Marketing Fortified Rice Effects of Aspirational Messaging and Association With Free Distribution

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Food Policy 112 (2022) 102346

Contents lists available at ScienceDirect

Food Policy
journal homepage: www.elsevier.com/locate/foodpol

Marketing fortified rice: Effects of aspirational messaging and association


with free distribution
Reajul Chowdhury a, *, Benjamin Crost b, Vivian Hoffmann c, d
a
Amazon, Seattle, USA
b
Department of Economics, University of Calgary, Canada
c
International Food Policy Research Institute, USA
d
School of Public Policy and Administration and Department of Economics, Carleton University, Canada

A R T I C L E I N F O A B S T R A C T

Keywords: Fortification of staple foods is among the most cost-effective public health interventions. In settings where
Aspirational messaging mandatory fortification is not feasible, providing fortified foods through public safety net programs and building
Willingness-to-pay private demand can be implemented in parallel to maximize coverage. However, little is known about what
Adoption
marketing messages are most effective at increasing the demand for fortified foods. Further, as governments scale
Reference price effects
Fortification
up fortification of foods distributed through safety net programs, understanding the effect of knowledge about
this on consumer demand will be important for crafting communication strategies. Drawing on insights from the
literature on marketing and consumer behavior, we hypothesize that a message linking fortified rice, an unfa­
miliar product in this setting, to a known, aspirational product, can increase demand. We test the individual and
joint impacts of an aspirational marketing message and information about the public distribution of fortified rice
through a framed field experiment eliciting willingness to pay for rice fortified with six micronutrients in
Chandpur District, Bangladesh. We find that a message which combines aspirational positioning and information
on public distribution increases the proportion of participants willing to pay a premium at least equal to the cost
of fortification by 19 percentage points. The results suggest that any negative impact on willingness to pay
associated with free distribution is outweighed by the credibility distribution lends to health claims about the
product.

1. Introduction As fortification of rice remains voluntary in most contexts, under­


standing how to build consumer demand for fortified rice is critical to
Micronutrient deficiency or ‘hidden hunger’ affects more than 2 realizing the potential impact of this technology. While the literature on
billion people worldwide (von Grebmer et al., 2017). Fortification of consumer willingness to pay for fortified foods shows the importance of
staple foods with vitamins and minerals is considered the most cost- providing nutritional information, evidence on the impact of comple­
effective way to address population-level micronutrient deficiency, mentary marketing messages is thin and inconclusive (Birol et al., 2015).
and has repeatedly been ranked by a global panel of experts as among This paper attempts to address this gap in the literature by offering
the top three solutions to advance global welfare, in terms of costs versus answers to two questions: can the demand for fortified rice be increased
benefits (Copenhagen Consensus, 2012, 2008). Fortification of rice – the by positioning it as an aspirational health product, and how does asso­
main staple food in Asia and the source of 80 percent of calories for the ciating the rice with a social safety net program affect demand.
population there living in poverty – is considered a cost-effective and In particular, we tested the effect of two messages on demand, as
sustainable solution to reduce the micronutrient deficiency in the region measured through an experimental auction procedure, for fortified rice
(Harding et al., 2018). However, as of 2017, only seven countries in a highly vulnerable population in rural Bangladesh. Drawing on the
globally mandate rice fortification through legislation (Mkambula et al., marketing literature, the first message was designed to facilitate con­
2020). sumers’ understanding of and demand for fortified rice, an unfamiliar

* Corresponding author at: Department of Agriculture and Consumer Economics, University of Illinois at Urbana-Champaign, 405 Mumford Hall, Urbana, IL 61801,
USA.
E-mail addresses: reajul.alamchy@gmail.com, mac7@illinois.edu (R. Chowdhury).

https://doi.org/10.1016/j.foodpol.2022.102346
Received 10 February 2021; Received in revised form 17 August 2022; Accepted 23 August 2022
Available online 10 September 2022
0306-9192/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
R. Chowdhury et al. Food Policy 112 (2022) 102346

concept in this population, by linking it to a known, aspirational prod­ farmers perceive an advantage to biofortified varieties (Vaiknoras et al.,
uct: a fortified nutritional drink. Several studies have argued that con­ 2019). In Bangladesh, a recent study found that consumer acceptance of
sumer learning about unfamiliar products can be facilitated through the bio-fortified rice was high and that information on its nutritional qual­
use of analogies to an existing and familiar product set (Kuijken et al., ities increased willingness to pay by 4.6 percent (Herrington et al.,
2017; Gregan-Paxton and Moreau, 2003; Hoeffler, 2003; Moreau et al., 2021). According to HarvestPlus estimates, 13 percent of all rice-
2001; Gregan-Paxton and John, 1997). Drawing analogies to a high- farming households were growing zinc biofortified rice in 2020 (Har­
value product can influence consumers’ monetary value assessment of vestPlus, 2021, as cited by Herrington et al., 2021).
a product (Kuijken et al., 2017) while also tapping into their aspirations. A disadvantage of biofortification based on conventional breeding
A recent review found significant aspirational consumption among so- techniques is that these varieties tend to target a single micronutrient,
called “base of the pyramid” (BoP) consumers living on between 2 and and may not reach the concentration required to reach the recom­
8 USD per day (Srivastava et al., 2020), and a survey of marketing mended daily intake based on typical consumption patterns (Van Der
practitioners targeting this population identified the use of aspirational Straeten et al., 2020). The biofortified rice varieties available in
messages – messages positioning a product as something the consumer Bangladesh only have elevated levels of zinc, just one of the six micro­
would wish to use but may not currently able to – as a common approach nutrients included in a typical industrial rice fortificant (Andrade et al.,
(Beninger and Robson, 2015). However, the impact of aspirational 2021).1 While it may be feasible to engineer crops with a stronger
product positioning on demand has not, to our knowledge, been tested micronutrient profile (Van Der Straeten et al., 2020), GM technologies
experimentally in a low and middle-income country (LMIC) setting. face strict regulatory control based on potential ecological and health
The second message informed consumers about the distribution of risks (Jeon et al., 2021; Kok et al., 2018). At the time of writing, only one
fortified rice through a major public safety net program in Bangladesh. country, the Philippines, had approved planting a GM bio-fortified crop,
Provision of fortified foods to vulnerable populations is a core strategy Golden Rice.2
used by the Government of Bangladesh and other governments in the While industrial fortification is able to provide a larger number of
region to address malnutrition (Department of Food & Public Distribu­ micronutrients at greater concentration, and typically faces fewer reg­
tion, Government of India, 2021; Andrade et al., 2021). This fact is at ulatory hurdles than biofortification, constraints in manufacturing ca­
odds with, and could potentially erode the effect of, aspirational posi­ pacity, quality control, and poor enforcement of regulatory standards
tioning of fortified foods by positioning this as food ‘for the poor’ – a limit the reach and quality of fortified foods in many low and middle-
perception explicitly avoided by a social marketing campaign for forti­ income countries (Olson et al., 2021). Also, because fortification re­
fied rice in Brazil, even though it targeted low-income consumers quires an additional input (fortificant) and specialized machinery to
(Milani et al., 2017). In addition to contradicting aspirational product blend this into the fortified food, it entails a cost, which consumers may
positioning, the free distribution could be interpreted by consumers as a not be willing to bear. In settings where mandatory fortification is not
signal of low product quality (Darke and Chung, 2005) or lead to a low feasible, designing effective social marketing strategies to promote de­
anchoring of consumers’ price expectations (Kőszegi and Rabin, 2006). mand for fortified foods is thus critical.
On the other hand, public distribution of fortified foods could signal
government endorsement and thus confer the importance of fortifica­ 1.2. Consumer demand for fortification
tion, perhaps by bolstering the credibility of claims about its health
benefits. Neither participants in the experiment, nor their social con­ While mandating fortification of staples through legislation is the
nections, were likely to have had prior experience of the health benefits primary strategy for increasing consumption of fortified foods, a
of consuming fortified rice, making it a credence good from their voluntary fortification combined with social marketing has been used in
perspective. Previous literature suggests that consumers tend to rely on some settings (Milani et al., 2017; Wang et al., 2009). The market
external cues such as brand name, price, or third party endorsements as penetration, and resulting public health impact, of voluntarily fortified
signals of the credibility of claims made about credence goods (Bloom foods varies widely (Lalani et al., 2019).
and Pailin, 1995; Zeithaml, 1988). Previous studies estimating consumer willingness to pay for either
In the remainder of this section, we provide a brief overview of op­ industrially or biofortified staples in low-income settings through
tions for addressing micronutrient malnutrition and the advantages and incentivized experimental auctions and choice experiments have found
disadvantages of each of these. We then review the literature on con­ substantial premiums for fortification (Chowdhury et al., 2011; De
sumer acceptance and willingness to pay for fortified foods. Finally, we Groote et al., 2011). Experimenter demand effects or social desirability
describe the contribution of this paper in the context of related bias may influence these estimates, and more recent work showing a
literature. decay in the effect of a health-related label on consumer food choice
over time (Hoffmann et al., 2021) implies that fortification premiums
1.1. Options for addressing micronutrient deficiencies estimated in this way – including through the present study – should be
interpreted with caution. In contrast, we estimate the effect of alterna­
Micronutrient malnutrition can be addressed by improving dietary tive marketing messages by comparing WTP among consumers who bid
diversity, by supplementing the diet with concentrated doses of micro­ on a fortified product, differencing out any overall bias associated with
nutrients, or through mass fortification of staple foods or regularly fortification.
consumed condiments. Mass fortification can be achieved either through A rapidly growing experimental literature evaluates social marketing
biofortification, which improves the bioavailability of micronutrients strategies to build consumer demand for biofortified crop varieties. Ef­
through agronomic practices, traditional plant breeding, or genetic fects on demand of varying different aspects of the information provided
modification (GM), or through industrial fortification, which entails to consumers, including content (nutrition benefits), framing (gains of
adding a fortificant during post-harvest processing. consumption vs losses from non-consumption), medium (radio vs com­
Both biofortification and industrial fortification have strong poten­ munity leaders), length (short vs detailed), and frequency, have been
tial to improve nutritional outcomes in Bangladesh (Ara et al., 2019; De tested. Pre-publication versions of several of these studies are reviewed
Moura et al., 2016), and each has advantages and disadvantages. A by Birol et al. (2015). The impact of providing nutrition information on
major advantage of biofortification is that it can reach populations that
rely on unprocessed grain, including those who produce the food they
consume. Further, the recurrent costs of biofortified varieties are 1
These are: vitamin-A, thiamin (B1), cobalamin (B-12), folic acid (B-9), iron
generally no higher than those of their non-fortified analogs (Olson and zinc.
et al., 2021). Adoption thus can increase rapidly in contexts where 2
https://www.goldenrice.org/.

2
R. Chowdhury et al. Food Policy 112 (2022) 102346

the willingness to pay is consistently strong and positive across studies. fortified with vitamin A, vitamin B1, vitamin B12, folic acid, iron, and
The source of this information is not important (Meenakshi et al., 2012), zinc to ultra-poor households in selected pilot districts through the
nor is the level of detail (Birol et al., 2015), but message repetition does Vulnerable Group Development program within the Public Food Dis­
increase demand (Oparinde et al., 2019). Results on the source of cer­ tribution System (PFDS) since 2014. A recent study found that after 12
tification or planting material are mixed, with consumers preferring months, anemia prevalence among women in households receiving this
international (versus state-level) certification in one context (Banerji rice dropped by 11 percent, and zinc deficiency dropped by 6 percent
et al., 2016), but no impact on demand of the source of planting material compared to those in non-beneficiary households (Ara et al. 2019). The
(national government versus an international authority) in another provision of fortified grains through the PFDS is set to expand, and
(Oparinde et al., 2016). Varying the framing of the effect of fortification policymakers have expressed interest in making fortified rice available
by using a gain (benefits of consumption) versus loss (harms of non- through the private market as well.
consumption) orientation had no effect on demand (Oparinde et al., Fortified rice in Bangladesh constitutes an ideal case for studying the
2019). effect of an analogical aspirational message on consumer demand. First,
consumers in this setting generally, and in particular those in the study
1.3. Contribution sample, are unfamiliar with fortified rice. To our knowledge, the only
company to sell fortified rice in Bangladesh did so for just a few months
We make two contributions to the existing literature. First, we pre­ in 2017 before discontinuing the product due to lack of demand. During
sent new evidence on the effect of aspirational messaging in stimulating this short time, the fortified brand was marketed as a premium product
demand for a health good. To our knowledge, the effect of aspirational in supermarkets frequented by relatively high-income consumers. Sec­
messages on consumer acceptance of or demand for fortified foods – or ond, rice is perceived as an essential staple rather than as a source of
indeed other preventive health products – has not previously been nutrients. Thus, fortified rice does not align with typical conceptions
studied experimentally, despite evidence that low-income populations about the product category of rice. Finally, since rice is the main staple
engage in significant aspirational consumption and that marketers tar­ in Bangladesh, the study population is familiar with the characteristics
geting these populations often use aspirational marketing strategies. and market price of unfortified rice and is, therefore, able to form a
While the literature on the promotion of demand for biofortified foods is meaningful assessment of the value of the fortified product.
growing, effective marketing messages beyond the description of The government’s dual strategy of distributing fortified rice through
nutritional benefits of these foods have not been clearly identified. We a public safety net program while simultaneously encouraging the
thus fill a gap in the literature on how to stimulate demand for fortified development of a private market raises the questions of (1) how
foods, and for preventive health products more generally. knowledge of free distribution might affect consumer demand, and (2)
Second, we address how knowledge about the free distribution of a how this reaction mediates the effectiveness of a marketing strategy
health product to targeted groups interacts with efforts to market it more based on aspirational product positioning.
broadly. Previous studies have examined the effect of free distribution
on subsequent demand among those to whom goods were provided and 3. Experimental and survey design
their social contacts (Bensch and Peters, 2020; Cohen and Dupas, 2010;
Omotilewa et al., 2019). In these studies, the impact of free distribution In this section, we describe study site selection, determination of
on demand is interpreted as the combined effect of learning through sample size, participant recruitment process, and data collection
direct or observed experience with the good, and that of anchoring methods.
consumer expectations on a reference price of zero. The evidence we
present differs from these studies in that participants informed that 3.1. Study site and population
fortified rice is distributed through a safety net program elsewhere in the
country do not observe the benefits of consuming this rice. We thus We aimed to recruit a study population with a high potential to
interpret the effect of this information as arising from the combination of benefit from the consumption of industrially fortified rice within
a credibility cue and a reference price effect. Bangladesh. This led us to draw from the functionally landless rural
population (defined by the Government as those who own<0.2 ha of
2. Study setting land). Prevalence of micronutrient deficiency is highest in rural areas
(Ministry of Health and Family Welfare and Bangladesh, 2020), and
Approximately half of the Bangladeshi children under five years of landlessness is an effective proxy for poverty (and thus vulnerability to
age (U-5), and 11 percent of school-aged children suffered from anemia micronutrient deficiency) in rural Bangladesh, as evidenced by its use as
as of 2013, the most recent year for which nationally representative an eligibility criterion for rural safety net programs (Chowdhury et al.,
micronutrient deficiency data are available (Khan et al. 2016, UNICEF 2020).
2013). More recent statistics indicate that the nutritional status of As resources did not allow for a more broadly representative study,
Bangladeshi children remains poor, with 31 percent of U-5 children we selected a high-poverty district, Chandpur, in Chittagong Division,
stunted (short for their age) (Ministry of Health and Family Welfare and which has the lowest per-capita rice production among the eight di­
Bangladesh, 2020). Anemia, to which iron deficiency is a major visions in the country. Households in low rice production areas rely at
contributor, is one of the leading causes of pregnancy-related compli­ least partially on markets – as opposed to subsistence production – for
cations and maternal mortality in the country, with approximately 50 the rice they consume, and thus constitute a potential market for
percent of reproductive-age women classified as anemic (UNICEF 2013). industrially fortified rice. Within the Chittagong division, Chandpur has
Zinc deficiency is also common, affecting 45 percent of U-5 children and the highest incidence of extreme poverty; 51 percent of households are
66 percent of non-pregnant non-lactating women (National Institute of below the poverty line, and 30 percent are below the extreme poverty
Population Research and Training (NIPORT), Mitra and Associates, and line (Bangladesh Bureau of Statistics 2017). By conducting the study in
ICF International, 2016; UNICEF, 2013). this district, we purposively selected a population expected to have low
Rice is the main staple food in Bangladesh and accounts for 70 dietary diversity and thus high potential to benefit from the consump­
percent of caloric intake (Sayeed and Yunus 2018), with an average tion of fortified rice.
daily per capita consumption at 426 g (Yunus et al. 2019). Fortified rice Four of the eight sub-districts in Chandpur district were excluded
is not currently marketed in Bangladesh, and the low manufacturing from the sample due to urban or semi-urban classification (associated
capacity for fortified kernels precludes mandatory fortification. Instead, with higher wealth status), flood risk (which could imply inaccessibility
the government has taken a more targeted approach, distributing rice during the study period), or involvement in other research activities.

3
R. Chowdhury et al. Food Policy 112 (2022) 102346

From the remaining four sub-districts, namely Haziganj, Fairdganj, the auction sessions, the total targeted sample size for the four experi­
Shahrasti, and Kachua, we randomly selected a total of ten unions for mental groups was set to 1,200(≈ 245 × 4 × 1.22), with 120 households
inclusion in the study. The data collection (household survey and auc­ from each of the ten unions. A total of 1,203 households were surveyed.3
tion sessions) was conducted between July and early September of 2019.
To identify poor, landless households with a high probability of both 3.3. Survey
reliance on purchased rice and micronutrient deficiency, we selected the
sample from beneficiary lists of the Food Friendly Program (FFP). Upon recruitment into the study, respondents were administered a
Through this program, which is budgeted to cover 25 million individuals structured, quantitative survey through a face-to-face interview. The
(GoB SSPS 2020), functionally landless rural households are able to survey included questions regarding household income, expenditures,
purchase up to 30 kg of rice at 20–30 percent of the market price during land ownership, rice cultivation, rice consumption, and recent rice
each of the five ‘lean season months’ (Chowdhury et al., 2020). This purchases. The survey also included questions regarding household food
subsidy is offered during March to April and September to November, security and dietary diversity, which were administered to a female
which does not overlap with the timing of the framed field experiment household member responsible for food preparation.
(July to August). Participation in this program signifies food insecurity Upon completion of the survey, the primary respondent was invited
as well as significant reliance on unsubsidized rice, as the quantity of to participate in an auction session to be held on a later date in a school
subsidized rice for which a household is eligible is low relative to total building within the community where he lived. Participants were not
annual consumption needs. Fortified rice is not currently available informed about the purpose of the session but were informed about its
through the FFP, and even if it were in the future, FFP beneficiaries expected duration and that they would receive compensation of BDT
could benefit from purchasing fortified rice during months of the year 200 (USD 2.42) for attending,4 which is equivalent to the value of 3 h of
when the program is not active. We note that fortified rice had not, at the unskilled work at the local wage rate.5 No compensation was paid for the
time of the study, ever been distributed through the FFP. completion of the household survey.6
Households that had received any free food transfer from the gov­
ernment in the last year were excluded due to the risk that receiving
another such transfer immediately prior to the experiment or expecting 3.4. Auction sessions and experimental treatments
to receive one shortly thereafter could affect their demand for rice.
Those who did not have at least one child aged between 0 and 16 years Respondents in each union were randomly assigned to one of four
were also excluded, as micronutrient fortification is especially beneficial treatment arms: a comparison group was given information about
to the health of children, and the nutritional drink to which fortified rice common micronutrient deficiencies and the health benefits of
was compared in the aspirational treatment stresses the benefits to consuming fortified rice (C); treatment group 1 (T-1), which in addition
children in its marketing. Finally, noting that approximately 15 percent was read an aspirational message using a premium nutritional drink mix
of FFP beneficiaries do not meet targeting criteria (Chowdhury et al., as an analog for fortified rice; treatment group 2 (T-2), which in addition
2020), to ensure a high risk of micronutrient malnutrition among the was read a message stating that fortified rice is distributed through the
subject pool and also their reliance on marketed rice, participants had to Public Food Distribution System (PFDS) in particular districts, and
satisfy at least two of the following five inclusion criteria, adopted from treatment group 3 (T-3), which received all three messages.7 In each
program targeting criteria used by the Bangladeshi NGO BRAC: (1) the union, auction sessions for the four groups were held simultaneously in
household purchases rice from the market for at least four months of the
year; (2) an adult female household member sells labor; (3) the main
3
income earner is disabled; (4) a school-aged child sells labor at least Sample size by union: Chitoshi Purba 120, Dwadashgram 119, Dakkhin
Rupsha 122, Gandharbapur Dakkhin 119, Kalachon Uttar 120, Balithuba East
occasionally; (5) no household member has a regular salaried job. Only
123, Purba Subidpur 125, Tamta Dakkhin 119, Tamta Uttar 119, and Uttar
3.5 percent of households screened were excluded from participation in
Kachua 117.
the study based on all these inclusion and exclusion criteria. 4
Calculated using the official exchange rate of 1 USD = 82.78 BDT.
To ensure that participants had the authority to make food purchases 5
The literature on how participation incentives affect behavior in economic
on behalf of the household, the male household head or another adult studies is mixed. Clark (2002) suggests that such “house money” effects are not
male who usually buys groceries for the household was designated as the significant in the context of experiments on contributions to public goods,
primary respondent. though Harrison (2007) has disputed this. If the participation fee was only just
enough to cover the cost of transportation back home, participants’ need for
cash could result in a downward bias. Participants were provided with BDT
3.2. Sample size
200, so that even if a participant purchased rice at the maximum possible price
of BDT 100 (over three times the market value of unfortified rice), and both
The sample size for the experiment was set to achieve a minimum practice goods at the maximum combined price of 35 BDT, he would still be left
detectable effect (MDE) in WTP of each marketing message against the with BDT 65, which is more than enough to cover the 30 BDT maximum round
comparison group of 25 percent. This MDE is 26 percent lower than the trip transportation cost, as estimated by field staff.Our results indicate that bids
premium for biofortified rice estimated using a similar methodology for unfortified rice in the comparison group, at 24.4 BDT, were slightly lower
among low-income households in rural China (De Steur et al., 2012). than the market price of the rice used in the experiment (30 BDT). This is ex­
Type I and II error rates were set to the conventional levels of 5 percent pected, as participants were not necessarily in need of rice at the time of the
and 80 percent, respectively. Based on these parameters, the sample size bidding sessions, and indicates that any upward bias due to a participation
required was 245 households for each intervention group. For the binary incentive was minimal overall. However, we are unable to rule out that the
participation incentive affected bids for fortified rice since there is no market
outcome of willingness to pay at least the cost of fortification, this
price for fortified rice in the study communities.
sample size implies an MDE of between 8.9 and 12.6 percentage points, 6
We note that it is typical to conduct surveys after experimental procedures,
depending on the proportion in the comparison group. to avoid priming effects. In our case, collecting data before differences across
Anticipating some attrition between recruitment and participation in treatment groups were introduced was important for the heterogeneity analysis,
as survey responses could have been influenced by differences across message
treatments. Further, any potential priming effects of the survey on WTP for
fortification would be uniform across treatment groups, and our primary
questions related not to the level of the fortification premium, but rather to
differences in WTP based on messages given during the sessions.
7
Experimental scripts are provided in the online appendix B.

4
R. Chowdhury et al. Food Policy 112 (2022) 102346

different classrooms within a schoolhouse to avoid any potential spill­ Eliciting bids for both types of rice was necessary because the messages
over effects from participants in different groups interacting between tested could potentially affect willingness to pay for non-fortified rice as
sessions. well as fortified rice through framing effects. We calculate the premium
The nutritional drink mix to which fortification was compared was that participants are willing to pay for fortification as the difference
Horlicks, a brand heavily promoted over the years through radio, tele­ between their bids for fortified and unfortified rice.
vision, and print media in Bangladesh. This promotion has resulted in a BDM auctions are commonly used in experimental economics to
strong aspirational image among consumers, which is confirmed by the induce participants to bid their actual valuation of a good. In our
data from our household survey (see Table A.1 in the appendix for more implementation of the BDM auction, each participant was asked to
details). submit the maximum price he was willing to pay for 1 kg of unfortified,
The PFDS message informed study participants that while fortified and for 1 kg of fortified rice. As unfortified rice is well known to par­
rice is not available for purchase in Bangladesh, the government dis­ ticipants, this was consistently offered first as a base case to anchor
tributes this rice to poor households through the PFDS in two districts, participants’ thinking.
not including Chandpur. PFDS is the primary social support program for After both bids were entered, the type of rice (fortified or non-
poor and vulnerable households in Bangladesh, and the FFP, as well as fortified) for which the auction was to be implemented was deter­
several other programs, are covered within this umbrella. For study mined through a random draw. Then, a random price was drawn from a
households, this message implied that fortified rice was being given for uniform distribution of prices ranging from 1 to 100. If a participant’s
free to poor people like them, elsewhere in the country. While the fact bid for the randomly selected type of rice was higher than or equal to the
that fortified rice is distributed to vulnerable populations in Bangladesh randomly drawn price, the participant had to purchase that rice at the
is publicly available online, it is not widely known. drawn price, using the compensation provided for attendance.9 Under
Table 1 below illustrates the combinations of messages delivered to this procedure, participants do best for themselves by bidding their true
each of the four experimental groups. The aspirational message read to willingness to pay. A bid below one’s actual valuation risks losing the
participants assigned to T-1 is expected to have a positive effect on WTP. good if the randomly drawn price is lower than the participant would be
If the association of fortified rice with the PFDS implies a low product willing to pay. On the other hand, if the participant bids too high, he
value, anchors participants on a low reference price, or creates an anti- may be forced to buy the good at a higher price than his actual valuation
aspirational impression of the product, the effect of T-2 could be nega­ of the good.
tive. On the other hand, if government distribution signals the impor­ To ensure participants understood the auction procedure, prior to
tance of consuming fortified food, or the credibility of health claims bidding on unfortified and fortified rice, two brands of soap and two
associated with fortification, the effect could be positive. The effect of brands of biscuits were offered through two practice auctions. Aside
combining messages in T-3 is likewise ambiguous. An anti-aspirational from the goods offered and the range of possible prices (1–10 BDT for
effect of associating fortification with the PFDS could weaken or even biscuits, 1–25 BDT for soap), the procedures for these auctions were
reverse the hypothesized positive effect of the aspirational marketing identical to those for rice, including the opportunity to inspect goods
message. However, if this message serves as a positive signal of impor­ prior to bidding. Purchases of these goods, for those who bid at or above
tance or credibility, combining the two messages could lead the com­ the drawn price, were implemented prior to the auction for rice. As
bined message (T-3) to have the largest positive impact on WTP, through purchases during the practice rounds reduced the funds available to
an additive or even multiplicative effect. spend on rice, we include in the regression models below indicators for
After the treatment-specific message was read, participants were whether the participant made a purchase in each practice round.
invited to inspect samples of fortified and unfortified rice; each pack­ Of the 1,203 primary respondents interviewed, 1,057 (≈ 88 percent)
aged in clear plastic bags. The fortified rice offered was created by participated in the auction sessions. The 12 percent attrition rate did not
mixing fortified kernels into the unfortified rice offered, so that the two come as a surprise, considering that 65 percent of respondents worked as
types differed only in terms of the presence of fortified kernels (Figure A- wage laborers, and the auction sessions were held during the day on a
1 in the online appendix).8 date after the baseline interview. We present the results of a regression
The participants were asked to bid on 1 kg of fortified rice and 1 kg of of an attrition indicator on the baseline variables and treatment group
unfortified rice using the Becker-DeGroot-Marschak (BDM) mechanism. indicators in Table A.2 in the online appendix. Overall, we do not find a
strong association between attrition and baseline characteristics. Our
balance check of the random assignment in section 5 also confirms
Table 1 similarities between the treatment groups.
Messages by experimental treatment group.
Information/Messages Comparison Treat.-1 Treat.-2 Treat.-3 4. Empirical framework
(25 %) (25 %) (25 %) (25 %)

Health risks of anemia and ✓ ✓ ✓ ✓ We estimate the effects of the message treatments on consumers’
zinc deficiency + willingness to pay a premium for fortified rice using the following linear
benefits of consuming
regression model:
fortified rice ∑
Aspirational message: × ✓ × ✓ WTPrequiredpremium ijk = β0 + 3j=1 βj Tij + βx Xi + γ k + eijk (1).
Analogy to Horlicks Where WTPrequiredpremium ijk is an indicator for whether the respondent i
Information about free ✓ ✓
in union (sub-district) k was the willingness to pay a premium at least as
× ×
distribution through
PFDS

9
A handful of participants did not want to buy the rice even though they won
the bid. These cases occurred during the first week of the experiment, and were,
according to field staff, due to participants’ desire to keep the larger de­
8
While it is possible to identify fortified kernels by their yellowish color, nominations provided as compensation, and avoiding the smaller de­
these constitute only 0.05 to 2 percent of the blend, resulting in little visible nominations provided as change for rice purchases, rather than anything to do
difference between fortified and unfortified rice. Nor does the fortification of with the bidding process. In response, the field team subsequently used smaller
rice in this way change its smell, taste, or other observable traits. Therefore, denominations to avoid this problem coming up, after which all winners pur­
participants’ willingness to pay for the fortified rice is not likely to be influ­ chased the rice they had won. The results are robust to excluding observations
enced by observational characteristics arising from fortification. from the first week of auctions.

5
R. Chowdhury et al. Food Policy 112 (2022) 102346

high as the cost of fortification. Tij are indicators for the three treatment of the aspirational good used in the messages delivered to groups T-1
groups (T-1, T-2, T-3). Xi is a vector of household and auction level and T-3.
covariates for participant i, γ k are union-level fixed effects, and ejik are
individual error terms, clustered to account for correlation that could 5.2. Acceptability of and willingness to pay for fortified rice
result from potential differences in the delivery of messages or imple­
mentation of the auction. Xi includes household demographic charac­ Column 1 of Table 2 shows the proportion of study participants
teristics, socio-economic status, involvement in rice cultivation, foods overall and by treatment group to whom fortified rice is acceptable,
perceived as nutritious, knowledge of Horlicks, indicators for whether defined as those who were willing to pay at least as much for fortified
the participant purchased each of the practice goods, and auction week rice as for unfortified rice. Overall, fortification was acceptable to 93
(as a measure of the experience of auction session moderators). The percent of participants. In the group that received information only on
coefficients of interest are β1 , β2 , and β3 , which respectively measure the the health benefits of fortified rice, 89 percent bid at least as much for
effects of the aspirational message, the link to public distribution, and fortified as unfortified rice, whereas this proportion was 91 percent of
the combination of the two. We inspect the sensitivity of the estimates those to whom the aspirational cue was communicated, and 93 percent
by estimating models both with and without covariates. of those to whom information about free distribution was given.
We also test whether the effects of the messages vary by households’ Acceptability was highest in group T-3, who received both messages; 99
poverty level (e.g., food security and asset holdings) and demographic percent of these participants bid at least as much for fortified rice as they
characteristics (e.g., presence of reproductive-aged women and U-5 did for the regular, unfortified rice.
children). In measuring treatment effect heterogeneity, we follow the Considering willingness to pay a premium for fortified rice (Column
conventional econometric approach of estimating the interaction be­ 2), we find that 76 percent of participants overall were willing to pay at
tween the treatments and the respective baseline characteristics through least BDT 3.96 (USD 0.048) more for fortified than unfortified rice,
the following linear regression model: reflecting the per-kg cost of fortification paid by the Government of
∑ ∑
WTPrequiredpremium ijk = β0 + 3j=1 βj Tij + βx Xi + 3j=1 αj Tij × xi + γk + Bangladesh for rice distributed through the PFDS. Again, this proportion
eijk (2). is highest among the group that received both messages, lowest among
where, xi is an indicator for whether the participant belongs to the the group that only received health information, and intermediate
relevant group (e.g., households with a child under 5). The coefficients among the two treatment groups that received either the aspirational
of interest are αj , which measure the differential effect of marketing message or information on the link to PFDS.
message j on the subgroup denoted by the indicator xi . Table 3 presents the means and confidence intervals of participants’
bids for both unfortified and fortified coarse rice. Overall, participants
5. Data and results were willing to pay an average of BDT 6.8 (USD 0.082) more for fortified
rice than comparable unfortified rice, representing a 25 percent pre­
In this section, we begin by describing balance across treatment mium. While the premium for fortified rice is positive and statistically
groups. We then turn to the potential influence of providing compen­ significant for all four experimental groups, it is highest in T-3 (com­
sation for participation on the results, before turning to a description of bined free distribution and aspirational cues) at (30 percent), followed
WTP for fortified rice by treatment group, and then formal analysis of by T-2 (free distribution) (27 percent). The premiums among T-1
the impact of the message treatments on this outcome, in a regression (aspirational cue) and comparison group subjects are the same, at 21
framework. percent above the average bid for unfortified rice.
We note that mean willingness to pay for unfortified rice follows a
similar pattern across the four groups as the fortification premium.
5.1. Balance across treatment groups While messaging treatments were not expected to affect WTP for this
rice, it is possible that information on fortified rice altered the reference
Table A.3 in the online appendix presents summary statistics on price for other rice. A number of studies, including by Thaler (1985) and
household demographic characteristics, socio-economic status, rice Lattin & Bucklin (1989), demonstrate that WTP for a product can be
cultivation, consumption, foods perceived as nutritious, and knowledge influenced by purchase context, for example, the prices of other prod­
of Horlicks. The table also presents the balance tests for random ucts available at the same shop (Mazumdar et al. 2005). Since our in­
assignment of the treatment groups. In sum, we find that our randomi­ terest is in the magnitude of the premium, we control for these reference
zation was effective in creating comparable treatment groups, and re­ effects in the next section by analyzing the effect of the experimentally
spondents across the four groups are mostly similar in terms of socio- varied messages on the difference in bids between fortified and unfor­
demographic characteristics, rice consumption, and their perceptions tified rice.

Table 2
Acceptability and willingness to pay for fortified rice.
Message Willing to pay at least as much for fortified Willing to pay a premium of at least USD N
rice as unfortified rice 0.048 (cost of fortification)

(1) (2) (3)


Full Sample proportion 93 % 76 % 1,057
95 % CI [0.91, 0.94] [0.73, 0.78]
Comparison Health benefits only proportion 89 % 69 % 264
95 % CI [0.84, 0.92] [0.63, 0.74]
T-1 Health benefits + Aspirational message proportion 91 % 72 % 264
95 % CI [0.86, 0.94] [0.66, 0.77]
T-2 Health benefits + Free distribution info proportion 93 % 77 % 275
95 % CI [0.90, 0.96] [0.72, 0.82]
T-3 Health benefits + Aspirational message + proportion 99 % 86 % 254
Free distribution info
95 % CI [0.97, 1.00] [0.81, 0.90]

Notes: Two-tailed exact binomial confidence intervals (CI) around the proportion means for corresponding samples are reported in brackets in columns 1 and 2 using
the Clopper-Pearson (1934) method.

6
R. Chowdhury et al. Food Policy 112 (2022) 102346

Table 3
Willingness to pay for unfortified and fortified Rice.
(1) (2) (3) (4) (5)

Message Fortified Rice Unfortified Rice Difference (std. Fortification N


error) Premium
Full Sample 33.91 [33.11 27.11 [26.56 6.80*** (0.492) 25 % 1,057
34.70] 27.65]
Comparison Health benefits 29.69 [28.2 31.2] 24.38 [23.25 5.31*** (0.949) 21 % 264
Group 25.50]
T-1 Health benefits + Aspirational message 33.23 [31.84 27.53 [26.59 28.5] 5.71*** (0.852) 21 %
34.63]
T-2 Health benefits + Free distribution info 34.34 [32.75 27.04 [26.01 7.30*** (0.962) 27 % 275
35.94] 28.06]
T-3 Health benefits + Aspirational + Free distribution 38.52 [36.79 29.57 [28.36 8.95*** (1.069) 30 % 254
info 40.25] 30.77]

Notes: Unconditional mean bids for each type of rice are shown. *, **, *** denote statistical significance of coefficients at the 10 percent, 5 percent, and 1 percent
confidence level, based on a paired t-test.

The distribution of WTP, plotted in Fig. 1, shows substantial As a robustness check, we also measure the effects of the messages by
dispersion in preferences for both unfortified and fortified rice, with a logit regression model and report the results in Table A.4 in the online
longer right tails for fortified rice in all groups. For instance, among appendix. For ease of interpretation and comparison with the linear
participants in the comparison group, the 90th percentile bid was BDT regression outcomes, we estimate and report the marginal effects of the
42 (USD 0.51), almost three times the 10th percentile bid. The disper­ logistic regression estimates. The logit model estimates consistently
sion in WTP for fortified rice is greatest in groups T-2 and T-3, the groups match the results from the linear regression model. Furthermore, we
for which the average fortification premium was highest, indicating that estimated the same models shown in Table 4 with the price premium for
the mean premium is to some extent driven by a subset of consumers fortified rice as the outcome variable. Results are presented in online
willing to pay much more for fortification than average. For this reason, appendix Table A.5. Results are consistent with those shown in Table 4;
and because the commercial viability of fortified rice depends on we find a significant effect for the combination of aspirational and public
whether a sufficient number of consumers are willing to pay a premium distribution messages. Participants in the combined message group were
that covers its cost, we focus below on the binary outcome of whether willing to pay an additional BDT 3.99 (USD 0.048) on average for for­
participants’ fortification premium is at least equal to the government’s tified rice relative to those in the health information-only group. The re­
cost of fortifying rice for the PFDS program. sults also show a mild positive effect (significant at the 10 percent level)
of the aspirational message cue on the premium for fortified rice.
5.3. Effect of the marketing messages on demand for fortified rice We note that associating fortified rice with the PFDS is unlikely to
have improved participants’ perceptions of this rice on dimensions other
In this section, we measure the effects of the marketing messages on than the credibility of health claims. In fact, data from the baseline
the participants’ WTP premium for the fortified rice using the empirical survey shows that of those who knew about the free distribution of rice,
model specified in equation 1. The results are presented in Table 4. In 94 percent mentioned that the government distributes low-value coarse
addition to the specification in equation 1, which is shown in Column 2 rice (see Table A.1). When asked if they knew of any special charac­
of Table 4, we also present results for specifications in which the cova­ teristics of the rice distributed through PFDS, only 4 percent of partici­
riates (Xi ) are excluded (Column 1). pants indicated that they did; of these, the majority stated that PFDS rice
Across specifications, we do not find a significant effect of adding is of poor quality or mentioned a negative characteristic (see Table A.1).
either the aspirational message or information about distribution
through the PFDS to the cross-cutting information about product attri­ 5.4. Heterogeneous effects
butes and health benefits. However, combining these two messages (in
group T-3) increases the proportion of participants willing to pay the We next test whether the effect of the messages depends on partici­
required premium by 19 percentage points relative to the health infor­ pant characteristics. Understanding heterogeneous effects can shed light
mation only group (p < 0.01). The effect of the joint message has a on which message interventions are most effective for inducing demand
stronger effect on demand than either of the individual messages. among subgroups of interest, for example, the poorest or most food-
The linear combination of the two individual message effects is insecure households. This analysis also allows us to develop insights
similar to that of the aspirational message, at around 9 percent, and not regarding the mechanism through which a message affects participants’
significantly different from zero (p-value = 0.473). While the effect of valuation of fortified rice.
the combined (T-3 group) message is not significantly different from the We begin by testing whether the effects of the messages depend on a
sum of the two individual effects (p = 0.168), its magnitude is over twice household’s level of food security and asset holdings. In Column 1 of
as great, suggesting a possible complementarity of messages that the Table 5, we include the interaction between treatment status and a
study was not powered to detect. dummy indicating whether the household has an above-median value of
To save space, coefficients on the baseline covariates are not re­ the Food Insecurity Access Score (FIAS). We find no evidence that either
ported in Table 4. Three of these were associated with WTP for fortifi­ WTP a premium, nor message effects, vary by FIAS.
cation, all at the 10 percent significance level. Respondents with higher In Column 2, we test whether treatment effects vary by whether a
household dietary diversity scores were less likely to be willing to pay household’s asset score is above versus below the median value.
for fortification (coeff. = -0.022, SE = 0.012), suggesting that they may Consistent with the analysis based on food insecurity status, asset wealth
have perceived fortification as less important to their family’s health. is not significantly correlated with WTP. We note a significant hetero­
Those who purchased the soap bar during the auction practice session geneity result for one of the messages: participants from households
were less likely to be willing to pay the fortification premium (coeff. = − with an asset score above the median are 11 percentage points more
0.070; SE = 0.038). Finally, the auction week covariate, which controls likely to be willing to pay the required premium for the rice when
the auction moderators’ effect, was positively associated with WTP for exposed to the aspirational cue (p < 0.05).
fortification (coeff. = 0.022; SE = 0.012). Next, we test whether the impact of messages depends on whether

7
R. Chowdhury et al. Food Policy 112 (2022) 102346

Fig. 1. Kernel distribution of the willingness-to-pay (WTP) for unfortified and fortified rice by treatment groups in Bangladeshi Taka (BDT).

participants live with at least one woman between the ages of 15 and 49
Table 4
or a child below the age of five. We focus on these two groups because
Determinants of consumer willingness to pay for the cost of fortification.
the auction scripts described the health risks of micronutrient deficiency
(1) (2) for expectant mothers and children under five. We find no evidence that
T-1: Aspirational message 0.012 0.011 the effect of the experimental messages differed for households that
(0.057) (0.057) included a woman of reproductive age versus those that did not (Column
T-2: Free distribution info 0.076 0.070
3). However, we do observe heterogeneity across households with and
(0.064) (0.064)
T-3: Asp. msg. + Free dist. 0.187 0.191
without children under five. Within the comparison group, a larger
(0.055) (0.057) proportion of respondents from households with at least one child under
*** *** this age were willing to pay the required premium for fortified rice
Union fixed effects Yes Yes compared to those without young children (Column 4). However, this
Practice round purchase indicators Yes Yes
subgroup was significantly less responsive to the aspirational cue than
Baseline controls, auction week No Yes
Treatment effect of Asp. Msg: T-3 (Comb.) – T-1 (Asp.) 0.175 0.180 others. This could be because Horlicks marketing targets school-aged
(0.043) (0.043) children and pregnant women and does not focus on young children,
*** *** hence these households may not have perceived any additional benefits
Treatment effect of Free dist. info: T-3 (Comb.) – T-2 (Free 0.111 0.121
due to the association with Horlicks, or may even have inferred that
dist.)
(0.051)** (0.052)**
fortification is less important for young children. Lower responsiveness
Sum of individual message effects: T-1 (Asp.) + T-2. (Free 0.088 0.080 to additional messages could also arise if there exists a ceiling in po­
dist. info) tential willingness to pay, which is reached by households with children
(0.110) (0.111) under five when they are provided with health information alone.
Complementarity test: T-3 (Comb.) – [T-1 (Asp.) + T-2. 0.099 0.111
Finally, in Column 5 we test whether participants’ responses to the
(Free dist.)]
(0.079) (0.079) message interventions differed based on educational attainment. We do
Constant 0.743 0.531 not find any heterogeneity in effects along this dimension. The heter­
(0.046) (0.271)* ogenous treatment effect analysis from the logit model also reveals the
***
similar findings (see Table A.6 in online appendix).
N 1,057 1,057
The total effect of the combined message (T-3) on demand for for­
Note: Coefficients shown are estimated via the linear regression model described tified rice among households especially vulnerable to micronutrient
in Equation (1). Willingness to pay a premium at least as high as the fortification malnutrition, whether defined as those with a high food insecurity score,
cost is the dependent variable in all models. Standard errors clustered by auction low asset holdings, inclusion of a member in a vulnerable demographic
session, the unit of randomization, are shown in parentheses. Union fixed effects
group, or low educational attainment, is always positive and significant
and indicators for purchases during practice auction rounds are included in both
at p < 0.1 (see online appendix table A.7).
models. Baseline variables included in the model reported in column (2) are
those shown in Table A.3. *, **, *** denote statistical significance of coefficients
at the 10 percent, 5 percent, and 1 percent confidence level.

8
R. Chowdhury et al. Food Policy 112 (2022) 102346

Table 5
Heterogenous Effects by Food Security Status, Assets Holdings, Household Demographics, and Education Level.
High Food Insecurity High Asset Holding HHs with Reproduc-tive HHs with U-5 Respondent with High School and
Access Score (1) Score (2) Aged Female (3) Children (4) Above Education (5)

T-1: Aspirational msg. − 0.027 − 0.069 0.010 0.102 0.041


(0.079) (0.065) (0.142) (0.070) (0.063)
T-2: Free dist. info 0.046 0.105 − 0.022 0.113 0.066
(0.086) (0.066) (0.171) (0.078) (0.069)
T-3: Aspirational + Free 0.185 0.189 0.175 0.220 0.205
dist.
(0.077)** (0.059)*** (0.204) (0.075)*** (0.061)***
Subgroup indicator − 0.038 − 0.039 0.096 0.147 0.095
(0.057) (0.049) (0.120) (0.072)** (0.074)
T-1 # Subgroup 0.075 0.114 0.001 − 0.256 − 0.124
(0.099) (0.054)** (0.150) (0.079)*** (0.077)
T-2 # Subgroup 0.043 − 0.045 0.096 − 0.108 0.003
(0.087) (0.060) (0.190) (0.086) (0.079)
T-3 # Subgroup 0.013 0.004 0.017 − 0.071 − 0.063
(0.077) (0.056) (0.205) (0.083) (0.084)
Constant 0.431 0.680 0.558 0.459 0.520
(0.241)* (0.252)** (0.264)** (0.270)* (0.277)*
Baseline controls, Yes Yes Yes Yes Yes
auction week
R2 0.09 0.10 0.09 0.10 0.10
N 1,056 1,056 1,056 1,056 1,056

Note: Coefficients shown are estimated via the linear regression model described in Equation (2). Willingness to pay a premium at least as high as the fortification cost
is the dependent variable in the models. Standard errors clustered by auction session, the unit of randomization, are shown in parentheses. Union fixed effects, in­
dicators for purchases during practice auction rounds, auction week, and baseline characteristics shown in Table A.3 are included in all models. *, **, *** denote
statistical significance of coefficients at the 10 percent, 5 percent, and 1 percent confidence level.

6. Discussion programs while also encouraging voluntary fortification by private firms


is that an association with free distribution could lead consumers to infer
6.1. Summary and interpretation of results low product quality or depress price expectations. Our findings
demonstrate that aspirational product positioning can be effective in
Results from a framed field experiment conducted among a low- such a context.
income population in rural Bangladesh suggest that a message posi­ Our findings stand out from prior experimental studies, which have
tioning fortified rice as a high-value product and linking it to a gov­ not identified marketing messages that significantly increase demand for
ernment program can increase the proportion of consumers willing to biofortified foods. Adopting an aspirational product positioning strat­
pay the cost of fortification. The lack of impacts from each message on its egy, while also informing consumers about government investment in
own, along with their strong joint impact when delivered together, this technology, could significantly increase private demand for fortified
suggests that each conveys distinct information, which may act on foods in contexts where free distribution is undertaken in parallel with
consumer perceptions in complementary ways. Our interpretation is development of a private market. Rather than downplaying the use of
that connecting fortified rice – a novel product in this context – to a fortified foods in safety net programs, our results suggest that incorpo­
known, high-value brand, increases its potential value in the minds of rating information about public distribution into social marketing
consumers – but that this is insufficient to induce higher willingness to campaigns could bolster their impact on private demand.
pay, possibly due to doubts about the credibility of the health claims We note that the Government of Bangladesh has been praised for
made about the rice. advancing population health through equitable and extensive programs
Informing consumers about the distribution of fortified rice through (El Arifeen et al. 2014, Ahmed et al. 2013, Chowdhury et al. 2013).
safety nets might be expected to counter the effect of aspirational Without this positive track record, the effect of associating fortified rice
product positioning, thus weakening, or even reversing the impact of with a public safety net program may not have been as effective.10
such a strategy. Social marketing campaigns in other settings have
avoided positioning fortified products as food ‘for the poor’. In this
6.3. Limitations and future directions
experiment, however, we observed the opposite: consumers who
received the aspirational message and were told that rice was provided
A limitation of the study is that it was conducted in only one district
for free through a government program elsewhere in the country were
of Bangladesh, with participants drawn from a specific population:
willing to pay significantly more than those given the aspirational
beneficiaries of a seasonal rice subsidy, who were verified to meet a set
message alone. We hypothesize that this may be due to the information
of commonly used targeting criteria. While this population has great
about government distribution lending credibility to health claims about
potential to benefit from the purchase and consumption of fortified rice,
the product. This finding complements previous work on the long-term
a more representative sample would have allowed us to draw broader
impact of limited-term subsidies for health products and points to a
conclusions. A second limitation is that the information about free dis­
novel mechanism through which subsidies could increase private
tribution mentioned that fortified rice was not available for purchase in
demand.
Bangladesh at the time. Thus, participants assigned to this treatment

6.2. Policy implications


10
Of course, the messaging interventions studied in this paper are only a small
Experience has shown that achieving high rates of take-up of part of the toolkit policy makers have available to promote fortified food. Other
voluntarily fortified foods is possible, but far from guaranteed (Lalani more direct interventions such as subsidies and regulatory standards are not
et al., 2019). A potential challenge faced by governments taking a dual- studied in this paper but are nevertheless likely to be crucial components of a
track approach of distributing fortified foods through social safety market solution for fortified food.

9
R. Chowdhury et al. Food Policy 112 (2022) 102346

may have seen the experiment as a unique opportunity to test a new Banerji, A., Birol, E., Karandikar, B., Rampal, J., 2016. Information, branding,
certification, and consumer willingness to pay for high-iron pearl millet: Evidence
product which may have increased their willingness-to-pay. The impact
from experimental auctions in Maharashtra, India. Food Policy. 1 (62), 133–141.
we attribute to a credibility effect could potentially be confounded by Bangladesh Bureau of Statistics, 2017, Statistical Year Book of Bangladesh 2016,
this effect. Retrieved September 5, 2021, from, http://www.bbs.gov.bd/site/page/29855dc1-
Future research could elucidate the mechanism through which the f2b4-4dc0-9073-f692361112da/-.
Beninger, S., Robson, K., 2015. Marketing at the base of the pyramid: Perspectives for
messaging interventions studied in this paper affect willingness-to-pay practitioners and academics. Bus. Horiz. 58 (5), 509–516.
for fortified rice. For instance, we hypothesize that the message about Bensch, G., Peters, J., 2020 Jan. One-Off Subsidies and Long-Run
governmental distribution affects willingness-to-pay by increasing the Adoption—Experimental Evidence on Improved Cooking Stoves in Senegal. Am. J.
Agric. Econ. 102 (1), 72–90.
credibility of the product. Future research could test this hypothesis by Birol, E., Meenakshi, J.V., Oparinde, A., Perez, S., Tomlins, K., 2015 Jun. Developing
collecting data on trust in the government and more detailed attitudes country consumers’ acceptance of biofortified foods: a synthesis. Food Security. 7
about different attributes of fortified rice. Furthermore, future research (3), 555–568.
Bloom, P.N., Pailin, J.E., 1995 May 1. Using information situations to guide marketing
conducted in a more broadly representative sample, which assesses strategy. Journal of Consumer Marketing. 12 (2), 19–27.
demand for fortified rice over a longer time horizon and in a more Chowdhury, A.M.R., Bhuiya, A., Chowdhury, M.E., Rasheed, S., Hussain, Z., Chen, L.C.,
natural setting, would provide a clearer picture of the true level of 2013. The Bangladesh paradox: exceptional health achievement despite economic
poverty. The Lancet 382 (9906), 1734–1745.
consumer demand for this product. Chowdhury SK, Hoque MM, Rashid S, Bin Khaled MN. Targeting errors and leakage in a
large-scale in-kind transfer program: The food friendly program in Bangladesh as an
Funding example (Vol. 1935). Intl Food Policy Res Inst; 2020 Jun 12.
Chowdhury, S., Meenakshi, J.V., Tomlins, K.I., Owori, C., 2011 Jan. Are consumers in
developing countries willing to pay more for micronutrient-dense biofortified foods?
We gratefully acknowledge the financial support from the Govern­ Evidence from a field experiment in Uganda. Am. J. Agric. Econ. 93 (1), 83–97.
ment of Bangladesh, Directorate General of Food, through the Modern Clark, J., 2002 Dec. House money effects in public good experiments. Experimental
Economics. 5 (3), 223–231.
Food Storage Facilities Project (Contract number SD-21), and the ADM Clopper, C.J., Pearson, E.S., 1934. The use of confidence or fiducial limits illustrated in
Institute for the Prevention of Postharvest Loss at the University of Il­ the case of the binomial. Biometrika 26 (4), 404–413.
linois, Urbana-Champaign. Cohen, J., Dupas, P., 2010. Free distribution or cost-sharing? Evidence from a
randomized malaria prevention experiment. Quart. J. Econ. 125 (1), 1.
Copenhagen Consensus. (2008) Copenhagen Consensus 2008 – Results, Retrieved
CRediT authorship contribution statement November 1, 2021, from, https://www.copenhagenconsensus.com/sites/default/
files/cc08_results_final_0.pdf.
Copenhagen Consensus Outcome Page. Outcome – The Expert Panel Findings. (2012).
Reajul Chowdhury: Conceptualization, Formal analysis, Investiga­ Retrieved November 1, 2021, from, https://www.copenhagenconsensus.com/sites/
tion, Software, Data curation, Writing – original draft. Benjamin Crost: default/files/outcome_document_updated_1105.pdf.
Methodology, Writing – review & editing, Supervision. Vivian Hoff­ Darke, P.R., Chung, C.M., 2005. Effects of pricing and promotion on consumer
perceptions: it depends on how you frame it. J. Retail. 81 (1), 35–47.
mann: Methodology, Writing – review & editing, Funding acquisition, De Groote, H., Kimenju, S.C., Morawetz, U.B., 2011 Jan. Estimating consumer
Supervision. willingness to pay for food quality with experimental auctions: the case of yellow
versus fortified maize meal in Kenya. Agricultural Economics. 42 (1), 1–6.
De Moura, F.F., Moursi, M., Donahue Angel, M., Angeles-Agdeppa, I., Atmarita, A.,
Gironella, G.M., Muslimatun, S., Carriquiry, A., 2016 Sep 1. Biofortified β-carotene
Declaration of Competing Interest rice improves vitamin A intake and reduces the prevalence of inadequacy among
women and young children in a simulated analysis in Bangladesh, Indonesia, and the
The authors declare that they have no known competing financial Philippines. The American journal of clinical nutrition. 104 (3), 769–775.
De Steur, H., Gellynck, X., Feng, S., Rutsaert, P., Verbeke, W., 2012 Sep 1. Determinants
interests or personal relationships that could have appeared to influence of willingness-to-pay for GM rice with health benefits in a high-risk region: Evidence
the work reported in this paper. from experimental auctions for folate biofortified rice in China. Food Qual. Prefer.
25 (2), 87–94.
Department of Food & Public Distribution, Government of India. 2021. Centrally
Acknowledgments Sponsored Pilot Scheme on “Fortification of Rice & Its Distribution under Public
Distribution System”. October 05, Accessed October 12, 2021.
W e express our gratitude to Alex Winter Nelson, Professor at Uni­ El Arifeen, S., Hill, K., Ahsan, K.Z., Jamil, K., Nahar, Q., Streatfield, P.K., 2014. Maternal
mortality in Bangladesh: a Countdown to 2015 country case study. The Lancet 384
versity of Illinois, for his comments on earlier drafts of the article. We (9951), 1366–1374.
thank Dr. Shahidur Rahman, Director for South Asia at IFPRI, for GoB SSPS, Government of Bangladesh, Social Security Policy Support (SSPS) Programme,
providing institutional support. We would also like to thank Razin Kabir 2020, Retrieved November 1, 2021, from, https://mof.portal.gov.bd/sites/default/
files/files/mof.portal.gov.bd/page/672e3d4d_09bb_4205_9afd_843de55481d1/
and Shahadat Hossain at IFPRI for their assistance with the selection of Safety%20Net%2020-21_English_2020-11-16.pdf.
study sites and the design of the sampling strategy. Special thanks are Gregan-Paxton, J., John, D.R., 1997. Consumer learning by analogy: A model of internal
due to Mr. Tofazzal Hossain, national consultant at IFPRI, for his support knowledge transfer. Journal of Consumer Research 24 (3), 266–284.
Gregan-Paxton, J., Moreau, P., 2003. How do consumers transfer existing knowledge? A
on logistics management. comparison of analogy and categorization effects. Journal of Consumer Psychology
13 (4), 422–430.
Harding, K.L., Aguayo, V.M., Webb, P., 2018 Mar. Hidden hunger in South Asia: a review
Appendix A. Supplementary material
of recent trends and persistent challenges. Public Health Nutr. 21 (4), 785–795.
Harrison, G.W., 2007 Dec 1. House money effects in public good experiments: Comment.
Supplementary data to this article can be found online at https://doi. Experimental Economics. 10 (4), 429–437.
org/10.1016/j.foodpol.2022.102346. HarvestPlus. (2021). Households reached model – Bangladesh. Internal HarvestPlus
Monitoring Report. International Food Policy Research Institute: Washington, DC.
Herrington CL, Maredia M, Ortega DL, Taleon V, Birol E, Sarkar AR. Bangladeshi
References Consumers’ Willingness to Pay for Rice with Improved Nutrition Via Biofortified
Zinc Rice and Decreased Milling Practices. 2021.
Hoeffler, S., 2003. Measuring preferences for really new products. J. Mark. Res. 40 (4),
Ahmed, S.M., Evans, T.G., Standing, H., Mahmud, S., 2013. Harnessing pluralism for
406–420.
better health in Bangladesh. The Lancet 382 (9906), 1746–1755.
Hoffmann, V., Moser, C.M., Herrman, T.J., 2021 Jan. Demand for Aflatoxin-Safe Maize in
Andrade, J.E., Ali, A.N., Chowdhury, R., Crost, B., Hoffmann, V., Mustafa, S., Shaima, N.
Kenya: Dynamic Response to Price and Advertising. Am. J. Agric. Econ. 103 (1),
A., 2021. Rice fortification in Bangladesh: Technical feasibility and regulatory
275–295.
requirement for introducing rice fortification in public modern storage/distribution
Jeon, D.W., Park, J.R., Jang, Y.H., Kim, E.G., Ryu, T., Kim, K.M., 2021 Dec. Safety
of fortified rice through PFDS channels. International Food Policy Research Institute
verification of genetically modified rice morphology, hereditary nature, and quality.
(IFPRI).
Environ. Sci. Eur. 33 (1), 1–2.
Ara, G., Khanam, M., Rahman, A.S., Islam, Z., Farhad, S., Sanin, K.I., Khan, S.S.,
Khan, J.R., Awan, N., Misu, F., 2016. Determinants of anemia among 6–59 months aged
Rahman, M.M., Majoor, H., Ahmed, T., 2019. Effectiveness of micronutrient-fortified
children in Bangladesh: evidence from nationally representative data. BMC
rice consumption on anaemia and zinc status among vulnerable women in
pediatrics 16 (1), 3.
Bangladesh. PLoS ONE 14 (1), e0210501.

10
R. Chowdhury et al. Food Policy 112 (2022) 102346

Kok AD, Yoon LL, Sekeli R, Yeong WC, Yusof ZN, Song LK. Iron biofortification of rice: Omotilewa, O.J., Ricker-Gilbert, J., Ainembabazi, J.H., 2019 Apr. Subsidies for
progress and prospects. F. Shah, ZH Khan and A. Iqbal (Norderstedt: IntechOpen. agricultural technology adoption: Evidence from a randomized experiment with
2018 Sep 5:25-44. improved grain storage bags in Uganda. Am. J. Agric. Econ. 101 (3), 753–772.
Kőszegi, B., Rabin, M., 2006 Nov 1. A model of reference-dependent preferences. Q. J. Oparinde, A., Banerji, A., Birol, E., Ilona, P., 2016 Mar. Information and consumer
Econ. 121 (4), 1133–1165. willingness to pay for biofortified yellow cassava: evidence from experimental
Kuijken, B., Gemser, G., Wijnberg, N.M., 2017. Categorization and willingness to pay for auctions in Nigeria. Agricultural Economics. 47 (2), 215–233.
new products: The role of category cues as value anchors. J. Prod. Innov. Manage 34 Oparinde, A., Birol, E., Murekezi, A., Katsvairo, L., Disressie, M.T., Nkundimana, J.D.,
(6), 757–771. Butare, L., 2019 Feb 15. Consumer acceptance of biofortified iron beans in rural
Lalani, B., Bechoff, A., Bennett, B., 2019 Jul. Which choice of delivery model (s) works Rwanda: Experimental evidence. Gates Open Res. 3 (367), 367.
best to deliver fortified foods? Nutrients. 11 (7), 1594. Sayeed, K. A., & Yunus, M. M., 2018. Rice prices and growth, and poverty reduction in
Lattin, J.M., Bucklin, R.E., 1989. Reference effects of price and promotion on brand Bangladesh. FAO: Rome, Italy.
choice behavior. J. Mark. Res. 26 (3), 299–310. Srivastava, A., Mukherjee, S., Jebarajakirthy, C., 2020 Mar. Aspirational consumption at
Mazumdar, T., Raj, S.P., Sinha, I., 2005. Reference price research: Review and the bottom of pyramid: A review of literature and future research directions. Journal
propositions. Journal of Marketing 69 (4), 84–102. of Business Research. 1 (110), 246–259.
Meenakshi, J.V., Banerji, A., Manyong, V., Tomlins, K., Mittal, N., Hamukwala, P., 2012 Thaler, R., 1985. Mental accounting and consumer choice. Marketing Science 4 (3),
Jan 1. Using a discrete choice experiment to elicit the demand for a nutritious food: 199–214.
Willingness-to-pay for orange maize in rural Zambia. Journal of health economics. UNICEF, B., 2013. National micronutrients status survey 2011-12. Institute of Public
31 (1), 62–71. Health and Nutrition. Accessed December, 2, 2020.
Milani, P., Carnahan, E., Kapoor, S., Ellison, C., Manus, C., Spohrer, R., van den Berg, G., Vaiknoras, K., Larochelle, C., Birol, E., Asare-Marfo, D., Herrington, C., 2019 Feb.
Wolfson, J., Kreis, K., 2017 Nov 17. Social marketing of a fortified staple food at Promoting rapid and sustained adoption of biofortified crops: What we learned from
scale: generating demand for fortified rice in Brazil. Journal of Food Products iron-biofortified bean delivery approaches in Rwanda. Food Policy 1 (83), 271–284.
Marketing. 23 (8), 955–978. Van Der Straeten, D., Bhullar, N.K., De Steur, H., Gruissem, W., MacKenzie, D.,
Ministry of Health and Family Welfare, Bangladesh 2020. Bangladesh Demographic and Pfeiffer, W., Qaim, M., Slamet-Loedin, I., Strobbe, S., Tohme, J., Trijatmiko, K.R.,
Health Survey 2017-18. Accessed August 15, 2021. https://dhsprogram.com/ 2020 Oct 15. Multiplying the efficiency and impact of biofortification through
publications/publication-FR344-DHS-Final-Reports.cfm. metabolic engineering. Nat. Commun. 11 (1), 1.
Mkambula, P., Mbuya, M.N., Rowe, L.A., Sablah, M., Friesen, V.M., Chadha, M., Osei, A. Von Grebmer K, Bernstein J, Hossain N, Brown T, Prasai N, Yohannes Y, Patterson F,
K., Ringholz, C., Vasta, F.C., Gorstein, J., 2020 Feb. The unfinished agenda for food Sonntag A, Zimmerman SM, Towey O, Foley C. 2017 Global Hunger Index: the
fortification in low-and middle-income countries: quantifying progress, gaps and inequalities of hunger. Intl Food Policy Res Inst; 2017 Oct 11.
potential opportunities. Nutrients. 12 (2), 354. Wang, B., Zhan, S., Sun, J., Lee, L., 2009 Oct. Social mobilization and social marketing to
Moreau, C.P., Lehmann, D.R., Markman, A.B., 2001. Entrenched knowledge structures promote NaFeEDTA-fortified soya sauce in an iron-deficient population through a
and consumer response to new products. J. Mark. Res. 38 (1), 14–29. public–private partnership. Public Health Nutr. 12 (10), 1751–1759.
National Institute of Population Research and Training (NIPORT), Mitra and Associates, Yunus, M., Rashid, S., & Chowdhury, S., 2019. Per capita rice consumption in
and ICF International., 2016. Bangladesh demographic and health survey 2014. Bangladesh: Available estimates and IFPRI’s validation survey results (Vol. 3). Intl
Olson, R., Gavin-Smith, B., Ferraboschi, C., Kraemer, K., 2021 Apr. Food Fortification: Food Policy Res Inst.
The Advantages, Disadvantages and Lessons from Sight and Life Programs. Zeithaml, V.A., 1988. Consumer perceptions of price, quality, and value: a means-end
Nutrients. 13 (4), 1118. model and synthesis of evidence. J. Market. 52 (3), 2–22.

11

You might also like