Skip to main content

Elaine Thompson/AP Photo

Excerpted from World 3.0: Global Prosperity and How to Achieve it by Pankaj Ghemawat. Reprinted by permission of Harvard Business Review Press. Copyright 2011, Pankaj Ghemawat. All rights reserved.

The Mystery of the Missing Trade

The U.S.-Canadian border is the world's longest undefended border. Trade across it accounts for the world's largest bilateral trading relationship, still larger than that between the United States and China. Two-way trade between the United States and Canada amounted to nearly $750-billion in 2008 before falling to $600-billion in 2009, thanks largely to the decline in energy prices and weakness in the auto sector; in both areas, Canada is the United States' largest foreign supplier. So important is this trading relationship for Canada that the Canadian government regards several industries as more susceptible to U.S. economic conditions than to domestic ones. Thus, Canada's $100-billion drop in exports to the United States between 2008 and 2009 was three times as large as the decline in Canada's GDP during that period.

Story continues below advertisement

All this suggests lots of trade, but we shouldn't just jump to the conclusion that the border doesn't matter. In fact, economists who've looked at U.S.-Canada trade in recent years haven't puzzled over why it is so large; rather, they've wondered why it isn't nearly as large as one would expect if the border didn't matter. There has been a spate of work on "the mystery of the missing trade," since the original finding that in 1988, when the United States and Canada signed a free trade agreement, merchandise trade between Canada's different provinces was twenty-two times as intense as their trade with the United States.

The 1988 free trade agreement did reduce the "home-bias multiple," as economists call it, by the mid-1990s, but only to twelve (and with the multiple remaining stuck at thirty to forty in the case of services). It is currently estimated to be between five and ten - lower than before but still significantly greater than the level of one that would correspond to zero home bias. Corroboration of significant border effects comes from the price differences between the United States and Canada. As so many border dwellers know, there's a reason to go on international shopping trips (although this type of "suitcase trade" amounts to only a small percentage of total trade, and is therefore insufficient to eliminate price differentials).

How do other borders stack up to this one? It's hard to tell, since very few countries maintain data on within-country trade of the sort available for Canada. However, we can get a sense of merchandise trade across regions within a country by examining regional transportation flows.

One study that does so concludes that German länder, or states traded four to six times as much with each other as with other EU countries in 2002, and that the corresponding home bias multiple for the French regions was about fifteen. More than three decades after the EU eliminated all formal trade barriers, such as tariffs and quotas, between member states, the German and French borders still matter a great deal.

If borders still matter so much between neighbors, they cast an even bigger shadow on trade between countries farther away from each other.

My analysis of Spanish regions' merchandise trade with other OECD countries over 1995-2005 found a home bias multiple ranging from fifteen with Portugal to 150 for Japan! As we know from other studies, Spain hasn't integrated with world markets as well as Germany or even France, and these numbers bear that out. And the variation in the home bias multiple reminds us that a border effect is a "bilateral effect"-that is, it depends on which country pair one is talking about - rather than a "unilateral effect," which depends on the attributes of just one country.

For even more evidence that national borders impede trade, we can look to situations where new borders have arisen or old ones have gone away.

Story continues below advertisement

In 1993, when the former Czechoslovakia broke into the Czech Republic and Slovakia, the two governments took significant measures to preserve open trading relations, including a customs union, a temporary payment mechanism to deal with delinked currencies, and an agreement stipulating free movement of labor. Yet, trade intensity between the two new countries fell from forty times the "normal" level of trade with other countries in 1991 to ten times by 1995.

Meanwhile, Germany's experience illustrates the effects of removing national borders. In the five years that followed the reunification of the former East and West Germany in 1990, trade between the two shot up sixfold, and the share of intra-German trade in their overall trade grew fourfold. These gains reflected large investments intended to facilitate integration, including spending on physical infrastructure like rail lines and highways, and the East's rapid development as a result.

Even more interesting than the rapid increase in trade, however, are estimates that it will take decades before effects of the former East-West border disappear. Erection of a new border can cause trade to collapse almost overnight, as in the Czechoslovak example, but removing a border has a much slower economic impact. This makes sense when you think about the relationships that accrete over time between buyers and sellers, the investment in familiar brands, the knowledge that locals have with local markets, tastes, preferences, and, of course, connective infrastructure. Removal of a barrier doesn't put outsiders on equal footing with locals - not for decades, at least.

Of course, there are also studies focused on emerging markets.

Although Brazil opened up to more international trade during the 1990s, Brazilian states still traded an estimated twenty-seven times more with each other than with foreign countries in 1999. China's estimated home bias in the late 1990s was also in the twenties. This figure would have been higher if Chinese provinces hadn't become significantly less integrated with each other: between 1987 and 1997, provincial border effects are estimated to have more than doubled.

The effects of borders between states or regions within the same country in limiting trade seem particularly large in the BRIC countries (Brazil, Russia, India, and China) because of their size, poor infrastructure (especially in the hinterlands), and administrative barriers to internal trade. In general, though, the effects of internal borders are an order of magnitude smaller than those of international borders. Thus, in 1999, Brazilian states "traded" internally more than ten times as intensely as with other Brazilian states but 280 times as intensely as with foreign countries; for Chinese provinces in 1997, those multiples were estimated at twenty-seven and more than 400 times. The point is not that internal trade flows or barriers to them are unimportant: in large countries, in particular, internal trade is often significantly larger overall than international trade and therefore even relatively small impediments to it can matter a great deal. Rather, the point is that if we want to "solve" the mystery of the missing trade, we ought to look at national borders, since that is where the really large drop-offs in trade are observed, not at state or regional borders. In other words, World 0.0, with its primary focus on the subnational level, turns out to be even less realistic a worldview than World 1.0.

Story continues below advertisement

Pankaj Ghemawat is the Anselmo Rubiralta Professor of Global Strategy at IESE Business School in Barcelona. He is the author of five books including the award-winning Redefining Global Strategy. For more visit:

Report an error
As of December 20, 2017, we have temporarily removed commenting from our articles as we switch to a new provider. We are behind schedule, but we are still working hard to bring you a new commenting system as soon as possible. If you are looking to give feedback on our new site, please send it along to If you want to write a letter to the editor, please forward to