This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
Author's personal copy
Economics Letters 119 (2013) 325–327
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/ecolet
‘Nudging’ hotel guests to reduce food waste as a win–win
Steffen Kallbekken 1, Håkon Sælen ∗
CICERO Center for International Climate and Environmental Research – Oslo, P.O. Box 1129 Blindern, 0318, Oslo, Norway
•Wasted food embodies large environmental impacts.
•We conduct a field experiment to test two ‘nudges’ to reduce food waste.
•Both measures reduce food waste in hotel restaurants by around 20%.
•The measures also reduce private costs, making a win–win outcome likely.
Received 20 December 2012
Received in revised form
23 February 2013
Accepted 8 March 2013
Available online 18 March 2013
We show that two simple and nonintrusive ‘nudges’ – reducing plate size and providing social cues
– reduce the amount of food waste in hotel restaurants by around 20%. The results are statistically
significant. They are also environmentally substantial as food waste is a major contributor to climate
change and other forms of environmental degradation. Given the magnitude of the contribution of food
waste to global environmental change, it is surprising that this issue has not received greater attention.
The measures reduce the amount of food the restaurants need to purchase, and there is no change in guest
satisfaction, making it likely that profits will increase. The measures thus constitute potential win–win
©2013 Elsevier B.V. All rights reserved.
The environmental impact of food provision is well-known.
Food accounts for 20% of global greenhouse gas emissions
(Hertwich and Peters, 2009), and 92% of the global water footprint
is related to agriculture (UNEP, 2012). In addition are issues such as
land degradation, overfishing and local air and water pollution. The
fact that roughly one-third of all food is lost or wasted (Gustavsson
et al., 2011) has received less attention. It therefore appears that
food waste is a substantial, but largely neglected contributor to
environmental change. The IPCC (Metz et al., 2007), for instance,
∗Corresponding author. Tel.: +47 22 85 85 63.
E-mail addresses: email@example.com (S. Kallbekken),
firstname.lastname@example.org,email@example.com (H. Sælen).
1Tel.: +47 22 85 87 58.
does not list any measures to reduce food waste among the more
than 50 mitigation policies highlighted in its summary for policy
Choice architecture can be used to alter people’s behavior in
predictable ways (Thaler and Sunstein, 2008). From obesity and
nutritional research we know that ‘‘the eating situation often (but
not always) provides clues allowing us to infer how much we can
eat without eating an inappropriately large amount’’ (Herman and
Polivy, 2005). These clues form a part of the choice architecture,
and can be used to alter behavior. Our focus is on reducing food
waste, and thereby also reducing emissions of greenhouse gases.
We take as our starting point the observation that some of the
measures aimed at reducing food intake have been reported to also
reduce food waste (Freedman and Brochado, 2010).
Our study includes a field experiment (Harrison and List,
2004) where we intervene to change two variables of interest,
and a complementary observational study, based on pre-existing
variation in one of the variables of interest. In the field experiment
0165-1765/$ – see front matter ©2013 Elsevier B.V. All rights reserved.
Author's personal copy
326 S. Kallbekken, H. Sælen / Economics Letters 119 (2013) 325–327
we test the effectiveness of two treatments in reducing the amount
of food waste generated. Both treatments rely on influencing
consumption norms through external cues.
The first treatment concerns the effect of plate size on the
amount of food waste. It has been argued that ‘‘plate shape and size
delineate norms for appropriate amounts of food to eat at a meal
(Sobal and Wansink, 2007) and it has been shown that’’. .. big bowls
lead to overserving, small bowls lead to underserving. .. ? (Ittersum
and Wansink, 2012). In addition to the social cue it provides, larger
plates might also contribute to people serving and consuming
more food due to visual illusions that lead to biased perceptions
of how much food is served or consumed (Ittersum and Wansink,
2012). Combining this with the finding that increased portion
size leads to both increased food intake and increased food waste
(Freedman and Brochado, 2010), it seems a reasonable hypothesis
that decreasing plate size will decrease the amount of food waste.
In the field experiment the typical plate size reduction was from
24 to 21 cm (the plates were used for breakfast buffets, and, if
applicable, also at lunch and dinner buffets). In the observational
study the average plate size ranged from 15 to 28 cm (average
The second treatment is to provide a more direct social cue by
displaying a sign at the buffet that encourages restaurant guests
to help themselves more than once. The text reads, in seven
different languages: ‘‘Welcome back! Again! And again! Visit our
buffet many times. That’s better than taking a lot once’’. The sign
is intended to make it salient that it is socially acceptable to
help yourself more than once from the buffet, which might affect
behavior as ‘‘just as people often look to portion size for guidance
in eating situations ... so they may rely on the example of others
for guidance, when such examples are salient’’ (Herman and Polivy,
2005). The hypothesis is that the sign will encourage guests to load
less food on their plates each time they serve themselves, in turn
reducing the amount left over.
The two treatments are tested in an experiment conducted
in collaboration with hotel restaurants. From a hotel chain we
recruited individual hotels to implement the two treatments, and
the remaining hotels made up the control group. A total of 52 hotels
delivered data which we could use in the final analysis. There
were 7 in each treatment group. The experiment was implemented
between June 1st and August 15th 2012. All hotel restaurants in
the study recorded and reported the amount of food waste daily
during the whole period. The treatment hotels implemented their
respective treatments from July 1st until August 15th.
We estimate a difference-in-difference model (Card and
Krueger, 1994) using a fixed effects panel regression to analyze the
treatment effects. The difference-in-difference method controls for
pre-treatment differences between the hotels and for trends over
time that are unrelated to the intervention. In addition, we control
for the number of guests staying at the hotel (expected to be a
good predictor of the number of breakfasts served), and food sales
revenue (in 1000s of NOK, a useful proxy for the number of meals
served, excluding breakfast as this is commonly included in the
price). Food waste from hotel iat time tis modeled as follows:
Wasteit =β11D1i+β12 D2i+ · · · + β1nDni +β2Guestsit
+β3Food salesit +γti+δ(Ti×ti)+eit
where D1i. . . Dni are hotel specific intercept dummy variables
defined so that
0 otherwise,D12 =1i=2
0 otherwise, . . . ,
Average amount of food waste (kg) per hotel in the control group (38 hotels) and
test groups (7 hotels in each group), before and after the treatment was introduced.
Standard deviations in brackets.
Group Pre-treatment food
waste (kg, average per
waste (kg, average per
Control 35.07 32.98
Reduced plate size 36.88 25.84
Salient sign 47.76 34.25
These control for average permanent differences between hotels,
ti=0 before treatment begins
1 after treatment begins
controls for time trend common to control and treatment groups.
Tiis defined so that
Ti=0 for hotels in control group
1 for hotels in treatment group.
The random error term eit is assumed to be i.i.d. normal. It is the
coefficient of the interaction between tiand Tithat measures the
effect of the treatment. A separate regression is estimated for each
of the treatments.
In the observational study, we utilize pre-existing differences
in plate size across hotels in the control group. We use a panel
regression also here. Food sales revenue and guests are still
controlled for. As the fixed effect model cannot handle variables
that are constant within hotels over time, which plate size is in this
case, we specify a random effects model:
Wasteit =β1+β2Plate sizeit +β3Guestsit
+β4Food salesit +eit +vi
where β1is the average intercept and viare random hotel-specific
deviations from the average.
The experiment indicates that reducing the plate size reduces
food waste by 19.5% (p<0.001), and that introducing the sign
pointing out that guests can help themselves more than once
reduces food waste by 20.5% (p<0.001). Descriptive statistics by
time period and treatment group are reported in Table 1, while
the regression results are shown in Table 2. All the estimated
regression coefficients are significant at 0.1% level or lower. The
percentage treatment effects are found by dividing the coefficient
by the mean pre-treatment level from Table 1. The hotel-specific
coefficients are not reported. Random effect estimations were also
run, giving essentially the same results, however Hausman tests
conclude that these estimators are inconsistent and we hence
report the fixed effect estimations. The R2values are 0.36 and 0.39
The plate size treatment in the field experiment is supported
by the observational study, where we measure the strength
of association between plate size and food waste among the
untreated hotels. The results from the latter are reported in Table 3.
The R2value is 0.42. The percentage effect is approximated by
dividing the coefficient by the overall mean waste level in the
control group (33.82 kg). The results suggest that a 1 cm reduction
in plate size reduces food waste by 2.5 kg (p<0.01), which is
7.4% of the overall mean in the control group. This implies that
a 3 cm reduction reduces waste by approximately 22%, which is
Author's personal copy
S. Kallbekken, H. Sælen / Economics Letters 119 (2013) 325–327 327
Estimated coefficients (and the associated standard errors) from the difference-in-
Plate size Salient sign
Guests 0.033 0.038
Food sales 0.138 0.171
Time trend −4.317 −4.428
Treatment effect −7.179 −9.772
well within the 95% confidence interval for the treatment effect.
While lacking the within-subject design of the experiment, the
observational study benefits from a larger number of observations
than contained in the treatment groups.
We have identified and quantified two no-regrets measures
that can substantially reduce the amount of food waste from
hotel restaurants. The findings, at least for the effect of reduced
plate size, are likely to be transferable to other contexts such as
food services at institutions (schools, hospitals, retirement homes,
prisons, workplace canteens, etc.) where buffet meals are served.
The results for the effect of plate size are the strongest, since the
experimental results are supported by an observational study.
The cost of the plate size measure is negative. All restaurants
regularly have to replace plates, and as smaller plates are cheaper
to purchase than larger one, the cost of the measure will be
negative as long as the rate of replacement remains the same.
The cost of the second measure is minimal (printing 10–30 small
posters per hotel). Reducing food waste represents a financial
saving to the hotel, estimated at around NOK 50/kg (USD 9/kg) by
one of the hotels. As 1 kg of food waste is responsible for lifecycle
emissions of around 1.9 kg of CO2e (European Commission, 2010),
the negative cost of each measure is perhaps as large as USD 4700
per ton of CO2.
One potential concern is that customer satisfaction could be
negatively affected, for instance because customers have to return
to the buffet more often to fill the smaller plates, or because
larger plates produce a more luxurious feeling. The hotels use
an online survey tool to record customer satisfaction with the
restaurants. Customer satisfaction with the buffet breakfast within
each treatment group was essentially constant from the pre-
treatment period to the treatment period. A simple difference-
in-difference with more than 45,000 observations shows no
significant change for the treatment hotels relative to the hotels
in the control group.
Our study leaves many important questions unaddressed, e.g.
the effect of implementing the two measures jointly or the optimal
plate size in terms of minimizing food waste, but it provides a
strong indication that using simple nudges to reduce food waste
might represent a very fruitful approach to achieving significant
greenhouse gas emission reductions. There is reason to believe
that the measures are also privately profitable as they reduce
food expenditure while having no effect on customer satisfaction
Estimated coefficients, standard errors and p-values for the observational analysis.
and requiring minimal costs of implementation. As such, they
represent an example of a strategy that makes both environmental
and business sense, giving anecdotal support to the controversial
Porter-hypothesis (Porter and van der Linde, 1995).
We thank Cathrine Dehli at Nordic Choice Hotels for substantial
help with planning and implementing the experiment, as well
as the hotel employees who recorded data on food waste, and
helped us implement the treatments. We are indebted to Bård
Romstad for helpful support with data management. Thanks to
Hege Westskog, Kristin Linnerud, Torben Mideksa and Todd Cherry
for valuable comments. The research was funded by the not-for-
profit organization GreeNudge.
Card, D., Krueger, A.B., 1994. Time-series minimum-wage studies: a meta-analysis.
American Economic Review 85, 238–243.
European Commission, 2010. Preparatory study on food waste across EU 27.
Technical Report 2010-054. European Commission, Brussels.
Freedman, M.R., Brochado, C., 2010. Reducing portion size reduces food intake and
plate waste. Obesity 18, 1864–1866.
Gustavsson, J., Cederberg, C., Sonesson, U., van Otterdijk, R., Meybeck, A., 2011.
Global food losses and food waste: extent, causes and prevention, In: FAO (Ed.),
Harrison, G.W., List, J.A., 2004. Field experiments. Journal of Economic Literature 42,
Herman, C.P., Polivy, J., 2005. Normative influences on food intake. Physiology &
Behavior 86, 762–772.
Hertwich, E.G., Peters, G.P., 2009. Carbon footprint of nations: a global, trade-linked
analysis. Environmental Science & Technology 43, 6414–6420.
Ittersum, K.V., Wansink, B., 2012. Plate size and color suggestibility: the Delboeuf
illusion’s bias on serving and eating behavior. Journal of Consumer Research
Metz, B., Davidson, O.R., Bosch, P.R., Dave, R., Meyer, L.A. (Eds.), 2007. Contribution
of Working Group III to the Fourth Assessment Report of the Intergovernmental
Panel on Climate Change. Cambridge University Press, Cambridge, United
Porter, M., van der Linde, C., 1995. Green and competitive. Harvard Business Review
Sobal, J., Wansink, B., 2007. Kitchenscapes, tablescapes, platescapes, and foodscapes.
Environment and Behavior 39, 124–142.
Thaler, R., Sunstein, C., 2008. Nudge: Improving Decisions About Health, Wealth and
Happiness. Yale University Press, New Haven, USA.
UNEP, 2012. Global environment outlook 5—summary for policy makers. UNEP,