Monday, September 16, 2013

Exchange Rates and Trade Flows in Asia

A bit more on shifting trade flows following the 1997 Asian Crisis.

Enno Schroeder, whose decomposition  of European trade flows I've mentioned here before, was kind enough to do a similar exercise for the four Asian crisis countries. His results are here; below I present them in graphical form below.

The conventional story, as we all know, is that relative prices drive trade flows. The Asian countries, in this view, moved from deficits to surpluses after 1997 because abandoning their currency pegs and devaluing made their exports cheaper and their imports more expensive. I've been suggesting a different story: Relative prices were relatively unimportant in the post-1997 move to surpluses, with the improved trade balance mostly or entirely a matter of lower imports resulting from the deep fall in income in the crisis. Looking at the picture in more detail suggests a more complex but in some ways even stronger version of my earlier story.

Some context: Suppose a country reduces its total import bill. As a matter of accounting, this reduction can be broken up into some mix of lower total quantity of goods bought, a smaller fraction of those goods being imports, and a lower price of the imported goods. Similarly for exports, any increase can be broken up into higher incomes in a country's export markets, a greater market share there for our exports, and higher export prices. So the overall trade balance -- here expressed as the ratio of total export value to total import value -- can be decomposed into the change in relative expenditure growth, the expenditure switch between the home country's goods and the rest of the world's; and the change in the relative price of home goods compared with foreign ones. (Note that relative price presumably affects trade volumes, but it also affects trade value directly -- for given trade volumes, if a country's imports are more expensive it will spend more on imports.) The cumulative contribution of each of these components is shown in the figures below, along with the nominal exchange rate. (The exchange rate is the nominal rate for July of each year, from the BIS.)

The heavy black line is the actual trade balance. Again, since the balance here is expressed as the ratio of exports to imports, a value of 1 means balanced trade. The other three solid lines show the cumulative contributions of each component to the changes in trade flows after 1996; the values represent how the trade ratio would have changed from that factor alone. Yellow is expenditure switch, from the rest of the world's goods to the home country's; this includes both home country switch from imports to domestic goods, and foreign country shift toward the home country's exports. Green is income growth in the country's trade partners relative to the home country. The solid red line is the terms of trade. The dotted red line shows the cumulative change in the nominal exchange rate; this isn't directly a contribution to the change in trade flows, but it's useful to know how closely the change in the terms of trade tracks the exchange rate.

It's convenient to think of the difference between the black line and the green line as the change in competitiveness.

The immediate effect of a devaluation is to make the home country's goods cheaper in the rest of the world, and the rest of the world's goods more expensive in the home country. The direct effect of this is to move the trade balance further toward deficit -- a descending red line in the figures below. But in the conventional story, the change in price leads to a more than proportionate change in quantity -- a rise in exports and/or a fall in imports -- so that the overall trade balance improves. This should show up here as a rise in the yellow line steeper than the fall in the red one. Income growth doesn't really come into the conventional story, so the green line should be flat.

This is not what we see. Even in terms of this simple decomposition, the post-crisis experience of each of the four Asian NICs was different, but none of them fit the standard story. Devaluations don't reliably translate into changes in the terms of trade, and changes in the terms of trade don't reliably translate into changes in trade flows. The income-trade balance link, on the other hand, looks quite reliable. In terms of the debate taking place elsewhere in econ blog land, this is a case where "hydraulic Keynesianism" looks pretty good.

Thailand is the clearest picture.

In the 1997 devaluation, the baht lost about a third of its value; the fall in the terms of trade -- the price of Thai exports relative to imports -- was less than proportionate, but still substantial. You can see this in the red lines at the bottom. But there was no expenditure switch at all. The flat yellow line shows that expenditure on foreign goods out of a given Thai income, and expenditure on Thai goods out of a given income elsewhere, did not change at all in the ten years after the crisis. (More precisely, expenditure in Thailand shifted toward domestic goods, while Thailand lost ground in its export markets; the two effects approximately canceled out.) Given that Thai goods were getting cheaper relative to foreign goods, the lack of net expenditure switch toward Thai goods should have led to deeper deficits. The only reason Thailand moved from deficit to surplus, is the decline in expenditure in Thailand relative to expenditure in its trading partners. The close match between the black and the green line in the figure, means that essentially the whole change in Thailand's trade balance is explained by the change in relative growth rates; there was no net switch toward Thai goods from foreign goods.

Indonesia is in some ways even a starker example:

Here we see a very deep devaluation, but again only a moderate change in the terms of trade, and an even smaller response of trade volumes. As in Thailand, the trade balance basically tracks relative income growth. The difference between these two cases is where the devaluation-trade flows link fails. In Thailand, the devaluation did reduce the price of exports relative to imports, but demand was not price-elastic enough for the change in prices to improve the trade balance. (In other words, the Marshall-Lerner-Robinson condition appears not to have been satisfied.) In Indonesia, the even larger devaluation -- the rupiah lost almost 80 percent of its value -- failed to change relative prices of traded goods, so demand elasticities did not come into play. This is partly because of high inflation in Indonesia following the devaluation, but not entirely - the rupiah fell by nearly half in real terms. But there was no change in the price of Indonesia's exports relative to its imports. If you want an example of a devaluation not working, this is a good one.

Korea, by contrast, looks superficially like the devaluation success story.

As I mentioned in the previous post, Korea was the only one of these countries where export growth in the decade after 1997 was as fast as in the decade before. And as the figure here shows, there was a substantial shift expenditure toward Korean goods following the crisis; alone among the four countries, Korea achieved its immediate post-crisis improvement in trade balance mainly through favorable expenditure switch rather than solely through a fall in income (though that contributed too.) But over time, Korea's terms of trade continued to deteriorate, without any further favorable expenditure switch; meanwhile, Korean growth slowed relative to its trade partners. By 2007, expenditure shares were back at 1997 levels; to the extent that Korea's trade balance was more favorable, it was only because spending was lower relative to its trade partners. Of course the surpluses it had run in the meantime had allowed the accumulation of substantial foreign exchange reserves. But if the goal is to use a lower exchange rate to achieve a permanent shift in trade balances, Korea post-1997 cannot be considered a success.

I should emphasize here: Slower relative expenditure growth in Korea does not mean slow growth in absolute terms. In fact, Korea (and, to varying degrees, the other three) did enjoy strong post-crisis recoveries. But because by far the largest trading partner for these countries is China -- taking about 25% of their exports -- even fairly strong growth translated into low relative growth. In other words, rapid growth in China implied growing exports in the NICs even in the absence of any competitiveness gains.

Finally, the one country that did achieve a lasting improvement in competitiveness, Malaysia.

In the immediate crisis period, Malaysia looks like Thailand and Indonesia: A deep devaluation fails to pass through to the relative prices of traded goods, and there is no expenditure switching; instead, the entire burden of raising the trade balance falls on slower growth in domestic expenditure. In the case of Malaysia, domestic expenditure fell by an astonishing 28 percent in 1997, a collapse in economic activity that has few precedents -- neither the US in the 1930s nor any Euro-crisis country comes close. But in the case of Malaysia, unlike the other three countries, growth subsequently accelerated relative to its trade partners, reflected in the downward sloping green line; at the same time, there was a continued expenditure switch in favor of Malaysian goods, reflected in the upward slope of the yellow line. What's especially striking about this competitiveness success story is that the favorable expenditure switch happened despite a rising price of of Malaysia's exports relative to its imports.

To continue with this analysis properly, one would want to disaggregate imports and exports by sector or industry. And would want to study, for each country, the institutional and legal changes that influenced trade flows in the decade after 1997. But failing that, it's at least worth understanding what the aggregate numbers are saying. It seems to me that they are saying this:

Even a very deep devaluation, as in Indonesia, is not guaranteed to change the relative prices of a country's imports and exports.

Even if a devaluation is passed through to relative prices, as in Thailand, price elasticities may not be large enough to produce a favorable change in the trade balance.

Even if a devaluation moves relative prices, and demand is price-elastic enough for the price change to move the trade balance in the right direction, as in Korea, a short-term improvement in competitiveness may not persist.

When countries do achieve a long-term improvement in competitiveness, like Malaysia, they don't necessarily do so through a relative cheapening of exports compared to imports. On the contrary: If the Marshall-Lerner condition is not satisfied, then a relative increase in the price of a country's exports will raise export earnings. In the case of Malaysia, improved terms of trade (that is, a rise in the price of its exports relative to its imports) account for about half the long-run improvement in its trade balance.

The Asian precedent does not make a Greek (or Spanish, or Portuguese, or Irish) devaluation look like an obviously good idea.


One other thing, if even real exchange rate changes are not passed through to traded-good prices in the destination country, then they must be showing up as changes in exporter profit margins. This shifts the focus from demand responses to supply responses, which I would argue are  more institutionally mediated. As you can tell if you've read this far, I am sympathetic to the "elasticity-pessimist" strand of Post Keynesian thought. But on the other side Robert Blecker has a strong argument for a strong effect of exchange rate changes, focusing on the role of export-industry profits in financing investment. Blecker's paper, in my opinion, is more convincing the straightforward "prices matter" view of exchange rate changes. But it also suggests a certain asymmetry: low profits induce exit from tradable sectors, especially for countries with Anglo-American market-based financial systems, more reliably than high profits encourage entry.

UPDATE: The fact that even large exchange rate changes produce relatively small movements in the relative prices of traded goods is well-known in the empirical trade literature. See for example here. I should have made this clearer.


  1. JW,

    I don't think that elasticity pessimism is inconsistent with exchange rates matter view. The elasticity pessimism will say that countries with income elasticities in its favour will have higher incomes as compared to the ones with the opposite. At the same times, the exchange rate gives some advantage than otherwise.

    The elasticity pessimism is actually inconsistent with the (strong) view that exchange rates adjusts to bring in full employment and convergence of living standards and so on and stories/conjectures of that kind.

    So Joan Robinson may be said to be an elasticity pessimist but she wrote on beggar-my-neighbour as I would imagine she thought it does beggar neighbours to some extent.

    (The above isn't disagreeing to any point of yours).

  2. "if even real exchange rate changes are not passed through to traded-good prices in the destination country, then they must be showing up as changes in exporter profit margins."

    I'm very symphathetic to this point of view, because I think that this is the major way nations compete one against the other (in general, not just during crises).
    This would explain the general increase in the share of income that goes into profits during the "globalisation" period.
    Slightly OT, but I think that it is important also the fact that developing economies try to climb the ladder of "high added value" industries by offering them an higer share of profits, and this is politically important.
    Have you some link whith something on this argument? pleez...

    "As you can tell if you've read this far, I am sympathetic to the "elasticity-pessimist" strand of Post Keynesian thought."

    When people speak of elasticity and competitiveness, I always have doubts about the timeframe of reference.
    For example, China trades with Japan and with Thailand. Thailand devalues, so that it would be in theory much more competitive than japan. However, a big part of trade is not in final goods but intra-industries, and Chinese industries have long term contracts with Japanese industries that can't be changed overnight; also the people who work in the Chinese industries maybe know some Japanese guys whith whom they worked until now, while maybe they don't know Thai guys, etc.
    In short, Chinese industries face a lot of costs to switch their contracts from Japan to thailand.
    On the short term most chinese industries are likely to keep their contracts with the Japanese, over the long term however Chinese companies will slowly switch to Thai suppliers.
    On top of this, some of the "competitiveness" is likely a consequence of the industrial developement of a nation, so that there might be a virtuous circle where a nation becomes more competitive, snatches some market share in (say) electronics, this causes an increase in know-how and infrastructure, which makes said nation even more competitive.
    Thus when we speak of elasticity of exports, we are speaking of three different aspects that arguably work on different timeframes:
    1) we could have a fall in prices that immediately affects "a priori competitiveness", but might not show in the stats because
    2) when one tries to evaluate competitiveness "a posteriori" from the actual changes in trade volumes, transaction costs look really big, so we might see a long and slow increase in "a posteriori competitiveness" that in reality is just the developement of a stark increase in "a priori competitiveness". In such a situation, saying that there is few elasticity is quite misleading IMHO;
    3) while this is beyond the point of the OP, there is likely also a very long term story of nations that try to create or retain a big industrial base, and a lot of the "austerity/moral hazard" rethoric becames more rational from this point of view.

    1. Very thoughtful comment as usual, RL. I will post a proper reply tonight, after I am done teaching....

  3. I agree with the majority of the points in this article and it’s great without any doubt. Really a wonderful post! I like it very much. Here I find everything in details. I hope I will see this type of post again in your blog.