The most recent entry is Mankiw's Ricardo vs Hecksher-Ohlin. He asks the question: "Which model is more useful in thinking through issues in trade policy: the Ricardian model or the Heckscher-Ohlin model?" After considering the implications of mobile capital, he then writes that "[a]s a tentative conclusion, therefore, I am inclined to think that in a world with significant capital mobility, the Ricardian theory of trade is more useful than Heckscher-Olin."
I'd like to contribute a different perspective on the issue. As every economist would surely agree, Mark Thoma is absolutely correct when he writes
Choose the model that best answers the question you are interested in (and that can vary with the question you are asking) and then, if there are weaknesses in the model you choose that may affect the conclusions, those ought to be clearly identified and explored.But that is not what Mankiw does when he reaches his tentative conclusion in favor of the Ricardian model. Rather than work from the question that needs to be answered, Mankiw seems to be comparing them based on which model is less unrealistic (after all, we must acknowledge that both models are - by design - grossly unrealistic). Since economic models are not meant to be at all realistic, but rather are intended to help teach us through analogy, I find Mankiw's answer unsatisfying.
When I teach international trade theory, my students find that the Ricardian model is the simplest and easiest-to-grasp illustration of the answer to one big question:can international trade leave both countries better off? The Ricardian model makes it easy to see that the answer is yes, which is what the model was designed to show, and is the single best thing about it.
But I don't really think that that is the only important - or even most important - question asked about trade policy today. Instead, trade policy now is at least as concerned with much more subtle questions like what specific form(s) will the gains from trade take if we pursue trade policy xyz?, or who will gain and who will lose from trade policy xyz?
For better or worse (and I think it's probably for the better), trade policy today must have as a crucial component some consideration for the distribution of the benefits of trade, and not just be concerned with whether trade liberalization will make some aggregate measure of US economic activity go up or down. These concerns are vitally important for moral, economic, and political reasons, and must be addressed when it comes to the creation of trade policy.
And when it comes to questions of the distribution of the gains from trade, the Hecksher-Ohlin (H-O) model is ideally suited to give us important answers - because that is exactly what it was designed to do. The H-O model provides us a simple and clean insight into how and why there are winners and losers from any given international trade policy, and exactly who they might be. That's why I find my mind automatically turning toward the H-O model for guidance whenever any new trade policy issue comes up, not the Ricardian model.
The Ricardian model is great for illustrating that countries as a whole can benefit from trade. But trade policy today must consider a whole lot more than just that simple calculation. So for my money, when it comes to current questions of trade policy, the H-O model is the one that has much more relevence.