Professional Documents
Culture Documents
http://dx.doi.org/10.1257/jel.53.1.5
5
6 Journal of Economic Literature, Vol. LIII (March 2015)
Developed
150
Developing
121
100
50 42.56
38.17
32.5
23 24.014
17.8333
a
es
a
ia
n
sia
il
di
y
in
pa
an
ss
at
az
In
ne
Ch
Ru
St
Ja
Br
m
do
er
d
te
In
G
ni
U
7.99
8
7.27
6.38
4
3.31
0
sia
es
ia
n
a
ia
il
in
pa
d
y
at
ss
az
ne
an
In
Ch
Ja
St
Ru
Br
do
m
d
er
In
te
G
ni
U
Notes: Panel A shows average particulate matter (PM 10) from urban centers, in μg / m 3, using data from the
World Health Organization (2011). Panel B shows dissolved oxygen, in mls/litre, using data from the United
Nations Environment Program (2001). The four most populous developed and developing countries are
shown, ranked according to the pollution measure (from least to most polluted).
Greenstone and Jack: Envirodevonomics: A Research Agenda for an Emerging Field 7
The developing countries are remarkably preference data, really 10,000 times lower
dirtier (higher particulates, lower dissolved than typical figures for the value of a statis-
oxygen).2 tical life in the United States? Is the current
These stark differences in environmen- level of environmental quality in developing
tal quality appear to have paradoxical con- countries optimal, leaving no room for policy
sequences. On the one hand, the available improvements (i.e., is poor environmental
evidence suggests that they lead to large quality just another dimension of poverty)?
health and productivity losses. For exam- Is it possible that the welfare loss from poor
ple, figure 2 shows the striking differences environmental quality in developed coun-
in the burden of disease from air and water tries is greater than in developing countries
pollution in developed and developing coun- in spite of the substantially cleaner condi-
tries, as calculated by the World Health tions in the former?
Organization.3 On the other hand, in spite of This paper argues that a series of critical
this large disease burden, the small handful economic and policy questions about envi-
of studies measuring marginal willingness ronmental quality in developing countries
to pay (MWTP) for environmental quality cannot be properly analyzed or understood
improvements indicate low valuations by with the tools of environmental economics
affected households. For example, using alone or the tools of development economics
households’ willingness to pay for access to alone. We believe that research is increas-
clean water in Kenya to impute the value of ingly producing credible answers to these
a statistical life (VSL) leads to an estimate of questions and a new field is emerging that
USD2013 860 (Kremer et al. 2011), while is at the intersection of these two larger and
typical VSL numbers from the United States more well-established fields.5 At the risk of
are on the order of USD2013 8.6 million excessive reductionism, we argue that this
(U.S. EPA 2010).4 field can be organized around a central ques-
These seemingly contradictory facts raise tion: Why is environmental quality so poor in
a series of compelling questions and puzzles. developing countries?
Given poor environmental quality and high This paper develops four potential expla-
health burdens in developing countries, why nations for the poor state of environmental
is MWTP for environmental quality seem- quality in developing countries that apply in
ingly so low? Put another way, is the value of varying degrees across contexts. These four
a life in Kenya, as suggested by the revealed explanations go beyond the traditional mar-
ket failures associated with the public goods
and externality characteristics of environ-
2 We follow the UN categorization of developed and
mental quality and are likely to be especially
developing countries (UN Statistics Division), though the
binary country-level classification overlooks large varia- important in developing countries. First and
tions both within category and within country. most obviously, environmental quality may be
3 Obtaining causal estimates of the health consequences
low because MWTP for environmental qual-
of environmental quality is challenging. Randomized con-
trol trials are unethical in many settings, and quasi-experi- ity is low. There are several possible causes
ments face limitations when it comes to long-run impacts. of seemingly low MWTP for environmental
In this paper, beginning in section 3, we emphasize the
findings from experimental and quasi-experimental studies
that we believe provide the most reliable causal estimates 5 In many respects, this research is an answer to
of the health impacts of pollution in developing countries. Horowitz’s call for increased credibility in empirical
4 As the conceptual framework makes explicit, MWTP research at the intersection of development and environ-
for environmental quality includes not only the valuation of mental economics in his review of the Dasgupta and Mäler
morbidity and mortality benefits of improved environmen- (1997) edited volume The Environment and Emerging
tal quality, but also aesthetic and income benefits. Development Issues (Horowitz 1998).
8 Journal of Economic Literature, Vol. LIII (March 2015)
3.39
Developed
3
Developing
1.91
2
1.33
1.14
a
n
a
il
ia
an
sia
at
di
in
pa
az
ss
St
m
In
Ch
ne
Br
Ja
Ru
er
do
te
G
In
ni
U
300
Developed
Developing
200
129.91
123.24
100
55.47
sia
n
zil
a
y
ia
a
in
pa
an
di
at
ss
ne
Ch
In
St
Br
Ja
m
Ru
do
er
d
In
te
G
ni
U
Notes: Panel A shows the burden of disease in deaths among children under five per 100,000 from outdoor air
pollution. Panel B shows the burden of disease in deaths among children under five per 100,000 from poor
water quality, sanitation, and hygiene. Data are from the World Health Organization (2004) for the four most
populous developed and developing countries.
Greenstone and Jack: Envirodevonomics: A Research Agenda for an Emerging Field 9
through spending on self protection, s, (scaled by the marginal utility of consump-
such as indoor air purifiers or water treat- tion) and the indirect benefit of environmen-
ment. Finally, environmental quality affects tal quality for health (again scaled by the
income (equation 3), which in turn affects marginal utility of consumption), as well as
utility via the budget constraint. Tourism the impact of environment on income and the
revenue from a national park or agricultural indirect impact of environmental quality on
income that is sensitive to water quality or income via the changes in health.8 Equation
the presence of pollinators are examples of (5) makes clear that if utility is concave in
environment-dependent income. consumption, low levels of income will cor-
In addition, by allowing consumption to respond to high marginal utility of income
affect experienced environmental quality (λ y) and low M
WTPe .9
(equation 4), the model acknowledges that Similarly, we can calculate marginal WTP
agents may affect environmental quality for self-protection s:
both directly, through Δe, and indirectly,
through cand s. We model the latter as a (c, s) 1 (__
(6) MWTPs ≡ __ ∂ u __
∂ a + __
( + ))
∂ u __
∂ h ∂ h __
__ ∂ a
λ y ∂ e ∂ s ∂ h ∂ s ∂ e ∂ s
to highlight consumption geared toward
self-protection, such as air conditioning or
∂ Δy ∂ Δy
+ ___ ∂ a + ___
__ ( ∂ h + __
__ ∂ h __
∂ a .
)
bottled water, that creates feedbacks between ∂ e ∂ s ∂ h ∂ s ∂ e ∂ s
self-protection, health, and environmental
quality. Note that in this perfect markets set The MWTP for self-protection is composed
up, environmental quality is affected only by of the indirect effect of self-protection on
the agent’s own choices, and does not include environmental quality and on health, both
the environmental quality, consumption, or of which are scaled by the marginal utility
self protection choices of others. of consumption, as well as its indirect effect
The agent chooses c, Δe, and sto equalize on income via production and health. If the
the marginal utility of these investments. Let marginal utility of consumption is decreas-
λ e be the marginal utility of environmental WTPs is higher at higher levels of
ing, then M
quality improvements and λ ybe the marginal consumption, as long as any negative impacts
utility of consumption. Individual marginal of self-protection on environmental quality
willingness to pay is therefore the marginal can be offset by compensatory investment
rate of substitution between income and in self-protection (in other words, the
environmental quality. second of these indirect effects can be used
λ
(5) MWTPe ≡ __
e
λ
y 8 A large literature investigates the hypothesis that there
is an inverted-U shaped relationship between income
≡ __ ∂ u + __
1 (__ ∂ u __
∂ h
)
and environmental quality—the so-called environmen-
λ y ∂ e ∂ h ∂ e tal Kuznets curve (e.g., Grossman and Krueger 1995;
Andreoni and Levinson 2001; for reviews of the litera-
∂ Δy ∂ Δy ture, see Dasgupta et al. 2002; Stern 2004). A number of
+ ___
+ ___
∂ h .
__ potential explanations underlie the proposed relationship,
∂ e ∂ h ∂ e including ones consistent with the micro model we present
here. The macro relationship does not, however, appear
robust across pollutants or time (Harbaugh et al. 2002).
This marginal WTP function expresses in 9 Note that λ also captures the indirect effects of
y
dollars how much the agent will be willing to consumption via environmental quality and its effect on
pay for a marginal increase in environmental ∂ u + (__
__ ∂ u + __
∂ u __
∂ h ) ∂ a
__
∂ c ∂ e ∂ h ∂ e ∂ c
production and health. λ y =
_____________
, which
quality. MWTPe is composed of the aesthetic 1 − (___
∂ h ∂ e ) ∂ c
∂ Δy ∂ Δy
∂ h __
+ ___
__ ∂ a
∂ u if __ ∂ e
benefit from improved environmental quality collapses to λ y = __ ∂ a = 0.
∂ c ∂ c
12 Journal of Economic Literature, Vol. LIII (March 2015)
to offset the first). Thus, agents will prefer developing countries are poor, and there-
a higher income with greater investment in fore the marginal utility of consumption is
self-protection even if self-protection has a high relative to the marginal utility of envi-
negative impact on environmental quality. ronmental quality. To put it simply: when
The marginal cost of improving environ- people are very poor, what little money they
∂ c have goes toward immediate consumption
mental quality is ___
e and the marginal cost
∂ Δe needs.
∂ c
of self protection is ___s . The representative The comparative statics of our frame-
∂ Δs
agent will therefore set the marginal costs of work illustrate this point. The agent trades
environmental quality improvements and of off consumption and environmental quality
self protection equal to their respective mar- by setting the marginal utility of environ-
ginal WTP(as defined in equations 5 and 6), mental quality equal to the marginal utility
such that the ratios are equal to each other: of consumption. If the marginal utility of
consumption is decreasing, then the agent is
___ ∂ c likely to forego investments in environmen-
MWTP e
(7) _______e ∂ ∂ Δc e .
= ___ tal quality in favor of consumption at low lev-
MWTPs ___
s
∂ Δs els. As the budget constraint is relaxed, the
value of an additional unit of consumption
With this set-up, individuals’ decisions about falls and the trade-off between consump-
c, Δe, and s will produce the first-best or tion and environmental quality becomes less
socially efficient outcomes. extreme, increasing demand for the latter.
Even in this setting, it is instructive to For example, even if improvements in envi-
introduce a social planner whose job it is to ronmental quality lead to large, measurable
choose the level of environmental quality. health gains that improve the quality of life,
For the social planner to achieve the socially these improvements may still be small rela-
efficient outcome by aggregating across the tive to the utility gains (also possibly through
agents, it is necessary to assume that she improved health) from an increase in con-
has a costless technology to raise revenue, sumption. This explanation does not result
has no preferences of her own, and is able from any market failure and suggests that the
to observe true MWTPe . The next subsection low levels of environmental quality in devel-
considers reasons that environmental quality oping countries are socially efficient.
is so poor in developing countries, with three Richer models provide related expla-
of the explanations relying on violations of nations for poor environmental quality in
these assumptions. We postpone modeling developing countries. Consider an extension
the consequences of the classic externality of the simple model above to one with two
and public goods characteristics of environ- periods. If the probability of living to the next
mental quality until section 2.3. period in time is affected by environmental
quality or health, then M WTPe will increase
2.2 Four Explanations for Poor
with income. Because later-period consump-
Environmental Quality in Developing
tion is increasing in income and the marginal
Countries
utility of consumption is decreasing, MWTPe
will increase as income increases. Hall and
2.2.1 Explanation 1: High Marginal Utility
Jones (2007) show that this holds for health
of Consumption
spending, and the intuition extends to any
We begin with the straightforward expla- good that increases the probability of living
nation that MWTPe is low because people in to the next period, provided that utility is
Greenstone and Jack: Envirodevonomics: A Research Agenda for an Emerging Field 13
accounts for externalities and public goods, substantially between developed and devel-
so it reflects the first-best allocation and is oping countries, but externalities may be
defined by exacerbated by, and interact with, other mar-
ket failures that are more prevalent in devel-
∂ u + ___
nSP (___ ∂ u ___
∂ h
λ y ∂ e ∂ h ∂ e )
(9) SMWTPe ≡ ___ oping country settings, as described above.
For example, in settings with multiple mar-
∂ Δy ∂ Δy ___ ket failures, an otherwise efficient correction
+ n ____
+ n ____ ∂ h . (e.g., a Pigouvian tax) to one market failure
∂ e ∂ h ∂ e may be suboptimal in the presence of others
(e.g., weak property rights).
SMWTPe will exceed private M WTPe
(as defined in equation 5) if environmental
3. Existing Evidence
quality investments create public goods or
positive externalities, or if consumption gen- A long history of theoretical and empiri-
erates negative externalities.10 Specifically, cal economic research on environmental
each agent’s choice of Δe i affects the util- issues in developing countries has generated
ity of other agents (typically positively), and considerable insight into these issues.11 The
because of these externalities, agents do papers we describe in the remainder of the
not make socially efficient investments in text stand on the shoulders of these pioneers
Δe i. For example, a farming household (for example, Dasgupta and Mäler 1991,
does not enjoy the full environmental ben- 1997). Our scope is limited to a relatively
efits associated with a switch to less harm- recent set of papers that prioritize causal
ful pesticides, and underinvests in the new inference and/or open new areas of inquiry,
technology even if it also improves the and still our coverage is incomplete. Our
agent’s own experienced environmental review of this literature leads us to conclude
quality. Consumption externalities may also that a confluence of trends has created the
affect the environmental quality experi- conditions for a full-fledged field to bloom
enced by other agents (typically negatively). that is defined by the question of why envi-
Social MWTP for self-protection (SMWTPs ) ronmental quality is so poor in developing
will also diverge from private MWTPs if countries. Its position at the intersection of
self-protection generates negative externali- environmental and development economics
ties. For example, air conditioning protects an leads us to label this field envirodevonomics.
individual from poor environmental quality This section describes these trends and the
but creates negative externalities from energy opportunity they create. It then reviews the
consumption and greenhouse gas emissions. empirical evidence that attempts to quantify
Thus, individuals may overinvest in self-pro- the impacts of poor environmental quality,
tection relative to the first best allocation. as well as the existing evidence on the four
Generally, the public goods nature of potential explanations for the low levels of
environmental quality does not differ environmental quality in developing coun-
tries. Throughout, we highlight the areas
where the opportunities for new research
10 The social marginal utility of consumption is are greatest. We conclude this section with
∂ u + n(__
∂ u + __
∂ u __
∂ h __ ∂ a
∂ h ∂ e ) ∂ c
__
∂ c ∂ e
λ SP
y = ______________
, which diverges from the
1 − n(___ ∂ h )
∂ Δy ∂ Δy __
+ ___ ∂ a
__ 11 The journal Environmental and Development
∂ e ∂ h ∂ e ∂ c
private marginal utility of consumption because of negative Economics is devoted to the topic and a recent special issue
consumption externalities. (vol 19 no 3) reflects on the publication’s first twenty years.
Greenstone and Jack: Envirodevonomics: A Research Agenda for an Emerging Field 17
a review of the theory and macroeconomics and empirical work on the determinants and
research that is relevant to envirodevonom- effects of environmental quality in develop-
ics and continue to point out areas where ing countries. In section 3.4, we summarize
further work would be especially fruitful. some of these theoretical advancements and
highlight areas with substantial opportunities
3.1 A Promising Confluence of Trends
for innovation. The best empirical work will
We believe that a confluence of at least contribute to the identification of relevant
four different trends has created the con- parameters in the social planner’s maximiza-
ditions necessary for a field to emerge that tion problem, and therefore requires a clear
answers questions of great social and eco- theoretical underpinning of how revealed-
nomic importance at the intersection of envi- preference measures are tied to a concep-
ronment and development economics. First, tual parameter.13 As the studies below make
and perhaps most obviously, there is a grow- clear, the greatest progress has been made
ing recognition of the extreme levels of pol- in quantifying the impact of environmental
lution in many developing countries. As one ∂ h ), rather than the mar-
quality on health (__
example, measured Chinese and Indian air ∂ e
ginal willingness to pay for such a change. In
pollution concentrations exceed those ever addition, the vast majority of studies related
recorded in any other country, which may be to MWTPe have focused on health channels,
rather than direct effects (__ ∂ u ) or other
partially explained by the absence of reliable
monitoring equipment in the first half of the ∂ e
indirect channels such as income (
∂ e )
∂ Δy
twentieth century in the United States and ___
.
Europe, but still highlights the extraordinary Presumably, this is largely due to the chal-
levels of pollution faced by the citizens of lenges of measuring the relevant outcomes
these countries today.12 Relatedly, the liter- or of credibly identifying the effects of envi-
ature on the health effects of pollution have ronmental quality on nonhealth parameters,
advanced greatly in the last two decades though continued theoretical advancements
although almost all of this research has been will facilitate new and better empirical
conducted in developed country settings, tests. Though our focus in this paper is on
where pollution levels are less extreme (for microeconomic topics and research, we also
example, Chay and Greenstone 2003; Currie use section 3.4 to highlight the emerging
and Neidell 2005; Currie and Walker 2011). macroeconomic work on growth and the
At least in terms of human health, the stakes environment.
appear high. Third, a number of breakthroughs in
Second, advances in modeling individual measurement and access provide research-
and firm behavior open the door to the esti- ers with new developing country datasets.
mation of parameters with a clear economic Many countries, including China and India
interpretation. Many of these advancements (see, for example, Greenstone and Hanna
have occurred outside of the development 2014), are beginning to open up to eco-
or environmental literatures, but lend nomic research on environmental quality.
themselves well to theoretical application
Technology, such as satellite imagery, allows
researchers to bypass local data collection
obstacles, which can be important where
12 See relevant articles in a recent Review of
Environmental Economics and Policy symposium on envi-
ronmental issues in China (e.g., Vennemo et al. 2009; Cao,
Garbaccio, and Ho 2009) and a recent paper on the geog- 13 Potentially, but by no means exclusively, the ones laid
raphy of pollution in China (Zheng et al. 2014). out in section 2.
18 Journal of Economic Literature, Vol. LIII (March 2015)
government officials have incentives to dis- 3.2 Quantifying the Impact of Poor
tort environmental data (e.g., Chen et al. Environmental Quality
2012 on gaming of air q uality data in China).
Indeed, a growing number of papers rely Pollution levels in developing countries
on satellite imagery to measure outcomes often exceed the standards set by regulators
including pollution (Foster, Gutierrez, in developed countries and by global health
and Kumar 2009; Jayachandran 2009), recommendations (as shown in figure 1).
deforestation (Burgess et al. 2012; Alix- Whether higher pollution levels translate
Garcia et al. 2013), and economic activity into worse health outcomes depends both
(Henderson, Storeygard, and Weil 2011). on the shape of the dose-response curve and
Others use geographic information system also behavioral adaptations to the high levels
maps to generate instruments for devel- of pollution, such as staying indoors on heav-
opment outcomes (Duflo and Pande 2007; ily polluted days and the purchase of air puri-
Dinkelman 2011; Lipscomb, Mobarak, and fiers. An emerging body of evidence focuses
Barham 2013). on the relationship between pollution and
Fourth, there is a greater appreciation health outcomes in developing countries,
of the opportunities for conducting studies and suggests that poor environmental quality
based on quasi-experimental and true experi- leads to sicker and shorter lives.14
mental variation in key parameters. Just as in Many of the relevant studies focus on the
developed countries, many developing coun- health impacts of air pollution (__ ∂ h ). For
tries implement policies in seemingly arbi- ∂ e
trary ways or use discrete rules to determine example, Almond et al. (2009) and Chen et
eligibility, which facilitates credible policy al. (2013) use the geographic discontinu-
evaluation. Examples include China’s Huai ity created by a Chinese policy to subsidize
River policy, which provided free coal for coal north of the Huai River to estimate a
winter heating to the north of the river but significant increase in total suspended par-
forbade winter heating with coal in the south ticulate matter (TSP). TSP has a large effect
(Chen et al. 2013), and Mexico’s appliance on mortality rates: increasing the long-term
buy-back program, which set subsidy levels exposure to TSP by 100 μg/m3 (or around
based on past energy consumption thresholds half a standard deviation) is associated with
(Davis, Fuchs, and Gertler forthcoming). In a decrease of three years of life expectancy.
addition, there has been a near explosion of This estimate is larger than the effect mea-
studies in developing countries that demon- sured in developed countries and is more
strate the power of randomized control trials than five times the conventional OLS esti-
or field experiments to identify key parame- mate. Because of China’s extreme policies
ters or relationships (e.g., Berry, Fischer, and on migration during the period of study, the
Guiteras 2011; Duflo et al. 2013, 2014; Jack authors are able to study pollution effects
2013; Kremer et al. 2011; Miller and Mobarak on life expectancy, which is typically diffi-
2013; Bennear et al. 2013, and others still in cult to identify given migration and other
progress). Experiments cannot, of course, self-selection responses to long run pollution
answer many important questions in environ- exposure.
mental and development economics, yet they
serve as a complement to other empirical
methods that may be better suited to under- 14 See also the review article by Pattanayak et al. (2009),
standing problems such as climate change which focuses on four environmental health challenges in
(Dominici, Greenstone, and Sunstein 2014). developing countries.
Greenstone and Jack: Envirodevonomics: A Research Agenda for an Emerging Field 19
Table 1
Health Impacts of Pollution on Environmental Quality
Notes: Summary of empirical findings on the impact of pollution on environmental quality. Pollutants are abbreviated as fol-
lows: Particulate matter (PM), carbon monoxide (CO), total suspended particulate (TSP), sulfur dioxide (SO 2). RCT refers
to a randomized controlled trial. Findings are as reported in the paper cited in the rightmost column. A lack of the relevant
information for a number of the studies precludes the translation of the health impacts into elasticities.
Other papers in this literature focus on that a 1 μg/m3 increase in particulate matter
infant and child health outcomes to mitigate increases infant deaths per 100,000 by 0.24,
concerns about unmeasured lifetime expo- while a 1 part per billion increase in carbon
sure. Jayachandran (2009) takes an innovative monoxide increases infant deaths per 100,000
approach to measuring the “missing children” by 0.0032 per week. Comparing their results
associated with extreme pollution exposure to other studies, they find that—in elasticity
during infancy and in utero due to forest terms—carbon monoxide has a larger effect
fires in Indonesia. She uses satellite aerosol on infant mortality in Mexico than in the
monitoring data to show that the particulate United States, while the impact of particu-
matter emitted by the fires led to a reduction late matter on infant mortality is similar to or
in the size of the exposed birth cohort by 1.2 smaller than in the United States.
percent, an effect that is largely explained by A number of papers have also documented
prenatal exposure. A recent study by Arceo- serious health impacts from poor water
Gomez, Hanna, and Oliva (2012) finds that quality in developing countries. For exam-
exposure to carbon monoxide and particulate ple, Ebenstein (2012) studies the effects of
matter increases infant mortality in Mexico industrial water pollution on stomach can-
City. They use the frequency of thermal cer in China. Like in Chen et al. (2013),
inversions, which trap pollutants close to the historical restrictions on mobility in China
ground, as a source of plausibly exogenous make it possible to estimate long-run health
variation in pollution exposure. Looking at outcomes with diminished concerns about
exposure over the previous week, they find selection. Specifically, Ebenstein (2012)
20 Journal of Economic Literature, Vol. LIII (March 2015)
finds that a decline in water quality by one effect of mining on the residuals of produc-
grade (on a scale of 1 to 6) is associated with tivity, which is shown to correspond to an
a 9.7 percent increase in deaths from diges- 18 percent increase in rural poverty. The
tive cancers. empirical strategy relies on geo-referenced
Other studies look at the child health data on the location of mines and satellite
outcomes of poor water quality. Field, imagery to measure air pollution (nitrogen
Glennerster, and Hussam (2011) find when dioxide). Second, some evidence suggests
households in Bangladesh switched from direct income effects from improved envi-
deep wells to surface wells contaminated ronmental quality associated with ecotour-
with fecal bacteria, infant and child mortal- ism. Sims (2010) uses placement rules for
ity from diarrhea disease increased by 27 protected areas in Thailand, together with
percent. Brainerd and Menon (2014) use satellite data and survey measures of pov-
the seasonality associated with the appli- erty, to document both a significant impact
cation of fertilizers across different crops of protection status on deforestation and a
and regions of India to identify significant positive impact on consumption and poverty
effects on numerous infant and child health reduction.15 In this case, the channel appears
outcomes including infant mortality, neo- to be increased local revenue from tourism.
natal mortality, height-for-age z-scores, and Indirect effects have been documented
via productivity and health (i.e., ___
∂ h ∂ e )
∂ Δy __
weight-for-age z-scores. In their data, the
∂ h .
effects are strongest in low socioeconomic
status households. Kremer et al. (2011) find Using a difference in difference strat-
that a spring protection investment in Kenya egy, with additional variation in exposure
reduced fecal contamination by 66 percent, from seasonal wind patterns, Hanna and
which led to a reduction in child diarrhea Oliva (forthcoming) examine the effect of a
of 25 percent. Duflo et al. (2014) find that decrease in pollution resulting from the clo-
providing communal water tanks and private sure of a refinery in Mexico City. They find
bathing facilities and toilets to households that a one percent decrease in sulfur diox-
decreases severe episodes of diarrhea by ide concentrations increased labor supply
30–50 percent over the long run. by 0.61 hours per week. They present sug-
As illustrated in the conceptual frame- gestive evidence that the effects are driven
work, better environmental quality may also by child health, which affects parental labor
have direct or indirect effects on income. supply. Pitt, Rosenzweig, and Hassan (2012)
While these relationships may be more find evidence that arsenic exposure low-
difficult to quantify than the direct health ers cognition and results in lower schooling
impacts, they are important for the devel- attainment using biological measures of arse-
opment of a comprehensive measure of the nic (i.e., toenail clippings) and variation asso-
benefits of improved environmental quality. ciated with genetic predisposition to store
To the best of our knowledge, only two stud- arsenic in the body. Kremer et al. (2011)
ies credibly show more direct effects of envi- also measure the effect of pollution on pro-
∂ Δy ductivity in their study of springs in Kenya
ronmental quality on income (i.e., ___
> 0).
∂ e and find that improvements to water quality
First, Aragón and Rud (2013) show that
the pollution associated with gold mining 15 Using matching methods, Andam et al. (2010) provide
in Ghana has a negative impact on income additional evidence from both Costa Rica and Thailand on
the poverty alleviation of protected areas. For a broader
from agriculture. They estimate an agricul- review of empirical work on forests and forest conservation
tural production function and analyze the in developing countries, see Ferraro et al. (2012).
Greenstone and Jack: Envirodevonomics: A Research Agenda for an Emerging Field 21
did not increase school attendance among States.17 In addition, the same households’
primary-school children. revealed MWTPe is substantially lower than
Taken as a whole, these papers indicate the valuations they give in a contingent valu-
that the health burden of air and water pol- ation survey. Other studies that measure the
lution in developing countries are substan- MWTP for health in developing countries
tial and that the productivity and income may offer estimates of the appropriate valua-
effects may also be important. There is still ∂ u . The empirical literature
tion to assign to __
∂ h
a great deal to learn about dose-response valuing improvements to health in develop-
functions, the distribution of impacts, and ing countries is reviewed by Dupas (2011)
the nonhealth outcomes. All of these are and is consistent with measured MWTPwell
promising areas for future research, with the below that in developed countries.
greatest contribution likely to come from Overall, what little evidence we have
studies that go beyond quantifying the health clearly suggests that observed levels of
impacts to investigate both economic conse- MWTPe are low. It is an open question as
quences and underlying mechanisms. to whether the low measured MWTPe in
these studies reveals a M WTPe that would
3.2.1 Willingness to Pay for Environmental
be low even if well-functioning markets or
Quality
if market failures reduce measured M WTPe
A high health burden from environmen- relative to its “true” value, as discussed in
tal quality in developing countries does not section 2.2.4. For example, the paucity of
directly imply a high MWTPe . Instead, the hedonic studies in developing countries
studies reviewed in the previous subsection could reflect imperfections in the land and
∂ Δy
∂ h or ___
aim to describe __ ∂ h . However, the
__ labor markets that make them ill-suited for
∂ e ∂ h ∂ e inference about MWTPe (although it surely
MWTPe associated with these changes is the also reflects the limited availability of high
relevant parameter for the social planner quality land price and labor market data).
(see equation 9) and central ingredient in the Even the Kremer et al. (2011) study relies
determination of optimal policy. on market wages to convert walking time to
Few studies attempt to develop revealed monetary units, in a setting where labor mar-
preference estimates of M WTPe .16 Kremer kets are highly imperfect. Our judgment is
et al. (2011) uses a randomized controlled that there is hardly a more important topic
trial to generate exogenous variation in water for future study than developing revealed
quality across springs in Western Kenya. preference measures of M WTPe that capture
They find that households are only willing to the aesthetic, health, and/or income gains
pay about $11 per year for clean water, where from environmental quality.
MWTPe is calculated from rural wage rates
and revealed willingness to walk to clean
water. This translates into a revealed pref-
erence value of a statistical life of USD2013
860, which is four orders of magnitude lower 17 Other estimates of VSL numbers for developing
than accepted VSL numbers in the United countries are highly variable depending on the methodol-
ogy employed (see Viscusi and Aldy 2003 for a summary
and León and Miguel 2012 and Bhattacharya, Alberini,
and Cropper 2007 for examples of different methods).
Many revealed preference studies of VSL rely on hedonic
16 The well-identified empirical literature on willing- regressions to isolate the risk–price trade-off, which relies
ness to pay for environmental quality in developed coun- on assumptions of well functioning markets that may not
tries is also extremely sparse. hold in developing country settings.
22 Journal of Economic Literature, Vol. LIII (March 2015)
3.2.2 Willingness to Pay for Self-Protection yet neither measures MWTPe for the ulti-
mate change in environmental quality or
A larger number of studies measure MWTP for health outcomes or beliefs about
WTPs , which identifies M
M WTPs as defined how the technology will affect health.
in equation (6) under the assumption that
3.3 Existing Evidence on the Explanations
market failures do not bias valuations. Berry,
for Poor Environmental Quality
Fischer, and Guiteras (2011) observe a
median willingness to pay for a water filter The limited evidence on MWTPe in devel-
in Ghana that ranges from $1.80 to $2.40, oping countries suggests extremely low valu-
depending on how the valuation was elic- ations. While more evidence is needed, this
ited. Overall, the filter appears to reduce finding raises questions about the underlying
self-reported diarrheal incidents after one
causes. In addition, while low M WTPe is a
month of use by 8 to 14 percent. They find sufficient condition for poor environmental
only weak evidence of a relationship between quality in developing countries, it is not the
measured demand and self-reported diar- only possible explanation. Next, we turn to
rheal reductions from the filter. Strategies the evidence on each of four possible expla-
for self-protection are likely to be diverse nations for poor environmental quality in
and more evidence on the potentially costly developing countries, including why M WTPe
choices that households make in the face might be low.
of pollution, such as adjustments to migra-
3.3.1 Explanation 1: High Marginal Utility
tion or fertility decisions, will help quantify
of Consumption
households’ M WTPs to avoid pollution.
MWTPs may also vary within the house- Recall from equations 5 and 9 that
hold if the health burden of poor environ- WTPe is determined not only by the
M
mental quality is unevenly distributed across endowment of environmental quality, but
household members. Miller and Mobarak also by the marginal utility of income.
(2013) show that women in Bangladesh have Increasing income or wealth may increase
a stronger preference for low-emissions MWTPe as the marginal utility of consump-
cookstoves than do men. In the study setting, tion decreases. An ideal experiment would
men often manage money and women are measure how MWTPe changes with an exog-
liquidity constrained, such that many women enous change in income. However, most
who express a preference for the low-emis- experiments generate only short-run shifts
sions stove are unable to purchase it, while in income, and it may be challenging to find
men are more likely to implement their a quasi-experimental design that credibly
stated preferences. Pitt, Rosenzweig, and identifies the effects of permanent income
Hassan (2010) also find meaningful gender shocks. Evidence on the individual-level
differences in the health impacts of indoor income–MWTPe relationship is extremely
air pollution and show that status within the scarce in both developing and developed
household is an important determinant of countries.
exposure. Thus, household-level measures Hanna and Oliva (2014) offer some direct
of MWTPs may not accurately capture with- evidence by examining the fuel choices of
in-household heterogeneities. Both Berry, households in India following the random-
Fischer, and Guiteras (2011) and Miller and ized roll out of a transfer program that had
Mobarak (2013) measure marginal willing- measurable effects on income and assets.
ness to pay for a (unfamiliar) product that While energy use by treated households
affects experienced environmental quality, increases substantially, it does not become
Greenstone and Jack: Envirodevonomics: A Research Agenda for an Emerging Field 23
much cleaner; the shift from dirtier ker- in income to estimate these impacts is gen-
osene to cleaner electricity is offset by an erated through a randomized cash trans-
overall increase in kerosene purchases, and fer program in Mexico in the studies by
the cooking technologies that are respon- Alix-Garcia et al. (2013) and Gertler et al.
sible for most indoor air pollution do not (2013). A lix-Garcia et al. (2013) find that
improve. Other studies that are relevant the additional income associated with the
to the relationship between income and Oportunidades program increased defor-
MWTPe or MWTPs provide mixed correla- estation due to higher consumption of
tional evidence. Jalan and Somanathan land-intensive goods, such as beef. Their
(2008) document higher expenditures on study implements a community level regres-
water quality among households in urban sion discontinuity design based on the
India with more assets. Consistent with rules that determine program eligibility to
this result, Kremer et al. (2011) find that study deforestation as measured by satellite
households with more education or assets images. They combine this community-level
are more willing to walk to access improved analysis with household survey measures
water sources. On the other hand, neither of consumption from the randomized pilot
Berry, Fischer, and Guiteras (2011) nor phase of the program.
Ashraf, Berry, and Shapiro (2010) find a Gertler et al. (2013) use a different source
correlation between revealed willingness to of identification under the same program
pay for water purification technologies and to find a similar impact on environmen-
assets. The relationship between income and tally harmful consumption. In their case,
MWTPs or M
WTPe in all of these studies is the outcome of interest is the purchase of
correlational, and further research on the energy-intensive durable goods, specifically
topic will help determine whether the high refrigerators, and they use a combination of
marginal utility of consumption that accom- the random variation of when communities
panies very low incomes drives low private were phased in to the program and house-
and social M WTPe .18 hold-level variation in the income flow due
As suggested by the conceptual frame- to household structure.19 They find that
work, higher incomes might be associated the income shock of the transfer increased
with both higher MWTPe and also larger refrigerator purchases, and therefore house-
impacts on the environment ( ∂ a > 0).
if __ hold energy consumption, with the larg-
∂ c est effects for households that received the
Several recent papers provide evidence that
income over a short period of time. They use
as incomes in developing countries increase,
these empirical results to simulate differ-
there is a negative causal impact on environ-
ent development pathways, and provide the
mental quality. Plausibly e xogenous variation
insight that economic growth that benefits
the poor increases energy consumption and
18 A stated preferences study by Israel and Levinson
the related negative externalities more than
(2004) uses a World Values Survey question to correlate less progressive patterns of growth. These
individual-level income and stated M WTPe across and findings do not, of course, directly imply that
within countries. They use the survey data to test different households have low M WTPe or that aggre-
mechanisms underlying the Environmental Kuznets Curve
hypothesis (the inverted-U shaped relationship between gate MWTPe does not increase with income.
GDP and environmental quality) and find a strong cor- They do, however, highlight the importance
relation between income and marginal stated MWTPs but
no systematic relationship between per capita GDP and
stated MWTPs . Country fixed effects explain a large share 19 The transfer amount differed by gender and grade of
of the variation. school-aged children.
24 Journal of Economic Literature, Vol. LIII (March 2015)
Table 2
Evidence for High Marginal Costs
Brazil Decentralization increases water pollution Fixed effects Lipscomb and Mobarak
(2011)
Mexico Policy loopholes undermine effectiveness Temporal discontinuity Davis (2008)
Mexico Voluntary certification lowers regulatory Structural identification Foster and Guiterrez
costs (2012)
Mexico Large inframarginal payments lower policy Fixed effects, RD Davis et al.
impacts (forthcoming),
Boomhower and Davis
(2014)
Bangladesh Policy has large unintended consequences Quasi-experiment Field et al. (2011)
Philippines Public and private provision are substitutes Fixed effects, IV Bennett (2012)
India Public support improves the effectiveness of Fixed effects Greenstone and Hanna
environmental policies (2014)
Notes: Summary of empirical findings on the marginal costs of environmental policies in developing countries. Further
details on the studies are described in the text.
tality. The paper also highlights the chal- increased, presumably because consumers
lenge of multiple e nvironmental risks, and could bypass the regulation by purchasing
an inability of households or policymakers an additional car with different plates.21 In
to accurately rank them. Szabo (2014) pro- Mexico City, implementation of the policy
vides another clear example of suboptimal and the costs to households are estimated at
policy making. Using a very different meth- over 300 million dollars annually and gen-
odological approach, she examines the wel- erated no improvements in environmental
fare implications of South Africa’s Free Basic quality. Policy implementers also have incen-
Water policy. Rich household panel data, tives to exploit loopholes in policy design.
together with variation in nonlinear tariffs Duflo et al. (2013) study the performance
across households and over time, identifies a of third-party auditors used by the environ-
structural demand model. Her results imply mental regulator in Gujarat, India to identify
that, without a loss of revenue, the social industrial plants that exceed the emissions
planner could reallocate the existing subsidy standards. When plants choose and pay the
to increase welfare. auditors, as is the norm, they find that the
Unanticipated effects of environmen- system is corrupted. Auditors systematically
tal regulations also arise through agents’
responses to the policies. For example,
Davis (2008) finds that a policy to restrict 21 Similar driving restrictions have been introduced
driving in Mexico City according to license and evaluated in other cities including Beijing, Santiago,
and São Paulo, with mixed success (Gallego, Montero, and
plate number had no effect on pollution lev- Salas 2013; Lin, Zhang, and Umanskaya 2011; Viard and
els. Instead, the number of registered cars Fu 2011).
26 Journal of Economic Literature, Vol. LIII (March 2015)
reported plant emissions just below the mobile. On the other hand, Kahn, Li, and
standard, although actual emissions were Zhao (forthcoming) provide clear evidence
typically higher. The social costs of these that interjurisdictional spillovers also depend
policy-making oversights are large. on incentives set further up the bureaucratic
Poor targeting also increases the marginal chain. They exploit a change in the rules for
cost of environmental quality improvements. promotion of local officials in China that
Another program in Mexico that subsidized reward pollution reductions along admin-
the purchase of energy-efficient appliances istrative boundaries. The incentives set by
performed very poorly, actually increasing the centralized government make progress
energy consumption for some appliances, toward internalizing the interjurisdictional
while ex ante engineering estimates indi- externalities. For pollutants that cross inter-
cated that the program would pay for itself national boundaries, the problem is more
in energy savings (Davis, Fuchs, and Gertler complex, and we return to it in the discus-
forthcoming). The extreme overestimate sion of climate change in section 4.
of the program’s effectiveness was due to a Other investments in the economy, by
combination of changes in the types of appli- policymakers, firms, and individuals, affect
ances purchased, usage patterns, and the the marginal cost of improving environmen-
inclusion of many inframarginal households tal quality, and privately optimal investment
(Boomhower and Davis 2014; Davis, Fuchs, choices change in response to new envi-
and Gertler forthcoming). ronmental regulations or improvements in
In settings with low enforcement capacity environmental quality. For example, the
and heterogeneous benefits from environ- marginal costs of improving environmen-
mental quality, decentralization may pose tal quality may appear higher to the social
a useful solution for improving monitoring planner if individual responses to public
and accountability (Bardhan 2002; Lemos investments are taken into consideration.
and Agrawal 2006). Decentralization of poli- While public (or privatized) provision itself
cies for environmental quality may therefore may improve efficiency, it will not improve
improve local welfare, while generating an experienced environmental quality if com-
incentive to divert pollution to neighboring pensatory changes in behavior reduce pri-
jurisdictions. Sigman (2002) analyzes this vate investments in self-protection. Bennett
problem with respect to international water (2012) studies such a case in the Philippines
pollution and shows that pollution levels and argues that improved water supply infra-
are elevated in rivers upstream of an inter- structure resulted in lower private sanitation
national boundary. Lipscomb and Mobarak investments. Consistent with this finding,
(2011) analyze a similar problem across Berry, Fischer, and Guiteras (2011) show
counties in Brazil. They use changes in dis- a negative correlation between access to
trict boundaries to generate evidence from an improved water source year-round and
Brazil that, within a jurisdiction, pollution measured willingness to pay for an in-home
levels are highest where the river is close to water filter. This suggestive evidence on the
entering a downstream jurisdiction. Without substitutability of public and private invest-
centralized enforcement capacity, these ments does not imply that public invest-
types of environmental externalities may go ments are inefficient, just that the marginal
unpenalized. While decentralization may costs may be underestimated if the relation-
lower the costs of improving environmen- ship between public and private investments
tal quality locally, it may simply shift these is ignored. Further research on the inter-
costs to other jurisdictions if pollutants are action between public and private resource
Greenstone and Jack: Envirodevonomics: A Research Agenda for an Emerging Field 27
to a voluntary audit in exchange for a two both structural and reduced form analyses.
year inspection exemption. This allowed the The results suggest that at least 9.6 percent
regulator to better target inspections, leading of old-car owners paid bribes of $20 to cir-
to pollution reductions by uncertified firms. cumvent the regulations.
These findings suggest that policies that offer Among elected officials, rent-seeking
some information revelation component can opportunities can also undermine policy
improve how regulations are targeted, which implementation. Compelling evidence to this
may be particularly important for cost effec- effect is presented by Burgess et al. (2012),
tive environmental quality improvements who find that corruption increases deforesta-
where monitoring and enforcement capacity tion in Indonesia. A combination of satellite
is weak. imagery, data on electoral cycles, oil prices,
and district boundaries are used to show that
3.3.3 Explanation 3: Political Economy and
in years when oil revenues are low, illegal
Rent-Seeking Behavior
logging increases. Enforcement of environ-
The conceptual framework distinguishes mental policies (in this case, forest laws) may
between high marginal costs associated therefore suffer when natural resources are
with weak capacity or missing infrastructure viewed as a source of rents. More broadly,
and political economy factors that distort if taxation provides an opportunity for rent
the social planner’s maximization problem. seeking (and it does, according to Olken and
Empirical studies on the effects of political Pande’s 2012 review of the literature on cor-
economy considerations and rent-seeking ruption), then the aggregation of individual
behavior on environmental quality provide preferences will be distorted. On the other
estimates of these distortions, which may hand, new evidence indicates that improving
take the form of an additional parameter incentives for bureaucrats can both reduce
in the welfare equation. For example, the corruption and improve monitoring and
policymaker’s and/or bureaucrat’s own util- enforcement outcomes in developing coun-
ity may enter the function that determines tries (for example, Duflo et al. 2013). It is
policy, or there may be unequal welfare apparent that, at least in some settings, part
weights assigned to specific groups. The sup- of the cost of environmental quality improve-
ply or withholding of environmental qual- ment is therefore increased by rent seeking
ity by the social planner may be one form and other corrupt behavior.
of affecting distributional outcomes across Firms may respond to regulatory settings
groups, as shown by Feler and Henderson distorted by opportunities for rent capture
(2011) for water connections in Brazil. On by modifying their objective function in ways
the other hand, utility weights in the social that undermine competitiveness and reduce
planner’s function may also help correct for environmental quality. For example, McRae
an unequal burden of poor environmental (2015) describes the impact of government
quality in developing countries. subsidies for public services, sewerage, and
There is an emerging empirical literature electricity, which undermines firm incen-
that finds evidence for rent seeking as an tives to improve infrastructure quality in
explanation for poor environmental quality. Colombia. Because firms receive transfers
Oliva (forthcoming) studies a pollution con- from the government in lieu of payments
trol policy in Mexico City and finds extensive from customers, they have little incentive
corruption in the smog emissions testing to improve infrastructure (and therefore
program for private vehicles. She uses data cost recovery) themselves. The “grabbing
from smog testing centers to implement hand” model of government suggests that
Greenstone and Jack: Envirodevonomics: A Research Agenda for an Emerging Field 29
Table 3
Evidence for Political Economy Distortions to Environmental Policy
Notes: Summary of empirical findings on political economy and rent seeking. Greater detail on the studies is provided in
the text.
corruption and state owned enterprises indoor air pollution and poor water quality (as
discourage entrepreneurship and distort firm discussed in section 3.2.2). And within many
entry and exit decisions (Shleifer and Vishny developing country settings, poor and margin-
2002). Some of the low observed total factor alized groups are often described as depen-
productivity among firms in countries such dent on natural resources, either as a primary
as India and China may be a direct result of source of income or for consumption smooth-
government intervention or other market ing (Hassan, Scholes, and Ash 2005). Thus,
failures that distort resources away from the the burden of poor environmental quality
most productive firms (Hsieh and Klenow may fall disproportionately on these already
2009). When rent-seeking politicians set vulnerable groups. Whether such outcomes
regulation, they may therefore undermine are avoided may depend on explicit weights
competition that would help increase the in the social welfare function, which carry
efficiency and lower the pollution intensity their own political economy considerations.24
of the average firm.23 Further research on the causal relationship
While utility weights in the social planner’s between environmental quality, social status,
maximization problem are most often associ- and economic vulnerability will help inform
ated with handouts and political favors (e.g., open questions about the incidence of the
Fisman 2001), they may also serve the func- pollution burden in developing countries.
tion of correcting socially undesirable distri-
butions of the burden of poor environmental 3.3.4 Explanation 4: Market Failures and
quality. Within the household, women and Behavioral Biases
children are more likely to bear the cost of
Information, credit, risk, or land/prop-
23 Ryan (2014) shows related evidence that reduced
erty rights market failures may distort mea-
competition harms electricity provision in India. Dasgupta, sured MWTPe away from its value in a world
Lucas, and Wheeler (1998) show evidence from Brazil and
Mexico that small firms are both more prevalent and have
much higher pollution levels in poorer regions, though 24 See, for example, the extensive discussion of indige-
these results can be partially explained by the sectors in nous rights to resource use in the UN’s State of the World’s
which small firms operate. Indigenous People (UN 2009).
30 Journal of Economic Literature, Vol. LIII (March 2015)
Table 4
Evidence on Market Failure-Based Distortions to MWTP
Property rights Rwanda Formal titling increased soil Spatial discontinuity Ali et al. (2014)
conservation
Information India Information increased investments RCT Jalan and Somanathan
in water filters (2008)
Information Bangladesh Information increased switching to Quasi-experiment Madajewicz et al.
clean water sources (2007)
Information Bangladesh Information content determined RCT Bennear et al. (2013)
effects on behavior
Credit Bangladesh Access to credit increases WTP for RCT Guiteras et al. (2014)
self protection
Labor No empirical evidence
Risk No empirical evidence
Notes: Summary of empirical findings on the effects of market failures on MWTP for environmental quality, classified
according to the market failures described in the text.
free from market failures (as in equation 5). show that intensive information designed to
Empirical evidence on the topic is scarce, generate social pressure as well as awareness
though the potential to evaluate the environ- increased latrine adoption in Orissa, India.
mental effects of interventions that improve Ashraf, Jack, and Kamenica (2013) focus
markets in developing countries is high. instead on information about the self-pro-
Some evidence suggests that informa- tection technology. They offer an unfamil-
tion provision can have substantial effects iar water purification solution at randomly
on measured MWTPe and on investments varied prices to urban consumers in Zambia
in self-protection, which implies that infor- and observe that households are more price
mation market failures may distort revealed sensitive when they have more information
preference measures. In a randomized about the product. In the absence of other
experiment, Jalan and Somanathan (2008) sources of information, households may
provide Delhi residents with information infer product quality from price. Another
about the quality of their tap water and find study in Bangladesh suggests that the effect
a significant change in expenditures follow- of information is sensitive to how it is deliv-
ing the treatment. Individuals who learn that ered (Bennear et al. 2013). The authors show
their water is dirty are 11 percentage points that providing households with complicated
more likely to purchase in-home treatment information about well contamination leads
and overall expenditures on in-home treat- to suboptimal decisions in water choice.
ment increase by 6.5 percent. Madajewicz These studies relate to a large literature in
et al. (2007) also find a significant response development economics that describes the
to information about water quality in information and learning challenges associ-
Bangladesh, while Pattanayak et al. (2009) ated with technology adoption (for a review of
Greenstone and Jack: Envirodevonomics: A Research Agenda for an Emerging Field 31
the technology adoption literature, see Foster the same time, credit constraints may inhibit
and Rosenzweig 2010). A household that has investment in environmentally damaging
never experienced clean water may not know production and consumption. For example,
the benefits of experimenting with technolo- Assunção et al. (2013) show that a restriction
gies or behaviors that improve water quality. on credit in Brazil lowered deforestation rates,
To the extent that peers and neighbors offer likely by decreasing land-intensive livestock
transferrable information through their own investments. These results are consistent with
actions, social learning is more likely to occur Alix-Garcia et al.’s (2013) finding that income
(Foster and Rosenzweig 1995; Conley and transfers under Oportunidades led to greater
Udry 2010). A trusted government agency deforestation in Mexico.
may be able to help overcome information Because land, risk, and capital market fail-
failures in developing countries. However, ures are prevalent in developing countries,
individuals may not trust official information more evidence is needed about how these con-
sources—in some cases, for good reason, as straints affect the measurement of MWTPe .
shown in the study by Field, Glennerster, and In particular, the size and direction of the gap
Hussam (2011) described above. One rea- between measured MWTPe and MWTPe in
son for low measured MWTPe in developing the absence of these constraints, and the costs
countries may therefore be a lack of clear and of narrowing the gap, are necessary policy
trustworthy information about the benefits of inputs. The substantial literature in develop-
improved environmental quality. ment economics documenting plausibly exog-
Evidence on the effects of market failures enous sources of variation in market function
other than information are less common. In should help facilitate future research.
one existing study in Rwanda, Ali, Deininger, The behavioral implications of defaults
and Goldstein (2011) use a spatial regression and cognitive depletion associated with the
discontinuity design to measure the impacts constant active decisions that residents of
of a land titling program on a number of out- developing countries must make to raise
comes, including investments in soil fertil- their experienced environmental quality
ity. They show that more secure land titles may partly explain low revealed MWTPe .
increased investments in environmental qual- Though studies of the behavioral effects
ity, particularly among female-headed house- of defaults have proliferated in other con-
holds, for whom the change in tenure security texts, they have received relatively little
was likely to be most dramatic. attention for environmental decision mak-
Missing capital markets may also reduce ing.25 Devoto et al. (2012) use a randomized
incentives for investments with long-run experiment to study the take-up and impacts
payoffs, which includes many environmental of a subsidized program to connect house-
quality investments. Guiteras et al. (2014) holds to the municipal water system in
randomly introduce different types of Morocco. They find that small administra-
credit for water purification filters in tive barriers have a large impact on take-up,
Bangladesh and document a positive rela- and that households who received tap water
tionship between measured willingness to experience a substantial improvement in
pay for a filter and credit availability. Their quality of life and mental health, in spite of
study offers clear evidence on the distor- the lack of health impacts. Also related to
tion to revealed preference measures of
MWTPe that emerge from settings with miss- 25 A developed-country exception is Lofgren et al.
ing credit markets for investments with long- (2012), who study the effect of defaults on the purchase
run payoffs or largely nonmonetary benefits. At of carbon offsets.
32 Journal of Economic Literature, Vol. LIII (March 2015)
demand for clean water, Ashraf, Berry, and of theory and methods from industrial eco-
Shapiro (2010) provide evidence against nomics to questions of optimal environmental
the sunk cost fallacy associated with paying regulation and pollution control (e.g., Fowlie
for self-protection. This study falls into the 2009, 2010; Ryan 2012). Extending this work
growing literature on measuring M WTP for to developing countries, where contracts are
health investments in developing countries difficult to enforce, information asymmetries
(Dupas 2011), some of which highlights are potentially large, and corruption affects
behavioral biases. Behavioral biases and psy- the entry and exit of firms, offers rich grounds
chological factors offer a promising direction for new innovation in applied theory. Finally,
for future research, which can help identify behavioral economics continues to generate
the gap between measured M WTPe and new theories relevant to individual deci-
MWTPe in the absence of market failures. sion making in developing countries (e.g.,
Bertrand, Mullainathan, and Shafir 2004;
3.4 Theory and Macroeconomics
Banerjee and Mullainathan 2010; Bryan,
The majority of the papers reviewed thus Karlan, and Nelson 2010). Environmental
far are solidly within the realm of empirical applications contain many of the ingredients
applied microeconomics. However, as we for interesting behavioral theory—low prob-
suggest in section 3.1, new developments ability outcomes, social spillovers, and future
in theory and modeling facilitate empirical impacts—and more complete descriptions
tests at the intersection of environmental and of individual and household decision mak-
development economics. At the same time, ing will help the field of environmental and
the challenges laid out in this paper present development economics progress.
opportunities for new applied theory, which Additionally, there is a small but growing
will help advance understanding of the chal- and insightful macroeconomics literature at
lenges at the core of this nascent subfield. the intersection of environment and devel-
An exhaustive review of theoretical prog- opment. Recent papers by Fischer and
ress relevant to environmental quality in Heutel (2013) and Smith (2012) provide
developing countries is well beyond the an overview of recent work and potential
scope of this paper. Instead, we highlight research directions at the intersection of tra-
three areas where new applied theory is ditional macroeconomics and environmen-
emerging or may be particularly important tal economics. Fischer and Huetel describe
for advancing our understanding of environ- a nascent literature on economic growth
ment and development issues. First, as sec- that considers induced innovation and path
tion 2.2.4 makes clear, the market failures of dependency for environmental technolo-
environmental economics may interact with gies. For example, recent theoretical work
the market failures of development econom- by Acemoglu et al. (2012) and Hemous
ics. While substantial theoretical literatures (2013) considers the effects of carbon taxes
describe each in isolation (see Stiglitz 1989; and research subsidies on innovation in a
de Janvry, Fafchamps, and Sadoulet 1991 single-economy or two-economy model,
for seminal works in development econom- respectively. The m odels presented in both
ics and Baumol and Oates 1988; Cornes and papers find support for the use of research
Sandler 1996 for environmental economics), subsidies in clean sectors, particularly when
the implications of these interactions require subsidies in a rich country generate tech-
additional modeling to derive specific predic- nological spillovers for a poor country. A
tions. Second, environmental economics has number of related papers have followed
benefitted enormously from the application this work, and are described by Fischer
Greenstone and Jack: Envirodevonomics: A Research Agenda for an Emerging Field 33
and Heutel (2013) in their review. The c ountry. Further, the benefits of mitigation
directed technological change literature is are both uncertain and in the future. The
not the only macroeconomic subfield that result is that any one actor has little incentive
has embraced environment–development to curb growth in the name of mitigation and
or environment–growth topics. Studies that no single government or entity has authority
use integrated assessment models of climate over all emitters. In these respects, climate
change necessarily consider growth–envi- change is the standard problem of externali-
ronment feedbacks (see Kelly and Kolstad ties and public goods on steroids.
1999; Nordhaus 2013; Hassler and Krusell This paper’s four explanations for the poor
2012) and the literature on trade and the state of environmental quality in developing
environment (for example, Copeland and countries all pose serious challenges to any
Taylor 2005; Shapiro 2013) has a history global effort to significantly limit climate
of work on pollution havens and environ- change. First, developing countries are likely
mental trade barriers that affect developing to place a relatively low present value on cur-
countries. rent greenhouse reductions. This is because
the marginal utility of current consumption
in these countries is very high, relative to
4. Climate Change
the marginal utility of future consumption,
We believe that climate change is the due to their low income levels and fast rates
most important topic in envirodevonomics. of growth. Further, today’s developed coun-
It is often referred to as an existential threat tries are richer than the developing ones and
and this is literally true for some developing experiencing slower growth, which means
countries (e.g., parts of Bangladesh are at they are likely to place a higher value on cur-
risk of disappearing due to sea-level rise). rent greenhouse gas reductions. Thus, both
More broadly, the greatest damages are the high current marginal utility of consump-
projected to occur in today’s developing tion in developing countries and the differ-
countries, especially those in the tropics. At ence in the marginal rates of substitution
the same time, today’s developing countries between current and future consumption
are expected to be the largest emitters of make it challenging for rich and poor coun-
greenhouse gases in the coming decades tries to find common ground on the value of
due to their projected growth in GDP and carbon reductions, at least without substan-
energy consumption (see Gertler et al. 2013; tial transfers that have their own political
Wolfram, Shelef, and Gertler 2012); China challenges (Becker et al. 2011; Deshpande
is currently the largest emitter of C O2 in the and Greenstone 2011). Second, the marginal
world and their emissions exceeded U.S. costs of reducing greenhouse gas emissions
emissions by 50 percent in 2012 (Olivier, are substantial. In the transportation sector,
Janssens-Maenhout, and Peters 2012). there is not currently a viable large-scale
Climate change is defined by many of alternative to petroleum, and in the elec-
the issues raised in section 2, and each of tricity sector, recent estimates suggest that
the potential barriers to optimal policy are the private costs of zero carbon sources of
even greater than they are for conventional electricity can be two-to-three times more
pollutants. In the climate case, production expensive than electricity generated from
and consumption choices generate global fossil fuels (Greenstone and Looney 2012).
externalities (e = e 0 + ∑ ni=1
(Δe i + a(c i, s i)), Slowing climate change may be difficult
rather than just externalities across house- even without distortions to social welfare
holds in the same city, region, or even maximization, yet the distortions discussed
34 Journal of Economic Literature, Vol. LIII (March 2015)
above are highly relevant. With respect to change may sharply reduce agricultural
the third explanation, political economy and yields and increase mortality rates in rural
rent-seeking forces can easily undermine
India. They also compare results from rural
efforts to reduce greenhouse gas emissions. areas to urban areas and to developed coun-
It is difficult to monitor carbon emissions, tries to show that those with opportunities
and this opens the door to claimed reduc- for self-protection are able to mitigate the
tions that exceed actual reductions. Indeed, negative health impacts.
the market for carbon offsets has been A recent line of research indicates that
greatly undermined by overstated claims that adaptation does occur, but not without
about impacts. Finally, a series of market its own drawbacks. Interestingly, air condi-
failures may distort revealed MWTPe for tioners can play a significant role in mod-
carbon reductions in developing countries. erating the effects of extreme weather on
For example, incomplete credit markets mortality (Barreca et al. 2013; Graff Zivin,
restrict opportunities to engage in long-term Hsiang, and Neidell 2013). However, they
mitigation projects and incomplete insur- are often run with C O2 intensive electric-
ance markets complicate efforts to protect ity and do little to protect against other cli-
oneself from uncertain climate damages. mate-related risks such as storm frequency.
Furthermore, climate change seems almost For example, Davis and Gertler (2013) doc-
to have been conceived by behavioral econo- ument the relationship between tempera-
mists as an ideal setting for behavioral biases; ture and energy consumption in Mexico,
it involves low probability events, impacts and show that long-term weather patterns
that are in the future, and trade-offs between drive air conditioner adoption. Additionally,
current and future generations. There is sub- areas that experience more regular storms
stantial opportunity for important research appear to experience less damage than do
about the roles of each of these four explana- less well-adapted places when hit by com-
tions in shaping climate mitigation policies. parable events, though adaptation is never
It is against this background that research sufficient to shrink damages to zero (Hsiang
on the impacts of climate change and oppor- and Narita 2012; Anttila-Hughes and Hsiang
tunities for adaptation have begun to emerge. 2012). Farmers also appear able to adapt to
A growing literature aims to estimate the changes in climate, with some evidence from
likely economic impacts of climate change extended periods of drought that may better
and shows projected increases in agricultural approximate the effects of climate change
losses, storm damages, civil conflicts, and than studies that rely on short-run weather
morbidity and mortality rates (Burke et al. shocks. By adjusting both crop choice and
forthcoming; Deschênes, Greenstone, and land under irrigation, farmers in India mit-
Guryan 2009; Deschênes and Greenstone igate losses due drought, however, only 14
2011; Feng, Krueger, and Oppenheimer percent of the substantial decline in profits
2010; Feng, Oppenheimer, and Schlenker are offset through adaptation (Taraz 2012).
2012; Schlenker and Lobell 2010; Graff Zivin Moreover, migration may also offer a via-
and Neidell 2014; IPCC 2001). The available ble adaptation strategy (Boustan, Kahn, and
evidence indicates these losses are all likely Rhode 2012), though external migration
to be of a greater magnitude in developing incentives may be necessary even when the
countries, where the resources available costs of not moving are very high (Bryan,
for investment in self-protection (adapta- Chowdhury, and Mobarak 2014).
tion) are limited. For example, Burgess et Evidence to date suggests that cli-
al. (2011) provide evidence that climate mate change is already underway and that
Greenstone and Jack: Envirodevonomics: A Research Agenda for an Emerging Field 35
2. What are the costs and benefits of pol- iii) What is the cost of abating a ton
icies to improve environmental quality of carbon through payments for
and access to energy? ecosystem services?
(a) How large are the health benefits iv) Through market-based emis-
that result from improvements in sions trading systems?
water and air quality? Which pro-
grams achieve these benefits most (b) What factors or design elements
cost-effectively? cause people in developing coun-
tries to make energy-efficiency
(b) Can providing information to the investments?
public change their behavior and
exposure to environmental risks? (c) Do transfers of funds or tech-
nologies from developed coun-
(c) What factors determine whether tries crowd out developing
environmental regulations are country investments in mitigation
effective in developing countries? and adaptation?
(d) How do issues of corruption and a (d) Will clean-energy products that
government’s capacity to enforce work in the lab have the same
regulation influence the impact of results when real people use them
specific environmental policies? in real world settings?
(e) How much are people willing to (e) What programs or policies can
pay for access to reliable energy best protect vulnerable popula-
sources? Or for reliable clean tions, like children, the elderly, and
energy sources? smallholder farmers, against the
effects climate change?
3. What policies can be effective for cli- These and many other topics are increas-
mate mitigation and adaptation? ingly feasible areas for economic research,
as data quality and availability improve.
(a) How do the costs of abating a ton of But even more importantly, finding reliable
carbon emissions compare across answers to these questions will advance eco-
different policies? nomic understanding and inform policy, with
the potential to enormously impact human
i) What is the cost of abat- welfare.
ing a ton of carbon through
References
energy-efficient investments
for consumers? Acemoglu, Daron, Philippe Aghion, Leonardo Bursz-
tyn, and David Hemous. 2012. “The Environment
and Directed Technical Change.” American Eco-
ii) What is the cost through nomic Review 102 (1): 131–66.
energy-efficiency policies tar-
Ali, Daniel Ayalew, Klaus Deininger, and Markus Gold-
geting manufacturers, espe- stein. 2014. “Environmental and Gender Impacts
of Land Tenure Regularization in Africa: Pilot Evi-
cially small- and medium-sized dence from Rwanda.” Journal of Development Eco-
ones? nomics 110: 262–75.
Greenstone and Jack: Envirodevonomics: A Research Agenda for an Emerging Field 37
Alix-Garcia, Jennifer, Craig McIntosh, Katharine R. overnance and Development.” Journal of Eco-
G
E. Sims, and Jarrod R. Welch. 2013. “The Ecologi- nomic Perspectives 16 (4): 185–205.
cal Footprint of Poverty Alleviation: Evidence from Bardhan, Pranab, and Christopher Udry. 1999. Devel-
Mexico’s Opportunidades Program.” Review of Eco- opment Microeconomics. Oxford and New York:
nomics and Statistics 95 (2): 417–35. Oxford University Press.
Alix-Garcia, Jennifer M., Katharine R. E. Sims, and Barreca, Alan, Karen Clay, Olivier Deschenes, Michael
Patricia Yañez-Pagans. Forthcoming. “Only One Greenstone, and Joseph S. Shapiro. 2013. “Adapt-
Tree from Each Seed? Environmental Effectiveness ing to Climate Change: The Remarkable Decline in
and Poverty Alleviation in Mexico’s Payments for the U.S. Temperature–Mortality Relationship over
Ecosystem Services Program.” American Economic the 20th Century.” National Bureau of Economic
Journal: Economic Policy. Research Working Paper 18692.
Almond, Douglas, Yuyu Chen, Michael Greenstone, Baumol, William J., and Wallace E. Oates. 1988. The
and Hongbin Li. 2009. “Winter Heating or Clean Theory of Environmental Policy, Second edition.
Air? Unintended Impacts of China’s Huai River Pol- Cambridge and New York: Cambridge University
icy.” American Economic Review 99 (2): 184–90. Press.
Andam, Kwaw S., Paul J. Ferraro, Katharine R. E. Becker, Gary S., Kevin M. Murphy, and Robert H.
Sims, Andrew Healy, and Margaret B. Holland. 2010. Topel. 2011. “On the Economics of Climate Policy.”
“Protected Areas Reduced Poverty in Costa Rica and B.E. Journal of Economic Analysis and Policy 10 (2).
Thailand.” Proceedings of the National Academy of Bennear, Lori, Alessandro Tarozzi, Alexander Pfaff,
Sciences 107 (22): 9996–10001. Soumya Balasubramanya, Kazi Matin Ahmed, and
Andreoni, James, and Arik Levinson. 2001. “The Sim- Alexander van Geen. 2013. “Impact of a Randomized
ple Analytics of the Environmental Kuznets Curve.” Controlled Trial in Arsenic Risk Communication on
Journal of Public Economics 80 (2): 269–86. Household Water-Source Choices in Bangladesh.”
Anttila-Hughes, Jesse K., and Solomon M. Hsiang. Journal of Environmental Economics and Manage-
2012. “Destruction, Disinvestment, and Death: Eco- ment 65 (2): 225–40.
nomic and Human Losses Following Environmental Bennett, Daniel. 2012. “Does Clean Water Make You
Disaster.” Unpublished. Dirty? Water Supply and Sanitation in the Philip-
Aragón, Fernando M., and Juan Pablo Rud. 2013. pines.” Journal of Human Resources 47 (1): 146–73.
“Modern Industries, Pollution and Agricultural Pro- Berry, James, Greg Fischer, and Raymond Guiteras.
ductivity: Evidence from Ghana.” Unpublished. 2011. “Eliciting and Utilizing Willingness to Pay:
Arceo-Gomez, Eva O., Rema Hanna, and Paulina Evidence from Field Trials in Northern Ghana.”
Oliva. 2012. “Does the Effect of Pollution on Infant Unpublished.
Mortality Differ between Developing and Devel- Bertrand, Marianne, Sendhil Mullainathan, and Eldar
oped Countries? Evidence from Mexico City.” Shafir. 2004. “A Behavioral-Economics View of Pov-
National Bureau of Economic Research Working erty.” American Economic Review 94 (2): 419–23.
Paper 18349. Besley, Timothy, and Torsten Persson. 2013. “Taxation
Arriagada, Rodrigo A., Paul J. Ferraro, Erin O. Sills, and Development.” In Handbook of Public Econom-
Subhrendu K. Pattanayak, and Silvia Cordero-San- ics, Volume 5, edited by Alan J. Auerbach, Raj Chetty,
cho. 2012. “Do Payments for Environmental Ser- Martin Feldstein, and Emmanuel Saez, 51–110.
vices Affect Forest Cover? A Farm-Level Evaluation Amsterdam and Boston: Elsevier, North-Holland.
from Costa Rica.” Land Economics 88 (2): 382–99. Bhattacharya, Soma, Anna Alberini, and Maureen L.
Ashraf, Nava, James Berry, and Jesse M. Shapiro. 2010. Cropper. 2007. “The Value of Mortality Risk Reduc-
“Can Higher Prices Stimulate Product Use? Evi- tions in Delhi, India.” Journal of Risk and Uncer-
dence from a Field Experiment in Zambia.” Amer- tainty 34 (1): 21–47.
ican Economic Review 100 (5): 2383–413. Boomhower, Judson, and Lucas W. Davis. 2014. “A
Ashraf, Nava, B. Kelsey Jack, and Emir Kamenica. Credible Approach for Measuring Inframarginal Par-
2013. “Information and Subsidies: Complements ticipation in Energy Efficiency Programs.” Journal of
or Substitutes?” Journal of Economic Behavior and Public Economics 113: 67–79.
Organization 88: 133–39. Boustan, Leah Platt, Matthew E. Kahn, and Paul W.
Assunção, Juliano, Clarissa Gandour, Romero Rocha, Rhode. 2012. “Moving to Higher Ground: Migration
and Rudi Rocha. 2013. “Does Credit Affect Defor- Response to Natural Disasters in the Early Twenti-
estation? Evidence from a Rural Credit Policy in the eth Century.” American Economic Review 102 (3):
Brazilian Amazon.” Unpublished. 238–44.
Banerjee, Abhijit, and Sendhil Mullainathan. 2010. Brainerd, Elizabeth, and Nidhiya Menon. 2014. “Sea-
“The Shape of Temptation: Implications for the Eco- sonal Effects of Water Quality: The Hidden Costs
nomic Lives of the Poor.” National Bureau of Eco- of the Green Revolution to Infant and Child Health
nomic Research Working Paper 15973. in India.” Journal of Development Economics 107:
Barbier, Edward B. 2007. Natural Resources and Eco- 49–64.
nomic Development. Cambridge and New York: Bryan, Gharad, Shyamal Chowdhury, and Ahmed
Cambridge University Press. Mushfiq Mobarak. 2014. “Underinvestment in a Prof-
Bardhan, Pranab. 2002. “Decentralization of itable Technology: The Case of Seasonal Migration in
38 Journal of Economic Literature, Vol. LIII (March 2015)
on Tap: Piped Water Adoption in Urban Morocco.” Foster, Andrew, and Emilio Gutierrez. 2012. “Direct
American Economic Journal: Economic Policy 4 (4): and Indirect Effects of Voluntary Certification: Evi-
68–99. dence from the Mexican Clean Industry Program.”
Dinkelman, Taryn. 2011. “The Effects of Rural Elec- Unpublished.
trification on Employment: New Evidence from Foster, Andrew, Emilio Gutierrez, and Naresh Kumar.
South Africa.” American Economic Review 101 (7): 2009. “Voluntary Compliance, Pollution Levels, and
3078–3108. Infant Mortality in Mexico.” American Economic
Dominici, Francesca, Michael Greenstone, and Cass Review 99 (2): 191–97.
R. Sunstein. 2014. “Particulate Matter Matters.” Foster, Andrew D., and Mark R. Rosenzweig. 1995.
Science 344 (6181): 257–59. “Learning By Doing and Learning from Others:
Duflo, Esther. 2012. “Human Values and the Design Human Capital and Technical Change in Agri-
of the Fight against Poverty.” Tanner Lectures on culture.” Journal of Political Economy 103 (6):
Human Values at Harvard University. 1176–1209.
Duflo, Esther, Michael Greenstone, Raymond Guit- Foster, Andrew D., and Mark R. Rosenzweig. 2010.
eras, and Thomas Clasen. 2014. “The Short- and “Microeconomics of Technology Adoption.” Annual
Medium-Run Impact of Clean Water and Sanitation Review of Economics 2: 395–424.
on Diarrhea in Rural India.” Unpublished. Fowlie, Meredith L. 2009. “Incomplete Environmental
Duflo, Esther, Michael Greenstone, Rohini Pande, Regulation, Imperfect Competition, and Emissions
and Nicholas Ryan. 2013. “Truth-Telling By Third- Leakage.” American Economic Journal: Economic
Party Auditors and the Response of Polluting Firms: Policy 1 (2): 72–112.
Experimental Evidence from India.” Quarterly Jour- Fowlie, Meredith L. 2010. “Emissions Trading, Elec-
nal of Economics 128 (4): 1499–1545. tircity Restructuring, and Investment in Pollution
Duflo, Esther, and Rohini Pande. 2007. “Dams.” Quar- Abatement.” American Economic Review 100 (3):
terly Journal of Economics 122 (2): 601–46. 837–69.
Dupas, Pascaline. 2011. “Health Behavior in Devel- Fowlie, Meredith L., Christopher R. Knittel, and Cath-
oping Countries.” Annual Review of Economics 3: erine Wolfram. 2012. “Sacred Cars? Cost-Effective
425–49. Regulation of Stationary and Nonstationary Pollution
Ebenstein, Avraham. 2012. “The Consequences of Sources.” American Economic Journal: Economic
Industrialization: Evidence from Water Pollution Policy 4 (1): 98–126.
and Digestive Cancers in China.” Review of Econom- Galiani, Sebastian, Paul Gertler, and Ernesto Schar-
ics and Statistics 94 (1): 186–201. grodsky. 2005. “Water for Life: The Impact of the
Feler, Leo, and J. Vernon Henderson. 2011. “Exclusion- Privatization of Water Services on Child Mortality.”
ary Policies in Urban Development: Under-Servicing Journal of Political Economy 113 (1): 83–120.
Migrant Households in Brazilian Cities.” Journal of Gallego, Francisco, Juan-Pablo Montero, and Christian
Urban Economics 69 (3): 253–72. Salas. 2013. “The Effect of Transport Policies on Car
Feng, Shuaizhang, Alan B. Krueger, and Michael Use: Evidence from Latin American Cities.” Journal
Oppenheimer. 2010. “Linkages among Climate of Public Economics 107: 47–62.
Change, Crop Yields and Mexico–US Cross-Border Gertler, Paul, Orie Shelef, Catherine Wolfram, and
Migration.” Proceedings of the National Academy of Alan Fuchs. 2013. ”How Pro-Poor Growth Affects
Sciences 107 (32): 14257–62. the Demand for Energy.” National Bureau of Eco-
Feng, Shuaizhang, Michael Oppenheimer, and Wol- nomic Research Working Paper 19092.
fram Schlenker. 2012. “Climate Change, Crop Graff Zivin, Joshua S., S. Hsiang, and Matthew Neidell.
Yields, and Internal Migration in the United States.” 2013. “Climate, Human Capital and Adaptation.”
National Bureau of Economic Research Working Unpublished.
Paper 17734. Graff Zivin, Joshua, and Matthew Neidell. 2014. “Tem-
Ferraro, Paul J., Kathleen Lawlor, Katrina L. Mullan, perature and the Allocation of Time: Implications for
and Subhrendu K. Pattanayak. 2012. “Forest Figures: Climate Change.” Journal of Labor Economics 32
Ecosystem Services Valuation and Policy Evaluation (1): 1–26.
in Developing Countries.” Review of Environmental Greenstone, Michael, and Rema Hanna. 2014. “Envi-
Economics and Policy 6 (1): 20–44. ronmental Regulation, Air and Water Pollution,
Field, Erica, Rachel Glennerster, and Reshmaan Hus- and Infant Mortality in India.” American Economic
sam. 2011. “Throwing the Baby Out with the Drink- Review 104 (10): 3038–72.
ing Water: Unintended Consequences of Arsenic Greenstone, Michael, and Adam Looney. 2012. “Pay-
Mitigation Efforts in Bangladesh.” Unpublished. ing Too Much for Energy? The True Costs of Our
Fischer, Carolyn, and Garth Heutel. 2013. “Environ- Energy Choices.” Daedalus 141 (2): 10–30.
mental Macroeconomics: Environmental Policy, Grossman, Gene M., and Alan B. Krueger. 1995. “Eco-
Business Cycles, and Directed Technical Change.” nomic Growth and the Environment.” Quarterly
Annual Review of Resource Economics 5: 197–210. Journal of Economics 110 (2): 353–77.
Fisman, Raymond. 2001. “Estimating the Value of Guiteras, Raymond, David I. Levine, Thomas H. Pol-
Political Connections.” American Economic Review ley, and Brian Quistorff. 2014. “Credit Constraints,
91 (4): 1095–1102. Discounting and Investment in Health: Evidence
40 Journal of Economic Literature, Vol. LIII (March 2015)
from Micropayments for Clean Water in Dhaka.” on Demand for Environmental Quality.” Journal of
Unpublished. Development Economics 87 (1): 14–28.
Hall, Robert E., and Charles I. Jones. 2007. “The Value Jayachandran, Seema. 2009. “Air Quality and Ear-
of Life and the Rise in Health Spending.” Quarterly ly-Life Mortality: Evidence from Indonesia’s Wild-
Journal of Economics 122 (1): 39–72. fires.” Journal of Human Resources 44 (4): 916–54.
Hanna, Rema, and Paulina Oliva. 2014. “Moving Up Kahn, Matthew E., Pei Li, and Daxuan Zhao. Forth-
the Energy Ladder: The Effect of an Increase in Eco- coming. “Water Pollution Progress at Borders: The
nomic Well-Being on Fuel Consumption Choices of Role of Changes in China’s Political Promotion
the Poor in India.” Unpublished. Incentives.” American Economic Journal: Economic
Hanna, Rema, and Paulina Oliva. Forthcoming. “The Policy.
Effect of Pollution on Labor Supply: Evidence from Kelly, David L., and Charles D. Kolstad. 1999. “Inte-
a Natural Experiment in Mexico City.” Journal of grated Assessment Models for Climate Change
Public Economics. Control.” In The International Yearbook of Environ-
Harbaugh, William T., Arik Levinson, and David Mol- mental and Resource Economics 1999/2000: A Sur-
loy Wilson. 2002. “Reexamining the Empirical Evi- vey of Current Issues, edited by Henk Folmer and
dence for an Environmental Kuznets Curve.” Review Tom Tietenberg, 171–97. New York: Elgar.
of Economics and Statistics 84 (3): 541–51. Kremer, Michael. 1993. “The O-Ring Theory of Eco-
Hassan, Rashid, Robert Scholes, and Neville Ash, eds. nomic Development.” Quarterly Journal of Econom-
2005. Ecosystem and Human Well-Being: Current ics 108 (3): 551–75.
State and Trends, Volume 1. Washington and Lon- Kremer, Michael, Jessica Leino, Edward Miguel, and
don: Island Press. Alix Peterson Zwane. 2011. “Spring Cleaning: Rural
Hassler, John, and Per Krusell. 2012. “Economics Water Impacts, Valuation, and Property Rights Insti-
and Climate Change: Integrated Assessment in a tutions.” Quarterly Journal of Economics 126 (1):
Multi-region World.” Journal of the European Eco- 145–205.
nomic Association 10 (5): 974–1000. Kumler, Todd, Eric Verhoogen, and Judith Frías. 2013.
Hemous, David. 2013. “Environmental Policy and “Enlisting Workers in Monitoring Firms: Payroll Tax
Directed Technical Change in a Global Economy: Compliance in Mexico.” Working Paper.
The Dynamic Impact of Unilateral Environmental Lemos, Maria Carmen, and Arun Agrawal. 2006. “Envi-
Policies.” Centre for Economic Policy Research Dis- ronmental Governance.” Annual Review of Environ-
cussion Paper 9733. ment and Resources 31: 297–325.
Henderson, Vernon, Adam Storeygard, and David León, Gianmarco, and Edward Miguel. 2012. “Trans-
N. Weil. 2011. “A Bright Idea for Measuring Eco- portation Choices, Fatalism and the Value of Life in
nomic Growth.” American Economic Review 101 (3): Africa.” Working Paper.
194–99. Lin, C.-Y. Cynthia, Wei Zhang, and Victoria I. Uman-
Horowitz, John K. 1998. “Review of The Environment skaya. 2011. “On the Design of Driving Restrictions:
and Emerging Development Issues. Volumes 1 and Theory and Empirical Evidence.” Working Paper.
2.” Journal of Economic Literature 36 (3): 1529–30. Lipscomb, Molly. 2008. “The Effect of Environmental
Hsiang, Solomon M., and Daiju Narita. 2012. “Adap- Enforcement on Product Choice and Competition:
tation to Cyclone Risk: Evidence from the Global Theory and Evidence from India.” Unpublished.
Cross-Section.” Climate Change Economics 3 (2). Lipscomb, Molly, and Ahmed Mushfiq Mobarak. 2011.
Hsieh, Chang-Tai, and Peter J. Klenow. 2009. “Misallo- “Decentralization and the Political Economy of
cation and Manufacturing TFP in China and India.” Water Pollution: Evidence from the Re-drawing of
Quarterly Journal of Economics 124 (4): 1403–48. County Borders in Brazil.” Unpublished.
Intergovernmental Panel on Climate Change. 2001. Lipscomb, Molly, Ahmed Mushfiq Mobarak, and Tania
Climate Change 2001: Impacts, Adaptation, and Vul- Barham. 2013. “Development Effects of Electrifica-
nerability: Contribution of Working Group II to the tion: Evidence from the Topographic Placement of
Third Assessment Report of the Intergovernmental Hydropower Plants in Brazil.” American Economic
Panel on Climate Change. Edited by James J. McCa- Journal: Applied Economics 5 (2): 200–231.
rthy et al. Cambridge and New York: Cambridge Lofgren, Asa, Peter Martinsson, Magnus Hennlock,
University Press. and Thomas Sterner. 2012. “Are Experienced People
Israel, Debra, and Arik Levinson. 2004. “Willingness to Affected By a Pre-set Default Option—Results from
Pay for Environmental Quality: Testable Empirical a Field Experiment.” Journal of Environmental Eco-
Implications of the Growth and Environment Litera- nomics and Management 63 (1): 66–72.
ture.” B. E. Journal of Economic Analysis and Policy Madajewicz, Malgosia, et al. 2007. “Can Information
3 (1). Alone Change Behavior? Response to Arsenic Con-
Jack, B. Kelsey. 2013. “Private Information and the tamination of Groundwater in Bangladesh.” Journal
Allocation of Land Use Subsidies in Malawi.” Amer- of Development Economics 84 (2): 731–54.
ican Economic Journal: Applied Economics 5 (3): McRae, Shaun. 2015. “Infrastructure Quality and the
113–35. Subsidy Trap.” American Economic Review 105, (1):
Jalan, Jyotsna, and E. Somanathan. 2008. “The Impor- 35–66.
tance of Being Informed: Experimental Evidence Miller, Grant, and Ahmed Mushfiq Mobarak. 2013.
Greenstone and Jack: Envirodevonomics: A Research Agenda for an Emerging Field 41