The Rationalist Conspiracy

Vegetarian Advocacy Is Ineffective

Most pro-vegetarian advocacy is not very effective. The problem isn’t the goal of making animals happy – it’s likely that farm animals have moral value, and almost everyone agrees that factory farm conditions are horrible. Instead, the problem is the most common strategy used to achieve that goal: namely, emotionally-charged rhetoric to convince people, either in person or on the Internet, that they should personally not eat meat.[1] This category of solution to animal rights problems is likely ineffective at best, and downright harmful at worst. As GiveWell says, non-profits shouldn’t “point to a problem so large and severe (and the world has many such problems) that donors immediately focus on that problem – feeling compelled to give to the organization working on addressing it – without giving equal attention to the proposed solution, how much it costs, and how likely it is to work.”

This essay doesn’t address whether animals have nonzero moral value, which has been thoroughly discussed elsewhere. Nor does it look at other solutions to factory farming, like government legislation, or scientific research to develop meat substitutes. It simply tries to show that a lot of pro-vegetarian advocacy, as it’s currently practiced, is ineffective or outright counterproductive. Since there’s a wide range of arguments to consider, this writeup has been broken up into chunks, of which this is the first. This chunk looks at the simplest sub-question: is vegetarian advocacy a cost-effective way of reducing the amount of meat eaten?

First, we should check: is any kind of activism ever cost-effective? The answer seems to be yes. Eg., everyone knows about the campaign for gay marriage, which won a full victory in the US in 2015 (though after many decades of work). However, there seems to be a clear pattern in which activism campaigns are successful, and which aren’t. Psychologist Daniel Kahneman divides the brain into two systems: System 1, which is fast and intuitive, and System 2, which is slow and reflective. System 1 evolved before System 2, and is more connected to the physical actions we take, while System 2 is more closely linked with what we say, write, and think. The historical record shows that activism aimed at System 2 is difficult, but can sometimes be effective. On the other hand, activism aimed at System 1 is usually a waste of effort.

[Edit: The pattern of successful vs. unsuccessful activism still seems real, but the distinction being drawn here is not what Kahneman meant by System 1 and 2. Apologies for the mistake, further clarification to follow.]

For example, consider the history of activism against racism. In 1955, Rosa Parks started the Montgomery Bus Boycott, which triggered a wave of anti-racist advocacy across the US. Though it was a tough battle, after nine years, Congress and President Lyndon Johnson passed the Civil Rights Act, a sweeping bill that outlawed almost all racial discrimination. Over the next fifty years, advocates kept pushing harder and harder for an end to racism. On a System 2 level, this campaign was so successful that virtually no public figure now advocates for segregation, a radical change from 1950. When businessman Donald Sterling was caught being racist, it was such a big deal that it became the focus of his entire Wikipedia page. However, System 1 has been much more stubborn. After sixty years of advocacy, a psychology metric called the Implicit Association Test shows that most white Americans still have negative System 1 assocations with black faces.

There have been many other campaigns to persuade people’s System 1s through rhetoric, advertising, peer pressure, graphic images, and so on, but they usually get negligible or marginal results, compared with the effort invested. Consider smoking as another test case. There’s near-universal agreement that smoking is very, very bad for health, in both the short term and long term, and there’s been enormous efforts to convince smokers to quit. On one side, most smokers themselves know darn well how bad smoking is, and many make heroic efforts to stop. On the other side, governments, nonprofits, and smoking-cessation-aids companies spend billions researching how to help people stop smoking. Thousands of studies have been done on the effectiveness of anti-smoking programs, so we’ve put a lot of effort into finding the very best strategies.

The results of this enormous, expensive, fifty-year effort have been modest at best. The US smoking rate has fallen from ~40% to ~20%, a decline of ~50%, or a bit over 1% per year. Of that decline, much of it was caused by fewer people taking up smoking in the first place. Much of the remainder was caused by laws that make cigarettes more expensive and difficult to use, such as taxes, restrictions on sales, public smoking bans, restaurant smoking bans, and so on. Hence, all anti-smoking programs, cessation aids, addiction research, PR campaigns, etc. combined have given us a few tenths of a percent decline per year.[2] More generally, Scott has a long essay on how these types of programs are ineffective:

“We figured drug use was “just” a social problem, and it’s obvious how to solve social problems, so we gave kids nice little lessons in school about how you should Just Say No. There were advertisements in sports and video games about how Winners Don’t Do Drugs. And just in case that didn’t work, the cherry on the social engineering sundae was putting all the drug users in jail, where they would have a lot of time to think about what they’d done and be so moved by the prospect of further punishment that they would come clean. And that is why, even to this day, nobody uses drugs. (…)

What about obesity? We put a lot of social effort into fighting obesity: labeling foods, banning soda machines from school, banning large sodas from New York, programs in schools to promote healthy eating, doctors chewing people out when they gain weight, the profusion of gyms and Weight Watchers programs, and let’s not forget a level of stigma against obese people so strong that I am constantly having to deal with their weight-related suicide attempts. As a result, everyone… keeps gaining weight at exactly the same rate they have been for the past couple decades.”

To create a quantitative model, we can look at the test case of online ads, which Animal Charity Evaluators (ACE) have done substantial research on. ACE says that “online vegetarianism and vegetarianism ads are currently our most cost-effective intervention recommendation”.[3] In economic terms, one should naively expect that one dollar of ad purchases causes about one dollar of money moved, where “money moved” equals the change in (retail price of goods purchased – marginal cost of goods sold), summed over all relevant goods. If each dollar of ads caused more than one marginal dollar of money moved, companies would just buy more ads to make more money, until decreasing marginal returns brought gains back down to $1.

Of course, that’s only a rough approximation. Any given ad campaign might be more or less effective, for any number of reasons. However, in this case, the sheer magnitude of the gap is cause for great concern. ACE estimates that the cost-per-click (CPC) of pro-vegetarian ads is about two to twenty cents, and that based on survey data, around 2% of ad clickers become vegetarian or vegetarian. American adults spend around $5,000 to $10,000 on food per year,[4] so total money moved through becoming vegetarian is on the order of $100,000. Hence, under the naive economic model, the chance of people becoming vegetarian because of an ad click is roughly 0.00002% – 0.0002%, a massive four to five orders of magnitude smaller than ACE’s estimate. A likely explanation for this, as ACE themselves note, is that people only click on the ads if they were thinking about becoming vegetarian anyway. About two thousand people typically see an online ad for each person who clicks, so even a very small number of existing proto-vegetarians in the ad audience fully account for the survey data.

[Edit: 0.00002% is inaccurate, even within this model. Two cents per click is only available in poorer countries, which have much less total money moved, bringing the estimated odds back to around 0.0002%. However, poorer countries also consume much less meat, which largely compensates for this effect in terms of benefit per dollar.]

What empirical data we have backs this model up. After decades of vegetarian advocacy, PETA says that “society is at a turning point” for veganism. But polling data suggests that only 5% of Americans are vegetarian, and that this percentage has gone down since the 90s. Only 2% consider themselves vegan. Further, these numbers are likely overestimates. Polls often show that a few percent will support any idea, no matter how crazy, like “all politicians are secretly alien lizards” (really!); in addition, most people who said “yes” to the vegan question said “no” to the vegetarian question, which suggests lots of confusion. Even among self-described vegetarians, more detailed surveys show that most still eat an average of one serving of meat per day, which nicely confirms the System 1/System 2 model. It’s easier to convince System 2 that vegetarianism is a good idea than System 1, creating a paradox where two-thirds of ‘vegetarians’ ate meat yesterday.

In addition, even if advocacy is successful, the benefits from one person going vegetarian are not very large compared to their cost. Statistics from vegetarian advocacy groups usually cite the large numbers of animals killed. However, each individual animal life is very short, because meat becomes cheaper when farmers breed animals for rapid growth. Consider chickens as an example. The average American eats 27.5 chickens per year; since a broiler chick takes about five weeks to grow, this gives us 2.64 chicken-years of life prevented by one year of vegetarianism. To evaluate the cost of not eating chicken, we must look at not the price of the chicken, but the “consumer surplus” – how much benefit the customer derives from the product. With some rough math, this comes out to around $18.84 per chicken;[5] this is averaged over both people who like chicken a lot, and people who only like it barely enough to buy it. That gives us a total value-from-chicken (after the cost of the chicken) of $518 per person per year, which can be given up to save 2.64 years of chicken suffering.

Comparing this to human charity, GiveWell estimates a cost-per-child-saved from malaria nets of $2,838. Since GiveWell’s numbers only count children saved, given developing-world life expectancy, each life saved creates about 60 extra person-years. That equals a cost of $47 per person-year, compared to the average cost of $196 per chicken-year from not eating chicken. Therefore, the person-years are a much cheaper buy, even if we assume that chicken lives are so incredibly bad that preventing one chicken-year is as good as saving one person-year.

In fact, this estimate is still biased in favor of chickens, for two main reasons. The first is that GiveWell’s estimate doesn’t include the benefits of mosquito nets beyond saving children; these include saving adults from death by malaria (though adults have a much lower fatality rate), preventing many more non-fatal cases of malaria, preventing mosquito-borne disease in general, and of course preventing mosquito bites, which (ignoring everything else) can be done at a cost of hundredths of a penny per bite. The second reason is that Against Malaria Foundation is selected for being extremely low-risk; given a donation to AMF of $X dollars, one can be extremely confident in at least Y lives being saved. GiveWell thinks it’s likely that if we take on riskier projects, like scientific research and policy reform, the expected cost per life saved will be even lower. Indeed, one of these riskier projects is actually US policy reform to improve farm animal welfare.

Several people have suggested this is an unfair comparison. Most donors have a limited “charity budget”, and $10 that they spend on one charity is $10 which they don’t spend on another. However, what if going vegetarian increases people’s willingness to do other kinds of good? Unfortunately, psychology research suggests this is unlikely – people often have a “do-gooding budget” in addition to their financial budget, and doing one altruistic act will decrease their willingness to do others. To quote Nick Cooney, founder of animal charity The Humane League:

“The contribution ethic refers to the feeling many people have that “I’ve done my part on issue A, so it’s okay for me to ignore issues B, C, and D.” During a Humane League campaign to get restaurants to stop purchasing products from a particularly cruel farm, owners would often tell us that they already do something to help animals (“We buy our eggs from a local farm,” or “I donate to the ASPCA”) so we shouldn’t be bothering them. This phenomenon worked across issues too, as we often had owners or chefs tell us how they supported some other social cause so we should leave them alone about this one.

In addition to feeling like they’ve done their part and therefore don’t need to do anything more, people often overestimate the amount of good they’ve done. Combined, these phenomena make it hard to move people beyond small actions for their one or two preferred causes (Thogersen and Crompton 2009).”

Footnotes

1. Of course, there are many different animal-rights diets. Other types of ethics-based dietary restrictions include “vegan”, “lacto-ovo vegetarian”, “pescatarian”, “flexitarian”, “reducetarian”, and so on. Since keeping track of all the different labels is unwieldy, for the most part I’ll simply say “vegeterian”, even though the main arguments also apply to many other ethics-based diets. I’ve seen people use “veg*n”, but that’s also unwieldy due to ‘*’ not being a letter.

2. One possible complication is that nicotine is chemically addictive. However, treatments to fight the chemical part of addiction have been available for decades, which I’d expect to largely cancel out this effect. In addition, similar problems (alcoholism, gambling, sugar consumption, etc.) have generally seen similar results.

3. Although I disagree with them on several issues, ACE should be commended for trying to be quantitative.

4. Data from this Gallup poll. Note that this data is based on self-reports, and is per-household rather than per-adult, so it is only approximate.

5. The cost of a whole chicken is roughly $1.50 per pound per this article, or $9 for a six-pound chicken. From this paper, the price elasticity of demand of chicken is around -0.8, so we can naively model the demand-price curve with the ODE dy/dx = -0.8*y/x, y(9) = 1, which has the solution y(x) = 5.8/x^0.8. Integrating from x = 9 to, say, 100, we get 27.84, which subtracting the $9 for the chicken’s price gives us a consumer surplus of $18.84. This is, of course, just a rough guess.