Economists on Ethics and Animals: Blackorby and Donaldson (1992)

Charles Blackorby and David Donaldson, "Pigs and Guinea Pigs: A Note on the Ethics of Animal Exploitation." The Economic Journal 102: 1345-1369, November 1992. 

The leading general economics journals are not exactly replete with articles centered on the treatment of nonhuman animals, but they are not totally bereft of such articles, either, thanks mainly, I think, to Blackorby and Donaldson (1992). Here are some notes about this outlier of an economics article...

Human use of animals for food and research imposes harms on animals -- but by increasing the demand for animals, these exploitative uses result in many more animals being bred than the numbers that otherwise would be brought into existence: if humans didn't eat cattle, the world would have a lot fewer cattle. (Right now, there are nearly one billion cattle in the world.) So perhaps, if the average animal has a life that is worth living, despite the human-imposed ravages, one could argue that human exploitation of animals is good, on net, for the animals. (This "logic of the larder" has come up before in Animals and Econ.) Blackorby and Donaldson offer a structured examination of the possible trade-off, where animal exploitation simultaneously imposes direct harms but indirectly increases animal populations and possibly improves some measures of overall welfare.

Comparing overall wellbeing across situations with different quantities of "beings" is not a straightforward exercise. Blackorby and Donaldson invoke "critical-level utilitarianism", which is one approach for making such comparisons. In doing so, they also employ Singeresque equal consideration, where the weight given to the wellbeing of a creature is independent of what species (human or nonhuman) the creature happens to be. So, two states of the world with equal numbers of creatures would be compared on a classical utilitarian basis, by the sum of the wellbeing of each creature.  

With different numbers of creatures or different lifespans, the evaluation of social states still involves summing lifetime utilities (so it is a form of aggregate wellbeing, not per capita wellbeing, that ranks states), but with some twists. A hedonically neutral life is normalized to a lifetime utility of zero. What is summed is the excess of individuals' lifetime utilities above some specified critical value. Consider two situations (A and B) that are identical, except that state B has one more individual than does state A -- all the individuals "shared" between the two states have the exact same lifetime utility whether they are in state A or state B. Is state B preferred to state A, because B has everything A has plus one more creature? The answer depends on whether that individual's lifetime utility (in state B, the only state that has that individual) exceeds the critical value: if the creature's utility exceeds the critical value, B is better than A. If the creature's utility falls short of the critical value, A is preferred to B. "Higher" critical values, then, make it more likely that additional creatures per se do not make social states more appealing. Indeed, if the average utility is below the critical value, adding more folks with such an average utility would make the situation worse. (Classic utilitarianism can always make situations better by adding more individuals with above-neutral hedonic evaluations.) If the critical utility value is 3, "a population of [size] one with a utility equal to 4 is preferred to a population of [size] two with utilities of 4 and 2 [p. 1348]." Nonetheless, killing the person with utility of 2 does not improve the social state: that person's lifetime utility is lowered by the death, and the utility of any creature alive for any part of a "state" counts towards that state's evaluation. (But the birth of someone with a lifetime utility of 2 should be avoided, as that would lower the evaluation of the second state: killings of existing creatures and births of new ones do not have identical effects on measuring overall wellbeing!) A lifetime utility can fall below the critical value if the per-period hedonic payoff is low, or if the lifetime is short (as it is with farm animals). In principle, this set-up can provide a basis for making claims about optimal population sizes as a function of the critical wellbeing level. The authors give an example where it is better to mistreat a single mouse three times for research purposes than it is to mistreat three mice, one time each. 

Further complex developments (dozens of equations, eight graphs) fill out much of the article, but from my perspective, they are of marginal interest. Section V (pages 1363-1365) and the Conclusion (1365-1367) are more pertinent; the former compares the critical-level utilitarianism framework with other ethical systems. The authors note the centrality of the critical utility level (possibly different for different species) to their framework, and suggest that for "simpler animals" who lead "less coherent lives," a sub-human critical utility level might be appropriate. 

The Conclusion features the author's answers to some fundamental questions at the human/animal nexus. "There is a case for raising animals for food, but it is possible that the ethical vegetarians are right that the optimal level of meat consumption is zero [p. 1366]." The case for exploiting animals for useful medical research is stronger, but replacement and reduction and refinement are called for here, too, up to a point. Also stronger is the case for reducing our consumption of meat, and for providing better conditions for animals than profit-maximization alone would offer. A tax on exploitative uses of animals is probably helpful in moving us closer to optimal population numbers.

PETA on a Meat Tax

People for the Ethical Treatment of Animals support a meat excise tax "to help cover the health and environmental costs that result from using animals for food." They detail the usual health and environmental impacts of meat consumption and production, and (as is usually the case, it seems) do not distinguish between internal and external health costs. PETA does, however, mention the potential cross-subsidy inherent through our health care financing: "It doesn’t make sense that the millions of meat-free Americans have to help pick up the tab (through taxes and health-insurance premiums) when meat-eaters get sick." On their "Tax Meat" FAQ page, PETA also notes the health impacts of the production of meat, the significant health and safety risks within the meatpacking industry -- an industry with working conditions that have been found further wanting during the Covid pandemic. The revenues also can be earmarked: "Revenue from the tax could be used to clean up areas polluted by animal agriculture, assist farms in transitioning away from animal-based agriculture, and increase access to healthy plant-derived foods in communities that need it most." I didn't see a specific tax rate in the PETA material, but they do indicate (on the FAQ page) that the tax payments from a typical (non-vegetarian!) family would come to about $5 per month. Given standard household meat expenditures in the US, that suggests something like a 5% meat (ad valorem) excise tax. 

PETA end their appeal for a meat tax by identifying the victims of the current system, the millions and millions of abused animals: "When it comes to Americans’ meat habit, animals are paying the biggest costs. A tax on meat could help persuade humans to save other animals (and their own bodies) from abuse."

Katare et al. (2020) on Optimal Meat Consumption

 Bhagyashree Katare, H. Holly Wang, Jonathan Lawing, Na Hao, Timothy Park, and Michael Wetzstein, "Toward Optimal Meat Consumption." American Journal of Agricultural Economics 102(2): 662-680, March 2020.

Meat consumption worldwide increased by more that a factor of five between 1992 and 2016, with concomitant negative effects on the environment and health. Many of these harms are of the external variety, and thus present the usual case for potentially welfare-improving policy interventions -- including a Pigovian meat tax. But the existence of external costs does not in itself establish the optimality of a Pigovian tax; perhaps improved consumer information or a campaign aimed at voluntary reductions in meat consumption would be preferable policy responses. Alternatively, changes in production methods could reduce the environmental impact of the meat industry.

The authors compare a meat tax with a mandatory labelling scheme, where the labels on meat products would indicate the environmental and health costs of meat production and consumption. The authors note that individuals enjoy behaving in what they see as a prosocial manner, providing an intrinsic motivation to reduce meat consumption. This intrinsic motivation, they argue, can be counteracted by the presence of external motivations to reduce meat consumption, via a meat tax, for instance -- this type of crowding out has been examined pretty broadly within behavioral economics. Green labelling (as opposed to the tax) could promote, not undermine, intrinsic incentives to reduce meat consumption. (Although the opposite effect, that being constantly lectured to cut back on meat could incentivize more meat consumption, seems possible, too.) Other issues to consider are the potential that the substitutes to meat consumption might involve external social costs of their own, and the revenues collected by a meat tax could be spent in ways to further reduce meat consumption.

In their baseline scenario, the authors find that green labelling has little impact on reducing meat consumption; they also find that the optimal meat tax is quite significant, at more than 67% of the pre-tax price. Of course, different parameterizations could alter the specific findings, but a significant meat tax combined with a supplementary informational campaign seems to me like a big improvement over the status quo.

Forero-Cantor, Ribal, and Sanjuán (2020) on Carbon Footprint Meat Taxes

Germán Forero-Cantor, Javier Ribal, and Neus Sanjuán, "Levying carbon footprint taxes on animal-sourced foods. A case study in Spain." Journal of Cleaner Production 243, 10 January 2020, 118668.

Meat production as practiced today is not friendly to the environment, with animal agriculture responsible for some 14.5% of greenhouse gas emissions. Other environmental harms associated with animal agriculture include water pollution and rain forest destruction. 

With respect to greenhouse gas emissions alone, Forero-Cantor et al. (2020) calculate (Pigovian?) taxes for pork, beef, chicken, turkey, lamb, eggs, and fish, using quarterly Spanish data from 2004 to 2015. (During this period, the total Spanish consumption of beef, eggs, and lamb decreased. The modal meat meal is fish -- that is, the weight of seafood consumed each year globally exceeds that of pork or poultry, the most common land-animal-sourced meats; fish is undifferentiated in the data, as opposed to the separate types of land-animal-sourced meats.) The researchers need to know not only the carbon footprint throughout the meat supply chain -- lamb and beef are the biggest carbon releasers on a per kilogram basis, while fish have the lowest carbon footprint -- they also have to understand demand elasticities, how changes in prices will alter the quantities of meat and other products that consumers purchase. 

The researchers simulate the effect of various taxes on greenhouse gas emissions, and find, counterintuitively, that a tax on fish does best at lowering the overall carbon footprint. A tax on pork, alternatively, given the induced product substitutions, would have a perverse effect, it would raise the overall carbon footprint. A tax on poultry (chicken or turkey) wouldn't help much in curtailing greenhouse gasses, either. Given the huge (1 trillion? -- pdf) number of fish killed each year for human and animal consumption, animal welfare considerations would seem to bolster the (relative?) case for taxing seafood.

The New Beyond Meat Ad...

 ...is here. [The original link stopped working, so this is a link to what might be the same(?) ad from summer 2020; still worth seeing!]

Springmann et al. (2018) on Health-Motivated Meat Taxes

 Springmann M, Mason-D'Croz D, Robinson S, et al. Health-motivated taxes on red and processed meat: A modelling study on optimal tax levels and associated health impacts. PLoS One. 2018;13(11):e0204139. Published 2018 Nov 6. doi:10.1371/journal.pone.0204139

The authors look at the health impacts of an increase of one serving per day in the consumption of red and processed meat. (Red meat here is beef, pork, and lamb, despite what the ads have told us; processed meats include sausage, hot dogs, and many deli-style lunchmeats.) They monetize these health impacts, and then examine taxes that would internalize the incremental health costs; here's a schematic representation of their analytic approach.

Red and processed meats come out of the study as fairly lethal, accounting for something like 4.4% of deaths. In high-income countries, "internalizing" the mortality and morbidity effects would raise prices of red meat by about one-fifth, and more than double the prices of processed meat. These price changes would result in lots of substitutions in terms of food consumption, and meat-related deaths worldwide would decline by some 9%. The remaining meat-related health care costs would exceed meat tax revenues.

While I learned a lot from the useful exercise presented in Springmann et al. (2018), I don't think that it sheds much light on "optimal" meat taxes, at least when that term is used in the Pigovian sense (which the authors reference). Their approach treats all incremental meat-related health care costs as external, that is, as if they are ignored entirely by the consumer. But people engage in lots of risky activities even when they know the risks and even know that they will bear the costs if the risks materialize. Is the additional utility that people get from that incremental serving of meat more valuable than the additonal (expected) health care costs? We can't know the answer to this question from the approach taken in the article. We don't know benefits, we don't know what portion of costs are truly external, and we don't know the extent of underweighting of non-externalized costs, so-called internalities. Hence, this approach does not possess the building blocks required to provide reliable estimates of Pigovian meat taxes -- a point similar to one I raised concerning the analysis in Meatonomics. I suspect the underlying issue (with Springmann et al. (2018)) is a reliance on a cost-of-illness approach, which is inadequate for capturing economic optimality. 

On Pigovian Taxes

In the previous post, I threw around the phrase "Pigovian tax" much too cavalierly. Looking at the external costs that an industry imposes on society, and then choosing a per-unit tax (so many cents per pound of meat, say, or a percentage tax imposed at the retail level) so that the tax revenues equal those external costs (and hence the consumers are now paying the full freight, as it were) is not equivalent to a Pigovian tax -- indeed, I expect that, in most circumstances, that procedure bears no relationship to a Pigovian tax.

To find a Pigovian tax, you should first identify all of the social costs and benefits associated with all potential levels of output of the good in question. The benefits typically are measured by consumer willingness-to-pay, and the costs are standard production costs (wages, prices of other inputs, rent on capital, and so on) as well as external non-monetized costs such as pollution or what have you (and any other adjustments where market prices for inputs do not reflect opportunity costs). Then you calculate the output level where the net benefit (all benefits minus all costs) are maximized, which typically is where marginal benefits equal marginal social costs. That output level is "socially optimal," producing more economic pie than any other output level. Having found the socially optimal output, you look at marginal social costs at that level of output and also the marginal costs actually facing private producers at that same output level. If there is a net negative externality, private marginal costs will be less than social marginal costs. The per-unit Pigovian tax, then, is given by the difference between these two marginal costs, measured at the socially efficient output. Such a tax (or the ad valorem version) does indeed bring home the full costs to the consumer, makes them pay the full freight of the goods they consume, but generally it has nothing to do with having tax revenues equal some level of external costs that were measured from the pre-tax status quo.

OK, even this more defensible use of the phrase "Pigovian tax" paints with pretty broad strokes, but I wanted at least to make some movement in the direction of accuracy.

As for the external costs associated with meat eating in the US, I believe that they are significant, even if we look only at the environmental subset of those costs. As soon as we treat harms to the animals themselves as cognizable costs (or even if we only count one-thousandth of those harms as social costs), then the size of the Pigovian tax is enormous -- indeed, it would be prohibitive, in that industrial farm operations as now undertaken in the US could not survive if animal harms were reflected in retail meat prices.

On Meatonomics, by David Robinson Simon

I am trying to put some thoughts together on meat taxation, so I thought a good place to start would be to comment a little bit on the 2013 book Meatonomics, by David Robinson Simon. The cover of Meatonomics includes the phrase "$414 Billion Reasons to Eat Less Meat," and the substance of the book explains where these extra (annual, 2012 dollars) costs of meat (not paid for directly by consumers or producers) come from; they are helpfully summarized in two pages (pp. 202-203) in Appendix B.

Kudos to Mr. Smith for taking on this ambitious task, and for documenting and attempting to quantify the many external costs of animal agriculture. The details undoubtedly are debatable, but what I take to be the main message -- there are huge external costs associated with meat, egg, fish, and dairy consumption in the US and that a significant tax on products containing these ingredients should be a major part of a set of policy reforms -- is one with which I agree. (And if you don't agree, your position might be altered by reading Meatonomics.)

Once the existence of sizable negative externalities is established, the economic case for corrective measures, including a so-called Pigovian tax, is fairly strong. Smith's $414 billion is more than 1.6 times consumer spending on animal foods, so a Pigovian tax would increase retail prices by something on the order of 2.5 times their existing (2012) level. [OK, this claim is actually a massive simplification or even distortion.] The higher retail prices would decrease the quantity demanded of animal foods, of course, but economic efficiency would be enhanced, as consumers began to shoulder the full social costs of their dining decisions. Mr. Smith does not propose a full Pigovian tax, however; instead, he suggests (page 172) a 50% ad valorem tax on domestic retail sales of animal-based foods, which would raise consumer prices for those products by something like 50%, too, presumably.

The precise nature of Mr. Smith's accounting of external costs is relevant for desirable policy reforms. First, some 9 percent of the external costs come in the form of subsidies to agriculture, including irrigation subsidies. (Since much of the corn crop is used for animal feed, subsidies to corn, for instance, end up supporting animal agriculture.) The externalities associated with these subsidies would best be handled by eliminating the subsidies (which are spread around to all the uses of corn, including ethanol), not through a meat tax. Second, the majority of the external costs come through impacts on health (especially increased heart disease), but it is not clear that these are externalities. (Some, such as building up of antibiotic resistance through farm antibiotic use, are classic externalities, however.) They are not direct physical externalities, like when your steel mill pollutes my air (or vice versa), but rather, they operate through the socialization of some healthcare costs -- the spillovers are financial, and would not exist if we chose not to provide some public subsidy to healthcare. Using such financial spillovers as justifications for further public interventions (here, a meat tax), is not necessarily a good idea. After all, once it is agreed that such externalities can drive policy, then any personal behavior (not getting enough exercise, say, or eating an extra dessert) seems ripe for public control: your irresponsibility is costing me money! This is not a road down which I am eager to travel, at least without a lot of consideration. (Though I do support some socialization of health care expenses -- I just don't want their availability to be made contingent on whether your behavior is judged to be sufficiently protective of the public purse.) Of course, even if there were no issues with fiscal spillovers, we should still be concerned with the health effects of consuming animal products or anything else. Do individuals understand the health risks they are running, or are these risks for some reason undercounted in individual decision making? That is, health risks can be internalities, even if they are not externalities, and taxes might be helpful to overcome internalities, too.

The environmental externalities identified in Meatonomics are, for the most part, good old-fashioned standard externalities, suggesting that the current size of the animal agriculture sector is socially excessive and a meat tax would move us in the right direction. Mr. Smith also notes the willingness-to-pay by people for less cruel methods of farming, so not only is there too much animal agriculture, it is insufficiently protective of animal interests -- even if those interests are directly ignored, and only taken into account through human willingness-to-pay.