Trading-off the Income Gains and theInequality Costs of Trade Policy

Erhan Artuc (World Bank, Development Research Group)Guido G. Porto (Universidad Nacional de La Plata)

Bob Rijkers (World Bank, Development Research Group)

IMF-WB-WTO Trade ConferenceJune 25, 2019

Motivation: the Trade-Off

Should free trade be opposed on inequality grounds?

This paperStudy aggregate gains from trade as well as theirdistribution in 54 low- and middle-income countries

Estimated using household survey data with verydisaggregated information on expenditures and incomes (FOapproximation)

Use an Atkinson index to explore the trade-off betweenincome gains and inequality costs

Main Findings - Preview

We find evidence of gains from trade in 45 out of the 54countries we study

Average gains are 1.9% of real income

Income gains are negatively correlated with equalitygains

Inequality costs arise because consumption gains are spreadbut income losses tend to be concentrated

Yet, trade-off is typically resolved in favor of free trade

Building Blocks

A Modeldescribes how trade policy affects real income at householdlevel

Datatrade data (tariffs)household survey data (household expenditures and incomes)

A shock: full tariff liberalizationuse the model, with the tariff and the household survey data,to measure the change in the real income of each household−→ pre- and post-liberalization income distribution

A Social Welfare Function a-la-Atkinson (Atkinson,1970)

it is a tool to assess the trade-off between equality and meanincome, or growth, across households (Deaton, 1997)

The model

Household welfare metric, V h, real expenditure:

V h = yh(p, vh)− e(p, uh)

yh(p, vh): income e.g.

labor income (wages)enterprise profitsnet transfers

e(p, uh) consumption

First Order Effects

Change in price of good i impacts welfare:

dV hi

eh= −

((φh

i − shi ) + φhwi

) τi1 + τi

+ Ψhi

whereI si budget share of good iI φi income share of sales of good iI φwi share of labor incomeI Ψh

i tariff revenue transfers

Summing up dV h

eh=∑

idV h

i

ehenables comparison of ex-ante

and ex-post distributions of real per capita income

Data: Expenditure Template

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 1112. Wheat 1113. Rice 1114. Other Cereals

1121. Beans 1122. Other

1131. Banana 1132. Grapes 1133. Citrus 1134. Apples 1135. Other Fruits

1141. Tomato 1142. Potato 1143. Greens 1144. Other

Vegetables

1151. Vegetable Oils 1152. Animal Fats 1153. Other oils/fats

1161. Fish 1162. Shrimp 1163. Other Crustacean

1171. Pork (Pig) 1172. Beef (Cattle) 1173. Poultry (Chicken) 1174. Other meat/animals

1181. Milk 1182. Eggs 1183. Cheese 1184. Other Dairy

1191. Other staple food 1192. Other processed food

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 1212. Beer 1213. Other alcohol

1221. Cigarettes 1222. Other tobacco

1231. Soya 1232. Other oil seeds

1241. Cloves 1242. Pepper 1243. Vanilla 1244. Saffron 1245. Qat (chat) 1246. Other spices

1251. Coffee 1252. Tea 1253. Cocoa

1261. Cashew 1262. Coconut 1263. Other nuts

127. Cotton 1281. Sugar (any kind) 1282. Other non-staple

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

Average Tariffs

Bhutan

CameroonBhutan

Burundi

Bhutan

010

2030

4050

Non-staple Agriculture ManufacturesStaple Agriculture

Country Non-Staple Agriculture Manufactures Staple Agriculture

Average tariffs 14.4 10.8 10.9

Budget Shares: shi

0.1

.2.3

.4.5

expe

nditu

re s

hare

2 4 6 8 10 12log per capita expenditure

Agriculture and Food ManufacturingServices OtherHome consumption

Staple Ag. Autocons Non-Tradeables Manuf Non-staple Ag. Other

Average budget share 37.0 16.6 17.4 15.1 8.0 5.8

Income Shares: φhi

0.2

.4.6

.8In

com

e sh

are

2 4 6 8 10 12log per capita expenditure

Agricultural sales WagesNon-farm enterprise Other/transfersHome production

Wages Autocons Other Family Enterprise Staple Agric. Non-staple Agric.

Average income share 29.0 22.8 18.6 12.5 12.9 4.2

Gains From TradeG(0)

(6.0,6.8](5.0,6.0](4.0,5.0](3.0,4.0](2.0,3.0](1.0,2.0](0.0,1.0](-1.0,0.0](-2.0,-1.0](-3.0,-2.0][-3.1,-3.0]No data

The Distributional Effects: Pro-PoorJordan

-10

010

20w

elfa

re e

ffect

s

3 4 5 6 7log per capita expenditure

The poor gain, the rich too, but to a lesser extent: inequality ↓

17 countries in which the bottom 20% gain more than the top 20%

The Distributional Effects: Pro-RichUzbekistan

-10

-50

510

15w

elfa

re e

ffect

s

8 9 10 11 12log per capita expenditure

The poor gain, but the rich gain proportionately more: inequality ↑

37 countries in which the top 20% gain more than the bottom 20%

Atkinson Social Welfare Index

W =1

H

∑h

(xh)1−ε

1− ε, ε 6= 1

ε measures inequality aversionwhen ε=0, every household counts the same and social welfareis just the sum (average) of per capita income/expendituresas ε increases, the weight towards the poorer householdsincreases

Can be decomposed into mean income µ and“inequality” (1− I)

W = µ ∗ (1− I)

I = 1−

(1

H

H∑h=1

(xh/µ)1−ε

)1/(1−ε)

Inequality Adjusted Gains from Trade

Define G(ε) measure of the gains from trade, including acorrection for the inequality costs (Antras, de Gortari,Itskhoki, 2016)

G(ε) =W1(ε)−W0(ε)

W0(ε)

This can be written as

G(ε) = G(0) +µ1µ0

I0(ε)− I1(ε)1− I0(ε)

The Inequality Adjusted From TradeG(0)

(6.0,6.8](5.0,6.0](4.0,5.0](3.0,4.0](2.0,3.0](1.0,2.0](0.0,1.0](-1.0,0.0](-2.0,-1.0](-3.0,-2.0][-3.1,-3.0]No data

The Inequality Adjusted From TradeG(0.5)

(6,7](5,6](4,5](3,4](2,3](1,2](0,1](-1,0](-2,-1](-3,-2][-4,-3]No data

The Inequality Adjusted Gains From TradeG(1)

(6,7](5,6](4,5](3,4](2,3](1,2](0,1](-1,0](-2,-1](-3,-2][-3.432549,-3]No data

The Inequality Adjusted Gains From TradeG(10)

(6,6.293945](5,6](4,5](3,4](2,3](1,2](0,1](-1,0](-2,-1](-3,-2](-4,-3](-5,-4](-6,-5][-6.90912,-6]No data

No Trade-Off CountriesIncome Gains & Equality Gains

Central African RepublicG(ε) ↑ as ε ↑

01

23

45

67

8

01

23

45

67

8in

equa

lity

adju

sted

gai

ns, G

(ε)

0 2 4 6 8 10inequality aversion, ε

Similar countries: Guinea Bissau, Jordan, Mongolia, Yemen

Liberalization is preferable ∀ ε

Trade-Offs Without Preference Reversals

Uzbekistan

01

23

4

01

23

4in

equa

lity

adju

sted

gai

ns, G

(ε)

0 2 4 6 8 10inequality aversion, ε

Trade-off: Income Gains & Equality Gains have oppositesigns

15 countries similar to Uzbekistan (e.g. G(ε) > 0)2 countries (Bhutan, Cambodia) have trade-offs but G(ε) < 0

Trade policy preference (liberalize/protect) does notvary with ε

Trade-Offs With Preference Reversals

Benin

-6-5

-4-3

-2-1

01

23

-6-5

-4-3

-2-1

01

23

ineq

ualit

y ad

just

ed g

ains

, G(ε

)

0 2 4 6 8 10inequality aversion, ε

Trade policy preference varies with ε

for low ε liberalization is preferredfor high ε protectionism is preferred

14 countries similar to Benin

Trade-Offs With Preference Reversals

Mali

-5-3

-11

35

-5-3

-11

35

ineq

ualit

y ad

just

ed g

ains

, G(ε

)

0 2 4 6 8 10inequality aversion, ε

Trade policy preference varies with ε

for low ε protectionism is preferredfor high ε liberalization is preferred

2 countries similar to Mali

Assessment G(1.5)

Central African Republic

Guinea Bissau

Indonesia

Jordan

Mongolia

NepalP.N. Guinea

YemenEcuador

Pakistan

Azerbaijan

Benin

Burkina FasoGambia

Bangladesh

Burundi

Togo

Mali

Sri Lanka

Cameroon

Cote d'IvoireEgypt

Guinea

Malawi

Nigeria

Sierra Leone

Tanzania

Zambia

Iraq

Ukraine

Bolivia

Guatemala

Ethiopia

Kenya Mozambique

Niger

South Africa

Uganda

Kyrgyz Rep.

Tajikistan

Vietnam

Uzbekistan

LiberiaGeorgia

Moldova

NicaraguaArmenia

Mauritania

Ghana

Bhutan

Cambodia

Comoros

Madagascar

Rwanda

-50

510

-50

510

G(1

.5)

-5 0 5 10G(0)

Concluding Remarks

Large effort to study the trade off between the income gainsand the inequality costs of trade in 54 countries using afirst-order approximation

Overwhelming evidence of gains from trade45 countries gain, 9 countries loseOn average, countries gain 1.9%

These real income gains are inversely correlated withequality gains

consumption gains are evenly spread whereas income lossestend to be concentrated

Yet, the trade-off typically resolves in favor ofliberalization

Implications

The majority of countries could liberalize trade andachieve income gains at “low” or no inequality cost

Yet, they protect. Why?because they are not maximizing the Atkinson We.g., rent-seeking, political contributions?need to raise government revenue?

This is very costly, in terms of Social Welfare