Saturday, December 29, 2007

Hoisted From Comments: Query On Foreign Trade

A reader asks:
"This is an unrelated discussion, but I was hoping that maybe you (or someone else reading this forum) could help me with a question that may also be of interest to you.

How does a standard neo-Walrasian CGE (computational general equilibrium) model predict economic losses from trade liberalisation?

I understand the logic behind the two goods and country case where welfare gains can be made through the elimination of trade barriers if there exists a comparative advantage (assuming the standard assumptions of perfect competition, constant returns to scale etc.), but how does the logic work with more than two countries and goods?

Is the answer simply that those countries which do not have a comparative advantage (in any or critical goods), experience welfare losses, because they are now importing more (due to lower prices abroad) and exporting less (trade is being diverted to other countries)?

Most (if not all) countries when getting the CGE treatment with full liberalisation scenarios, seem to make welfare gains. Is this because most countries have at least some comparative advantage in some goods, and the theory neglects any adjustment and transaction costs?

Some background on this question: I'm not an economist by trade, but rather study development studies, for which I'm writing a Master's Thesis. My strategy has been to use the Lakatosian methodological framework for studying the hardcore assumptions behind general equilibrium analysis. This has led me to study for example Keen's work, and Fabio Petri's (on the capital controversies) among others. Since I'm not concentrating on heuristics (and I'm not particularly mathematically inclined), I'm not familiar with all the implications of the standar neo-Walrasian theory (e.g. Arrow-Debreu) that the criticism, quite rightly it seems, is able to destroy by attacking its inconsistent or absurdly narrowed assumptions. To the standard unacquainted reader (for example my supervisor) it is however perhaps not only enough to attack the assumptions, but also be able to reduce the theory via its implications. I'm thinking that for a conclusion, I could write up a kind of characterisation of what the theory implies in the narrow sense and a wish list of what problems a perfect economic theory would be able to cover and account for as a contrast (perhaps a distant dream, but why not?)

Any ideas on the above would be appreciated.."
I don't know much about CGE models. I understand that commodities can be indexed in the Arrow-Debreu model on space, and this provides a theory of foreign trade. In the Arrow-Debreu model, the initial endowments of all commodities are taken as given data. This makes it a very short run theory. I was under the impression that the traditional argument about comparative advantage takes place in the Heckscher Ohlin Samuelson (HOS) model, which is a long run model.

Given the presence of produced capital goods and a positive interest rate, many supposed theorems in the HOS model are simply incorrect. In particular, trade according to the pattern of comparative advantage can move a country's Production Possibilities Frontier inward. Ian Steedman is a good author to read here. Perhaps your university library has a copy of the New Palgrave, and Steedman wrote the entry on "Foreign Trade". (I talked to Keen about these ideas before the publication of his book. He said that if he writes another edition, he might put in a chapter on foreign trade.)

I don't think comparative advantage explains the pattern of trade. A theory based on the Keynesian multiplier is another possibility. Literature here includes Luigi Pasinetti's Structural Economic Dynamics: A Theory of the Economic Consequences of Human Learning (Cambridge, 1993). Paul Davidson has proposed some reforms of the international monetary system. I have not read Ha-Joon Chang, neither his Bad Samaritans nor his Kicking Away the Ladder.

Two PDFs that I have downloaded from somewhere or other in the past year and have never got around to reading might be of interest. I am talking about Roberto Mangabeira Unger's Free Trade Reimagined and Robert Driskill's "Deconstructing the Argument for Free Trade".

Another body of literature that I have never explored is new trade theory, as presented by Paul Krugman. Given that comparative advantage fails to justify free trade, I don't see the point of that theory. As I understand it, new trade theory asserts incorrectly that comparative advantage would provide this justification if it were not for increasing returns or oligopoly or something. According to Barkley Rosser (in a 1996 book review in the Journal of Economic Behavior and Organization), Krugman claims more originality for his presentation than can be justified.


Gabriel M. said...

Robert, could you please elaborate on this?

>> "In the Arrow-Debreu model, the initial endowments of all commodities are taken as given data. This makes it a very short run theory."

That initial endowments are given is true of ANY more. Would you say that the Solow growth model, to pick a traditional example, is short-run?

As a first-order approximation I would say that if by "long run" you mean steady states, then model will generally converge to some steady state or another, regardless of the initial endowments.

How can a model that gives us all prices and quantities for all future time periods and world states be "short-run"?

Beyond that, a few thoughts on "free trade" and why it's justified...

1) Because in a free society you justify restrictions not freedom, freedom of (international) association being at stake. In other words, it's up to protectionists to justify their policy proposals.

2) Larger markets, less market power.

3) Division of labor. Free trade might allow smaller economies to reach low-cost production scales.

4) You can eliminate most or all (real or imaginary) ills of "free trade" with fiscal policy so don't throw the baby with the bath water.

Have an awesome 2008!

Anonymous said...

Regarding the inapplability of the law of comparative advantage to real world trading patterns, see Paul Davidson's book , JOHN MAYNARD KEYNES (Palgrave/Macmillan, 2007, pp. 128-137.) For comapartive advantage to hold and be applicable requires the following assumptions : (1) full employment before and after trade in all trading partners, (2) no mobility across national borders for either capital or labor.

If there is not full employment and capital is free to cross national borders, as those who promote free international capital flows demand, then trade will depend on absolute cost advantge -- not comparative advantage (as demonstrated in the above cited pages of Davidson's book).

Except fr natural resources most cost advantages depend on relative wages of workers in the trading countries-- e.g., wage differentials between China India and the USA.

Absolute cost advantage predicts that jobs will be outsourced to the lowwage country (as long as technology is the same crosss borders -- and with multinational corporations technology is easily transportable over borders -- as long as transportation and communication costs are not sufficient to offset the difference in money wages when wages in each country is expressed in a common currency.

We have already seen this affecting most US blue collar jobs -- as our industrial base disappears, as well as call centers in India, as well as white collar technical jobs like computer programers in Asia,
Thus in the long run, this means for the US that all jobs can be outsourced except for those that require personal attendance (e.g., waiters, newspaper deliverers, health care attendants, college professors, etc.), or military defencse industry jobs, while the average wage of American workers sinks to the level of a Chinese . See Krugman's article on Trade in the NY TIMES, December 28. where he predicts the continual collapse of US wages. [

YouNotSneaky! said...

Well, in the HO model endowments of K and L are taken as given so in that sense it looks like the static AD model. Of course you can allow for capital accumulation and even for changes in L (Malthusian models) but then things get a lot more complicated with the possibility of immiserizing growth and so on (dynamic competitive advantage).

In fact, I'm pretty sure when people talk about forecasting the effects of trade agreements using CGE models they're talking (mostly) HO.

To anon's question about when welfare can go down in those models - that'd probably be the optimal tariff literature where you don't assume that a country is 'small' relative to the world. In general these tariffs are estimated to be very low and in some cases even negative (i.e. you should be subsidizing imports).

To Robert's comment that "comparative advantage doesn't explain the pattern of trade" - that's a bit of a silly thing to say. Of course if it's modified to "comparative advantage doesn't explain ALL pattern of trade" then that's fine. After all, what do you think Venezuela's #1 export is? But it could very well be that a decreasing share of world trade is explained by CA today.

As far as the New Trade Theory. Well, it's all based on increasing returns to scale. Hence it is substantially different from traditional CA models. In fact it is sort of antithetical to it. CA results from specialization. NTT trade results from larger market size. Whether Krugman deserves all the credit here that he gets is another question.

Oh yeah. In HO with more than two countries, CA may become indeterminate. In general the number of countries vs. number of goods matters. Obviously.

I would be interested in seeing how the Keynesian multiplier (for existence of which there is even less empirical support than for CA based trade) can explain the pattern of trade. I can sort of see how it would explain the volume of trade but "pattern"? Do various goods have different multipliers?

YouNotSneaky! said...

"Paul Davidson's book , JOHN MAYNARD KEYNES (Palgrave/Macmillan, 2007, pp. 128-137.) For comapartive advantage to hold and be applicable requires the following assumptions : (1) full employment before and after trade in all trading partners, (2) no mobility across national borders for either capital or labor."

I do respect Paul Davidson so I'm guessing that this is a misreading. These two conditions are at best sufficient, not necessary. If 1) fails it is true that it may be possible for a small tariff to increase welfare (this is why Brad DeLong always caveats all his support of free trade with "as long as the Fed is doing its job"). However this isn't really that CA isn't applicable. It's just that other factors can matter more sometimes.
Wrt 2) If capital is mobile across countries then it will move until its marginal product is equalized (we're working within the neoclassical model here) and the only productivity differences between countries will be due to technology. But as long as technologies differ, comparative advantage still applies. Similarly if only labor is mobile. If both are perfectly internationally mobile then it's possible that all productivity differences are eroded - and in that case there's no CA and no reason to trade.

So note that while there's some truth to 1), 2) is essentially irrelevant and only applicable (in the sense that it results in 'no CA, no reason to trade' - but then why do you care?) if 2a) labor is PERFECTLY mobile, 2b) capital is PERFECTLY and 2c) there are no TECHNOLOGICAL differences. In contrast to 1) and 2) being sufficient and not necessary, 2a), 2b), 2c) are all necessary for 2) to matter (in an irrelevant sort of way).

Gabriel M. said...

YNS!, of course you want to compare steady states but my point was that there's nothing stopping you from having transitional dynamics towards that steady states and have that steady states imply whatever quantity of capital.

Back in my day (2 years ago ;-) we used a 2 country Solow-ish model and you had all the traditional conclusions. I'll try to dig that up and make a blog post out of it.

Peter H said...


You might appreciate these pieces by William Milberg "Say’s Law in the Open Economy: Keynes’s Rejection of the Theory of Comparative Advantage” & an entry on Globalization in the Elgar Encylcopedia on Post-Keynesian Economics.

Peter H said...

I'd particularly recommend the Milberg entry on Keynes' rejection of the theory of comparative advantage. However, it's important to note that one can reject traditonal approaches to comparative advantage and still oppose protectionism. See Milberg & Jan Kregel's comment in the Financial Times.

Robert Vienneau said...


Thanks for the suggestions. I'm downloading the more strongly recommended piece.

Blogus Pokus said...

Thanks for the comments on this section Robert and sorry for a belated answer (the New Year and related adjustments provide some background towards understanding the reason for this late answer).

I have read Ha-Joong Chang's Kicking Away the Ladder, and found it a good informative read. His work is more general on the merits of foreign trade and protectionism, so for my current analysis it is not that useful. But I do recommend it to anybody interested in trade policy.

I haven't yet got my hands on the New Palgrave. I've read some criticism from Mark Blaug however that this piece of work is generally biassed towards the Sraffian account of General Equilibrium theory. I have to say though that I do not yet understand how or whether the Sraffian strand of GE has been taken up by applied CGE economists. I've got a book from Walsh & Gram "Classical and Neo-Classical Theories of General Equilibrium", which might help to clear up this matter..

As for the other references, I have to take a look.

I share the impressions here that H-O is applied on top of the edifice of the Walrasian GE-framework, although I yet do not know this for sure. Whether the applied CGE-analysis is then basically akin to H-O with more than two countries, and factors is also something I need to dig up.

On the instant nature of the Walrasian equilibrium adjustments and offshoots into steady states I propose reading an article called Equilibrium and Disequilibrium in Walrasian and Neo-Walrasian Economics by Michel de Vroey in the Journal of History of Economic Thought (2002). This paper clarifies some of the conceptual issues.

Another more complicated and mathematically oriented work is Fabio Petri's "General Equilibrium, Capital and Macroeconomics". I would generally be interested to know what people think of this Magnum Opus. It took me quite a long time to understand his point, but when it clicked, it has helped to contextualise other literature in the GE-fold.

Robert Vienneau said...

Thanks for the de Vroey reference.

Given my focus, it's no surprise I like Petri's book. I think what a definitive Sraffian critique of G.E. theory looks like is still open.

Some time ago, I mentioned textbooks treating the Cambridge Capital Controversy. Walsh and Gram is in this list, and some of the others try to get at the nature of Sraffa's alternative theory of value and distribution.

I worry that this sort of list delays your thesis rather than helps.