Content uploaded by Justus H.H. Wesseler
Author content
All content in this area was uploaded by Justus H.H. Wesseler on Jul 28, 2014
Content may be subject to copyright.
Environment and Development
Economics
http://journals.cambridge.org/EDE
Additional services for Environment and
Development Economics:
Email alerts: Click here
Subscriptions: Click here
Commercial reprints: Click here
Terms of use : Click here
The economic power of the Golden Rice
opposition
Justus Wesseler and David Zilberman
Environment and Development Economics / FirstView Article / January 2014, pp 1 - 19
DOI: 10.1017/S1355770X1300065X, Published online: 22 January 2014
Link to this article: http://journals.cambridge.org/abstract_S1355770X1300065X
How to cite this article:
Justus Wesseler and David Zilberman The economic power of the Golden Rice
opposition . Environment and Development Economics, Available on CJO 2014
doi:10.1017/S1355770X1300065X
Request Permissions : Click here
Downloaded from http://journals.cambridge.org/EDE, IP address: 137.224.252.10 on 05 Feb 2014
Environment and Development Economics, page 1 of 19. © Cambridge University Press 2014
doi:10.1017/S1355770X1300065X
The economic power of the Golden Rice
opposition
JUSTUS WESSELER
Technische Universit¨at M¨unchen, Center of Life and Food Sciences
Weihenstephan, Weihenstephaner Steig 22, 85354, Freising, Germany.
Tel: +49 8161 715632. Fax: +49 8161 713030.
Email: justus.wesseler@wzw.tum.de
DAVID ZILBERMAN
Department of Agricultural and Resource Economics, University of
California, Berkeley, USA. Email: zilber11@berkeley.edu
Submitted 4 March 2013; revised 18 October 2013; accepted 23 October 2013
ABSTRACT. Vitamin A enriched rice (Golden Rice) is a cost-efficient solution that can
substantially reduce health costs. Despite Golden Rice being available since early 2000,
this rice has not been introduced in any country. Governments must perceive additional
costs that overcompensate the benefits of the technology to explain the delay in approval.
We develop a real option model including irreversibility and uncertainty about perceived
costs and arrival of new information to explain a delay in approval. The model has
been applied to the case of India. Results show the annual perceived costs have to be
at least US$199 million per year approximately for the last decade to explain the delay in
approval of the technology. This is an indicator of the economic power of the opposition
towards Golden Rice resulting in about 1.4 million life years lost over the past decade
in India.
1. Introduction
‘So, if introduced on a large scale, golden rice can exacerbate malnutrition
and ultimately undermine food security.’ This statement by Greenpeace
(2012: 3) is in strong contradiction to the reported impacts of vitamin A
deficiency and the nutritional impacts of vitamin A enriched diets. More
than 125 million children under five years of age suffer from vitamin A
deficiency (VAD). Dietary VAD causes 250,000–500,000 children to go blind
each year.
More than half of the children who lose their sight because of VAD
die within a year of becoming blind. Additionally, VAD compromises
the immune systems of approximately 40 per cent of children under
the age of five in the developing world, greatly increasing the risk
The online version of this article is published within an Open Access environment
subject to the conditions of the Creative Commons Attribution licence http://
creativecommons.org/licenses/by/3.0/
2Justus Wesseler and David Zilberman
of severe illnesses from common childhood infections. Further health
consequences include anemia, increased susceptibility to infection, and
poorer growth (West and Darnton-Hill,2008). Additionally, early child-
hood malnutrition has long-lasting effects that are difficult to reverse by
better nutrition in later life (World Health Organization,2001). Address-
ing VAD and zinc deficiency in nutrition has been ranked by the
Copenhagen Consensus (2008) as the number one problem in developing
countries for sustainable economic development and was reemphasized in
2012 (Copenhagen Consensus,2012).
There are two major channels through which nutrition may affect eco-
nomic development. Malnutrition limits (labor) productivity and effects
human capital accumulation. Firstly, there is convincing evidence from
past experience that nutritional status affects labor outcomes, particu-
larly productivity. While the exact mechanisms underlying these rela-
tionships are not entirely clear, experimental and observational studies
have documented sizeable effects of nutrition on productivity indicators
(Thomas and Frankenberg,2002). Fogel’s (1994) study of the economic and
health history of Europe illustrates the aforementioned relationship. At the
end of the 18th century in England and France, agricultural production,
and therefore food provision, was so low that approximately 20 per cent of
the population was incapable of working more than a few hours of light
work per day due to lacking food resources. Many people were chroni-
cally malnourished and died young, which resulted in premature losses of
their human capital. Only the agricultural productivity increases in the 19th
century permitted an escape from this developmental trap, and enabled
individual productivity gains as well as increases in productivity on a
macroeconomic level (Deaton,2003). Similarly, Doblhammer and Vaupel
(2001) investigate the relationship between month of birth and remaining
life expectancy at age 50 in Austria, Denmark and Australia. Their results
indicate that health appears to depend on factors like nutrition in utero or
early childhood. These effects are attributed to the effects of intrauterine
nutrition more than half a century ago, particularly to the seasonal avail-
ability of high-quality food like fresh fruit and vegetables (Deaton,2003).
Secondly, malnutrition may also affect economic development indi-
rectly via hampering cognitive abilities and human capital develop-
ment. Richards et al. (2002), for example, use birth weight and cognitive
indicators of the British 1946 birth cohort and find that birth weight
was positively associated with cognition up to age 26, and with the
likelihood of obtaining advanced educational qualifications. Similarly,
Currie and Hyson (1999) find that malnutrition, measured by low birth
weights, has significant long-term effects on health status, educational
attainments and labor market outcomes. Moreover, Case and Paxson (2006)
use stature of US and UK citizens as an indicator for their quality of diet
during childhood and find that nutrition is positively associated with cog-
nitive ability, which is rewarded in the labor market. They show that taller
children have higher average cognitive test scores, and that these test scores
explain a large portion of the height premium in earnings. In an historical
study, Baten et al. (2010) show that nutrition during early childhood mat-
tered for labor market outcomes: individuals who grew up in times and
Environment and Development Economics 3
places characterized by high food prices tended to acquire fewer cognitive
skills and were more likely to pursue occupations with limited intellectual
requirements.
Vitamin A enriched rice has the potential to address the problem of
micronutrient deficiencies in early childhood. Rice produces β-carotene
in the leaves but not in the grain, where the biosynthetic pathway is
turned off during plant development. In Golden Rice, two genes have been
inserted into the rice genome by genetic engineering, which would not
have been possible by traditional breeding, to restart the carotenoid biosyn-
thetic pathway leading to the production and accumulation of β-carotene
in the grains. The intensity of the golden color is an indicator of the con-
centration of β-carotene in the endosperm. Since a prototype Golden Rice
was developed in 1999, new lines with higher β-carotene content have
been generated. The breeding goal is to provide the recommended daily
allowance of vitamin A in 100–200 g of β-carotene containing rice. This
corresponds to the amount of rice eaten daily by children living in rice-
based societies, such as India, Vietnam and Bangladesh. In other countries,
Golden Rice could still be a valuable complement to children’s diets, thus
contributing to the reduction of clinical and sub-clinical VAD-related dis-
eases. The different forms of Golden Rice contain between 1.6 and 35 μg
β-carotene per gram of rice. A recent study with children has shown that
the bioavailability of provitamin A from Golden Rice is as effective as
pure β-carotene in oil, and far better than spinach in providing vitamin
A to children. A daily intake of 60 g of rice (half a cup) would provide
about 60 per cent of the Chinese Recommended Nutrient Intake of vita-
min A for 6–8-year-old children and be sufficient to prevent vitamin A
malnutrition (Tang et al.,2009,2012).
Despite the expected nutritional benefits of Golden Rice, the technology
faces strong opposition. Opponents to the technology argue that a sufficient
daily vitamin A supply would require a two-year-old child to eat about 3 kg
of rice per day; alternative and cheaper strategies would exist to address
VAD and in the end the project is only an industry ploy to open doors for
other genetically modified (GM) crops (Greenpeace,2005). The argument
about the daily intake has been dismissed as this assumes Golden Rice
would be the only source of vitamin A (Enserink,2008). Suggested alterna-
tives such as alternative diets, industrial fortification or supplements have
been available for many years but have not solved the problem (Stein et al.,
2008). Hence, the possibility exists that Golden Rice may reduce VAD and
save millions of lives, while other strategies have not been so successful.
Proof of concept of enhancing the vitamin A content of rice has existed
since the late 1990s and the expectation was that by 2002 the first commer-
cial rice varieties would be available. Developing a commercial product
has not been delayed by solving a number of intellectual property right
issues. The major stumbling block was – and still is – the regulation of
genetically engineered (GE) crops. In 2000, the Golden Rice project started
a public–private partnership with Syngenta to use their expertise in devel-
oping dossiers for the approval requirements of GE crops to ease problems
caused by national regulations. Ten years have passed and still Golden Rice
has neither been approved for cultivation in India and Bangladesh, two of
4Justus Wesseler and David Zilberman
the target countries of the Golden Rice Humanitarian Project, nor in other
countries where its varieties are under development (Potrykus,2010a,b).
Many argue that the additional regulations for approval required for
GE crops compared with non-GE crops, and Golden Rice in particular,
are not supported by science, and hence are overregulated. Justification
for current regulation is based on the notion that the technology leads to
‘unpredictable and uncontrolled modification of the genome’; this argu-
ment ignores the fact that all traditional breeding has been and is doing
exactly the same (Baudo et al.,2006;Shewry et al.,2007;Batista et al.,2008;
Herman et al.,2009). Although there have been studies showing the safety
consequences of GM crops for rats, a careful assessment of those studies
did not confirm the claims being made. Further, extensive reviews found
no support for adverse human health effects from GM crops (Bennett et al.,
2013). Developing Golden Rice requires the use of selectable marker genes.
Regulatory authorities prefer that antibiotic selectable marker genes be
deleted (e.g., Committee on Agriculture,2012). This is technically possible,
but requires substantial time and effort despite the fact that there is a wealth
of scientific literature documenting that the antibiotic marker genes in
use have no effect on consumer and environmental safety (Ramessar et al.,
2007). Furthermore, requests for additional regulation are often based on
the precautionary principle, while a strong interpretation of the precaution-
ary principle has been shown to be a logically inconsistent line of reasoning
for extra regulation (van den Belt,2003),1while there is wide agreement
among economists that optimal regulations related to food and environ-
mental safety should be based on benefit-costs analyses (Arrow et al.,1996).
Even the definition of what constitutes a GE crop is controversial from a
scientific point of view (Herring,2007).
Ex ante assessments have been done for Golden Rice in India (Stein et al.,
2008) and the Philippines (Zimmermann and Qaim,2004). These studies
show that in India the costs per disability-adjusted life year (DALY) saved
between US$3.1 and US$19.4: much lower than for alternative intervention
strategies. The net social benefits for the Philippines have been calculated
to be in the range of US$16 to US$88 million per year. These numbers are
now expected to be on the lower side of this range as only direct health
impacts have been considered and improvements in vitamin A enhanced
rice breeding since 2004 were not included.
Despite the wealth of information available about the health benefits
of Golden Rice, national governments, which are bound by the rules and
1According to Pascal’s wager: ‘Given known but nonzero probability of God’s exis-
tence and the infinity of the reward of an eternal life, the rational option would be
to conduct one’s earthly life as if God exists’ (van den Belt,2003: 1124). The con-
tradiction is the many gods example: ‘Consider the possible existence of another
deity than God, say Odin. If Odin is jealous, he will resent our worship of God,
and we will have to pay an infinite price for our mistake. Never mind that Odin’s
existence may not seem likely or plausible to us. It is sufficient that we cannot
exclude the possibility that he exists with absolute certainty. Therefore, the same
logic of Pascal’s wager would lead us to adopt the opposite conclusion not to
worship God. Pascal’s argument, then, cannot be valid’ (van den Belt,2003: 1124).
Environment and Development Economics 5
regulations on the use of GE crops, are reluctant to approve its introduction
(Potrykus,2010a,b).
The objectives of this paper are to identify the costs and benefits of regu-
lating Golden Rice, paying particular attention to possible overregulation,
and to calibrate the model for Golden Rice in India. The simple model we
present considers uncertainties related to the perceived costs by national
governments of the technology and approval time, i.e., over time, addi-
tional information becomes available. This allows us to identify the costs
of delaying the approval and in particular the perceived costs by national
regulators equivalent to the costs imposed on society by the opposition to
Golden Rice. Alternatively, ex ante assessments show substantial benefits
of Golden Rice, but governments are still reluctant to approve the techno-
logy. Hence, there must be additional perceived costs for governments not
approving the technology. This is the first study calculating those costs for
the case of India. These perceived costs have been treated as being irre-
versible for being cautious. Hence, the opportunity to introduce Golden
Rice has been valued following the quasi-option-value approach intro-
duced by Arrow and Fisher (1974). Arrow and Fisher, among others (e.g.,
Dixit and Pindyck,1994;Wesseler,2009), show that decisions under uncer-
tainty and irreversible benefits and costs either understate or overstate net
benefits if irreversibility effects are ignored.
Results show that it pays to bring forward new arguments against
the introduction of Golden Rice when the technology is close to being
approved. The Indian case shows that the perceived costs by national gov-
ernments are substantially larger than the costs of introducing Golden Rice,
and has caused and continues to cause the death of thousands of children.
2. The real option model of perceived costs
The problem is described as follows. A benevolent developer of a vita-
min A biofortification strategy in the form of Golden Rice, henceforth
called the Golden Rice strategy (GRS) such as, for example, the ‘Human-
itarian Golden Rice’ project, provides the technology for free. The benefits
of the program are known to the developer as well as the national gov-
ernment and depend on the government’s implementation strategy, the
costs of which are known. The acceptance of the technology depends on
national governments. The introduction of the GRS (the implementation
strategy) costs the national government one-off administrative set-up costs
and annual costs of running the strategy. This set-up serves as the reference
model. If the benefits outweigh the costs of the GRS, a welfare maximiz-
ing regulator would implement the GRS. It is well known in the literature
that regulators, other government agencies or governments in general do
not follow this simple benefit-cost metric. Their objective function may not
only be based on the direct benefits and costs of the GRS. A number of polit-
ical economy models have been developed to analyze policies related to
the introduction of GE crops in more detail considering two or more lobby
groups (e.g., Graff et al.,2009;Swinnen and Vandemoortele,2010) and the
role of media (Vigani and Olper,2014). Apel (2010) has argued that an
6Justus Wesseler and David Zilberman
anti-genetically modified organism (GMO) strategy has been a successful
fundraising strategy for environmental lobby groups such as Greenpeace
or Friends of the Earth. The role of anti-GMO lobby groups and their strong
influence on decision making, particularly in developing countries, has
been well described by political scientists (e.g., Paarlberg,2008;Herring,
2010). Lobby groups with capital stocks at risk in the face of agricultural
innovations use the political systems to protect their interests (Graff et al.,
2013). Decision makers may consider the arguments raised by opponents
for reasons of re-election, administrative power, budgetary power, side-
payments and more. In the end, whatever the specific reasons might be
and whatever the detailed political economy process may look like, the
final result is that decision makers take additional costs into account that
result in the delay of the approval.2These costs we call the perceived costs
– a term we use as those costs cannot be observed directly, and are a result
of the political economy process of the regulatory decision.
If the inclusion of the perceived costs results in the delay of an approval,
benefits of the GRS are foregone. We call the foregone benefits caused by a
delay in approval the economic power of the opposition to a GRS.
We now develop the model to assess the economic power in more detail.
The approval decision by the national government is exogenous to the
developer as are the set-up and annual costs. At time t=0 the govern-
ment’s view is that perceived costs, Gc, of introducing the GRS exist and
they are high, Gc0, while other benefits and costs discussed in more
detail below are assumed to be known. This assumption is a simplifica-
tion but can be justified by the studies investigating the costs and benefits
of introducing the GRS. Hence, all remaining uncertainty is summarized
under perceived costs. Over time, further information about the perceived
costs arrive and at time κeither the strategy will be successful and per-
ceived costs not be confirmed, Gc=0 with probability q, or confirmed to
be high, Gc0 with probability (1 – q). Hence, the introduction mainly
depends on the perceived costs of implementing the GRS. Based on that,
the national government may decide the strategy will be introduced imme-
diately (I) or postponed (P). Uncertainty about perceived costs will be
resolved at some future point in time κ>0. At κ>0 the national govern-
ment will know whether or not their perceived costs have been confirmed.
At time t=0 the arrival of κ>0 is uncertain to the government and
hence a random variable. k∈(0,∞)is the date at which uncertainty about
perceived costs is resolved; it follows from the exponential distribution
f(κ) =he−hκ(Taylor and Karlin,1984), with E(κ) =1/h, where his the
hazard rate.3
2A delay is equivalent to a temporary ban as commonly observed in many coun-
tries. Regulatory decisions are almost never final. The decisions can be challenged
in court and in particular if new evidence challenging the current status is pre-
sented. Also, public opinion may change over time, influencing the view of the
regulatory body.
3The exponential distribution has a couple of attractive features for models with
arrival of information; an important feature is that it allows analytical tractability
of the model.
Environment and Development Economics 7
This specification is not that restrictive as it will allow identifying
threshold levels for perceived costs as shown below. The specification of
the model considers the inherent uncertainty decision makers face as well
as the fact that new information arrives over time, but that the specific point
in time when information will be available is uncertain.
The health benefits of introducing the GRS are in the form of improved
vitamin A supply. These benefits increase over time via the distribution
of the seeds. As commonly done, we assume a logistic functional form
for the diffusion of the Golden Rice seed (t)=ρmax
1+exp(−αρ−βρt), where the
slope parameter βρis known as the natural rate of diffusion, as it measures
the rate at which adoption ρincreases with time t. The parameter αρis a
constant of integration and the ceiling ρmax is the long-run upper limit on
adoption. Benefits do follow a similar logistic pattern to the adoption curve
ρ(t)with parameters αb,βband ρmax . The health benefits are divided into
annual irreversible, ib, and reversible, rb, health benefits with subscript b
for benefits. Irreversible benefits refer to the benefits for children that in
case of VAD could not be reversed in later life, such as premature death,
blindness, stagnant growth, and cognitive capabilities, while reversible
benefits refer to VAD symptoms that can be cured or at least reduced at
every stage of life via an increase in vitamin A supply, such as diarrhea
or infections like measles (West and Darnton-Hill,2008). We have for the
annual irreversible benefits ibt (t)=ibmax
1+exp(−αb−βbt)and annual reversible
benefits rbt (t)=rbmax
1+exp(−αb−βbt). We get for the total irreversible benefits
Ib=∞
0i(t)e−μtdt and for the total reversible benefits Rb=∞
0r(t)e−μtdt,
where μis the discount rate. From this we can deduct the average annual
irreversible health benefits, ib=Ibμand the average annual reversible
health benefits, rb=Rbμ.
The present value of the GRS benefits can be written as
B(Ib,Rb,t)=Ib(ib,t)+Rb(rb,t)=∞
0
ibe−μtdt +∞
0
rbe−μtdt.(1)
The costs of introducing the GRS from the government perspective include
irreversible one-time administrative set-up costs for starting the GRS cam-
paign, Ic, average annual reversible costs for running the campaign, rc,and
additional irreversible perceived costs of the GRS, Gc, with subscript cfor
indicating costs:
C(Ic,Rc,Gc,t)=Ic+Rc+Gc=Ic+Gc+∞
0
rce−μtdt.(2)
The state of nature and related benefits and costs the government faces can
be summarized as follows:
The net-present-value (NPV) of immediate introduction at t=0 with
subscript Ifor immediate:
NPV(GRSI)=Ib−Ic−Gc+∞
0
((rb−rc)e−μtdt)(3a)
8Justus Wesseler and David Zilberman
NPV(GRSI)=Ib−Ic−Gc+(rb−rc)
μ.(3b)
Postponed introduction but perceived costs are not correct, valued at t=0
with subscript Pfor postponed and subscript efor error:
NPV(GRSPe)=q∞
0(Ib−Ic)e−μκ +∞
κ
(rb−rc)e−μtdtf(κ)dκ, (4a)
=q(B−Ic−Rc)h
μ+h.(4b)
Postponed introduction but perceived costs are correct with subscript ne
for no error:
NPV(GRSPne)=((1−q)[0]|B<Cne)=0,(5)
where subscript ne indicates that the perceived costs in this case are at least
as high that B<Cne holds.
The value of a postponed GRS valued at t=0is
NPV(GRSP)=max[NPV(GRSPe), 0].(6)
Considering this setting, the developer faces two possibilities: either the
GRS will be introduced immediately or postponed until time κhas arrived.
On the one hand, the immediate introduction bears the risk that the GRS
results in high perceived costs Gc. On the other hand, postponing the GRS
might cause foregone benefits of saved lives and health, NPV(GRSI)−
NPV(GRSP). If the decision is postponed, the regulator gains additional
information about the perceived costs and will know the true costs of
implementing the GRS. Hence, the net benefits depend on weighing the
benefits and costs of immediate against postponed introduction. Under this
setting, whether or not the GRS will be introduced at time κprovides the
following option value of the GRS strategy:
F[NPV(GRE)]=max[NPV(GRSI), NPV(GRSP)].(7)
The set-up allows us to identify the threshold when a national gov-
ernment might immediately introduce the GRS, which is NPV(GRSI)−
NPV(GRSP)>0, and yields:
B−C−q(Ib−Ic+Rb−Rc)h
μ+h>0.(8)
Solving for B:
B>(Ic+Rc)+μ+h
μ+(1−q)hGc=B∗.(9)
And solving for Gc:
Gc<NPVG
μ+(1−q)h
μ+h=G∗
c,(10)
where NPVG=B−Ic−Rc.
Environment and Development Economics 9
According to the result of equation (9), immediate introduction of the
GRS will be economical if the reversible and irreversible benefits are larger
than the irreversible and reversible costs plus the irreversible perceived
costs of introducing GRS weighted by the leverage factor μ+h
μ+(1−q)h>1. The
first part of the right-hand side of equation (9) captures the standard costs
as part of a benefit-cost analysis; the second part adds the perceived costs.
One unit of perceived irreversible costs of introducing the GRS weighs
more than one unit of other costs. This can be explained by the fact that
Gcis uncertain, and due to the uncertainty about the perceived costs those
costs weigh more than the other costs. This is an effect well known within
the real option literature on regulation (e.g., Hennessy and Moschini,2006;
Ansink and Wesseler,2009).
Equivalently, equation (10) shows that fewer perceived costs will be
needed to outweigh the benefits minus the other reversible and irreversible
costs for explaining a delay in approval. Using this equation, we can iden-
tify a minimum value for the economic power of the opposition towards
the GRS.
We can compare these results with three alternative specifications for
identifying the relative importance of uncertainty about perceived costs
and uncertainty about the arrival date of information or date of decision: no
uncertainty, uncertainty about approval date only, and uncertainty about
perceived costs only. The first specification excludes any uncertainty, the
approval will only be delayed, and at approval time a,GCwill be zero.
As in the previous specification, known benefits and costs remain con-
stant over time. In this case the threshold value G∗
cn with subscript nfor
no uncertainty yields:
G∗
cn =NPVGeμa−1
eμa.(11)
The second specification adds uncertainty over the delay of approval. In
this case the threshold value G∗
ch with subscript hfor uncertainty about
approval date yields:
G∗
ch =NPVGμ
μ+h.(12)
The third specification includes uncertainty about perceived costs only, the
approval decision will only be delayed to a future point in time a, but from
today’s perspective the perceived costs will not be confirmed with prob-
ability q. This yields, for the threshold value of G∗
cq with subscript qfor
uncertainty about the presence of perceived costs,
G∗
cq =NPVGeμa−q
eμa.(13)
Comparing equations (10)to(13) we can observe that including uncertainty
about the approval date results in the lowest threshold value for per-
ceived costs as long as eμa>μ+h
hfollowed by excluding any uncertainty,
10 Justus Wesseler and David Zilberman
whereas considering uncertainty about perceived costs only results in a
higher threshold than considering uncertainty about perceived costs in
combination with uncertainty about the arrival time of additional infor-
mation about whether or not the perceived costs will be confirmed. While
the differences between equation (11) and (12), and (10) and (13)aresub-
stantial for q=0.5, the differences between equation (10) and (13), and
equation (11) and (12) are less pronounced and for some parameter values
may even be reversed. In essence, ignoring any uncertainty substantially
undervalues the perceived costs, while ignoring uncertainty about deci-
sion dates slightly overvalues the perceived costs. This will become more
obvious in section 3 where the model will be calibrated for the case of India.
2.1. Regulations and delay strategies
The result of equation (9) also provides economic insight into the success
of delay strategies of opponents to the GRS. One might be surprised by
the fact that technology of this kind will not immediately be introduced
but delayed by several means. Many developing countries either have or
are discussing additional regulations for the approval of GE seed varieties.
Most countries including those targeted for VAD have stringent biosafety
regulations for the approval of GM crops (Paarlberg,2008). The compliance
with those regulations costs additional time and delays introduction. Take
as an example India, whose government implemented a working group
to assess biotechnology, which recommended that the Indian Government
ban all GM food crops from cultivation, thereby stirring a debate about the
approval process of GM crops.
By looking at equation (9), an increase in h(∂( μ+h
μ+(1−q)h)/∂h>0), equiv-
alent to availability of information about the perceived costs will soon
become available, increases the weight of the perceived costs. In this sense,
it pays for opponents to again raise concerns about perceived costs via dif-
ferent forms of protest or new but often unfounded claims about negative
implications, when a decision to introduce the GRS is soon to be made.
This is observed not only for Golden Rice, but in general for the approval
of GE crops in many developing countries (e.g., Paarlberg,2008 for Africa;
Herring,2010,2012 for India) as well as the European Union (Herring,2007;
Wesseler et al.,2012).
3. Calibration of the model: the foregone benefits
The previous results will be used to quantify the foregone benefits of a
delayed introduction of the GRS as well as the perceived costs by the Gov-
ernment of India. The model will be calibrated for India thanks to the study
by Stein et al. (2008), in which detailed information about potential benefits
and costs of the GRS for India are available. We consider the time period
from 2002 when the technology was available and could have been intro-
duced. For the purpose of our analysis, irreversible and reversible benefits
and costs have been calculated based on Stein (2006; personal communi-
cation, 2013) and Stein et al. (2008), while the numbers employed for the
‘pessimistic scenarios’ have been on the lower side.
Environment and Development Economics 11
The benefits of the GRS are the reduced health costs of VAD. Those
benefits, similar to other health benefits, are commonly assessed by cal-
culating the DALYs (Murray and Lopez,1996). The current burden of VAD
has been calculated with about 2.3 million DALYs per year (over a 10-year
period: about 23 million DALYs ignoring changes in the Indian popula-
tion) (Stein et al.,2008, table 4). These include the burden of night blindness,
corneal scars, blindness caused by corneal scars, measles and mortality of
children five years old and younger, and night blindness for pregnant and
lactating women. Corneal scars and blindness caused by corneal scars of
children five years old and younger are VAD-related burdens that cannot
be reversed and are considered to be irreversible. Reducing those burdens
through the GRS is an irreversible benefit. Further, reducing child mortality
through the GRS has been considered an irreversible benefit too; accord-
ing to the numbers provided by Stein (2006, table A3), this amounts to
about 71,600 child deaths annually.
The other health burdens can be reduced through the GRS. The benefits
can be considered to be reversible as they normally do not result in health
problems in later life. Using the information provided by Stein (2006), the
share of irreversible and reversible health benefits can be calculated. About
74.4 and 25.6 per cent of the health benefits can be considered to be irre-
versible and reversible, respectively. Please note that the absolute numbers
are on the lower side as a number of health effects for which a causal link
has not yet been fully established have not been considered, such as stunted
growth. Further, as an economic value of a DALY, a rather low value of
US$500 has been used in the study. A low value can be justified in order
not to overestimate the benefits of the GRS, but places a low value on the
average statistical value of life and higher values would further increase
the economic benefits of a GRS.4
On the costs side, a number of cost items have been discussed within
the literature. These include research and development costs at the inter-
national and national level but also costs for launching a media campaign
to introduce the GRS and annual costs for maintenance breeding. While
the costs for the media campaign will definitely arise and will be addi-
tional costs for the government, the annual costs for maintenance breeding
are less obvious if they are additional. The media campaign, called social
marketing, will be needed at the beginning when the GRS is introduced.
The costs are treated as irreversible as they are sunk and will not matter at
a later stage whether or not to continue the GRS, while the maintenance
breeding costs still matter. Maintenance breeding by national agricul-
ture research centers is a regular activity. If the Golden Rice trait were
to be been introduced in Indian rice lines, they would be part of the
ongoing maintenance breeding and additional costs would not arise. On
the other hand, one can argue that the efforts for maintenance breeding
4A more detailed discussion on calculating DALYs within the debate on a GRS
and other biofortification strategies has been provided by Qaim et al. (2007). A
criticism on the use of DALYs to allocate resources for health projects can be found
in King and Bertino (2008).
12 Justus Wesseler and David Zilberman
Table 1. DALYs, benefits and costs of a Golden Rice Strategy to address Vitamin
A deficiency in India, costs of delay, and minimum perceived costs
Benefits and costs in present values DALY (’000) US$ (’000)
Irreversible health benefits (74.42%) 2,071,601
Reversible health benefits (25.58%) 711,999
Irreversible costs (social marketing) 15,554
Reversible costs (maintenance breeding) 4,292
Net-present-value (as of beginning of 2002) 2,763,755
Net-present-value (as of beginning of 2012,
valued at beginning of 2002)
2,056,493
DALY saved if introduced in 2002 5,567
DALY saved (as of beginning of 2012, valued
at beginning of 2002)
4,142
DALY lost over the past decade 1,425 (2,041)a
Notes:aDALY not discounted.
might increase as the number of rice lines has increased. Therefore cost
for maintenance breeding has been considered and treated as annual
reversible costs. According to Stein (personal communication, 2013), these
costs have been assessed to be about US$15,553,610 and about US$125,000
for the social marketing and the additional annual maintenance breeding
costs, respectively. Table 1summarizes the benefits and costs used for the
assessment.
Using the benefits and costs provided under the ‘pessimistic scenario’,
implying a low impact of the GRS, provides a net-present-value for
NPV(GRSI)(number in thousand US$):5
NPV(GRSI)=2,071,601 −15,554 −Gc+711,999 −4,292
=2,763,755 −Gc.(14)
The value for NPV(GRSP)=q(B−Ic−Rc)h
μ+h, using a value for hof 0.1, equiv-
alent to an expected decision being made after 10 years, value for qof 0.5,
and a discount rate of three per cent in discrete time (2.955 per cent in con-
tinuous time) as also used by Stein (2006) and commonly done for such
kinds of health benefit assessments, provides the following result in US$:
NPV(GRSP)=0.5(2,071,601 −15,554 +711,99 −4,292)0.1
0.02955 +0.1
NPV(GRSP)=1,066,602,416.(15)
5Please note that we use an infinite stream of annual benefits and costs while Stein
(2006) uses a 30-year period. Even for the 30-year period the benefits of the tech-
nology are larger than the costs in the ‘pessimistic’ scenario. In this case, gains
from waiting do not exist if perceived costs are zero.
Environment and Development Economics 13
Using equation (10), the critical value G∗
Cin US$:
G∗
c=(B−Ic−Rc)μ+(1−q)h
μ+h
=2,763,7550.02955 +0.5·0.1
0.03 +0.1=1,697,152,214.(16)
This threshold level of about US$1.7 billion is substantial, or if annualized
over a 10-year period is about US$199 million per year, and is an indicator
of the economic power of GMO opposition in India. This is a minimum
value as this only indicates what perceived costs at least have to be in order
to explain a delay in approval following the specification of the model.
A comparison with the two alternative specifications discussed in
section 2 provides the following threshold value in case of no uncertainty
by using equation (11):
G∗
cn =NPVGeμa−1
eμa=2,763,755 e0.2955 −1
e0.2955 =707,261,628 (17)
In case of uncertainty with respect to the approval date only provides the
following threshold value in US$ by using equation (12):
G∗
ch =NPVGμ
μ+h=2,763,755 0.02995
0.02995 +0.1=630,549,798.(18)
In case of uncertainty with respect to the perceived costs only provides the
following threshold value in US$ by using equation (13):
G∗
cq =NPVGeμa−q
eμa=2,763,755 e0.2955 −0.5
e0.2955 =1,735,508,129.
(19)
The differences in the weighing factors between the four specifications
are substantial when comparing the results of equation (16)to(19). The
weighing factor in equation (17) is 0.26 and 0.23 in equation (18), while in
equation (16) the factor is 0.61 and 0.63 in equation (19). The difference
in the weighing factors between equation (17) and (18), and equation (16)
and (19) is a factor ranging between 2.40 and 2.75; the difference between
these sets of equations is less pronounced with a factor of 1.12 and 1.02,
respectively. A summary and comparison of the results with respect to the
different model specification is provided in table 2.
The results also illustrate the importance of information timing. The
importance of perceived costs increases the sooner the date of new infor-
mation arrives (figure 1). We can observe a sharp increase in the leverage
effect for values of hbelow about 0.5 (two years). Figure 1and equation (10)
show that the leverage effect increases with an increase in q.
The reported results are the minimum amount based on the pessimistic
or worst case scenario reported by Stein et al. (2008), who also report ben-
efits and costs for an optimistic scenario. In that case the threshold level
14 Justus Wesseler and David Zilberman
Table 2. Results of different model specifications
ModelaG∗
ch G∗
cn G∗
cG∗
cq
Threshold level (US$) 630,549,798 707,261,628 1,697,152,214 1,735,508,129
Weighing factor of Gc0.2281 0.2559 0.6141 0.6280
Comparison of weighing factors
G∗
cn,G∗
c,G∗
cq over G∗
ch 1.1217 2.6915 2.7524
G∗
c,G∗
cq over G∗
cn 2.3996 2.4538
G∗
cq over G∗
c1.0226
Notes:aG∗
ch, alternative specification with uncertainty about approval date and
no uncertainty about perceived costs; G∗
cn, alternative specification with no
uncertainty about perceived costs and approval date; G∗
c, specification with
uncertainty about perceived costs and uncertainty about arrival date of infor-
mation about perceived costs; G∗
cq , alternative specification with uncertainty
about perceived costs but certain arrival date of confirmation.
0.8000
1.3000
1.8000
2.3000
2.8000
3.3000
3.8000
0.0000 0.5000 1.0000 1.5000 2.0000
Leverage Factor
h -value
Leverage Facto
r
(q=0.3)
Leverage Facto
r
(q=0.5)
Leverage Facto
r
(q=0.7)
Figure 1. Increase in leverage factor with a decrease in arrival time of new information
for different q-values (μ=0.03)
increases to about US$11.6 billion, or if annualized over a 10-year period to
about US$1,360 million per year.6
4. Discussion and conclusion
Nutritional and economic ex ante assessment studies of a GRS have shown
that Golden Rice can reduce VAD-related mortalities and diseases at less
cost than alternative strategies discussed in the literature. Previous studies
for India have shown that about 204,000 life years can be saved annu-
ally. Golden Rice was expected to be introduced in 2002. Golden Rice has
6Details of the optimistic calculation are available upon request from the authors.
Environment and Development Economics 15
not yet been approved in any country, including India. According to our
calculations, the delay over the last 10 years has caused losses of at least
1,424,680 life years for India, ignoring indirect health costs of VAD.
The differences in net present value from a 10-year delay are about
US$707 million. This difference does not fully capture the minimum
amount of perceived costs that the Government of India places on the intro-
duction of a GRS. Considering uncertainty and irreversibility substantially
increases the minimum amount of these perceived costs. Our calculation
shows that the additional perceived costs by the Government of India are
at least US$1.7 billion (about US$199 million annually).
This is a substantial amount and reflects the economic power of the
opposition against the introduction of Golden Rice and explains why it is
more difficult to convince regulators when a strong vocal opposition exists
that mainly stirs uncertainty about the GRS.
The comparison with alternative specifications shows that uncertainty
about the irreversible costs is economically substantially more important
than uncertainty about a specific date. Understanding and quantifying
the causes of uncertainty about irreversible costs seems to be economi-
cally more important for assessing decision making than uncertainty with
respect to the timing of decision making or the arrival of new information
in general. Uncertainty with respect to the arrival of new information κ
has been modeled by using an exponential distribution, which is mem-
oryless and hence constant with respect to time. Other functional forms
such as a Weibull or log-normal distribution are not necessarily memory-
less (Billingsley,2012). Nevertheless, we expect that the quality of the result
will not change, but leave it for further research to assess the implications
in more detail.
The size of the perceived costs is substantially larger, 85 times, than the
cost of implementing the GRS. Having a better understanding of the polit-
ical economy behind the perceived costs and how to reduce them seems to
be economically much more important than additional investigations into
the costs of social marketing and maintenance breeding.
The results further show that it pays for those opposing the GRS to
raise concerns about the technology the sooner a decision by regulators
is expected. The leverage factor of the perceived costs increases the closer
the point of decision making is. This explains why the opposition to the
GRS has substantial power and indicates that it will be difficult for those
supporting the technology to change the view on perceived costs. In this
context it is not so important to provide factual evidence, but to raise
uncertainty. The opposition to GRS in India was able to link the GRS
with the overall debate over GMOs in general (Enserink,2008). Narratives
about farmer suicides and dead sheep linked to GM cotton cultivation,
environmental damages of Bt eggplant, and health damages linked to
antibiotic marker genes are used to stir uncertainty among decision mak-
ers. Never mind that the narratives have not been correct (Herring,2010),
keeping them alive is sufficient. It has also been argued that the GRS is
only an industry public relations strategy as the multinational company
Syngenta is involved in further developing the technology to convince a
skeptical society about the benefits of GM crops. If Golden Rice receives
16 Justus Wesseler and David Zilberman
approval, other GM crops can be introduced more easily. In this case
uniform regulation – i.e., to ban the cultivation of all GM food crops –
might be a cost-effective approach to avoid the introduction of potentially
more ‘dangerous’ crops. While this argument has received support within
the environmental economics literature (e.g., Kolstad,1987), as identifying
the marginal benefits and damage costs of a technology can be very costly,
an important difference needs to be considered. Real damages to the envi-
ronment from technologies assessed within the environmental economics
literature have been demonstrated, such as damages from sulfate dioxide
emission from steel factories or nitrogen emissions from intensive agricul-
ture, while in the case of GM crops, net environmental benefits have been
reported (e.g., Wesseler et al.,2011;Bennett et al.,2013).
One may then ask how new technologies can be introduced, if it is so
easy to block them. There will always be a small, vocal group oppos-
ing a new technology. On the one hand, this is correct. Introducing new
technologies will become more difficult. The advances being made over
the past decade in information and communication systems reduce the
costs of organizing the opposition as well as the costs of communicating
technology-related uncertainties. On the other hand, uncertainties over a
new technology can be balanced by uncertainties over not having access
to a new technology stressed by a group of stakeholders as powerful as
the opposition. This seems to explain why transgenic crops have been
introduced within the US but not in Europe (Graff et al.,2013).
A countervailing power supporting the GRS and stirring uncertainties
about not introducing the GRS has not been explicitly considered within
the model presented, as this is currently not present within the GRS debate.
The GRS has been developed by scientists. Scientists have the tendency to
argue based on facts and not fiction, making it more difficult to be a coun-
tervailing power against the opposition to Golden Rice. Also, uncertainties
and irreversibilities about the benefits and the other costs of the GRS can be
included in the model. As a worst case scenario has been employed with
respect to those benefits and costs, modifications of the model in the afore-
mentioned direction will increase the calculated perceived costs. Further,
recent nutritional studies show the factors translating β-carotene of Golden
Rice into vitamin A are larger than expected (Tan g et al.,2012).
One question that remains to be answered within this debate is: what are
the incentives of the opposition to the GRS in India? This has not yet been
well investigated empirically. Apel (2010) argues that this is driven by the
financial support the opposition receives. A small industry has developed
around the opposition to transgenic crops that survives mainly on dona-
tions and has to keep the debate about the risks of the technology alive.
This strategy seems to be a successful strategy albeit, as the case of Golden
Rice shows, at the cost of the lives of several thousand children.
References
Ansink, E. and J. Wesseler (2009), ‘Quantifying type I and type II errors in decision-
making under uncertainty: the case of GM crops’, Letters in Spatial and Resource
Sciences 2: 61–66.
Environment and Development Economics 17
Apel, A. (2010), ‘The costly benefits of opposing agricultural biotechnology’, New
Biotechnology 27: 635–640.
Arrow, K.J. and A.C. Fisher (1974), ‘Environmental preservation, uncertainty, and
irreversibility’, Quarterly Journal of Economics 88: 312–319.
Arrow, K., M. Cropper, G. Eads, et al. (1996), ‘Is there a role for benefit-
cost analysis in environmental, health, and safety regulation?’, Science 272:
221–222.
Baten, J., D. Crayen, and H.-J. Voth (2010), ‘Poor, hungry and ignorant: numeracy
and the impact of high food prices in industrializing Britain, 1780–1850’, Univer-
sity of Tuebingen Working Paper, [Available at] http://www.econ.upf.edu/docs/
papers/downloads/1120.pdf.
Batista, R., N. Saibo, T. Lourenco, and M. Oliveira (2008), ‘Microarray analyses
reveal that plant mutagenesis may induce more transcriptomic changes than
transgene insertion’, Proceedings of the National Academy of Science USA 105:
3640–3645.
Baudo, M., R. Lyons, S. Powers, G. Pastori, K. Edwards, M. Holdsworth, and
P. Shewry (2006), ‘Transgenesis has less impact on the transcriptome of wheat
grain than conventional breeding’, Plant Biotechnology Journal 4: 369–380.
Bennett, A.B., C. Chi-Ham, G. Barrows, S. Sexton, and D. Zilberman (2013), ‘Agri-
cultural biotechnology: economics, environment, ethics, and the future’, Annual
Review of Environment and Resources 38: 19.1–19.31.
Billingsley, P. (2012), Probability and Measure, Hoboken, NJ: John Wiley.
Case, A. and C. Paxson (2006), ‘Stature and status: height, ability, and labor market
outcomes’, Journal of Political Economy 116: 499–532.
Committee on Agriculture (2012), Cultivation of Genetically Modified Food Crops –
Prospects and Effects, New Delhi: Lok Sabha Secretariat.
Copenhagen Consensus (2008), Copenhagen Consensus 2008, [Available at] http://
www.copenhagenconsensus.com/Projects/Copenhagen%20Consensus%202008-
1.aspx.
Copenhagen Consensus (2012), Copenhagen Consensus 2012, [Available at] http://
www.copenhagenconsensus.com/Projects/CC12.aspx.
Currie, J. and R. Hyson (1999), ‘Is the impact of health shocks cushioned by
socioeconomic status? The case of low birth weight’, American Economic Review
89: 245–250.
Deaton, A. (2003), ‘Health, inequality, and economic development’, Journal of Eco-
nomic Literature 41: 113–158.
Dixit, A. and R. Pindyck (1994), Investment Under Uncertainty, Princeton, NJ: Prince-
ton University Press.
Doblhammer, G. and J. Vaupel (2001), ‘Lifespan depends on month of birth’,
Proceedings of the National Academy of Sciences USA 98: 2934–2939.
Enserink, M. (2008), ‘Tough lessons from Golden Rice’, Science 320: 468–471.
Fogel, R.W. (1994), ‘Economic growth, population theory and physiology: the bear-
ing of long-term processes on the making of economic policy’, American Economic
Review 84: 369–395.
Graff, G., G. Hochman, and D. Zilberman (2009), ‘The political economy of agricul-
tural biotechnology policies’, AgBioForum 12: 34–46.
Graff, G., G. Hochman, and D. Zilberman (2013), ‘The political economy of tech-
nological regulation in the face of creative destruction: the case of agricultural
biotechnology’, in R. Herring (ed.), Oxford Handbook on Food, Politics and Society,
Oxford: Oxford University Press.
Greenpeace (2005), Golden Rice: All Glitter, No Gold, Amsterdam: Greenpeace Inter-
national.
Greenpeace (2012), Golden Illusion: The Broken Promises of ‘Golden’ Rice, Amsterdam:
Greenpeace International.
18 Justus Wesseler and David Zilberman
Hennessy, D. and G. Moschini (2006), ‘Regulatory actions under adjustment costs
and the resolution of scientific uncertainty’, American Journal of Agricultural
Economics 88: 308–323.
Herman, R., B. Chassy, and W. Parrott (2009), ‘Compositional assessment of trans-
genic crops: an idea whose time has passed’, Trends in Biotechnology 27: 555–557.
Herring, R. (2007), ‘Opposition to transgenic technologies: ideology, interests and
collective action frames’, Nature Review Genetics 9: 458–463.
Herring, R. (2010), ‘Epistemic brokerage in the bio-property narrative: contribu-
tions to explaining opposition to transgenic technologies in agriculture’, New
Biotechnology 27: 614–622.
Herring, R. (2012), ‘State science and its discontents: why India’s second transgenic
crop did not follow the path of Bt cotton’, paper presented at the Weihenstephaner
Socio-Economic Seminar, Center of Life and Food Sciences Weihenstephan, Tech-
nische Universit¨
at M ¨
unchen, 13 June.
King, C.H. and A.M. Bertino (2008), ‘Asymmetries of poverty: why global burden of
disease valuations underestimate the burden of neglected tropical diseases’, PLoS
Neglected Tropical Diseases 2: e209.
Kolstad, C. (1987), ‘Uniformity versus differentiation in regulating externalities’,
Journal of Environmental Economics and Management 14: 386–399.
Murray, C. and A. Lopez (1996), The Global Burden of Disease,Vol.1,Cambridge,MA:
Harvard University Press.
Paarlberg, R. (2008), Starved for Science: How Biotechnology is Being Kept Out of Africa,
Cambridge, MA: Harvard University Press.
Potrykus, I. (2010a), ‘Regulation must be revolutionized’, Nature 466: 561.
Potrykus, I. (2010b), ‘Lessons from the “Humanitarian Golden Rice” project: regula-
tion prevents development of public good genetically engineered crop products’,
New Biotechnology 27: 466–472.
Qaim, M., A. Stein, and J.V. Meenakshi (2007), ‘Economics of biofortification’,
Agricultural Economics 37(s1): 119–133.
Ramessar, K., A. Peremarti, S. Gomez-Galera, S. Naqvi, M. Moralejo, P. Munoz,
T. Capell, and P. Chritou (2007), ‘Biosafety and risk assessment framework for
selectable marker genes in transgenic crop plants: a case of the science not
supporting the politics’, Transgenic Research 16: 261–280.
Richards, M., R. Hardy, D. Kuh, and M. Wadsworth (2002), ‘Birth weight, postnatal
growth and cognitive function in a national UK birth cohort’, International Journal
of Epidemiology 31: 342–348.
Shewry, P., M. Baudo, A. Lovegrove, and S. Powers (2007), ‘Are GM and conven-
tionally bred cereals really different?’, Trends in Food Sciences and Technology 18:
201–209.
Stein, A. (2006), Micronutrient Malnutrition and the Impact of Modern Plant Breeding
on Public Health in India: How Cost-effective is Biofortification?,G
¨
ottingen: Cuvillier
Verla g .
Stein, A., H. Sachdev, and M. Qaim (2008), ‘Genetic engineering for the poor: Golden
Rice and public health in India’, World Development 36: 144–158.
Swinnen, J.F.M. and T. Vandemoortele (2010), ‘Policy gridlock or future change?
The political economy dynamics of EU biotechnology regulation’, AgBioForum 13:
291–296.
Tang, G., J. Qin, G. Dolnikoawski, R. Russell, and M. Grusak (2009), ‘Golden Rice
is an effective source of vitamin A’, American Journal of Clinical Nutrition 89:
1176–1183.
Tang, G., Y. Hu, S. Yin, Y. Wang, G. Dallal, M. Grusak, and R. Russell (2012),
‘β-Carotene in Golden Rice is as good as β-carotene in oil at providing vitamin
A to children’, American Journal of Clinical Nutrition 96: 658–664.
Environment and Development Economics 19
Taylor, H. and S. Karlin (1984), An Introduction to Stochastic Modeling, Orlando, FL:
Academic Press.
Thomas, D. and E. Frankenberg (2002), ‘Health, nutrition and prosperity: a microe-
conomic perspective’, Bulletin of the World Health Organization 80: 106–113.
van den Belt, H. (2003), ‘Debating the precautionary principle: “guilty until proven
innocent” or “innocent until proven guilty”?’, Plant Physiology 132: 1122–1126.
Vigani, M. and A. Olper (2014), ‘GM-free private standards, public regulation of GM
products and mass media’, Environment and Development Economics, this issue.
Wesseler, J. (2009), ‘The Santaniello theorem of irreversible benefits’, AgBioForum 12:
8–13.
Wesseler, J., S. Scatasta, and E. Fall (2011), ‘Environmental benefits and costs of GM
crops’, in C. Carter, G.C. Moschini, and I. Sheldon (eds), Genetically Modified Food
and Global Welfare, Bingley: Emerald Group Publishing, pp. 173–199.
Wesseler, J., S. Leimgruber, and R. Smart (2012), ‘Comparison of time periods in
the approval processes for GM crops in the EU, USA, Canada & South Africa’,
Paper presented at the 16th ICABR Conference, 128th EAAE Seminar, Ravello,
Italy, 24–27 June.
West, K. and I. Darnton-Hill (2008), ‘Vitamin A deficiency’, in R.D. Semba and M.W.
Bloem (eds), Nutrition and Health in Developing Countries, Totowa, NJ: Humana
Press, pp. 377–433.
World Health Organization (2001), Macroeconomics and Health: Investing in Health
for Economic Development. Report of the Commission on Macroeconomics and Health,
Geneva: World Health Organization.
Zimmermann, R. and M. Qaim (2004), ‘Potential health benefits of Golden Rice: a
Philippine case study’, Food Policy 29: 147–168.
