As discussed in McFadden, et al., 2022, when focusing on policies and behaviors related to climate change and sustainability we should consider actual abatement potential and plasticity.
A lot of important work needs to be done to understand which solutions are technically correct but also the most impactful at scale from a behavior change perspective. And importantly - ethical and honest. It is important when designing for behavior change that choice architectures reflect the science and honestly represent tradeoffs that are relevant to the context and particulars of circumstances and place. Often behavioral designers are challenged by the fact that nudges are sensitive to context, so may be less impactful when context changes in different environments. Good science and good business practices and good ethics mandate testing in different contexts to understand the impact. We must also recognize, as I discuss below, that the science and tradeoffs that can be implicitly baked into choice architecture also depend on context and must be considered. When we design choice architectures it should reflect this in ways that are honest and transparent, and not let politics and personal biases get mixed into the batter. If nudges are successful, we want to make sure we are not doing more harm than good at scale. We don't want to unintentionally embed misinformation into product design, and certainly don't want to do this willfully (see here and here for posts discussing misinformation getting been baked into 'GMO' labeling).
Blondin et al. (2022) investigate adding descriptive messages to nudge consumers to choose plant based food choices. They find that the most impactful framing was related to a 'small changes big impact' frame:
When it comes to measuring climate impact the important human, sustainability, and nutrition tradeoffs related to the metrics we use matter. In Blondin et al. (2022) it was not clear to me how they defined the relationship between food choices and greenhouse gas emissions. In the article above De-loyde et al. (2022) they cite Espinoza-Orias & Azapagic (2012) in relation to greenhouse gas emissions. While that analysis was super detailed and rigorous, I'm not sure it represents to most accurate or transparent information when it comes to making these choices as it relates to CO2 equivalents.
When considering beef and dairy specifically, understanding the differences in the way CO2 vs. methane behaves is fundamental to understanding their respective roles impacting climate change, and personal and policy decisions related to mitigating future warming. Understanding this can help direct attention to those areas where we can make the biggest difference in terms impacting climate change over the long term. As discussed in Allen et al. (2018):
It's not clear that the work cited from 2012 cited in De-loyde et al. (2022) or the work in Blondin (2022) appropriately accounts for the biogenic carbon cycle and the role of methane as a flow vs. stock gas in their estimates of GHG emissions and warming potential.
Are consumers scanning QR codes on their smartphones (pumping new permanent sources of GHG into the atmosphere) to read a menu nudging them away from beef being distracted from focusing on other seemingly arbitrary choices they could be making to positively impact the climate in more meaningful ways? (This post compares GHG emissions and warming potential of internet use vs. beef consumption)
Additionally, even an accurate measure of GHG emissions and warming potential isn't enough to transparently relate all of the tradeoffs involved.
In a 2010 Food and Nutrition Research article, authors introduce the Nutrient Density to Climate Impact (NDCI) index. Metrics like this could add some perspective. According to their work:
Authors Drewnowski, Adam et al. apply this more nuanced approach to 34 different food categories including meat and dairy:
While it does not impact the major findings related to how to influence consumer choices about food and sustainability, it is not clear to me that Blondin et al. (2022) or De-loyde et al. (2022) sufficiently consider these tradeoffs in the choice architectures that they proposed for nudging consumers to make decisions about food consumption.
If appropriately accounted for in their specific contexts, it does not necessarily imply that these tradeoffs would be appropriately accounted for in different contexts which is what I turn to next.
Context Matters
When it comes to behavior change, context and environment matter immensely.
As discussed in Kanemoto et al. (2019)
"most global-scale models have an important shortcoming; they do not consider subnational variations in food production and consumption. This lack of subnational detail is significant because subnational detail could be more important than global coverage and may show the opportunity to promote and use different subnational policy"
Using combined microdata on 60,000 households collecting information about diet, income, and demographics and a subnational input-output model for production and trade across 47 prefectures in Japan they find:
"higher-CF [carbon footprint] households are not distinguished by excessive meat consumption relative to other households but rather have higher household CF intensity because of elevated consumption in other areas including restaurants, confectionery, and alcohol."
De-loyde et al. (2022) start off their article with a global framing of the impact of livestock on GHG emissions:
"Livestock production contributes an estimated 14.5% of human-induced global greenhouse-gas emissions."
Despite some of the questions above about the metrics used to relate food to climate impact and the nutritional tradeoffs involved when presenting consumers with options, maybe this global framing is appropriate for consumers in the U.K. but is this the most honest framing to use if we try to transport these results to other consumers like U.S. consumers? Are the tradeoffs the same?
Kanemoto (2019) was based on Japanese consumers who consume relatively low amounts of beef compared to U.S. consumers. When considering U.S. consumers, there are important differences we should consider when attempting to adopt choice architectures used in other contexts designed to influence consumer choice.
Ignoring change in context ignores important differences between technological capabilities and production practices but also differences in incomes, tastes, and preferences. As stated in a recent article in Foreign Policy:
"Generalizations about animal agriculture hide great regional differences and often lead to diet guidelines promoting shifts away from animal products that are not feasible for the world’s poor....A nuanced approach to livestock was endorsed in the latest mitigation report of the U.N. Intergovernmental Panel on Climate Change (IPCC)."
When we are designing choice architectures to influence dietary choices related to climate and sustainability, a nuanced approach is also necessary.
A lot of the criticism of U.S. beef misappropriates or conflates the environmental footprint of beef produced and consumed in the U.S. with beef from other parts of the world. These criticisms may not be appropriate when applied to U.S. consumers. When we drill into agriculture and focus on beef in the U.S. we find that it accounts for about 4% of total emissions. But on a global scale, which matters most to climate change, overall, total GHG emissions related to U.S. beef consumption are 10X lower in the U.S. than beef produced in other places in the world and accounts for less than 1/2 of 1% (i.e. .5%) of global GHG emissions. (EPA GHG Emissions Inventory, Rotz et al, 2018). See also Allen, M.R., Shine, K.P., Fuglestvedt, J.S. et al., 2018.
Similar to Kanemoto et al. (2019) and McFadden, et al., 2022 when it comes to U.S. consumers it may be the case that the real meat on the bone so to speak (what has the most potential impact given consumer plasticity and realistic assessments of climate impact) as it relates to climate may have more to do with where the food is consumed than what is consumed. It may be the case that once you are already in the restaurant reading the menu that having a salad vs. steak sourced in the U.S. is inconsequential when looking at the global impact. It could even be the opposite case, considering nutritional density and climate tradeoffs, when appropriately quantified, the salad in the restaurant may not be the best choice.
Scaling and Ethics
In their paper Monetizing disinformation in the attention economy: The case of genetically modified organisms (GMOs) Ryan, Schaul, Butner and Swarthout provide an in depth background on the attention economy, disinformation, the role of the media and marketing as well as socioeconomic impacts. They articulate how how rent seekers and special interests are able to use disinformation in a way to create and economize on misleading but coherent stories with externalities impacting business, public policy, technology adoption, and health. These costs, when quantified can be substantial and should not be ignored:
"Less visible costs are diminished confidence in science, and the loss of important innovations and foregone innovation capacities"
One of the big challenges presented by misinformation and disinformation is its clever use of our fast thinking system 1, its use of social proof and confirmation bias, magnified by technology and social media that focuses on capturing attention and dissemination vs. communication and persuasion. To say the least it can be dangerous at scale. In today's information age and polarized political climate we need to be careful about the things we do at scale. Low cost scalable interventions can be dangerous.
If lots of businesses used nudges and choice architectures that did not accurately reflect the tradeoffs for a given context, or government became involved as an enforcer, or it became a common strategy for corporations to juice up their ESG reporting, these issues could have negative impacts at scale. The harm would be greatest if this becomes a distraction and diverts attention and resources away from more impactful behavior change or reduces investment in greener food technologies.
We can look at what has happened to the use of rBST in milk production for instance. According to Perkowski (2013) "more than two-thirds of dairy farmers who have ever treated their cows with rBST have stopped using it" despite the fact that rbST in dairy (just one technology) has the equivalent impact of removing 300,000 cars from the road annually (see Capper, 2008). Would consumers favor milk produced from cows supplemented by rBST (leading to an economic premium to encourage its adoption by producers) if they were made more aware of these tradeoffs? Are current labels related to rBST doing consumers justice by providing them better information or are they reinforcing false negative perceptions?
We could ask similar questions regarding other technological advances in food production and manufacturing like finely textured beef which was so maligned by social media that its developer Beef Products Incorporated (BPI) forced a settlement with NBC for spreading misinformation characterizing it as unsafe pink slime (Mclaughlin, 2017).
Do current GMO labeling regulatory requirements and food packaging accurately reflect the tradeoffs involved? Would consumers view GMOs differently if they were presented with knowledge that their adoption currently has offset the emissions equivalent to 15 million cars annually. That is more than the total new cars typically sold in the U.S. every year, and about 20% of all new cars sold globally. Or 3X the number of EVs currently being sold in the U.S. annually.
Whether intentionally designed or not we are always nudging. Are we nudging consumers in the wrong direction (in terms of environmental impact) with labels claiming 'no rBST' on milk cartons and non-GMO labels on other food products?
Damage can be done when attention and perceptions are manipulated and consumer sentiment leads us to abandon promising technologies that could really make a difference when it comes to climate.
Misinformation and disinformation have played a major role driving vaccine hesitancy and may have even been accelerated during the COVID19 pandemic. See Johnson et al. (2020) & Vaidyanathan (2020). Both misinformation and disinformation play large roles in other areas like climate denialism and GMO hesitancy and act as bottlenecks to adopting better technologies and policies. Behavioral designers often strive to identify nudges that are both as effective and scalable. In the research above, De-loyde et al. (2022) and Blondin et al. (2022), social nudges seem to fit the bill here. But misinformation and disinformation unfortunately represent some of most effective and scalable social nudges you will find. Even with the best of intentions, we need to be careful in the design and deployment of social nudges.
When it comes to developing nudges and choice architectures related to food choices, we want to be careful that what we do is based on sound science and transparently reflects tradeoffs vs. simply manipulates consumers to act according to our own biases and preferences that may not necessarily represent the optimal choice in every context.
A Guide for Ethical Nudging
The Observatory of Public Innovation has recently published a document titled Good Practice Principles For Ethical Behavioural Science In Public Policy that includes checklists and questions that may speak to many of the issues above.
They provide a checklist and prompting questions focused on four areas: Scope, Design, Research and Evaluation, and Policy Implementation. Below are some relevant prompting questions based on the discussion above:
- Did you establish clear criteria for why the behavioural change has a positive outcome for the affected population? Are these criteria monitored and evaluated regularly? (related to Scope)
-Set up protocols to identify and mitigate ethical risks (such as unintended negative side-effects, both in general and to particular groups (related to Design)
- Have you considered new ethical concerns resulting from scaling and adapting in new contexts?
The criteria for the nudges mentioned above seem to be based on measures of CO2e. But we know that those metrics don't necessarily reflect the latest science and don't necessarily encompass all the relevant tradeoffs implied in the choice architectures being presented to consumers. As a consequence, if these nudges were scaled, we might nudge people to make choices that may not be as optimal as advertised when the goal is climate impact, especially for certain groups in given contexts. These are important side effects to consider.
Perhaps in the research and design phase when there is debate in the literature or the metrics are complex (as they are in climate science and nutrition and health) it is hard to claim that any of the papers discussed above represent a marked violation of ethics. No standard of ethics should require human infallibility or perfect knowledge - especially when the goals are learning. I don't think this guide is necessarily meant to catch all of the nuances discussed in this post. But when it comes to adoption and scaling of interventions or policies, the ethics involved probably merit greater attention. I think the checklist and prompting questions as they are may be quite useful and can be refined to better reflect certain domains.
Conclusion
Learnings from studies like Blondin et al. (2022) and De-loyde et al. (2022) are very important for understanding how choice architecture and nudges can be used to influence behaviors that may represent impactful solutions for important societal problems like climate change. We should take these learnings and test them in other contexts to confirm they work in different environments before scaling. We also want to make sure that as we learn about nudging and behavior change to improve choices related to food and sustainability in non-coercive ways, that we do not inadvertently do more harm than good on a grander scale in terms of loss of trust in the science, institutions, and innovations that can have the greatest impact. We need to make sure that we are using choice architecture responsibly, and the ingredients are based on sound science and transparent in the tradeoffs they represent and sensitive to the context in which choices are being made.
Additional and Related Reading
More recent related posts and updates:
Nudging Back: Turning off your Zoom Camera May Be Good for the Climate. https://ageconomist.blogspot.com/2023/02/nudging-back-turning-off-your-camera.html
Picture This: Putting Beef and Climate into Perspective. https://ageconomist.blogspot.com/2023/03/putting-beef-and-climate-into.html
Rational Irrationality and Behavioral Economic Frameworks for Combating Vaccine Hesitancy https://ageconomist.blogspot.com/2021/08/rational-irrationality-near.html
Facts, Figures, or Fiction: Unwarranted Criticisms of the Biden Administration's Failure to Target Methane Emissions from Livestock. https://ageconomist.blogspot.com/2021/12/facts-figures-or-fiction-unfair.html
The Limits of Nudges and the Role of Experiments in Applied Behavioral Economics. https://ageconomist.blogspot.com/2022/06/the-limits-of-nudges-and-role-of.html
Will Eating Less U.S. Beef Save the Rainforests? http://realclearagriculture.blogspot.com/2020/01/will-eating-less-us-beef-save.html
Modern Sustainable Agriculture Annotated Bibliography (updated) http://ageconomist.blogspot.com/2011/02/modern-sustainable-agriculture.html
References
Allen, M.R., Shine, K.P., Fuglestvedt, J.S. et al. A solution to the misrepresentations of CO2-equivalent emissions of short-lived climate pollutants under ambitious mitigation. npj Clim Atmos Sci 1, 16 (2018) doi:10.1038/s41612-018-0026-8
Blondin, Stacy & Attwood, Sophie & Vennard, Daniel & Mayneris, Vanessa. (2022). Environmental Messages Promote Plant-Based Food Choices: An Online Restaurant Menu Study. World Resources Institute. 10.46830/wriwp.20.00137.
Graham Brookes & Peter Barfoot (2020) Environmental impacts of genetically modified (GM) crop use 1996–2018: impacts on pesticide use and carbon emissions, GM Crops & Food, 11:4, 215-241, DOI: 10.1080/21645698.2020.1773198
The environmental impact of recombinant bovine somatotropin (rbST) use in dairy production Judith L. Capper,* Euridice Castañeda-Gutiérrez,*† Roger A. Cady,‡ and Dale E. Bauman* Proc Natl Acad Sci U S A. 2008 July 15; 105(28): 9668–9673
De-loyde, K., Pilling, M., Thornton, A., Spencer, G., & Maynard, O. (2022). Promoting sustainable diets using eco-labelling and social nudges: A randomised online experiment. Behavioural Public Policy, 1-17. doi:10.1017/bpp.2022.27
Drewnowski, Adam et al. “Energy and nutrient density of foods in relation to their carbon footprint.” The American journal of clinical nutrition 101 1 (2015): 184-91 .
Johnson, N.F., Velásquez, N., Restrepo, N.J. et al. The online competition between pro- and anti-vaccination views. Nature 582, 230–233 (2020). https://doi.org/10.1038/s41586-020-2281-1
Keiichiro Kanemoto, Daniel Moran, Yosuke Shigetomi, Christian Reynolds, Yasushi Kondo,Meat Consumption Does Not Explain Differences in Household Food Carbon Footprints in Japan, One Earth, Volume 1, Issue 4, 2019,Pages 464-471, ISSN 2590-3322, https://doi.org/10.1016/j.oneear.2019.12.004.
Private costs of carbon emissions abatement by limiting beef consumption and vehicle use in the United States. McFadden BR, Ferraro PJ, Messer KD (2022) Private costs of carbon emissions abatement by limiting beef consumption and vehicle use in the United States. PLOS ONE 17(1): e0261372. https://doi.org/10.1371/journal.pone.0261372
ABC TV settles with beef product maker in 'pink slime' defamation case. By Timothy Mclaughlin https://www.reuters.com/article/us-abc-pinkslime/abc-tv-settles-with-beef-product-maker-in-pink-slime-defamation-case-idUSKBN19J1W9
Dairymen reject rBST largely on economic grounds. Mateusz Perkowski Dec 10, 2013 Updated Dec 13, 2018 .https://www.capitalpress.com/dairymen-reject-rbst-largely-on-economic-grounds/article_335ce36b-ec75-5734-9aad-80f11cae1d43.html
Camille D. Ryan, Andrew J. Schaul, Ryan Butner, John T. Swarthout, Monetizing disinformation in the attention economy: The case of genetically modified organisms (GMOs), European Management Journal, Volume 38, Issue 1, 2020, Pages 7-18, ISSN 0263-2373
C. Alan Rotz et al. Environmental footprints of beef cattle production in the United States, Agricultural Systems (2018). DOI: 10.1016/j.agsy.2018.11.005
News Feature: Finding a vaccine for misinformation. Gayathri Vaidyanathan. Proceedings of the National Academy of Sciences Aug 2020, 117 (32) 18902-18905; DOI: 10.1073/pnas.2013249117 https://www.pnas.org/content/117/32/18902
