The Causal Inefficacy Objection is Provably Wrong

This can be shown both by evidence and by basic economics

Part 1: Explaining the causal inefficacy objection

Michael Huemer often argues for moral realism by pointing to the existence of very obvious moral truths. Often his go-to is the following: it’s wrong to torture babies for trivial pleasure. Well, it turns out that we do that—it’s called eating meat.

Why babies, you ask? Well, most of the animals we eat are, in fact, babies.

Factory farming is very obviously very horrible. This is perfectly clear—it causes more suffering, very plausibly, than all other actions in human history, every few years.

May The Factory Farms Burn

It is very difficult for anyone of sense and decency to defend factory farms—that’s why the odious task of defending them is left to dishonest industry shills and Timothy Hsiao. However, lots of smart and sane people like Danny Shahar think that it’s fine to eat factory-farmed meat because you’ll have no effect on the industry. The industry is so complicated, or so they claim, that one individual consumer won’t have any impact. Thus, if you abstain from meat, you won’t have any effect.

My preferred argument against eating meat is the following.

1. Eating meat causes vast amounts of suffering for the sake of trivial gains

2. It’s wrong to cause vast amounts of suffering for the sake of trivial gains

Therefore, eating meat is wrong.

Premise 2 is very obvious. But premise 1 is the one that this argument contests—even factory farms cause tons of suffering, the argument claims that the industry is too complex for you to have any effect.

Here, I will explain why the causal inefficacy objection is utterly unpersuasive—it is provably wrong. I will go on to provide various other reasons to doubt the causal inefficacy objection, before explaining why even if it’s probably true you still shouldn’t eat meat, before explaining why even if you think it’s definitely true, you still shouldn’t eat meat. With a sophisticated understanding of economics, the objection can be refuted purely from the armchair—and here, I’ll explain how. I will try to explain the economic logic as simply and clearly as possible; by the end of this, you should have a clear picture of why the causal inefficacy objection not merely does not work, but cannot work, in a much deeper sense. This article may be long and a bit dry—sorry about that—but I can promise you one thing: by the end of it, the causal inefficacy objection will be totally debunked. While causal inefficacy sounds plausible at first, it is easy to prove that it’s wrong.

Part 2: The basic reason that it’s false

Why? But it’s totally sound. I’m going to go and kill an antelope. And every time I plunge my knife into its heart, I shall scream, with blood dripping from my face, “No single person can affect the supply chain!”

—Amos Wollen, who just started a substack that you should all immediately read.

1. On average, if a consumer has no special reason to think that their action will not trigger any thresholds, a consumer should expect to have the average marginal impact of consumers.

2. The average marginal impact of consumers consuming one chicken is increasing chicken production by more than one, multiplied by the value of the cumulative elasticity factor.

3. Therefore, on average, if a consumer has no special reason to think that their action will not trigger any thresholds, a consumer who consumes one extra chicken should expect to increase chicken production by more than one, multiplied by the value of the cumulative elasticity factor

4. The cumulative elasticity factor of chicken consumption is around .75

5. Therefore, on average, if a consumer has no special reason to think that their action will not trigger any thresholds, a consumer who consumes one extra chicken should expect to increase chicken production by more than .75 chickens.

This is, as stated, quite a confusing mess that will be hard to understand for most people. In the next section, I’ll clarify what it means. Here, I talk about chicken in the example, but it generalizes to all animal products. Thus, here I take what’s called the expected value response—while it’s true that most times people eat chicken it has no effect, sometimes it has a huge effect, such that the average effect is roughly the same as it would be if one extra chicken was produced each time a person consumed chicken.

1

On average, if a consumer has no special reason to think that their action will not trigger any thresholds, a consumer should expect to have the average marginal impact of consumers.

The average marginal effect is the average amount of change that an increase of consumption of chicken by 1 will have on the production of chicken. The claim is the following: if you have no knowledge about where the thresholds are, you should expect, on average, for an increase in chicken consumption of 1 to have the average marginal impact of an extra chicken consumption.

Let’s imagine that you know that, on average, each of ten people who consume chicken will result in one extra chicken being produced. You don’t know if there’s some specific threshold—maybe it’s the case that at particular points you trigger a threshold, maybe it’s just a smooth continuum. The claim is that if you have no extra information beyond what the average impact will be from one extra chicken consumed, you should expect to have the average impact.

This is a pretty general logical principle—if you have no specific information beyond the average impact, and no reason to think either your contribution will be more or less than average, you should expect to have the average impact. If you know that the average person rides 10 trains in their life, and you are assigned a random person who you have no information about beyond that they exist, on average, you should expect them to ride 10 trains.

If you have no special information about an event, beyond the fact that they are a member of a set which has an average impact of N, you should expect them, on average, to have an impact of N. This principle is very obvious.

2

The average marginal impact of consumers consuming one chicken is increasing chicken production by more than one, multiplied by the value of the cumulative elasticity factor.

The claim here is the following: on average, people who consume an extra chicken increase the amount of chicken produced by a value that’s some number greater than one multiplied by the value of the cumulative elasticity factor. Let me start by explaining the cumulative elasticity factor, and then I’ll explain why the number that you multiply it by is greater than 1.

If you decrease consumption of chicken by 1, that can have two effects. The first one is that it will lower the price of something. The second is that it will lower production of that thing. To illustrate this, let’s imagine that I’m producing one cart an hour and selling them for fifteen dollars. If more people are buying them, I’ll expand my operations, and raise the price, because they’re more profitable. If no one is buying them, I’ll lower the price, in order to be able to sell them. Thus, when one fewer person consumes chicken, it will decrease price by some amount and also decrease production by some amount. Thus, you might worry that if you don’t have chicken someone else will—and thus you’ll have no effect.

The cumulative elasticity factor is a measure of how much chicken consumption will decrease if one marginal person consumes one less chicken. For example, if the cumulative elasticity factor is .5, then for every two people who decide to stop eating chicken, the number of chickens consumed will decrease by 1. It’s unlikely that the cumulative elasticity factor will be 1, for example, because they’ll just lower the price and still sell some chickens, though probably less than the ones sold. Well, it turns out that very sophisticated studies have been done and have concluded that the cumulative elasticity factor for meat tends to be around .75—so for every chicken you don’t eat, you are reducing the number of chickens sold by around .75. Thus, this means your impact is roughly 75% of what it would be if each chicken you didn’t consume reduced total chicken consumption by about 1.

So this explains the whole multiplying by the cumulative elasticity factor effect. But what of the claim that “The average marginal impact of consumers consuming one chicken is increasing chicken production by more than one.” Well, we know that it must be at least one, because if you add together all of the marginal effects of chicken consumption, it has to be at least as great, in total, as chicken production. If no one ate chicken there would obviously be no chickens produced for food. Thus, somewhere along the consumption of chickens, people’s collective marginal consumption triggers the total number of chickens that are produced.

Let’s illustrate this with one chart. Ignoring the cumulative elasticity factor—which must be taken into account, but which we already have taken into account—let’s look at a simple situation in which ten people consume chickens. We know that there must be at least ten chickens produced, and if no one was consuming chicken, there would be no chickens produced. Thus, together their marginal impacts have to add up to resulting in ten extra chickens being produced, because the total production, which is entirely a function of the amount consumed, is just equal to the added values of all of the marginal effects of consumption.

Let’s illustrate this with a chart. It doesn’t matter what the marginal effects are at the individual points—we know that, the total amount produced is a function of the added effects of all of the marginal effects of consumption. If 10 units are consumed, the marginal effect of each unit of production on consumption is 1—because as the consumption increased by 10, so too did production.

Given that each increase in chicken consumption of 1 requires at least one extra chicken produced, we know that chicken production must be at least chicken consumption, and so the average marginal impact of chicken consumers on consumption, and thus production, must be at least 1.

But it’s going to be greater than 1. Remember, the average marginal impact will be the total amount of production divided by the total amount of consumption. Thus, if production is greater than consumption—which it must be—the marginal impact on the marginal number of animals bred into existence and given horrific lives as a result of a 1 unit increase in consumption will be greater than 1.

Let’s start by just giving an intuition pump for this. Ancient kings used to have hundreds of meals prepared—and they’d pick one of them to eat. It’s obvious that in this case, their consumption causes more than just the number of animals that they eat to be killed, because their consumption requires, as a means to facilitate it, lots of other animals to be killed.

Remember, the average marginal effect of an increase in consumption by 1 will be production/consumption. If the number of animals produced is greater than the number of animals consumed, then the average marginal effect of an increase in consumption by 1 will be an increase in production by greater than 1. If production is twice as great as consumption, then the average marginal impact on production of an increase in consumption of 1 will be 2.

Now, we don’t know exactly what production divided by consumption is, but we know that it’s greater than one by quite a bit. I’ll just describe a few mechanisms.

1. Even after meat is produced, a decent portion of it is lost or wasted before it gets sold. All of that must be taken into account. As Torella says

Harish Sethu, a data scientist and author of the blog Counting Animals, says America’s meat waste problem means we’re raising about a billion chickens, more than 100 million other land animals (mostly turkeys, pigs, and cows), as well as capturing around 25 billion fish and 15 billion shellfish (mostly shrimp), only to have them wind up in a landfill.

This chart from Torella shows that a significant amount is wasted at the retail level.

2. Lots of animals die before they are sold. These animals are part of the animals produced on the supply side, and thus factor into the average impact of the marginal consumer. This effect is very significant. For farmed fish, for example, around 40% of them die before they are sent to slaughter, meaning that an increase in consumption of 6 fish will result in an increase in production of around 10. The number is more like 9% for animals in general. I don’t think this takes into account transport, because, once you do, the situation becomes even starker. As one source notes “The poor living conditions results in a 26% death rate among hens5 and a 15% among male chickens during the transport to the slaughterhouse.”

There are other ways in which production outstrips consumption that I can’t get into—like the grinding up of billions of baby male chicks—but the basic idea is clear: a decrease in consumption by 1 decreases production of animals by considerably more than 1. In fact, this effect is sufficiently dramatic to almost certainly outstrip the cumulative elasticity factor. Thus, if you consume one extra chicken, you are causing, on average, more than one extra chicken to be bred into existence and tortured. This effect generalizes to all animal products.

3

Therefore, on average, if a consumer has no special reason to think that their action will not trigger any thresholds, a consumer who consumes one extra chicken should expect to increase chicken production by more than one, multiplied by the value of the cumulative elasticity factor

This follows from the previous two premises. Remember, the first premise established that your expected effect is the average marginal effect of an increase in consumption of 1, and the second established that the average marginal effect of an increase in consumption of 1 is an increase in chicken production of more than one, multiplied by the value of the cumulative elasticity factor.

4

The cumulative elasticity factor of chicken consumption is around .75

I mentioned this briefly before—this claim is empirical and has been backed up by rigorous studies.

5

Therefore, on average, if a consumer has no special reason to think that their action will not trigger any thresholds, a consumer who consumes one extra chicken should expect to increase chicken production by more than .75 chickens.

This follows from the previous premises. In fact, was we discussed before, you should probably expect that, for every chicken you consume, on average, you’re causing more than one chicken to be bred into existence in horrible conditions.

Concluding thoughts

It doesn’t matter how complicated and interconnected the industry is: the basic idea is the same. As long as you have no special information about the thresholds and producers wouldn’t just produce animal products to sell to no one, for no reason, the objection cannot work.

Just to illustrate that this is the correct way of thinking about the industry, let’s imagine an industry where the causal inefficacy objection holds—and we’ll see that the various assumptions have to be false. Here’s one example of an industry. Let’s imagine there’s a machine that costs a lot but will create 200 billion chicken sandwiches. It’s only worth using it if there are at least 5 billion things of chicken ordered. In this case, you know that you have no significant effect, because you know that no matter what you do, the extra chickens will be produced, because you’ll be above the threshold. In this case, 1,2,3,4 and 5 would hold—so isn’t this a counterexample? No, it’s not. While all of those would hold, remember, 5 just talks about your expected impact if you don’t know where the thresholds are, and thus don’t have any special reason to think that you won’t affect thresholds. In this case, you do have special knowledge, so this wouldn’t apply.

Part 3: Even if you didn’t understand economics based knock-down argument against the causal inefficacy objection, you should still think it’s false, based on the empirical evidence

The previous section established that the causal inefficacy objection not only does not but cannot work. Thinking it does is bad economics. But even if you didn’t follow the last section and think that it must be more complicated than that, you should still think that the objection doesn’t work.

Halteman and McMullen produced the most comprehensive and systematic report on the way that actual markets work and conclude the following.

The picture that emerges from economic studies of these types of markets is one in which, despite the scale involved, individual consumer actions are reliably, if probabilistically, translated into changes in the quantity supplied to the market.

While it’s true that the individual consumer will probably have no effect, they will have a significant average effect. The authors conclude that each time you consume chicken there’s about a 1/900 chance of leading to 900 extra chickens being consumed.

Part 4: Responding to objections

There are lots of things that people say about why individual consumers won’t have any effect. There are two main claims about why consumers won’t have an effect. The first is to claim that markets are very complicated, the second is to claim that there are buffers.

Let’s start with the complexity claim. It’s certainly true that the market is very complicated. But even though it is complicated, as the arguments above show, there must be buffers of some sort, and these mean that the expected effects turn out a wash. There is no way to design the system so that collectively consumers have a significant effect but no one has any significant effect. The early analysis explains why.

The second main claim is that there are thresholds. Again, I could just respond to this by pointing out that this contradicts none of the premises and thus the conclusion would thus necessarily be true. But there’s a separate problem: even if there are thresholds, you still have an expected effect.

The basic idea of thresholds is that the companies want to produce more meat than they think will be consumed, because the cost of running out is greater than the costs of overproducing and wasting some animals. So, if they think that there will be 1 billion chickens consumed, they’ll produce maybe a billion extra chickens.

Now, as I explained before, this actually increases one’s marginal effect—it means that the average marginal effect is greater, because they collectively comprise overall production. But even setting that aside, if they want to remain a billion chickens above the total number of chickens consumed, then they’ll increase chicken production when there are more chicken producers—the threshold rises as the thing that the threshold is supposed to be greater than rises.

Part 5: Explaining Michael Huemer’s Argument

Michael Huemer has a debate with Danny Shahar about the causal inefficacy objection to eating meat. I think that Huemer wins the debate quite decisively. Here, I’ll reproduce Huemer’s argument, explaining why one does have a significant effect, in expectation.

Huemer starts out by noting that one can’t believe the following three things.

1. A major change to X will have a huge effect on Y.

2. Small changes of X will never have a huge effect on Y.

3. Big changes are comprised of lots of small changes.

In this case, X can be number of individual meat purchases, and Y can be meat production. Huemer’s argument is the following.

1. If 1 million people gave up meat, meat production will decrease by roughly 1 million.

2. The average impact of one person giving up meat is a reduction in meat production by 1 million portions/1 million people=1 portion per person.

3. You’re not special—you’re not more or less likely to trigger a change in meat production than any other person.

4. Therefore, (from 2 and 3), your expected impact from giving up meat is a reduction of 1 meat portion produced.

Remember, all of the million productions added together collectively result in all meat consumption, so the marginal meat consumption of a million people is greater than increasing production by 1 million. Thus, all of the premises are on quite firm foundation—largely for the reasons I described.

Part 6: Even if this is probably wrong, you should still go vegan

Remember, eating meat plausibly causes thousands of times more suffering than the benefits that it produces for you, if the causal inefficacy objection is unsuccessful. Thus, even if you think that the causal inefficacy objection is probably wrong, even if there’s a 1% chance that it’s right, you still should abstain from eating meat. As Huemer shows in the debate, eating meat causes, assuming the causal inefficacy objection is wrong, about 245 years worth of suffering. Suppose you think that animal pain is 10% as bad as human pain and there’s a 10% chance that eating meat is wrong. Well, then eating meat over the course of your life is still equivalent in wrongness to torturing someone for two years. If it’s 1% as bad, then it’s still equivalent to torturing someone for .245 years, so only torturing someone for several months. Thus, even by conservative estimates, assuming the causal inefficacy is probably right, eating meat is still the worst thing you’ll ever do—unless you torture people for more than two years.

Part 7: Even if the causal inefficacy objection is wrong, it’s still probably wrong to eat meat

Let’s imagine that a bunch of soldiers are going to go into a town and burn everyone’s home down, causing everyone to burn to death. Let’s imagine that they’ll do it if more than 50 people agree to set some people on fire. There have already been 60 people that have agreed to do it, so everyone in the village will be burned to death either way. It still seems wrong to agree to do it.

Here’s a plausible ethical principle: it’s wrong to engage in a collective practice that is severely wrong, where it would be better if everyone abstained from it, even if you engaging in the practice will not be wrong. There are some rule utilitarian justifications for it, as well as act utilitarian justifications based on heuristics, as well as clear deontological judgments. The intuition is supported by lots of cases.

1. The case I gave above about burning down a village.

2. Your friends have agreed that they’ll set a homeless man on fire. You think it would be fun to do it, and you know that even if you don’t do it, your friends still would. It still seems wrong to do it. You cannot get help or prevent it.

3. Your spouse will cut your child’s arm off with a steak knife unless you do it. You cannot get help or prevent it. It still seems wrong to cut off your child’s arm.

There are enormous numbers of cases like this—cases which support this fundamental intuition.

Part 8: Even if the causal inefficacy objection has a 100% chance of being wrong and the previous objection does not succeed, you still shouldn’t eat meat

Credit to Huemer for this argument.

Even if it’s fine to engage in bad collective actions and the causal inefficacy objection fails, eating meat is still wrong. The following principle is plausible and supported by lots of specific cases.

It’s wrong to support an evil industry even if supporting them has no marginal effect.

This is supported by lots of specific intuitions.

1. ISIS is holding a bake sale to fund their WMD program. However, you know that you buying their cookies will give you enjoyment and will also not help ISIS, because they already have enough money. Still, it seems wrong.

2. You can purchase a baby from the worst industry in history that slaughters 78 billion babies. However, it would be tasty and you think—for causal inefficacy reasons—that you’ll have no effect on baby production. Should you?

3. You can pay to eat a slice of a trafficked child prostitute who was killed by the cartel. She would taste good. Should you?

The answer is, intuitively, of course not. But this would also apply to eating meat, at least, if you buy the basic case against eating meat.

Now, I’m a consequentialist, so I think that it’s plausible that if you really have no effect the act wouldn’t be intrinsically bad. But it is plausible that, even for a consequentialist, one should have a disposition that would prevent them from doing it. Additionally, we should all have moral uncertainty—I only think there’s like a 70% chance that consequentialism is correct; thus, based on the offchance of it being seriously immoral, there’s still a good reason not to do it

Conclusion

The causal inefficacy objection fails for lots of different reasons. Fundamentally, supply and demand works—when more people want meat, more is produced. There are powerful systematic arguments that show beyond a shadow of a doubt that the argument is empirically wrong. There are many such arguments. However, even if there were not, the moral claims behind the objection are dubious. Those who raise the objection are (inadvertently) peddling bad philosophy and even worse economics.