economic article

economic article

read the following paper and answer the following questions

During the 19th century there was a dramatic reductions in transportation costs. One consequence of this was increased trade in grains, the so-called “grain-invasion”,
from land-abundant Americas to labor-abundant Europe. O’Rourke et al. (1996)1 examine factor-price convergence between these countries in the late 19th century. Answer
the following questions related to the paper.

a) In a 2x2x2 Heckscher-Ohlin world in which land and labor are the two factors of production and agriculture and manufacturing are the two goods, explain what you
would predict for initial (high trade cost) factor price differences across Europe and the Americas. Make sure to explain relative factor abundances between countries
and factor intensities across industries.

b) In what ways would you expect factor prices to evolve as trade costs reductions allow the two regions to become more integrated (think of this as a gradual move
from autarky to free trade)?

c) Does the data support these predictions (Figures 1-3)? Why would you expect the effect on factor prices to be weaker in European countries that raised tariff
barriers?

File #1
Institute of Social and Economic Research, Osaka University
Factor Price Convergence in the Late Nineteenth Century
Author(s): Kevin H. O’Rourke, Alan M. Taylor and Jeffrey G. Williamson
Source: International Economic Review, Vol. 37, No. 3 (Aug., 1996), pp. 499-530
Published by: Wiley for the Economics Department of the University of Pennsylvania and
Institute of Social and Economic Research, Osaka University
Stable URL: http://www.jstor.org/stable/2527439
Accessed: 23-03-2017 23:55 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Wiley, Institute of Social and Economic Research, Osaka University are collaborating with
JSTOR to digitize, preserve and extend access to International Economic Review
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
INTERNATIONAL
ECONOMIC
August 1996
REVIEW Vol. 37, No. 3
FACTOR PRICE CONVERGENCE IN THE LATE
NINETEENTH CENTURY*
BY KEVIN H. O’ROURKE, ALAN M. TAYLOR,
AND JEFFREY G. WILLIAMSON1
We examine a dramatic historical episode of factor price convergence in the
late nineteenth century. Our focus is convergence between Old World and
New, and the analysis centers on land and labor. Wage-rental ratios boomed in
the Old World and collapsed in the New, moving the resource-rich, labor-scarce
New World closer to the resource-scarce, labor-abundant Old World. We use
econometrics and simulations to identify pro-convergence forces which include
commodity price convergence, factor accumulation, and factor-saving biases.
The results confirm that open-economy characteristics and international mar-
ket integration are important sources of convergence.
1. FACTOR PRICES AND CONVERGENCE IN THE LONG RUN
Today’s journalists fill the media with references to the global economic village,
politicians talk in terms of world competitiveness, and Americans fear the loss of
productivity and living-standard leadership. We often forget that this process of
global economic integration and convergence has a very long history. Under the
leadership of William Baumol (Baumol, Blackman, and Wolff 1989), Robert Barro
(1991, Barro and Sala-i-Martin 1991) and many others, the literature on post-World
War II economic convergence has reached enormous proportions, but few economists
in this new tradition pay serious attention to history. This seems surprising: after all,
the literature was initiated by economic historians like Alexander Gerschenkron
* Manuscript received October 1994; revised July 1995.
A number of scholars have helped us construct the international database on land values used
in this paper. These have our grateful thanks: Juan Carmona, Roberto Cortes Conde, Antoni
Estevadeordal, Giovanni Federico, Barry Howarth, Peter Lindert, Ian McLean, David Pope,
Leandro Prados, Graeme Snooks, and Vera Zamagni. We are grateful for the research assistance of
Kimiko Cautero and Boris Simkovich. In addition, we acknowledge the helpful comments of Cormac
6 Grada, Don Davis and participants at the European Historical Economics Society Conference on
Market Integration (Lerici, Italy, April 1993), the NBER Workshop on Macroeconomic History
(April 1993) and the Harvard Workshop in Economic History (April 1993). We also gratefully
acknowledge the financial support of the National Science Foundation SES-9021951 and
SBR-9223002.
499
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
500 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
(1952) and Moses Abramovitz (1986) with a keen view of the long run. Furthermore,
few economists pay much attention to the role which international commodity-,
labor-, and capital-market integration has played in the process. In contrast,
economic historians of the late nineteenth century pay considerable attention to
trade, capital flows, and migration, but pay little attention to their impact on
convergence. This paper bridges the gap.
When measured by GDP per capita, labor productivity per worker-hour, or real
wages, currently-industrialized nations have converged on one another, at least since
1870 (Maddison 1982; Abramovitz 1986; Baumol, Blackman, and Wolff 1989; De
Long 1988; Williamson 1995). However, the convergence did not take place without
interruption: dramatic convergence took place from 1870 to 1913 as international
trade boomed, capital flows became enormous, and international migrations rose to
levels large enough to be called “mass”; long-run convergence slowed down and
eventually ceased during and between the World Wars while world commodity trade
and capital markets collapsed and international migrations slowed to a trickle in the
face of quotas and a Great Depression; and convergence resumed after World War
II while international trade and capital flows gradually regained pre-World War I
levels of integration, and guest workers and illegal immigration pushed migrations
back towards pre-quota levels.
A central question underlying this convergence experience is: What role did
international migrations, capital flows, and commodity trade play? True, all the
currently industrialized countries underwent different experiences with human capi-
tal accumulation and technical progress, but most shared something in
common-their integration into world factor and commodity markets. To under-
stand the contribution of factor- and commodity-market integration to the conver-
gence process, we need far better evidence than simply GDP per capita and labor
productivity. In addition, we need evidence on the wages of common labor, the rents
on land, the returns to capital, and the premia on skills. That is, we need evidence
to better document factor-price convergence.
Why should factor-price convergence matter for the broader convergence debate,
a debate usually concerned with the convergence properties of an aggregate indica-
tor like GDP per worker? It matters crucially, since aggregates like GDP are, by
definition, a collection of endowments weighted by factor prices. In a more disag-
gregative view, the absolute convergence in the levels of variables like GDP per
worker or real wages depends on the relative convergence of the whole set of factor
prices. For example, let Y be GDP, P the price level of GDP, vi the endowment of
factor i (where VL = L, the endowment of homogeneous labor), and let wi be the
price of factor i (where WL = W, the wage of labor). The factor-income definition of
GDP implies that PY= Eiwivi. By rearrangement of this identity,
(1) i
(1) ~~~~~~L P ( i L WL vL)
Thus, the convergence of absolute labor-productivity (Y/L), so commonly studied in
the literature, may be accounted for by absolute convergence in real wages (W/P),
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 501
or by relative convergence in factor endowments (Vi/VL) or factor prices (wi/wL).2
This decomposition has a natural appeal when considering the process of conver-
gence in the greater Atlantic economy of the late nineteenth century. Mass migra-
tions were the result of global labor-market integration which brought about
absolute real wage convergence. Meanwhile, international capital and labor mobility
entailed both factors chasing abundant resources at the New World frontier, an
international factor reallocation which raised labor-land and capital-land ratios in
the New World toward those in the Old World-that is, it caused relative factor
endowment convergence, and, ceteris paribus, a convergence in relative factor prices
associated with the leveling of factor scarcities across countries (Williamson 1995;
O’Rourke and Williamson 1994; Taylor and Williamson 1994).
This paper uses standard trade-theoretic approaches to understand the determi-
nants of factor price convergence. The period of interest is the late nineteenth
century, when economic convergence among the current OECD countries (and in
the “greater Atlantic economy”) was dramatic; the focus is on the convergence
between Old World and New, by far the biggest participants in the global conver-
gence during the period; and the analysis centers on land and labor, the two most
important factors of production in the nineteenth century. Section 2 establishes the
facts: wage-rental ratios boomed in the Old World and collapsed in the New,
moving the resource-rich and labor-scarce New World closer to the resource-scarce
and labor-abundant Old World. Section 3 confronts theory: What did Ricardo,
Malthus and the frontier-staple histories have to say about the contribution of labor
and capital transfers to wage-rental convergence? What did Heckscher and Ohlin
have to say about the contribution of commodity-price equalization to the wage-
rental convergence? The remainder of the paper brings evidence to bear on theory:
Section 4 attacks the problem with econometrics and Section 5 complements the
argument with simulations using applied computable general-equilibrium models.
We conclude with an assessment and a research agenda.
2. THE WAGE-RENTAL RATIO CONVERGENCE FACTS
We all know that farm land was abundant and cheap in the New World while
scarce and expensive in the Old World. And we all know that labor was scarce and
expensive in the New World while abundant and cheap in the Old World. Thus, we
know that the wage-rental ratio was high in the New World and low in the Old.
What we don’t know is how big the gap was between the two and how the gap
evolved over time. The latter is resolved by the evidence on the ratio of wages to
land values in Figures 1-3; the former is unlikely ever to be resolved-while we can
say something about wages for comparable work and comparable workers across
countries (Williamson 1995), we cannot say much about rent on comparable land
2 In a standard profit-maximizing framework with constant returns to scale, we expect
wA(v1, v2, . . . ) to be homogeneous of degree zero. We may then write Wi/WL =
WiIWz ( V IVL, U /V2 .V … ).
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
502 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
700
600
500 U
400 / \%//
300
200′
100
1870 1875 1880 1885 1890 1895 1900 1905 1910 1915
–. A rgentina — Australia –C anada 0 USA
FIGURE 1
RATIO OF WAGES TO LAND VALUES 1870-1910, NEW WORLD COUNTRIES (1911 = 100)
140
120
100
1870 1875 1880 1885 1890 1895 1900 1905 1910 1915
-| n- Britain -De-Dnmark -Ireland -0–wSeden
FIGURE 2
RATIO OF WAGES TO LAND VALUES 1870-1910, OLD WORLD “FREE TRADE” COUNTRIES (1911 = 100)
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 503
since the latter varied so much in quality (a point of which Heckscher was well
aware: Flam and Flanders 1991, p. 48).
The figures draw on a late-nineteenth century panel data-set we have constructed
documenting wage-rental convergence among eleven countries. Four New World
countries are plotted in Figure 1-Argentina, Australia, Canada and the U.S.A.;
four “free trade” Old World countries in Figure 2-Denmark, Great Britain,
Ireland and Sweden; and three “protectionist” Old World countries in Figure
3-France, Germany and Spain. The data is fully documented in the Appendix. A
word is in order concerning the Old World labels on Figures 2 and 3, “free trade”
and “protectionist.” The impact of the invasion of New World grains on Old World
wage-rental ratios must have been muted where tariffs were raised in defense. As
Charles Kindleberger (1951) pointed out long ago, and as the new theories of
endogenous tariffs predict, the response was especially strong on the continent. But
comparative measures of late nineteenth century protection are hard to construct.
Ideally, we’d like effective rates of protection-especially to distinguish their impact
on grain producers and feedgrain-using livestock producers-but these are only
available for a few years and for a few countries. Alternatively, tariff rates disaggre-
gated by sector would be helpful, but even the classic study by Liepmann (1938)
suffers serious flaws (Tracy 1989, pp. 22-23) and, in any case, excludes six of our
eleven countries. We have no choice, therefore, but to fall back on the crude
measures of protection offered by Bairoch (1989). Based on his evidence for 1913,
reproduced in Table 1, the “protectionist” label applied to France, Germany and
Spain, and the “free trade” label applied to Britain, Denmark and Ireland. Sweden
3As the Appendix points out, we also have no choice but to use land values as a proxy for land
rents for the late nineteenth century. While we are hardly the first to do so, the underlying
assumptions linking the two should be made explicit. If land is an economic asset with infinite life,
and if the land markets of that time simply projected current rents into the future, and if global
financial markets were well enough integrated so that interest rates were pretty much the same
everywhere across our eleven countries, then land values should serve as an effective proxy for land
rents. However, there are reasons to suspect that these assumptions may not hold for the late
nineteenth century. First, Offer (1991) has recently argued that land was a “positional” asset in
Britain, offering social status independent of economic value. If so, land values would have been
higher than that based simply on rents. More important to the issue of factor-price convergence,
Offer asserts that the positional value of British farm land probably declined in the late nineteenth
century. If so, the fall in British land values would have been steeper than the fall in land rents, in
which case the rise in Britain’s wage-rental ratio in Figure 2 is exaggerated. There is no evidence
available to test Offer’s plausible assertion, and, more importantly, we do not know whether
positional value had greater or lesser influence in Britain than elsewhere. Second, naive projections
of current rents into the future were unlikely to have characterized land markets in the late
nineteenth century, but we have no evidence documenting the impact of speculative behavior on
relative trends in wage-rental ratios. Third, while global financial markets linking northwest Europe
to the New World were well integrated by the 1890s (Zevin 1992; Obstfeld 1995), they may have
been less so vis-a-vis the periphery (like Spain). Furthermore, global financial markets became better
integrated as the late nineteenth century progressed, lowering interest rates among New World
capital-importers relative to Old World capital-exporters. This implies that the wage-rental ratio
convergence is exaggerated by our use of land values. Our assumptions therefore serve to overstate
the case for wage-rental ratio convergence, but we have no way of knowing how much. Our
intuition is that the spectacular late nineteenth century wage-rental ratio convergence documented
in Figures 1-3 would persist if properly measured by land rents.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
504 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
140
120
100
80
60 0_sa/_\
40 .
1870 1875 1880 1885 1890 1895 1900 1905 1910 1915 | * France — Germany – Spain (1907= 100)|
FIGURE 3
RATIO OF WAGES TO LAND VALUES 1870-1910, OLD WORLD “PROTECTED”COUNTRIES (1911 = 100)
TABLE 1
OLD WORLD IMPORT TARIFF LEVELS IN 1913*
Average level of duties (%)
Country Manufactures Wheat
Austria-Hungary 20 35
Belgium 9 0
Denmark 0
Finland 28 0
France 21 38
Germany 13 36
Italy 20 40
Netherlands 0
Norway 4
Portugal prohibitive
Spain 34 43
Sweden 25 28
Switzerland 8 2
United Kingdom 0 0
* Source: Bairoch (1989).
lay somewhere in between, but since protectionist policy was implemented there
relatively late in the period, we throw Sweden into the “free trade” group. While
these categories could be, and have been, debated (Nye 1991, Irwin 1993, O’Rourke
forthcoming), they serve well enough to motivate what follows.
What does the evidence reveal? Relative factor-price convergence certainly char-
acterized these four decades (Figures 1-3, Table 2). In the New World, the
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 505
TABLE 2
TRENDS IN THE RATIO OF WAGES TO LAND VALUES, 1870-1910 (1901 100)*t
Country 1870 1890 1910
Old World, Free Trade
Britain 42.28 84.99 115.42
Denmark 32.89 62.06 101.09
Ireland 12.61 66.86 70.31
Sweden 41.41 70.02 108.88
Average 32.30 70.98 98.93
Old World, Protected
France 59.97 112.97 122.36
Germany 67.51 86.47 95.57
Spain 102.55 123.21 67.52
Average 76.68 107.55 95.15
New World
Argentina 167.58 106.45 31.95
Australia 289.74 118.54 75.64
United States 127.99 103.23 64.07
Average 195.10 109.41 57.22
Ratio of New World to:
Old World, Free Trade 6.04 1.54 0.58
Old World, Protected 2.54 1.02 0.60
* Index numbers are not comparable across countries.
tSource: Predicted values from regressions run on time and time squared, from time series
underlying Figures 1-3. Group averages are unweighted. New World excludes Canada since the
latter has data only for 1901-1911.
wage-rental ratio plunged. By 1913, the Australian ratio had fallen to one-quarter
of its 1870 level, the Argentine ratio had fallen to one-fifth of its mid-1880 level, and
the U.S.A. ratio had fallen to less than half of its 1870 level. In the Old World, the
wage-rental ratio surged. According to the trend values in Table 2, the British ratio
in 1910 had increased by a factor of 2.7 over its 1870 level, while the Irish ratio had
increased even more, by a factor of 5.5. The Swedish and Danish ratios had both
increased by a factor of 2.3. The surge was less pronounced in the “protectionist”
than in the “free trade” group. The ratio had increased by a factor of 1.8 in France,
1.4 in Germany, and not at all in Spain. The last two lines of Table 2 summarize
wage-rental trends in the New World relative to the Old for 1870 through 1910.
They must, however, be treated with caution: they are both based on indices
1901 = 100, but we are not sure that the underlying wage-rental ratios refer to
quality-comparable units of land in the denominator. Subject to that word of
caution, one index drops by a factor of ten, from about 6 to about 0.6; and the other
by a factor of four, from about 2.5 to about 0.6.
3. THEORY
What explains this impressive relative factor-price convergence?
First, one might appeal to the discovery and exploitation of land and resources at
the open frontiers in the New World. As Lindert (1988) has reminded us, classical
theories of pre-industrial performance argued that population growth would cause
the relative price of land to rise as long as land scarcity did not choke off that
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
506 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
growth. Certainly Malthus saw it that way: demographic events pushed up man-land
ratios, lowered real wages, raised land rents, and caused the wage-rental ratio to
fall to a new long-run equilibrium where zero population growth was restored.
Ricardo and John Stuart Mill told similar, but more explicit, stories. Yet what of the
“safety valve” of emigration? Since Europeans moved to New World land in massive
numbers in the late nineteenth century, wage-rental ratios should have fallen by
much more in the New World than in Europe, a Mill-Ricardo argument that should
help account for the opposing factor-price trends in these two parts of the world.
Latin American scholars have argued thus to explain the late nineteenth century fall
in wage-rental ratios there.4 In the Australian literature, Sinclair (1976, see espe-
cially p. 5) sketched a model based explicitly on relative resource abundance and
international factor transfer between core and periphery. The paradigm, dating back
to Mill (1848), places emphasis on international factor reallocation as the mechanics
by which a fundamental disequilibrium is resolved: capital and labor chase higher
returns (and each other) migrating from the Old World to the New.5
Second, one could explore accumulation forces. To the extent that physical and
human capital (skills) are used more intensively in industry than agriculture, rapid
accumulation (associated with the industrial revolution) should favor the relative
expansion of industry and the relative demand for labor, thus raising the wage-
rental ratio. Together, accumulation and factor-mobility effects such as those just
mentioned play a key role in today’s multi-sector trade models. Endowments (or
factor intensities)-whether measured by land labor ratios or capital-labor
ratios-constitute a key exogenous force in all variants of the Ricardo-Viner
(sector-specific-factor) and Heckscher-Ohlin (mobile-factor) models of trade.
Third, one might invoke other (non-endowment) economic forces associated with
the industrial revolution. Industrial revolutions typically embody productivity growth
which favors industry, even when one takes account of the fact that such unbalanced
4 Diaz-Alejandro (1970) wrote that the labor supply in Argentina was highly elastic (p. 23): he
thought that the labor supply before 1930 was “perfectly elastic at the going wage rate (plus some
differential) in the industrial centers of Italy and Spain, the main sources of emigration to
Argentina” (pp. 21-22). However, recent evidence on Argentine immigration (Taylor 1994) and
European emigration (Hatton and Williamson 1994) is inconsistent with this thesis. Certainly, elastic
labor supplies have crept into the language of Latin American economic history (e.g., Reynolds
1985, p. 87). Given a long-run boom in the relative price of export staples, elastic immigrant-aug-
mented labor supplies combined with less elastic land supplies in the Pampas is certainly consistent
with falling wage-rental ratios in Argentina (along lines suggested by Harley 1986). Leff (1992, p. 6)
believes the same was true of Brazil and that elastic labor supplies accounted for falling wage-rental
ratios in the Sdo Paulo and Santos areas from the 1880s onwards: “The similarities between Brazil’s
historical experience in the nineteenth century and W. A. Lewis’ celebrated model .., are evident.”
5 Sinclair’s model drew its inspiration from Butlin (1964). Much of existing Australian literature
follows in this Mill-Ricardo tradition. While McLean (1990, p. 8) offers a reference to commodity
trade by acknowledging the impact of the decline in long-distance freight rates, commodity-price
convergence and the terms of trade generally receive little emphasis by Australian historians. The
Mill-Ricardo staple model has, of course, been applied to other New World examples. Innis (1927)
used it effectively for Canada, and it has found a welcome home in writings on other settler
economies (Green and Urquhart 1976, Schedvin 1990).
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 507
productivity advance tends to lower the relative price of industrial goods.6 Since
industrial output makes little use of farmland, industrialization tends to be land-sav-
ing, raising instead the relative demands for labor and capital. Such industrial
revolutionary events should, therefore, tend to raise the wage-rental ratio. Accord-
ing to this prediction, more rapid industrialization in Europe relative to the New
World should also have served to raise the wage-rental ratio by more in Europe.
Such events should have contributed to factor-price convergence, including the rise
of real wages in the labor-abundant Old World relative to the labor-scarce New
World. This prediction would be reinforced if productivity advance in the late
nineteenth century New World was labor-saving and land-using, as an induced-
innovation hypothesis would suggest (Hayami and Ruttan 1971), and as economic
historians generally believe (Habakkuk 1962; David 1974; Williamson and Lindert
1980; Di Tella 1982). The prediction would be further reinforced if productivity
advance in the Old World was land-saving and labor-using, as we also generally
believe.7
Fourth, one might have recourse to trade theory, where the factor-price-equaliza-
tion theorem has been a durable tool for seventy years.8 The Heckscher-Ohlin
paradigm has it that countries tend to export commodities using intensively the
factors with which they are well endowed while they tend to import commodities
using intensively the factors with which they are poorly endowed. Should falling
transport costs (unimpeded by any protective reaction from importing countries)
tend to equalize prices of traded commodities, then countries will tend to export
more of the goods which exploit their favorable factor endowment; the demand for
the abundant and cheap factor booms while that for the scarce and expensive factor
slumps. Thus, commodity-price convergence tends to produce factor-price conver-
gence, although theory is ambiguous about how much.9 Moreover, as a classic article
by Mundell (1957) argued, countries attempting to insulate themselves from such
forces via protection might stimulate additional capital and labor flows, which would
have the same ultimate effect.10 When Heckscher was writing in 1919 and Ohlin in
1924, they were motivated by the commodity-price equalization trends which they
thought had taken place between the Old World and the New in the late nineteenth
6 The supply-side forces listed above tended (on net) to favor the relative expansion of industry
and thus to lower its price relative to agriculture. Demand-side forces are likely to have muted those
effects. That certainly seems to be the implication of Engel effects, demand-side forces which
appeal to the relatively high income elasticity of demand for industrial goods compared with
agricultural goods. Such demand-side forces should, by themselves, have tended to lower the
relative price of agricultural goods world-wide.
7The intertemporal application of the induced-innovation hypothesis reappears in cross-sectional
accounts of trade patterns in the 1980s (Trefler 1993).
8 The remainder of this section draws on O’Rourke and Williamson (1994).
9Given the attention which trade theory has paid to the factor-price equalization theorem, it is
surprising that a recent survey by Rassekh and Thompson (1993, pp. 11-12) could report that the
theorem had received no empirical attention prior to the 1980s.
10Although it should be noted that in the context of the late nineteenth century, when capital
and labor both chased New World land, pushing back the frontier in the process, trade and factor
flows may actually have been complements rather than substitutes.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
508 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
century. Their economic metaphor was driven by foodstuffs: what economic histori-
ans now call the invasion of grains from the New World, an inflow driven by the
sharp decline in transport costs that served to raise the relative price of grains in the
New World relative to the Old. What occurred in the late nineteenth century was
exactly the kind of exogenous relative price shock which is supposed to set factor-
price equalization in motion. According to the theorem, the invasion of grains
should have raised the wage-rental ratio in the Old World while lowering it in the
New World, ceteris paribus. In Ohlin’s words, “trade increased] the price of land in
Australia and lowere[d] it in Europe, while tending to keep wages down in Australia
and up in Europe” (Flam and Flanders 1991, pp. 91-92). Did it?
This list of potential sources of factor price convergence provokes several ques-
tions. We can think of three key issues. Were factor endowments major deter-
minants of trade and factor prices in the late nineteenth century? Was there
pronounced commodity-price convergence in the late nineteenth century? If the first
two propositions hold, did factor-endowment and commodity-price trends make a
significant contribution to the observed factor price (wage-rental) convergence? The
last question is the main focus of our paper, but answers to the first two questions
certainly bear on the validity of our approach.
Consider the first question. Several recent papers have analyzed the determinants
of comparative advantage in British and American manufacturing in the late
nineteenth century. Crafts and Thomas (1986) found that endowments explain well
the pattern of trade in British manufacturing between 1910 and 1935, as well as the
U.S.A. in 1909. Wright (1990) found the same in accounting for the evolution of
U.S. trade patterns between 1879 and 1940, a result reinforced more recently by
Nelson and Wright (1992). Estevadeordal (1993) has found more support based on a
large sample of 18 countries around 1913. Indeed, the 1913 evidence is far more
supportive of the hypothesis than Leamer (1984) was able to report using post-
World-War-II data.11
Consider the second question. Economic historians have long been aware of the
revolutionary decline in transport costs underlying overseas trade in the late
nineteenth century. North (1958, p. 537) called the decline “radical” both for
railroads and ocean shipping. Since Europe imported foodstuffs and raw materials,
and since these bulk commodities “were fundamental beneficiaries of the cheapen-
ing transport costs” (p. 544), North thought it was clear that it contributed in Europe
to “lower priced foodstuffs and therefore rising real wages, and to lowering in the
cost of industrial raw materials,” (p. 545) and therefore, we take it, rising rates of
industrialization. Although North didn’t say so, symmetry suggests that the wage-
rental ratio must have been lowered in the U.S. while industrialization must have
been suppressed, ceteris paribus. When deflated by a general price index, North’s
11 Also note that Whitney (1968) found no evidence of a Leontief paradox in the United States
using 1899 data. Leader (1980) has argued, of course, that there was never a paradox even in
Leontief’s data.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 509
two freight rate indices dropped from 41 to 53 percent between 1870 and 1910.
Similar evidence has been offered more recently by Harley (1988).12
The answers to the first two questions, then, are broadly in the affirmative. What
remains is the third question. How much of the late nineteenth century wage-rental
convergence documented in Figures 1-3 can be accounted for by commodity-price
convergence (Heckscher-Ohlin factor-price equalization effects), how much by trends
in land-labor and capital-labor ratios, and how much by other forces like productiv-
ity bias?
4. AN ECONOMETRIC APPROACH
Insights from the Heckscher-Ohlin and Ricardo-Viner open-economy models
concerning the determinants of wage-rental-ratio convergence across the late
nineteenth century can be applied by estimating factor-price equations. For exam-
ple, consider any multi-sector model in a standard open-economy revenue-maximiz-
ing framework (cf. Dixit and Norman 1980, ch. 2). With constant returns to scale, we
would expect general factor-price functions of the form
(2) wj = ajfj(v>,pi), with fj homogeneous degree 0 in v;, degree 1 in pi,
where the v; and wj are vectors representing the endowments and prices of factors
j, pi is a vector of prices of (traded) final goods i, and aj represents a factor-specific
shift term for each j. Given factor endowments relative to a base factor endowment
v Ivo and final good prices relative to a base final good price pi/po, it is then
straightforward to derive a general relative factor price function of the form
(3) (wjl/w,) = (a,1/a, )gj( vj/vo, Pi/Po),
with gjj, homogeneous degree 0 in v; and in pi.
Our estimation relies on such a model, where (wj/wj) is the ratio of the wage to
the value of land. In our interpretation, (aj/aj) represents factor-saving effects
arising from productivity bias, for example. The arguments v1/vo include land-labor
and capital-labor ratios, and pi/po measures the terms of trade between agricul-
tural and manufacturing goods. Formally, given the data available, we adopted a
12 In assessing the radical decline in overseas freight rates and the cost reductions along the rails,
what mattered, of course, was its impact on the price convergence of tradables: while Liverpool
grain prices exceeded Chicago prices by 60 percent in 1870, the spread was only 14 percent in 1912
(O’Rourke and Williamson 1994). The price gap for meat and animal fats declined from 93 percent
to 18 percent over the same period, the gap for iron products fell from 80 to 20 percent, for cotton
textiles from 14 to 1 percent, and so on. Clearly there was dramatic price convergence in global
commodity markets between 1870 and World War I, especially among those Old World countries
who chose not to raise tariffs on New World farm products.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
510 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
specification of the form
(4) WGRENTit = P0i + p1 LANDLABit + 02 CAPLABit
+ p3i PAPMit + 34 PRODit,
where for each country i, in period t, we define the variables in natural logarithms:
WGRENTit = log of the wage-rental ratio (nominal wage divided by nominal value
of land);
LANDLABit = log of the land-labor ratio (quantity of land divided by labor force);
CAPLABit = log of the capital-labor ratio (capital stock divided by labor force);
PAPMUt = log of the terms of trade (agricultural goods price divided by manu-
facturing goods price);
PRODit = a Solovian residual (log of output per worker minus 0.4 times
CAPLAB minus 0.1 times LANDLAB).13
Since relative factor prices appear on the left-hand side of equation (4), relative
commodity prices and relative factor endowments (land-labor and capital-labor
ratios) are appropriate for the right-hand side. We have been able to augment our
1870-1914 wage-rental ratio database to include factor endowments, relative com-
modity prices and productivity for all but four of our eleven countries (see Ap-
pendix; data scarcity forces us to exclude Argentina, Canada, Ireland and Spain).
Such data are not entirely comparable internationally (for example, we have no way
to correct for varying land quality), so an index-number interpretation must be given
to each series, and the estimation of (4) in log-levels must include fixed-effect
coefficients to mop up the proportionality constants for each country.
Given the stress which contemporaries, historians, Heckscher and Ohlin have
placed on commodity prices, it might be useful to say a few words about PAPM.
According to that literature, the link between the relative price of tradable agricul-
tural goods (principally, grains) and the wage-rental ratio was often broken by two
events. First, tariffs often raised domestic farm prices far above world prices,
breaking down the influence of commodity-price convergence. Second, relatively
non-tradable farm products (like meat, butter and vegetables) increased their share
of farm output while undergoing a less dramatic price convergence, and thus the
output switch often served to mute the impact of the more tradable farm-commodity
(grain) price convergence. PAPM should include both influences: the impact of
tariffs should be embedded in the relative price and so too should the impact of any
switch in farm output away from grains. However, we are not sure that the imperfect
country PAPM time series in our database always obey these rules. One way to find
13 That is, we take land’s share as 10% of income, capital’s share as 40%, and labor’s share
implicitly as 50%. These shares are not unreasonable for the “greater Atlantic economy” of the late
nineteenth century. See Taylor and Williamson (1994), where evidence on factor shares is discussed.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 511
TABLE 3
THE DETERMINANTS OF THE WAGE-RENTAL RATIO IN THE NEW AND OLD WORLDS, 1875-1914
Regression 1 2 3
Sample ALL NEWWORLD OLDWORLD
LANDLAB 1.09** 1.16** 0.77**
(6.88) (11.39) (3.53)
CAPLAB 1.26** 1.19** 0.83**
(5.37) (3.43) (3.17)
PROD 0.71** -0.85** 1.05**
(3.66) (3.60) (8.79)
AUS x PAPM 0.76 0.58
(1.20) (1.21)
USA x PAPM -6.09** -1.94*
(10.66) (2.08)
FRA x PAPM -4.78** -4.74**
(7.17) (8.79)
GER x PAPM -0.93* -0.91*
(1.82) (1.76)
GBR x PAPM -1.64** –1.26**
(3.68) (3.28)
DEN x PAPM 1.19 0.14
(0.92) (0.14)
SWE x PAPM -0.45 -0.63*
(1.42) (2.15)
R2 .834 .936 .879
Standard Error of Estimate 0.12 0.10 0.10
Number of Observations 56 16 40
Degrees of Freedom 39 9 27
Durbin-Watson 2.10 2.60 1.83
Restrictions p = 0.00** p = 0.02* p = 0.00**
Cointegration tests:
Durbin-Watson p < 0.01** p < 0.01** p < 0.01**
Dickey-Fuller (0 lags): ZDF – 51.43** – 19.02** -38.07**
Phillips-Perron (4 lags): Zpp – 43.45 * * – 14.74* – 34.43**
Bayes: t2 56.19 24.27 51.27
F-Test, (column 1) versus (columns 2, 3): F(3, 36) = 7.91, p = 0.00.
*Significant in one-tailed test at 5% level; ** significant in one-tailed test at 1% level.
t Dependent variable is WGRENT. Estimation is panel OLS with fixed effects (variables have
country means removed prior to regression). Absolute t-statistics in parentheses. Restrictions is the
test that the PAPM coefficients are equal across countries. Durbin-Watson cointegration test
follows Sargan-Bhargava testing for DW= 0. Dickey-Fuller and Phillips-Perron test for unit root in
the residuals and include a constant term but no trend. All regressions and tests are implemented
using the RATS econometrics software. NEWWORLD = (AUS, USA); OLDWORLD =
(FRA, GER, GBR, DEN, SWE).
out is to allow country dummy variables to interact with PAPM in the regression,
yielding country-specific estimates in the form of the 03i coefficients.14
14 Of course, one could allow country dummies to interact with all the right-hand side variables
-except that scarce degrees of freedom would be completely exhausted. An alternative is to
interact each right-hand side variable (LANDLAB, CAPLAB, PAPM, PROD) individually with
country dummies, and test for pooling. The results suggest that the PAPM interaction is the most
significant of all in terms of cross-country slope-coefficient variation.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
512 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
What about the expected signs of the coefficients in equation (4)? The Ricardo-
Viner (specific factors) model, or, more generally, the three-factor two-good model,
is the natural framework within which to answer the question. In this 3 X 2 context,
land and capital are the two ‘extreme’ factors associated with agriculture and
manufacturing, respectively, and labor is the ‘middle’ factor. With more goods than
factors, factor-price equalization does not hold, and endowments influence factor
prices. Ruffin (1981) shows that in general land and capital are ‘enemies,’ and that
both are ‘friends’ with labor. Thus, increases in land and capital endowments
increase wages and reduce rents, while increases in labor endowments lower wages
and increase rents: both p1 and 02 should be positive.
Theory is ambiguous about the sign of the f3i coefficients, however. It seems
intuitive that increasing the price of agricultural goods should raise rents more than
wages, and lower the wage-rental ratio, but Thompson (1985, 1986) has shown that
commodity price changes can have counterintuitive effects on factor prices in a 3 X 2
setting: an increased price of food could actually lower rents, rather than increase
them. The sign of 84 is also indeterminate, and depends on the workings of PROD,
a Solovian residual, introduced as a proxy for productivity-enhancing technological
forces. If the forces were land-saving (as seems likely in the land-scarce Old World),
then we would expect 04 > 0; if, instead, the forces were labor-saving (as seems
likely in the labor-scarce New World), then we ought to see 04 < 0.
Table 3 presents the econometric evidence, where the panel data is drawn from a
sample of seven countries using five-year period averages from 1870 to 1914.
Estimation is by OLS using fixed effects-the variables had country means removed
before the regression. The first column reports the results for the pooled sample of
all seven countries. In all cases the PAPM variable is allowed to interact with a
country dummy since F-tests of restrictions indicate that the PAPM coefficients vary
significantly across countries. An F-test clearly indicates that the New World and
the Old World have different structures, and thus should be treated separately, as in
columns 2 and 3. The results are quite good: of the 23 estimated coefficients, 19
have the correct sign; most of the 19 pass conventional significance tests; and those
with the wrong sign (PAPM coefficients for Australia and Denmark) are not even
weakly significant. The variance explained is quite high: country dummies (inter-
cepts) have already been removed but still the R2 is over 0.8 in all cases. Panel
Durbin-Watson statistics (DW) do not differ significantly from 2; thus serial correla-
tion is not an issue, and cointegration problems do not arise.15 The cointegration
properties of the estimated equation (4), as always only weakly testable, justify our
specification and an interpretation of the results as estimates of long-run equilib-
15 Several diagnostic tests were undertaken to verify the cointegration property, although the
small sample meant that the tests were inevitably weak. First, following the Sargan-Bhargava
procedure, we found that the Durbin-Watson statistic is significantly greater than zero in all cases
(the 1% critical value is DW = 0.46; see Mills 1990, p. 272). Second, we implemented the standard
Dickey-Fuller and Phillips-Perron tests, both of which strongly rejected the null hypothesis of a unit
root (we used tests with a constant but no time trend; and we used the test statistic T( p – 1); see
Hamilton 1994, Ch. 17). Third, we applied a standard Bayes test that provided further evidence in
favor of accepting the cointegration hypothesis (we used the standard RATS implementation of this
test).
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 513
rium factor-price equations. Moreover, if the variables are truly cointegrated we can
finesse the question of simultaneity bias: although the right-hand side variables in
(4) are surely endogenous-and in ways that are certainly of interest for questions
of trade and factor-flow analysis-the standard properties of a cointegrating regres-
sion allow us to accept our ,3 estimates as consistent (Hamilton 1994, p. 588).
Broadly speaking, we interpret these results as strong support for both the
Ricardo-Viner and Heckscher-Ohlin theories: not only do relative commodity prices
matter, but so too do relative factor endowments. Capital-deepening and land-
deepening both raise the wage-rental ratio, although the impact is larger in the New
World (where agriculture was bigger) than in the Old (where agriculture was
smaller). A rise in the relative price of agricultural goods favors returns to land over
returns to labor, and the impact is usually greater than unity when significant-a
corollary of what trade theorists call the magnification effect (Jones 1965).16 Econ-
omy-wide productivity growth plays a significant role, and one that conforms to
qualitative economic histories: that is, while productivity growth was land-saving in
the full sample (+ 0.71, column 1), it was labor-saving in the New World (-0.85,
column 2) and land-saving in the Old World (+ 1.05, column 3), a finding consistent
with the induced-innovation hypothesis.
So much for statistical significance, but what of quantitative significance? How
important were the various forces in contributing to the dramatic convergence in
wage-rental ratios? Table 4 offers some surprising answers. Panel A reports the
actual trends (log change per decade) in the wage-rental ratio and all four right-hand
side variables for each country. Panel B lists the regression coefficients from the
sub-samples (Table 3, columns 2 and 3), used in the subsequent decomposition
analysis (reported in Panels C and D). The decomposition analysis simply multiplies
the changes in the right-hand side variables (Panel A) by the regression coefficients
(Panel B), allowing us to infer what forces were doing most of the work in driving
trends in wage-rental ratios across the late nineteenth century.
Consider first a pairwise comparison-the results for the U.S.A. and Britain. The
increase in PAPM in America (+ 0.059 per decade) and its decrease in Britain
(- 0.065 per decade) reveal a dramatic price convergence, one that accounts for a
large share of the Anglo-American wage-rental convergence. Almost two thirds of
the fall in the American wage-rental ratio is explained by PAPM (Panel C: 61.3
percent), while it accounts for about one third of the rise in the British ratio (Panel
C: 36.9 percent). Combining the two, we find that about half of the Anglo-American
convergence in the wage-rental ratio is explained by commodity-price convergence
(Panel D: 48.1 percent). Thus, the Heckscher-Ohlin factor-price-equalization insight
is confirmed by Anglo-American experience. Differences in land-labor ratio trends,
however, contribute nothing to wage-rental ratio convergence (Panel D: – 0.3
percent): there is no support for the closing-the-frontier thesis here. Capital-deepen-
ing also contributes nothing to convergence: in fact, it makes for divergence (Panel
D: – 23.2 percent). There is, however, a strong factor-saving bias underlying
16 Note, however, the results for Australia and Denmark, where a rise in the relative price of
agricultural goods favors labor. We have no explanation for the perverse Australian result. We do,
however, have an explanation for the perverse Danish result, and we will discuss it at length below.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
514 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
TABLE 4
DECOMPOSITION OF CHANGING WAGE-RENTAL RATIOS, 1875-19140
A: Underlying data 1870-1914 (log change per decade)
WGRENT LANDLAB CAPLAB PAPM PROD
AUS – 0.255 -0.128 -0.035 -0.026 0.048
USA -0.188 -0.070 0.130 0.059 0.155
FRA 0.105 -0.100 0.075 0.006 0.052
GER 0.064 -0.134 0.160 0.024 0.066
GBR 0.220 -0.107 0.072 -0.065 0.090
DEN 0.248 -0.026 0.080 0.015 0.209
SWE 0.231 -0.012 0.127 0.050 0.167
B: Regression Coefficients (from Table 3, regressions 2 and 3)
LANDLAB CAPLAB PAPM PROD
AUS 1.161 1.187 0.583 -0.852
USA 1.161 1.187 -1.941 -0.852
FRA 0.766 0.826 -4.745 1.046
*GER 0.766 0.826 -0.914 1.046
GBR 0.766 0.826 -1.260 1.046
DEN 0.766 0.826 0.137 1.046
SWE 0.766 0.826 -0.627 1.046
C: Explaining changes (log change per decade)
change in due to due to due to due to residual
WGRENT LANDLAB CAPLAB PAPM PROD
AUS – 0.255 -0.148 -0.042 -0.015 -0.041 -0.008
shares 100.0% 58.2% 16.4% 6.0% 16.1% 3.3%
USA -0.188 -0.081 0.154 -0.115 -0.132 -0.014
shares 100.0% 43.0% -82.2% 61.3% 70.4% 7.5%
FRA 0.105 – 0.076 0.062 – 0.027 0.055 0.092
shares 100.0% – 72.9% 58.9% -26.3% 52.2% 88.1%
GER 0.064 -0.103 0.132 -0.022 0.069 -0.012
shares 100.0% -161.4% 207.0% -34.7% 108.5% -19.3%
GBR 0.220 -0.082 0.060 0.081 0.094 0.067
shares 100.0% -37.2% 27.0% 369% 42 7% 30.5%
DEN 0.248 -0.020 0.066 0.002 0.218 -0.019
shares 100.0% -8.1% 26.8% 0.8% 88.1% -7.6%
SWE 0.231 -0.009 0.104 -0.032 0.174 -0.007
shares 100.0% -3.9% 45.2% -13.6% 75.5% -3.2%
NEWWORLD (avg.) -0.221 -0.115 0.056 -0.065 – 0.087 -0.011
shares 100.0% 51.8% -25.5% 29.5% 39.1% 5.1%
OLDWORLD (avg.) 0.324 -0.128 0.126 0.074 0.164 0.089
shares 100.0% -39.6% 38.8% 22.8% 50.6% 27.5%
D: Explaining convergence (log change per decade)
change in due to due to due to due to residual
WGRENT LANDLAB CAPLAB PAPM PROD
NEWWORLD minus – 0.545 0.014 -0.069 -0.139 -0.250 -0.100
OLDWORLD
shares 100.0% -2.6% 12.7% 25.5% 45.9% 18.4%
US minus GB – 0.408 0.001 0.095 – 0.196 – 0.226 – 0.081
shares 100.0% -0.3% -23.2% 48.1% 55.4% 20.0%
* See text and Table 3.
Sources: see Appendix.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 515
Anglo-American productivity advance: a labor-saving bias accounted for 70.4 per-
cent of the fall in the American wage-rental ratio and a land-saving bias accounted
for 42.7 percent of the rise in the British wage-rental ratio (Panel C); the combina-
tion accounted for about half of the convergence (Panel D: 55.4 percent), leaving a
residual of one fifth unexplained. The powerful Anglo-American factor-saving
effects are consistent with the qualitative assertions of economic historians.
A factor-saving bias in productivity change and commodity-price convergence
were the prime movers underlying the spectacular late nineteenth century wage-
rental ratio convergence on the Anglo-American axis. Experience was, however, a
bit different elsewhere. Commodity-price convergence didn’t matter as much, and
factor accumulation mattered more. Small values under PAPM in Panel C could, of
course, simply reflect protectionist policies, at least on the continent; but not only
are the values small, they are of the “wrong” sign (for example, wage-rental ratios
were rising in France, Germany and Sweden, but trends in PAPM were serving to
lower them; Denmark has a perverse PAPM trend and a badly signed coefficient,
but two wrongs don’t make a right). Such results could be explained by more-than-
offsetting tariffs. They could also be explained by a switch out of grain-producing
activities and in to grain-using activities (Denmark being the canonical case). It is a
great irony that PAPM has negative entries for Sweden (Panel C: -0.032 per
decade), given that the factor-price-equalization theorem was constructed by two
Swedes.17 While we need more detailed analysis of continental response to grain
invasions, note that trends in PAPM also contributed very little to the Australian
collapse in the wage-rental ratio: Ohlin was correct in asserting that “trade
increased the price of land in Australia while tending to keep wages down” (Flam
and Flanders 1991, pp. 91-92), but incorrect in asserting that it was an important
force. We need to learn more about this New World case, and that of Argentina.
For the New World and Old World overall, we have the following results (Panel
D, row 1). Changing land-labor ratios accounted for none of the wage-rental ratio
convergence (indeed, these effects were countervailing: – 2.6 percent), while
capital-deepening accounted for about one eighth (12.7 percent). These endowment
influences offer mild support for the Ricardo-Viner model. Commodity-price con-
vergence accounted for about a quarter of the wage-rental ratio convergence (25.5
percent), offering stronger support for Heckscher and Ohlin. Factor-saving produc-
tivity advance exerted a strong influence on wage-rental ratio convergence (45.9
percent), highlighting the importance of innovation induced by factor scarcities.
5. ANGLO-AMERICAN CONVERGENCE: A COMPUTABLE GENERAL
EQUILIBRIUM ACCOUNTING
How robust might our results be? A complementary way to explore the determi-
nants of wage-rental ratio convergence is to apply counterfactual simulation analysis
to multi-sector, open-economy, computable general equilibrium (CGE) models. This
17 The true test of the factor-price-equalization theorem is, of course, the Swedish coefficient on
PAPM in Table 4 (columns 3 and 5). Significant negative coefficients support the theorem, but it is
still an irony that PAPM was moving over time in Sweden such as to cause divergence.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
516 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
section will now apply such models to Anglo-American experience.18 True, Britain
and the U.S.A. are only two countries of the eleven for which we have wage-rental
ratio data, and we have seen that these two countries may have been unusual.
Nevertheless, it seems useful to compare the implications of the CGE simulation
approach with the econometric results in Section 4.
The British and the U.S. models include tradable commodities (agricultural and
nonagricultural) and nontradable services.19 Britain imports, and the U.S. exports,
food and raw materials such as cotton, the latter a key input into manufacturing in
both economies. Britain exports, and the U.S. imports, manufactured goods. Factor
endowments and technologies are taken as given, as are the relative prices of traded
goods.20 The two models are calibrated to data for 1870-1871.
Table 5 reports the results; for the sake of comparison, panel A reports the actual
movements in real wages, rents, and the wage-rental ratio. We start with the
influence of commodity prices and the Heckscher-Ohlin hypothesis. Panel C reports
the impact of the Anglo-American commodity-price convergence discussed in Sec-
tion 3 on nominal factor returns in Britain and the U.S.A. Commodity-price
convergence raised agricultural prices in the U.S.A. and lowered them in Britain,
while having the opposite effects on manufactured goods prices in the two countries.
Not surprisingly, the result was that rents fell in Britain and rose in the U.S.A.,
while returns to capital (not shown) increased in Britain relative to the U.S.A. The
effect in both countries on nominal wages was positive. However, food was a key
consumption good in this period, and so these commodity price shocks reduced the
consumer price index in Britain and increased it in the U.S.A. The net effect was to
increase real British wages by 20.3 percent, while leaving U.S. real wages relatively
unaffected (Panel D). Commodity price shocks increased the British wage-rental
ratio by 152.3 percent, and lowered the U.S. ratio by 10.6 percent: they can thus
explain over two thirds of the rise in the British wage-rental ratio and a bit less than
a fifth of the fall in the U.S. wage-rental ratio. Relative to the U.S.A., the British
wage-rental ratio increased by 674.9 percent in the late nineteenth century (Panel
E), and commodity price convergence explains a fair share of this impressive factor
price convergence: roughly one quarter (182.2 percent).21
18 The models are fully documented in a paper by two of the present authors which isolates the
impact of late nineteenth century commodity-price convergence on the shrinking Anglo-American
real wage gap (O’Rourke and Williamson 1994).
19 In the U.S. model, agricultural intermediate goods such as cotton are separated out from the
rest of agriculture to take account of U.S. market power in these commodities.
20 Cotton exports face a downward sloping world demand curve; and U.S. manufactures substi-
tute imperfectly with imported manufactures in consumption.
21 Why does commodity-price convergence have a bigger impact on the British economy than on
the U.S. economy? We can think of at least three reasons. First, Britain was a much more open
economy than was the U.S.: imports accounted for roughly a quarter of British GDP in 1870,
whereas they accounted for less than ten percent of U.S. GDP. Second, the U.S. was more
technologically dynamic than Britain, and so it makes sense that the forces we have been exploring
explain a smaller proportion of the U.S. experience. Third, we may have understated American
commodity-price convergence by focusing on price gaps between big midwestern cities and Europe.
Price gaps between the farmgate and Chicago or Cincinnati probably declined by more than price
gaps between those cities and London.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 517
TABLE 5
THE ESTIMATED IMPACT OF ANGLO-AMERICAN COMMODITY-PRICE CONVERGENCE AND FACTOR
ENDOWMENT CHANGES ON WAGE-RENTAL RATIOS, 1870-1913: CGE SIMULATION*
Movements (in percent):
Wage Rate Farm Rent Wage-Rental C.P.I.
Ratio
A: Actual, real returns
USA + 47.0 + 258.3 -59.0
Britain +43.1 – 55.0 +217.7
B: Impact of input growth (land, labor and capital combined) on real returns
USA + 20.0 + 36.7 -12.2
Britain + 24.0 + 10.1 + 12.6
C: Impact of commodity price convergence on nominal returns
USA + 13.3 + 26.7 -10.6 + 13.0
Britain +11.7 -55.7 +152.3 -7.1
D: Impact of commodity price convergence on real returns
USA +0.3 +12.1
Britain + 20.3 – 52.3
E: Relative movements in Anglo-American wage-rental ratios
Actual Due to: Input Growth Commodity-
Price
Convergence
+ 674.9 + 28.2 + 182.2
* Panels A, C and D taken from O’Rourke and Williamson (1994), Table 3 and erratum thereto.
Panel B uses the footnote table input growth estimates and the CGE models for the USA and
Britain. Panel E reports movements in the ratio of the British to the American wage-rental ratio.
At least as far as Anglo-American wage-rental convergence was concerned,
Heckscher and Ohlin were right: commodity-price convergence made a powerful
contribution to factor-price convergence. But it was hardly the only force at work.
Some might argue, for example, that the closing of the American frontier mattered.
However, that piece of conventional wisdom is unlikely to have the predicted affect
once we note that land-labor ratios diminished faster in Britain than in America
(see Appendix).22 Relative trends in land-labor ratios on their own should thus have
led to wage-rental ratios falling in Britain relative to the U.S.A., not rising. What
about capital accumulation and the rate of capital deepening, as well as the export
of capital and labor from Britain to the U.S.A. and elsewhere? While capital
accumulation was faster in America, the differences do not seem to be very large.
Capital-labor ratios actually moved very similarly in the two countries in the late
22 The relevant data is as follows, for 1912 relative to 1872 (1872 = 100):
Labor Land Land/Labor Capital Capital/Labor
Britain 152.0 104.0 68.4 218.1 143.5
U.S.A. 285.4 232.9 81.6 415.5 145.5
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
518 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
nineteenth century. In short, it looks like changing factor endowments cannot
account for much of the wage-rental convergence. Looks can be deceiving, however:
these two economies had vastly different structures, agriculture being very big in
America and very small in Britain. Panel B in Table 5 nets out the influences of
these changing endowments. The combined increase in capital, labor and land in the
U.S.A. served to raise the wage by 20 percent and rents by 36.7 percent, thus
lowering the wage-rental ratio. In Britain, endowment trends led to wages increas-
ing by 24 percent and rents by 10.1 percent, implying an increase in the wage-rental
ratio. Endowment trends did therefore lead to wage-rental convergence. How can
this paradoxical finding be explained? In both countries, falling land-labor ratios
had the standard classical effect of lowering the wage-rental ratio, while rising
capital-labor ratios had the opposite effect. In Britain, agriculture was insignificant
enough that the former effect was swamped by the latter: in the U.S.A., however,
agriculture mattered enough that land-labor effects dominated.
In summary, according to the CGE accounting, endowment changes explain
almost 6 percent of the increase in the British wage-rental ratio, 21 percent of the
decline in the U.S. wage-rental ratio, and a little more than 4 percent of the
increase in the wage-rental ratio in Britain relative to the U.S.A. Together,
commodity price and endowment trends can explain 85 percent of the increase in
the British wage-rental ratio, 36 percent of the fall in the U.S. ratio, and 39 percent
of the relative increase in the British ratio. The residual unaccounted for by these
forces might plausibly be attributed to cross-country biases in technological change,
an exogenous determinant omitted from the CGE analysis but apparent in the New
World-Old World econometric results. Overall, the CGE results are consistent with
the econometric findings presented earlier, and exhibit similar orders of magnitude
for commodity-price effects (explaining 20 to 50 percent of factor price convergence)
and factor-endowment effects (explaining less than 10 percent).
6. ASSESSMENT AND AGENDA
Economic convergence within the currently industrialized OECD countries has
been taking place for at least a century, and the late nineteenth century was one
important part of that experience. Convergence has been manifested in absolute
form by aggregates like GDP per capita and GDP per man-hour. More importantly,
it was also manifested by relative factor-price convergence. Furthermore, the most
important part of this historical experience involved a collapse of factor-price
differentials between the Old World and New. This paper has focused on the
convergence of wage-rental ratios, and the relative scarcities of land and labor. In
1870, labor was very scarce and land was very abundant in the New World, while the
opposite was true of the Old World. By 1913, the gaps had narrowed. The
wage-rental ratio fell sharply everywhere in the New World, while it rose sharply
almost everywhere in the Old World.
What accounted for the convergence? Our results confirm the empirical relevance
of standard trade models in explaining the evolution of factor prices in the
late-nineteenth century world economy. While factor markets were becoming in-
creasingly integrated during this period, trade and technological change acted as
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 519
substitutes for factor migration in driving the world economy toward factor price
equilibrium. This conclusion also applies to absolute convergence in GDP per
worker or GDP per capita levels since these variables, used so often in the
convergence debate, are just a weighted average of individual factor prices. The
late-nineteenth century trade boom saw the convergence of commodity prices, and
the factor-price-equalization theorem predicts that some of the factor-price conver-
gence should have been driven by commodity-price convergence. It turns out that
Heckscher and Ohlin were right, but more right for Anglo-America than for other
participants in the convergence process. Some of this can be explained, of course, by
protective tariff responses in the Old World, and some of it by a rapid switch from
grain-producing to grain-using activities in agriculture. It also turns out that the
stress which Mill, Ricardo and Viner placed on changes in factor endowments (land,
labor and capital) is correct, but the magnitudes were relatively weak. We expected
a bigger contribution given the mass migrations to the New World, but apparently
these forces were partially offset by a quickening in accumulation and land settle-
ment.23 A third force contributing to factor-price convergence was a strong labor-
saving bias in the New World and a strong land-saving bias in the Old World, an
endogenous response to relative factor scarcities.
Future research on these issues might proceed in several directions. We think it
would be desirable to augment our late nineteenth century database to include more
countries, like Argentina in the New World and the Mediterranean countries in the
Old. We would also welcome more detailed assessments of individual country
experience, including growth and structural change in response to commodity-price
convergence and factor-saving bias, the accumulation of labor and capital, and
endogenous frontier dynamics, the latter being particularly relevant in the New
World. We also see a natural extension of the analysis to the interwar years when
open-economy pro-convergence forces collapsed-with a move toward protectionist
trade policies, migration restrictions, and turbulent international capital
markets-and when long-run convergence trends were reversed, at least for a time.
Finally, we think it is surely time to implement the same kind of analysis for the past
four decades of rising globalization.
University College, Dublin, Ireland, and Centre for Economic Policy Research,
U. K.
Northwestern University and National Bureau of Economic Research, U.S.A.
Harvard University and National Bureau of Economic Research, U.S. A.
APPENDIX
The derived variables used in the econometric analysis in this study are shown in
Table 6, and this appendix documents the sources of the underlying data.
23 Endogenous expanding New World frontiers and capital stocks clearly represented a source of
divergence in the late nineteenth century (Taylor 1995; Taylor and Williamson 1994; Williamson
1995).
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
520 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
TABLE 6
BASIC DATA*t
A: Panel Data Series
Country Years WGRENT LANDLAB CAPLAB PAPM PROD
AUS 1875-1879 0.561 6.471 – 0.052 0.167 -3.721
1880-1884 0.516 6.593 – 0.098 0.132 -3.664
1885-1889 0.395 6.565 -0.130 0.108 – 3.587
1890-1894 -0.044 6.285 -0.152 -0.031 -3.674
1895-1899 0.036 6.137 – 0.204 0.111 -3.766
1900-1904 -0.108 6.056 – 0.225 0.134 -3.709
1905-1909 -0.371 6.028 – 0.223 0.216 – 3.640
1910-1914 – 0.330 6.024 -0.175 0.075 – 3.353
USA 1875-1879 0.051 2.772 1.388 -0.120 – 7.245
1880-1884 0.018 2.787 1.553 -0.063 -7.143
1885-1889 -0.017 2.767 1.465 -0.090 – 7.032
1890-1894 – 0.065 2.736 1.654 -0.014 – 6.986
1895-1899 -0.057 2.687 1.723 – 0.067 – 6.959
1900-1904 – 0.076 2.632 1.756 -0.003 – 6.845
1905-1909 – 0.347 2.567 1.781 0.015 – 6.775
1910-1914 – 0.606 2.529 1.843 0.087 – 6.703
FRA 1875-1879 – 0.236 0.116 -5.320 0.006 2.412
1880-1884 -0.140 0.108 -5.281 0.060 2.422
1885-1889 0.081 0.082 -5.223 0.011 2.315
1890-1894 0.035 0.056 -5.161 – 0.009 2.345
1895-1899 0.141 -0.071 -5.204 -0.031 2.294
1900-1904 0.244 -0.161 -5.178 -0.067 2.340
1905-1909 0.224 – 0.206 -5.132 – 0.054 2.426
1910-1914 0.130 – 0.233 – 5.059 0.026 2.594
GER 1875-1879 -0.312 0.302 -5.382 -0.074 2.224
1880-1884 – 0.285 0.254 -5.313 -0.042 2.161
1885-1889 – 0.240 0.190 -5.247 -0.117 2.236
1890-1894 -0.110 0.129 -5.163 -0.042 2.291
1895-1899 -0.010 0.060 -5.072 – 0.089 2.382
1900-1904 -0.018 -0.012 – 4.980 -0.121 2.381
1905-1909 – 0.045 – 0.090 -4.889 – 0.045 2.449
1910-1914 – 0.088 -0.169 -4.822 0.011 2.456
GBR 1875-1879 – 0.554 0.945 – 1.676 0.185 – 4.889
1880-1884 – 0.257 0.922 – 1.615 0.168 – 4.858
1885-1889 – 0.150 0.869 – 1.609 0.085 – 4.713
1890-1894 -0.116 0.809 -1.608 0.147 -4.639
1895-1899 -0.114 0.744 -1.573 0.061 -4.539
1900-1904 0.000 0.682 -1.482 0.000 -4.589
1905-1909 0.079 0.619 -1.428 -0.071 -4.608
1910-1914 0.218 0.571 – 1.423 -0.041 -4.574
DEN 1875-1879 – 0.907 0.852 1.889 -0.102 – 0.657
1880-1884 – 0.878 0.870 1.891 -0.061 – 0.493
1885-1889 – 0.647 0.867 1.904 – 0.064 -0.423
1890-1894 – 0.480 0.850 1.927 – 0.005 -0.296
1895-1899 – 0.204 0.847 1.963 – 0.034 – 0.166
1900-1904 – 0.036 0.885 2.043 – 0.082 – 0.104
1905-1909 – 0.076 0.822 2.112 – 0.054 – 0.015
1910-1914 -0.040 0.760 2.170 – 0.051 0.073
SWE 1875-1879 – 0.733 – 0.150 -3.897 – 0.085 1.364
1880-1884 – 0.590 – 0.115 – 3.832 – 0.076 1.380
1885-1889 – 0.456 – 0.071 -3.784 -0.127 1.436
1890-1894 – 0.339 – 0.047 – 3.733 – 0.036 1.517
1895-1899 -0.151 – 0.055 -3.679 – 0.060 1.655
1900-1904 – 0.045 -0.083 – 3.608 0.032 1.741
1905-1909 0.010 – 0.145 -3.511 0.100 1.826
1910-1914 0.075 – 0.191 -3.454 0.091 1.947
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 521
TABLE 6 continued
B: Sample Statistics
DWGRENT DLANDLAB DCAPLAB DPAPM DPROD
(i) By Region
1875-1914 ALL 0.061 -0.082 0.087 0.009 0.112
NEWWORLD -0.221 -0.099 0.047 0.016 0.102
OLDWORLD 0.173 -0.076 0.103 0.006 0.117
1875-1894 ALL 0.096 -0.047 0.078 0.003 0.102
NEWWORLD -0.241 -0.074 0.055 -0.030 0.102
OLDWORLD 0.231 -0.036 0.086 0.017 0.102
1895-1914 ALL 0.034 – 0.109 0.094 0.013 0.120
NEWWORLD -0.206 -0.117 0.041 0.052 0.101
OLDWORLD 0.131 -0.106 0.115 -0.002 0.128
(ii) By Country
1875-1914 AUS -0.255 -0.128 -0.035 -0.026 0.048
USA -0.188 -0.070 0.130 0.059 0.155
FRA 0.105 -0.100 0.075 0.006 0.052
GER 0.064 – 0.134 0.160 0.024 0.066
GBR 0.220 -0.107 0.072 -0.065 0.090
DEN 0.248 -0.026 0.080 0.015 0.209
SWE 0.231 -0.012 0.127 0.050 0.167
1875-1894 AUS -0.404 -0.124 -0.067 -0.132 0.031
USA -0.077 -0.024 0.178 0.071 0.173
FRA 0.180 -0.040 0.106 -0.010 – 0.044
GER 0.134 -0.115 0.146 0.021 0.045
GBR 0.292 -0.091 0.045 -0.025 0.167
DEN 0.284 -0.001 0.026 0.065 0.241
SWE 0.263 0.068 0.110 0.033 0.102
1895-1914 AUS -0.143 -0.130 -0.011 0.053 0.061
USA -0.270 -0.104 0.094 0.050 0.142
FRA 0.048 -0.144 0.051 0.018 0.125
GER 0.011 -0.149 0.171 0.026 0.082
GBR 0.167 -0.119 0.093 -0.094 0.033
DEN 0.220 -0.045 0.122 -0.023 0.185
SWE 0.207 -0.072 0.139 0.063 0.215
* For variable definitions see text. D denotes a first difference of a log-level (growth rate). Means
are averages of five-year period averages, expressed as log-level change per decade. Sub-samples are
NEWWORLD = {AUS, USA}; OLDWORLD = {FRA, GER, GBR, DEN, SWE}.
t Source: see Appendix.
1. DATA SOURCES FOR WAGE-RENTAL RATIOS
What follows are the sources for the wage-rental (WGRENT) ratios reported in
the text. The data cover eleven late nineteenth century “countries” (Ireland is
treated as a separate country): Australia, Argentina, Canada, and the U.S.A. in the
New World; and Britain, Denmark, France, Germany, Ireland, Spain, and Sweden
in the Old World. Unless otherwise noted, the wage in the numerator is a nominal
urban unskilled wage rate and it is taken from the database underlying Williamson
(1995). The rent in the denominator always refers to farmland, and almost always
refers to nominal land values or prices per unit. The exceptions are Britain, Ireland,
and Spain, where they refer to nominal rents per unit. The wage-to-land-value ratio
is indexed on 1911 = 100 (except Spain, 1907 = 100) and a three-year moving
average is used (where possible) when the series are displayed in Figures 1-3.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
522 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
Argentina. Land Values: Prices from sales of rural land in Buenos Aires province,
per hectare. R. Cortes Conde, El progresso argentino, 1880-1914 (Buenos Aires:
Editorial Sudamericana, 1979), Table 3.8, p. 164 and Table 3.10, p. 166.
Australia. Land Values: Local government assessments of rural land, implicit
price, pounds per acre, Victoria. A. M. Taylor, “The Value of Land in Australia
Before 1913,”Australian National University, Source Papers in Economic History, no.
19, 1992, p. 20.
Wages: 1870-1900: Urban unskilled nominal wages in pounds per year,
G. Withers, unpublished database, n.d. 1901, and 1906-1914: Nominal industrial
wage index (1955 = 100). B. R. Mitchell, International Historical Statistics, The
Americas and Australasia, (Detroit: Gale Research Company, 1983), Table C4,
column j, p. 184, index of money wages in industry.
Britain. Land Values: 1870-1880: Norton, Trist, and Gilbert, “A Century of
Land Values: England and Wales,” in E. M. Carus-Wilson, ed., Essays in Economic
History (London: Edward Arnold, 1966), pp. 129-131. Average prices of a “repre-
sentative” portion of marketable agricultural estates 30-6,000 acres, purely agricul-
tural, excluding “fancy property.” 1881-1920, land rents from H. A. Rhee, The Rent
of Agricultural Land in England and Wales, 1870-1943 (Oxford: Institute for Re-
search in Agricultural Economics, 1949; for the Central Landowners’ Association),
Table 2, pp. 44-46, based on a weighted average of nominal rents per acre of
farmland, ten series. The land rents were converted to land values using the
expression, value = rent/interest rate, using the British consol rate, taken from
B. Mitchell and P. Deane, Abstract of British Historical Statistics (Cambridge:
Cambridge University Press, 1962), Table 8, p. 455.
Canada. Land Values: Average values of land per acre. Seventh Census of
Canada, 1931, Agriculture, vol. VIII, Department of Trade and Commerce, Domin-
ion Bureau of Statistics, Canada, 1936, Table 5, pp. 6, 46, 76, 146, 199, 383, 539, 591,
663, and 730. The 1921 census counts as farms only holdings larger than one acre.
The 1901 and 1911 censuses include smaller holdings, but these do not significantly
affect the results. Farms operated by share tenants and by cash and share tenants
are included.
Denmark. Land Values: 1845-1900: Land prices per T6nder Hartkorn, in K. J.
Christensen, Landbostatistik: Handbog I Dansk Landbohistorisk Statistik 1830-1900,
Landbohistorisk Selskab, Copenhagen, 1985, Table VII.2, pp. 106-107. 1885-1912:
Various issues of Statistik Aarbog, the Danish statistical yearbook published by
Danmarks Statistik. These contain the same information as the Christensen source.
The two sources report exactly the same numbers for the overlapping years
1885-1900, except for 1889, where the Statistical Yearbook’s number is used.
Wages: Industrial wages for unskilled males. H. C. Johansen, Danmarks Historie,
Bind 9, /)konomisk Statistik, 1814-1980 (Gyldendal: Copenhagen, 1985), Table 7.5,
pp. 294-295, column 2, Ore per hour.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 523
France. Land Values: Land price series for Argences, a sub-region: M. L6vy-
Leboyer, Le revenu agricole et la rente fonciere en Basse-Normandie: Etude de
croissance regional (Editions Klincksieck: Paris, 1972), Table K-56, pp. 186-189.
Germany. Land Values: Price index for agricultural land, 1913 = 100. W. G.
Hoffmann, Das Wachstum der deutschen Wirtschaft seit der mitte des 19. Jahrhunderts
(Berlin: Springer-Verlag, 1965), Table 139, pp. 569-570.
Wages: B. R. Mitchell, European Historical Statistics 1750-1975, Second Revised
Edition (New York: Facts on File, 1981), Table C4, pp. 193-194, index of money
wages in industry, 1900 = 100.
Ireland. Land Values: 1854-1908: Agricultural rent income in 1854, 1876, and
1908 from C. 6 Grada, Ireland Before and After the Famine: Explorations in Economic
History, 1800-1925 (Manchester: Manchester University Press, 1988), Table 33, p.
130; total acreage of agricultural land under cultivation for the same years from
Mitchell (British Historical Statistics, pp. 190-191). Dividing the former by the latter
gives an estimate of rent per acre as: 1854, 0.663; 1876, 0.763; and 1908, 0.545.
1901-1913: D. Nunan, “Price Trends for Agricultural Land in Ireland 1901-1986,”
Irish Journal of Agricultural Economics and Rural Sociology 12 (1987): 51-77, Ap-
pendix A, Table 1, pp. 69-71) reports conacre rents in the Limerick area. We link
the Nunan series with the 6 Grada series at 1908. The rents were not converted to
values by any interest rate adjustment.
Wage Rates: 1870-1913: We use F. D’Arcy, “Wages of Labourers in the Dublin
Building Industry 1667-1918,” Saothar 14 (1989), pp. 17-32. An alternative for the
1870-1901 period would be to use Wilson Fox’s farm wage series (Board of Trade,
Thirteenth Abstract of Labour Statistics of the United Kingdom 1907-1908, London,
1910: H. M. Stationery Office, Cd. 5041, pp. 73 and 137). Fox’s series rises somewhat
less steeply up to 1901 than does D’Arcy’s, and thus the wage rental ratio based on
D’Arcy would rise a bit less steeply too. However, since we have been using urban
unskilled wages throughout in our eleven-country panel, we use the D’Arcy series
here.
Spain. Land Values: Land rents from J. Carmona, El comportamiento economic
de la nobleza espahola en el siglo XIX: La Casa de Alcanices 1790-1910 (Doctoral
thesis, Universidad Complutense de Madrid, 1991), Table 2. First, an unweighted
average was taken of columns (4) and (5), pasture land in the Toledo area and large
estates. Second, an unweighted average was taken of this created series and column
(1), large wheat-growing estates in the Seville and Cordoba areas. (1900/04 = 100)
The land rents were converted to land values using the expression, value =
rent/interest rate. Interest rate for 1867-1873 is British consol rate: Mitchell and
Deane, Abstract (1962), Table 8, p. 455; for 1874-1882 is Spanish discount rate:
Estadisticas historicas de Espaha: Siglos XIX-XX (Madrid: Fundacion Banco Exte-
rior, 1989), Table 9.8, p. 388; for 1883-1920 is Spanish interest rate: P. M. Acenia
and L. Prados de la Escosura (eds.), La nueva historia economic en Espaha (Madrid:
Editorial Tecnos, 1985), Table 9, p. 278. These three series were linked together and
then used to convert land rents into land values.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
524 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
Sweden. Land Values: kronor per hectare. E. Lindahl, E. Dahlgren, and
K. Kock [Staff of the Institute for Social Sciences, University of Stockholm], Wages,
Cost of Living and National Income in Sweden 1860-1930; Volume III: National
Income of Sweden 1861-1930, part two (London: P. S. King and Son, 1937), Table
126, p. 393.
Wages: Day rates of male workers in agriculture, markegang returns, Ore.
G. Bagge, E. Lundberg, and I. Svennilson [Staff of the Institute for Social Sciences,
University of Stockholm], Wages, Cost of Living and National Income in Sweden
1860-1930; Volume II: Wages in Sweden 1860-1930, part two (London: P. S. King
and Son, 1935), Table 169, col. a, pp. 113-114.
United States. Land Values: Nominal, purchase-value of farmland per acre, in
dollars. P. H. Lindert, “Long-run Trends in American Farmland Values,” Working
Paper no. 45, Agricultural History Center, University of California, Davis (February
1988), Table 1, following p. 5.
2. DATA SOURCES FOR EXPLANATORY VARIABLES
We sought data for ten countries: Australia (AUS), Argentina (AGN), Canada
(CAN), and the USA in the New World; and Britain (GBR), Denmark (DEN),
France (FRA), Germany (GER), Ireland (IRL), Spain (SPA), and Sweden (SWE) in
the Old World. However, due to missing information, the following are excluded
from the analysis: Argentina, Canada, Ireland and Spain. The data is annual
1870-1914 wherever possible. Missing years were subject to interpolation along an
exponential trend. Variables are:
W: nominal wage index (see Appendix, Section 1)
VLAND: nominal land value index (see Appendix, Section 1)
LAND: total useable land stock index
K: total capital stock index
L: total labor force index
PA: price index of agricultural goods
PM: price index of manufactured goods
Y: real output index
The following were derived:
WGRENT = ln(W/VLAND)
LANDLAB = ln(LAND/L)
CAPLAB = ln(K/L)
PAPM = ln(PA/PM)
PROD = a Solovian productivity residual calculated according to the formula
ln(Y/L) – 0.4 ln(K/L) – 0.1 ln(LAND/L)
Australia. LAND: Land use in all colonies/states, excluding territories.
W. Vamplew, ed., Australians: Historical Statistics (Broadway, NSW: Fairfax, Syme &
Weldon, 1987), p. 73.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 525
K: Aggregate capital stock in constant 1910/11 million pounds. A. M. Taylor,
“External Dependence, Demographic Burdens and Argentine Economic Decline
after the Belle Epoque,” Journal of Economic History, vol. 52, December, 1992,
Appendix. Extended to 1870 similarly.
L: Total work force. Vamplew, p. 147.
PA: Geometric mean of the price index of pastoral and agricultural components
of GDP, weights corresponding to current price GDP share of each component.
Vamplew, pp. 133 and 217.
PM: Price index of the manufacturing component of GDP. Vamplew, p. 217.
Y: Real GDP index. A. Maddison, Phases of Capitalist Development, (Oxford:
Oxford University Press, 1982), pp. 172, 174.
Britain. LAND: Acreage of crops, rotation grasses, permanent pasture, and
fallow. B. R. Mitchell and P. Deane, Abstract of British Historical Statistics
(Cambridge: Cambridge University Press, 1962), pp. 78-79.
K: Net stock of capital in millions of pounds at 1900 prices. Charles H. Feinstein
and Sidney Pollard, eds., Studies in Capital Formation in the United Kingdom,
1750-1920 (Oxford: Clarendon Press, 1988), Table VIII, column 7, pp. 441-443.
L: Total labor force. Mitchell and Deane, Table 1, p. 60.
PA: The Rousseaux Price Index for total agricultural products. Mitchell and
Deane, Prices Table 3, column 3, pp. 471-473.
PM: Price index for principal industrial products, average of 1865 and 1885 equals
100. Mitchell and Deane, Table 3, column 4, pp. 471-473.
Y: Total net national income at 1900 prices, in millions of pounds. Mitchell and
Deane, Table 2, column 3, pp. 367-368.
Denmark. LAND: Total agricultural area. H. C. Johansen, Danmarks Historie,
Bind 9: Dansk /ikonomisk Statistik 1814-1980 (Copenhagen: Gyldendalske Boghan-
del, 1985), Table 2.2, column 7, p. 129.
K: Total capital in millions of 1929 kroner. Niels Kaergard, /ikonomisk Vaekst: En
/ikonometrisk Analyse af Danmark 1870-1981 (Copenhagen: Jurist-og Obkonomfor-
bundets Forlag, 1991), Table 7, column 6, p. 516.
L: Labor force in thousands. Svend Aage Hansen, /)konomisk Vaekst i Danmark,
Bind 2: 1914-1983 (Copenhagen: University of Copenhagen, 1983), Table 1, column
5, pp. 229-231.
PA: Total agricultural production price index. 1870-1900 from Hansen, Table 18,
column 11, pp. 323-324. 1900-1914 from Kaergard, Table 1, column 4, pp. 578-579.
The two price indices have been linked together and indexed on 1901 = 100.
PM: Wholesale price index. Johansen, Table 8.1, pp. 298-301, column 4 is used
for 1876 to 1913. For 1870 to 1875, column 1 (consumer price index) was linked with
column 4.
Y: GNP in million kroner at constant 1929 prices. Mitchell, European Historical
Statistics, Table K1, column 3, pp. 817 and 820.
France. LAND: Arable land is used as a proxy. B. R. Mitchell, European
Historical Statistics 1750-1975, 2nd revised edn. (New York: Facts on File, 1981).
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
526 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
Total of columns in Table D1, area of main cereal, potato, and sugar beet crops,
pp. 213 and 224, and area of vineyards from Table D3, pp. 298 and 300, in
thousands of hectares.
K: The capital stock series was created by deflating Levy-Leboyer and
Bourguignon’s agricultural and nonagricultural fixed capital stock series (in thou-
sand million francs) in Table 7.8 with their price series in Table A-IV, and then
weighting the series by their respective GDP shares, taken from Table A-I.
M. Levy-Leboyer and F. Bouguignon, The French Economy in the Nineteenth Century:
An Essay in Econometric Analysis (Cambridge: Cambridge University Press, 1990),
Table 7.8, p. 277; Table A-I, pp. 312-16; Table A-IV, pp. 327-331.
L, selected dates: Mitchell, European Historical Statistics, Table C1, p. 163. These
economy-wide labor force figures were computed by summing over all the occupa-
tional categories listed by Mitchell. Also, Levy-Leboyer and Bouguignon, Table 7.16,
p. 294, active population in millions. The above two series were interpolated.
PA: Price index for agriculture, 1908-12 base. Levy-Leboyer and Bouguignon,
Table A-IV, column 6, pp. 327-331.
PM: Price index for industry, 1908-12 base. Levy-Leboyer and Bouguignon, Table
A-IV, column 7, pp. 327-331.
Y: Real GDP index. Maddison, pp. 172 and 174.
Germany. LAND: Land planted with crops plus fallow, pasture, and miscella-
neous land areas. W. G. Hoffmann, Das Wachstum der Deutschen Wirtschaft seit der
Mitte des 19. Jahrhunderts (Berlin: Springer-Verlag, 1965), Table 48, column 19, pp.
272-273.
K: Capital stock in constant 1913 prices. Hoffmann, Table 39, column 7,
pp. 253-254.
L: Hoffmann, Table 20, column 9, pp. 204-206. These employment figures
include the self-employed. Most of those counted are men, but some female family
members who help are also included among the employed. Unfortunately, the
female workers were not accurately reported before 1907, so the employment
figures for 1870-1907 underreport the total number of employed workers in the
economy.
PA: Agricultural producer price index. Hoffmann, Table 137, column 4,
pp. 561-562.
PM: Deflator for NNP at market prices using the capital balance, in 1913 prices.
Hoffmann, Table 148, column 15, pp. 598-601.
Y: NNP in constant 1913 prices, in millions of marks. Mitchell, European
Historical Statistics, Table K1, pp. 817 and 821.
Sweden. LAND: Arable land is used as a proxy. Mitchell, European Historical
Statistics, Table D1, sum of the seven columns, pp. 219 and 235.
K: Capital stock in current prices, 1871-1910. U. Karlstrom, Economic Growth
and Migration During the Industrialization of Sweden, Ph.D. dissertation, Stockholm
School of Economics, 1985, Appendix B, Table 9, sum of columns 1 through 5,
p. 187. This series was deflated using an investment price index from 0. Krantz and
C. A. Nilsson, Swedish National Product 1861-1970 (Kristianstad: CWK Gleerup,
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 527
1975). The index was created by linking the domestic investment deflator series for
total domestic investment in machinery and buildings on pp. 48, 56, 66, and 76. The
linked series has been indexed on 1901 = 100.
L: Mitchell, European Historical Statistics, Table C1, p. 170. The economy-wide
labor force figures were computed by summing over all the occupational categories
listed by Mitchell. Since it is not clear that Mitchell’s categories cover all employed
workers, these figures may not be reliable. The labor force figures include those in
the armed forces; according to footnote 49 on p. 173, the 1870 figure “probably”
includes some “inactive persons.”
PA: Agricultural price index. Krantz and Nilsson. PA was created by linking the
agricultural deflator series on pp. 105, 111, 117, and 124.
PM: Manufacturing price index. Krantz and Nilsson. PM was created by linking
the manufacturing deflator series on pp. 107, 113, 118, and 126.
Y: GDP in constant 1913 prices, in million kroner. Mitchell, European Historical
Statistics, Table K1, pp. 818 and 825.
United States. LAND: Acres of improved land in farms. 1870-1900: Twelfth
Census of the United States 1900, Vol. V, Agriculture, Part I, (Washington, D.C.:
United States Census Office, 1902) Table II, p. xviii. 1900: Fourteenth Census of the
United States, 1920, Vol. V, Agriculture, General Report and Analytical Tables
(Washington, D.C.: Department of Commerce, Bureau of the Census, Washington,
1922), Table 6, pp. 36-37.
K: Real domestic capital stock in millions of 1929 dollars. The numbers for
1869-78 and for 1879-88 are the annual averages for the decades. John W.
Kendrick and M. R. Pech, Productivity Trends in the United States (Princeton, N.J.:
Princeton University Press, 1961), Table A-XV, column 3, pp. 320-322.
L: Total persons 10 years of age and over engaged in gainful occupations.
1870-1900: Occupations at the Twelfth Census, 1900, Special Reports (Washington,
D.C.: Department of Commerce and Labor, Bureau of the Census, 1904), Table IV,
p. 1. 1910 & 1920: Fourteenth Census of the United States, 1920, Vol. IV, Population,
Occupations, Table 2, p. 34.
PA: Historical Statistics of the United States. PA was created by linking the “Farm
products” series E42 (BLS) and E53 (Warren and Pearson).
PM: Historical Statistics of the United States. PM was created by linking the “All
commodities” series E40 (BLS) and E52 (Warren and Pearson).
Y: GNP in constant 1958 prices, in billions of dollars. B. R. Mitchell, International
Historical Statistics, The Americans and Australasia (Detroit: Gale Research Com-
pany, 1983), Table K1, pp. 887 and 889.
REFERENCES
ABRAMOVITZ, M. A., “Catching Up, Forging Ahead, and Falling Behind,” Journal of Economic
History 46 (1986), 385-406.
BAIROCH, P., “European Trade Policy, 1815-1914,” in P. Mathias and S. Pollard, eds., The
Cambridge Economic History of Europe, vol. III (Cambridge: Cambridge University Press, 1989).
BARRO, R. J., “Economic Growth in a Cross Section of Countries,” Quarterly Journal of Economics
106 (1991), 407-443.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
528 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
BARRO, R. J., AND X. SALA-I-MARTIN, ” Convergence across States and Regions,” Brookings Papers on
Economic Activity 1 (1991), 179-182.
BAUMOL, W. J., S. A. B. BLACKMAN, AND E. N. WOLFF, Productivity and American Leadership: The
Long View (Cambridge: MIT Press, 1989).
BuTLIN, N. G., Investment in Australian Economic Development 1861-1900 (Cambridge: Cambridge
University Press, 1964).
CRAFTS, N. F. R. AND M. THOMAS, “Comparative Advantage in UK Manufacturing Trade,”
1910-1935, Economic Journal 96 (1986), 629-645.
DAVID, P. A., Technical Choice, Innovation and Economic Growth (Cambridge: Cambridge University
Press, 1974).
DELONG, J. B., “Productivity Growth, Convergence, and Welfare: A Comment,” American Eco-
nomic Review 78 (1988), 1138-1154.
DITELLA, G., “The Economics of the Frontier,” in C. P. Kindleberger and G. DiTella, eds.,
Economics in the Long View, vol. I (New York: New York University Press, 1982).
DiAZ-ALEJANDRO, C. F., Essays on the Economic History of the Argentine Republic (New Haven: Yale
University Press, 1970).
DIxIT, A. K. AND V. NORMAN, Theory of International Trade (Cambridge: Cambridge University
Press, 1980).
ESTEVADEORDAL, A., “Historical Essays on Comparative Advantage, 1913-1938,” Ph.D. disserta-
tion, Harvard University, 1993.
FLAM, H. AND M. J. FLANDERS, Heckscher-Ohlin Trade Theory (Cambridge: MIT Press, 1991).
GERSCHENKRON, A., “Economic Backwardness in Historical Perspective,” in B. F. Hoselitz, ed., The
Progress of UnderdevelopedAreas (Chicago: University of Chicago Press, 1952).
GREEN, A. AND M. C. URQUHART, “Factor and Commodity Flows in the International Economy of
1870-1914: A Multi-Country View,” Journal of Economic History 36 (1976), 217-252.
GREENE, W. H., Econometric Analysis, 2nd ed. (New York: Macmillan, 1993).
HABAKKUK, H. J., American and British Technology in the Nineteenth Century (Cambridge: Cambridge
University Press, 1962).
HAMILTON, J. D., Time Series Analysis (Princeton: Princeton University Press, 1994).
HARLEY, C. K., “Late Nineteenth Century Transportation, Trade and Settlement,” in W. Fischer,
R. M. McInnis, and J. Schneider, eds., The Emergence of a World Economy 1500-1914,
Part II: 1850-1914 (Wiesbaden: Steiner, 1986).
,”Ocean Freight Rates and Productivity, 1740-1913: The Primacy of Mechanical Invention
Reaffirmed,” Journal of Economic History 48 (1988), 851-876.
HAITON, T. J. AND J. G. WILLIAMSON, “Latecomers to Mass Emigration: The Latin Experience,” in
T. J. Hatton and J. G. Williamson, eds., Migration and the International Labor Market
1850-1939 (London: Routledge, 1994).
HAYAMI, Y. AND V. W. RuTrAN, Agricultural Development: An International Perspective (Baltimore:
The Johns Hopkins Press, 1971).
INNIS, H. A., The Fur Trade in Canada: An Introduction to Canadian Economic History (Toronto:
University of Toronto Library, 1927).
IRWIN, D. A., “Free Trade and Protection in Nineteenth-Century Britain and France Revisited: A
Comment on Nye,” Journal of Economic History 53 (1993), 146-152.
JONES, R. W., “The Structure of Simple General Equilibrium Models,” Journal of Political Economy
73 (1965), 557-572.
KINDLEBERGER, C. P., “Group Behavior and International Trade,” Journal of Political Economy 59
(1951), 30-46.
LEAMER, E. E., “The Leontief Paradox, Reconsidered,” Journal of Political Economy 88 (1980),
495-503.
, Sources of International Comparative Advantage (Cambridge: MIT Press, 1984).
LEFF, N. H., “Economic Development in Brazil, 1822-1913,” First Boston Working Paper Series
FB-92-02, Columbia University, 1992.
LEWIS, W. A., “Economic Development with Unlimited Supplies of Labour,” Manchester School of
Economic and Social Studies 22 (1954), 139-191.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
FACTOR PRICE CONVERGENCE 529
LIEPMANN, H., Tariff Levels and the Economic Unity of Europe (London: Allen and Unwin, 1938).
LINDERT, P. H., “Long-Run Trends in American Farmland Values,” Agricultural History 62 (1988),
45-85.
MADDISON, A., Phases of Capitalist Development (Oxford: Oxford University Press, 1982).
MCLEAN, I. W., “Notes on Growth in Resource-Rich Economies,” Department of Economics,
Harvard University, 1990.
MILL, J. S., Principles of Political Economy (London: Longmans, 1848 [1929]).
MILLS, T. C., Time Series Techniques for Economists (Cambridge: Cambridge University Press, 1990).
MUNDELL, R. A., “International Trade and Factor Mobility,” American Economic Review 47 (1957),
321-335.
NELSON, R. R. AND G. WRIGHT, “The Rise and Fall of American Technological Leadership,” Journal
of Economic Literature 30 (1992), 1931-1964.
NICHOLS, D. A., “Land and Economic Growth,” American Economic Review 60 (1970), 332-340.
NORTH, D. C., “Ocean Freight Rates and Economic Development, 1750-1913,” Journal of Economic
History 18 (1958), 537-555.
NYE, J. V., “The Myth of Free-Trade Britain and Fortress France: Tariffs and Trade in the
Nineteenth Century,” Journal of Economic History 51 (1991), 23-46.
O’RoURKE, K. H., “Measuring Protection: A Cautionary Tale,” Journal of Development Economics,
forthcoming.
AND J. G. WILLIAMSON, “Late Nineteenth Century Anglo-American Factor-Price Conver-
gence: Were Heckscher and Ohlin Right?” Journal of Economic History 54 (1994), 892-916.
OBSTFELD, M., “International Capital Mobility in the 1990s,” Understanding Interdependence: The
Macroeconomics of the Open Economy (Princeton: Princeton University Press, 1995).
OFFER, A., “Farm Tenure and Land Values in England, c. 1750-1950,” Economic History Review 44
(1991), 1-20.
RASSEKH, F. AND H. THOMPSON, “Factor Price Equalization: Theory and Evidence,” Journal of
Economic Integration 8 (1993), 1-32.
REYNOLDS, L. G., Economic Growth in the Third World, 1850-1980 (New Haven: Yale University
Press, 1985).
RUFFIN, R. J., “Trade and Factor Movements with Three Factors and Two Goods,” Economics
Letters 7 (1981), 177-182.
SCHEDVIN, C. B., “Staples and Regions of Pax Britannica,” Economic History Review 43 (1990),
533-559.
SINCLAIR, W. A., The Process of Economic Development in Australia (Melbourne: Longman Cheshire,
1976).
TAYLOR, A. M., “Mass Migration to Distant Southern Shores: Argentina and Australia, 1870-1939,”
in T. J. Hatton and J. G. Williamson, eds., Migration and the International Labor Market,
1850-1939 (London: Routledge, 1994).
, “Peopling the Pampa: On the Impact of Mass Migration to the River Plate, 1870-1914,”
Working Paper Series on Historical Factors in Long Run Growth No. 68, National Bureau of
Economic Research, Cambridge, May 1995.
AND J. G. WILLIAMSON, “Convergence in the Age of Mass Migration,” Working Paper Series
No. 4711, National Bureau of Economic Research, Cambridge, May 1994.
THOMPSON, H., “Complementarity in a Simple General Equilibrium Production Model,” Canadian
Journal of Economics 18 (1985), 616-621.
, “Free Trade and Factor-Price Polarization,” European Economic Review 30 (1986), 419-425.
TRACY, M., Government and Agriculture in Western Europe, 1880-1988, 3rd ed. (London: Harvester
Wheatsheaf, 1989).
TREFLER, D., “International Factor Price Differences: Leontief Was Right!” Journal of Political
Economy 101 (1993), 961-987.
WHITNEY, W. G., “The Structure of the American Economy in the Late Nineteenth Century,”
Ph.D. dissertation, Harvard University, 1968.
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms
530 KEVIN H. O’ROURKE, ALAN M. TAYLOR AND JEFFREY G. WILLIAMSON
WILLIAMSON, J. G., “The Evolution of Global Labor Markets Since 1830: Background Evidence and
Hypotheses,” Explorations in Economic History 32 (1995), 141-196.
AND P. H. LINDERT, American Inequality: A Macroeconomic History (New York: Academic
Press, 1980).
WRIGHT, G., “The Origins of American Industrial Success, 1879-1940,” American Economic Review
8 (1990), 651-668.
ZEVIN, R. B., “Are World Financial Markets More Open? If So Why and With What Effects?” in
T. Banuri and J. Schor, eds., Financial Openness and National Autonomy (Oxford: Oxford
University Press, 1992).
This content downloaded from 128.59.222.107 on Thu, 23 Mar 2017 23:55:05 UTC
All use subject to http://about.jstor.org/terms