Bairoch hypothesis

Elhanan Helpman - The mystery of economic growth (The Belknap Press of Harvard University Press, 2004) 72-79

Evidence on Trade Policies

Trade volumes depend on endowments, technologies, preferences, and market structures, and on how these characteristics differ across countries. As a result, the trade volumes of some countries would be high and of others low even if all of them engaged in free trade. For this reason it is not apparent that growth rates should be positively correlated with trade volumes across countries. Moreover, even if one believed that trade promotes growth, one would not necessarily conclude from this premise that larger trade volumes promote faster growth. As a result, studies that examine correlations between growth rates and trade volumes cannot provide fully satisfactory evidence on the effects of trade on growth. It would have been more informative to study the mechanisms through which trade influences growth. But data limitations greatly restrict research of this type. As a result, growth economics—like many other areas of economics that face similar problems—has turned to the study of indirect relationships instead.

A great many studies have examined the impact of trade policies on growth. We have seen that growth theory does not predict a simple relationship between trade policies and growth. In some countries a restrictive trade policy may accelerate the growth of income per capita, in others it may slow it down. The way trade policy affects an economy’s growth depends on the economy’s characteristics, such as the type of products it trades on foreign markets or the human-capital intensity of its importcompeting sectors. Nevertheless, empirical studies do not provide estimates of the growth effects of trade policies conditioned on these characteristics. Therefore estimates that exploit crosscountry variations are best interpreted as average effects of trade policies on growth, similarly to the estimates of the effects of trade volumes on growth that were discussed above.

Bairoch (1993, chap. 4) argued that the European experience in the late nineteenth century does not support the view that protection is bad for growth. According to Bairoch, the liberal phase of European trade policies lasted from 1860 to 1892. In response to an inflow of cheap grain from Russia and the New World, some countries raised their impediments to trade. France went protectionist in 1892. The growth rate of its GNP increased from an annual average of 1.2 percent in the decade preceding the policy shift to 1.3 percent in the decade following the policy shift. Germany changed its policy in 1885, experiencing a rise in the growth rate of its GNP from 1.3 percent in the decade preceding the rise of protection to 3.1 percent in the subsequent decade. Sweden also experienced an acceleration of GNP growth around its policy shift toward more protection in 1888, while Italy experienced a slowdown in GNP growth around 1887, the year in which it went protectionist. In view of this evidence Bairoch noted that “it remains generally true that in all countries (except Italy) the introduction of protectionist measures resulted in a distinct acceleration in economic growth during the first ten years following a change in policy, and that this took place regardless of when the measures were introduced” (1993, 50).

O’Rourke (2000) examined more carefully the relationship between average tariffs and growth in the late nineteenth century. Estimating a growth equation with data for ten countries between 1875 and 1914, he found a positive effect of tariffs on the rate of growth of real income per capita, thereby confirming Bairoch’s argument. 26 Allowing for fixed country effects, his panel estimates imply that an increase of one standard deviation in the average tariff rate raised the annual growth rate by 0.74 percent.

Clemens and Williamson (2002) confirmed O’Rourke’s finding for a sample of more than thirty countries between 1870 and 1913. But they also found that the relationship was reversed in the post–World War II period. That is, in the postwar period high-tariff countries grew more slowly than low-tariff countries. Clemens and Williamson suggested that the reversal might be related to the average level of protection in the world economy. When a country’s trade partners have high tariffs, it can speed up its own growth by adopting a higher rate of protection. When a country’s trade partners have low tariffs, however, higher protection harms growth.

Figure 5.2 portrays the evolution of the average tariff rate of thirty-five countries from the late nineteenth century to the late twentieth century. 27 Tariffs were higher before World War I than after World War II, and they hit record levels between the wars. This intertemporal pattern of tariffs is at the heart of Clemens and Williamson’s explanation of the reversal of the relationship between protection and growth. Although they also provided econometric evidence in support of their hypothesis, note that in view of our theoretical discussion other interpretations of the evidence are possible as well.

The economies in the post–World War II period were very different from the economies in the late nineteenth century and the beginning of the twentieth century. In each of these eras there were important structural differences among countries. The structure of some countries could have produced a positive effect of tariffs on the growth of income per capita; the structure of others could have produced a negative effect. In each of these periods the econometric estimates measure the average response across countries of the growth rate to the rate of protection. Therefore we may interpret the evidence as stating that in the post–World War II period the channels through which protection hindered growth dominated, while in the late nineteenth century and the beginning of the twentieth century the channels through which protection promoted growth dominated. This is a reasonable interpretation, but it does not help to understand exactly what were the dominant channels of influence in each of these periods. An understanding of this issue requires studying the relationship between protection and growth conditional on the characteristics that affect the nature of this relationship.

Apart from these difficulties, the study of trade policies is also plagued by other hardships. Although in the late nineteenth century and the beginning of the twentieth century protection was predominantly in the form of tariffs, the nature of protection changed in the post–World War II period. As tariffs were reduced in the various negotiating rounds of the General Agreement on Tariffs and Trade (GATT), countries erected ever higher non-tariff barriers. 28 For this reason the average tariff rate displayed in Figure 5.2 does not provide an accurate measure of protection in the late twentieth century. This fact led scholars of the post-World War II period to use a variety of additional indicators as proxies for levels of protection. These indicators include measures of real exchange-rate distortions, the size of the black-market premium on foreign exchange, the fraction of imports covered by nontariff barriers, and various institutional features of the economic regimes. 29 Other scholars used outcome indicators—such as the deviation of trade volumes from the predictions of trade theory—to measure the restrictiveness of trade regimes. 30 They all found negative effects of trade restrictions on growth.

There are nagging problems with these studies. Many of the problems were discussed by Rodríguez and Rodrik (2000). Trade policies are not entirely exogenous, they are often highly correlated with other policies, and they are too complex to be adequately represented by a single index of trade restrictiveness. The Sachs-Warner index is a good example (see Sachs and Warner 1995). It is a binary index, which assigned the value 1 when an economy was deemed to be open and 0 when it was deemed to be closed. An economy was considered to be closed if its average tariff exceeded 40 percent, or nontariff barriers covered more than 40 percent of its imports, or it had a socialist economic system, or much of its exports were controlled by a state monopoly, or its black-market premium exceeded 20 percent during the 1970s or the 1980s.

The Sachs-Warner index was found to be positively correlated with the growth rate of income per capita. According to the estimates, countries that were open grew faster—at a rate of 2.44 percent per annum—than countries that were closed. This impact is large indeed. As Rodríguez and Rodrik showed, however, the Sachs-Warner index is dominated by the criteria applied to state monopolies and the black-market premium. At the same time, black-market premiums are highly correlated with other government policies. Black-market premiums tend to be high in countries with lax macroeconomic policies, tight capital and exchange controls, and high levels of corruption. For this reason the estimated impact of this index on growth may not properly isolate the effects of trade policies, but may rather reflect the broader impact of government policies on economic growth.

Wacziarg (2001) corroborated this hypothesis. He developed a simultaneous equations model that allowed him to estimate the impact of trade policies on growth via six different channels: the quality of macroeconomic policies, the size of government, price distortions, factor accumulation, technology transmission, and foreign direct investment. That is, he estimated the effects on growth of the variables representing the six channels, as well as the effects of trade policies on each one of these variables.31 Combining these estimates enabled him to assess the impact of trade policies on growth

In the first stage Wacziarg estimated the impact of average tariffs, the nontariff coverage ratios, and the timing of trade liberalization according to the Sachs-Warner index, on the trade shares. He also estimated this equation without the timing of trade-liberalization variables. He then used the predicted impact of the trade-policy variables on the trade share as a measure of the restrictiveness of trade policies, in order to estimate its impact on the variables representing the various channels of influence on economic growth. When he used the timing of trade liberalization based on the Sachs-Warner index, he found that 63 percent of the effects of trade policies on growth were through investment, with technology transmission and the quality of macroeconomic policies constituting the other two important channels of transmission. Without the timing variables of trade liberalization based on the Sachs-Warner index, the effects of trade policies through the quality of macroeconomic policies disappear. In terms of the overall impact, Wacziarg estimated that in this case an increase of one standard deviation in the restrictiveness of trade policies reduces the growth rate of income per capita by 0.264 percent annually, a significant impact.

My view is that despite the many difficulties that exist in the literature, it is fair to conclude that the evidence favors a negative effect of protection on rates of growth in the post–World War II period. Importantly, there is no real evidence of a positive link for this era. But I also share the view expressed by Rodríguez and Rodrik that “it might be productive to look for contingent relationships between trade policy and growth. Do trade restrictions operate differently in low- vs. high-income countries? In small vs. large countries? In countries with a comparative advantage in primary products vs. those with comparative advantage in manufactured goods?” (2000, 317). To this list of contingencies I would only add structural features, which have been found in theoretical models to affect the link between trade policies and growth.

Notas

26. O’Rourke’s sample consists of developed European and non-European countries: Australia, Canada, Denmark, France, Germany, Italy, Norway, Sweden, the United Kingdom, and the United States.

27. The countries are: Argentina, Australia, Austria, Brazil, Burma, Canada, Ceylon, Chile, China, Colombia, Cuba, Denmark, Egypt, France, Germany, Greece, India, Indonesia, Italy, Japan, Mexico, New Zealand, Norway, Peru, the Philippines, Portugal, Russia, Spain, Sweden, Thailand, Turkey, United Kingdom, United States, Uruguay, Yugoslavia (Serbia). I thank Michael Clemens and Jeffrey Williamson for the data for this figure.

28. Little progress was achieved in the Dillon Round of trade negotiations in 1960–1961, but tariffs were reduced by 35 percent in the Kennedy Round (1962–1967), by an additional 33 percent in the Tokyo Round (1973–1979), and they were further reduced to just a few percent on trade in manufactures in the Uruguay Round (1986–1994).

29. See, for example, Dollar (1992), Ben-David (1993), and Sachs and Warner (1995).

30. See Leamer (1988).

31. Apart from the accumulation variables, all other variables affect growth through TFP. Although some of Wacziarg’s variables closely approximate the desired channel of influence, such as the share of government consumption in GDP as a measure of the size of government, others do not. Particularly unsatisfactory is the use of manufactured exports in total merchandise exports as a proxy for technology transmission.

Bibliografía

Bairoch, Paul. 1993. Economics and World History. Chicago: University of Chicago Press.

Clemens, Michael, and Jeffrey G. Williamson. 2002. “Why Did the Tariff-Growth Correlation Reverse after 1950?” NBER Working Paper no. 9181.

O’Rourke, Kevin. 2000. “Tariffs and Growth in the Late 19th Century.” Economic Journal 110: 456–483.

Rodríguez, Francisco, and Dani Rodrik. 2000. “Trade Policy and Economic Growth: A Skeptic’s Guide to the Cross-National Evidence.” NBER Macroeconomic Annual 2000, vol. 15, 261–325.

Sachs, Jeffrey D., and Andrew Warner. 1995. “Economic Reform and the Process of Global Integration.” Brookings Papers on Economic Activity, vol. 1, 1–118.

Wacziarg, Romain. 2001. “Measuring the Dynamic Gains from Trade.” World Bank Economic Review 15: 393–429.