World Development Vol. 51, pp.

119131, 2013
2013 Elsevier Ltd. All rights reserved.
0305-750X/$ - see front matter
www.elsevier.com/locate/worlddev
http://dx.doi.org/10.1016/j.worlddev.2013.05.013

Are Carbon Taxes Good for the Poor? A General Equilibrium

Analysis for Vietnam
IAN COXHEAD
University of Wisconsin-Madison, USA

ANAN WATTANAKULJARUS
National Institute of Development Administration, Bangkok, Thailand

and
CHAN V. NGUYEN *
National Economics University, Hanoi, Viet Nam
Summary. We evaluate eects of an environmental tax using a general equilibrium model linked to a household database. The burden
of the tax, applied mainly to energy, is passed forward by non-tradable industries and backward by tradable industries facing xed world
prices. The tax is thus equivalent to a real exchange rate appreciation, and since export industries are labor-intensive, reduces employ-
ment, and increases poverty, especially when labor supply is responsive to wages. The use of revenues to increase transfers to households
can oset poverty increases, but does not create jobs; thus the tax will likely conict with other development policy objectives.
2013 Elsevier Ltd. All rights reserved.

Key words carbon tax, environmental tax, poverty, labor market, general equilibrium, Vietnam

1. INTRODUCTION income and poverty. This in turn requires a discussion of some

In December 2012 the Government of Vietnam imple-
mented its rst law on environmental taxation. 1 The law man-
dated new taxes on coal, gasoline and other fossil fuels,
pesticides, and some other products. As such it bears a strong
similarity to carbon taxes being implemented or debated in
other countries. Such taxes have far-reaching economic eects,
so as in other countries, there is considerable uncertainty over
their economic implications. It is likely that reaching their
nominal objectivesreduced growth rates of emissions from
carbon-based fuelswill also require changes in the structure
of production, employment, wages, and prices, and that these
changes in turn will aect the levels and distribution of house-
hold incomes. In trade-dependent low-income economies like
Vietnam, therefore, the eort to reduce emissions raises addi-
tional concerns at points where it intersects with other primary
targets of development policy. Do energy taxes alter the com-
petitiveness of industries producing exports and import substi-
tutes? Do they alter prospects for employment growth? What
eects do they have on poverty? If adverse eects are su-
ciently large, policymakers may be compelled to choose
among development policy priorities.
In this paper we evaluate the likely economic impacts and
incidence of Vietnams environmental tax law. The scope of
the law ensures that its impacts will be felt throughout the econ-
omy. Accordingly, we adopt a general equilibrium approach,
taking account of known economic linkages among production
activities, employment, wages, household incomes, consumer
expenditures, trade, government revenues, and other important
macroeconomic variables. This methodology facilitates a rela-
tively complete assessment of the impacts and incidence of the
tax. We quantify the consequences of the tax for production,
employment, and wagesand consequently for household
119
important parameters about which relatively little is known.
Among these are elasticities of labor supply. The environmental
tax aects real wages through a number of channels, and the ef-
fect of real wage changes on labor supply determines aggregate
employment eects as well as contributing to distributional and
poverty outcomes. Our examination of the labor supply issue is
threaded throughout the paper.
Environmental policy issues in Vietnam have received lim-
ited prior attention from a general equilibrium perspective.
El Obeid, Van der Mensbrugghe, and Dessus (2002, chap. 9)
used a small general equilibrium model to examine the struc-
tural eects of trade policy reforms and environmental policy
initiatives, focusing on measuring changes in emissions of air
and water pollution and solid waste, but did not explore wel-
fare and distributional questions. In a concurrent study
Willenbockel (2011), using a smaller version of the database
that we do, explores structural impacts of the environmental
tax law. His study includes a helpful exploration of the long-
er-run issue of tax-induced substitution among energy sources
as well as careful predictions of environmental outcomes. We
complement this approach by focusing on welfare issues aris-
ing from the incidence of the tax as transmitted through factor
* The authors thank Le Dong Tam, Kim Tran Dung, and Kim Pho Chi
for research assistance, The Ford Foundation and the Graduate School of
the University of Wisconsin-Madison for nancial support, and confer-
ence/seminar participants at the University of Waikato, the National In-
stitute of Development Administration (Bangkok), the Midwest
International Economic Development Conference, Madison, WI, and
two anonymous reviewers for helpful comments on earlier drafts. Address
for correspondence: coxhead@wisc.edu. Final revision accepted: May 15,
2013.
120 WORLD DEVELOPMENT
markets and alternative scal mechanisms, and map results to
a complete household survey database. In doing so, as noted
above, we draw attention to incipient tradeos among devel-
opment policy goals.
The remainder of the paper proceeds as follows. In Section 2
we review relevant tax incidence concepts and theory. Section 3
introduces the model and its main datasets. We describe the
environmental tax simulation experiment in Section 4, and dis-
cuss the results in Section 5. In a nal section we draw some
conclusions and identify areas for future research.
2. TAX INCIDENCE IN AN OPEN ECONOMY
The economic incidence of a tax diers from its statutory
(direct) incidence. Those on whom the statutory burden falls
pass the cost of the tax on, through markets, to consumers
and the suppliers of labor and other factors (Atkinson &
Stiglitz, 1980). The extent to which tax burden is passed for-
ward (to consumers) or backward (to factor owners) depends
on behavioral and technological responses to the taxfor
example the elasticity of consumer demand for a product, or
the extent to which a less highly taxed input to production
may be substituted for a more highly taxed one. In general,
tax burden is redistributed according to relative magnitudes
of relevant elasticities of demand or supply. The less elastic is
demand by purchasers of a taxed item, the more likely it is that
the tax will be passed forward. In the event that demand is elas-
tic, the burden will instead be passed backward to factors. In
that case, the distribution of the burden among factors will de-
pend, in part, on relative values of their supply elasticities. The
values of such parameters are thus crucial to the economic inci-
dence of a tax. 2 Moreover, so long as agents are heterogeneous
in terms of the services that they supply to factor markets, the
conditions under which they supply them (e.g., reservation
wages), and/or the preferences they exhibit as consumers, then
the shifting of the tax burden has consequences for income
distribution in addition to its impacts on aggregate welfare.
These general observations come into sharp focus in small
open economies, where the dierences between domestic and
foreign markets become important to overall tax incidence.
Suppose (for simplicity) that the supply of an exportable good
produced by homogeneous domestic rms is inelastic, so de-
mand (D) is the main determinant of its price. Denoting for-
eign and home markets by f and h, respectively, we have
D = Df(pf) + Dh(ph), with oDi/opi < 0, i = f, h. Using the total
dierential to nd the change in total demand,
@Df @Dh
dD dpf dp :
@pf @ph h
If we multiply and divide the rst right-hand side term by Dfpf
and the second by Dhph then divide the whole by D, we have
b dD ef hf ^pf eh 1  hf ^
D ph ;
D the tax aects not only the structure of production and trade,
where a hat denotes the proportional change in a variable;
ei = (oDi/opi)/(pi/Di) is the absolute value of the demand elas-
ticity in the ith market, and hf is the foreign share in total de-
mand. The change in total demand is a weighted sum of price
changes in each market, where the weights are price elasticities
and market shares. More intuitively, if we suppose that the
good in question has close substitutes in both markets so that
price changes are correlated, we can write ^ pf ^ph ^
p, which
(after more rearranging) yields
^p 1 sources have inherently general equilibrium eects; stylized
:
b
D ef  e h hf eh
The change in the price of a good with respect to demand
growth will be equal to 1/eh when there are no foreign sales,
and 1/(ef  eh) when there are no domestic sales. In the small
country case foreign demand is elastic, and therefore ef > eh.
In industries whose output is sold primarily to the world mar-
ket, price is unresponsive to demand shocks and the tax bur-
den cannot be passed forward. In contrast, for industries
producing non-traded or little-traded goods and services, de-
mand is less elastic and prices are more responsive. These con-
ditions apply particularly to suppliers of domestic services
such as transportation, storage and trade, which are inputs
to all other productive activities and have few substitutes. If
the law of one price holds, then any tax aecting costs in these
intermediate services will be passed forward until it reaches the
border (that is, domestic industries competing directly with
foreign producers), at which point it will bounce back onto
factors of production. This is easily seen by recalling that un-
der constant returns to scale and competitive markets, price is
equal to unit cost. Given a vector of primary factors x with
prices w and an intermediate input z with tax-inclusive price
q(p) = q(1 + p), unit cost is p = w0 x + q(p)z. Choose units
of z such that q = 1. Then in proportional change form the
zero-prot condition is:
X
p
^ ^ i /z p
/i w ^;
i
P
in which the / are cost shares and i /i /z 1: When the
rate of the tax increases, p must rise, and/or some wi must fall
to maintain zero pure prots. If p is constrained from increas-
ing by limit prices set in world markets, the burden of adjust-
ment falls instead on factors. In its turn, the distribution of
this loss among factors will be strongly inuenced by their rel-
ative supply elasticities. We return to this point below.
If non-traded industries can pass the tax burden forward
while those competing at the border cannot, then in the mac-
roeconomy, a tax on a widely used input such as energy has
eects analogous to those of a real exchange rate appreciation.
It reduces the relative protability of producing tradable
goods and services, and so diminishes the countrys competi-
tiveness in world markets. This point is strikingly absent from
standard carbon tax models, which omit or minimize the role
of international trade (Fullerton & Heutel, 2007; Metcalf,
2008). The loss of tradable sector competitiveness opens (or
widens) a balance of trade decit, with a matching excess of
domestic aggregate expenditure over income. For given inter-
national capital ows, 3 the elimination of these decits re-
quires a combination of expenditure reductions and a fall in
the relative level of domestic prices so as to depreciate the real
exchange rate. Among tradable industries, higher costs and
lower prots cause tax burden to be passed back in the form
of lower factor prices, and ultimately to household incomes.
These are likely to be important components of the adjustment
to a new equilibrium. Accordingly, the shifting incidence of
but also factor prices and employment, and ultimately the dis-
tribution of household income and welfare.
(a) Environmental taxes
Environmental taxes are politically controversial, in part
due to doubts over their ecacy, and in part because there
is much uncertainty over their economic incidence. Econo-
mists have also struggled to identify incidence, whether in the-
oretical models or in quantitative analyses. Taxes on energy
models have been used to explore structural issues (Copeland
 ARE CARBON TAXES GOOD FOR THE POOR? A GENERAL EQUILIBRIUM ANALYSIS FOR VIETNAM
& Taylor, 2003), but for applications to welfare and policy the
use of applied general equilibrium (AGE) models is required.
Until very recently, AGE analyses of the incidence of envi-
ronmental taxes were focused on industrialized economy
cases. The majority of these studies nd that taxes on energy
are fully passed forward to consumers; as a result, they are
usually found to be regressive (Metcalf, 2008; OECD, 1994,
1995). Possible exceptions to this regressivity exist when the
tax is applied to natural resources, since the (short-run) inelas-
ticity of their supply causes part of the tax burden to be shifted
back, reducing rents that often accrue to more wealthy mem-
bers of society.
For developing countries, however, the nature of any trade-
o between environmental and eciency or equity objectives
remains unresolved. Most studies of carbon taxes in low-in-
come economies use partial equilibrium methods (Datta,
2010; Sterner, 2012; Ziramba, Kumo, & Akinbiade, 2009).
Therefore they do not fully capture the trade competitiveness
and employment linkages we have described. This is a large
source of potential bias in results (Parry, Sigman, Walls, &
Williams, 2005). Among general equilibrium studies, Coxhead
(2000) showed that in a developing-country setting, imposing a
pollution tax need not reduce real wages. The most pollution-
intensive industries are typically capital-intensive; a tax that
reduces their protability has general equilibrium eects
through the real exchange rate that cause labor demand in
cleaner and more labor-intensive industries to expand. The
key to this result is that polluting industries produce tradables,
and so are constrained by the elasticity of demand in world
market from passing costs forward in higher prices. Similarly,
Yusuf (2008) showed that a reduction in fuel subsidies need
not be regressive in Indonesia, depending on compensation
schemes implemented alongside the tax reform. In general,
however, when tax burden is passed backward rather than for-
ward, new sources of distributional concern arise, and these
are linked directly to the elasticity of factor supply. 4
3. METHODS AND DATA
(a) The model and database
To evaluate the eects of Vietnams environmental tax we
use an AGE model of the Vietnamese economy. Such models
represent the entire economy in numerical form. They com-
bine baseline data from national accounts and other sources
about the activities of rms, households, enterprises, and gov-
ernment with theory-based specications about market opera-
tion, factor supplies, trade balances, and other constraints,
and the assumed behavior of foreign partners in trade and
investment. They thus provide a consistent interface between
macro and micro phenomena, at least in the realm of the real
economy. 5
We seek to observe the eects of tax policy shocks on prices,
production and factor demand, and through these, on markets
for labor, land, and capital and thus on household incomes
and wellbeing. Because households are heterogeneous in terms
of asset ownership, income, and expenditure we can also mea-
sure eects on income distribution and poverty. The model is
based on the well-known IFPRI model (Lofgren, Harris, &
Robinson, 2002), modied to t Vietnamese data. Here, to
save space we merely summarize the models main features.
The model identies 63 sectors, each producing a single
commodity. There are three aggregate primary factors: land,
labor, and capital. Land is used only in agriculture and natural
resource sectors. Capital has two components: part of the
121
stock is mobile among sectors, while another part is specic
to each sector. Labor is divided into six categories by location
(urban/rural) and skill (low/medium/high). 6 These categories
are based on data in the 2007 Vietnam Social Accounting Ma-
trix (SAM), which are fully documented by Arndt, Garcia,
Pham, McCoy, Tarp, and Thurlow (2010). Labor demands
are derived in the usual way from prot-maximizing choices
made by a representative rm in each industry. The model
posits a nested factor demand structure, with composite factor
demand decisions at the top level and demands for each type
of labor determined at the next level; these are governed by
constant elasticity of substitution (CES) production technol-
ogy. Similarly, intermediate inputs used in each industry are
aggregated into a composite input using CES technology. In
the long run this structure allows for the possibility of interfuel
substitution when tax increases energy prices at dierent rates.
The possibility of substitution among fuel types is an impor-
tant feature for any study of the eects of a tax in the long
run. However, a suitable model for that analysis must allow
for capital investment and depreciation whereas our model is
focused on a shorter-run incidence analysis. Finally, the model
allows for two-way trade in each good, although the data con-
tain some pure non-traded goods such as domestic trade and
transportation, public administration, and personal services.
Despite two decades of rapid growth, Vietnam remains a
small player in global trade. 7 Our model reects this by
assuming xed world prices of Vietnams imports and highly
elastic global demand for its exports. This small open econ-
omy representation distinguishes the Vietnam model from
most environmental tax models for the US and other large
industrialized economies, in which domestic markets typically
dominate trade. In those models, all virtually all tax burden is
passed forward to as higher prices to domestic consumers
(Bovenberg & Goulder, 1996; Fullerton & Metcalf, 1997).
Vietnams labor force consists mainly of unskilled or low-
skilled workers. In the 2006, VHLSS 44% of workers had pri-
mary or lower education; 33% had completed lower secondary
school; 19% upper secondary school, and just 4.2% some form
of tertiary education. This labor force structure ensures that
the countrys comparative advantage lies, for now, rmly in
the agricultural and light manufacturing sectors. Slightly less
than half the Vietnamese labor force is employed in agricul-
ture, about one-fth in manufacturing, and the remainder in
services (World Bank, WDI Online).
The model database contains 20 representative household
types, distinguished by location (urban/rural), primary occu-
pation (farm/nonfarm), and quintile in the aggregate expendi-
ture distribution. Households earn income from their
ownership of factors, and may receive transfers from govern-
ment or enterprises as well as remittances from the rest of the
world. They consume goods and services, both those produced
domestically and those imported from abroad. Consumption
is assumed to take a Stone-Geary form, in which expenditure
is allocated rst to satisfaction of minimum subsistence quan-
tities of goods, and then to supernumary, or discretionary,
expenditures.
(b) Links to national household survey data
For the purpose of welfare analysis, we augment this repre-
sentative household structure by linking it recursively to the
individual records of the 2006 Vietnam Household Living
Standards Survey (VHLSS), which contains information on
a representative sample of 9,185 households nationwide. As
in the SAM, each household in the VHLSS dataset can be
identied by its location, primary occupation, and position
122 WORLD DEVELOPMENT
in the expenditure distribution. The results of a simulation
experiment in the AGE model are transmitted to individual
household eects by mapping predicted changes in prices
and supply quantities of factors, consumer prices, and trans-
fers from private and government sources onto information
about household endowments, consumption patterns, and un-
earned income. This yields household-specic predictions of
changes in income, expenditure, and price indices associated
with the ceteris paribus eects of a given policy shock applied
to the model.
This link, from macro model to micro data, makes it pos-
sible simultaneously to conduct two types of experiments. One
is macrosimulation, or experiments in which we examine the ef-
fects of a growth or policy shock on macroeconomic aggre-
gates such as GDP, CPI, wages, employment, and industry
activity levels. The other type is microsimulation, in which
we trace the eects of the same shock(s) to the incomes and
expenditures of individual households, or to regional and
other aggregates. The latter method enables us to draw conclu-
sions about the eects of the shock on household real incomes
and poverty, both nationally and for subsets of the population
such as urban and rural households and residents of dierent
administrative regions. 8
(c) Solution method
The model is implemented using GEMPACK software
(Harrison & Pearson, 1996) and as such relies heavily on code
from the well-known ORANI-G model (Horridge, 2005). The
model is implemented in percentage changes of variables rela-
tive to a no-tax counterfactual. Thus, for example, an increase
of 0.5% in the CPI means that imposing the tax raises the rate
of ination (that is, the growth rate of the CPI) by 0.5% during
the period encompassed by the simulation, i.e., about one
year.
4. ENVIRONMENTAL TAX POLICY EXPERIMENT
The Vietnamese legislation mandates specic taxes on the
volume sold of coal, gasoline and other fossil fuels, and pesti-
cides. These taxes are intended to replace fees currently levied
at the point of sale. 9 Using current market prices, we convert
these fees and specic taxes to their ad valorem equivalents.
Table 1 shows basic data on the two main products targeted
by the tax, coal, and rened fuels. As legislated, these taxes
span a range of values. The last column of the table shows
the sales tax equivalent of the median tax rate for each prod-
uct. These are the counterfactual rates we apply in our exper-
iments.
The environmental taxes are not strictly construed as taxes
on carbon; 10 however, they are equivalent to a tax of about
$US0.71/tonne of carbon dioxide equivalent (CO2e) on coal,
Table 1. Environmental tax rates
Domestic base price
(VND per ton or liter)
(1)
Coal (lignite) 450,000
Fuel 15,000
Source of tax rate data: http://bit.ly/12IHzSc, accessed May 1, 2013.
a
For fuel, tax rates are computed as 0.71 * diesel rate + 0.29 * gasoline rate based on import shares.
about $20/tonne on diesel, and about $47/tonne on gasoline. 11
The latter two gures are quite high by international stan-
dards, and the striking disparity between these and the implied
tax rate on coal is a further indication that environmental mo-
tives were not dominant in the design of this tax structure.
(a) Labor supply
In Vietnam, the labor market is the most important market
in the economy from the point of view of household income,
income distribution, and poverty. The countrys rapid growth
over the past 20 years has seen enormous movement of labor
across industries and locations, and (increasingly, from a low
base) across skill types. The large numbers of workers chang-
ing both occupation and location in any period suggest a very
elastic labor market response to changing opportunities, with
internal migration by workers estimated at between 0.5 m and
1 m per year (Phan & Coxhead, 2010). In particular, there has
been a surge of predominantly semi-skilled workers (i.e., those
with a completed junior high school education) into urban la-
bor markets. Young women are predominant among internal
rural-urban migrants, perhaps reecting a lower opportunity
cost of their labor in source households and communities.
Yet beyond these observations, surprisingly little is known
quantitatively about the responsiveness of labor supply to real
wage changes.
We adapt the model to incorporate a simple labor supply
structure. The representative household supplies labor that is
dierentiated by location and by education (primary, second-
ary, and tertiary), a proxy for skill. Each of these characteris-
tics is likely to be associated with a dierent supply elasticity,
and the mix of labor endowments within each household type
further dierentiates them from one another. For a worker in
region i with skills j, we posit the labor supply function
LSij LTij  wij =CPI j bij ;
where superscripts S and T denote labor supplied to the mar-
ket and total labor endowment respectively; the CPI is specic
to location (either rural or urban), and bij P 0 is the elasticity
of the labor supply with respect to the real wage. This styliza-
tion ignores income eects and demographic sources of heter-
ogeneity. 12
Other than the initial distribution of labor endowments
across households, the critical labor supply parameters are
the elasticities, b. We have very little information on these.
Classical development theory often posits a horizontal labor
supply curve (b = 1), but this is rarely supported by the data.
Standard trade models typically assume full employment and
perfectly inelastic labor supply (b = 0), but this too is unreal-
istic for developing countries, especially in rural areas where
households typically exhibit both underemployment and
unemployment. Individual workers may supply additional
Legislated environmental tax Environmental tax
(VND/tonne or liter) as% of base pricea
(2) (3)
10,00030,000 Minimum: 2.20
Median: 4.40
1,0004,000 (Gasoline) Minimum: 4.30
5002,000 (Diesel) Median: 10.75
 ARE CARBON TAXES GOOD FOR THE POOR? A GENERAL EQUILIBRIUM ANALYSIS FOR VIETNAM 123

Table 2. Assumed values of labor supply elasticities labor either at the intensive margin (hours worked) or the
Labor type Rural Urban extensive margin (jobs taken by those previously unem-
ployed). The total elasticity of labor supply is the sum of these,
Unskilled (primary education) 1.0 0.5 but most estimates for developing countries appear to apply
Medium-skilled (secondary ed) 0.5 0.1
only to the intensive margin. 13 Household-level labor supply
High-skilled (tertiary ed) 0.1 0.1
is also more elastic than for individuals. If the household
consists only of one or two adult members, they are likely to

Table 3. Environmental taxes: macroeconomic impacts

Variables A: Elastic labor B: Elastic labor C: Fixed labor
supply, no transfers supply, transfers supply, no transfers
Real GDP, % change 0.63 0.46 0.35
Real absorption, % change 0.39 0.29 0.16
CPI, % change 0.72 0.59 0.73
Government tax revenues, % change 3.79 4.05 4.10
Government revenues, change (VND mn) 10,920 11,661 11,806
Govt transfers to households, % change 2.98 18.13 3.02
Output by aggregate sector, % change
Agriculture 2.53 1.83 1.69
Manufacturing 1.47 0.98 1.26
Services 0.50 0.74 0.29
Employment by aggregate sector, % change
Agriculture 1.90 1.30 0.52
Manufacturing 1.27 0.31 0.95
Services 0.12 0.30 0.11
Note: Complete tables of simulation results available from authors.

Table 4. Environmental tax impact on selected sectors (base case, percentage changes)
Selected sectors Output Price Trade
Energy Domestic price
Coal mining 4.12 3.41
Crude oil 8.56 7.07
Petroleum products 5.12 8.14
Electricity and gas 1.65 1.50
Margins and non-traded goods
Retail and wholesale trade 1.97 1.92
Road transport 3.10 6.40
Air transport 2.08 4.80
Other transport 2.17 6.81
Paddy rice 1.93 0.81
Fisheriesa 5.11 7.89
Export goods Fob export price Exports
Fish processing 4.64 0.69 6.62
Textiles 0.61 0.09 0.92
Furniture 0.91 0.00 0.11
Footwear 0.80 0.00 0.02
Vegetable and fruit processing 1.21 0.07 0.72
Wood products 0.35 0.01 0.05
Other food processing 2.38 0.23 2.23
Import-competing goods Cif import price Imports
Vehicles and transport equipment 0.51 0.00 0.48
Other chemicals 1.73 0.00 1.45
Basic metals 1.09 0.00 0.85
Textiles 0.61 0.00 0.57
Metal products 1.09 0.00 0.85
Machinery and equipment 0.87 0.00 0.82
Paper products 1.35 0.00 1.03
Note: Complete tables of simulation results available from authors.
a
Just 3% of sheries output is exported.
124 WORLD DEVELOPMENT

supply additional labor only at the intensive margin. Rural
households, however, are commonly larger and may have
some adult members (in Vietnam, often young women) who
are openly unemployed. These households are likely to exhibit
more elastic supply at both the intensive and extensive mar-
gins.
Ocial sources in Vietnam report urban unemployment
rates to be roughly steady at around 56%, and the ratio of
rural hours worked to hours available at about 75% (Brassard,
2004). About 13% of rural workers are reported in ocial sta-
tistics as working less than full-time and seeking more hours of
work in a week (Coxhead et al., 2010, Table 6). From our eval-
uation of these admittedly fragmentary data, we surmise that
labor supply by rural workers is in general more elastic than
that for urban workers and that supply is more elastic for un-
skilled labor than for medium or high-skilled labor. In the
model, we build a vector of b values from these conjectures.
The assumed values (see Table 2) range from 1 for the most
elastic supplies, to 0.1 for the least. The estimation of labor
supply parameters from household and individual data is the
subject of ongoing research.
(b) Macroeconomic closure assumptions
In order to conduct tax policy experiments we must make
assumptions about the way in which the economy responds
to a shock, choosing which variables and policy settings are
exogenously xed, and which may endogenously adjust.
We assume investment-driven savings, and a xed current
account balance with no endogenous transfers to or from
the rest of the world. The latter assumption requires that the
balance of trade must remain constant. This eliminates unac-
counted intertemporal borrowing, forcing all adjustment into
the single period of the simulation. The nominal exchange rate
of VND for USD (assumed xed) is the models numeraire
price. Therefore, when a shock applied to the model creates

Table 5. Environmental tax: Impact on factor employment and real returns (percent change)
Variables A: Elastic labor supply, no B: Elastic labor supply, C: Fixed labor supply, no
transfers transfers transfers
Wage Employment Wage Employment Wage Employment
Labor
Urban, tertiary ed 0.38 0.02 1.92 0.11 1.35 0
Urban, secondary ed 0.97 0.03 1.27 0.13 2.61 0
Urban, primary ed 1.45 0.14 0.97 0.49 4.02 0
Rural, tertiary ed 0.38 0.02 1.92 0.14 1.35 0
Rural, secondary ed 0.97 0.87 1.27 0.60 2.61 0
Rural, primary ed 1.45 1.38 0.97 0.93 4.02 0
Other factors
Capital, agriculture, xed 4.05 0 2.73 0 2.54 0
Capital, agriculture, mobile 4.46 0 3.23 0 2.87 0
Capital, non-agriculture, xed 3.52 0 3.17 0 2.95 0
Capital, non-agriculture, mobile 1.68 0 1.63 0 1.38 0
Note: Nominal price changes deated by CPI change. The model species distinct urban and rural labor endowments and supply elasticities (see Table 2);
we assume mobility across locations. Changes in returns to xed capital are averages over multiple sectors.

Table 6. Real consumption changes by household type and quintile (% change from base)
Quintile and HH type Percent A: Elastic labor supply, B: Elastic labor supply, C: Fixed labor supply,
rural (%) no transfers transfers no transfers
Quintile 1 (Lowest) 93.5 1.90 0.17 1.65
Urban 2.50 0.94 2.45
Rural 1.85 0.11 1.60
Quintile 2 88.1 1.97 0.49 1.65
Urban 2.58 1.35 2.31
Rural 1.88 0.38 1.56
Quintile 3 77.8 2.06 0.68 1.69
Urban 2.40 1.18 2.03
Rural 1.96 0.53 1.59
Quintile 4 65.5 1.91 0.65 1.47
Urban 2.26 1.15 1.77
Rural 1.73 0.38 1.31
Quintile 5 (Highest) 50.4 1.61 0.50 1.08
Urban 1.70 0.71 1.09
Rural 1.53 0.30 1.08
All Households 74.9 1.89 0.50 1.51
Urban 2.10 0.99 1.62
Rural 1.82 0.33 1.47
 ARE CARBON TAXES GOOD FOR THE POOR? A GENERAL EQUILIBRIUM ANALYSIS FOR VIETNAM
pressures for an imbalance on the trade account, domestic
spending, and (where applicable) factor employment must ad-
just to restore external balance.
In each closure we assume that half the capital in each
industry is xed (immobile), while the other half is mobile (it
can be reallocated across industries or sectors). 14 We also as-
sume that labor of a given skill level is mobile across loca-
tionsthat is, we assume costless rural-urban migration
within the equilibrium timeframe of the model.
Early discussions of Vietnams environmental tax law
claimed that revenues were to be spent on environmental pro-
grams, but the legislation itself makes no mention of this, and
we can nd no evidence that this issue has received serious pol-
icy attention to date. 15 Therefore, and in order to maintain fo-
cus on the incidence of the taxes themselves, we abstract from
environment-specic expenditures. In the base case (Case A)
we assume that additional revenues from the tax increases are
spent in an equiproportional, across-the-board increase in
government consumption of goods and services. Most govern-
ment expenditure is on relatively skill-intensive services such
as public administration and education, these are the indus-
tries and associated factors that will gain most in this case.
For comparison of welfare outcomes, we assume alternatively
that net changes in government revenue are redistributed as an
across-the board increase in transfers to households, again
leaving the governments net budget decit unchanged. In this
case, all existing transfer recipients will receive an equal per-
centage increase in payments (this is Case B). However, trans-
fers contribute diering shares of initial income (Figure 1), so
the impact will vary across households. It should also be noted
that an across-the-board increase in transfer payments is the
same as an increase in payments to the poor. Government
transfers include anti-poverty transfers along with education
subsidies, pensions paid to retirees and invalids, and other
transfers identied in social programs. Rural households in
the poorest two quintiles of the expenditure distribution earn,
on average, 2.5% of their income from government transfers.
Some other household groups earn a lot more, the highest
Figure 1. Sources of household income (percent), by type and quintile. Source: Vietnam SAM 2007 (Arndt et al., 2010).
125
being urban households in the richest two quintiles, who earn
almost 6.5% of income from this source. Our experiment in
Case B, therefore, simply asks whether the existing system of
transfer payments provides a feasible mechanism for compen-
sating households aected by the environmental tax.
Finally, Figure 1 also conrms that households are hetero-
geneous in terms of factor income shares. Labor earnings
are by far the most important single income source, and as ex-
pected, poorer households earn relatively higher shares of in-
come from low-skilled labor. Policy shocks that operate
through the labor market will thus have dierential eects
on household income.
5. SIMULATION RESULTS
We are interested in both structural and distributional as-
pects of the incidence of the environmental tax. The experi-
ment consists of imposing new taxes at the median rates
shown in column 6 of Table 1, and we run three variants. In
the base case (A), as just described, we assume that transfers
from government to households increase only at the same rate
as other components of public spending. Also in this closure,
labor supplies are elastic as shown in the rst column of Ta-
ble 2. We compare this with Case B, in which government uses
the additional tax revenues to increase transfer payments to
poor households. For comparison with the base case, we also
adopt a closure (Case C) in which the supply of all labor types
is assumed xed. Relative to Case A, this highlights the dier-
ential macroeconomic and tax incidence consequences of alter-
nate labor supply assumptions. To save space, the tables
report only the most relevant results in each case; full simula-
tion results are available from the authors.
(a) Macroeconomic impacts
Table 3 shows the main macroeconomic results for all three
cases. 16 In Case A, the model predicts that the environmental
126 WORLD DEVELOPMENT
taxes reduce GDP growth by 0.63% and increase the CPI by
0.72%. The taxes raise government revenues by about 3.8%
(VND 10.9bn, equivalent to about $US640m). Real absorp-
tion falls by 0.39%. At the level of aggregate sectors, the tax
raises costs and reduces output in manufacturing by 2.5%, in
services (including energy-intensive transport services) by
0.12% in services, and in agriculture and resource sectors by
1.9%. Aggregate sectoral employment falls by somewhat more
(1.47% in manufacturing, 0.5% in services, 2.5% in agricul-
ture). Because the supply of labor is somewhat elastic, prices
of capital and land, which by contrast are assumed xed in
supply, fall further than wages in general and this induces sub-
stitution away from labor in production.
Comparing Cases A and C shows the macro implications of
elastic labor supply. With xed factor supplies (Case C) the
aggregate production possibilities frontier remains unaltered
and overall GDP falls by much less (0.35%), whereas in Case
A elastic labor supply permits a larger response. In both cases
A and C, additional tax revenues are used by the government
to increase its own consumption; given base values, this favors
the aggregate services sectors relative to agriculture and man-
ufacturing. In Case B, by contrast, it is households whose
spending pattern is inuential at the margin, and this dier-
ence is reected in a smaller decline in agriculture (i.e., food)
output in Case B relative to Case A, and a larger decline in ser-
vices output in Case B relative to Case A.
Among individual industries, the range of impacts is wide,
and industry-level variation in production and prices illus-
trates the means by which the incidence of the taxes is shifted.
Dierences among all cases are small in this respect, so we re-
port results only for Case A, in Table 4. The industries directly
targeted by the tax are of course seriously aected. Domestic
output of coal falls by 4.1%, and that of fuels by 5%. Gas-
oline imports, which account for most of the domestic supply,
fall by 5.1%. Electricity and gas output falls: this industry
consists mainly of coal-burning electrical power generation
plants, so the tax drives up production costs and this eect
dominates any substitution among energy sources, which by
itself would increase electricity demand.
Other industries, especially those that are energy-intensive,
experience cost increases. Among these, transport industries
are the most severely aected since gasoline and other fuels ac-
count for between 2235% of their costs. 17 The prices they
charge to users rise, by 6.4% (road transport), 4.8% (air trans-
port) and 6.8% (other transport, i.e., rail and water). Higher
transport costs mean more costly production and distribution
(domestic and export margins), and so raise costs throughout
the economy.
Among industries that are especially badly hurt are produc-
ers of some key natural resource products and exports derived
from them. An important case is that of sheries, in which fuel
accounts for 44% of input costs. When the taxes are imposed,
the sheries industry experiences a large cost increase; produc-
tion falls by 5.1% and its price rises by almost 8%. This sec-
tor sells 97% of its output in domestic markets; of this, 66%
goes to seafood processors, for whom purchases from sher-
ies, aquaculture, and upstream seafood processors make up
81% of total cost22% of it directly from sheries. Conse-
quently, even though fuel is only around 1% of total produc-
tion cost in seafood processing, this industry suers a major
indirect cost increase though its purchases from sheries, and
as a result its output falls by 4.6% and exports by 6.6%.
Similarly, output and exports of other food processing indus-
tries fall by 3.4% and 2.2% respectively, while other major ex-
port industries suer declines of smaller magnitude. Because
Vietnam is a small supplier to global markets, however, border
prices of these tradable sectors outputs hardly change. Final-
ly, many import-competing manufacturers are also hurt, with
relatively large output cuts in energy-intensive sectors like
chemicals and paper products.
An important insight from these results is that some indus-
tries experience much greater adjustment pressure than others.
In general, energy generation and energy-intensive activities
are most heavily aected, as expected. But these include some
industries which, because they supply domestic markets virtu-
ally free from international competition, have discretion to
raise prices, thereby passing part of their input cost increase
forward to other industries and nal consumers. Transport
services and sheries exemplify this property.
Other industries produce goods for which output prices are
set in the world market; processed seafood, fruit and vegetable
products, clothing, furniture, and coee are examples. Among
these are several industries that are large employers, especially
of low-skilled labor. In Figure 2, which for tradable industries
plots their share in total employment against their export inten-
sity, these industries are all net exporters whose shares in total
employment are at or above the median. Faced with higher
costs and elastic foreign demand, these industries cannot raise
prices and so must instead cut productionand jobs. There-
fore, their contraction has relatively large impacts on employ-
ment and wages. These industries inherit the environmental
tax burden from upstream suppliers and pass it back to factor
owners. They are not necessarily energy-intensive in produc-
tion, but are aected by energy prices in other waysmost
especially through their use of domestic transport services to
source inputs and to move outputs to markets and ports.
We see the consequence of this adjustment in the wage and
employment data. Wages and other factor prices are reduced
by the taxes (Table 5). Due to migration, urban and rural wage
changes are the same for each labor type. When labor supply is
inelastic (Case C), dierential wage changes are driven by the
impacts of the tax. With the largest contractions taking place
among highly labor-intensive industries, wages for low and
medium skill workers fall sharply. More elastic supply means
smaller wage declines, but accompanied by falls in employ-
ment, as in Cases A and B. Adding wage and employment
changes, the earnings of unskilled labor fall in Case A by
1.6% (urban) to 1.8% (rural); medium skilled by 1% (urban)
to 1.8% (rural), while high-skilled earning rises by 0.4%. In
Case B the income declines are somewhat larger, though
now the earnings of tertiary-educated workers also fall as
the increase in government expenditures is redirected from ser-
vices to transfers.
Other factor returns also decline in both nominal and real
terms. Although our model does not track investment deci-
sions, lower unit prices for xed and mobile capital signal that
expected investment returns in the medium run will be lower
than in the absence of the tax. Unless there are compensatory
adjustments, investment at the margin will be diverted to other
uses, quite likely in other economies. Environmental consider-
ations aside, this investment reaction threatens to undermine
the basis for Vietnams economic growth.
As seen, the taxes increase overall prices due to the pervasive
eects of energy cost hikes. Higher prices for consumer goods
reduce households purchasing power, and, by lowering real
wages, also reduce aggregate employment when labor supply
is price-responsive. Table 6 summarizes key results of the tax
on the real consumption expenditures of household groups.
The pre-transfer impact of the environmental taxes is to re-
duce average real incomes for all household types and quin-
tiles. The biggest losses are by urban household groups,
whose factor incomes depend relatively heavily on medium-
 ARE CARBON TAXES GOOD FOR THE POOR? A GENERAL EQUILIBRIUM ANALYSIS FOR VIETNAM 127

Figure 2. Employment shares and net trade orientation of industries. Vertical line shows median employment share. Source: Vietnam SAM 2007
(Arndt et al., 2010).

skill labor and non-agricultural capital returns. When trans-
fers take place in Case B, the picture is very dierent. For some
groups, including the largest (rural households in quintiles 1
and 2) higher transfers are almost sucient to compensate
for their factor income losses. Their average real income
changes remain negative but are small. For others, however,
the transfers diminish but do not eliminate the eect of factor
market losses. The taxes are somewhat regressive in their im-
pact: In Case A, the average losses of all households, urban
or rural, in the lowest three quintiles are equal to or greater
than the average for all households. The transfers, in contrast,
have weakly progressive eects among the poorer three quin-
tiles, but also confer larger than average gains on the richest
quintile.
We also impute the overall eects of the environmental taxes
on changes in household-level poverty and economic welfare,
by mapping changes from the simulation model to the VHLSS
database as described above. Table 7 shows percentage
changes headcount poverty for the country as a whole and
for urban and rural household groups. 18 In calculating
poverty changes from VHLSS, we apply the predicted changes
in factor prices and supplies, transfers, and consumer prices to
each household according to its endowments, initial transfer
income, and expenditure pattern. This gives household-specic
poverty change results. Table 7 shows summaries of these,
constructed by weighting each household result to ensure that
poverty change measures are representative at regional and
national levels.
The poverty change predictions show that without transfers,
the tax sharply raises the numbers of those counted as poor.
Both the national headcount rate and rural poverty rise by
about two-thirds of one percentage point. Inclusive of trans-
fers, these increases in the numbers of the poor are smaller
by about half. However, the gains from increased transfers
are unequally distributed, with the regions that are home to
Vietnams two largest cities, Hanoi and Ho Chi Minh City,
experiencing large gains relative to the no-transfer case, while
more remote northern regions and the Mekong Delta, which
together account for 69% of poor households, gain much less.
These are regions with large concentrations of low-skill labor.

Table 7. Implied changes in headcount poverty due to environmental tax

Baseline Share of Increase (percentage points from base)
povertya (%) poor HHs (%)
A: Elastic labor supply, no transfers B: Elastic labor supply, no transfers
National 15.86 0.62 0.27
Urban 6.64 0.35 0.16
Rural 19.22 0.72 0.32
Regions
Red River Delta (Inc. Hanoi) 11.78 8 0.44 0.02
North East 20.75 22 0.55 0.42
North West 45.20 19 1.04 1.12
North Central Coast 26.70 17 0.52 0.22
South Central Coast 15.71 5 0.67 0.37
Central Highlands 25.20 13 0.94 0.03
South East (inc. HCM City) 5.28 5 0.24 0.11
Mekong River Delta 11.35 11 1.02 0.56
a
Calculated from 2006 VHLSS using national poverty lines of VND 265,000/person/month for rural and VND 280,000/person/month for urban areas.
128 WORLD DEVELOPMENT

Figure 3. Change in household real income due to environmental tax: Case A, no transfers. Vertical line shows national poverty line.

Figure 4. Change in household real income due to environmental tax: Case B, transfers. Vertical line shows national poverty line.

The Mekong Delta region, moreover, is the main location for
Vietnams aquaculture and seafood processing industries.
A comparison of Figures 3 and 4 provides more insight into
the household welfare impacts of the environmental taxes with
and without a compensating increase in transfers. These g-
ures plot real income changes for all households against their
initial income (shown in log form for easier exposition). In
Figure 3 panel a, we see that while most rural households
experience declining real income due to the taxes, the propen-
sity for losses is especially acute among initially poor house-
holds (those to the left of the poverty line marked on the
gure). The corresponding panel of Figure 4 shows that com-
pensating transfers generate moderate to large real income
gains for many households, but leave many among the poor
relatively unaected. A similar pattern can be seen in the data
for urban households, shown in panel b of each gure.
6. DISCUSSION AND CONCLUSIONS
Does the desire to contribute to a cleaner local and global
environment conict with other important development policy
goals? In this paper we predict the broad economic impacts of
a set of environmental taxes recently legislated in Vietnam.
These predictions are based on simulation experiments with
an applied general equilibrium model of the Vietnamese econ-
omy. Our approach has been to simulate the environmental
taxes as ad valorem sales taxes on coal, gasoline, and other
fuels. This approach is consistent with the most important fea-
tures of the environmental tax legislation.
Results from the model show counterfactual changes in
variables such as prices, employment, incomes, and poverty.
In order to evaluate the incidence of the taxes on household
welfare, we have also posited that additional revenues be
either spent on increased government consumption, or re-
turned to households in the form of increases in the existing
pattern of government transfers.
Because these are taxes on energy, their eects on economic
activity are pervasive. Transport sectors are especially aected,
but these largely non-traded industries pass the tax burden
forward to others. As a result, energy-dependent, export-ori-
ented industries like seafood processing experience cost in-
creases. Since their output prices are set in the world
market, they also suer declines in production, exports, and
 ARE CARBON TAXES GOOD FOR THE POOR? A GENERAL EQUILIBRIUM ANALYSIS FOR VIETNAM
employment. These losses are then passed back to factor
owners.
Among factor markets, that for labor is the most important
channel for translating changes in economic activity into
changes in household welfare. We investigate this channel un-
der the assumption that the supply of certain types of labor is
responsive to real wage changes, and also (for comparison) un-
der the standard assumption of full employment and zero
elasticity of labor supply response. The resulting dierences
in predicted changes in wages and aggregate employment
map into contrasting household welfare changes. More impor-
tant still is the decision by the government to return all addi-
tional tax revenues to households in the form of proportional
increases in existing transfer payments. This leads to an out-
come in which the environmental taxes reduce real GDP and
aggregate employment by somewhat less than when the pat-
tern of government expenditures remains unchanged. How-
ever, the transfer system is not strongly progressive, so the
poorest households still experience substantial losses.
The experiments thus yield mixed results for policy. The uni-
lateral adoption of environmental taxes hurts a small econ-
omys competitiveness in global markets for a wide range of
products. In the case of Vietnam, a low-income, labor-intensive
economy, the losers include some of its largest and most labor-
intensive export industries, even through these are not them-
selves highly energy-intensive in production. The losses they
experience impede job growth, which is an important consider-
ation in an economy where the labor force is still growing at
more than 1.7% per year, adding about 900,000 new job-seekers
each year. 19 Signicantly higher revenues enable government,
in principle, to compensate poor households for losses shifted
forward to them (as consumers) and backward to them (as fac-
tor owners). However, increasing the rate of existing transfers
to households does not eliminate the losses of the poorest
households, because it does not compensate for lost jobs.
In the longer run, an economy responds to higher energy
prices in multifarious ways; these include interfuel substitution,
improvements in technology and eciency, and many smaller
adjustments in policy. Therefore the experimental results re-
ported here, in which many other variables are held constant,
are indicative but not conclusive predictions of real-world out-
comes. However, they reveal some important but indirect chan-
nels of tax incidence, and because of this it is our view that they
merit careful consideration at policy level. An initial reaction
might be to regard the impending environmental taxes as a
costly threat to continued economic growth and development
in a lower-income economy. We do not support this conclusion,
because other consequences of the taxes, not captured or valued
in this model, have yet to be taken into account. To frame ongo-
ing debate, we oer the following thoughts.
First, although we can estimate likely reductions in the
growth of industrial emissions, we lack the information neces-
sary to assign valuations of the environmental benets of the
new taxes. These may take the form of reduced abatement
costs (that is, costs associated with remediation of pollution,
clean-up of water supplies, waste disposal, and so on), and
also lower incidence of pollution-related damages to health.
Studies in other developing countries indicate that particulate
matter and gaseous emissions from industries and vehicles
have large and costly impacts on human health, as well as
reducing the productivity of labor. If environmental taxes re-
duce emissions growth, then they also deliver benets in the
form of a more healthy and productive workforce. These ben-
ets should be taken into account when evaluating the net
gains and losses from an environmental tax. Because we do
not, our results may understate the gains from the new taxes.
129
Second, in the absence of better information, we have as-
sumed that revenue gains from the taxes are either spent on
additional government consumption or on increases in existing
transfers from government to households. These are useful
mechanisms for the purpose of evaluating the incidence of
the taxes themselves. But they are not by any means the only
or the best way to dispose of additional revenues. The double
dividend literature identies improvements in the eciency of
the tax system as a second gain beyond the environmental ben-
ets themselves (Bovenberg & Goulder, 1996). Distortionary
taxes reduce social welfare; therefore, if revenue gains from
environmental taxes (which themselves reduce distortions, by
helping to correct pollution externalities) are used in budget-
neutral fashion to reduce the rate of more distortionary taxes
such as import taris or factor income taxes, then social wel-
fare may improve. Once again, because we have not included
such possibilities, our experiments may underestimate the po-
tential social gains from the environmental taxes.
Third, new tax revenues could also be used to oset cost in-
creases in energy-intensive and/or employment-intensive indus-
tries. Our experiments show that tradable sectors in particular
suer contractions and job losses because they are unable to
pass on tax-related cost increases. To minimize the impact on
these industries, and therefore on national employment and ex-
port competitiveness, the Vietnamese government might con-
sider using environmental tax receipts to compensate them,
for example by means of a corporate tax rebate proportionate
to their input cost increases. This revenue-recycling strategy is
more narrowly targeted than that discussed in the previous par-
agraph, and as such bears substantial risks (for rent-seeking
and corrupt practices) along with its potential gains. On the
other hand, its potential for benecial impacts on employment
growth and export revenues may well justify such risks.
Finally, since the lower emissions create a global externality
(lower GHG emissions), there is a compelling case for com-
pensation from the international community. In principle, this
could be sucient to fully compensate for the loss of aggregate
income. The UN Clean Development Mechanism (CDM),
established under the Kyoto Protocol, would seem to provide
a vehicle for such compensation. The CDM allows Kyoto Pro-
tocol Annex I states (mainly OECD members) to sponsor
emissions-reducing projects in non-Annex I states (mainly
developing countries) as a way of helping meet their own
GHG reduction commitments. More than 4,000 projects have
been funded to date, and by raising about $18bn in direct car-
bon revenues over its 200112 implementation period, the
CDM is the dominant emissions reduction mechanism for
developing countries (World Bank, 2010: 262). Two issues re-
main, however. First, by themselves any transfers to Vietnam
from the international community would no more address
non-environmental development targets (such as job creation)
than would compensation paid to households by the Vietnam-
ese governmentas discussed above. Second, to date all pro-
jects funded under the CDM have been linked to specic
activities such as reaorestation, introduction of new technol-
ogies, fuel substitution, etc. We know of no project as yet
funded that rewards a country for implementing emissions-
reducing incentive mechanisms such as energy taxes. Indeed,
the CDM Methodology Booklet (UNFCCC, 2010), which lists
and describes hundreds of CDM-permissible methodologies
for emissions reduction, contains no mention at all of taxes
or other mechanisms based only on incentives to alter behav-
ior. So deployment of the CDM in support of an environmen-
tal tax such as that legislated in Vietnam would seem also to
require a substantial change in the mode of operation of that
international mechanism.
130 WORLD DEVELOPMENT
NOTES
1. The law passed the National Assembly in November 2010. For an
English language version see http://bit.ly/12IHzSc.
2. Fullerton and Metcalf (2002) provide a helpful model illustrating these
points.
3. We return to this point in the concluding section of the paper.
4. Some more recent contributions focus on labor markets using a
HarrisTodaro migration structure. See, for example, Daitoh and Omote
(2011).
5. For recent surveys of AGE models applied to the Vietnamese
economy, see Coxhead et al. (2010) and Abbott, Bentzen, & Tarp, 2009.
6. Earlier editions of the Vietnam SAM identied male and female
workers separately, but the 2007 SAM makes no such distinction.
7. Vietnams shares in world exports and imports are 0.53% and 0.58%
respectively (WTO, http://stat.wto.org/CountryProle/
WSDBCountryPFView.aspx?Language=E&Country=VN, accessed
March 1, 2013).
8. In structure, our model is a hybrid. It is simultaneous in the behavior
of the household categories embedded in the social accounting matrix, but
recursive in the mapping from these to the full sample of VHLSS
households. This structure means that we can calculate unique household
impacts of each shock in the VHLSS sample, but imposes the assumption
that all households within each category respond identically to a given
shock.
9. Vietnam uses a variety of instruments to pursue fuel price stabiliza-
tion, including a price stabilization fund and point of sale price limits
(Kojima, 2013). In the longer run, however, domestic price changes have
tracked trends in global benchmark prices (see data at http://energype-
dia.info/wiki/Fuel_Prices_Vietnam, accessed May 8, 2013).
10. Only EU, Australia, and New Zealand currently have carbon taxes.
Several developing countries have implemented or are proposing taxes on
energy similar to the Vietnamese tax.
11. Calculated at the rate of VND18,000:US$1, and with emissions
coecients of (coal) 2,336 kg CO2e/tonne; (diesel) 2.68 kg CO2e/l;
(gasoline) 2.32 kg CO2e/l. Source for CO2e conversions: http://www.car-
bontrust.co.uk/cut-carbon-reduce-costs/calculate/carbon-footprinting/
pages/conversion-factors.aspx.
12. The denition of labor supply and the means of measuring it are both
controversial (e.g. Blundell & McCurdy, 2008). Among many key issues
relevant to our study are the distinction between supply at the intensive
REFERENCES
Abbott, P., Bentzen, J., & Tarp, F. (2009). Trade and development:
Lessons from Vietnams past trade agreements. World Development,
37(2).
Arndt, C., Garcia, A., Pham, Hoang Ha, McCoy, S., Tarp, F.,
Thurlow, J. (2010). A 2007 social accounting matrix (SAM) for
Vietnam. In Study Prepared under the CIEM-Danida Project
Strengthening the Development, Research and Policy Analysis
Capacity of CIEM (mimeo).
Atkinson, A. B., & Stiglitz, J. E. (1980). Lectures on public economics. New
York: McGraw-Hill.
margin (hours worked) and at the extensive margin (labor force
participation). Moreover, family labor supply diers substantively from
that by individuals. Preference heterogeneity within familiesand across
the labor force as a wholeis associated with more elastic labor supply.
13. Empirically, our search for estimates of labor supply elasticities in
developing countries returned few results. Rosenzweig (1980), using
individual data, found inelastic o-farm supply of labor hours by men and
women with land, but considerably more elastic supply from landless
workers. Using data from Sri Lanka, Sahn and Alderman (1988) nd the
elasticity of rural labor supply w.r.t. wages of 0.135 (men) and 0.149
(women), with smaller values (0.07 and 0.03) for urban workers. These are
based on individuals and apply only to the intensive margin, and so are at
the lower bound of such estimates. In more complete analyses, supply
elasticities at the extensive margin dominate (in magnitude) those at the
intensive margin (Blundell & McCurdy, 1999; Heckman, 1993). The
latters surveys of empirical estimates range very widely, with some
empirical models yielding values of 1 and even more, especially for
women.
14. The Vietnam SAM provides no guidance on xed and mobile capital,
and our 50-50 division is merely an ad hoc assumption. However,
robustness checks using 25-75 and 72-25 splits show that the results are
not very sensitive to the exact percentage allocations.
15. The lack of a direct link to environmental measures is a distinctive
feature of Vietnams tax when compared with similar policies in other
countries (see Gale, Brown, & Saltiel, 2013 for a survey). The most likely
reason for this disconnect is institutional: the tax is designed and
implemented by the Ministry of Finance, whereas responsibility for
environmental policy rests with the Ministry of Environment and Natural
Resources, and there are few meaningful mechanisms for interagency
coordination. There are no mentions of the environmental tax law
accessible on the MONRE website http://www.monre.gov.vn/, accessed
March 1, 2013).
16. More complete details of the model, database and results are
available from the authors.
17. These and other baseline data reported in the text are calculated from
the 2007 SAM; see Arndt et al. (2010).
18. It is important to notice that although these predictions are
aggregated into the same set of household groups as in Table 6, they
are obtained from a dierent data set and are not precisely comparable.
19. Data from World Bank: World Development Indicators Online,
accessed on March 1, 2013.
Blundell, R., & McCurdy, T. (1999). Labour supply: A review of
alternative approaches. In O. Ashenfelter, & D. Card (Eds.).
Handbook of labor economics (Vol. 3, pp. 15601695). Amsterdam:
North-Holland.
Blundell, R., & McCurdy, T. (2008). Labour supply. In S. N. Durlauf, &
L. E. Blume (Eds.), The new Palgrave dictionary of economics (2nd ed.).
Palgrave Macmillan.
Bovenberg, L., & Goulder, L. (1996). Optimal environmental policies in
the presence of other taxes: General equilibrium analyses. American
Economic Review, 86(4), 9851000.
 ARE CARBON TAXES GOOD FOR THE POOR? A GENERAL EQUILIBRIUM ANALYSIS FOR VIETNAM
Brassard, C. (2004). Wage and labour regulation in Vietnam within the
poverty reduction agenda. Public Policy Programme, National Univer-
sity of Singapore, manuscript.
Copeland, B. R., & Taylor, M. S. (2003). Trade and the environment:
Theory and evidence. Princeton, NJ: Princeton University Press.
Coxhead, I. (2000). Trade and tax policy reform and the environment in
developing economies: Is a double dividend possible? FEEM Working
Papers No. 88.2000. Milan.
Coxhead, I., Nguyen, Chan V., Wattanakuljarus, Anan, Kim, Tran Thi
Dung, Kim, Pho Chi, et al. 2010. A general equilibrium model for the
study of globalization, poverty and the environment in Vietnam.
University of Wisconsin-Madison and National Economics University,
Hanoi, Vietnam. Accessible at: <http://www.aae.wisc.edu/coxhead/
papers/VNCGE-ModelDescription.pdf>.
Daitoh, I., & Omote, M. (2011). The optimal environmental tax and urban
unemployment in an open economy. Review of Development Econom-
ics, 15(1), 168179.
Datta, A. (2010). The incidence of fuel taxation in India. Energy
Economics, 32(Suppl. 1), S26S33.
El Obeid, A., Van der Mensbrugghe, D., & Dessus, S. (2002). Outward
orientation, growth, and the environment in Vietnam. In J. Beghin, D.
Roland-Holst, & D. van der Mensbrugghe (Eds.), Trade and the
environment in general equilibrium: Evidence from developing economies.
Boston: Kluwer Academic Publishers.
Fullerton, D., & Heutel, G. (2007). The general equilibrium incidence of
environmental taxes. Journal of Public Economics, 91(34), 571591.
Fullerton, D., & Metcalf, G. (1997). Environmental taxes and the double-
dividend hypothesis: Did you really expect something for nothing? NBER
Working Papers No. 6199.
Fullerton, D., & Metcalf, G. (2002). Tax incidence. NBER Working
Papers No. 8829.
Gale, W. G., Brown, S., & Saltiel, F. (2013). Carbon taxes as part of the scal
solution. Manuscript, Brookings Institution. <http://www.brookings.edu/
research/papers/2013/03/12-carbon-tax-gale> Accessed 31.03.13.
Harrison, W. J., & Pearson, K. R. (1996). Computing solutions for large
general equilibrium models using GEMPACK. Computational Eco-
nomics, 9, 83127.
Heckman, J. J. (1993). What has been learned about labor supply in the
past twenty years?. American Economic Review, Papers and Proceed-
ings, 83(2), 116121.
Horridge, M. (2005). ORANI-G: A generic single-country computable
general equilibrium model Edition prepared for the Practical GE
Modelling Course, February 711, 2005. Australia: Centre of Policy
Studies and Impact Project, Monash University.
Kojima, M. (2013). Petroleum product pricing and complementary policies:
Experience of 65 developing countries since 2009. Washington, DC:
World Bank.
131
Lofgren, H., Harris, R. L., & Robinson, S. (2002). A standard computable
general equilibrium (CGE) model in GAMS. Washington, DC: Inter-
national Food Policy Research Institute.
Metcalf, G. (2008). Environmental taxation: What have we learned in this
decade?. Conference on tax policy lessons from the 2000s, May 30 2008.
Washington, DC: American Enterprise Institute.
OECD (1994). The distributive eects of economic instruments for
environmental policy. Paris: OECD.
OECD (1995). Climate change, economic instruments and income distribu-
tion. Paris: OECD.
Parry, I., Sigman, H., Walls, M., & Williams, R. C., III (2005). The
incidence of pollution control policies, NBER Working Papers No.
11438.
Phan, Diep, & Coxhead, I. (2010). Interprovincial migration and
inequality during Vietnams transition. Journal of Development Eco-
nomics, 91(1), 100112.
Rosenzweig, M. R. (1980). Neoclassical theory and the optimizing
peasant: An econometric analysis of market family labor supply in
a developing country. Quarterly Journal of Economics, 94(1),
3155.
Sahn, D. E., & Alderman, H. (1988). The eects of human capital on
wages, and the determinants of labor supply in a developing country.
Journal of Development Economics, 29(2), 157183.
Sterner, T. (Ed.) (2012). Fuel Taxes and the Poor: The Distributional
Eects of Gasoline Taxation and Their Implications for Climate Policy.
Oxford, UK and Washington DC, USA: RFF Press.
UNFCCC (United National Framework Convention on Climate
Change). (2010). CDM Methodology Booklet. <http://cdm.unfcc-
c.int/methodologies/documentation/meth_booklet.pdf> Accessed
23.07.11.
Willenbockel, D. (2011). Environmental tax reforms in Vietnam: An ex
ante general equilibrium assessment. In Paper for presentation at
Ecomod 2011, University of the Azores, Ponta Delgada, July 2011.
<http://ecomod.net/system/les/EcoMod2011_VietnamEcoTax_0.pdf>
Accessed 01.08.11.
World Bank (2010). World Development Report 2010: Development and
climate change. Washington, DC: World Bank.
Yusuf, A. A. (2008). The distributional impact of environmental policy: The
case of carbon tax and energy pricing reform in Indonesia. Report No.
2008-RR1. Singapore: Environment and Economy Program for
Southeast Asia Research.
Ziramba, E., Kumo, W. L., & Akinbiade, O. A. (2009). Economic
instruments for environmental regulation in Africa: An analysis of the
ecacy of fuel taxation for pollution control in South Africa. CEEPA
Discussion Paper No. 44. <http://www.ceepa.co.za/Discussionpaper-
no44.pdf> Accessed 02.05.11.