برندگان و بازندگان در قرعه کشی کالا: تاثیر رشد و بی ثباتی قوانین و مقررات تجارت در کشورهای پیرامونی 1870-1939

Most countries in the periphery specialized in the export of just a handful of primary products for most of their history. Some of these commodities have been more price volatile than others, and those with more volatility have grown much more slowly relative to the industrial leaders and to other primary product exporters. This fact helps explain the growth puzzle noted by Easterly, Kremer, Pritchett and Summers more than a decade ago: that the contending fundamental determinants of growth—institutions, geography and culture—exhibit far more persistence than do the growth rates they are supposed to explain. Using a new panel database for 35 countries, this paper estimates the impact of terms of trade volatility and secular change on country performance between 1870 and 1939. Volatility was much more important for growth than was secular change and accounts for a substantial degree of the divergence in incomes within the sample of small, commodity-dependent ‘Periphery’ nations as well as under-performance of the Periphery as a whole relative to such nations as the US and Western Europe, or ‘Core’. One channel of impact seems to be the adverse effect of volatility on foreign investment. It appears that the terms of trade effects were asymmetric between Core and Periphery.

This essay explores an underappreciated aspect of long term growth: differences in price trends and volatility across primary commodities explain much of the global income divergence observed in the last century and a half. We show that most countries outside Western Europe and the US have been specialized in the export of the same handful of primary commodities for most of their history. Moreover, some commodity prices have proven more volatile than others, and some have enjoyed better secular growth. A look at commodity dependent price-taking economies (the Periphery) yields two main results. First, those countries with more volatile primary product prices grew more slowly relative to other primary product exporters as well as to the industrial leaders (or Core). Second, after controlling for volatility, rising terms of trade are associated with higher growth in Core countries, but not so in the price-taking Periphery. We establish these facts using a new database of commodities, prices, and incomes for 35 countries over eight decades—a coverage of more than 85% of the world population and nearly all of world GDP in 1914. Fig. 1 illustrates the simple bivariate relationship between income and terms of trade volatility over the entire period, 1870 to 1939. Examining the price-taking Periphery, simple OLS estimates will suggest that a one standard deviation increase in terms of trade volatility was associated with a 0.4 percentage point decrease in the rate of income growth per annum—a 40% decrease from the mean. This estimate seems to be a lower bound, with the upper bound a 66% decrease from the mean of income growth. Our results are extremely robust to alternative time periods, definitions of volatility, and exclusion of larger exporters where one might fear the terms of trade are less likely to be exogenous. Meanwhile, Fig. 2 illustrates the bivariate relation between income and terms of trade growth, suggesting a mild positive correlation. A multivariate treatment will show that after controlling for volatility income growth is in fact not robustly related to secular terms of trade changes in the Periphery. Full-size image (26 K) Fig. 1. 1939 GDP per capita and terms of trade volatility 1870–1939. Figure options Full-size image (23 K) Fig. 2. 1939 GDP per capita and mean terms of trade growth 1870–1939. Figure options We investigate one prime channel through which volatility could have impacted growth—foreign investment. Eichengreen (1996) argues that in much of the Periphery capital inflows shrank as price shocks reduced the attractiveness of investment. We attempt to quantify this response using the only source of investment data for the period in question—British capital flows from 1870 to 1913. While we do not find evidence that changes in terms of trade trends influenced capital flows, we do find evidence that volatility mattered. Fig. 3 illustrates this bivariate relationship graphically. In OLS regressions we find that a one standard deviation increase in terms of trade volatility was associated with a 33% to 58% decrease in average capital flows to the Periphery (although the results are not as robust as in the case of the income growth effects). While this impact is substantial, foreign investment flows were small relative to output in most countries, so it is perhaps only one of many channels of impact. If one assumes that domestic savings and investment respond in similar fashion to volatility, however, the investment channel would appear to be an important one. Full-size image (21 K) Fig. 3. Capital flows and terms of trade volatility. Figure options Finally, our results are consistent with asymmetry between the Core and Periphery. In contrast to the Periphery results reported here, elsewhere we show that income growth in Core countries bears little relationship to the volatility of the terms of trade, but is positively correlated with secular terms of trade growth (Blattman et al., 2004). We believe that the source of the asymmetry is diversification and industrialization in the Core. Countries with a broader base of exports, especially manufactures rising in value over time, can be expected to grow faster and be better insured against price fluctuations in individual commodities. Similarly, countries having governments with a broader tax base are less vulnerable to terms of trade, tariff revenue and project volatility, thus are capable of faster growth. Our results speak to three longstanding puzzles in the economic growth literature. First, although specialization in the production and export of primary commodities has proven to be one of the most robust determinants of poor economic growth, economists have yet to account fully for the power of commodity-specialization in predicting economic performance.1 We suggest that commodity price volatility is one reason why. Second, in spite of the explanatory power of commodity specialization, economic performance among the resource-rich before 1940 varied enormously. Some grew remarkably fast (e.g., Canada and Norway), some grew very slowly (e.g., Colombia and India), while some lay in between (e.g., Turkey and Brazil). We explain the variation by noting that all primary products were not created equal. Third, we seek to explain the volatility of growth rates over time using an equally volatile variable, commodity prices. As argued by Easterly, Kremer, Pritchett, and Summers (1993), the usual determinants of income growth—institutions, geography and culture—exhibit far more persistence than do the growth rates they are supposed to explain. In Table 1 we look at two time periods and compare the fraction of within-country variation to total variation for GDP growth and its common determinants. As noted by Pritchett (2000), growth within a country over time varies more than between countries, while the opposite is true of most explanatory variables. Therefore growth determinants that change over time, like commodity prices, are both intuitively and econometrically more satisfying. Table 1. Ratio of within-country variance to total variance for typical explanatory variables in growth regressions Variable 1870–1909 1960–1999 Growth of GDP per capita 0.85 0.73 Terms of trade growth 0.96 0.90 Population growth 0.37 0.29 Investment 0.31 0.28 Level of primary education 0.05 0.09 Primary products in exports 0.02 n.a. Number of countries 35 125 Data for 1960–1999 come from the Penn World Tables v6.1 (for GDP, population, and the investment rate), the Barro–Lee Data Set of International Measures of Schooling Years and Schooling Quality (for education), and the World Bank's 2004 World Development Indicators (for the terms of trade). Countries with fewer than 20 annual observations (or 4 quinquennial ones) were dropped from the analysis. Data for 1870–1939 are discussed in the data appendix. The investment figure is not investment rates (as in 1960–99) but rather is capital flows from the UK to the receiving country. Table options We conclude that two important determinants of growth—commodity price volatility and secular growth—have been overlooked in the contest between constitutions, cultures and coastlines. Our findings are reminiscent of what Carlos Diaz-Alejandro (1984) called the “commodity lottery”. Each country's exportable resources, he explained, were determined in large part by geography and chance, and differences in later economic development were a consequence of the economic, political and institutional attributes of each commodity. Differences in the secular growth of each primary product’s relative price mattered, but not, as we shall see, in the way Prebisch and Singer saw it. Furthermore, the exogenous price volatility of each primary product mattered even more by generating internal instability, reduced investment, and diminished economic growth.

Our analysis suggests that the terms of trade needs to be more central to our thinking about long term development. Commodity choice, dependence and the associated price trends and their volatility were especially powerful determinants of growth in the Periphery up to 1940. In focusing upon the forces that led to industrialization and diversification in Western Europe and the US, scholars may have forgotten that the rest of the world-rich and poor-took a different path. Some of the primary-product producing nations did quite well by this strategy—witness Canada, Australia, or Norway. Some, like India or Colombia, did quite poorly. Others were somewhere in the middle. The commodity-specialized were not uniformly slow-growing and poor. Neither were the prices of their commodities uniformly volatile and decreasing. In reconstructing nearly a century of terms of trade experience from 1870 to 1939 and assessing its impact on economic performance, we see that some commodities proved more volatile in price than others, and that those countries with more volatile terms of trade grew more slowly than other commodity-specialized nations. Countries with just one standard deviation higher volatility, moreover, grew on average more than half a percentage point per annum slower. The same difference in volatility seems to be associated with a third to two-thirds lower capital inflows. While our capital flows results are not large or as robust as our growth results, they still point to an important channel of impact. Other channels are likely to have been important too, such as the effect of terms of trade shocks on the incidence of civil conflict (Rodrik, 1999 and Miguel et al., 2004). The economic effects of our estimates are big. To illustrate the impact, consider an example from the primary product-exporting Periphery. Per capita income in Canada grew faster than in Indonesia by about 1% per annum. The difference in terms of trade volatility between the two countries was just under one half of one standard deviation. Our estimates imply that if, through good fortune, Indonesia had experienced the smaller terms of trade volatility of Canada, then Indonesia would have grown faster by about 0.3 percentage points, reducing the growth rate gap between them by a third. These magnitudes suggest that terms of trade fluctuations are a major force behind the big divergence in income levels among Periphery nations. Up to now, most development economists would explain the nineteenth and twentieth century take-off of Chile relative to Colombia in terms of geography, institutions, or culture. We argue that commodities probably mattered at least as much. By our estimates, producing and exporting a commodity with half the average level of volatility was associated with an annual increase in the rate of growth of a third to a half of a percentage point—a thirty to fifty percent improvement—as well as an increase in UK capital flows of roughly the same magnitude. What is also notable about our results is the asymmetry between Core and Periphery. Where terms of trade volatility was present in the Periphery, it created a significant drag on output growth. Elsewhere we show that this was not true of the Core—where it experienced the same high price volatility, it did not experience the same drag on growth. This asymmetry is consistent with two growth narratives. One, positive price shocks reinforced comparative advantage, and so they induced more industrialization in the Core and less in the Periphery. Two, it may be that rich countries with more sophisticated institutions and markets were better able to insure against price volatility than poor countries, so that terms of trade instability had a far bigger negative impact in the Periphery than in the Core. We await better data on institutional quality to test this hypothesis. In the meantime, our results are suggestive that terms of trade fluctuations were also a reason for divergence in income levels between the Core and Periphery. Volatility mattered little for the larger, diversified industrial nations, but volatility seems to have impacted the commodity-dependent nations adversely. The gap in growth rates in per capita income between Core and Periphery in our sample is 0.44 percentage points. If the Periphery had experienced the same volatility as the Core, ceteris paribus, this would have added 0.2 percentage points to average GDP per capita growth rates in the Periphery. This alone accounts for nearly half of the output per capita growth gap. Could countries in the Periphery have lessened the impact of volatile prices through judicious use of policy? Questions like this often arise in the resource curse literature. Given the adverse impact of resource-dependence, either through deindustrialization or price volatility, why did these countries not shift resources away from commodity production into something else? While we make no statements about the larger question raised by the resource curse literature, we believe resource-shifting was not really an option as a response to price volatility. As we argued earlier, the choice of commodities was better described as a “lottery” than as a policy choice. Nations seldom ‘switched’ commodities unless forced to do so. For example, Colombia only began as a producer of coffee in the late nineteenth century once cheap Indonesian exports undermined the Colombian tobacco export trade, forcing it to scramble for a decade to find a new product and a footing in world markets. Even if such a choice was possible, it would have been difficult to predict future price volatility with any degree of accuracy. Hence our finding for the period up to 1940 agrees with a view of the post-war period that good economic performance was as much the result of good luck as it was of good policy. More importantly still, we believe that commodities deserve at least as much attention as the current triumvirate of cultures, coastlines and constitutions.