Globalization,Convergence,and History 281 Q 0.15 Full Sample Less North America Full Sample 0.1 0.05 0 1870 1890 1913 FIGURE 2 GDP PER WORKER-HOUR DISPERSION,1870-1913 Source:Table 2. implies that real wage gaps would still have persisted well into the present century even had the convergence not been interrupted:for example,wage gaps in 1940 would still have been half of the big 1870 gaps.Large initial gaps take a long time to erase,even when convergence is persistent.But it was not persistent:an anticonvergence regime intervened,which stopped convergence between 1914 and 1950.Figure 3 documents the interruption. The Great War produced real wage divergence,both the 1920s and the 1930s produced stability in real wage dispersion,and World War II
Globalization, Convergence, and History 281 0.15 Full Sample Less North America Full SampleX 0.1 0.05 0 1870 1890 1913 FIGURE 2 GDP PER WORKER-HOUR DISPERSION, 1870-1913 Source: Table 2. implies that real wage gaps would still have persisted well into the present century even had the convergence not been interrupted: for example, wage gaps in 1940 would still have been half of the big 1870 gaps. Large initial gaps take a long time to erase, even when convergence is persistent. But it was not persistent: an anticonvergence regime intervened, which stopped convergence between 1914 and 1950. Figure 3 documents the interruption. The Great War produced real wage divergence, both the 1920s and the 1930s produced stability in real wage dispersion, and World War II This content downloaded from 211.80.95.69 on Sun, 20 Oct 2013 19:57:15 PM All use subject to JSTOR Terms and Conditions
282 Williamson 0.35 0.3 Full Sample 0.25 C(17) 0.2 金 年C(16) c(月-e 9-0 Full Sample Less 0.15 North America C(14) AA A 0.1 d Full Sample Less C(13) North America and Iberia Q 0.05 0 1914 1919 1924 1929 1934 1939 FIGURE 3 REAL WAGE DISPERSION 1914-1939 Notes:C(14)and C(16)exclude Portugal,which enters in 1927,giving rise to C(15)and C(17). Source:Williamson,“Evolution,”table A2.2. produced more divergence.Figure 4 tells a similar,but less dramatic tale: GDP per worker-hour convergence slowed down sharply between 1913 and 1938.Once again,real wage dispersion exhibits more dramatic behavior than GDP per worker-hour,and the rest of this article will offer some explanations for the difference. There are instructive country performances hidden by these summary statistics,especially the big North American outliers,Canada and the United States,both of which bucked the convergence tide.As Gavin
282 WilliaTmon 0.35 0.3 - Full Sample 0.25 C(17) 0.2- C(16) 0.5 -> EFull Sample Less 0.15 North America C(14) Full Sample Less 0.1 C(13) North America and Iberia 0.05 - 1914 1919 1924 1929 1934 1939 FIGURE 3 REAL WAGE DISPERSION 1914-1939 Notes: C(14) and C(16) exclude Portugal, which enters in 1927, giving rise to C(15) and C(17). Source: Williamson, "Evolution," table A2.2. produced more divergence. Figure 4 tells a similar, but less dramatic tale: GDP per worker-hour convergence slowed down sharply between 1913 and 1938. Once again, real wage dispersion exhibits more dramatic behavior than GDP per worker-hour, and the rest of this article will offer some explanations for the difference. There are instructive country performances hidden by these summary statistics, especially the big North American outliers, Canada and the United States, both of which bucked the convergence tide. As Gavin This content downloaded from 211.80.95.69 on Sun, 20 Oct 2013 19:57:15 PM All use subject to JSTOR Terms and Conditions
Globalization,Convergence,and History 283 0.15 Full Sample 0.1 G Full Sample Less North America 0.05 1913 1929 1938 FIGURE 4 GDP PER WORKER-HOUR DISPERSION,1913-1938 Source:Table 2. Wright,Moses Abramovitz,and Paul David have argued,North America enjoyed a spectacular leap into industrial superiority after the early 1890s.5 The great leap forward is manifested by the rich North American New World improving its advantage over the poorer industrial Old World after 1890:real wages in the United States were 72 percent higher than in Britain in 1870.That wage advantage had diminished to 63 percent by 5 Abramovitz and David,“Convergence'";and Wright,,“Origins
Globalization, Convergence, and History 283 0.15- Full Sample 0.1 Full Sample Less North America 0.05 0 1913 1929 1938 FIGURE 4 GDP PER WORKER-HOUR DISPERSION, 1913-1938 Source: Table 2. Wright, Moses Abramovitz, and Paul David have argued, North America enjoyed a spectacular leap into industrial superiority after the early 1890s.5 The great leap forward is manifested by the rich North American New World improving its advantage over the poorer industrial Old World after 1890: real wages in the United States were 72 percent higher than in Britain in 1870. That wage advantage had diminished to 63 percent by S Abramovitz and David, "Convergence"; and Wright, "Origins." This content downloaded from 211.80.95.69 on Sun, 20 Oct 2013 19:57:15 PM All use subject to JSTOR Terms and Conditions
284 Williamson 1890,supporting convergence;but by 1913 the United States regained everything it had lost.Canada offers an even better example of North American resistance to convergence.Canada improved its real wage superiority from 48 percent above Britain in 1870 to 57 percent in 1900 and,riding the prairie wheat boom,to 123 percent in 1913. This deviant North American behavior tended to retard the rate of convergence in the late nineteenth century,and Figures 1 and 2 show just how much.The "full sample less North America"converges faster than the full sample itself.Indeed,although the former traces out an abrupt switch from real wage convergence to divergence around the turn of the century, the latter continues the post-1850 convergence at about the same rate (Figure 1).The GDP per worker-hour evidence suggests the same(Figure 2) Some might argue that this deviant behavior would be even more pronounced if the industrial North were treated separately.Maybe yes, maybe no.After all,regional inequality was not simply an American problem,for it applied with equal drama to European countries like Italy. Although I will stick to national definitions in what follows,I am aware of regional experience with convergence and divergence,and of an older literature that distinguished between regional“backwash,.”“polarization,” and"spread"effects.There is reason to believe that the globalization and convergence forces that operated at the national level also operated at the regional level,but I do not have the space here to pursue the issue in depth.? What about Europe?Given the great debate about Britain's loss of industrial leadership to her close competitors,most of us would look for evidence of,say,German catch-up on the leader.We would be looking in the wrong place.What matters far more to European convergence is the performance of poor countries around the European periphery.Over the thirty years following 1870,four of these poor countries dramatically improved their real wages relative to Britain:Denmark rose from 54 to 85 percent;Ireland,from 73 to 89 percent;Sweden,from 42 to 82 percent; and Norway,from 42 to 65 percent.Italy also made gains,but they were more modest and were centered in the North.The Iberians lost ground: Portugal fell from 48 to 42 percent of Britain,and Spain,from 76 to 48 percent. If convergence was relatively slow in Europe in the late nineteenth century,it was the rise in the historically persistent wage gap between the Latin south and the non-Latin north that accounts for it-and this in spite of so much attention to an alleged late Victorian and Edwardian failure in England.Late Victorian and Edwardian failure helps explain continued convergence in the north of Europe,but what dominated European Williamson,"Regional Inequality";Hirschman,Strategy;and Myrdal,Economic Theory. See Barro and Sala-i-Matin,"Convergence,"for a modern look by two macroeconomists
284 Williamson 1890, supporting convergence; but by 1913 the United States regained everything it had lost. Canada offers an even better example of North American resistance to convergence. Canada improved its real wage superiority from 48 percent above Britain in 1870 to 57 percent in 1900 and, riding the prairie wheat boom, to 123 percent in 1913. This deviant North American behavior tended to retard the rate of convergence in the late nineteenth century, and Figures 1 and 2 show just how much. The "full sample less North America" converges faster than the full sample itself. Indeed, although the former traces out an abrupt switch from real wage convergence to divergence around the turn of the century, the latter continues the post-1850 convergence at about the same rate (Figure 1). The GDP per worker-hour evidence suggests the same (Figure 2). Some might argue that this deviant behavior would be even more pronounced if the industrial North were treated separately. Maybe yes, maybe no. After all, regional inequality was not simply an American problem, for it applied with equal drama to European countries like Italy. Although I will stick to national definitions in what follows, I am aware of regional experience with convergence and divergence, and of an older literature that distinguished between regional "backwash,"polarization," and "spread" effects.6 There is reason to believe that the globalization and convergence forces that operated at the national level also operated at the regional level, but I do not have the space here to pursue the issue in depth.7 What about Europe? Given the great debate about Britain's loss of industrialeadership to her close competitors, most of us would look for evidence of, say, German catch-up on the leader. We would be looking in the wrong place. What matters far more to European convergence is the performance of poor countries around the European periphery. Over the thirty years following 1870, four of these poor countries dramatically improved their real wages relative to Britain: Denmark rose from 54 to 85 percent; Ireland, from 73 to 89 percent; Sweden, from 42 to 82 percent; and Norway, from 42 to 65 percent. Italy also made gains, but they were more modest and were centered in the North. The Iberians lost ground: Portugal fell from 48 to 42 percent of Britain, and Spain, from 76 to 48 percent. If convergence was relatively slow in Europe in the late nineteenth century, it was the rise in the historically persistent wage gap between the Latin south and the non-Latin north that accounts for it-and this in spite of so much attention to an alleged late Victorian and Edwardian failure in England. Late Victorian and Edwardian failure helps explain continued convergence in the north of Europe, but what dominated European 6Williamson, "Regional Inequality"; Hirschman, Strategy; and Myrdal, Economic Theory. 7 See Barro and Sala-i-Matin, "Convergence," for a modern look by two macroeconomists. This content downloaded from 211.80.95.69 on Sun, 20 Oct 2013 19:57:15 PM All use subject to JSTOR Terms and Conditions
Globalization,Convergence,and History 285 experience was not Britain's failure(which hastened convergence),but the failure of the Latin economies (which retarded convergence).Figure 1 shows this clearly:real wage convergence in the OECD club is consider- ably greater when the two Iberian countries are removed from the sample. Three countries illustrate the convergence best:Ireland,Sweden,and the United States.In 1854,real wages in Sweden were only 48 percent of those of Britain,whereas in 1913 they were at par,an impressive catch-up by any standard.In 1854,and shortly after the Famine,real wages in Ireland were only 60 percent of those of Britain,a figure that had hardly changed at all over the previous three decades.Real wages in Ireland started a dramatic convergence on those across the Irish Sea during the 1850s (and notably,in the absence of any Irish industrialization),so that they were 73 percent of Britain's by 1870.By 1913 they were 92 percent of Britain's.Ireland was transformed over this period of convergence from a poverty-stricken,peasant economy that had served as a source of elastic labor supply for Britain's booming cities to an economy at the start of the twentieth century that boasted urban wages close to those prevailing in English cities.Irish and Swedish wages even converged on those of the New World between 1854 and 1913:as a percentage of U.S.wages,Irish wages rose from 38 to 53 and Swedish wages rose from 24 to 53.8 Now,why do I think globalization accounts for most of this conver- gence? GLOBALIZATION IN COMMODITY MARKETS:THE FACTOR PRICE CONVERGENCE THEOREM AT WORK The factor price equalization theorem has been a durable tool for trade theorists ever since Eli Heckscher and Bertil Ohlin made their seminal contributions in 1919 and 1924,although it was convergence not equaliza- tion that held the interests of these two Swedes.The Heckscher-Ohlin paradigm argues that countries export commodities that use intensively the factors in which they are well endowed,whereas they import commod- ities that use intensively the factors in which they are poorly endowed.Let falling transport costs tend to equalize prices of the traded commodities, encouraging more trade.Countries will now export more of the goods that exploit their favorable factor endowment.The demand for the abundant and cheap factor booms,while that for the scarce and expensive factor falls.Thus,commodity price convergence tends to produce factor price convergence:for example,wages should rise in poor countries relative to 8As it turns out,the average wage gap between New World and Old drives a large share of the convergence over the half century from 1854 to 1913(Williamson,"Evolution"). Flam and Flanders,Heckscher-Ohlin Trade Theory.This section draws heavily on a recent collaboration with O'Rourke on late nineteenth-century Scandinavian catch-up ("Open Economy Forces"and "Education"),as well as earlier joint work on Anglo-America("Were Heckscher and Ohlin Right?")
Globalization, Convergence, and History 285 experience was not Britain's failure (which hastened convergence), but the failure of the Latin economies (which retarded convergence). Figure 1 shows this clearly: real wage convergence in the OECD club is considerably greater when the two Iberian countries are removed from the sample. Three countries illustrate the convergence best: Ireland, Sweden, and the United States. In 1854, real wages in Sweden were only 48 percent of those of Britain, whereas in 1913 they were at par, an impressive catch-up by any standard. In 1854, and shortly after the Famine, real wages in Ireland were only 60 percent of those of Britain, a figure that had hardly changed at all over the previous three decades. Real wages in Ireland started a dramatic convergence on those across the Irish Sea during the 1850s (and notably, in the absence of any Irish industrialization), so that they were 73 percent of Britain's by 1870. By 1913 they were 92 percent of Britain's. Ireland was transformed over this period of convergence from a poverty-stricken, peasant economy that had served as a source of elastic labor supply for Britain's booming cities to an economy at the start of the twentieth century that boasted urban wages close to those prevailing in English cities. Irish and Swedish wages even converged on those of the New World between 1854 and 1913: as a percentage of U.S. wages, Irish wages rose from 38 to 53 and Swedish wages rose from 24 to 53.8 Now, why do I think globalization accounts for most of this convergence? GLOBALIZATION IN COMMODITY MARKETS: THE FACTOR PRICE CONVERGENCE THEOREM AT WORK The factor price equalization theorem has been a durable tool for trade theorists ever since Eli Heckscher and Bertil Ohlin made their seminal contributions in 1919 and 1924, although it was convergence not equalization that held the interests of these two Swedes.9 The Heckscher-Ohlin paradigm argues that countries export commodities that use intensively the factors in which they are well endowed, whereas they import commodities that use intensively the factors in which they are poorly endowed. Let falling transport costs tend to equalize prices of the traded commodities, encouraging more trade. Countries will now export more of the goods that exploit their favorable factor endowment. The demand for the abundant and cheap factor booms, while that for the scarce and expensive factor falls. Thus, commodity price convergence tends to produce factor price convergence: for example, wages should rise in poor countries relative to 8 As it turns out, the average wage gap between New World and Old drives a large share of the convergence over the half century from 1854 to 1913 (Williamson, "Evolution"). 9 Flam and Flanders, Heckscher-Ohlin Trade Theory. This section draws heavily on a recent collaboration with O'Rourke on late nineteenth-century Scandinavian catch-up ("Open Economy Forces" and "Education"), as well as earlier joint work on Anglo-America ("Were Heckscher and Ohlin Right?"). This content downloaded from 211.80.95.69 on Sun, 20 Oct 2013 19:57:15 PM All use subject to JSTOR Terms and Conditions