Journal of International Economics 120 (2019) 1–45

Contents lists available at ScienceDirect

Journal of International Economics

journal homepage: www.elsevier.com/locate/jie

Full length articles

Trading off the income gains and the inequality costs of trade policy☆
Erhan Artuc a, Guido Porto b, Bob Rijkers a,⁎
a
Development Economics Research Group, Trade and Integration, The World Bank, Argentina
b
Universidad Nacional de La Plata, Departamento de Economia, Calle 6 e/ 47 y 48, 1900 La Plata, Argentina

a r t i c l e i n f o a b s t r a c t

Article history: This paper characterizes the trade-off between the income gains and the inequality costs of trade using survey
Received 16 February 2018 data for 54 developing countries. Tariff data on agricultural and manufacturing goods are combined with house-
Received in revised form 19 April 2019 hold survey data on detailed income and expenditure patterns to estimate the first-order effects of the elimina-
Accepted 3 May 2019
tion of import tariffs on household welfare. The paper assesses how these welfare effects vary across the
Available online 17 May 2019
distribution by estimating impacts on the consumption of traded goods, wage income, farm and non-farm family
Research data related to this submission: enterprise income, and government transfers. For each country, the income gains and the inequality costs of trade
https://data.mendeley.com/datasets/ liberalization are quantified and the trade-offs between them are assessed using an Atkinson social welfare index.
7p69kfmghh/draft?a=112f4293-3a03-4249- The analysis finds average income gains from import tariff liberalization in 45 countries and average income
8216-e420775fb67b losses in nine countries. Across countries in the sample, the gains from trade are 1.9% of real household expendi-
ture on average. We find overwhelming evidence of a trade-off between the income gains (losses) and the in-
Keywords: equality costs (gains), which arise because trade tends to exacerbate income inequality: 45 countries face a
globalization trade-off, while only nine do not. The income gains typically more than offset the increase in inequality. In the
inequality
majority of developing countries, the prevailing tariff structure thus induces sizable welfare losses.
welfare
© 2019 Published by Elsevier B.V.
trade integration
tariffs
poverty

1. Introduction In this paper, we combine these two questions and assess the income gains
relative to the inequality costs of trade policy. Using survey data for 54 de-
The recent wave of ‘new’ trade models has rekindled interest in the veloping countries, we explore the potential trade-off between the gains
gains from trade. The results and theorems on the aggregate gains from from trade and the distribution of those gains and we provide a quantifica-
trade in Dixit and Norman (1980, 1986) have been extended by Arkolakis tion of the inequality-adjusted welfare gains from trade. The evaluation of
et al. (2012), Costinot et al. (2012) and Costinot and Rodriguez-Clare this trade-off is important, especially because free trade is often opposed
(2014).1 Concurrently, there has also been a renewed interest in the distri- on inequality grounds.
bution of those gains. These are the focus of Porto (2006), Fajgelbaum and We develop a comprehensive model that describes how trade policy
Khandelwal (2016), Atkin and Donaldson (2015), and Atkin et al. (2018).2 affects the real income of different households. Tariffs determine do-
mestic prices which affect households both as consumers and as income
☆ We thank R. Adao, I. Brambilla, A. Deaton, L. Gasparini, A. Mattoo, B. McCaig, A. Nicita earners. As consumers, households are affected through the cost of the
and M. Olarreaga for comments and discussion, and C. Arkolakis and two anonymous entire bundle of traded consumption goods. Similarly, household in-
referees for constructive feedback. J. Angbazo, S. Fernandez, W. Kassa, H. Liu, A. Luo and come is affected through changes in the returns to household produc-
M. Saleh provided excellent research assistance. This research was supported by the
World Bank's Research Support Budget, the ILO-World Bank Research Program on Job
tion activities, crop growing, family businesses, labor earnings, and
Creation and Shared Prosperity, the Knowledge for Change Program, and the World government transfers. Our model encompasses all these mechanisms.
Bank’s Umbrella Facility for Trade. The findings, interpretations, and conclusions Following Deaton (1989), we use a first-order approximation to mea-
expressed in this paper are entirely those of the authors. They do not necessarily sure how changes in tariffs impact real income.
represent the views of the International Bank of Reconstruction and Development/
We then combine tariff data on various goods with household sur-
World Bank and its affiliated organizations, or those of the Executive Directors of the
World Bank or the countries they represent. All errors are our responsibility. vey data on detailed income and expenditure patterns to estimate
⁎ Corresponding author. these first order welfare effects for 54 low and middle income countries.
E-mail addresses: [email protected] (E. Artuc), With estimates of the welfare effects of import tariff liberalization for
[email protected] (G. Porto), [email protected] (B. Rijkers). each household, we study the aggregate gains from trade (as in
1
See also Artuc et al. (2015), Caliendo and Parro (2015), Melitz and Redding (2015),
Arkolakis et al. (2019), and Caliendo, Dvorkin and Parro (2018).
Arkolakis et al., 2012) and the distribution of the gains from trade (as
2
See also Nicita et al. (2014), Faber (2014), Goldberg and Pavcnic (2007), Topalova in Porto, 2006). Using an Atkinson Social Welfare function (Atkinson,
(2010), Kovak (2013), Autor et al. (2013), Dix-Carneiro and Kovak (2017). 1970), we assess the trade-off between the income gains and the

https://doi.org/10.1016/j.jinteco.2019.05.001
0022-1996/© 2019 Published by Elsevier B.V.
2 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45

inequality costs of trade. Our joint study of the gains from trade and gains, presents robustness tests, and assesses the trade-offs that would
their distribution across households contributes to an incipient strand arise if countries were to undertake protectionist trade reforms instead
of literature including Antras et al. (2017) and Galle et al. (2017). of liberalizing. Section 6 concludes.
It is useful to put our methodological approach into context.
Arkolakis et al. (2012) quantify the aggregate gains from trade by deriv- 2. Tariffs and household welfare
ing a sufficient statistic to compare autarky with the status quo. Subse-
quent literature has developed extensions allowing for imperfect In this section, we develop a model to study the welfare effects of
competition (Arkolakis et al., 2019), labor market frictions (Caliendo tariff changes. We adopt an extended agricultural household model to
et al., 2019), and productivity advantages (Melitz and Redding, 2015). define household welfare (Singh et al., 1986; Benjamin and Deaton,
Work on the distributional effects identifies instead winners and losers 1993) and we derive the welfare effects using first order approxima-
from trade. Much of this literature builds on Deaton's (1989) first-order tions (Deaton, 1989; Porto, 2006; Nicita et al., 2014).
effects approach. Porto (2006) studies the distribution of the household We begin by discussing production decisions. We assume that
welfare effects across the income distribution, Nicita et al. (2014) ex- households are endowed with a fixed amount of resources vh, which in-
plore the poverty bias of trade policy (the welfare effects of the poor rel- clude land or capital, and labor Lh. There is no leisure choice but house-
ative to the welfare effects of the rich), and Atkin et al. (2018) holds can differ in the labor endowment because of differences in family
investigate the distribution of the household welfare effects from FDI. size and composition. Assume for now that these factors can be allo-
Another branch of the literature examines distributional effects in an cated to the production of one (composite) agricultural good or to the
Arkolakis et al. (2012) setting. Fajgelbaum and Khandelwal (2016) in- labor market. The agricultural good i is produced with a constant return
troduce non-homothetic preferences and focus on expenditures only. to scale production function Fi(vh, Lh). The household takes goods prices
Costinot et al. (2012) and Galle et al. (2017) adopt a Ricardo-Roy pi and wages w as exogenous. There is no market for land or capital v,
model and focus on both expenditures and wages. but labor can be traded, i.e., it can be hired in-farm, sold off-farm, or
A distinctive feature of this paper is that we merge these two ap- sold to the labor market. These different types of labor are perfect sub-
proaches by looking at both average gains from trade and their distribu- stitutes. Agricultural profit maximization requires using labor in-farm
tional impacts. As is standard in the literature, we rely on first order (own or hired) up to a point where pi∂Fi/∂L = w. The profit function as-
approximations. We offer a flexible model with extensive household sociated with this optimization problem is π ~ i ðpi ; w; vÞ. In this formula-
heterogeneity in incomes and expenditures. As in Nicita et al. (2014), tion, household income yh is the sum of maximized profits π ~ hi and the
we allow for a more comprehensive set of sources of income heteroge- value of the labor endowment wLh. To simplify the exposition that fol-
neity than in most other papers. As in Fajgelbaum and Khandelwal lows, let wh be the labor income that household h derives only from
(2016), we allow for non-homothetic preferences and heterogeneity the labor market and let πh be maximized profits defined net of hired
in expenditure patterns. To operationalize this flexible framework, we off-farm labor only.3 Allowing for many goods (Singh, Squire and
need to impose some structure on our model. Our setting is compatible Strauss, 1986), total maximized household income yh is
with perfect competition, constant returns to scale, and homogeneous

 X
products.
yh p; vh ¼ wh þ π hi ðpÞ−T h þ Ωh ; ð1Þ
We find average income gains from import tariff liberalization in 45 i
countries and average income losses in the remainder 9 countries. On
average, the developing countries in our study enjoy gains from trade where p is the vector of prices pi, πhi are farm enterprise profits obtained
equivalent to 1.9% of real household expenditure. This is mostly because from the sales of good i (such as cotton, tobacco, beans or maize), and Th
the consumption gains from lower prices dominate the income losses are taxes paid to (or transfers received from) the government. All other
from reduced protection. potential sources of income, such as remittances, gifts, other types of
The distributional impacts of import tariff liberalization are public transfers (e.g. pensions) and income from non-traded household
highly heterogeneous, across both countries and households. We enterprises are included in Ωh.4
find that the equality gains, the change in social welfare associated Household h maximizes a utility function defined over a vector of
with these distributional impacts, are negatively correlated with consumer goods c, uh = u(c), subject to a vector of given prices p and
the average income gains. Inequality costs arise primarily because total household income. Assuming households are price takers in con-
trade exacerbates nominal income inequality, while the consump- sumer markets, in production and in the labor market, the optimization
tion gains tend to be more evenly spread. This creates trade-offs be- problem is recursive because production decisions are independent of
tween the income gains and the equality gains in 45 of the 54 consumption decisions. Thus, households maximize u(⋅), subject to p
countries in our sample. The income gains typically more than offset and maximized income yh. The solution to this optimization problem
the increase in inequality. In 39 countries, liberalization of import delivers a demand function for each good. Optimal consumption of
tariffs would result in inequality-adjusted welfare gains for a wide good i is chi and, given required utility uh, the household expenditure
range of empirically plausible values of inequality aversion (between function is
1 and 2). In 9 countries that face trade-offs, protectionism would in-

 X
stead be welfare enhancing for plausible values of inequality aver- e p; uh ¼ pi chi ðpÞ ð2Þ
sion. Finally, there are 6 countries where the trade-offs are acute, in i
which the presence of welfare gains or losses depends crucially on
the presumed level of inequality aversion and policy prescriptions To derive the welfare effects of trade, we use the concept of compen-
are consequently more equivocal. These results imply that in the ma- sating variation CVh. For price changes, this is generally done using the
jority of developing countries in our study, the prevailing pattern of expenditure function eh. In our case, we need to consider the fact that
protection induces sizeable welfare losses.
The rest of the paper is organized as follows. Section 2 sets up the 3
In particular, let Lhired be hired labor in-farm, Lmarket the supply of household labor to
model and derives the formulas for the welfare effects of trade policy. the labor market, and let Lfarm be the own-farm family labor. Then, the labor endowment
Section 3 uses the tariff data and the survey data to estimate those wel- is Lh = Lmarket + Lfarm, while farm employment is LD = Lfarm + Lhired. In addition, πh = piqi −
fare effects in 54 countries. Section 4 discusses the gains from trade and wLhired and household income is wh + πh. See Benjamin (1993) for a full derivation.
4
Because of data constraints, we do not deal with savings, debt, inventories, and other
their distribution. Section 5 evaluates and quantifies the trade-off be- intertemporal considerations. Because very few surveys include detailed input expendi-
tween income gains and inequality costs of trade. It also decomposes ture data in the income modules, we do not deal with imported input prices (seeds, fertil-
equality gains into consumption equality gains and income equality izer) either.
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 3

trade affects nominal income yh as well. Consequently, we follow Dixit is perfectly immobile across sectors; (v) the loss of public revenue due
and Norman (1980) and Anderson and Neary (1996) and use the to tariff cuts is compensated with increases in income taxes. Then, the
trade expenditure function, Vh: estimable welfare effects are given by:




V h p; vh ; uh ¼ yh p; vh −e p; uh ; ð3Þ
h
dV i

  τ
i
¼ − ϕhi −shi þ ϕhwi þ Ψhi ; ð5Þ
eh 1 þ τi
which depends on prices p via the maximized nominal income function
yh(⋅) and the expenditure function e(⋅). Note that the trade expenditure where now ϕhwi is the share of labor income derived only from wages in
function is usually defined in the Hicksian tradition as eh − yh but we sector i (and not other sectors) and Ψhi is the increase in income tax
work with (3) instead so as we can interpret the results as changes in accrued by the household.
real household income (Porto, 2006). This expression is the welfare effect of a simulated full unilateral tar-
We proceed in two steps. We first derive general welfare effects of iff liberalization. This means that the country reduced its own tariffs in-
price changes in Proposition 1. Then, in Proposition 2, we derive estima- dividually. Under full import tariff liberalization, so that dτi = − τi, the
ble welfare effects of trade policy. In particular, we list additional perfect pass-through assumption implies that
assumptions that we need to impose in order to obtain estimates of
the welfare effects that are compatible with our data. This proposition d ln pi ¼ −τ i =ð1 þ τi Þ: ð6Þ
imposes restrictions on the price changes and on household impacts
that we can accommodate in our data. Changes in these assumptions The unitary pass-through elasticity requires constant returns to scale
allow for different responses and different welfare effects. We discuss in the production of the traded goods and perfect competition.7 This is a
below some salient alternative model formulations, and present robust- simplification of our analysis that is rooted in the lack of data needed to
ness tests in Section 5.6. estimate the pass-through elasticities for a broad range of products and
countries (see Nicita (2009), Ural Marchand (2012), Atkin et al. (2018),
Proposition 1. Assume the household is a price taker in consumer, pro-
for estimates of imperfect pass-through for selected countries). Note
ducer and labor markets. Given the income generating function y(p, vh)
that by simulating cases of own unilateral tariff liberalization, we are
in Eq. (1) and the expenditure function e(p, uh) in Eq. (2), the impact of a
not accounting for the effects of foreign tariff reductions and market ac-
price change on household welfare dVhi (as a share of household initial
cess effects.
expenditure eh) is given by
The sector specificity of labor allows us to derive a simple wage-
h 
  h price elasticity, because with fixed labor d ln w = d ln pi for wages in
dV i ∂wh pi dT
¼ ϕhi −shi þ ϕhw d lnpi − h ð4Þ sector i and d ln w = 0 for wages in all sectors j ≠ i. It is in principle pos-
eh ∂pi wh e
sible to accommodate different assumptions on how labor markets
work. Under full labor mobility, for example, labor would reallocate
where shi is the share of traded good i in the consumption bundle of
until a new equilibrium is reached. In the literature, such a model has
household h, ϕhi is the income share derived from the sales of good i,
been estimated by Porto (2006), Nicita (2009) and Nicita et al. (2014).
and ϕhw is the share of labor income. dVhi is the negative of the compen-
In our model, furthermore, labor is homogeneous and, in particular,
sating variation CVh, the monetary transfer that would allow household
there is no skill differentiation. Nicita et al. (2014) estimate
h to attain the same utility uh before the trade shock and the price
wage-price elasticities for skilled and unskilled labor separately. An-
change.
other possibility is to assume imperfect labor mobility and equilibrium
Eq. (4) follows from taking the derivative of (3) with respect to pi inter-industry wage differences. In this case, a price shock can trigger
and expressing the resulting expression in proportional terms. Thus, labor reallocation responses across sectors and, consequently, there
the proportional price change is d ln pi = dpi/pi. Hotelling's Lemma can be sector-specific wages elasticities. Artuc et al. (2015) provide esti-
(i.e., the envelope theorem applied to the profit function) implies that mates of such a model for a wide range of countries. In Section 5.6 we
(∂πi/∂pi)dpi = qhi dpi, where qhi is the quantity produced of good i. Multi- assess how our results change if we do not account for labor market re-
plying and dividing by pi and expressing the result relative to total in- sponses to tariff changes.
come yh gives (piqhi /yh)d ln pi = ϕhi d ln pi. From Shephard's Lemma, The interpretation of Eq. (5) is straightforward. After a price change
the derivative of the expenditure function with respect to pi is the quan- caused by tariff cuts d ln pi = − τi/(1 + τi), the first order effects on real
tity consumed chi so that (∂e/∂pi)dpi = chi dpi and (pichi /eh)d ln pi = shi d ln income can be well-approximated with the corresponding expenditure
pi.5 This accounts for the first term within brackets in (4). The second and income shares. In the language of Deaton (1989), because we are
term captures the labor income impacts, which are given by the product working with tariff cuts and price declines, net-consumers benefit
of the share of income derived from labor, ϕhw, and the wage elasticity while net-producers suffer. In our setting, the net position of a house-
with respect to the price (∂wh/∂pi)/(pi/wh). This formulation allows for hold is defined in an extended model including not only consumption
different assumptions about the functioning of labor markets that deter- and production of traded goods but also labor income and government
mine the nature of the labor income elasticities.6 The last term in (4) ac- transfers.
counts for any impacts on government transfers received or on taxes As in Nicita et al. (2014), we want a measure of the welfare effects
paid by household h. generated by the entire structure of tariff protection. To obtain it, we
sum the changes in welfare in (5) over all traded goods i to get:
Proposition 2. Let τi be the level of tariff protection in sector i. Assume:
(i) goods are homogeneous; (ii) the country is small and thus faces h X dV h
exogenously given international prices pi∗; (iii) perfect price transmis- ^ h ¼ dV ¼
V i
; ð7Þ
eh eh
sion from tariffs to domestic prices; (iv) labor is specific, that is, labor i

h
^ is the proportional change in household real income. In the re-
where V
mainder of the paper, we estimate the different components of Eq. (7)
5
This assumes that, ex-ante, household expenditures equal household income. This
rules out savings and intertemporal considerations.
6
Since we do not have information on input use, Eq. (4) omits the indirect effects of
wages to hired in-farm labor, Lhired on farm profits. See Porto (2005) for a study of those 7
See Goldberg and Knetter (1997) for a exhaustive discussion of the conditions needed
effects in Moldova. for perfect pass-through. Note also that there is no role for entry and exit in our model.
4 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45

and study them in detail. We also use Eq. (7) to build ex-post counter- we use and improve upon the templates and concordances developed
factual distributions. Let xh0 be the observed, ex-ante level of real house- by Nicita et al. (2014). In short, we first aggregate goods in the house-
hold income (from the data compiled in the household surveys). The hold surveys to 2-digit and 4-digit categories. We then aggregate tariff
h
counterfactual real income ^x1 is and trade data from COMTRADE to those categories. See Appendix A
for details.


^h :
^xh1 ¼ xh0 1 þ V ð8Þ
3.1. Trade policy and price changes

Much of what we do below hinges on the comparison of the ex-ante In the empirical application that follows, good i represents one of the
and ex-post distributions of income. product classifications from the expenditure, income and home-
Since our propositions are derived from an agricultural household consumption templates modules of the household surveys. Each of
model, which is not often used in the trade literature, it is important these classifications includes many finer product groups from the HS
to bear in mind some of the key features of our approach when classification. We compute weighted average tariff rates τi for each of
interpreting the results. If we take the standard models at face our survey categories:
value, it is clear that several data limitations would prevent us from
X mc;n
fully accurately estimating the welfare effects of trade stemming τi ¼ τc;n X ; ð9Þ
from those models. These limitations are inherent to household sur- c;n∈i mc;n
veys. For example, information on returns to capital, especially cor- c;n∈i

porate profits and so on, is often patchy and typically missing

altogether in the household surveys. Consequently, it is very difficult where n is an HS-category that belongs to survey-category i and mc, n are
to map the results from those standard models to our empirical re- imports of good n from country c. The results are shown in Table 1. We
sults which are based on household surveys. There might therefore report the average tariff for our 2-digit classification, Staple Agriculture,
be some discrepancies between some of our results and standard Non-Staple Agriculture, and Manufactures. Average tariffs are highest
trade theory. For example, Stolper-Samuelson effects resulting in dif- for non-staple agricultural goods (14.4%). They are lower for staple ag-
ferential impacts on the returns to capital vs labor, or to skilled vs un- ricultural goods (10.8%) and manufactures (10.9%). These averages
skilled labor, may not be captured by our analysis. However, most of mask substantial variation in trade barriers across countries. Average
the discussion about poverty, inequality and household welfare is (to tariffs on non-staple agricultural goods range from as high as 46.1% in
a large extent) based on household surveys, which remain the dom- Bhutan to as low as 1.9% in Indonesia. Countries with higher tariffs in ag-
inant instrument for measuring household consumption and income riculture (staple and non-staple) tend to have higher tariffs on manufac-
portfolios. Such surveys are thus a natural starting point for analyz- tures as well.
ing the distributional impacts of trade policy. Using the full price transmission assumption (Eq. 6), we calculate
Our results also have interesting implications for trade theory. We the price changes induced by the elimination of tariffs as follows:
argue that the most innovative feature of our approach is household
heterogeneity, which is often missing in standard trade models. pi −pi ð1 þ τi Þ τi
Δ ln pi ¼ ¼− : ð10Þ
Heterogeneity takes two forms. One is consumption heterogeneity pi ð1 þ τi Þ 1 þ τi
and non-homothetic preferences. In particular, since the poor and the
rich allocate their expenditure very differently across goods (most nota-
bly, food), this can have implications for the overall gains from trade as 3.2. Expenditure and income shares
well as for the inequality costs of trade. These differences and their im-
plications are not captured by standard models based on homothetic To measure the first order welfare effects we retrieve the expendi-
preferences. There are, of course, papers that rely on non-homothetic ture shares shi and the income shares ϕhi and ϕhw from the household sur-
preferences, and the conclusions from those models are consistent veys. Appendix Table A1 lists the countries included in the analysis,
with the conclusions from our paper. together with the corresponding household survey, the year of the sur-
The other important dimension of heterogeneity is income hetero- vey and the sample size. Our analysis covers all low income countries for
geneity. In our model, households earn income from various sources which appropriate household survey data were available, as well as the
not limited to labor markets. In particular, income from sales of different majority of lower middle income countries. To minimize the role of
agricultural products plays a crucial role in agrarian economies and measurement error, we exclude households in the top and bottom
these sales include returns to non-labor household factors such as 0.5% of the status quo expenditure distribution in all our analyses. All
land and assets such as tractors, ploughs, and more humble tools such statistics derived from household surveys presented in the remainder
as shovels, carts, and wheelbarrows. The “distribution” of these factors of the paper are weighted using survey weights. For the relatively few
across households in the surveys is highly heterogeneous. Households surveys for which survey weights are not available, we simply assume
thus have very different sources of incomes and this matters for welfare each household has the same weight.
and inequality. This income heterogeneity is missing in many neoclassi- Expenditure shares are reported in Table 2. We show averages for six
cal models. major expenditure aggregates, namely Staple Agriculture, Non-Staple
Agriculture, Manufactured Goods, Non-Traded Goods, Other Goods,
3. Estimating the welfare impacts of trade policy and Home Consumption. Expenditure on food is the dominant expendi-
ture category, accounting on average for 45% of all household spending
To estimate the welfare impacts of trade policy, we need to measure across countries, which is not surprising since the bulk of countries in
the different components of Eqs. (5) and (7). The data needed to esti- our sample are low income countries with an average poverty rate of
mate impacts on consumption and production of traded goods, labor in- 35% (using national poverty lines) and an average GDP per capita of
come and home enterprise income can be found in standard household US$ 1879. This focus on poor countries also helps explain why home
surveys. Trade and trade policy data come from United Nations consumption is important, accounting for an average budget share of
COMTRADE and UNCTAD TRAINS, which classify goods using the Har- 17% across countries and for more than a third of all expenditure in
monized System (HS) so that tariffs and imports are available at HS-6 Ethiopia, Madagascar, Mali and Uzbekistan. Spending on manufacturing
level. Household surveys use different nomenclatures of goods pro- goods on average accounts for 17% of overall household expenditure,
duced and consumed. To match trade data and household survey data, and spending on non-tradables accounts for 15%.
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 5

Table 1 Table 2
Average tariffs. Expenditure shares.

Country Staple Agric. Non-Staple Agric. Manufactures Country Staple Non-Staple Manuf. Non-Traded Other Home
Agric. Agric. Cons.
Benin 12.2 16.9 10.8
Burkina Faso 12.0 18.3 9.3 Benin 34.4 3.8 23.3 10.7 6.1 21.6
Burundi 23.8 21.6 10.8 Burkina Faso 24.4 12.3 16.1 8.8 8.3 30.1
Cameroon 13.8 22.5 23.0 Burundi 41.8 9.9 20.2 12.7 10.8 4.6
Central African Rep. 16.6 23.7 21.8 Cameroon 46.8 6.1 17.1 14.7 5.9 9.4
Comoros 1.8 10.4 8.9 Central African 40.4 18.5 21.3 7.9 0.2 11.7
Côte d'Ivoire 10.4 10.2 9.2 Rep.
Egypt, Arab Rep. 7.1 28.0 18.0 Comoros 48.1 9.5 10.8 17.3 5.2 9.2
Ethiopia 10.1 13.3 12.4 Côte d'Ivoire 35.7 3.9 22.2 20.5 6.5 11.3
Gambia, The 6.6 13.5 13.9 Egypt, Arab Rep. 45.5 4.9 13.8 31.4 2.0 2.4
Ghana 16.4 11.6 14.3 Ethiopia 23.1 9.1 17.0 2.9 10.1 37.7
Guinea 13.9 18.9 9.5 Gambia, The 45.3 11.5 11.3 12.0 10.4 9.6
Guinea-Bissau 13.5 15.7 12.8 Ghana 7.7 1.4 30.8 33.0 15.5 11.5
Kenya 18.7 25.1 11.0 Guinea 33.0 11.9 18.3 12.6 5.0 19.2
Liberia 6.3 5.6 16.4 Guinea-Bissau 50.7 6.3 6.6 7.4 4.2 24.7
Madagascar 8.3 9.6 14.8 Kenya 30.2 9.7 23.4 24.9 2.4 9.4
Malawi 8.2 22.0 9.3 Liberia 47.1 7.2 12.4 15.2 2.5 15.6
Mali 11.2 16.8 8.8 Madagascar 37.2 7.2 12.0 3.6 0.7 39.4
Mauritania 9.2 14.8 15.9 Malawi 25.8 5.7 29.1 6.9 0.7 31.8
Mozambique 8.8 13.9 7.4 Mali 25.6 7.6 4.2 4.9 0.5 57.1
Niger 12.2 17.6 9.3 Mauritania 47.2 11.5 14.6 6.7 0.7 19.3
Nigeria 11.3 19.8 11.0 Mozambique 44.7 5.3 14.7 3.9 1.5 29.9
Rwanda 21.0 30.1 11.0 Niger 35.7 8.8 17.1 6.5 10.2 21.7
Sierra Leone 11.8 16.2 9.7 Nigeria 47.9 3.6 18.0 9.4 0.5 20.6
South Africa 7.1 6.4 16.8 Rwanda 24.3 4.9 10.6 9.0 29.0 22.2
Tanzania 12.6 29.1 10.7 Sierra Leone 46.2 10.4 12.4 10.8 4.4 15.8
Togo 11.6 18.6 9.5 South Africa 31.6 8.3 31.8 16.4 11.8 0.1
Uganda 11.4 29.7 10.0 Tanzania 29.4 6.6 19.1 9.9 6.2 28.8
Zambia 17.1 19.7 6.8 Togo 39.0 7.8 15.1 26.2 5.8 6.1
Armenia 6.9 7.3 6.7 Uganda 24.2 7.6 16.4 17.9 2.0 31.9
Bangladesh 7.4 4.9 18.8 Zambia 53.5 4.8 6.3 10.1 0.6 21.8
Bhutan 43.7 46.1 23.5 Armenia 55.5 8.0 7.1 21.2 0.0 8.2
Cambodia 13.0 6.4 10.1 Bangladesh 45.3 9.0 14.1 16.2 4.4 11.0
Indonesia 6.0 1.9 6.1 Bhutan 26.9 7.2 25.5 15.8 12.4 12.3
Iraq 5.0 5.0 5.0 Cambodia 31.2 12.4 16.0 18.8 8.5 13.0
Jordan 7.9 18.6 8.3 Indonesia 29.3 11.7 11.4 22.8 13.5 11.3
Kyrgyz Republic 6.1 6.1 4.0 Iraq 32.3 5.2 35.2 23.0 3.4 0.8
Mongolia 5.3 6.5 4.9 Jordan 35.1 15.2 19.1 29.1 1.2 0.2
Nepal 9.0 11.7 13.9 Kyrgyz Republic 42.3 5.5 25.5 13.6 3.4 9.7
Pakistan 3.7 8.1 17.4 Mongolia 47.6 8.9 14.3 8.8 1.1 19.3
Papua New Guinea 4.7 12.4 0.9 Nepal 27.3 4.8 11.7 27.6 4.7 23.9
Sri Lanka 7.8 16.3 15.3 Pakistan 28.2 7.7 23.4 12.9 6.6 21.3
Tajikistan 7.4 5.8 8.3 Papua New 36.2 12.2 5.8 5.0 13.7 27.1
Uzbekistan 14.8 11.4 8.5 Guinea
Vietnam 11.1 6.3 9.8 Sri Lanka 32.6 10.2 9.4 19.4 21.7 6.7
Yemen, Rep. 4.4 7.6 7.7 Tajikistan 37.8 5.5 24.8 14.8 3.2 13.9
Azerbaijan 5.7 4.0 10.4 Uzbekistan 36.5 5.1 7.5 10.5 1.9 38.5
Georgia 6.0 6.4 0.5 Vietnam 37.3 6.5 19.6 15.3 10.6 10.7
Moldova 7.9 10.7 3.3 Yemen, Rep. 39.2 20.4 17.5 15.5 4.4 3.1
Ukraine 4.8 5.1 4.8 Azerbaijan 51.1 5.9 20.9 11.6 1.6 9.0
Bolivia 11.0 12.6 15.1 Georgia 34.1 7.8 23.7 27.6 4.7 2.1
Ecuador 14.4 15.4 14.0 Moldova 16.4 2.2 32.1 15.3 7.0 27.1
Guatemala 10.3 10.2 7.4 Ukraine 44.8 11.4 20.0 16.3 0.1 7.4
Nicaragua 12.1 9.8 9.1 Bolivia 44.0 7.7 16.8 23.9 1.3 6.4
Average 10.8 14.4 10.9 Ecuador 42.2 3.9 16.8 21.5 8.5 7.2
Pop. weighted average 9.0 12.1 11.8 Guatemala 37.7 5.3 19.8 17.8 4.8 14.6
GDP weighted average 8.1 10.2 10.9 Nicaragua 40.8 4.9 16.6 19.1 1.0 17.7
Average 37.0 8.0 17.4 15.1 5.8 16.6
Notes: Authors' calculations based on United Nations COMTRADE and UNCTAD TRAINS
Pop. weighted av. 35.4 7.9 17.5 15.9 6.6 16.6
data. The average tariff is expressed in percentage points.
GDP weighted av. 35.9 8.0 18.5 18.3 7.5 11.7

Notes: Authors' calculations based on household survey data. The average expenditure
shares are expressed in percentage points.
Average income shares for staple Agriculture income, Non-Staple
Agricultural income, Wages, Family Enterprise Income, Other income, and heterogeneity in survey design (including coverage of different
and Own Home Production are reported in Table 3. Wage income is sources of incomes and expenditures).
the single most important source of income, accounting on average for
29% of household income across countries, and for 39% if we weigh 3.3. Labor income and transfers
countries by their GDP, suggesting that the importance of this source
of income increases with development. The value of autoconsumption As established in Proposition 2, we assume that labor is sector spe-
accounts for 23% of household income. Profits from running farms and cific. This is consistent with a short-run model, in which households
other family businesses account for 17 and 13% of household income, do not adjust labor to the trade shock. To implement this assumption
respectively. These averages hide important heterogeneity across coun- empirically, we consider 10 different sectors (see the Income Template
tries, which reflects differences in structural features of their economies in the Appendix). This is convenient because in this setting the changes
6 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45

Table 3 Lastly, we need to derive the cost of tax payments needed to com-
Income shares. pensate for the tariff revenue loss, Ψhi . We assume that the government
Country Staple Non-Staple Wages Family Other Home imposes a proportional income tax to do so at the moment it liberalizes.
Agric. Agric. Enterp. Cons. Denoting import quantity by mi, we can approximate the loss of tariff
Benin 14.2 10.0 13.1 0.0 40.9 21.8 revenue as dRi = − τipi∗mi (ignoring production and consumption
Burkina Faso 19.1 2.9 13.4 17.7 12.9 34.0 responses). With a proportional income tax, the change in income tax
Burundi 39.5 29.4 8.1 7.5 11.0 4.5 paid by household h is dTh = dψyh, where dψ is the compensatory
Cameroon 15.4 0.1 27.3 23.1 0.0 34.1
change in the tax rate. Consequently,
Central African 42.5 9.3 2.4 3.4 4.3 38.1
Rep,
τi M
Comoros 24.2 3.7 26.7 16.4 11.1 17.8 Ψhi ¼ − Xi ; ð11Þ
Côte d'Ivoire 7.1 13.7 17.0 28.4 15.7 18.2 1 þ τi yh
Egypt, Arab Rep. 6.9 6.9 41.1 15.1 29.8 0.2 h
Ethiopia 14.3 0.5 5.2 24.3 10.9 44.8
Gambia, The 2.7 6.7 45.7 21.9 8.6 14.4 where Mi = pi∗(1
+ τi)mi is the value of imports. In the robustness
Ghana 8.5 5.5 58.5 0.0 12.0 15.6
exercises, we consider two additional cases, one where there is no com-
Guinea 17.5 3.2 7.0 18.2 13.5 40.5
Guinea-Bissau 5.5 21.9 21.7 7.8 10.4 32.8 pensation of the revenue losses via the income tax and another where
Kenya 21.8 3.1 35.4 5.3 17.8 16.6 there is progressivity in the income tax system.
Liberia 10.2 3.4 22.2 29.2 9.9 25.1
Madagascar 27.1 3.0 23.0 13.1 5.1 28.8 4. Income gains and inequality costs of trade policy
Malawi 18.4 4.6 21.4 12.7 3.8 39.0
Mali 8.7 2.8 8.3 10.5 15.4 54.3
Mauritania 13.4 0.0 3.7 10.1 30.8 42.0 In this section, we investigate the potential income gains (or losses)
Mozambique 10.4 7.1 15.1 10.4 10.0 46.9 and the potential inequality costs (or gains) from import tariff liberali-
Niger 17.3 3.1 4.0 1.5 38.2 35.9 zation. The next section (Section 5) investigates the potential trade-off
Nigeria 15.5 5.6 33.2 10.3 4.6 30.8
between the two.
Rwanda 10.5 3.7 24.7 2.5 11.8 46.7
Sierra Leone 18.7 4.6 11.0 13.3 19.6 32.7
South Africa 0.6 0.0 54.9 0.0 43.6 0.8 4.1. Income gains from trade
Tanzania 10.9 3.0 23.0 5.5 11.6 46.0
Togo 8.7 6.5 30.2 37.2 9.7 7.7 To be consistent with the literature (e.g., Arkolakis et al., 2012), the
Uganda 9.7 2.9 21.6 18.7 13.9 33.1
Zambia 5.7 1.7 21.0 13.3 18.8 39.5
gains from trade G are defined as the proportional change in aggregate
Armenia 9.2 0.1 35.1 6.5 40.6 8.5 household real expenditures, after import tariff liberalization:
Bangladesh 33.0 2.1 31.4 14.3 12.1 7.1
X

Bhutan 12.9 0.0 44.2 9.3 8.6 25.1 xh1 −xh0
Cambodia 24.2 0.5 30.8 23.6 5.3 15.6 X xh ^ h
Indonesia 4.6 1.2 38.2 0.6 20.9 34.4 G¼ h
X ¼ X0 V ; ð12Þ
Iraq 8.1 1.6 49.2 11.9 28.4 0.8 xh0 h xh0
Jordan 1.7 2.1 45.2 8.9 41.0 1.0 h h
Kyrgyz Republic 12.2 1.4 40.4 12.0 27.4 6.6
h
Mongolia 10.1 0.3 38.1 8.7 31.6 11.1 ^ is the proportional change in real expenditures of household h
where V
Nepal 4.1 1.2 25.8 11.1 22.1 35.7
Pakistan 7.6 3.1 45.9 12.1 13.8 17.5
which we estimate with Eq. (7). Thus, G is a weighted average of the
Papua New 13.8 6.5 14.8 9.6 17.9 37.2 welfare effects V^ h.
Guinea
Table 4 reports G for 45 countries with positive aggregate gains from
Sri Lanka 13.1 4.6 48.8 19.3 0.0 14.2
Tajikistan 0.9 1.5 38.7 8.5 22.4 28.0 trade (G N 0). On average, the net gain from import tariff liberalization is
Uzbekistan 7.4 0.2 20.3 11.2 20.7 40.2 a 2.5 percentage point increase in real expenditures. The highest gains
Vietnam 21.1 3.5 35.3 19.6 13.1 7.4 accrue to Cameroon and Zambia (6.9 and 5.9% of real expenditure, re-
Yemen, Rep. 7.8 9.2 43.7 15.3 21.2 2.8
spectively). The smallest gains, for Bangladesh, Burundi, and Mongolia,
Azerbaijan 28.8 1.9 26.1 2.9 26.2 14.1
Georgia 7.3 1.9 29.2 7.8 51.9 1.9
are about 0.5, 0.4 and 0.1% of initial expenditures, respectively.
Moldova 5.5 2.1 30.4 1.7 26.3 33.9 Table 5 reports 10 countries in which import tariff liberalization
Ukraine 2.8 0.0 43.5 0.1 48.0 5.6 causes losses (G b 0) which average − 0.9% of real expenditures. In
Bolivia 6.3 7.6 36.1 27.3 16.2 6.5 Cambodia, the country with the largest loss, households are estimated
Ecuador 10.6 1.1 48.4 16.7 17.3 5.8
to lose 3.1 percentage points of real expenditure. There are also in-
Guatemala 6.4 2.9 45.2 18.0 14.1 13.4
Nicaragua 10.9 2.8 40.4 18.4 13.5 14.0 stances of very small, almost negligible, losses as in Rwanda.
Average 12.9 4.2 29.0 12.5 18.6 22.8 Across all countries in the sample, the average gain from trade liber-
Pop. weighted av. 12.8 3.3 33.0 11.2 17.0 22.7 alization is equal to 1.9% of real expenditures. The developing world
GDP weighted av. 9.4 2.7 39.0 8.1 21.6 19.3 seems to gain from trade.
Notes: Authors' calculations based on household survey data. The average income shares To establish the sources of the gains from trade, we decompose the
are expressed in percentage points. average gains into different channels in columns 2–8 of Tables 4 and
5. Households gain on the expenditure side, but they lose on the income
in prices transmit one to one to nominal wages and the elasticity of the side. The consumption gains come from lower prices of tradables, which
wage in sector i with respect to its own price pi is one, while the elastic- on average result in (gross) real income gains of 6.4% for the winners
ities with respect to other prices j is zero. In robutsness tests, we also (Table 4) and 5.3% for the losers (Table 5). About two-thirds of these
consider a model without labor income responses (as in Deaton, 1989).8 gains, on average, are due to lower prices of agricultural goods and
one-third to lower prices of manufacturing goods. This is a consequence
both of the higher expenditure shares on food items in developing
8
In additional robustness exercises that are not reported here to conserve space we also countries, and the comparatively high tariffs on agricultural products.
estimate a model with perfectly mobile labor (as in Porto, 2006) as well as a model with
imperfect labor mobility (as in Artuc et al., 2010). The results are qualitatively consistent
Households lose nominal income. Agricultural income losses account
with those obtained using the baseline model with specific labor and the alternative for average real income declines of 1.5% across countries with gains
model without labor income responses. and 2.0% across countries with losses. Wage income effects create losses
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 7

Table 4
Gains from trade - winners.

Gains Expenditure Income

Agric. Manuf. Total Agric. Wage Enter. rev. Total

Cameroon 6.9 8.9 3.6 12.5 −1.7 −1.3 −0.7 −1.9 −5.6
Zambia 5.9 7.8 1.2 9.0 −1.0 −0.5 −0.7 −0.8 −3.0
Sierra Leone 4.3 5.8 1.7 7.4 −1.5 −0.1 −0.1 −1.5 −3.1
Tanzania 4.2 4.7 4.1 8.8 −1.7 −1.1 −0.2 −1.7 −4.6
Central African Rep, 4.2 7.1 3.7 10.8 −4.5 −0.0 0.0 −2.1 −6.6
Jordan 4.0 6.2 2.1 8.3 −0.3 −0.4 −0.1 −3.4 −4.2
Mozambique 3.7 6.0 1.2 7.2 −1.1 −0.3 −0.1 −2.0 −3.6
Uzbekistan 3.5 5.0 1.9 7.0 −0.7 −1.1 −0.4 −1.2 −3.4
Côte d'Ivoire 3.4 4.6 2.6 7.2 −1.7 −0.6 −0.4 −1.1 −3.8
Nigeria 3.3 6.2 2.2 8.3 −1.6 −1.8 −0.3 −1.3 −5.0
Ukraine 3.2 3.7 0.9 4.6 −0.2 −0.2 −0.0 −0.9 −1.3
Ecuador 3.0 5.9 1.5 7.3 −1.6 −1.4 −0.3 −1.0 −4.4
Kenya 2.9 6.1 2.5 8.6 −2.8 −1.2 −0.1 −1.6 −5.7
Egypt, Arab Rep. 2.9 4.7 1.9 6.6 −1.3 −1.1 −0.4 −1.0 −3.7
Guinea 2.8 4.9 2.9 7.8 −1.9 −0.1 −0.2 −2.9 −5.0
Bolivia 2.8 4.2 2.3 6.5 −1.2 −0.6 −0.6 −1.3 −3.7
Yemen, Rep. 2.7 4.1 1.3 5.4 −0.8 −0.4 −0.1 −1.5 −2.8
Azerbaijan 2.5 3.9 2.3 6.2 −2.3 −0.1 0.0 −1.3 −3.7
Armenia 2.5 3.8 0.4 4.1 −0.5 −0.3 −0.0 −0.9 −1.7
South Africa 2.5 1.2 2.9 4.2 −0.0 −1.0 0.0 −0.7 −1.7
Malawi 2.4 4.1 2.8 6.9 −2.4 −0.6 −0.3 −1.2 −4.5
Pakistan 2.4 2.0 3.7 5.7 −1.4 −0.8 −0.3 −0.8 −3.3
Benin 2.3 4.8 2.9 7.7 −2.0 −0.2 −0.0 −3.2 −5.4
Ethiopia 2.2 3.7 3.6 7.3 −2.9 −0.0 −0.7 −1.4 −5.1
Togo 2.1 5.3 1.8 7.1 −0.9 −1.0 −0.9 −2.3 −5.0
Guinea-Bissau 2.0 4.7 0.8 5.5 −0.9 −0.3 −0.0 −2.3 −3.5
Tajikistan 2.0 3.3 1.4 4.7 −0.2 −0.6 −0.0 −2.0 −2.8
Uganda 2.0 5.4 1.1 6.6 −2.0 −1.0 −0.3 −1.4 −4.6
Niger 1.9 4.3 2.0 6.3 −2.4 −0.0 −0.0 −1.9 −4.4
Nicaragua 1.9 5.0 1.1 6.1 −1.7 −0.9 −0.3 −1.3 −4.1
Gambia, The 1.9 6.5 1.5 8.0 −0.7 −1.1 −0.6 −3.7 −6.1
Guatemala 1.9 3.6 1.2 4.8 −0.8 −1.0 −0.2 −0.9 −2.9
Indonesia 1.9 2.8 0.5 3.2 −0.2 −0.6 −0.0 −0.6 −1.4
Papua New Guinea 1.7 4.3 0.4 4.7 −2.3 −0.2 −0.0 −0.5 −3.0
Iraq 1.6 1.5 2.0 3.5 −0.3 −0.3 −0.1 −1.2 −1.8
Liberia 1.6 3.3 1.3 4.6 −0.8 −0.4 −0.5 −1.3 −3.0
Nepal 1.4 2.7 1.6 4.4 −0.5 −0.3 −0.1 −2.0 −3.0
Vietnam 1.1 5.1 2.0 7.1 −2.8 −1.0 −0.4 −1.8 −5.9
Georgia 1.0 2.1 0.1 2.2 −0.6 −0.0 −0.0 −0.6 −1.2
Moldova 0.7 1.4 1.5 2.9 −0.6 −0.1 −0.0 −1.4 −2.1
Burkina Faso 0.7 3.8 2.3 6.1 −2.4 −0.6 −0.5 −1.9 −5.4
Kyrgyz Republic 0.6 1.8 1.4 3.2 −0.7 −0.2 −0.0 −1.6 −2.6
Bangladesh 0.5 4.9 2.3 7.2 −3.8 −1.5 −0.1 −1.3 −6.7
Burundi 0.4 6.9 2.2 9.0 −6.2 −0.5 −0.0 −1.8 −8.6
Mongolia 0.1 2.7 0.6 3.4 −0.6 −0.2 −0.1 −2.4 −3.3
Average 2.4 4.5 1.9 6.4 −1.5 −0.6 −0.2 −1.6 −3.9
Pop. weighted av. 2.3 4.1 2.1 6.2 −1.6 −0.9 −0.2 −1.2 −3.9
GDP weighted av. 2.4 3.7 1.8 5.4 −1.0 −0.9 −0.2 −1.0 −3.1

Notes: Authors' calculations. The gain from trade, expressed in percentage points, is the population weighted average of the proportional change in household real expenditure.

of 0.6% in countries with gains and 1.1% in countries with losses. The re- capita expenditure), and we estimate bivariate kernel densities of the
duction in income from enterprises producing tradable goods is small joint distribution of the gains from trade and household per capita
on average; −0.2% of income among winners and −0.1% among coun- expenditure.
tries that lose. The biggest driver of income losses is the reduction in For the sake of exposition, we divide countries into two groups using
government revenue: this channel accounts for 1.6 of the 3.9 percentage the pro-poor index of Nicita et al. (2014). In our application, the pro-
points loss in income among winners and 3.2 of the 6.4 percentage point poor index is the difference between the average gains for the poor—
loss among losers. the bottom 20% of the income distribution—and the rich—the top 20%.
If the index is positive the poor gain proportionately more (or lose pro-
4.2. The distributional effects of trade portionately less) than the rich, while the opposite happens when the
index is negative. According to this classification, import tariff liberaliza-
We now turn to the distribution of the gains from the trade, which tion would be pro-poor in 17 countries, while it would be pro-rich in the
have been the focus of Porto (2006), Nicita et al. (2014), Fajgelbaum remaining 37 countries.
and Khandelwal (2016), Atkin et al. (2018), Faber (2014) and Atkin We illustrate the case of pro-poor bias in Fig. 1 for the cases of the
and Donaldson (2015). Indeed, the average impacts just discussed Central African Republic (panel (a)) and Mauritania (panel (b)).
mask significant heterogeneity across households. This is because the Appendix B provides figures for all countries. In the Central African Re-
net welfare impact is determined by a combination of initial tariffs as public, the kernel average is positive everywhere, so that there are aver-
well as income and consumption portfolios. We combine two tech- age gains from trade across the income distribution, but the slope of the
niques to explore the distributional effects. We estimate kernel averages kernel regression is negative (so that the poor gain proportionately
of the gains from trade, conditional on household initial well-being (per more than the rich). This pro-poor bias with positive kernel average
8 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45

Table 5
Gains from trade - losers.

20
Gains Expenditure Income

agric. manuf. total agric. wage enter. rev. total

Cambodia −3.1 4.4 0.9 5.4 −4.5 −0.8 0.0 −3.1 −8.4

10
Ghana −1.9 1.0 2.9 3.9 −1.2 −2.8 0.0 −1.8 −5.8
Mauritania −1.3 4.5 1.8 6.3 −1.1 −0.1 −0.0 −6.5 −7.6
Madagascar −1.1 3.2 0.8 3.9 −2.3 −0.9 −0.1 −1.7 −5.0

welfare effects
Bhutan −0.8 8.5 5.3 13.8 −3.1 −2.8 0.0 −8.7 −14.6
Mali −0.3 2.4 0.2 2.6 −0.9 −0.0 −0.0 −2.0 −3.0

0
Sri Lanka −0.3 3.3 0.8 4.1 −1.4 −1.2 −0.7 −1.0 −4.4
Comoros −0.3 1.6 1.3 3.0 −0.6 −0.3 −0.3 −2.0 −3.2
Rwanda −0.1 3.7 1.4 5.1 −2.5 −1.1 −0.0 −1.6 −5.2
Average −1.0 3.6 1.7 5.3 −2.0 −1.1 −0.1 −3.2 −6.4

-10
Pop. −1.2 2.9 1.3 4.2 −1.9 −1.2 −0.2 −2.0 −5.4
weighted
av.
GDP −1.0 2.8 1.4 4.2 −1.7 −1.5 −0.3 −1.8 −5.3

-20
weighted
av. 6 8 10 12
log per capita expenditure
Notes: Authors' calculations. The gain from trade, expressed in percentage points, is the
population weighted average of the proportional change in household real expenditure.

gains also appears in Azerbaijan, Central African Republic, Ecuador,

Indonesia, Moldova, Nepal, Pakistan, Papua New Guinea, Rwanda, the
Republic of Yemen and Zambia (see Appendix B). In these countries, im-
port tariff liberalization raises incomes across the income distribution 5
and may reduce inequality. Liberalization would not be Pareto improv-
ing, however. The bivariate density of the welfare effects and initial in-
0

come illustrates the dispersion in the welfare effects and the existence
welfare effects

of winners and losers in all segments of the per capita expenditure spec-
trum. A more extreme version of this pattern is shown in panel (b) for
-5

the case of Mauritania, where there are average gains for the poor but
average losses for the rich. This implies a very strong pro-poor bias. Sim-
ilar patterns are also observed in Guinea-Bissau, Mali, Mongolia and Sri
-10

Lanka (see Appendix B). In all these countries, import tariff liberaliza-
tion raises average income and may reduce inequality significantly.
We illustrate the pro-rich bias in Fig. 2. In Uzbekistan (panel (a)), the
-15

kernel average is always positive at all levels of per capita expenditure,

and the slope of the kernel regression is positive, indicating that, on av- 7 8 9 10 11 12
erage, the rich gain proportionately more than the poor. Again, some in- log per capita expenditure
dividual households stand to lose, as the underlying bivariate kernel
density graph shows. Import tariff liberalization lifts incomes through- Fig. 1. Patterns of distributional impacts, pro-poor bias. Notes: The solid curve is the non-
parametric kernel regression of the welfare effects and the initial level of per capita
out the income distribution, but at the expense of potentially higher in-
household expenditure. The dotted curves are the corresponding 95% confidence bands.
equality. This pattern is found in Armenia, Bolivia, Cameroon, Côte The contour lines are level curves of the non-parametric kernel bivariate density of
d'Ivoire, the Arab Republic of Egypt, Ethiopia, Georgia, Guatemala, these two variables. Liberalization is classified as having a pro-poor bias if the average
Guinea, Iraq, Kyrgyz Republic, Liberia, Malawi, Mozambique, proportional real income gains accruing to households in the the bottom 20% of the pre-
Nicaragua, Niger, Sierra Leone, South Africa, Tajikistan, Tanzania, liberalization income distribution exceed the average proportional real income gains
accruing to households in the top 20% of the pre-liberalization real income income
Uganda, and Ukraine (see Appendix B). The case of pro-rich bias with distribution.
average gains for the richest households and losses for the poorest is il-
lustrated in panel (b) for Togo. Import tariff liberalization is strongly
pro-rich and inequality significantly exacerbated. Similar patterns (Atkinson, 1970):
arise in Bangladesh, Benin, Burkina Faso, Burundi, The Gambia, Kenya,
Nigeria, and Vietnam. In panel (c), we show the case of average losses  1−ε
1 X xh
and a pro-rich bias for Ghana. In this country, as well as in Bhutan, W¼ ; ð13Þ
H h 1−ε
Cambodia, Comoros and Madagascar, the poor lose proportionately
more than the rich. This is a scenario with average losses as well as in-
creased inequality. where W is social welfare and ε ≠ 1 is a parameter that measures the dis-
like for inequality.9 When ε=0, every household counts the same and
5. The trade-off between income gains and inequality costs social welfare is just the sum (average) of per capita expenditures. As
ε increases, the weight attached to the well-being of poorer households
Given the patterns of gains from trade and of the distribution of increases. In the limit, as ε approaches infinity social welfare is deter-
those gains, we now assess whether there is a trade-off between the in- mined by the well-being of the very poorest household (as in a Rawlsian
come gains and the inequality costs of import tariff liberalization. This social welfare function). It is very important to interpret the Atkinson
necessarily involves value judgments because different societies, indi- social welfare function correctly. As Deaton (1997) explains, W in (13)
viduals or policy makers may value the gains or losses of some house- is not necessarily (and more precisely, it seldom is) the object that
holds differently. A tool to describe the trade-off between income
inequality and average incomes is the Atkinson social welfare function 9
For completeness, when ε = 1, we define lnW = (1/H)∑h ln xh.
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 9

15

10
10

5
welfare effects

welfare effects
5

0
0

-5
-5

-10
-10

-15
8 9 10 11 12 6 8 10 12 14
log per capita expenditure log per capita expenditure
10
5
welfare effects
-5 -10
-150

8 10 12 14 16
log per capita expenditure

Fig. 2. Patterns of distributional impacts, pro-rich bias. Notes: The solid curve is the non-parametric kernel regression of the welfare effects and the initial level of per capita household
expenditure. The dotted curves are the 95% confidence bands. The contour lines are level curves of the non-parametric kernel bivariate density of these two variables. Liberalization is
classified as having a pro-rich bias if the average proportional real income gains accruing to households in the the top 20% of the pre-liberalization income distribution exceed the
average proportional real income gains accruing to households in the bottom 20% of the pre-liberalization real income income distribution.

policy makers maximize when choosing among policy options. Rather, Using W(ε), we can define a measure of the gains from trade that in-
it provides a means of quantifying potential tensions between mean in- cludes a correction for the inequality costs:
come and its distribution across households.
An important property of the Atkinson social welfare function is that W 1 ðεÞ−W 0 ðεÞ
GðεÞ ¼ ; ð16Þ
it can be decomposed in a way that is conducive to the assessment of W 0 ðεÞ
this trade-off. Concretely, we can write
where W0 is the ex-ante social welfare, calculated with the observed
(xh0) income distribution in the presence of trade protection and W1(ε)
W ¼ μ  ð1−I Þ; ð14Þ is the counterfactual social welfare under import tariff liberalization (
^xh1 ). Given the initial situation and the post-liberalization situation, we
where μ is mean income and can compare W0 and W1 using (16). For ε = 0, this is a comparison of
mean income, that is, the calculation of the gains from trade
!1=ð1−εÞ (Arkolakis et al., 2012; Costinot et al., 2012; Costinot and Rodriguez-
1 H
h 1−ε Clare, 2014; Artuc et al., 2015; Caliendo and Parro, 2015; Melitz and
I ¼ 1− ∑ x =μ ; ð15Þ
H h¼1 Redding, 2015; Arkolakis et al., 2019; and Caliendo et al., 2015). For ε
N 0, this comparison involves the calculation of the gains from trade
with an implicit correction for inequality (Antras et al., 2017; and
is an implicit measure of income inequality. Social welfare thus depends Galle et al., 2017). With estimates of G(ε) for different ε, we can establish
on average income μ and on the aggregate level of “equality” (1 − I(ε)). whether there is a trade-off between the gains in average incomes and
This measure of inequality I(ε) (or the measure of equality (1 − I(ε))) the costs of inequality in its distribution, we can quantify this trade-off,
depends on ε and nests a whole family of inequality measures. and we can assess it.
10 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45

For the discussion that follows, we exploit the decomposition of W in

8
Eq. (14). Note that we can write

7
μ 1 I 0 ðεÞ−I1 ðεÞ
GðεÞ ¼ Gð0Þ þ : ð17Þ
μ 0 1−I0 ðε Þ

inequality adjusted gains, G(ε)

6

6
5

5
The inequality-adjusted gains from trade are thus equal to the in-
come gains from trade G(0) plus a correction for changes in inequality,

4
which we will refer to as equality gains. The gains from trade G(0) can
be positive or negative, as shown in Section 4.1. The correction for in-

3
equality is governed by the Atkinson inequality index I(ε), which may
depend non-monotonically on ε. If inequality increases for some εa N 0

2
so that I1(εa) N I0(εa), then G(εa) incorporates a downward correction
for these inequality costs. Conversely, if I1(εb) b I0(εb) at some εb, then

1
the gains from trade are amplified. Note that G(ε) N 0 does not imply
no inequality costs per se but rather that their welfare impacts are dom-

0
0 2 4 6 8 10
inated by the income gains. inequality aversion, ε
Trade-offs arise when income gains and equality gains move in op-
μ 1 I0 ðεÞ−I 1 ðεÞ
posite directions, i.e. when G(0) and μ 0 1−I 0 ðεÞ have opposite signs. Fig. 3. No trade-off, income gains and equality gains. Notes: The solid line depicts how the
inequality adjusted welfare gains associated with liberalization G(ε) vary with inequality
This is the case in countries where trade exacerbates inequality but im-
aversion ε. The dotted lines represent 95% confidence intervals based on 1000 bootstrap
proves average income, and in countries where it reduces inequality at replications.
the expense of lowering mean income. In some countries, these trade-
offs can even result in reversals of trade policy preferences, in the
sense that for certain levels of inequality aversion ε, the inequality ad-
justed gains from trade may be negative (positive) even though import
tariff liberalization leads to an increase (reduction) in average income.
0

0
Since the sign and magnitude of the equality gains can vary with ε
both the existence and acuteness of the trade-offs depend on the level
-1

-1
of inequality aversion. No trade-offs occur in countries where import
inequality adjusted gains, G(ε)

tariff liberalization leads to both income and equality gains (for all ε)
-2

-2
or in countries where it leads to lower income and higher inequality
(for all ε).
-3

-3
One of the main findings of our paper is the high prevalence of trade
policy trade-offs between average incomes and income inequality in the
-4

-4
developing world. Among the 54 countries in our sample, 45 face a
trade-off and only 9 do not. In 27 of the 45 countries the trade-offs
-5

-5
can be severe enough to generate (potential) reversals in the ranking
of trade policy preferences. We present countries without trade-offs
-6

-6
(Section 5.1), countries with trade-offs but without trade policy prefer-
ence reversals (Section 5.2) and with reversals (Section 5.3). Sections
5.4 and 5.5 evaluate such trade-offs and some of the underlying factors.
-7

-7
0 2 4 6 8 10
Finally, we run robustness tests in Section 5.6 and we explore protec- inequality aversion, ε
tionists scenarios in Section 5.7.
Fig. 4. No trade-off, income losses and inequality costs. Notes: The solid line depicts how
5.1. No trade-off countries the inequality adjusted welfare gains associated with liberalization G(ε) vary with
inequality aversion ε. The dotted lines represent 95% confidence intervals based on 1000
bootstrap replications.
When average income gains emerge together with equality gains,
there is no trade-off. The case of the Central African Republic is shown
in Fig. 3, which plots G(ε) for ε ∈ [0, 10].10 To obtain confidence intervals At the other end of the spectrum lie 4 countries, Comoros, Ghana,
for G(ε) we resample from the observed distribution and bootstrap Madagascar and Rwanda, which are characterized by average income
using 1000 replications. In the Central African Republic, import tariff lib- losses and inequality costs for all ε. In these countries import tariff liber-
eralization leads to average welfare gains with a pro-poor bias (see alization would be unambiguously social welfare depressing. Fig. 4 illus-
Fig. 1). The gains in average incomes of 4.2% are independent of ε and trates the case of Ghana. Since income losses are disproportionately
the pro-poor bias implies that liberalization also leads to equality borne by the poor, the inequality adjusted gains from trade are negative
gains. As ε increases and more weight is put on the poor, these equality and decreasing with ε. For instance, the aggregate losses of −1.9% (for ε
gains actually get bigger. As a result, the inequality adjusted welfare = 0) are augmented to −4.3% when inequality aversion is high.
gains are increasing in the inequality aversion parameter ε, and exceed
6% for large ε. Other countries in which import tariff liberalization yields
both equality gains and lifts average incomes are Mongolia, Guinea-Bis- 5.2. Trade-off countries without trade policy preference reversals
sau, Jordan and Yemen. In these countries, import tariff liberalization is
unambiguously social welfare enhancing. There are 45 countries with evidence of a trade-off. In 18 countries,
this trade-off is not strong enough to generate reversals of trade policy
10
For presentational purposes, we examine G(ε) for a limited range of ε ∈ [0,10] but the
preferences because import tariff liberalization dominates protection
results hold more generally. Results are available upon request but omitted to conserve at all levels of inequality aversion (in 16 countries) or because protec-
space. tion dominates liberalization (in only 2 countries, notably Bhutan and
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 11

Cameroon, the Arab Republic of Egypt, Guinea, Indonesia, Iraq, the Kyr-
gyz Republic, Moldova, Pakistan, Tajikistan, Uganda, Ukraine,
4

4
Uzbekistan, South Africa, and Zambia. Since G(ε) is positive and
statistically significant for all ε, import tariff liberalization would unam-
biguously lead to higher social welfare.
inequality adjusted gains, G(ε)
3

3
Plots of G(ε) for all these countries are given in Appendix C. We sum-
marize the information contained in Figs. 3, 4 and 5 in Table 6, which re-
ports the income gains from trade G(0) (column 1) as well as the
equality gains ((μ1/μ0)(I0(ε) − I1(ε))/(1 − I0(ε)) for several values of in-
2

2
equality aversion ε. To illustrate, consider the case of Guinea-Bissau (a
country without a trade-off) in panel (a). The gains from trade are
2.0% (column 1). Because inequality declines with import tariff liberali-
zation, the equality gains increase with ε. The correction is thus positive
1

1
and increasing with ε. At ε = 0.5, for instance, the correction is 0.5 per-
cent and the inequality-adjusted gains are 2.5%. At ε = 1 (ε = 10), the
correction is 0.7 (0.8) percent and the total inequality-adjusted gains
go up to 2.7 (2.8) percent. Another interesting example is Madagascar,
0

0
0 2 4 6 8 10
inequality aversion, ε where there are losses from trade of −1.1 percent and increases in in-
equality costs so that, at ε = 1, the inequality-adjusted losses from
Fig. 5. Trade-off without policy preference reversal, income gains and inequality costs. trade drop to −1.8 percent and at ε = 10, to −3.4 percent. To illustrate
Notes: The solid line depicts how the inequality adjusted welfare gains associated with a country with trade-offs (Panel (b)), consider Ukraine. The gains from
liberalization G(ε) vary with inequality aversion ε. The dotted lines represent 95% trade are 3.2%, but inequality increases and consequently there is a
confidence intervals based on 1000 bootstrap replications. downward correction to G. At ε = 1, this correction is very small,
−0.1% and the inequality adjusted gains drop to 3.2; at ε = 10, the cor-
Cambodia). Fig. 5 illustrates the case of Uzbekistan, where liberalization rection is −0.7% and the inequality adjusted total gains are 2.5%. The
creates average income gains at the expense of inequality costs. In table reports many other interesting patterns.
Uzbekistan inequality increases smoothly with ε and, as a consequence,
the inequality adjusted welfare gains G(ε) decrease as inequality aver- 5.3. Trade-off countries with trade policy preference reversals
sion rises. The gains from trade are G(0) = 3.5 percent, while the
inequality-adjusted gains for large ε can go down to about 1.1%. Other In the remaining 27 countries in our sample, we find evidence of a
countries exhibiting a similar pattern are Armenia, Azerbaijan, stronger trade-off which may induce a potential reversal of the ranking

Table 6
Income gains and inequality costs without trade policy preference reversals.

μ 1 −μ 0 μ 1 I0 ðεÞ−I1 ðεÞ
Income Gains Equality Gains
μ0 μ 0 1−I0 ðεÞ

ε = 0.5 ε=1 ε = 10

A) Countries without Trade-offs

Guinea-Bissau 2.0 0.5 0.7 0.8
Central African Republic 4.2 0.4 0.8 2.1
Jordan 4.1 0.4 0.7 0.3
Yemen, Rep. 2.7 0.2 0.3 0.1
Mongolia 0.1 0.1 0.2 0.5
Comoros −0.3 −0.1 −0.2 −1.8
Rwanda −0.2 −0.2 −0.2 −4.9
Madagascar −1.1 −0.3 −0.7 −2.4
Ghana −1.9 −0.4 −0.8 −2.3

B) Countries with Trade-offs

Indonesia 1.9 0.1 0.2 −0.4
Pakistan 2.4 0.0 0.3 −0.2
Azerbaijan 2.5 0.0 0.0 −0.6
Zambia 5.9 −0.0 −0.1 −0.0
Moldova 0.7 −0.0 −0.0 0.6
Kyrgyz Republic 0.6 −0.0 −0.1 −0.0
Ukraine 3.2 −0.0 −0.1 −0.7
Egypt, Arab Rep. 2.9 −0.1 −0.2 −0.8
Tajikistan 1.9 −0.1 −0.2 −0.6
Iraq 1.6 −0.1 −0.1 −0.3
Armenia 2.5 −0.1 −0.2 −1.0
South Africa 2.5 −0.1 −0.4 −1.2
Guinea 2.8 −0.3 −0.6 −0.8
Uganda 1.9 −0.3 −0.5 0.3
Uzbekistan 3.5 −0.4 −0.7 −2.4
Cameroon 6.8 −0.5 −1.1 −5.0
Cambodia −3.0 0.0 −0.1 −0.8
Bhutan −0.8 −0.1 −0.5 −7.5

Notes: Authors' calculations. The table presents the decomposition of the inequality-adjusted gains from trade G(ε). The first column reports the average income gains from trade (the
proportional change in real household expenditures). The three remaining columns show the equality gains (due to changes in inequality) for different values of inequality aversion (ε
= 0.5,ε = 1, and ε = 10). The inequality-adjusted gains from trade is the sum of the income gains and the equality gains.
12 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45
3

5
2

2
1

1
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

3

3
0

0
-1

-1

1
-2

-2

-1

-1
-3

-3
-4

-4

-3

-3
-5

-5
-6

-6

-5

-5
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε

Fig. 6. Trade-off with trade policy preference reversal, income gains and inequality costs. Fig. 7. Trade-off with trade policy preference reversal, income losses and equality gains.
Notes: The solid line depicts how the inequality adjusted welfare gains associated with Notes: The solid line depicts how the inequality adjusted welfare gains associated with
liberalization G(ε) vary with inequality aversion ε. The dotted lines represent 95% liberalization G(ε) vary with inequality aversion ε. The dotted lines represent 95%
confidence intervals based on 1000 bootstrap replications. confidence intervals based on 1000 bootstrap replications.

of trade policy preferences. This reversal occurs when G(ε) changes sign,
going from positive to negative or from negative to positive, as ε in- relatively light weights on the inequality costs are enough to offset the
creases. This means that, depending on the value judgement parameter gains from trade. It is important to note that while the value of ε ∗ de-
ε, the social welfare function points to welfare gains associated with im- scribes the nature of the trade-off, it is silent about whether this trade-
port tariff liberalization or with trade protection. Fig. 6 and Fig. 7 show off is socially acceptable.
two examples of the existence of such trade-offs. In Benin (Fig. 6), Table 7 presents estimates of the trade-ε ∗ (column 1) and its 95%
there are significant average income gains of 2.2% so that G(0) N 0. How- confidence interval (columns 2 and 3). Since the interpretation of the
ever, as ε increases, import tariff liberalization creates larger and larger trade-ε ∗ depends on the sign of the gains, we report results separately
inequality costs so that, eventually, G(ε) becomes significantly negative. for countries that enjoy income gains in panel (a) and countries that suf-
At very large ε, the inequality-adjusted gains are −4.4 percent. It follows fer income losses in panel (b). The select few countries characterized by
that free trade dominates protection when ε is low, whereas protection multiple (potential) reversals are presented in panel (c).
dominates free trade when ε is high. Other countries that exhibit similar Among the countries with gains (Panel (a)), the trade-ε ∗ vary a lot.
trade-offs are Bangladesh, Burkina Faso, Burundi, Ethiopia, The Gambia, In some cases, the cutoff can be as low as 0.1 (Burundi), or 0.57
Guatemala, Kenya, Liberia, Malawi, Mozambique, Nigeria, Papua New (Burkina Faso and Bangladesh). In other cases, it can be much larger,
Guinea, Togo, and Vietnam.11 as in Malawi (7.1) or Guatemala (7.0). To put these numbers in perspec-
Mali (Fig. 7) exhibits the opposite pattern; There are small but statis- tive, we canvassed the literature for guidance on what a reasonable
tically significant average losses from trade (G(0) = − 0.3) but, as ε in- value for ε is. Deaton (1997) recommends exploring values of ε ∈ [0, 2]
creases, the equality gains from liberalization end up strictly dominating when doing policy evaluations. Using experiments, Carlsson et al.
those losses and the inequality-adjusted gains G(ε) approach 3%. Conse- (2005) estimate ε ∈ [1, 2] and Layard et al. (2008) estimate a value of
quently, protection dominates free trade at low ε, while free trade dom- ε of 1.3. A high ε ∗ consequently suggests that the trade-offs are soft
inates protection at high ε. This also happens in Mauritania and Sri in the sense that one would have to be implausibly inequality averse
Lanka. in order not to prefer liberalization. This implies a strong presumption
To quantify these policy preference reversals, we define the cutoff in favor of lower tariffs. By contrast, in Burundi, Bangladesh or Burkina
value ε ∗ such that G(ε ∗) = 0. The cutoff ε ∗, which we refer to as Faso, the trade-off would be quite stark. Since the gains from trade are
trade-ε ∗, is a measure of the inequality aversion to import tariff liberal- positive but very small, even at low levels of inequality aversion one
ization. It is a sufficient statistic to describe the trade-off between mean would prefer protection. In the remaining countries, the trade-off ap-
income and inequality in the presence of trade policy preference rever- pears to be more moderate, with a substantial number of the estimates
sals. Defining the trade-off in terms of the gains, the value of ε ∗ shows of trade-ε ∗ lying in the [1,2] interval (1.2 in The Gambia, 1.2 in Togo, 1.5
how intolerant towards inequality a society would have to be in order in Benin, 1.9 in Nigeria, and 2.0 in Vietnam). Kenya (2.5), Ethiopia (3.1),
to make the gains from trade not worthwhile from a social welfare Mozambique (3.5), Papua New Guinea (4.4), and Liberia (4.5) are coun-
perspective.12 A high value of ε ∗ implies a soft trade-off: a society tries with relatively high trade-ε ∗, but not quite as extreme as Malawi or
needs to put a heavy weight on the cost of higher inequality to be will- Guatemala.
ing to forgo the gains (always in a social welfare function sense). In the There are countries with trade-offs where the evidence on trade pol-
limit case, when trade-ε ∗ tends to infinity or when trade-ε ∗ does not icy reversals is not so compelling. This occurs when the inequality-
exist (as in the countries discussed in Sections 5.1 and 5.2), there is no adjusted gains from trade are not statistically indistinguishable from
reversal in trade policy preference rankings and, given gains from zero, that is the null hypothesis that G(ε) = 0 cannot be rejected for a
trade, import tariff liberalization leads to higher social welfare for any range of ε. We report these cases in columns 4 and 5 of Table 7. In
ε. By contrast, a low trade-ε ∗ implies a very hard trade-off because Sierra Leone, for instance, for ε N 4.0, there are inequality adjusted
gains from trade (G(ε) N 0) that are not statistically different from 0.
11 Consequently, we cannot rule out a potential reversal (from preferring
See below for a more detailed discussion of some of these countries and their trade-off.
12
Alternatively, the ε ∗ shows how much a society would have to value equality to forgo liberalization to preferring protection). Similar scenarios emerge
the average gains from trade. in Bolivia (ε N 4.1), Niger (ε N 6.0), Nicaragua (ε N 6.3), Côte d'Ivoire
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 13

Table 7
Income gains and inequality costs with trade policy preference reversals.

10

10
Trade Policy Preference Potential Reversals
Reversals

ε∗ Lower Upper Lower Upper Zambia

Bound bound Bound Bound Central African Republic Cameroon

5
Jordan

A) Countries with Income Gains

Ecuador

Burkina Faso 0.6 0.5 0.7

Ukraine Tanzania
Pakistan Yemen

G(1.5)
Guinea BissauAzerbaijan Egypt
Cote d'Ivoire Sierra Leone
Bolivia Uzbekistan

Bangladesh 0.6 0.5 0.7 Indonesia Armenia

Nicaragua
P.N. Guinea
Guinea
South Africa
Iraq Tajikistan Kenya
NigeriaMozambique
Nepal Guatemala

Gambia, The 1.2 1.1 1.3 Mali

Sri Lanka Georgia

Moldova
Liberia
Uganda
Niger Malawi
Ethiopia
Kyrgyz Rep.

Togo 1.2 1.2 1.3 Mongolia

Vietnam

0
Mauritania Benin
Rwanda

Benin 1.5 1.4 1.6 Comoros

Bangladesh

Burkina Faso
Togo
Gambia

Nigeria 1.9 1.8 2.0 Madagascar

Bhutan

Vietnam 2.0 1.8 2.1 Burundi

Ghana

Kenya 2.5 2.4 2.7 Cambodia

Ethiopia 3.1 2.7 3.6

-5

-5
Mozambique 3.5 2.9 8.5
Papua New 4.4 3.1 –
Guinea -5 0 5 10
Liberia 4.5 3.6 – G(0)
Guatemala 7.0 5.2 –
Malawi 7.1 4.1 – Fig. 8. Trade-off resolution, notes: scatter plot of the inequality-adjusted gains from trade
Sierra Leone 4.0 – G(ε), at ε = 1.5, against the gains from trade G(0). The symbols represent the trade-off
Bolivia 4.1 – resolution: : no trade-off, liberalize; : soft trade-off, liberalize; : policy reversal,
Niger 6.0 –
liberalize; : no trade-off, protect; : policy reversal, protect; : soft trade-off, protect.
Nicaragua 6.3 –
Côte d'Ivoire 7.0 –
Georgia 7.1 –
Tanzania 8.9 – is, in general, close to the Gini coefficient. Since the Gini is often used
Nepal 9.3 – in discussions about inequality, this is a useful benchmark. If there
Ecuador 9.9 – were only small corrections for inequality, then the pairs (G(1.5), G
B) Countries with Income Losses (0)) would lie along the 45 degree line, with larger corrections for
Mali 0.4 0.2 0.5 those pairs further away. Orthant I hosts countries with average gains
Mauritania 1.6 1.5 1.8 as well as gains after inequality corrections; orthant III hosts countries
C) Countries with multiple (potential) reversals with average losses with and without inequality corrections. In orthant
Countries with Income Gains II, we see countries with losses from trade that turn into gains after the
Burundi 0.1 0 0.2 5.6 7.1 inequality adjustments, and, in orthant IV, those countries with gains
Countries with Income Losses from trade that turn into losses with inequality considerations.
Sri Lanka 0.3 0.2 0.4
For an inequality aversion parameter of ε = 1.5, 17 countries would
8.9 7.2 –
not face a trade-off. Eleven of them would unambiguously benefit from
Notes: Authors' calculations. The table presents estimates of the trade-ε ∗, the cut-off value
liberalization as they enjoy both income and inequality gains. These
of inequality aversion at which there is a reversal of trade policy preference in terms of so-
cial welfare. The standard errors are bootstrapped using 1000 replications. countries, which lie in orthant I, above the 45 degree line, are
Azerbaijan, Central African Republic, Ecuador, Guinea-Bissau,
Indonesia, Jordan, Mongolia, Pakistan, Papua New Guinea, Nepal, and
(ε N 7.0), Georgia (ε N 7.1), Nepal (ε N 9.3), Tanzania (ε N 8.9), and Yemen. The remaining six countries, Bhutan, Cambodia, Comoros,
Ecuador (ε N 9.9). In all these countries, however, the trade policy pref- Ghana, Madagascar and Rwanda, would unambiguously prefer protec-
erence reversal would come about only for levels of inequality aversion tionism as trade liberalization leads to both income losses and inequal-
that are arguably implausibly large. ity costs (they lie in orthant III, below the 45 degree line).
Among the countries with aggregate losses (Panel (b)) of Table 7, A total of 37 countries would exhibit trade-offs (for ε = 1.5). In 30 of
the estimated trade-ε ∗ tend to be low. For instance, we get ε ∗ = 0.3 in these countries, the trade-off is resolved in favor of liberalization.
Sri Lanka (at the first reversal) and ε ∗ = 0.4 in Mali. Note that the inter- Twenty-eight countries would show income gains and inequality
pretation in these cases is different because for low ε, trade protection is costs, but inequality-adjusted gains from trade liberalization. These
preferred to liberalization, and, conversely, liberalization is preferred to are the countries in orthant I, below the 45 degree line. Two countries
protection for higher ε. In these countries, a low ε ∗ thus implies a pre- (Mali and Sri Lanka) would show instead income losses but sufficiently
sumption in favor of lower tariffs, too. In Mauritania (ε ∗ = 1.6), trade high equality gains so that there are inequality-adjusted gains from
protection would be preferred under more moderate values of inequal- trade in the end (for ε = 1.5). These are the countries in orthant II.
ity aversion making it harder to infer trade policy prescriptions. In 7 countries the trade-off is instead resolved in favor of protection
because tariffs lead to higher inequality-adjusted welfare. One country,
5.4. Assessment Mauritania (orthant III, above the 45 degree line), would face income
losses and equality gains, which, for ε = 1.5, are not enough to compen-
While our results attest to highly heterogeneous welfare impacts of sate for those losses. In six countries, Bangladesh, Benin, Burkina Faso,
trade liberalization across households and countries, overall the analysis Burundi, The Gambia and Togo, the inequality costs dominate the in-
provides overwhelming evidence of a trade-off between income gains come gains. These are in orthant IV.
and inequality costs of trade policy. In most cases, however, the income It turns out that the resolution of the trade-off is very stable for dif-
gains outweigh the inequality costs, suggesting countries are better off ferent values of plausible inequality aversion. In Fig. 9, we reproduce
with freer trade. We summarize these observations and results in Fig. 8 for ε = 1 (panel (a)) and ε = 2 (panel (b)). As it can be seen,
Fig. 8. We plot the value of the inequality-adjust gains from trade G(ε) there are only a few countries where the trade policy prescriptions are
against the gains from trade G(0). For our assessment, we use ε = 1.5 more equivocal. For ε = 1, Benin, The Gambia, and Togo jump from
because it is in the middle of the empirically plausible interval [1, 2] orthant IV to orthant I (thus preferring liberalization instead of protec-
and because it yields a measure of the Atkison inequality index I that tion). For ε = 2, Vietnam and Nigeria jump from orthant I to orthant
14 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45

dynamic considerations. Whatever the reason for it, our findings show
that protection can be very costly in terms of the social welfare
10

10
aggregator W.

Zambia
Cameroon
5.5. Underlying factors: expenditure and income household heterogeneity
Central African Republic
5

5
Jordan

Tanzania
How do the gains from trade and trade-offs described above
Ecuador

emerge? We argue that a distinctive element of our approach is the

Ukraine
Yemen Sierra Leone
Cote
Pakistan Egypt Bolivia Nigeria d'Ivoire
Uzbekistan
Guinea BissauAzerbaijan
G(1)

Kenya
Guinea Mozambique

vast heterogeneity in household expenditures and incomes. We showed

Nicaragua Armenia
Indonesia South Africa
P.N. Guinea
Tajikistan Guatemala
Iraq Uganda
Nepal
Liberia Malawi
Niger Ethiopia

Mali
Sri Lanka
Moldova
Kyrgyz Rep.
Mongolia
Georgia
Vietnam Benin
Togo
Gambia
in Section 4.1 that this heterogeneity helps explain the gains from trade.
0

0
Across-household differences in the consumption gains and in the in-
Bangladesh
Rwanda
Mauritania Burkina Faso
Comoros
Bhutan

Madagascar
Burundi
come losses of the elimination of tariffs show that countries are more
Cambodia
Ghana
likely to benefit from liberalization if food expenditure shares are
large, relative to agricultural production income shares.
Household heterogeneity underlies the patterns of trade-offs as well.
-5

-5
In Table 8, we decompose the equality gains (or losses) into consump-
-5 0 5 10 tion equality gains and (nominal) income equality gains. To calculate
G(0) these, we estimate two counterfactual scenarios; one in which liberali-
zation solely impacts consumption (and does not impact income), and
one in which it solely impacts income (and does not impact consump-
tion). We compute the consumption and income equality components
10

10
using Eq. (17) (recall that equality gains are equal to μμ 1 I01−I
ðεÞ−I 1 ðεÞ 13
0 ðεÞ
).
0

As much as there is heterogeneity in the trade-offs, there is a marked

heterogeneity in the income and consumption equality components.
Central African Republic
Jordan
Zambia
Note, however, that the consumption equality gains are positive yet
5

Cameroon
small in the majority of countries. As consumers, the poor seem to ben-
Pakistan Ecuador
Guinea Bissau Ukraine
efit disproportionately from liberalization, in part because they spend a
larger share of their budget on food items, which are subject to compar-
Yemen Tanzania
Azerbaijan
G(2)

Egypt Uzbekistan
Indonesia
Armenia Sierra Leone
Bolivia Cote d'Ivoire
Mali P.N. Nicaragua
Guinea Guinea

atively high tariffs. By contrast, the income component is overwhelm-

Nepal
South Africa
Iraq Tajikistan
Uganda
Guatemala
Sri Lanka Georgia Kenya Mozambique
Niger
Mongolia Moldova Liberia Malawi

ingly negative across countries, reflecting the fact that trade

Kyrgyz Rep.
Ethiopia
Mauritania
0

Vietnam
Rwanda

liberalization creates income inequality costs that are disproportion-

Nigeria
Benin
Comoros Bangladesh
Burkina Faso
Togo

ately borne by poorer households. Whereas the consumption equality

Gambia

Madagascar
Bhutan Burundi

Cambodia Ghana
gains on average increase only slightly as inequality aversion rises, the
average income inequality costs tend to increase sharply (i.e., become
-5

-5

more negative) with ε. The trade-offs between the aggregate gains

and aggregate inequality costs are thus predominantly driven by nomi-
-5 0 5 10
G(0) nal income inequality. This finding shows that the income losses associ-
ated with trade liberalization are borne disproportionately by the
Fig. 9. Trade-Off Resolution, Notes: scatter plot of the inequality-adjusted gains from trade poorer segment of the income distribution, whereas the consumption
G(ε) against the gains from trade G(0), for ε = 1 (panel a) and ε = 2 (panel b). The gains are more widely spread.
symbols represent the trade-off resolution: : no trade-off, liberalize; : soft trade-off, The role of household heterogeneity is typically underplayed in
liberalize; : policy reversal, liberalize; : no trade-off, protect; : policy reversal, much of trade theory. There are theories that postulate non-
protect; : soft trade-off, protect.
homothetic preferences and expenditure heterogeneity, but income
heterogeneity is often ignored. This can help rationalize potential dis-
IV (thus preferring protection) while Mauritania jumps from orthant III crepancies between our results and the intuitions regarding the gains
to orthant II (thus preferring liberalization). from trade and their distributive impacts that stem from many trade
A fundamental conclusion of this analysis is therefore that many of models. For example, the Stolper-Samuelson result could imply a reduc-
the countries that face a trade-off between mean income and its distri- tion in inequality for low-income, unskilled intensive countries that in-
bution are better off with lower tariffs (liberalization) than with higher tegrate with the world. Our model shows that this effect can be offset
tariffs (protection). Concretely, for empirically plausible levels of in- and fully dominated by impacts on other sources of income also affected
equality aversion, liberalization is expected to enhance welfare in 39 by trade.14
countries and to reduce it in 9 countries. Only in the remaining 6 coun-
tries are the policy implications more equivocal.
These results raise questions about why these countries protect their 5.6. Robustness
economies. Perhaps countries are not maximizing the Atkinson social
welfare function when setting tariffs. Alternatively, they may be maxi- To assess the robustness of our results, we use two different permu-
mizing this function, but subject to constraints. These could include po- tations of our model, shown in Table 9. To start with, we show results of
litical economy considerations such as rent re-distribution to non-labor
13
income (capital) and tariff revenue capture (Grossman and Helpman, Consumption and income impacts may interact, such that the total equality gains from
trade are not simply equal to the sum of the consumption equality gains and income
1994; Krueger, 1974). In addition, tariffs may be an appealing means equality gains; to assess the importance of these types of interaction effects, we calculated
of collecting revenue in a context in which income (and other) taxes “residual” equality gains as the difference between total equality gains and the sum of in-
are difficult to collect (Besley and Persson, 2013). Theories of social mo- come and consumption equality gains. The residual was typically very small and is there-
bility and redistribution that combine psychology and political economy fore not presented here.
14
Of course, the heterogeneity at the household level is in itself a consequence of addi-
may also rationalize the political constraints faced by policy makers
tional underlying factors, such as endowments, policies, institutions and so on. However,
(Piketty, 1995; Benabou and Tirole, 2006). Finally, there may be the analysis of these factors is beyond the scope of our paper.
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 15

Table 8 Table 9
Decomposing equality gains. Robustness tests.

Consumption Equality Income Equality Gains Baseline No Wage No tariff Alternative

Gains Response redistribution redistribution

ε = 0.5 ε=1 ε = 10 ε = 0.5 ε=1 ε = 10 N winners 45 50 53 45

N losers 9 4 1 9
Benin 0.9 1.4 2.9 −0.4 −0.7 −1.1
Burkina Faso 0.8 1.4 2.8 −0.3 −0.6 −3.0 Average (Inequality Adjusted) Gains from trade reform
Burundi 0.6 1.3 6.7 −0.1 −0.3 −3.5 G(0) 1.9% 2.5% 3.7% 1.9%
Cameroon 0.5 1.0 3.3 −0.1 −0.1 0.7 G(1) 1.5% 2.4% 3.3% 2.3%
Central African Republic 0.3 0.5 0.2 0.1 0.2 0.6 G(1.5) 1.3% 2.2% 3.1% 2.2%
Comoros 0.4 0.8 0.8 −0.0 −0.1 −0.5 G(2) 1.1% 2.1% 3.0% 2.1%
Côte d'Ivoire 0.3 0.6 1.4 −0.2 −0.3 −1.2 G(10) −0.4% 0.9% 1.5% 0.7%
Egypt, Arab Rep. 0.1 0.2 0.8 0.0 0.0 −0.4 Countries without tradeoffs 11 11 8 12
Ethiopia 0.6 1.0 1.6 −0.4 −0.9 −4.4 of which prefer freer trade 5 10 8 12
Gambia, The 0.2 0.4 0.3 −0.1 −0.2 −0.7 Countries with tradeoffs 43 43 46 42
Ghana 0.2 0.4 1.0 −0.1 −0.3 −2.2 of which prefer freer trade
Guinea 0.2 0.4 0.6 −0.1 −0.3 −9.4 ϵ=1 39 39 42 38
Guinea-Bissau 0.3 0.5 0.3 −0.2 −0.4 −2.4 ϵ = 1.5 36 39 42 37
Kenya 0.2 0.6 1.8 −0.2 −0.3 −2.0 ϵ=2 35 38 42 35
Liberia 0.4 0.6 0.5 −0.3 −0.6 −1.3 ϵ = 10 27 27 30 23
Madagascar 0.2 0.3 0.6 −0.1 −0.3 −1.2
Malawi 0.0 0.0 −0.3 −0.0 −0.1 −0.5 Total number of countries that prefer freer trade
Mali −0.0 −0.0 0.4 −0.0 −0.0 0.2 ϵ=0 45 50 53 45
Mauritania 0.1 0.2 0.5 −0.1 −0.3 −0.7 ϵ=1 44 49 50 50
Mozambique −0.0 −0.0 −0.1 −0.0 −0.0 0.1 ϵ = 1.5 41 49 50 49
Niger −0.0 −0.0 −0.3 −0.0 −0.0 −0.4 ϵ=2 40 48 50 47
Nigeria 0.2 0.4 1.0 −0.3 −0.5 −1.8 ϵ = 10 32 37 38 35
Rwanda −0.0 −0.0 −0.2 −0.1 −0.1 −0.4 Notes: The table summarizes the results of various alternative models to assess the robust-
Sierra Leone 0.2 0.4 −0.1 −0.3 −0.6 −2.7 ness of the results obtained using our baseline model present in column 1. The alternative
South Africa −0.0 0.0 0.4 −0.1 −0.1 −0.7 model presented in column 2 does not allow for labor market responses. Column 3 shows
Tanzania −0.0 −0.1 −1.0 −0.0 −0.1 0.0 the results of a model without tariff redistribution, i.e. in which governments do not in-
Togo 0.1 0.1 −0.4 −0.2 −0.3 −1.4 crease taxes to make up for the loss of government revenue. Column 4 shows the results
Uganda 0.9 1.5 2.0 −1.0 −2.0 −11.7 of model in which the government makes up the loss in tariff revenue by means of pro-
Zambia −0.0 −0.2 −1.5 −0.1 −0.2 0.3 gressive taxes. Both of these last two models are discussed in greater detail in the Appen-
Armenia −0.1 −0.2 −0.2 −0.1 −0.2 −1.8 dix, which also presents country specific results for all these models.

Consumption Equality Income Equality Gains

Gains
a model that does not allow for labor market responses to tariff changes.
ε = 0.5 ε=1 ε = 10 ε = 0.5 ε=1 ε = 10 With this alternative model, we find income gains in 50 countries and
Bangladesh 0.1 0.2 0.0 −0.3 −0.5 −2.2 losses in 4 countries. The estimated average income gains from trade
Bhutan 0.3 0.7 1.7 −0.4 −0.9 −6.7 are 2.5%. Yet, the estimated pattern of inequality adjusted gains is very
Cambodia 1.0 1.6 −0.2 −1.2 −2.3 −9.7 similar. Gains from trade are negatively correlated with equality gains,
Indonesia 0.1 0.2 0.7 −0.3 −0.7 −3.8
which tend to become lower (i.e. more negative) as inequality aversion
Iraq 0.2 0.4 −0.6 −0.5 −0.9 −2.9
Jordan −0.1 −0.3 −4.3 −0.2 −0.5 −5.0 increases. Nonetheless, in spite of the widespread prevalence of trade-
Kyrgyz Republic −0.1 −0.1 −0.3 −0.2 −0.4 −0.7 offs, countries are typically better off when pursuing free trade policies.
Mongolia −0.0 −0.1 1.4 −0.3 −0.5 −2.5 Our results are thus robust to using this alternative model (see Table E1
Nepal −0.1 −0.1 2.3 −0.2 −0.4 −2.4 for country-specific estimates).
Pakistan −0.1 −0.2 0.3 −0.2 −0.5 −2.8
Papua New Guinea −0.4 −0.8 −2.6 0.0 0.0 −0.3
Second, we re-evaluate our model using two alternative assumptions
Sri Lanka 0.0 −0.0 −0.9 −0.4 −0.8 −2.8 about tariff redistribution, notably (i) that no tax response is observed
Tajikistan 0.2 0.4 0.7 −0.6 −1.3 −3.2 and (ii) that the government makes up for lost revenue by imposing addi-
Uzbekistan 0.1 0.1 −0.1 −0.6 −1.0 −3.7 tional personal income taxes which respect the progressivity of the
Vietnam 0.2 0.3 0.1 −0.7 −1.4 −6.1
existing personal income tax system (see Appendix E for details). Exclud-
Yemen, Rep. −0.4 −0.7 −1.2 −0.2 −0.4 −1.5
Azerbaijan −0.0 −0.2 −0.5 −0.6 −1.3 −6.4 ing the loss of tariff revenue from the model nearly doubles the estimated
Georgia −0.2 −0.4 −1.1 −0.4 −0.7 −3.3 average income gains from trade to 3.7% on average across countries (rel-
Moldova −0.3 −0.5 −1.9 −0.4 −0.7 −2.4 ative to 1.9% in the main model). Moreover, income gains are now positive
Ukraine −0.1 −0.2 −4.3 −0.5 −1.1 −0.5 in 53 out of the 54 countries. Yet, equality gains continue to be negatively
Bolivia −0.5 −1.1 −3.8 −0.2 −0.4 −2.7
Ecuador 0.1 0.1 −0.7 −0.8 −1.8 −5.3
correlated with income gains and trade-offs remain widespread.
Guatemala −0.4 −0.7 −3.3 −0.4 −0.8 −3.4 If instead of assuming away tariff redistribution we assume that the
Nicaragua 0.9 1.9 0.5 −2.6 −4.5 −9.4 government makes up for the loss in tariff revenue by imposing progres-
Average 0.2 0.3 0.1 −0.3 −0.6 −2.6 sive taxes (calibrated using the World Tax Indicators (WTI) database
Pop. weighted av. 0.1 0.2 0.4 −0.3 −0.6 −2.8
(Andrew Young School of Policy Studies, 2010)), the inequality costs of
GDP weighted av. 0.1 0.2 0.2 −0.3 −0.6 −2.7
trade liberalization show some minor difference, but the overall pattern
Notes: The table presents the decomposition of the equality gains from trade G(ε) − G(0).
of results is not impacted. Income gains are negatively correlated with
The first three column report the average consumption equality gains from trade for dif-
ferent values of inequality aversion (ε = 0.5, ε = 1, and ε = 10). These consumption
equality gains. Yet, the latter typically are dominated by the former such
equality gains are calculated by assuming that liberalization only impacts consumption that countries facing trade-offs are typically better of with freer trade.15
and not income. The three remaining columns report the income equality gains from
trade for different values of inequality aversion (ε = 0.5, ε = 1, and ε = 10), calculated
by assuming that liberalization only impacts income but not consumption.

15
See Appendix E for details.
16 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45

Table 10 the results in Table 10 which presents estimates of the (inequality ad-
Protectionist scenarios. justed) gains from trade, the number of countries that exhibit trade-
Baseline 10% 10% Increase offs, and the trade policy reversals. Country-specific estimates of the
(Liberalization) relative absolute to gains from trade are presented in Appendix F.17
increase increase 62.4% As expected, the average gains from these protectionist trade re-
N winners 45 6 11 11 forms are now negative in the vast majority of countries. While there
N losers 9 48 43 43 are 45 winners under liberalization, we find between 43 and 48 losers
Average (Inequality Adjusted) Gains from trade reform with increased protection. The estimated average income gains (G(0))
G(0) 1.9% −0.2% −1.3% −5.7% across countries in the three scenarios are, respectively −0.2%, −1.3%
G(1) 1.5% −0.2% −1.3% −7.2% and −5.7% (Panel A of Table 10). More importantly, the prevalence of
G(1.5) 1.3% −0.2% −1.3% −7.8%
trade-offs is widespread (Panel B). There are only 8–10 countries with-
G(2) 1.1% −0.2% −1.3% −8.3%
G(10) −0.4% −0.2% −1.7% −17.9% out trade-offs and, of these, between 6 and 9 prefer the status quo to
Countries without 11 8 9 10 more protection. There are between 44 and 46 countries with
tradeoffs trade-offs. The resolution of these trade-offs is fairly stable in favor of
of which prefer freer 5 6 7 9 the status quo for plausible levels of ε (1, 1.5 or 2). That is, in all three
trade
protectionist scenarios the vast majority of countries attain higher levels
Countries with tradeoffs 43 46 45 44
of which prefer freer of welfare with the pre-existing structure of trade protection rather
trade than with more protection. Most countries would be hurt by protection-
ϵ=1 39 42 39 37 ist trade reforms, even after inequality impacts are taken into
ϵ = 1.5 36 43 38 37
consideration.
ϵ=2 35 41 36 37
ϵ = 10 27 33 36 42
6. Conclusion
Total number of countries that prefer freer trade
ϵ=0 45 48 43 43
ϵ=1 44 48 46 46 Using household survey data for 54 low and middle income coun-
ϵ = 1.5 41 49 45 46 tries harmonized with trade and tariff data, this paper offers a quantita-
ϵ=2 40 47 43 46 tive assessment of the income gains and inequality costs of trade
ϵ = 10 32 39 43 51 liberalization and the potential trade-off between them.
Notes: The table summarizes the results of three alternative protectionist trade reforms. A stylized yet comprehensive model that allows for a rich range of
Column 1 replicates the results of our baseline exercise in which a country eliminates its first-order effects on household consumption and income is used to
own import tariffs as a benchmark. Column 2 shows what would happen if countries
quantify welfare gains or losses for households in different parts of the
were to increase their tariffs by 10% in relative terms, i.e. if all tariffs were multiplied by
1.1. Column 3 shows what would happen if all countries were to increase all of their im- expenditure distribution. These welfare impacts are subsequently ex-
port tariffs by 10 percentage points. Column 4 demonstrates the gains from trade and plored by deploying the Atkinson social welfare function that allows
trade-offs between income gains and inequality costs if countries were to increase all of us to decompose inequality adjusted gains into aggregate gains and
their tariffs to 62.4%, or leave them at their pre-existing levels in case tariffs were already equality (distributional) gains.
in excess of 62.4%. See the Appendix for country specific results.
Liberalization is estimated to lead to income gains in 45 countries in
our study, and to income losses in 9 countries. The developing world as
5.7. Protectionist scenarios a whole would enjoy gains of about 1.9% of real household expendi-
tures, on average. These income gains are negatively correlated with
Thus far, we have analyzed the inequality-adjusted gains from liber- equality gains, such that liberalization typically entails a trade-off be-
alization, but our framework also lends itself to evaluating trade-offs tween average incomes and income inequality. In fact, such trade-offs
that might arise if countries become more protectionist. To explore arise in 45 out of 54 countries, and are primarily the result of trade ex-
this, we evaluate the welfare impacts associated with three protection- acerbating income inequality. By contrast, consumption gains tend to
ist scenarios that move the economy closer to autarky: (i) a uniform in- be more evenly spread across households.
crease in tariffs of 10%age points (i.e. adding 10% to all existing tariffs); While trade-offs are prevalent, our findings also suggest that liberal-
(ii) a relative increase in tariffs of 10 percent (i.e. multiplying all pre- ization would be welfare enhancing in the vast majority of countries in
existing tariffs by 1.1); and (iii) increasing all tariffs to 62.4%.16 The our study: in a large part of the developing world, the current structure
aim of this exercise is to establish whether the trade-offs that may of tariff protection is inducing sizable welfare losses. Explaining what
arise under protectionist scenarios are in general consistent with drives these patterns is beyond the scope of this paper but an interesting
those derived when countries liberalize. Accordingly, we summarize avenue for future research.

16
Following Ossa (2014), this is the level of tariffs that would prevail if countries did not fear retaliation from the rest of the world.
17
These country-specific losses/gains from trade under these various scenarios are not necessarily the mirror image of the results obtained when liberalizing. This is because the autarky
scenarios may have different impacts on prices and because the household heterogeneity noted above implies that the inequality implications of a positive welfare effect may be quite
different from those of negative effects.
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 17

Appendix A. Household surveys and data harmonization

Table A1 displays basic information on the household surveys used in the analysis. We report the name of the survey, the year when the data were
collected and sample sizes (number of households). The harmonization of the household surveys and the trade and trade policy data was done in two
steps. First, all household surveys product and income sources were standardized to common templates, which are shown in Fig. A1, Fig. A2, Fig. A3
below. Second, these harmonized household survey data were merged with HS6 tariff and trade data using custom-made concordances.

Table A1
Household surveys.

Country Year Obs Survey

Benin 2003 5296 Questionnaire Unifié sur les Indicateurs de Base du Bien-Être
Burkina Faso 2003 8413 Enquête sur les Conditions de Vie des Ménages
Burundi 1998 6585 Enquête Prioritaire, Etude Nationale sur les Conditions de Vie des Populations
Cameroon 2001–2002 10,881 Deuxième Enquête Camerounaise Auprès des Ménages
Central African Republic 2008 6828 Enquête Centrafricaine pour le Suivi-Evaluation du Bien-être
Comoros 2004 2929 Enquête Intégrale auprès des Ménages
Côte d'Ivoire 2008 12,471 Enquête sur le Niveau de Vie des Ménages
Egypt, Arab Rep. 2008–2009 23,193 Household Income, Expenditure and Consumption Survey
Ethiopia 1999–2000 16,505 Household Income, Consumption and Expenditure Survey
The Gambia 1998 1952 Household Poverty Survey
Ghana 2005–2006 8599 Living Standards Survey V
Guinea 2012 7423 Enquête Légère pour l'Evaluation de la Pauvreté
Guinea-Bissau 2010 3141 Inquerito Ligeiro para a Avalicão da Pobreza
Kenya 2005 13,026 Integrated Household Budget Survey
Liberia 2014–2015 4063 Household Income and Expenditure Survey
Madagascar 2005 11,661 Permanent Survey of Households
Malawi 2004–2005 11,167 Second Integrated Household Survey
Mali 2006 4449 Enquête Légère Intégrée auprès des Ménages
Mauritania 2004 9272 Enquête Permanente sur les Conditions de Vie des Ménages
Mozambique 2008–2009 10,696 Inquérito sobre Orçamento Familiar
Niger 2005 6621 Enquête Nationale sur les Conditions de Vie des Ménages
Nigeria 2003–2004 18,603 Living Standards Survey
Rwanda 1998 6355 Integrated Household Living Conditions Survey
Sierra Leone 2011 6692 Integrated Household Survey
South Africa 2000 25,491 General Household Survey
Tanzania 2008 3232 Household Budget Survey
Togo 2011 5464 Questionnaire des Indicateurs de Base du Bien-être
Uganda 2005–2006 7350 National Household Survey
Zambia 2004 7563 Living Conditions Monitoring Survey IV
Armenia 2014 5124 Integrated Living Conditions Survey
Bangladesh 2010 12,117 Household Income and Expenditure Survey
Bhutan 2012 8879 Living Standards Survey
Cambodia 2013 3801 Socio-Economic Survey
Indonesia 2007 12,876 Indonesian Family Life Survey
Iraq 2012 24,895 Household Socio-Economic Survey
Jordan 2010 11,110 Household Expenditure and Income Survey
Krygyz Republic 2012 4962 Intergrated Sample Household Budget and Labor Survey
Mongolia 2011 11,089 Household Socio-Economic Survey
Nepal 2010–2011 5929 Living Standards Survey
Pakistan 2010–2011 16,178 Social and Living Standards Measurement Survey
Papua New Guinea 2009 3776 Household Income and Expenditure Survey
Sri Lanka 2012–2013 20,335 Household Income and Expenditure Survey
Tajikistan 2009 1488 Tajikistan Panel Survey
Uzbekistan 2003 9419 Household Budget Survey
Vietnam 2012 9306 Household Living Standard Survey
Yemen, Rep. 2005–2006 12,998 Household Budget Survey
Azerbaijan 2005 4797 Household Budget Survey
Georgia 2014 10,959 Household Integrated Survey
Moldova 2014 4836 Household Budget Survey
Ukraine 2012 10,394 Sampling Survey of the Conditions of Life of Ukraine's Households
Bolivia 2008 3900 Encuesta de Hogares
Ecuador 2013–2014 28,680 Encuesta de Condiciones de Vida
Guatemala 2014 11,420 Encuesta Nacional de Condiciones de Vida
Nicaragua 2009 6450 Nicaragua - Encuesta Nacional de Hogares sobre Medición de Niveles de Vida
18 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45

Expenditure
1. Agriculture/Food
11. Staple Food

111. Cereals 112. Legumens 113. Fruits 114. Vegetables 115. Oils/Fats 116. Fish 117. Meat/Livestock 118. Dairy/Eggs 119. Other staple food

1111. Corn 1121. Beans 1131. Banana 1141. Tomato 1151. Vegetable Oils 1161. Fish 1171. Pork (Pig) 1181. Milk 1191. Other staple food
1112. Wheat 1122. Other 1132. Grapes 1142. Potato 1152. Animal Fats 1162. Shrimp 1172. Beef (Cattle) 1182. Eggs 1192. Other processed food
1113. Rice 1133. Citrus 1143. Greens 1153. Other oils/fats 1163. Other Crustacean 1173. Poultry (Chicken) 1183. Cheese
1114. Other Cereals 1134. Apples 1144. Other 1174. Other meat/animals 1184. Other Dairy
1135. Other Fruits Vegetables

12. Non Staple

121. Alcohol 122. Tobacco 123. Oil seeds 124. Spices/herbs 125. Coffee/tea/cocoa 126. Nuts 127. Cotton 128. Other non-staple food

1211. Wine 1221. Cigarettes 1231. Soya 1241. Cloves 1251. Coffee 1261. Cashew 127. Cotton 1281. Sugar (any kind)
1212. Beer 1222. Other tobacco 1232. Other oil seeds 1242. Pepper 1252. Tea 1262. Coconut 1282. Other non-staple
1213. Other alcohol 1243. Vanilla 1253. Cocoa 1263. Other nuts
1244. Saffron
1245. Qat (chat)
1246. Other spices

2. Manufacturing/Household Items
21. Energy
22. Textiles/Apparel
23. Electric/Electronics
24. Household items/Furniture
25. Other physical goods

3. Services
31. Transportation
32. Health
33. Education
34. Communication
35. Other Services

4. Other Expenditures
41. Remittances/transfers given
42. Investment of any sort
43. Festivities
44. Other Disbursement

Fig. A1. Expenditure template.

Autoconsumption
1. Agriculture/Food
11. Staple Food

111. Cereals 112. Legumens 113. Fruits 114. Vegetables 115. Oils/Fats 116. Fish 117. Meat/Livestock 118. Dairy/Eggs 119. Other staple food

1111. Corn 1121. Beans 1131. Banana 1141. Tomato 1151. Vegetable Oils 1161. Fish 1171. Pork (Pig) 1181. Milk 1191. Other staple food
1112. Wheat 1122. Other 1132. Grapes 1142. Potato 1152. Animal Fats 1162. Shrimp 1172. Beef (Cattle) 1182. Eggs 1192. Other processed food
1113. Rice 1133. Citrus 1143. Greens 1153. Other oils/fats 1163. Other Crustacean 1173. Poultry (Chicken) 1183. Cheese
1114. Other Cereals 1134. Apples 1144. Other 1174. Other meat/animals 1184. Other Dairy
1135. Other Fruits Vegetables

12. Non Staple

121. Alcohol 122. Tobacco 123. Oil seeds 124. Spices/herbs 125. Coffee/tea/cocoa 126. Nuts 127. Cotton 128. Other non-staple food

1211. Wine 1221. Cigarettes 1231. Soya 1241. Cloves 1251. Coffee 1261. Cashew 127. Cotton 1281. Sugar (any kind)
1212. Beer 1222. Other tobacco 1232. Other oil seeds 1242. Pepper 1252. Tea 1262. Coconut 1282. Other non-staple
1213. Other alcohol 1243. Vanilla 1253. Cocoa 1263. Other nuts
1244. Saffron
1245. Qat (chat)
1246. Other spices

2. Other goods
21. Energy (wood, coal)
22. Gathering (forest, mushrooms, berries, etc.)
23. Other goods collected for free
24. Other goods produced and consumed within the household

Fig. A2. Auto-consumption template.

 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 19

Income
1. Agriculture/Food
11. Staple Food

111. Cereals 112. Legumens 113. Fruits 114. Vegetables 115. Oils/Fats 116. Fish 117. Meat/Livestock 118. Dairy/Eggs 119. Other staple food

1111. Corn 1121. Beans 1131. Banana 1141. Tomato 1151. Vegetable Oils 1161. Fish 1171. Pork (Pig) 1181. Milk 1191. Other staple food
1112. Wheat 1122. Other 1132. Grapes 1142. Potato 1152. Animal Fats 1162. Shrimp 1172. Beef (Cattle) 1182. Eggs 1192. Other processed food
1113. Rice 1133. Citrus 1143. Greens 1153. Other oils/fats 1163. Other Crustacean 1173. Poultry (Chicken) 1183. Cheese
1114. Other Cereals 1134. Apples 1144. Other 1174. Other meat/animals 1184. Other Dairy
1135. Other Fruits Vegetables

12. Non Staple

121. Alcohol 122. Tobacco 123. Oil seeds 124. Spices/herbs 125. Coffee/tea/cocoa 126. Nuts 127. Cotton 128. Other non-staple food

1211. Wine 1221. Cigarettes 1231. Soya 1241. Cloves 1251. Coffee 1261. Cashew 127. Cotton 1281. Sugar (any kind)
1212. Beer 1222. Other tobacco 1232. Other oil seeds 1242. Pepper 1252. Tea 1262. Coconut 1282. Other non-staple
1213. Other alcohol 1243. Vanilla 1253. Cocoa 1263. Other nuts
1244. Saffron
1245. Qat (chat)
1246. Other spices

2. Wages
20. Agriculture, forestry, and fishing
21. Mining, oil, and gas extraction
22. Manufacturing
23. Construction
24. Transportation, communications, electric, gas, and sanitary services
25. Wholesale and retail trade
26. Finance, insurance, and real estate
27. Entertainment Services (Restaurant, entertainment, hotels, etc.)
28. Professional Services (Education, health, other professional occupations)
29. Public Administration

3. Sales of Goods/Services
30. Agriculture, forestry, and fishing (n.e.c.)
31. Mining, oil, and gas extraction
32. Manufacturing
33. Construction
34. Transportation, communications, electric, gas, and sanitary services
35. Wholesale and retail trade
36. Finance, insurance, and real estate
37. Entertainment Services (Restaurant, entertainment, hotels, etc.)
38. Professional Services (Education, health, other professional occupations)
39. Public Administration

4. Transfers
41. Remittances/transfers received (friend, relative)
42. Profits of investment (rent, interests)
43. Government transfers
44. Non-governmental transfers
45. Other

Fig. A3. Income template.

Appendix B. Distributional effects

This Appendix includes plots of the distributional effects (kernel regressions and bivariate kernel densities) for each of the 54 countries. We first re-
port 17 cases with a pro-poor bias (Figs. B1, B2, B3), then show another 37 cases with a pro-rich bias (Figs. B4, B5, B6, B7, B8, B9, B10).
20 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45
10

20
10
5
welfare effects

welfare effects
0

0 -10
-5
-10

-20
11.5 12 12.5 13 13.5 14 6 8 10 12
log per capita expenditure log per capita expenditure

10
10

5
0
welfare effects

welfare effects
-10

0 -5
-20

-10
-30

3 4 5 6 7 8 6 8 10 12 14
log per capita expenditure log per capita expenditure
15

20
10

10
welfare effects

welfare effects
5

0
0

-10
-5

4 5 6 7 8 9 3 4 5 6 7
log per capita expenditure log per capita expenditure

Fig. B1. Pro-poor bias. Notes: The red curve is the non-parametric kernel regression of the welfare effects and the initial level of per capita household expenditure. The contour lines are
level curves of the non-parametric kernel bivariate density of these two variables. Liberalization is classified as having a pro-poor bias if the average proportional real income gains accruing
to households in the the bottom 20% of the pre-liberalization income distribution exceed the average proportional real income gains accruing to households in the top 20% of the pre-
liberalization real income income distribution. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 21

5
10

0
5
welfare effects

welfare effects
-5
0

-10
-5
-10

-15
8 9 10 11 12 13 7 8 9 10 11 12
log per capita expenditure log per capita expenditure
10

5
5

0
welfare effects

welfare effects
0

-5
-5

-10
-10
-15

-15

6 7 8 9 9 10 11 12 13
log per capita expenditure log per capita expenditure
10

10
5

5
welfare effects

welfare effects
0

0
-5

-5
-10

-10

7 8 9 10 11 6 8 10 12
log per capita expenditure log per capita expenditure

Fig. B2. Pro-poor bias. Notes: The red curve is the non-parametric kernel regression of the welfare effects and the initial level of per capita household expenditure. The contour lines are
level curves of the non-parametric kernel bivariate density of these two variables. Liberalization is classified as having a pro-poor bias if the average proportional real income gains accruing
to households in the the bottom 20% of the pre-liberalization income distribution exceed the average proportional real income gains accruing to households in the top 20% of the pre-
liberalization real income income distribution. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
22 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45
20

20
10

10
welfare effects

welfare effects
0

0
-10

-10
-20

-20
2 4 6 8 6 8 10 12
log per capita expenditure log per capita expenditure
20

10
5
10
welfare effects

welfare effects
0
0

-5
-10

-10
-20

-15

7 8 9 10 11 12 7 8 9 10 11 12
log per capita expenditure log per capita expenditure
15
10
welfare effects
0 -5
-10 5

9 10 11 12 13
log per capita expenditure

Fig. B3. Pro-poor bias. Notes: The red curve is the non-parametric kernel regression of the welfare effects and the initial level of per capita household expenditure. The contour lines are
level curves of the non-parametric kernel bivariate density of these two variables. Liberalization is classified as having a pro-poor bias if the average proportional real income gains accruing
to households in the the bottom 20% of the pre-liberalization income distribution exceed the average proportional real income gains accruing to households in the top 20% of the pre-
liberalization real income income distribution. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 23

10
5

5
0
welfare effects

welfare effects
-5 0
-5

-10
-15
-10

9 10 11 12 13 -1 0 1 2 3
log per capita expenditure log per capita expenditure

20
10

10
5
welfare effects

welfare effects
0

0
-5

-10
-10

-20
-15

-30

7 8 9 10 11 12 7 8 9 10 11
log per capita expenditure log per capita expenditure
10
10

5
5

welfare effects
welfare effects

0
0

-5
-5

-10
-10

-15

4 6 8 10 7 8 9 10 11 12
log per capita expenditure log per capita expenditure

Fig. B4. Pro-rich bias. Notes: The red curve is the non-parametric kernel regression of the welfare effects and the initial level of per capita household expenditure. The contour lines are level
curves of the non-parametric kernel bivariate density of these two variables. Liberalization is classified as having a pro-rich bias if the average proportional real income gains accruing to
households in the the top 20% of the pre-liberalization income distribution exceed the average proportional real income gains accruing to households in the bottom 20% of the pre-
liberalization real income income distribution. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
24 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45
20

10
10

0
welfare effects

welfare effects
-10
0-10

-20
-20

-30
4 6 8 10 12 4 5 6 7 8 9
log per capita expenditure log per capita expenditure
20

10
10

5
welfare effects

welfare effects
0

0
-10

-5
-20

-10

8 9 10 11 12 13 6 8 10 12 14
log per capita expenditure log per capita expenditure
15
10

10
5
welfare effects

welfare effects
5
0

0
-5

-5
-10
-15

-10

6 8 10 12 14 4 5 6 7 8
log per capita expenditure log per capita expenditure

Fig. B5. Pro-rich bias. Notes: The red curve is the non-parametric kernel regression of the welfare effects and the initial level of per capita household expenditure. The contour lines are level
curves of the non-parametric kernel bivariate density of these two variables. Liberalization is classified as having a pro-rich bias if the average proportional real income gains accruing to
households in the the top 20% of the pre-liberalization income distribution exceed the average proportional real income gains accruing to households in the bottom 20% of the pre-
liberalization real income income distribution. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 25

20

10
10

5
welfare effects

welfare effects
0
0

-5
-10

-10
-20

-15
3 4 5 6 7 2 4 6 8
log per capita expenditure log per capita expenditure

10
5

5
welfare effects

welfare effects
0

-5 0
-5

-10
-15
-10

-4 -3 -2 -1 0 8 10 12 14 16
log per capita expenditure log per capita expenditure
10

10
5
5

welfare effects
welfare effects

0
0

-5
-5

-10
-10

5 6 7 8 9 4 5 6 7 8
log per capita expenditure log per capita expenditure

Fig. B6. Pro-rich bias. Notes: The red curve is the non-parametric kernel regression of the welfare effects and the initial level of per capita household expenditure. The contour lines are level
curves of the non-parametric kernel bivariate density of these two variables. Liberalization is classified as having a pro-rich bias if the average proportional real income gains accruing to
households in the the top 20% of the pre-liberalization income distribution exceed the average proportional real income gains accruing to households in the bottom 20% of the pre-
liberalization real income income distribution. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
26 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45
4

20
10
2
welfare effects

welfare effects
0

0 -10
-2

-20
-4

10 11 12 13 14 15 4 6 8 10 12
log per capita expenditure log per capita expenditure
4

5
2

0
welfare effects

welfare effects
0

-5
-2

-10
-4

-15
-6

6 7 8 9 10 7 8 9 10 11
log per capita expenditure log per capita expenditure
20
10
5

10
welfare effects

welfare effects
0

0
-5

-10
-10
-15

-20

8 9 10 11 12 5 6 7 8 9 10
log per capita expenditure log per capita expenditure

Fig. B7. Pro-rich bias. Notes: The red curve is the non-parametric kernel regression of the welfare effects and the initial level of per capita household expenditure. The contour lines are level
curves of the non-parametric kernel bivariate density of these two variables. Liberalization is classified as having a pro-rich bias if the average proportional real income gains accruing to
households in the the top 20% of the pre-liberalization income distribution exceed the average proportional real income gains accruing to households in the bottom 20% of the pre-
liberalization real income income distribution. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 27

20
10

10
5
welfare effects

welfare effects
0

0
-5

-10
-10
-15

-20
4 6 8 10 5 6 7 8 9
log per capita expenditure log per capita expenditure

20
10

10
welfare effects

welfare effects
0

0
-10

-10
-20

-20

7 8 9 10 11 12 2 4 6 8 10
log per capita expenditure log per capita expenditure
15

10
10

5
welfare effects

welfare effects
5

0
0

-5
-5

-10
-10

-15

8 10 12 14 16 2 4 6 8 10
log per capita expenditure log per capita expenditure

Fig. B8. Pro-rich bias. Notes: The red curve is the non-parametric kernel regression of the welfare effects and the initial level of per capita household expenditure. The contour lines are level
curves of the non-parametric kernel bivariate density of these two variables. Liberalization is classified as having a pro-rich bias if the average proportional real income gains accruing to
households in the the top 20% of the pre-liberalization income distribution exceed the average proportional real income gains accruing to households in the bottom 20% of the pre-
liberalization real income income distribution. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
28 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45

20
6

10
4
welfare effects

welfare effects
2

0
0

-10
-2

-20
-4

-30
4 5 6 7 8 8 10 12 14
log per capita expenditure log per capita expenditure

20
10

10
5
welfare effects

welfare effects
0

0
-5

-10
-10
-15

-20

6 8 10 12 14 8 9 10 11 12 13
log per capita expenditure log per capita expenditure
15
6

10
4

welfare effects
welfare effects

5
2

0
0

-5
-2

-10

6 6.5 7 7.5 8 8.5 8 9 10 11 12

log per capita expenditure log per capita expenditure

Fig. B9. Pro-rich bias. Notes: The red curve is the non-parametric kernel regression of the welfare effects and the initial level of per capita household expenditure. The contour lines are level
curves of the non-parametric kernel bivariate density of these two variables. Liberalization is classified as having a pro-rich bias if the average proportional real income gains accruing to
households in the the top 20% of the pre-liberalization income distribution exceed the average proportional real income gains accruing to households in the bottom 20% of the pre-
liberalization real income income distribution. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 29

20
10
welfare effects
0 -10
-20

5 6 7 8 9
log per capita expenditure

Fig. B10. Pro-rich bias. Notes: The red curve is the non-parametric kernel regression of the welfare effects and the initial level of per capita household expenditure. The contour lines are
level curves of the non-parametric kernel bivariate density of these two variables. Liberalization is classified as having a pro-rich bias if the average proportional real income gains accruing
to households in the the top 20% of the pre-liberalization income distribution exceed the average proportional real income gains accruing to households in the bottom 20% of the pre-
liberalization real income income distribution. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
30 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45

Appendix C. Inequality adjusted welfare gains

8

7
7

6
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

6

5
5

4
4

3
3

2
2

1
1

0
0

0
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
6

1
5

5
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

4

4
3

3
2

2
1

1
0

0
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
4

4
inequality adjusted gains, G(ε)
3

3
2

2
1

1
0

0 2 4 6 8 10
inequality aversion, ε

Fig. C1. No trade-off, income gains and equality gains, no trade policy preference ranking reversals. Notes: The solid red line depicts how the inequality adjusted welfare gains associated
with liberalization G(ε) vary with inequality aversion ε. The dotted blue lines represent 95% confidence intervals based on 1000 bootstrap replications. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this article.)
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 31

0
-1

-1
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

-1

-1

-2

-2
-3

-3
-2

-2

-4

-4
-5

-5
-3

-3

-6

-6
-4

-4

-7

-7
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
0

1
0

0
-1

-1

-1

-1
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

-2

-2
-2

-2

-3

-3
-4

-4
-3

-3

-5

-5
-6

-6
-4

-4

-7

-7
-5

-5

-8

-8
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε

Fig. C2. No trade-off, income losses and inequality costs, no trade policy preference ranking reversals. Notes: The solid red line depicts how the inequality adjusted welfare gains associated
with liberalization G(ε) vary with inequality aversion ε. The dotted blue lines represent 95% confidence intervals based on 1000 bootstrap replications. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this article.)
32 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45
3

3
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

2

2
1

1
0

0
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
8

3
7

7
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

6

2
5

5
4

4
3

1
2

2
1

1
0

0
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
3

3
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

2

2
1

1
0

0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε

Fig. C3-A. Trade-offs, income gains and inequality costs, no trade policy preference ranking reversals. Notes: The solid red line depicts how the inequality adjusted welfare gains associated
with liberalization G(ε) vary with inequality aversion ε. The dotted blue lines represent 95% confidence intervals based on 1000 bootstrap replications. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this article.)
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 33

1
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

1

1
0

0
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
2

4
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

3

3
1

2
1

1
0

0
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
3

3
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

2

2
1

1
0

0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε

Fig. C3-B. Trade-offs, income gains and inequality costs, no trade policy preference ranking reversals. Notes: The solid red line depicts how the inequality adjusted welfare gains associated
with liberalization G(ε) vary with inequality aversion ε. The dotted blue lines represent 95% confidence intervals based on 1000 bootstrap replications. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this article.)
34 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45
4

4
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

3

3
2

2
1

1
0

0
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
4

8
7

7
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

3

6
5

5
2

4
3

3
1

2
1

1
0

0
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε

Fig. C3-C. Trade-offs, income gains and inequality costs, no trade policy preference ranking reversals. Notes: The solid red line depicts how the inequality adjusted welfare gains associated
with liberalization G(ε) vary with inequality aversion ε. The dotted blue lines represent 95% confidence intervals based on 1000 bootstrap replications. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this article.)
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 35

3
2

2
0

1
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

0

0
-1

-1

-1

-1
-2

-2
-2

-2

-3

-3
-4

-4
-3

-3

-5

-5
-4

-4

-6

-6
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
1

1
0

0
0

0
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

-1

-1
-1

-1

-2

-2
-2

-2

-3

-3
-3

-3

-4

-4
-4

-4

-5

-5
-5

-5

-6

-6
-6

-6

-7

-7
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
3

3
2

2
2

1
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

1

0
-1

-1
0

-2

-2
-1

-1

-3

-3
-4

-4
-2

-2

-5

-5
-3

-3

-6

-6

0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε

Fig. C4-A. Trade-offs, income gains and inequality costs, trade policy preference ranking reversals. Notes: The solid red line depicts how the inequality adjusted welfare gains associated
with liberalization G(ε) vary with inequality aversion ε. The dotted blue lines represent 95% confidence intervals based on 1000 bootstrap replications. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this article.)
36 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45
2

3
1

1
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

1

-1

-1
-3

-3
-5

-5
0

-7

-7
-9

-9
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
2

4
3

3
2

2
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

1

1
0

0
0

-1

-1
-2

-2
-1

-1

-3

-3
-4

-4
-2

-2

-5

-5
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
3

2
2

2
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

1

-1

-1
0

-4

-4
-1

-1
-2

-2

-7

-7
-3

-3

-10

-10
-4

-4

-13

-13
-5

-5

0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε

Fig. C4-B. Trade-offs, income gains and inequality costs, trade policy preference ranking reversals. Notes: The solid red line depicts how the inequality adjusted welfare gains associated
with liberalization G(ε) vary with inequality aversion ε. The dotted blue lines represent 95% confidence intervals based on 1000 bootstrap replications. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this article.)
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 37

3
4

2
1

1
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

1

1
-2

-2

0
-1

-1
-5

-5

-2

-2
-8

-8

-3

-3
-11

-11

-4

-4
-14

-14

-5

-5
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
2

2
1

1
inequality adjusted gains, G(ε)
0

0
-1

-1
-2

-2
-3

-3
-4

-4

0 2 4 6 8 10
inequality aversion, ε

Fig. C4-C. Trade-offs, income gains and inequality costs, trade policy preference ranking reversals. Notes: The solid red line depicts how the inequality adjusted welfare gains associated
with liberalization G(ε) vary with inequality aversion ε. The dotted blue lines represent 95% confidence intervals based on 1000 bootstrap replications. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this article.)
38 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45
4

4
3

3
3

3
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

2

2
1

1
2

0
1

-1

-1
-2

-2
0

-3

-3
-4

-4
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
2

4
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

3

3
1

2
1

1
0

0
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
2

3
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

2

2
1

1
0

0
0

0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε

Fig. C5-A. Trade-offs, income gains and inequality costs, potential trade policy preference ranking reversals. Notes: The solid red line depicts how the inequality adjusted welfare gains
associated with liberalization G(ε) vary with inequality aversion ε. The dotted blue lines represent 95% confidence intervals based on 1000 bootstrap replications. (For interpretation of
the references to colour in this figure legend, the reader is referred to the web version of this article.)
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 39

6
2

5
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

1

4
0

3
-1

-1

2
-2

-2

1
-3

-3

0
-4

-4

-1

-1
-5

-5

-2

-2
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
5

5
4

4
inequality adjusted gains, G(ε)
3

3
2

2
1

1
0

0 2 4 6 8 10
inequality aversion, ε

Fig. C5-B. Trade-offs, income gains and inequality costs, potential trade policy preference ranking reversals. Notes: The solid red line depicts how the inequality adjusted welfare gains
associated with liberalization G(ε) vary with inequality aversion ε. The dotted blue lines represent 95% confidence intervals based on 1000 bootstrap replications. (For interpretation of
the references to colour in this figure legend, the reader is referred to the web version of this article.)
40 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45

4
5

3
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

3

2
1

1
-1

-1

0
-3

-3

-1

-1
-5

-5

-2

-2
0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε
1

1
inequality adjusted gains, G(ε)
-2

-2
-5

-5
-8

-8
-11

-11
-14

-14

0 2 4 6 8 10
inequality aversion, ε

Fig. C6. Trade-offs, income losses and equality gains, trade policy preference ranking reversals. Notes: The solid red line depicts how the inequality adjusted welfare gains associated with
liberalization G(ε) vary with inequality aversion ε. The dotted blue lines represent 95% confidence intervals based on 1000 bootstrap replications. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article.)
0

0
-2

-2

-1

-1
inequality adjusted gains, G(ε)

inequality adjusted gains, G(ε)

-4

-4

-2

-2
-6

-6

-3

-3
-8

-8

-4

-4
-10

-10

-5

-5

0 2 4 6 8 10 0 2 4 6 8 10
inequality aversion, ε inequality aversion, ε

Fig. C7. Trade-offs, income losses and equality gains, no trade policy preference ranking reversals. Notes: The solid red line depicts how the inequality adjusted welfare gains associated
with liberalization G(ε) vary with inequality aversion ε. The dotted blue lines represent 95% confidence intervals based on 1000 bootstrap replications. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this article.)
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 41

Appendix D. Alternative model - no labor markets

Table D1
Inequality adjusted gains from trade and trade ϵ, no wage responses.

G(0) G(0.5) G(1) G(1.5) G(2) G(10) ϵ

Countries without trade-offs

Jordan 4.4 4.8 5.2 5.5 5.7 4.8
Central Afr. Rep. 4.2 4.6 5.0 5.3 5.6 6.3
Pakistan 3.2 3.5 4.0 4.5 4.8 4.5
Yemen, Rep. 3.1 3.3 3.4 3.5 3.6 3.5
Nicaragua 2.8 3.1 3.3 3.5 3.7 3.3
Indonesia 2.4 2.6 2.8 2.8 2.9 2.6
Guinea-Bissau 2.3 2.9 3.2 3.3 3.4 3.9
Nepal 1.7 1.9 2.1 2.2 2.3 2.7
Sri Lanka 1.0 1.8 2.6 3.1 3.5 4.3
Mongolia 0.3 0.4 0.6 0.7 0.8 0.8
Madagascar −0.2 −0.4 −0.7 −0.9 −1.1 −2.5

Countries with trade-offs

Nigeria 5.1 4.9 4.4 3.2 1.1 −5.0 2.25
Kenya 4.2 4.1 3.6 3.0 2.2 −6.1 2.96
Mozambique 4.0 3.3 2.5 1.8 1.2 −2.2 4.20
Togo 3.1 2.8 2.3 1.8 1.2 −0.6 3.84
Gambia, The 3.0 2.5 1.8 1.0 0.3 −0.9 2.21
Benin 2.5 1.7 0.9 0.3 −0.4 −4.1 1.71
Ethiopia 2.2 1.6 1.1 0.7 0.4 −1.5 3.06
Vietnam 2.1 1.9 1.7 1.4 1.1 −1.9 4.09
Bhutan 2.0 2.3 2.3 2.0 1.7 −4.6 4.11
Bangladesh 2.0 1.7 1.4 1.1 0.9 −1.1 4.24
Papua New Guinea 1.9 2.0 2.0 1.9 1.8 −8.5 4.49
B. Faso 1.3 0.9 0.4 0.2 −0.1 −2.6 1.86
Burundi 0.9 −0.9 −1.9 −2.2 −2.3 −4.5 0.25
Ghana 0.9 0.9 0.8 0.7 0.5 −1.5 2.90
Comoros 0.1 −0.1 −0.2 −0.4 −0.6 −1.9 0.30
Mali −0.3 0.1 0.7 1.3 1.9 3.0 0.37
Mauritania −1.2 −0.8 −0.4 −0.0 0.3 2.7 1.52
G(0) G(0.5) G(1) G(1.5) G(2) G(10) ϵ

Countries with trade-offs

Cameroon 8.1 7.8 7.5 7.2 6.9 4.7
Zambia 6.4 6.4 6.4 6.4 6.4 6.3
Tanzania 5.3 5.1 5.0 5.0 5.1 5.1
Uzbekistan 4.6 4.2 3.8 3.5 3.2 1.8
Ecuador 4.4 4.7 5.0 5.2 5.3 3.5
Sierra Leone 4.3 3.7 3.0 2.6 2.2 0.6
Côte d'Ivoire 4.0 3.7 3.3 2.9 2.5 0.9
Egypt, Arab Rep. 3.9 3.9 3.8 3.7 3.6 3.1
Ukraine 3.5 3.4 3.4 3.4 3.3 2.9
South Africa 3.5 3.4 3.3 3.0 2.7 1.8
Bolivia 3.4 3.5 3.4 3.1 2.8 0.3
Malawi 3.0 2.3 1.8 1.5 1.2 0.5
Guinea 2.9 2.7 2.4 2.1 2.0 2.0
Uganda 2.9 2.7 2.7 2.8 3.0 6.0
Guatamala 2.9 2.9 2.9 2.9 2.9 2.8
Armenia 2.7 2.6 2.5 2.4 2.4 1.9
Azerbaijan 2.6 2.7 2.7 2.7 2.7 2.1
Tajikistan 2.5 2.5 2.5 2.5 2.4 2.1
Liberia 2.0 1.9 1.7 1.5 1.3 0.3
Niger 2.0 1.7 1.3 1.1 0.9 1.2
Iraq 1.9 1.8 1.8 1.7 1.7 1.5
Georgia 1.1 1.0 1.0 1.0 0.9 0.5
Rwanda 0.9 1.1 1.5 2.0 2.6 4.0
Kyrgyz Republic 0.9 0.8 0.8 0.8 0.8 0.8
Moldova 0.8 0.8 0.8 0.8 0.8 1.4
Cambodia −2.2 −2.1 −2.1 −2.2 −2.3 −2.6
Average 2.5 2.5 2.4 2.2 2.1 0.9 2.61
Population weighted average 3.0 2.9 2.8 2.6 2.4 0.8 3.14
GDP weighted average 3.1 3.1 3.1 2.9 2.6 1.2 2.95

Notes: this table shows the gains from trade G(0), the inequality adjusted gains from trade, G(0.5), G(1), G(1.5), G(2) and G(10), and trade-ϵ when wages do not respond to the tariff
liberalization.
42 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45

Appendix E. Robustness to alternative tariff redistribution schemes Table E1

inequality adjusted gains from trade and trade ϵ, no tariff redistribution.

To assess the robustness of our results we consider two alternative tariff G(0) G(0.5) G(1) G(1.5) G(2) G(10) ϵ
redistribution scenarios: (1) the government does not make up the bud- Countries without trade-offs
get loss (i.e. tariff redistribution is ignored) and (2) the government Jordan 7.5 7.9 8.2 8.5 8.7 7.8
makes up for lost revenue by imposing additional personal income Central Afr. Rep. 6.3 6.7 7.1 7.4 7.7 8.4
taxes which respect the progressivity of the existing personal income Mauritania 5.2 5.6 6.1 6.4 6.8 9.2
Guinea-Bissau 4.3 4.8 5.1 5.1 5.2 5.2
tax system. To proxy the progressivity of taxes we use the World Tax In-
Yemen, Rep. 4.1 4.3 4.4 4.4 4.5 4.2
dicator (WTI) database (Andrew Young School of Policy Studies, 2010). Mongolia 2.5 2.6 2.7 2.8 2.9 3.0
This is the most comprehensive and comparable measure of tax pro- Mali 1.7 2.1 2.7 3.3 3.9 5.0
gressivity available (see e.g. Heathcote et al., 2017). More specifically, Sri Lanka 0.7 1.2 1.6 1.8 2.0 0.8
it offers measures of the average tax rates faced by households in differ- Ghana −0.1 −0.5 −0.9 −1.3 −1.6 −2.4

ent segments of the income distribution.18 Countries with trade-offs

For each country we calculate the increase in segment j specific tax Mozambique 5.6 5.0 4.2 3.5 3.0 −0.5 8.5
rates τj required to balance the budget to compensate for the revenue Benin 5.5 4.7 3.9 3.2 2.5 −1.1 5.0
Nigeria 4.6 4.3 3.8 2.7 0.7 −5.5 2.2
loss resulting from liberalization. To do so, we first calculate the scaling
Kenya 4.5 4.3 3.8 3.2 2.5 −4.7 3.3
factor λ.. Togo 4.4 3.7 2.7 1.8 0.9 −1.3 2.6
Ethiopia 3.7 3.0 2.5 2.1 1.9 −0.0 9.9
J X
X N Vietnam 2.9 2.7 2.4 2.1 1.8 −0.6 6.8
dT ¼ λ  Iij yi r j ð18Þ Burkina Faso 2.6 1.9 1.4 1.0 0.7 −1.9 3.9
j¼1 i¼1 Papua New Guinea 2.3 2.4 2.4 2.3 2.2 −8.0 4.8
Burundi 2.2 0.6 −0.4 −0.7 −0.6 −2.6 0.8
where dT is the anticipated tariff revenue loss associated with the liber- Bangladesh 1.7 1.3 0.9 0.6 0.3 −1.8 2.5
alization, Iij is a dummy variable that takes the value 1 if household i be- Comoros 1.7 1.6 1.5 1.3 1.1 −0.1 8.7
Rwanda 1.5 1.3 1.3 1.3 1.2 −3.4 4.1
longs to tax segment j and 0 otherwise, yi is a measure of (pre- Madagascar 0.7 0.3 −0.0 −0.3 −0.5 −1.7 1.0
additional tax) income, and rj is the average tax rate paid by households Cambodia 0.0 0.1 −0.0 −0.1 −0.3 −0.7 0.9
in segment j. After solving for λ we can calculate the “segment” specific
Notes: this table shows the gains from trade G(0), the inequality adjusted gains from
tax increase as τj = λ ∗ rj. These tax increases in turn are part of the in- trade, G(0.5), G(1), G(1.5), G(2) and G(10), and trade-ϵ when the loss in tariff revenue in-
come losses associated with liberalization imposing that the progressiv- curred when countries liberalize and the welfare consequences of the attendant loss in tar-
ity of the tax system is respected.19 iff revenue is not taken into consideration.
Tables E1 and E2 presents estimates of the inequality adjusted gains
Table E1
from trade using these alternative tariff redistribution scenarios. Not Inequality adjusted gains from trade and trade ϵ, no tariff redistribution.
considering tariff revenue losses almost doubles the average income
gains from trade which are now 3.7% on average across countries (as op- G(0) G(0.5) G(1) G(1.5) G(2) G(10) ϵ

posed to 1.9% in our main model). Moreover, only 1 of the 54 countries Countries with trade-offs (continued)
(Ghana) experiences (very modest) income losses when trade is liberal- Cameroon 8.7 8.2 7.6 7.1 6.7 3.8
Bhutan 7.9 7.9 7.5 6.9 6.3 0.8
ized. Yet, trade-offs remain widespread, as they are prevalent in 45
Zambia 6.7 6.7 6.7 6.6 6.6 6.7
countries. However, in the vast majority of cases social welfare would Tanzania 5.8 5.4 5.0 4.7 4.5 2.9
be higher in case a country would liberalize its own tariffs than in the Sierra Leone 5.7 5.1 4.5 4.0 3.6 2.1
status quo. Guinea 5.7 5.4 5.1 4.9 4.7 4.9
When imposing that the tax hikes introduced to compensate for the tar- Gambia, The 5.7 5.0 4.1 3.1 2.1 0.4
Uzbekistan 4.8 4.4 4.0 3.8 3.6 2.3
iff revenue loss respect the progressivity of the existing income tax sys-
Côte d'Ivoire 4.5 4.3 4.0 3.6 3.2 1.7
tem, the estimated income gains from trade do not change of course. Bolivia 4.2 4.1 3.9 3.7 3.4 1.5
Yet, as expected, inequality costs are somewhat lower when using pro- Ukraine 4.2 4.1 4.1 4.0 4.0 3.5
portional income taxes, as these tend to reduce inequality. Crucially, Ecuador 4.0 4.2 4.2 4.2 4.1 1.8
Tajikistan 3.9 3.9 3.8 3.7 3.6 3.4
however, the income and equality gains of trade continue to be nega-
Egypt, Arab Rep. 3.9 3.8 3.7 3.6 3.5 3.1
tively correlated. Yet, income gains tend to dominate inequality costs Niger 3.8 3.5 3.2 2.9 2.7 3.1
for plausible levels of inequality aversion. Azerbaijan 3.8 3.8 3.8 3.8 3.8 3.2
Put differently, our main conclusions appear robust to different ways of Malawi 3.6 3.0 2.5 2.1 1.9 1.1
modeling the response to the loss of tariff revenue that results from Nepal 3.4 3.5 3.6 3.5 3.5 2.4
Armenia 3.4 3.3 3.2 3.1 3.0 2.4
trade liberalization.
Uganda 3.3 3.0 2.8 2.7 2.7 3.6
Nicaragua 3.3 3.3 3.3 3.3 3.1 1.5
Pakistan 3.2 3.2 3.4 3.7 3.9 3.0
South Africa 3.1 3.0 2.8 2.5 2.4 1.9
Liberia 2.9 2.8 2.6 2.4 2.2 1.0
Iraq 2.8 2.7 2.7 2.6 2.6 2.5
Guatamala 2.8 2.6 2.4 2.2 2.0 0.7
Indonesia 2.5 2.6 2.7 2.7 2.7 2.1
Kyrgyz Republic 2.2 2.2 2.2 2.2 2.2 2.2
Moldova 2.2 2.1 2.1 2.1 2.1 2.7
Georgia 1.6 1.6 1.5 1.5 1.4 0.8
18 Average 3.7 3.5 3.3 3.1 3.0 1.5 4.3
These measures are adjusted for allowances/deductions, tax credits, significant local
Population weighted average 3.4 3.2 3.0 2.8 2.5 0.7 4.3
taxes and other main rules of the tax code. They are not, however, adjusted for deductions,
GDP weighted average 3.3 3.2 3.1 2.9 2.6 1.0 3.5
exemptions, and credits that depend on taxpayer specific characteristics (for example, no
adjustment is made for child credits). They also do not account for evasion and/or Notes: this table shows the gains from trade G(0), the inequality adjusted gains from
avoidance. trade, G(0.5), G(1), G(1.5), G(2) and G(10), and trade-ϵ when the loss in tariff revenue in-
19
This formula can be thought of as a crude approximation to the tax function T(y) = y curred when countries liberalize and the welfare consequences of the attendant loss in tar-
− λy1−τ. iff revenue is not taken into consideration.
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 43

Table E2 Appendix F. Protectionist scenarios

Inequality adjusted gains from trade and trade ϵ, alternative tariff redistribution scheme.

G(0) G(0.5) G(1) G(1.5) G(2) G(10) ϵ Table F1

Inequality adjusted gains from trade and trade ϵ, 10% absolute increase in tariffs.
Countries without trade-offs
Zambia 6.0 6.2 6.4 6.5 6.5 6.7 G(0) G(0.5) G(1) G(1.5) G(2) G(10) ϵ
Central Afr. Rep. 4.2 5.5 6.6 7.2 7.6 8.4 Countries without trade-offs
Jordan 4.1 4.7 5.3 5.7 6.1 5.3 Vietnam 0.9 1.0 1.2 1.3 1.5 2.0
Guinea 2.8 3.1 3.2 3.1 3.1 3.4 Burundi 0.2 0.8 1.2 1.4 1.5 1.6
Yemen, Rep. 2.7 3.0 3.2 3.4 3.4 3.2 Rwanda −0.2 −0.4 −0.8 −1.1 −1.4 −2.4
Pakistan 2.4 2.8 3.3 3.7 3.9 3.0 Jordan −0.4 −0.6 −0.7 −0.9 −1.0 −1.9
Guinea-Bissau 2.0 3.7 4.6 4.9 5.1 5.2 Kyrgyz Republic −0.7 −0.8 −0.8 −0.9 −1.0 −1.7
Uganda 2.0 2.3 2.5 2.6 2.7 3.6 Nicaragua −0.8 −0.9 −1.0 −1.1 −1.2 −1.5
Iraq 1.6 1.7 1.8 1.8 1.8 1.7 Tanzania −0.8 −0.8 −0.8 −0.8 −0.9 −1.3
Nepal 1.4 2.5 3.1 3.4 3.5 2.4 Bangladesh −1.2 −1.3 −1.3 −1.4 −1.5 −2.2
Moldova 0.7 0.8 0.9 1.0 1.0 1.7 Nepal −1.2 −1.4 −1.6 −1.7 −1.9 −2.5
Mongolia 0.1 0.4 0.7 0.9 1.0 1.2 Indonesia −1.6 −1.9 −2.2 −2.4 −2.6 −3.6
Ghana −1.9 −2.0 −2.2 −2.4 −2.7 −3.5 Ecuador −1.7 −1.9 −2.0 −2.2 −2.3 −2.4
Georgia −1.8 −1.8 −1.9 −2.0 −2.0 −2.7
Countries with trade-offs
Yemen, Rep. −1.8 −2.1 −2.4 −2.5 −2.6 −3.2
Mozambique 3.6 3.0 2.2 1.5 1.0 −2.5 3.5
Guatamala −1.9 −2.1 −2.2 −2.3 −2.4 −3.5
Nigeria 3.3 3.2 2.7 1.7 −0.2 −6.4 2.0
Guinea-Bissau −1.9 −2.5 −2.9 −3.0 −3.2 −5.2
Kenya 2.9 2.8 2.4 1.9 1.1 −6.1 2.6
Azerbaijan −2.0 −2.2 −2.3 −2.5 −2.6 −3.6
Benin 2.3 2.7 2.8 2.6 2.3 −1.1 5.0
Armenia −2.3 −2.3 −2.4 −2.4 −2.4 −3.2
Ethiopia 2.2 2.1 2.0 1.9 1.7 −0.0 9.9
South Africa −2.5 −3.4 −4.6 −5.5 −6.1 −6.8
Togo 2.1 2.3 2.1 1.5 0.8 −1.3 2.6
Zambia −3.0 −3.0 −3.0 −3.0 −3.1 −3.7
Gambia, The 1.9 2.1 1.8 1.1 0.3 −1.3 2.2
Tajikistan −3.0 −3.1 −3.1 −3.2 −3.2 −3.1
Papua New Guinea 1.7 2.1 2.2 2.3 2.2 −8.0 4.8
G(0) G(0.5) G(1) G(1.5) G(2) G(10) ϵ
Liberia 1.6 1.5 1.3 1.1 0.9 −0.4 4.5
Vietnam 1.1 1.2 1.2 1.0 0.8 −1.5 3.8 Countries with trade-offs
Burkina Faso 0.7 0.5 0.1 −0.1 −0.4 −2.9 1.28 Mauritania 1.0 0.6 0.1 −0.2 −0.5 −2.6 1.2
Bangladesh 0.5 0.7 0.6 0.4 0.2 −1.8 2.4 Sri Lanka 0.9 0.3 −0.2 −0.6 −1.0 −3.0 0.8
Burundi 0.4 −0.3 −0.7 −0.8 −0.7 −2.6 0.3 Mali 0.8 0.4 −0.1 −0.7 −1.2 −2.6 0.9
Rwanda −0.1 0.3 0.8 1.1 1.2 −3.4 0.2 Moldova 0.4 0.4 0.4 0.4 0.4 −0.4 6.4
Comoros −0.2 −0.0 0.1 0.1 −0.0 −1.2 0.5 Ghana 0.4 0.4 0.3 0.3 0.3 −0.4 4.2
Sri Lanka −0.3 0.8 1.4 1.8 2.0 0.8 0.1 Mongolia 0.3 −0.0 −0.3 −0.6 −0.8 −1.5 0.5
Mali −0.3 0.8 1.8 2.7 3.5 4.8 0.2 Madagascar 0.0 0.1 0.1 0.2 0.2 −1.2 4.6
Bhutan −0.7 1.6 2.8 3.1 2.8 −2.5 0.1 Papua New Guinea −0.1 −0.1 −0.1 −0.0 0.1 0.7 1.6
Mauritania −1.3 0.2 1.4 2.1 2.7 5.4 0.5 Burkina Faso −0.6 −0.3 −0.1 0.0 0.1 −0.1 1.3
Central Afr. Rep. −0.8 −0.6 −0.4 −0.2 −0.0 0.1 2.3
G(0) G(0.5) G(1) G(1.5) G(2) G(10) ϵ Togo −0.9 −0.7 −0.4 −0.2 0.1 0.1 1.8
Benin −1.1 −0.7 −0.3 0.1 0.5 1.6 1.4
Countries with trade-offs (continued)
Kenya −1.3 −1.4 −1.3 −1.1 −0.8 2.1 2.8
Cameroon 6.9 6.7 6.3 5.9 5.6 2.7
Comoros −1.4 −1.2 −1.0 −0.7 −0.4 0.3 2.9
Sierra Leone 4.3 4.5 4.3 3.9 3.6 2.1
Gambia, The −1.5 −1.6 −1.4 −1.1 −0.7 0.4 3.1
Tanzania 4.2 4.6 4.7 4.6 4.5 2.9
Bhutan 1.2 1.2 1.2 1.2 1.2 1.3
Uzbekistan 3.5 3.2 2.9 2.7 2.5 1.2
Cambodia 1.1 1.2 1.2 1.3 1.3 0.8
Côte d'Ivoire 3.4 3.4 3.3 3.0 2.7 1.2
Uganda −0.6 −0.7 −0.8 −0.9 −1.1 −3.8
Ukraine 3.2 3.2 3.1 3.1 3.1 2.5
Uzbekistan −0.9 −0.8 −0.7 −0.7 −0.7 −2.1
Ecuador 3.0 3.6 4.0 4.1 4.1 1.8
Ethiopia −1.3 −1.0 −0.8 −0.6 −0.5 −0.2
Egypt, Arab Rep. 2.9 3.0 3.0 2.9 2.8 2.5
Guinea −1.5 −1.3 −1.2 −1.1 −1.0 −1.7
Bolivia 2.8 3.4 3.6 3.5 3.3 1.5
Côte d'Ivoire −1.5 −1.6 −1.5 −1.4 −1.3 −2.5
Azerbaijan 2.5 2.6 2.6 2.7 2.7 2.1
Malawi −1.8 −1.6 −1.4 −1.3 −1.3 −0.6
Armenia 2.5 2.4 2.3 2.2 2.1 1.5
Egypt, Arab Rep. −1.9 −2.0 −2.0 −1.9 −1.9 −2.4
South Africa 2.5 2.6 2.6 2.5 2.3 1.9
Mozambique −2.2 −1.7 −1.1 −0.7 −0.4 −0.4
Malawi 2.4 2.3 2.1 2.0 1.8 1.1
Niger −2.3 −2.2 −2.0 −1.8 −1.6 −2.3
Nicaragua 2.0 2.6 3.0 3.1 3.1 1.5
Bolivia −2.4 −2.4 −2.4 −2.2 −2.0 −1.2
Tajikistan 1.9 1.9 1.9 1.8 1.7 1.5
Ukraine −2.5 −2.5 −2.5 −2.5 −2.5 −2.3
Niger 1.9 1.7 1.4 1.1 0.9 1.3
Pakistan −2.8 −2.7 −2.9 −3.2 −3.5 −4.3
Indonesia 1.9 2.0 2.1 2.2 2.2 1.5
Nigeria −3.0 −2.9 −2.6 −2.0 −1.2 −4.0
Guatamala 1.9 1.9 1.8 1.6 1.5 0.2
Cameroon −3.1 −3.1 −3.0 −2.9 −2.9 −3.0
Georgia 1.0 1.0 1.0 0.9 0.8 0.2
Iraq −3.2 −3.1 −3.0 −3.0 −3.0 −3.2
Kyrgyz Republic 0.6 0.6 0.6 0.6 0.6 0.7
Sierra Leone −3.4 −3.0 −2.8 −2.6 −2.4 −1.1
Madagascar −1.0 −0.7 −0.5 −0.6 −0.7 −1.7
Liberia −3.7 −3.6 −3.5 −3.3 −3.2 −2.7
Cambodia −3.0 −2.3 −2.0 −1.9 −1.9 −2.3
Average −1.3 −1.3 −1.3 −1.3 −1.3 −1.7 2.4
Average 1.9 2.2 2.3 2.2 2.1 0.7 2.4
Population weighted average −1.6 −1.6 −1.7 −1.7 −1.7 −2.4 2.5
Population weighted average 2.1 2.3 2.3 2.2 1.9 0.1 3.4
GDP weighted average −1.8 −2.0 −2.2 −2.3 −2.3 −3.2 2.2
GDP weighted average 2.2 2.4 2.4 2.3 2.0 0.4 2.6
Notes: this table shows the gains from trade G(0), the inequality adjusted gains from
Notes: this table shows the gains from trade G(0), the inequality adjusted gains from
trade, G(0.5), G(1), G(1.5), G(2) and G(10), and trade-ϵ when all tariffs are increased by
trade, G(0.5), G(1), G(1.5), G(2) and G(10), and trade-ϵ when the loss in tariff revenue in-
10% in absolute terms terms.
curred when countries liberalize is compensated for by an increase in income taxes that
respects the progressivity of the existing labor income tax system.
44 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45

Table F2 Table F3
Inequality adjusted gains from trade and trade ϵ, 10% relative increase in tariffs. Inequality adjusted gains from trade and trade ϵ, tariffs increase to 62.4%.

G(0) G(0.5) G(1) G(1.5) G(2) G(10) ϵ G(0) G(0.5) G(1) G(1.5) G(2) G(10) ϵ

Countries without trade-offs Countries without trade-offs

Madagascar 0.1 0.1 0.1 0.1 0.1 0.1 Bhutan 8.0 8.2 8.4 8.5 8.7 8.5
Comoros 0.0 0.0 0.0 0.0 0.1 0.2 Jordan −0.2 −1.3 −2.4 −3.4 −4.3 −11.6
Mongolia −0.0 −0.0 −0.0 −0.0 −0.1 −0.1 Tanzania −0.2 −0.7 −1.1 −1.3 −1.6 −6.3
Sri Lanka −0.0 −0.1 −0.1 −0.2 −0.2 −0.3 Madagascar −0.5 −1.3 −2.3 −3.3 −4.5 −26.9
Rwanda −0.1 −0.1 −0.1 −0.2 −0.3 −0.5 Rwanda −0.8 −2.1 −3.9 −5.7 −7.4 −13.1
Bhutan −0.1 −0.1 −0.2 −0.2 −0.2 −0.1 Uganda −1.2 −1.8 −2.8 −4.1 −5.6 −22.5
Nepal −0.2 −0.2 −0.2 −0.2 −0.2 −0.3 Uzbekistan −2.1 −2.4 −2.8 −3.2 −3.7 −27.8
Indonesia −0.2 −0.2 −0.2 −0.2 −0.3 −0.2 Nicaragua −2.5 −3.5 −4.6 −5.5 −6.5 −15.1
Guinea-Bissau −0.2 −0.3 −0.3 −0.3 −0.4 −0.6 Kyrgyz Republic −3.7 −4.8 −5.8 −6.7 −7.5 −15.9
Nicaragua −0.2 −0.3 −0.3 −0.3 −0.3 −0.4 Côte d'Ivoire −5.9 −6.9 −7.5 −7.9 −8.3 −24.3
Azerbaijan −0.2 −0.2 −0.3 −0.3 −0.3 −0.3 Nepal −5.9 −7.2 −8.3 −9.2 −10.0 −18.4
Guatamala −0.2 −0.3 −0.3 −0.3 −0.3 −0.3 Bangladesh −6.5 −7.9 −9.4 −10.8 −12.2 −27.6
Yemen, Rep. −0.3 −0.3 −0.3 −0.3 −0.4 −0.4 Ecuador −7.0 −8.1 −9.2 −10.1 −10.9 −15.8
Ecuador −0.4 −0.4 −0.4 −0.5 −0.5 −0.5 Indonesia −8.1 −10.5 −12.7 −14.6 −16.3 −30.8
Jordan −0.4 −0.4 −0.5 −0.5 −0.5 −0.6 Yemen, Rep. −8.4 −10.6 −12.4 −13.7 −14.8 −24.0
Central Afr. Rep. −0.4 −0.5 −0.5 −0.6 −0.6 −0.7 Egypt, Arab Rep. −9.3 −9.7 −10.0 −10.3 −10.5 −17.2
Zambia −0.6 −0.6 −0.6 −0.6 −0.6 −0.7 Guatamala −9.3 −10.8 −12.1 −13.2 −14.1 −25.8
Guinea-Bissau −9.8 −13.4 −15.6 −17.0 −18.3 −30.5
Countries with trade-offs Georgia −9.9 −11.0 −12.1 −13.1 −14.1 −23.8
Mauritania 0.1 0.1 0.0 −0.0 −0.0 −0.3 1.4 Azerbaijan −10.3 −11.8 −13.4 −14.9 −16.4 −27.9
Mali 0.0 −0.0 −0.1 −0.1 −0.2 −0.4 0.4 Armenia −11.8 −12.7 −13.5 −14.3 −15.2 −24.0
Burundi −0.1 0.1 0.2 0.2 0.2 0.2 0.2 Ukraine −12.1 −12.6 −13.1 −13.4 −13.7 −16.6
Bangladesh −0.1 −0.1 −0.0 −0.0 −0.0 0.0 2.6 Zambia −12.3 −12.8 −13.3 −13.8 −14.3 −21.7
Burkina Faso −0.1 −0.1 −0.0 −0.0 0.0 0.1 1.7 South Africa −12.8 −19.6 −28.0 −34.0 −37.6 −44.4
Vietnam −0.1 −0.1 −0.1 −0.1 −0.1 0.0 5.3 Tajikistan −16.5 −17.4 −18.0 −18.4 −18.7 −21.4
Papua New Guinea −0.2 −0.2 −0.2 −0.2 −0.2 0.3 6.7
Ethiopia −0.2 −0.2 −0.1 −0.1 −0.1 0.1 5.1 Countries with trade-offs
Benin −0.2 −0.2 −0.1 −0.0 0.0 0.3 1.9 Sri Lanka 5.4 1.9 −1.7 −4.7 −7.4 −23.6 0.8
Gambia, The −0.2 −0.2 −0.1 −0.1 −0.0 0.1 2.2 Mauritania 5.2 2.2 −0.5 −2.6 −4.4 −14.7 0.9
Togo −0.3 −0.3 −0.2 −0.2 −0.1 −0.0 5.3 Mali 4.7 2.1 −1.4 −5.0 −8.7 −27.3 0.8
Mozambique −0.4 −0.3 −0.2 −0.2 −0.1 0.1 6.8 Cambodia 4.0 4.1 4.2 4.1 4.0 −2.3 7.1
Kenya −0.4 −0.4 −0.4 −0.3 −0.3 0.3 3.5 Moldova 3.3 3.1 2.8 2.5 2.1 −6.2 4.3
Nigeria −0.4 −0.4 −0.4 −0.3 −0.1 0.4 2.3 Ghana 2.1 1.5 1.0 0.5 −0.2 −18.4 1.9
Mongolia 1.8 −1.3 −4.1 −6.4 −8.4 −15.0 0.30
G(0) G(0.5) G(1) G(1.5) G(2) G(10) ϵ Burundi 1.6 3.6 4.8 5.2 5.2 −11.1 5.5
Papua New Guinea 1.1 0.6 0.6 0.6 0.6 −10.7 2.7
Countries with trade-offs
Central Afr. Rep. −0.7 0.2 1.5 2.5 3.0 −3.3 0.4
Cambodia 0.3 0.3 0.3 0.3 0.3 0.2
Togo −2.9 −2.5 −1.7 −0.9 −0.1 −6.0 2.1
Ghana 0.0 0.0 0.0 0.0 0.0 0.0
Kenya −4.5 −5.1 −5.1 −4.5 −3.4 4.2 3.0
Kyrgyz Republic −0.1 −0.1 −0.1 −0.1 −0.1 −0.1
Benin −4.7 −3.5 −2.2 −1.0 0.1 −4.8 2.0
Moldova −0.1 −0.1 −0.1 −0.1 −0.1 −0.1
Georgia −0.1 −0.1 −0.1 −0.1 −0.1 −0.1
G(0) G G(1) G G(2) G(10) ϵ
Iraq −0.2 −0.2 −0.1 −0.1 −0.1 −0.1
(0.5) (1.5)
Liberia −0.2 −0.2 −0.1 −0.1 −0.1 −0.1
Niger −0.2 −0.2 −0.1 −0.1 −0.1 −0.3 Countries with trade-offs
Tajikistan −0.2 −0.2 −0.2 −0.2 −0.2 −0.2 Vietnam 6.8 6.9 7.1 7.4 7.7 4.6
Armenia −0.3 −0.2 −0.2 −0.2 −0.2 −0.2 Burkina Faso −2.4 −1.8 −1.5 −1.6 −2.0 −14.4
South Africa −0.3 −0.2 −0.2 −0.2 −0.2 −0.2 Ethiopia −5.7 −5.1 −4.6 −4.4 −4.4 −14.8
Malawi −0.3 −0.2 −0.2 −0.1 −0.1 −0.1 Guinea −6.3 −6.0 −5.6 −5.4 −5.3 −15.0
Pakistan −0.3 −0.3 −0.3 −0.4 −0.4 −0.4 Gambia, The −7.0 −8.1 −8.2 −7.5 −6.2 −0.5
Uganda −0.3 −0.3 −0.3 −0.3 −0.3 −0.7 Malawi −8.8 −8.5 −8.4 −8.6 −8.9 −17.7
Guinea −0.3 −0.3 −0.2 −0.2 −0.2 −0.3 Comoros −9.0 −9.2 −9.1 −8.9 −8.8 −29.3
Bolivia −0.3 −0.3 −0.3 −0.3 −0.2 −0.1 Mozambique −10.0 −8.2 −6.2 −4.8 −4.1 −20.2
Egypt, Arap Rep. −0.3 −0.3 −0.3 −0.3 −0.3 −0.3 Cameroon −11.9 −11.8 −11.7 −11.6 −11.6 −15.4
Uzbekistan −0.3 −0.3 −0.3 −0.3 −0.2 −0.2 Bolivia −12.4 −13.0 −13.5 −13.6 −13.4 −23.8
Ukraine −0.3 −0.3 −0.3 −0.3 −0.3 −0.3 Niger −12.7 −12.6 −12.1 −11.6 −11.1 −20.7
Côte d'Ivoire −0.4 −0.3 −0.3 −0.3 −0.3 −0.3 Nigeria −14.1 −14.4 −14.0 −12.6 −10.7 −43.6
Sierra Leone −0.4 −0.4 −0.3 −0.3 −0.2 −0.2 Pakistan −15.0 −15.0 −16.4 −18.0 −19.6 −27.0
Tanzania −0.5 −0.5 −0.5 −0.5 −0.5 −0.7 Sierra Leone −16.7 −15.7 −15.4 −15.4 −15.6 −17.2
Cameroon −0.8 −0.7 −0.7 −0.7 −0.7 −0.7 Iraq −18.5 −18.3 −18.4 −18.6 −19.0 −23.9
Average −0.2 −0.2 −0.2 −0.2 −0.2 −0.2 3.2 Liberia −21.2 −21.4 −21.4 −21.4 −21.3 −26.1
Population weighted average −0.2 −0.2 −0.2 −0.2 −0.2 −0.2 3.4 Average −5.7 −6.4 −7.2 −7.8 −8.3 −17.9 2.4
GDP weighted average −0.3 −0.3 −0.3 −0.2 −0.2 −0.2 3.1 Population weighted −7.5 −8.4 −9.4 −10.2 −10.8 −23.1 2.6
average
Notes: this table shows the gains from trade G(0), the inequality adjusted gains from
GDP weighted average −9.0 −10.7 −12.5 −13.8 −14.8 −27.3 2.1
trade, G(0.5), G(1), G(1.5), G(2) and G(10), and trade-ϵ when all tariffs are increased by
10% in relative terms (i.e. then tariffs are pre-multiplied by 1.1). Notes: this table shows the gains from trade G(0), the inequality adjusted gains from
trade, G(0.5), G(1), G(1.5), G(2) and G(10), and trade-ϵ, when all tariffs are increased to
62.4%, with tariffs already in excess of 62.4% left unaltered.
 E. Artuc et al. / Journal of International Economics 120 (2019) 1–45 45

References Costinot, A., Donaldson, D., Komunjer, I., 2012. What goods do countries trade? A quanti-
tative exploration of Ricardo's ideas. Rev. Econ. Stud. 79 (2), 581–608.
Anderson, J., Neary, P., 1996. A new approach to evaluating trade policy. Rev. Econ. Stud. Deaton, A., 1989. Rice prices and income distribution in Thailand: a non-parametric anal-
63 (1), 107–125. ysis. Econ. J. 99, 1–37 Supplement.
Andrew Young School of Policy Studies, 2010. Andrew Young School World Tax Indica- Deaton, A., 1997. The Analysis of Household Surveys - A Microeconometric Approach to
tors. vol. 1. Development Policy. Johns Hopkins Press, Baltimore.
Antras, P., de Gortari, A., Itskhoki, O., 2017. Globalization, inequality, and welfare. J. Int. Dix-Carneiro, R., Kovak, B., 2019. Trade Liberalization and Regional Dynamics. Am. Econ.
Econ. 108, 387–412. Rev. 107 (10), 2908–2946.
Arkolakis, C., Costinot, A., Rodriguez-Clare, A., 2012. New trade models, same old gains? Dixit, A., Norman, V., 1980. Theory of international trade: a dual. General Equilibrium Ap-
Am. Econ. Rev. 102 (1), 94–130. proach. Cambridge University Press.
Arkolakis, C., Costinot, A., Donaldson, D., Rodriguez-Clare, A., 2019. The elusive pro- Dixit, A., Norman, V., 1986. Gains from trade without lump-sum compensation. J. Int.
competitive effects of trade. mimeo 86 (1), 46–80. Econ. 21, 111–122.
Artuc, E., Chaudhuri, S., McLaren, J., 2010. Trade shocks and labor adjustment: a structural Faber, B., 2014. Trade Liberalization, the Price of Quality, and Inequality: Evidence from
empirical approach. Am. Econ. Rev. Econ. 100 (3), 1008–1045. Mexican Store Prices. mimeo. Department of Economics, University of California at
Artuc, E., Lederman, D., Porto, G., 2015. A mapping of labor mobility costs in the develop- Berkeley.
ing world. J. Int. Econ. 95 (1), 28–41. Fajgelbaum, P., Khandelwal, A., 2016. Measuring the unequal gains from trade. Q. J. Econ.
Atkin, D., Donaldson, D., 2015. Who's getting globalized? The size and implications of 131, 1113–1180.
Intranational trade costs. mimeo https://uce2adbaa412557f3300255d7ec8.dl. Galle, S., Rodriguez-Clare, A., Yi, M., 2017. Slicing the Pie: Quantifying the Aggregate and
dropboxusercontent.com/cd/0/inline2/Ah3lkrqhraiEM88TuiFngs50f9LDeQNQRzL Distributional Effects of Trade. mimeo Berkeley University.
GF6NpqjyDlYrrAP9_qpCg4bTKoccdR56xDztV9z1VemuGmoDvsbcJUNjRaJnRl1qci Goldberg, P., Knetter, M., 1997. Goods prices and exchange rates: what have we learned?
YBLljGa1DSkuthZ0d5sdwF1FZoYZbSk9fG4l8EgsIlxXoatYqdkWbZbS3royHBYr7tKEls- J. Econ. Lit. vol. XXXV, 1243–1272.
qv2iHd9PLKFrvVaU_ETjucYi9qbSY1n4QnskHAU_NgEqs4NLQYmfayIpg31g–u6K91 Grossman, G., Helpman, E., 1994. Protection for sale. Am. Econ. Rev. 84 (4), 833–850.
W1V-KsqfYOYcPuM01we9JvsFNPVTIXtNFuOkpYAOgaPnye6h5YpTls6qwq1NBFqN3_ Heathcote, J., Storesletten, K., Violante, G., 2017. Optimal tax progressivity: an analytical
ymORIB9g-Soqr-DVMTlA_uqntqO-7C0lvOYrs5bi/file#. framework. Q. J. Econ. 132 (4), 1693–1754.
Atkin, D., Faber, B., Gonzalez-Navarro, M., 2018. Retail globalization and household wel- Kovak, B., 2013. Regional efects of trade reform: what is the correct measure of liberaliza-
fare: evidence from Mexico. J. Polit. Econ. 126 (1), 1–73. tion? Am. Econ. Rev. 103 (5), 1960–1976.
Atkinson, A., 1970. On the measurement of inequality. J. Econ. Theory 2 (3), 244–263. Krueger, A., 1974. The political economy of the rent-seeking society. Am. Econ. Rev. 64
Autor, D., Dorn, D., Hanson, G., 2013. The China syndrome: local labor market effects of (3), 291–303.
import competition in the United States. Am. Econ. Rev. 103 (6), 2121–2168. Layard, R., Mayraz, G., Nickell, S., 2008. The marginal utility of income. J. Public Econ. 92
Benabou, R., Tirole, J., 2006. Belief in a just world and redistributive politics. Q. J. Econ. 121 (8), 1846–1857.
(2), 699–746. Melitz, M., Redding, S., 2015. New trade models, new welfare implications. Am. Econ. Rev.
Benjamin, D., Deaton, A., 1993. Household welfare and the pricing of cocoa and coffee in 105 (3), 1105–1146.
Côte d'Ivoire: lessons from the living standards surveys. World Bank Econ. Rev. 7, Nicita, A., 2009. The Price effect of trade liberalization: measuring the impacts on house-
293–318. hold welfare. J. Dev. Econ. 89 (1), 19–27.
Besley, T., Persson, T., 2013. Taxation and development. Chapter 2. In: Auerbach, Alan J., Nicita, A., Olarreaga, M., Porto, G., 2014. Pro-poor trade policy in sub-Saharan Africa. J. Int.
Chetty, Raj, Feldstein, Martin, Saez, Emmanuel (Eds.), Handbook of Public Economics. Econ. 92 (2), 252–265.
Vol. 5, pp. 55–110. Piketty, T., 1995. Social mobility and redistributive politics. Q. J. Econ. 110, 551–583.
Caliendo, L., Parro, F., 2015. Estimates of the trade and welfare effects of NAFTA. Rev. Econ. Porto, G., 2005. Informal export barriers and poverty. J. Int. Econ. 66, 447–470.
Stud. 82 (1), 1–44. Porto, G., 2006. Using survey data to assess the distributional effects of trade policy. J. Int.
Caliendo, L., Dvorkin, M., Parro, F., 2019. Trade and Labor Market Dynamics: General Equi- Econ. 70, 140–160.
librium Analysis of the China Trade Shock. Econometrica 87 (3), 741–835. Singh, I., Squire, L., Strauss, J., 1986. A Survey of Agricultural Household Models: Recent
Caliendo, L., Dvorkin, M., Parro, F., 2015. Trade and Labor Market Dynamics (NBER Work- Findings and Policy Implications. World Bank Econ. Rev. 1 (1), 149–179.
ing Paper No. 21149). Topalova, P., 2010. Factor immobility and regional impacts of trade liberalization: evi-
Carlsson, F., Daruvala, D., Johansson-Stenman, O., 2005. Are people inequality-averse or dence on poverty from India. Am. Econ. J. Appl. Econ. 2 (4), 1–41.
just risk-averse. Economica 72 (287), 375–396. Ural Marchand, B., 2012. Tariff pass-through and the distributional effects of trade liber-
Costinot, A., Rodriguez-Clare, A., 2014. Trade theory with numbers: quantifying the con- alization. J. Dev. Econ. 99 (2), 265–281.
sequences of globalization. In: Gopinath, G., Helpman, E., Rogoff, K. (Eds.), Handbook
of International Economics, pp. 197–262 Chapter 4.