Irwin on trade and economic growth.

pdf

Page 1
http://www.nber.org/papers/w25927
Page 2
Recent research has found that trade reforms that significantly reduce import
barriers lead to faster economic growth, on average, although the effect is
heterogeneous across countries.
Douglas A. Irwin and NBER.
Page 3
The decade from 1985 to 1995 was a remarkable period of trade reform in
developing countries, with some papers finding support for the idea that openness to
trade was associated with better economic outcomes. Twenty years have elapsed
since Rodrguez and Rodrik last surveyed the field, and there are several reasons
why this question deserves reexamination. First, more countries have undertaken
trade reforms since the early 1990s, giving us a larger sample of country
experiences.
Page 4
Recent work has employed new and varied empirical methods to study the link
between trade reform and economic growth, including within-country growth
following a reform episode, synthetic control methods, and focus on the channels
through which reductions in trade barriers might improve economic performance. A
consistent finding is that trade reforms have a positive impact on economic growth,
but the effects differ considerably across countries. Research shows that trade
reforms can improve economic outcomes in developing countries, and that the
relationship between policy reform and economic outcomes has improved
considerably since the 1990s. This paper examines the links between trade reform
and economic performance using cross-country regressions, synthetic control, and
empirical country studies.
Page 6
The paper focuses on the impact of trade reforms, meaning unilateral reductions in
trade barriers, on national income. It does not discuss the relationship between trade
and the level of national income, or the impact of free trade agreements or regional
free trade areas. The evidence for increased participation in world trade comes from
developing countries and emerging markets, rather than from advanced countries.
Page 7
2. The Trade Liberalization Wave in Developing Countries
From the 1950s to the 1980s, most developing countries had in place an extensive
array of policies that restricted imports. In the late 1980s and early 1990s, however,
a wave of trade reforms began, with many countries cutting tariffs on imports. The
number of countries that changed their policies from being closed to being open
increased dramatically in the decade after 1985. Figure 2 shows that the average
tariff in developing countries declined steadily in the 1980s, dropped more sharply in
the early 1990s, and continued to fall at a slower pace thereafter. The reduction in
tariffs was not symmetric across regions. Developing countries have a history of
governments allocating foreign exchange to control imports. These foreign exchange
controls are inherently difficult to measure, but the World Bank and the International
Monetary Fund strongly supported these moves by developing countries to open up
their markets. Academic economists were more guarded about the benefits of trade
reform in the 1980s, and noted that systematic attempts at quantification failed to
single out trade policy as a major factor in economic growth. Standard theory
suggests that reducing trade barriers should lead to efficiency gains, but why might it
be expected to increase economic growth as well?
Page 11
Several factors affect a country's economic growth rate, and it is difficult to isolate
individual factors. To understand the impact of trade policies, economists have
compiled empirical evidence, identified trade reform episodes in different ways, and
made various kinds of counterfactual comparisons.
A seemingly straightforward way to evaluate the impact of trade policy on economic
performance is to compare an outcome variable (growth in real per capita income)
under different trade regimes.
Page 12
Sachs and Warner (1995) reduce trade policy to a single binary variable, and tally
the number of new countries that become "open" by this tally. Sachs and Warner
estimate a simple cross-sectional regression that relates economic growth in real per
capita income between 1970 and 1989 to the initial level of income in 1970, the
openness dummy variable (in the 1980s), and other political and economic control
variables. In theory, a tariff reduction should increase welfare or real consumption,
not necessarily real GDP, but it might actually decline if the changing production mix
is evaluated at prereform prices.
Page 13
The measure is crude but useful for placing countries into two different and
meaningful categories. However, it does not capture the quantitative significance of a
country's trade restrictions or the extent to which a country reforms its policy at a
given point in time. The Sachs-Warner indicator is based on five components: tariffs,
quotas, exchange rate distortions, black-market exchange rate premium, and state
monopoly of exports. None of these factors clearly represents a trade policy variable.
Page 14
They conclude that trade policy is not clearly linked to growth outcomes, and that the
black-market premium is not simply representative of macroeconomic distortions. In
countries with overvalued currencies, the use of trade measures to safeguard the
balance of payments and prevent the loss of foreign exchange reserves is a
common occurrence. A currency devaluation encourages exports and discourages
imports, which permits the removal of import licenses and the relaxation of other
quantitative import restrictions.
Page 15
The Sachs-Warner variable no longer separates high-growth from low-growth
countries, but the dating of reform episodes can be used to estimate the within-
country impact of reform on growth and investment. Wacziarg and Welch find that
trade reform had a positive, economically large, and statistically significant impact on
growth and investment.
Page 16
The coefficient of liberalization is 0.6 for 1950-70, 1.8 for 1970-90, and 2.5 for 1990-
98. Countries that opened up in the 1980s and 1990s saw large payoffs. Wacziarg
and Welch find that liberalization increases capital investment and the trade share in
GDP, and that the average effect of liberalization on growth is positive. However,
about half of the countries did not experience more rapid growth after opening.
Page 17
Growth effects could differ markedly across countries for a variety of reasons, and
complementary reforms may be needed to maximize the gains from trade. Feyrer
and Irwin (2019) expand the SWWW sample to include new liberalizing countries,
and include continuous time indicators of current account openness and other
measures of trade policy that reflect the speed and depth of such episodes. Although
many countries adopted reform packages after the SWWW dates, not just their trade
and exchange rate policies, making it difficult to attribute the growth outcomes
exclusively to trade, Wacziarg and Welch focus on 22 relatively "pure" trade
liberalization episodes. Prati, Onorato, and Papageorgiou (2013) use different
indicators for real (trade, agriculture, and networks) and financial (banking, finance,
capital account) policy to compare economic performance.
Page 18
The baseline regression indicates that trade reform and current account reform
independently improve growth prospects, although it is difficult to interpret the
precise meaning of these coefficients.
Page 19
In a cross-country panel regression with country and time fixed effects, researchers
find that economic growth is 1.2 percentage points faster after a trade reform than
before. Overvalued real exchange rates limit the supply response to trade reform.
Estevadeordal and Taylor (2013) argue that not all tariff reductions increase growth,
and focus on the differential growth impact of these different tariffs.
Page 20
They use data on actual tariff changes to compare liberalizing and nonliberalizing
countries and find that reduced tariffs on capital and intermediate goods result in a
0.75 - 1 percentage point increase in economic growth for liberalizers compared with
nonliberalizers. Their results survive several placebo checks and the endogeneity of
the treatment. They also find that the growth acceleration cannot be attributed to
other policy reforms. The Estevadeordal and Taylor findings confirm that using actual
tariff data on intermediate goods leads to similar results. Hausmann, Pritchett, and
Rodrik (2005) identify 80 episodes of rapid acceleration in economic growth that
were sustained for at least eight years during the period 1957 - 92. Most instances of
economic reform do not produce growth accelerations. Jong-A-Pin and De Haan
(2011) argue that the method Hausmann, Pritchett, and Rodrik use to select the
starting dates of the growth accelerations leads to anomalous results and that
economic liberalization precedes growth accelerations.
Page 22
Freund and Pierola (2012) find that periods of rapid export growth are more likely to
occur in open or liberalizing countries, that they are usually preceded by a large
depreciation of the real exchange rate, and that new exports account for much of the
acceleration. Countries that reduce high import barriers usually experience a pickup
in economic growth. The results may not hold for every country, but on average the
results are positive.
Page 23
Winters and Masters (2013) make a critical point: researchers focus on statistical
significance, whereas policymakers care about the distribution of possible outcomes.
In this hypothetical scenario, a 10 percentage point tariff reduction would raise
income by 10 percent.

4. Synthetic Control Methods

Focusing on within-country growth is an improvement over cross-sectional
comparisons, but it still does not get at the key issue of whether growth performance
is stronger than it would have been in the absence of a reform. The synthetic control
method allows for the effect of unobservable confounding factors to vary with time.
Billmeier and Nannicini (2013) apply the synthetic control method to 30 trade
liberalization episodes from 1963 - 2005 and find that trade reforms had a positive
impact on income, but with much heterogeneity across country and time. Indonesia's
economic performance improved dramatically after 1966, when increased growth
could not be attributed to high oil prices, and was 40 percent higher than the
estimated counterfactual after five years and 76 percent higher after 10 years.
Page 26
The results are generally positive in Latin America, with per capita income rising after
economic liberalization in Barbados, Colombia, Costa Rica, and Mexico. The results
are mixed in Africa, with only the early liberalizations having a positive impact. In
Middle East and North Africa, trade reforms were generally positive, but late
liberalizations had a lower payoff.
Page 27
Billmeier and Nannicini (2011) use synthetic control to examine five transition
economies from the former Soviet Union. They find that ten years after liberalization,
real per capita GDP was 44 percent higher in Georgia and almost 100 percent higher
in Armenia compared with each country's synthetic control.
5. Channels of Impact: Tariffs on Intermediate Goods
Rodrguez and Rodrik (2000) suggested that microeconomic evidence could reveal
the channels by which trade policy might affect productivity at the industry or firm
level. Researchers can use firm-level data in developing countries to determine the
economic adjustments that were made in the wake of trade reforms or the opening of
export opportunities. This approach does not rely on an aggregate indicator variable
as a measure of trade policy. De Loecker (2011) argues that the standard method of
measuring productivity (based on revenue deflated by a price index) may be flawed
and that new methods have improved upon past practices.
Page 29
Most of this literature focuses on whether lower tariffs lead to within-industry
efficiency gains as domestic producers face intensified competition. Lower tariffs on
intermediate goods can affect final goods producers through lower prices, increased
quality, and increased variety of inputs. Several papers document the productivity
gains from lower tariffs on final goods, including Pavcnik ( 2002), Tovar (2012), and
Fernandes (2007). The productivity gains are linked to increases in imports of
intermediate inputs, skill intensity, and machinery investments, and to reallocations
of output from less to more productive plants. Brazil's early liberalization in 1988 - 90
led to a 6 percent increase in total factor productivity and a similar impact on labor
productivity. The productivity gains were due to lower input tariffs and increased
exposure to competition.
Page 31
The authors find that importing foreign intermediate goods increases productivity by
2.6 percent, and also find evidence of a positive dynamic effect from the use of
imported materials. Several detailed studies look at India's trade reforms in 1991,
and find that reduced tariffs increased competition in general and forced firms facing
that competition to improve their efficiency. The availability of cheaper inputs was a
more significant driver of productivity than increased final goods competition.
Goldberg et al. (2010) link lower input tariffs to greater product scope and improved
firm performance. They separate changes into a "price" and a "varieties" channel
and find substantial gains from access to new varieties of imported inputs. Many
papers focus on China, another big reforming country. They find that lower tariffs
increase competition in the industry directly facing foreign competition, and reduce
markups, and that lower input tariffs increase efficiency in downstream purchasing
industries. Amiti et al. (2017) found that lower input tariffs increased productivity in
China, and that domestic producers benefited from lower prices, improved quality,
and greater variety of imported inputs. This literature has not succeeded in providing
an aggregate measure of the productivity improvement resulting from this channel. In
contrast, the benefits of achieving greater productivity in labor-abundant countries
that export manufactured goods have been readily apparent.