Globalization: What Did We Miss?

Paul Krugman

March 2018
Concerns about possible adverse effects from globalization aren’t new. In particular, as U.S.

income inequality began rising in the 1980s, many commentators were quick to link this new

phenomenon to another new phenomenon: the rise of manufactured exports from a group of

newly industrializing economies.

Economists – trade economists, anyway – took these concerns seriously. After all, standard

models of international trade do say that trade can have large effects on income distribution:

the famous 1941 Stolper-Samuelson analysis of a two-good, two-factor economy showed how

trading with a labor-abundant economy can reduce real wages, even if national income grows.

There was every reason to believe that the same principle applied to the emergence of trade

with low-wage economies exporting not raw materials but manufactured goods.

And so during the 1990s a number of economists, myself included (Krugman 1995), tried to

assess the role of Stolper-Samuelson-type effects in rising inequality. Inevitably given the

standard framework, such analyses did in fact find some depressing effect of growing trade on

the wages of less-educated workers in advanced countries. As a quantitative matter, however,

they generally suggested that the effect was relatively modest, and not the central factor in the

widening income gap.

Meanwhile, the political salience of globalization seemed to decline as other issues came to the

fore. So academic interest in the possible adverse effects of trade, while it never went away,

waned.

1
In the past few years, however, worries about globalization have shot back to the top of the

agenda, partly due to new research, partly due to the political shocks of Brexit and Trump. And

as one of the people who helped shape the 90s consensus – that the income distribution effects

of rising trade were real but modest – it seems appropriate to ask now what we missed. What

aspects of rising trade did we either fail to see at the time or fail to anticipate?

1. The 90s consensus

This is a short paper, not a literature review, so I don’t want to go through all the various 90s

analyses that tried to assess the distributional effects of trade. Instead, let me summarize the

methodological and quantitative conclusions that became fairly orthodox by the mid-1990s.

In terms of methodology: there was, for a time, some confusion (and a bit of heated debate)

over how to use data on trade to assess wage impacts. Most studies focused on the volume of

trade, and in particular on the factor content of trade – the labor and other resources

embodied in exports, the labor and other resources that would have been required to produce

imports. Some economists vehemently objected to this approach, since Stolper-Samuelson is

strictly speaking about prices rather than quantities – that is, it’s about the relationship

between goods prices and factor prices. Yet goods prices are endogenous; when trying to

assess the impact of globalization, it made no sense to treat prices as a causal variable.

2
What eventually emerged from this debate was a “but-for” approach: asking how different

wages would have been but for the rise of manufactured exports from developing countries,

which were minimal in 1970 but significant by the mid-1990s. It turned out that this but-for

approach was also consistent with a factor content calculation: the effect of North-South

manufactures trade on advanced economies was, in simple models, equivalent to what would

have happened if the OECD had been a closed economy experiencing immigration of less-

educated workers and emigration of more-educated workers corresponding to the factors

embodied in the goods being traded.

And the basic fact in the mid 1990s was that imports of manufactured goods from developing

countries, while much larger than in the past, were still small relative to the size of advanced

economies – around 2 percent of GDP. Given reasonable estimates of factor intensities and

elasticities of substitution, this wasn’t enough to cause more than a few percent change in

relative wages. This number wasn’t trivial, but it wasn’t big enough to be a central economic

story either.

This was a moderately comforting result for free-trade advocates. But what did the 90s

consensus miss?

3
2. Hyperglobalization

It is, I’d argue, quite possible, even likely, that assessments of the impact of trade made circa

1995, which inevitably relied on data from a couple of years earlier, were right in finding

modest effects. In retrospect, however, trade flows in the early 1990s were just the start of

something much bigger, what Arvind Subramanian and Martin Kessler (2013) have dubbed

“hyperglobalization.”

Until the 1980s, it was arguable that the growth of world trade since World War II had mainly

reflected a dismantling of the trade barriers erected during the interwar period; world trade as

a share of world GDP was only slightly higher than it had been in 1913. Over the next two

decades, however, both the volume and the nature of trade moved into uncharted territory.

Figure 1 shows one indicator of this change: manufactured exports from developing countries,

measured as a share of world GDP. As you can see, what seemed in the early 1990s like a major

disturbance in the trade force was just the beginning.

What caused this huge surge in what was, in the 1990s, still a fairly novel form of trade? The

answer probably includes a combination of technology and policy. Freight containerization was

not exactly a new technology, but it took time for business to realize how the reduction in

trans-shipping costs made it possible to break up value chains, moving labor-intensive parts of

the production process overseas. Meanwhile, there was a broad move away from import-

4
substituting industrialization toward outward-looking policies; and of course China made a

dramatic shift from central planning to a market economy focused on exports.

Figure 1

Developing country manufactures exports as

% of world GDP
6

0
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
2014
2016

Source: World Bank

Since manufactured exports from developing countries, measured as a share of the world

economy, are now triple what they were in the mid-1990s, should we conclude that the effect

on income distribution has also tripled? Probably not, for at least two reasons.

First, a significant part of the increase in developing-country exports reflects the rapid growth

of South-South trade, which is an important story but not relevant to the impacts on advanced-

country workers. Even more important, however, the nature of this trade growth – involving a

5
breaking up of the value chain – means that the factor content of North-South trade hasn’t

risen nearly as fast as the volume.

Consider two cases: imports of apparel from Bangladesh, and imports of iPhones from China. In

the first case, we are in effect importing the services of less-educated workers, putting

downward pressure on the demand for such workers in the U.S.. In the second case, most of

the value of that iPhone reflects inputs from high-wage, high-education countries like Japan; we

are in effect importing skilled as well as unskilled labor, so the impact on income distribution

should be much smaller.

Despite these qualifications, It’s clear that the impact of developing-country exports grew much

more between 1995 and 2010 than the 90s consensus imagined possible, which may be one

reason concerns about globalization made a comeback.

3. Trade imbalances

One major contrast between most economic analyses of globalization’s impact and those of the

broader public – including, of course, Donald Trump – is focus, or lack thereof, on trade

imbalances. The public tends to see trade surpluses or deficits as determining winners and

losers; the general-equilibrium trade models that underlay the 90s consensus gave no role to

trade imbalances at all.

6
The economists’ approach is almost certainly right for the long run, both because countries

must pay their way eventually and because trade imbalances mainly affect the relative shares

of traded and nontraded sectors in employment, with no clear effect on the overall demand for

labor. However, in the long run we are all dead, and rapid changes in trade balances can cause

serious problems of adjustment – a broader theme I’ll return to shortly.

Consider, in particular, the developments shown in Figure 2, which compares the U.S. nonoil

trade balance (which is overwhelmingly manufactured goods) with U.S. manufacturing

employment.

Figure 2
0.00000 18000
92
94
96
98
00
02
04
06
08
10
12
14

-0.50000 16
17000
19
19
19
19
20
20
20
20
20
20
20
20
20

-1.00000 Manufacturing employment (right)

16000
-1.50000
15000
-2.00000
14000
-2.50000
13000
-3.00000
12000
-3.50000
Nonoil trade
-4.00000 balance as 11000
% of GDP (left)
-4.50000 10000

7
Until the late 1990s employment in manufacturing, although steadily falling as a share of total

employment, had remained more or less flat in absolute terms. But manufacturing employment

fell off a cliff after 1997, and this decline corresponded to a sharp increase in the nonoil deficit,

of around 2.5 percent of GDP.

Does the surge in the trade deficit explain the fall in employment? Yes, to a significant extent. A

trade deficit doesn’t produce a one-for-one decline in manufacturing value added, since a

significant share of both exports and imports of goods include embodied services. But a

reasonable estimate is that the deficit surge reduced the share of manufacturing in GDP by

around 1.5 percentage points, or more than 10 percent, which means that it explains more than

half of the roughly 20 percent decline in manufacturing employment between 1997 and 2005.

Again, this is over a relatively short time period and focuses on absolute employment, not the

employment share. Trade deficits explain only a small part of the long-term shift toward a

service economy. But soaring imports did impose a significant shock on some U.S. workers,

which may have helped cause the globalization backlash.

And trade deficits are, as I said, part of a broader story of adjustment issues.

8
4. Rapid globalization and disruption

When trade theorists talk about the distributional effects of trade, they tend to use one of two

models (or classes of models.) Heckscher-Ohlin models treat factors of production as fungible

across industries, so that possible adverse effects involve broad classes of workers, such as

workers without college. Specific-factor models, by contrast, treat factors – definitely capital,

but perhaps labor as well -- as being stuck in particular industries.

It’s possible, and probably even correct, to think of specific factors as representing the short run

while Heckscher-Ohlin represents the long run. How long is the long run? Good question.

The 90s consensus, however, focused almost entirely on Heckscher-Ohlin-type analysis, asking

how the growth of trade had affected the incomes of broad labor classes, as opposed to

workers in particular industries and communities. This was, I now believe, a major mistake –

one in which I shared.

The thing is, anyone who worked on the political economy of trade policy knew that fights over

tariffs look very much as if they come out of a specific-factors world: labor and capital within a

given industry are generally on the same side in trade policy disputes, not on opposite sides as

they would be if they were thinking about the broad factoral distribution of income. It should

have been obvious that the general politics of globalization would reflect that same reality.

9
That is, never mind the question of how trade affects the blue-collar/white-collar wage gap, or

the aggregate Gini coefficient; the politics of globalization were likely to be much more

influenced by the experience of individual sectors that gained or lost from shifting trade flows.

This is where the now-famous analysis of the “China shock” by Autor, Dorn, and Hanson (2013)

comes in. What ADH mainly did was to shift focus from broad questions of income distribution

to the effects of rapid import growth on local labor markets, showing that these effects were

large and persistent. This represented a new and important insight.

To make partial excuses for those of us who failed to consider these issues 25 years ago, at the

time we had no way of knowing that either the hyperglobalization shown in Figure 1 or the

trade deficit surge shown in Figure 2 were going to happen. And without the combination of

these developments the “China shock” would have been much smaller. Still, we missed an

important part of the story.

5. A case for protectionism?

What did the 90s consensus that the adverse effects of globalization were modest miss? A lot.

Developing-country exports of manufactures grew far beyond their level at the time that

consensus emerged. The combination of this rapid growth and surging trade imbalances meant

that globalization produced far more disruption and cost for some workers than the consensus

had envisaged.

10
So does this mean that – not to put too fine a point on it – Trump is right, that a trade war

would be in the interest of workers hurt by globalization?

The answer is, as you might guess, no. This answer is based not so much on some rigid

commitment to free trade at all costs as on the nature of the losses globalization imposed.

Basically, the big problem with surging globalization wasn’t so much changing demand for

broadly defined factors of production as the disruption caused by rapid change. And that rapid

change appears to be largely behind us: many indicators suggest that hyperglobalization was a

one-time event, and that trade has more or less stabilized relative to world GDP. You can even

see that leveling off in Figure 1.

As a result, major disruptions now would be more likely to come from an attempt to reverse

globalization than from leaving the current trade regime in place. At this point millions of

decisions about where to put plants, where to move and take jobs, have been made on the

assumption that the open world trading system will continue. Making that assumption false, by

raising tariffs and forcing a contraction of world trade, would force a whole new wave of

disruption, a whole new set of winners and losers.

So while the 90s consensus on the effects of globalization hasn’t stood the test of time very

well, one can acknowledge that without accepting the case for protectionism now. We might

11
have done things differently if we had known what was coming, but that’s not a good reason to

try turning back the clock.

REFERENCES

Autor, David H., David Dorn, and Gordon H. Hanson. 2013. "The China Syndrome:
Local Labor Market Effects of Import Competition in the United States." American
Economic Review, 103w(6): 2121-68.

Krugman, Paul, 1995: “Growing World Trade: Causes and Consequences,” Brookings Papers on
Economic Activity 1995, no. 1, 327-62.

Subramanian, Arvind and Kessler, Martin. 2013. “The Hyperglobalization of Trade and Its
Future,” Peterson Institute for International Economics Working Paper 13-6

12