving down the relative price of the home goods. But once one drops the assumption that preferences are identical across countries, and allows for home bias toward consumption of home-produced goods, it may be the case that the original effect is mitigated or overcome. Consider that the higher income will induce a biased increase in the consumption of home goods, thereby driving up the relative price of home goods. If in addition, home households hold disproportionate shares of the equity in the firms producing home goods, this too will lead to an exchange rate appreciating effect.

The puzzle in my mind is that this pattern is not found for Italy and the UK. Of course, if one’s maintained hypothesis is to find depreciation in response to a productivity shock, then the there is no puzzle with respect to these countries.

I should mention that this is a very active research area. Returning to the traded/nontradaded framework, one avenue under consideration involves the entry/exit of monopolistically competitive firms into the export market in response to productivity and other shocks (as in Ghironi and Melitz). With changing bundles of goods, persistent deviations from purchasing power parity can be predicted; whether these patterns exactly mimic the time series behavior of real world counterpart variables remains to be seen.

Technorati Tags: dollar,
real exchange rate,
productivity,
terms of trade,
Harrod-Balassa-Samuelson,
tradables, and
nontradables