Does economic "globalization," understood as easier movement of goods and factors
across national boundaries, mean that states lose sovereignty and converge in their institutions and
policies? In particular, does globalization entail a "race to the bottom" and a substantial
dismantling of welfare states? Both Left and Right seem to agree that it does (a point attested,
e.g., by Economist 1997, 5, and Fligstein 1997, 8), or more precisely that "economic theory"
implies that it must. Dissenters mostly argue that this depressing theory is not borne out in
practice, particularly among the industrialized states of the Organization for Economic Co-operation and Development (OECD): no clear temporal link can be established between trade-dependence (or the expansion of trade) and policy changes; and major differences of policy persist
(Fligstein 1997, Garrett 1998).
Both Left and Right are wrong, and dissenters err in so readily abandoning to them the
theoretical ground. The relevant theories of economics and political economy say something
quite different and more nuanced about the effects of "globalization:" e.g., that a more open
world economy leads not at all to convergence, but to greater variety of national policies. Yet
the dissenters, accepting this tissue of distortions and vulgarizations as serious theory, have been
almost equally wrongheaded in their efforts at empirical refutation. Most of their cross-national
"tests" have failed to grasp marginalist logic or to incorporate elementary gravity-model controls
(above all: that smaller economies inevitably trade more); worse, they have wholly ignored (so far
as I am aware) the most telling evidence, namely the policies and politics of regimes whose
subjects already trade and move with complete freedom, namely the states or provinces of
existing federations (e.g., Canada, Germany, the U.S.A., Switzerland).
Section One recapitulates briefly the relevant theory. Section Two raises the issues of
marginalist logic and gravity models. Section Three surveys, very superficially, some recent work
on federations. Section Four concludes. To foreshadow some of the main results: globalization
1. Theory
Classical political economy offers two powerful models of the effects of increased (at
the limit: total) freedom of trade and mobility of factors among political jurisdictions. On the
economic side, we have the Heckscher-Ohlin-Samuelson (H-O-S) theory of trade, migration, and
investment based on differences in factor endowments; on the political side, the Tiebout model of
"voting with one's feet." For both, it is easiest to start with absolutes: imagine that we move
from a world in which labor and capital are entirely "trapped" within specific jurisdictions to one
in which they can move, freely and costlessly, wherever they wish. (As a convenient shorthand,
call the first the "closed," the second the "open" economy.) What happens, first economically,
second politically?
Under the closed economy, locally abundant factors were underpaid, locally scarce
ones overpaid. Workers in Japan, landowners in Argentina, capitalists in New York, "knowledge
workers" in Silicon Valley -- all owners of locally abundant factors(1) -- received far less than their
"world" wage; while landowners in Japan and workers in Argentina received far more. As
barriers fall, between nations as formerly within them, there is necessarily convergence toward the
"law of one price."(2) At the limit, just as in any competitive market, each factor commands only its
marginal product. Hence inevitably, greater openness and mobility mean that locally abundant
factors gain, locally scarce factors lose.
At the same time, movement from a closed to an open economy bestows on both
"world" welfare and the welfare of every individual nation a one-time dramatic improvement. As
every undergraduate learns to visualize in International Economics 101, the shift in price line
entailed by an opening to trade moves every nation all at once onto a higher indifference curve.
Essentially each country abandons much of what it does badly (e.g., agriculture in Japan), expands
what it does well (e.g., agriculture in Argentina); and the improvements thus attained come much
more rapidly than do those from capital- or efficiency-driven economic growth.(3) This implies,
accurately, that the abundant factors' gains outweigh the scarce factors' losses. As a whole, each
nation is better off, and owners of locally abundant factors are much better off; but owners of
scarce factors within each nation lose, at least in the short run.
Concretely, this means that unskilled labor loses in developed countries (where such
labor is less abundant) but gains in less developed countries (LDCs); while owners of physical or
human capital gain in developed countries (where such capital is abundant) but lose in LDCs.
Where returns to factors can adjust with relative freedom, they do so; where they cannot (e.g.,
local regulations or cartels hold returns above their marginal product), factors wind up un- or
underemployed. We should indeed expect rising inequality in advanced economies, declining
inequality in backward ones; coupled with (depending on local cartel power) high
unemployment of unskilled labor in advanced economies, of skilled labor in LDCs.
Politically, if we think again in absolutes, complete mobility of factors and complete
freedom of trade amount to an alternative realization of Charles Tiebout's (1956) model of
federalism (whose salient characteristic was costless mobility of factors among jurisdictions). As
Tiebout observed, such mobility creates (via "voting with one's feet") a competitive market
among jurisdictions, with two standard (and clearly welfare-improving) results:(4) (a) waste and
rent-seeking are eliminated, in the sense that public goods (like private ones) must be priced at
their exact cost of production; and (b) variety increases, i.e. some jurisdictions levy high taxes and
provide high levels of public service, while others keep taxes low but provide few services -- for
much the same reason that the retail market supports both Nordstrom's and K-Mart, i.e. different
consumers value such amenities differently. To Tiebout's results we may add a third, almost a
corollary of (a), namely (c) that in each jurisdiction taxation will be optimally non-distortionary.
The logic that supports these results is as simple as that of any competitive market: a
jurisdiction that prices its public services above cost (i.e., levies unnecessarily high taxes to
support them) will immediately lose all its taxpayers to some other jurisdiction that prices the
same services lower; and one whose taxes are more distortionary than they need be to yield the
given revenue will lose its entire tax base to a less distortionary regime. But a "K-Mart"
jurisdiction will actually attract "cheap" subjects and can thrive happily alongside one that offers
more luxurious (and pricier) services. To speak in a shorthand that locals in both cases will
immediately understand, Culver City can easily co-exist with Beverly Hills, or Somerville with
Newton.
Applied to a world in which factors can move costlessly among nations, the Tiebout
model predicts both greater efficiency and greater variety of public policy: all governments will
ruthlessly pare away waste in their provision of public services; but some will provide lavishly,
others stingily. (As we shall see below, one secure empirical finding -- often called "Wagner's
Law" -- seems to be that the income elasticitiy of demand for public services is greater than one;
hence, all else equal, rich jurisdictions will spend a higher share of income on public services than
poor ones.) Concretely, the greater efficiency probably means greater use of market
mechanisms (e.g., sale of pollution rights, grants via vouchers) and more "contracting out" of
governmental services.
Tiebout, however, was thinking of public goods in the classical sense of high external
benefits: public order, defense, enforcement of contracts, education and physical infrastructure.
What of the extensive redistributive and social insurance mechanisms of the modern welfare state,
or corporatist efforts to compress wages (i.e., essentially to pay the most skilled workers less, the
least skilled ones more, than their marginal product)? Most transfers and wage compression must
indeed fall away under assumptions of total mobility: workers (and other factor owners) will
migrate from locations that pay less than their marginal product, and will flood into ones that pay
above that level, until returns equalize (or the policies that try to sustain such market-defying
prices collapse of their own weight).(5) Governments faced with global openness cannot defy the
"law of one price."
Genuine social insurance, however, can survive and flourish under economic
openness, assuming that it avoids the obvious pitfalls -- observed consistently in competitive
markets for private insurance -- of moral hazard and adverse selection.(6) Indeed, a state that
socializes some of the risks, e.g. of trade-related sectoral unemployment, in incentive-compatible
ways may actually be more attractive both to workers and to employers than one that does not.(7)
Certainly, as Rodrik (1996) argues, the observed (positive) correlation between trade openness
and size of government (controlling for other relevant variables, such as size) seems most
consistent with a social insurance explanation.
Turn, finally, consider two further bugbears of the "globalization" debate, namely
"deindustrialization" (of the advanced economies) and environmental deregulation. No theory of
economics or political economy dictates that greater economic openness must lead to either. If
services were uniformly more skill-intensive than manufacturing, then the advanced economies (by
definition skill-abundant) would indeed shift from manufacturing to services (becoming, nota
bene, vastly better off in the process) as international trade became easier and cheaper. But I
know of no evidence for that premise. Rather, some kinds of manufacturing (e.g., high-end
apparel) are more skill-intensive than others (e.g., cheap textiles), and advanced countries will
focus on these (and, of course, on services).(8) Similarly, that First-World publics have a more
pronounced taste for environmental purity constrains only the kinds of economic activity that they
can pursue, not their intensity or profitability. What the Tiebout results say is only that
governments will choose maximally efficient ways to pursue whatever level of environmental
health they choose, e.g. auctionable pollution credits rather than uniform regulatory limits.
For greater verisimilitude, suppose that capital becomes mobile but labor does not:
workers are constrained against moving both by policy (e.g., immigration controls) and by nature
(e.g., language barriers). Do any of these results change? The economic implications change
hardly at all, the political ones considerably (but in one crucial respect quite counterintuitively).
As Mundell, Samuelson, and others demonstrated long ago, trade in goods substitutes fully, in all
but the most extreme cases, for trade in factors; e.g., the same economic results, up to and
including factor price equalization, follow theoretically from exporting labor-intensive goods as
from exporting labor. So a "globalization" that involved only easier trade, but no easier mobility
of either capital or labor, would suffice to increase inequality in advanced countries, decrease it in
backward ones, and improve welfare everywhere.
Politically, the central Tiebout result, that government waste must disappear, is
actually strengthened if only capital can move easily. Waste and rent-extraction will still induce
capital to flee; but this will unambiguously lower the wage of labor if the size of the labor pool
remains constant.(9) In the government's eyes, the only thing worse than an immiserated workforce
that can leave is one that cannot, and remains at home to vote, or to riot, against it. But if labor
cannot move there is less reason to expect variety in government(10) and every reason to expect that
taxation will be biased in favor of mobile, against immobile factors -- i.e., in favor of capital, and
against labor.
Summarizing, then, we can say that "globalization" in the "global" sense -- easier
movement of goods, services, and all factors -- would, according to the best available theories of
economics and political economy:
To the extent that only goods and capital, but not labor, move more easily, we would still expect
all of the above to hold, except
So much for the theory. Is it empirically accurate?
Efforts to examine empirically the link between "globalization" and -- choose your
poison -- inequality, deregulation, welfare cuts, decreases in high-bracket taxation, usually
consist only of looking for correlations between countries' levels of trade (typically, trade share in
GDP) and the alleged social or political effect. More rarely, scholars look instead, or also, at
changes in these levels: e.g., has inequality increased as trade share has increased?
As the renowned trade economist Edward Leamer has noted somewhat acerbically, we
can learn almost nothing from such tests (Leamer 1996), for at least two reasons. First, it is not
flows but prices that count; and prices are set at the margin, by potential more than by actual
flows. No one doubts, for example, that world financial markets are integrated: the compelling
evidence is that the real price of money -- the "covered" interest rate -- is the same everywhere.
It simply does not matter that most countries' savings are invested at home, i.e. that international
investment flows are still relatively small (Feldstein and Horioka 1980). High flows, in fact, are
usually evidence that prices have not (yet) converged: the flows of goods between East and West
German cities were doubtless greatest immediately after the Wall came down, when prices
differed most; as convergence was achieved, the flows diminished drastically.
Worse, the one flow that is most often studied, trade share in GDP, is overwhelmingly
a function of country size. "Gravity models" of international trade imply straightforwardly that
smaller countries will trade more, a result known to many political scientists as "Taagepera's
Law" (Taagepera and Hayes 1977); hence in most analyses country size must enter as a control
variable.
Suppose, for the sake of argument, that welfare cuts in the advanced economies were
caused chiefly by competitive pressures from abroad. How would we establish that fact
empirically? Certainly not by observing that little countries like the Netherlands both trade a lot
and have cut welfare. Rather, we would want to see a link between price convergence and
welfare cuts; but this test is rarely undertaken.(11)
Does a world in which people can move easily among competing jurisdictions in fact
have the effects that the Heckscher-Ohlin and Tiebout models predict? Our best evidence
probably comes, not from the nation-states among which mobility is growing but still quite
limited, but rather from governments constitutionally required to permit their subjects freedom of
movement, namely the states and provinces of federal regimes. Five empirical findings seem
particularly relevant.
First, federal systems offer no support for the view that greater mobility implies
"convergence" to a single level of spending. Among the U.S. states between 1960 and 1990, for
example,
a healthy cross-sectional variation is evident. The standard deviations [of
total spending] hover at about 20 percent of the mean, and the maximum
spending level is typically more than twice the minimum (Matsusaka 1995,
598).
Were spending levels converging, we would of course see instead a diminution of the standard
deviation.
Second, the high mobility among subunits of federal systems is associated neither with
lower taxation nor a smaller state. Overall, to be sure, federal Australia, Switzerland, and the
U.S. devote a considerably below-average share of GDP to government spending (all have
percentages in the mid-30s, vs. an overall OECD average of about 45 per cent); but federal
Canada spends at, federal Germany well above, the average level. More to the point, spending by
the subunits -- states or provinces -- in such systems continues to rise, constitutes a significant
share of total government expenditure, and is predicted by exactly the same variables as obtain
among separate countries. In the U.S., for example, spending by states and localities rose
between 1960 and 1990 by a factor of 2.5 in constant dollars, while real GDP over the same
period increased by a factor of about 2.4 (Matsusaka 1995, 598; World Bank 1992, Table 2).
Nor is spending dominated by the central governments (whose subjects of course are more
"captive") in federal countries: state and local outlays constituted 28 per cent of total government
expenditure in the U.S. in 1990, 35 per cent in Germany, 49 per cent in Canada.(12) Perhaps most
importantly, the pattern of spending, e.g. among the U.S. states, is extremely well predicted (R2 =
.93) by "Wagner's Law" (that government spending rises with income) and two other standard
variables in country analyses, population density and the share of population living in metropolitan
areas (Matsusaka 1995, Table 4).(13) In other words, both in the level and the determinants of their
spending, states in federal regimes, whose subjects can easily "vote with their feet" or transfer
their assets, strongly resemble nation-states. Mobility appears to make little difference.
On the other hand -- and this is the third point -- subunits of federal systems engage
in much less redistribution than do their central counterparts. Among the U.S. states in 1995, for
example, 28 per cent of total outlays went to subsidies and transfers (excluding transfers to other
levels of government); but for the U.S. national government, the comparable figure was 47 per
cent.(14) This, I suspect, can be traced precisely to the greater ease of mobility among the subunits:
an overly generous policy of redistribution in any of them would drive out the productive, and
pack in the unproductive, elements.
Fourth, and on the taxation rather than the expenditure side, we observe among the
subunits in almost every federal system, and with particular vividness in the historical evolution of
the nineteenth-century German Reich (Hallerberg 1996), exactly the kind of bias predicted
theoretically, namely lighter burdens on more mobile factors, heavier ones on immobile factors
(cf. Economist 1997, 30).
Finally, exactly as one would expect among any set of units where completely free
trade in goods and factors obtains, production and trade exhibit almost exactly the pattern that
Heckscher-Ohlin theory predicts, i.e. concentration on (and "export" of) products intensive in the
factors with which the given subunit is abundantly endowed (for the U.S. case see Kim 1995).
This end-state strongly suggests that, in the transition to it -- i.e., as trade opened among the
U.S. states, furthered by such innovations as canals and railroads -- locally scarce factors
suffered, locally abundant ones gained.
Most of the theory that supposedly bears on the issue of globalization and
convergence is misunderstood; most of the evidence, misapplied. Viewed accurately, both the
theory and the relevant evidence suggest that globalization will induce:
In short, globalization has powerful and important effects, but they are not the ones
usually asserted. We have been looking for the wrong things, in the wrong places, by the wrong
methods.
REFERENCES
Dickens, William T.; Lawrence F. Katz; Kevin Lang, and Lawrence H. Summers. 1989.
Employee Crime and the Monitoring Puzzle. Journal of Labor Economics, 7: 331-47.
The Economist. 1995. Concentrating the Mind: Coats Viyella Profile. 18 February.
The Economist. 1997. The Future of the State: A Survey of the World Economy. 20
September.
Feldstein, Martin, and Charles Horioka. 1980. Domestic Saving and International Capital Flows.
Economic Journal 90: 314-29.
Fligstein, Neil. 1997. Is Globalization the Cause of the Crises of the Welfare State. Paper
prepared for delivery at the Annual Meetings of the American Sociological Assocation, Toronto.
July.
Garrett, Geoffrey. 1998. Global Markets and National Politics: Collision Course or Virtuous
Circle? International Organization, 52 (50th Anniversary Special Issue), forthcoming; Fall.
Hallerberg, Mark. 1996. Tax Competition in Wilhelmine Germany and Its Implications for the
European Union. World Politics 48: 324-57.
International Monetary Fund (IMF). 1997. Government Finance Statistics Yearbook, vol. 21.
Washington: IMF.
Kim, Sukkoo. 1995. Expansion of Markets and Geographic Distribution of Economic Activities:
The Trends in U.S. Regional Manufacturing Structure, 1860-1987. Quarterly Journal of
Economics 110: 881-908.
Leamer, Edward E. 1987. Paths of Development in the Three-Factor, n-Good General
Equilibrium Model. Journal of Political Economy 95: 961-99.
Leamer, Edward E. 1996. In Search of Stolper-Samuelson Effects on U.S. Wages. Working
Paper no. 5427, National Bureau of Economic Research.
Matsusaka, John G. 1995. Fiscal Effects of the Voter Initiative: Evidence from the Last 30
Years. Journal of Political Economy 103: 587-623.
Moene, Karl Ove, Ragnar Nymoen, and Michael Wallerstein. 1996. The Persistence of Slack and
Tight Labor Markets. Mimeo. December.
Rodrik, Dani. 1996. Why Do More Open Economies Have Bigger Governments? NBER
Working Paper no. 5537. Cambridge, MA: National Bureau of Economic Research.
Taagepera, Rein; and James P. Hayes. 1977. How Trade/GNP Ratio Decreases with Country Size. Social Science Research 6: 108-132.
Tiebout, Charles. 1956. A Pure Theory of Local Expenditures. Journal of Political Economy
64: 416-24.
World Bank (International Bank for Reconstruction and Development). 1992. World Development Report 1992: Development and the Environment. New York: Oxford University Press.
1. Recall that it is local vs. world ratios of factors that define relative scarcity or abundance; hence the usefulness of such pictorial constructs as Leamer's "endowment triangles" (or, for a four-factor universe, "endowment tetrahedrons"). Leamer 1987.
2. Mostly as an amusement, some German newspapers plotted this phenomenon in some detail in the months following the opening of the Wall. Initially, of course, Western televisions and automobiles cost far more in Dresden than in Dortmund; but inevitably, within a few months they cost virtually the same in both cities.
3. China is probably Exhibit A for this phenomenon. Between 1965 and 1990, exports of goods and nonfactor services rose from 4 to 18 per cent of GDP; over the same period, GDP grew on average by 7.9 per cent annually (9.5 per cent annually between 1980 and 1990), a rate hardly achievable only by investment or improvements in productive efficiency. World Bank 1992, Tables 2 and 9.
4. Crucial to all these results, but seemingly self-evident, is some link between politicians' welfare and the tax base of their jurisdiction. I.e., we must assume that a political leader who drives all productive resources out of her jurisdiction suffers some personal loss.
5. That competitive firms often differ in the wages they pay seemingly identical employees -- e.g., janitors and secretaries in legal firms make more than their counterparts employed by struggling retailers -- seems to me irrelevant to this discussion. Those differentials are usually explained as "efficiency wages" (e.g. Dickens et al. 1989, and a vast ensuing literature) and apply only to firms that can be selective in whom they hire. Governments, at least to this point, must accept the citizens they inherit.
6. A state that insures too generously, e.g. against illness, will encourage the event insured against (people will indulge in risky behaviors or will fake illness) and progressively drive out those at least risk (notably the healthy young). The first phenomenon is known as moral hazard, the second as adverse selection.
7. Absent such collective insurance, employers are tempted to hoard labor as a form of private insurance. Cf. Moene, Nymoen, and Wallerstein 1996.
8. Precisely this strategy has been pursued, with considerable success, by the multinational textile and clothing firm Coats Viyella: Economist 1995.
9. Recall that, in a competitive market, the wage of each factor equals its marginal product; and in almost any reasonable specification of a production function, the cross-partial of output with respect to labor and capital is positive, i.e. the wage of labor varies directly with the supply of capital. In the simplest Cobb-Douglass production function, for example, Y = AKL1-, Y/L = A(1-)(K/L): more capital implies a higher wage for labor, less capital a lower wage.
10. I.e., we should observe about the same variety among governments as we do among college or company cafeterias, which also rely on a "captive" clientele.
11. Jagdish Bhagwati has looked for evidence of such a link between "globalization" and growing inequality in the U.S. and finds none: The Economist 1997, 41.
12. Calculated from The Economist 1997, 8 (total government expenditure), and World Bank 1992, Table 11 (central government expenditure).
13. All else equal, metropolitan population expands, population density diminishes, government spending. Metropolitan areas demand more services, but a dense population makes these cheaper to deliver. Matsusaka's central point, echoing earlier work on Switzerland, is that direct democracy (in particular: the initiative) restrains government spending.
14. Calculated from IMF 1977, 394 and 396: "Expenditure and Lending minus Repayments, by Economic Type"; Category 3, "Subsidies and Other Current Transfers" minus Subcategory 3.2, "Transfers to Other Levels of Government." Comparable or lower levels of transfers appear to obtain for subunits of other federal systems: e.g., among Swiss cantons, 22 per cent of total outlays; among Australian states, 13 per cent; among German Länder, 10 per cent. IMF 1997, 28, 159, 369.