Trade, environment, and income inequality: An optimal taxation approach

• DOI: http://doi.org/10.1111/jpet.12288
• Date: 01 August 2018
• Published date: 01 August 2018

Received: 30 November2016 Accepted: 11 December 2017

DOI: 10.1111/jpet.12288

ARTICLE

Trade, environment, and income inequality: An

optimal taxation approach

Philippe Bontems1Estelle Gozlan2

1ToulouseSchool of Economics

2INRAand AgroParisTech

PhilippeBontems, Toulouse School of Economics,

INRA,IDEI, Université Toulouse Capitole, 31000

Toulouse,France(philippe.bontems@inra.fr).

EstelleGozlan, Economie Publique,

AgroParisTech,INRA,Université Paris-

Saclay,78850 Thiverval-Grignon, France

(estelle.gozlan@inra.fr).

Weare grateful to the editor and two anonymous

reviewers,whose comments greatly improved

thepaper. We also thank the seminar partici-

pantsat the University of Queensland, at Paris

EnvironmentalEconomics Seminar, and at INRA,

Rennes.

In a small open economy, how should a government pursuing both

environmental and redistributive objectives design domestic taxes

when redistribution is costly? And how does trade liberalization

affect the economy's levels of pollution and inequalities, when taxes

are optimally and endogenously adjusted? Using a general equi-

librium model under asymmetric information with two goods, two

factors (skilled and unskilled labor), and pollution, this paper char-

acterizes the optimal mixed tax system (nonlinear income tax and

linear commodity and production taxes/subsidies) with both pro-

duction and consumption externalities. While optimal income taxes

are not directly affected by environmental externalities, conditions

are derived under which under- or over-internalization of social

marginaldamage is optimal for redistributive considerations. Assum-

ing that redistribution operates in favor of the unskilled workers

and that the dirty sector is intensive in unskilled labor, simulations

suggest that trade liberalization involves a clear trade-off between

the reduction of inequalities and the control of pollution when the

source of externality is only production; this is not necessarily true

with a consumption externality.Finally, an increase in the willingness

to redistribute income toward the unskilled results paradoxically in

less pollution and more income inequalities.

1INTRODUCTION

The social and environmental consequences of trade liberalization have receivedc onsiderableattention both in civil

society and in the economic literature. Governments in developed countries regularly face antiglobalization move-

ments, due to concerns about job losses, growing inequalities, increased pressure on the environment, and national

sovereigntylosses.1Conversely, environmental protection is often seen as a threat to international competitiveness in

political debates.

1The success of protectionist parties in the 2014 elections for the European parliament, and activism against the US-EU negotiations on the Transatlantic

FreeTrade Area, are recent illustrations of such concerns. Recently, Donald Trump'selection has been seen by many analysts as the result of political claims

resurgencefrom losers of globalization.

Journal of Public Economic Theory.2018;20:557–581. wileyonlinelibrary.com/journal/jpet c

2018 Wiley Periodicals,Inc. 557

558 BONTEMS AND GOZLAN

It is sometimes argued that many negative outcomes imputed to trade liberalization are rather the consequence

of missing or badly designed environmental and redistributive policies. Among trade economists, there is wide agree-

ment that domestic distortions should be addressed through domestic policy intervention, and that trade restrictions

for domestic policy objectives can at best be second best (Bhagwati, 1994). In particular, when local environmen-

tal externalities are the only market failure, a move toward free trade with optimal environmental policy is always

welfare-improving, eventhough it may involve an increase in pollution: with endogenous policy, any increase/decrease

in pollution then reflects an optimal trade-off between pollution and income at the country level (Copeland & Taylor,

2003). However,these results are obtained within a representative–agent framework. Yet,the view that the efficiency

gains brought by trade liberalizationallow governments to leave everyone better off with well-targeted redistributive

policies has been objected in modern normative economic theory: because individual characteristics are not directly

observable by governments, individualized lump-sum transfersare not feasible in practice, which challenges the sepa-

ration of efficiency and redistributiveconsiderations (Guesnerie, 2001; Naito, 1996; Tuomala, 2016).

On the other hand, environmental policyitself c anhave important distributive effects across heterogeneous house-

holds, whether it be through the regressive effect of consumption taxes (if low-income individuals are affected more

heavily), or be it through changes in relative returns to factors resulting from production taxes.2In other words,

the internalization of both consumption and production externalities might interfere with any given redistribution

objective.

Keeping this in mind calls for a fresh look at the optimal tax system as a whole in an open economy (i.e., where the

pattern of specialization matters for environmental and inequality-related outcomes), when the government pursues

both environmental and redistributive objectives. In particular,one would like to know under which conditions reduc-

ing pollution and redistributing wealth are conflicting tasks if any. Also, one would liketo know how the government

should adapt its fiscal system including environmental taxes when facing exogenousshocks like increasing globaliza-

tion or changes in social preferences with respect to wealth redistribution.

To address these issues, we consider a Heckscher–Ohlin model of a small open economywith two sectors (clean

and polluting) and two factors (skilled and unskilled labor). Assuming that emissions may arise either from the produc-

tion or from the consumption of the polluting good, we characterize the optimal tax system when policy is constrained

only by the information available to the regulator.The government maximizes a weighted sum of skilled and unskilled

agents'utility,using nonlinear income taxes (because skills are not observable) and linear taxes/subsidies on consump-

tion and production of the polluting good (because purchases are anonymous). Income redistribution is socially costly

because of incentive compatibility. While optimal income taxesare not directly affected by environmental externali-

ties, we show that optimal distortion of the consumer and the producer prices in the polluting sector consist in a Pigo-

vian term and a redistributive term. It is then optimal to over- or underinternalize the social marginal environmental

damage whenever it helps decreasing the cost of wealth redistribution due to incentive compatibility. FromPirttilä

and Tuomala (1997), we know that, for consumption externalities, this depends on the way pollution, consumption,

and leisure interact within individual preferences. In the present paper,because the nonlinear technology with imper-

fect substitutability between skilled and unskilled labor types makes wages endogenous, it is also optimal to distort

producer prices for redistributiveconsiderations; if the sector making intensive use of unskilled labor is also character-

ized by production externalities,then the Pigovian and redistributive components of the production tax/subsidy are of

opposite signs (underinternalization).

We then explore the consequences on the level of pollution and inequality of exogenousshocks on this economy,

when taxes are optimally and endogenously adjusted. We show that (i) under openness to trade, the source of the

externality (consumption or production) matters for redistribution, while it is not the case in autarky; (ii) whether the

economyspecializes into the clean or the dirty sector, trade liberalization involvesa clear trade-off between the control

of income inequalities and the control of pollution when the source of externality is production; this is not necessarily

true with a consumption externality; (iii) an increasing willingness to redistribute income corresponds to a shrinking

2SeeFullerton (2011) for a complete review of the mechanisms at stake, including the possibility that different types of agents might face different damages

fromemissions (a possibility not considered in this paper).