Quadrillion-dollar bills on the sidewalk? Quantifying animal welfare changes

How important is animal welfare as an issue? And how should it be traded against other concerns (if we must)?

Work over the last few years has begun to highlight the case for quantifying animal welfare impacts in dollar terms – for the purpose of comparing different causes and costs and benefits – and arguably made progress with the necessary foundations to do that, including assessing how we should weight the experience of different animals. This post summarises some of that work, particularly 'Monetizing Animal Welfare Impacts for Benefit–Cost Analysis' by @Bob Fischer and others, and tries to impatiently rush to one end of this process by putting some $ numbers on animal suffering changes. Hopefully this post helps provide a useful walk-through of some of the concepts, makes the case for further work, and demonstrates why some people think farm animal welfare is not just a 'nice-to-have' consideration but one of the most important topics in the world. (I’d like to do more on animal welfare economics, and this is my first foray, so feedback would be very welcome! Apologies to the real experts for any butchering, so to speak.)

Quantifying health changes

Our goal here is to put a $ or £ value on changes in animal welfare – and I'll be focusing here on farmed chickens given how numerous our dinosaur cousins are. The idea is that quantifying animal welfare, sharing a common unit with many other areas of life and public policy, can help people gauge the relative importance of the issue, and to include these changes in cost-benefit analyses, alongside many other considerations which are already included, such as impacts on incomes, health and global warming. (The inclusion of estimated climate impacts is a good example of how it may be worthwhile to put a value on some things that were previously excluded from such analysis, but also creates significant risks where emissions and animal welfare considerations are in conflict but only the former is included.)

To begin, it’s worth giving a human example. UK Treasury guidance says that the societal value of a ‘Quality-Adjusted Life Year’ is £70,000 in 2020-21 prices, so around £89,000 in 2025-26. Extending someone’s life in full health by a year may then, on paper, be valued at £89,000. Going from half-health to full-health for two years is also worth £89,000 ('0.5' health-improvement * 2 years = 1 QALY). Doing this for 1,000 people would be worth £89,000,000. (This kind of valuation is standard practice, even if it sounds funky, and can be based on the public's actual views about potential income versus health trade-offs.)

'Monetizing animal welfare...' makes the case that we can and should do similar calculations for other species, and to do this we may 'only' need 4 numbers that can then be multiplied to quantify the scale of an impact.

1. ('Easy') The value of a human QALY. Standards for valuing human QALYs will vary over time and by country and are of course debateable, but we don't need to reinvent them here. As a Brit, I'll use £89,000. The most common value used in US research now appears to be $100,000 per QALY (around £80,000) while for US impact assessments the standard might be $591,000. (I think £89,000 makes sense as an order of magnitude: for me specifically, an extra year of healthy life would definitely be worth more than £9,000 of my money, but £900,000 would be unaffordable.)
2. (Easy) The number of animals involved.
3. (Hard!) A factor for comparing each species’s 'welfare potential' to that of humans. This could optionally include some arbitrary speciesism if you like, but for now let’s try to take a utilitarian, bird's god's eye view (e.g. a veil of ignorance approach in which you don't know whether you are going to be incarnated as a human or a chicken) and assess only "how good and bad the life of an individual of a given species can be relative to a human". Bob Fischer and others have put a lot of work into this. Using a range of different approaches and uncertainties, this suggests that perhaps a chicken’s welfare potential is only 3% that of a human; or maybe it’s 95%; or it could well be 100%. While this is a big, uncertain and philosophically debateable range, for the purpose of illustration I will use their average value across a few methods of 40%. (This is not to say that we should be prepared to kill 1 human to save 3 chickens: only that both humans and chickens are very likely sentient and that experiencing severe pain as a chicken is not too disimilar from experiencing severe pain as a human, i.e. it sucks.)
4. (Hard!) How bad is a particular experience compared to full health, and how much can public or private policy change that? Can policy take animals from 'half health' to full health or well-being (0.5 QALYs/year), or is the potential change bigger or smaller than that? While this is hard to gauge, such assessments are made for human injuries and the examples below hopefully give a sense that making plausible estimates is possible.

In the discussion section I look at some of the extra-tricky questions that I've skipped over here, but for now let's look at what numbers this method gives us if we try to value farm animal welfare improvements or harms. And bear in mind that if you think any of the values I've chosen as inputs are too high or too low, you can of course scale down/up the results accordingly.

An impact assessment example

In the UK, official policy impact assessments are common but – even when the policy is about animal welfare – the costs and benefits for other animals are not quantified. As an example, the relevant government department has done an impact assessment of the costs and benefits of adding mandatory animal welfare labels to some chicken, egg and pig products. (This is a very good policy proposal and I hope the new UK government is not going to kill it off, having vaguely stated that they "will introduce the biggest boost for animal welfare in a generation".)

In this example: the assessment suggests that reforming labels will come with one-off transitional costs for businesses of around £12 million (in 2025-26 terms); annual compliance costs of around £4 million; a negative £20 million per year impact from increased greenhouse gas emissions associated with higher chicken welfare; and a positive £53 million per year gain for British farmers (at the expense of lower-welfare imports). Don't worry about whether these numbers are good guesses or not, but the example shows the kinds of things that are valued, and some typical numbers. Overall, these modelled benefits already exceed the modelled costs of the policy. But, as noted, the potential impact on animal welfare (the whole point of the policy!) is not quantified in this official assessment.

But armed with the 4-number approach above we can have a go! The impact assessment estimates that 1 in 10 UK broilers (plus a smaller fraction of laying hens) might move into higher welfare categories as a result of greater transparency for consumers, which equates to around 12 million at each point in time. With a value for human QALYs and a chosen relative welfare potential figure for chickens, we therefore only need to know by how much those chickens' lives will be improved.

To be brief, I think the life of a typical broiler (meat) chicken is extremely bad. Instead of living in the jungle, they are in barns of maybe 50,000 other chickens, with few features. Lights may be on for 20 hours a day. They never see their parents or any other older bird. If they are not given enough food (as will be the case for breeders and before slaughter), they will be hungry. If they have access to food they will grow enormous. Stress on their hearts and lungs causes a range of health problems. The skeleton struggles to keep up with this fast growth, causing bone deformities, fractures and infections: 90% of birds have definite gait defects. Birds will increasingly spend their time just sitting, contributing to ulcers and lesions. As their litter is never changed over their 40-day lives, the ground becomes increasingly dominated by excrement, leading to a build-up of ammonia that burns the skin. The birds may be pecked by others, while the presence of dead birds will be common. Then one day they are stuffed into crates and transported to the abbatoir where they will be grabbed by their legs and hung upside down from their possibly-broken feet, unable to breath properly. If they are lucky, a blade will quickly slit their throat. If not they may be boiled alive. (Or if mass culling is required, they may be fatally overheated or suffocated in foam.)

Even without radical change, this quality of life can be improved! The Welfare Footprint Project helps give some numbers here. Remarkably, while a chicken could naturally live for 7+ years, a typical broiler's entire life (including sleep) is now under 1,000 hours. With that rough point of comparison, the 'Better Chicken Commitment' can deliver an estimated reduction of at least 33 hours of Disabling pain, 79 hours of Hurtful pain and 25 seconds of Excruciating pain: "a reduction of approximately 66%, 24% and 78% , respectively".

As a rough but conservative guess, then, let's say that labelling reform could push up life quality for affected chickens by 0.1 QALY/year, where 1 would be the difference between zero and full-health (though see later for a discussion of negative QALYs). In other work that tries to convert the 'Five-Freedoms' animal welfare framework into QALYs, I think such a 0.1 change is – under some assumptions – equivalent to going from "very severe" to merely "severe" violations in one of the five freedoms (e.g. "freedom from pain, injury, and disease").   

The annual chicken health welfare gain from such a labelling change then is worth £89,000 * 0.4 (species factor) * 12 million chickens * 0.1 (average QALY change), and this equals... £44 billion a year ($54 billion). This is about 1000x bigger than the economic and environmental considerations considered in the existing impact assessment, and indeed is about 1.5% of UK GDP. And this is the impact just from a small change in UK food packaging. So 'how much good' could more radical change achieve, or 'how much harm' does the status quo entail? 

What is the maximum potential welfare impact of changes to chicken farming?

Going further, what if we could raise standards by 0.1 QALY/year not just for 10% of UK broiler chickens through better labelling, but for (say) 95% of all chickens?

In the UK, that would be worth £560 billion a year – around a fifth of UK GDP. 

In the US, where the chicken population is nearing 2 billion, the potential gain is around $7 trillion – over a fifth of US GDP. 

With 28 billion farmed chickens alive globally at any one point, the global equivalent would be £95 trillion or $118 trillion a year – similar to global GDP (although here you might argue that British or American QALY values are not affordable on a global scale).

To reiterate, that is an estimate based on raising most chickens' well-being by 0.1 QALY/year on an (arguably) 0-1 scale; assuming chicken welfare changes are only worth 0.4x  the equivalent changes for humans; and ignoring all other species. If, for example, you thought you could make changes worth 1 QALY/year – as I explore below – and removed the 0.4x factor, you'd be up to $3 quadrillion a year from improved chicken welfare. On the other hand, if you were far less ambitious, or gave chickens less moral weight, maybe you could get down to a mere $1 trillion a year of beneficial impact. But a separate estimate from 2021 (which I discovered very late when writing this), puts the animal welfare cost of a non-vegetarian American diet at $123,000 per year, which globally would add up to $100s of trillions: similar to my numbers.

To be clear, reducing or ending chicken suffering would not necessarily make us any richer: it is not going to be a significant boost to GDP and may even make humans worse off. But expressing the potential gains in $ terms can be one lens through which to assess the importance of the topic. And these rough results show (for those who need more persuading) that chicken welfare is extremely important! Few policy decisions are going to matter as much: to pick a couple of big ones for comparison, trade policy decisions might make countries or humanity a few % of GDP better off (or worse off), while – at the extreme – open borders might in economic theory add around 100% to global GDP ("Trillion-Dollar Bills on the Sidewalk?").

We can also compare global chicken welfare changes (e.g. the $118 trillion for marginal improvements) to the global value of the poultry industry – around $400 billion – to quantify why we should prioritise welfare concerns over the industry's concerns. This is a similar conclusion (and through similar methods) to work calculating what 'animal welfare levies' might be appropriate: which suggests that a levy of up to €3,000/kg on chicken might be justified, which is again saying that factory chicken farming does vastly more harm than good. And this also matches the non-quantitative view that this suffering is simply very bad and worth making radical change to prevent.

Discussion / a few possible points of disagreement

There are a lot of big questions that can be asked about these calculations and some of the assumptions I've skirted around, and it is worth briefly pointing to a few of them.

Welfare potential values for different species 

There is a whole free book on comparing well-being across species. A basic consideration is whether (in this case) chickens are sentient at all, but this project's median probability was an 89% chance. And in the specific UK context of potentially including animal welfare in impact assessments, this question has perhaps already been settled as the Animal Sentience Act states that all vertebrates are sentient beings. Conditional on sentience, that project suggests the relevant value for chickens is in the range of 4% to 100%, depending on what matters for welfare comparisons. Again, this is a big range and there is lots more scientific and philosophical work to be done here, but you could choose 4% rather than 40% and still come to the same basic conclusion that animal welfare concerns dominate.

Thought experiments may also be able to help here. For example, if you came across a body-swapping machine, would you rather be a person with a newly broken leg for 1 minute or a chicken with a newly broken leg for 10 minutes, or how about 100 minutes? Would it would be worse to be suffocated once as a human or ten times as a chicken? Again, it seems extreme to weight moment-to-moment chicken welfare at less than 1% of a human's.  

Questioning the use of human-based QALY values 

While I've included a cross-species adjustment (40% for chickens), an initial reaction might be "with an £89,000 human QALY and saying chickens have 40% of the welfare potential of a human, you're implying that regulators should value a year of full-health chicken life at £36,000: that seems... high... The value of a QALY has been determined based on human valuations, but no-one one has asked people how much they value a chicken life. What gives?" Here are 6 jumbled points in response: 

1. Yes, this approach is deliberately non-speciesist and utilitarian. It's not meant to be how much humans are willing to pay for animal welfare. Rather, as noted above, it is what a perfectly altruistic utilitarian might want to pay, or (equivalently) someone behind the veil of ignorance who didn't know what species they were going to be a part of.
2. This paper discusses one way in which using human QALY numbers is conservative. These may be designed to reflect the value of perfect health relative to death, but not to reflect total well-being: only the health component of that. If we are talking about changing farmed animals' well-being in a broader sense than health, the values we should use should perhaps be even higher. On the other hand, I wonder: even if willingness-to-pay for a QALY is driven by selfish individual rather than society-wide valuations, perhaps people value their health for a broader set of reasons such as ability to work, ability to socialise and do activities, and wanting not to be a burden on others. And at least some QALY valuations seem to be based on the value of a year of life overall, which is about more than avoiding ill-health.
3. We know that many pet owners are willing to spend thousands of dollars on surgery or health insurance for their animals. And while poll responses should be taken with a grain of salt, 43% of Brits say that animal lives are worth the same as (or more than) a human life!
4. QALYs may be driven by real-world affordability. Even if this line of calculation suggested a spider's health was worth £4,000 a year, would we commit to spending that to help each wild spider?: no, even rich countries could not afford that, suggesting that (while we may be far richer and more capable in the future) here in 2025,  animal QALY values should be lower than I've suggested or perhaps even that a firm utilitarian should put a lower value on human QALYs, given the need to consider non-human QALYs too.
5. The UK values a QALY at £89,000, yet its National Health Service can provide a marginal QALY for £15,000 (in 2016 prices). So, shouldn't the UK spend far more on healthcare? Arguably yes! But my point is that there may be a distinction (in practice) between valuing something at £X/outcome on paper and actually having a policy of being willing to spend £X/outcome. So, perhaps we wouldn't actively spend £36,000 for a chicken QALY, but perhaps we should still be £36,000-worth of happy when an extra chicken QALY is delivered.
6. Above all (and trying to avoid discounting chickens twice – through both the QALY value and the species adjustment factor), note that even with much lower QALY values the conclusions would hold: that farm animal welfare will – if it is included in considerations – tend to absolutely dominate the costs of making changes to food and farming practices.

Population ethics and states worse than death

As 'Monetizing Animal Welfare Impacts for Benefit–Cost Analysis' discusses, it's one thing to use QALYs to value states of health, but it's trickier to know how to proceed if you need to compare not just better or worse lives but also non-existence. i.e. Ignoring effects on humans, it would superficially be better for the 28 billion chickens if we made their lives as good as possible, but would it be better if those standards were so expensive that in reality their population were reduced to 1 billion? This is the classic 'logic of the larder' argument for eating meat. I think there are very strong counterarguments – see 'the illogic of the larder' – including the fact that farm animals are not merely wished into existence: in reality the crops required to feed chickens mean that we are really arguing about is what usage of land is optimal, bearing in mind the welfare of humans, human-reared animals and wild animals. This is a tricky topic in all sorts of ways. It is something for animal welfare valuations to grapple with, but it's not clear that it undermines any of the above.

A related question is how to treat states that are worse than death. In the examples above I have discussed raising chicken standards by 0.1 on a 0 to 1 scale, or even an apparent extreme of going from 0 to 1. Human QALY scores do usually range from 0 to 1, ranging from death-equivalent to perfect health. But in principle at least they can be negative, for 'states worse than death' and you can see how that might make sense for extreme suffering, e.g. being tortured. This consideration adds further complexity to quantifying relevant welfare changes. But it also makes it even easier to achieve changes of +0.1QALY/year or +1QALY/year, if existing chicken welfare ranges from some negative number up to 1, rather than only 0 to 1. And in cases where lives are not worth living, there is broader philosophical agreement that it is better not to bring those lives into existence.

This is another case where thought experiments might be useful. Imagine you're age 80 and you're all set to live for another 20 years in perfect health before dying at age 100. Then the devil comes along and gives you the chance to live an extra year at the end, but as a series of typical broiler chickens (suspending disbelief that this transfer of the self can meaningfully happen). Having done all the research about broiler lives you can, do you accept this bonus year on earth? I don't think I would: those lives have negative value. Indeed, I am so concerned about these lives being torture that – if the devil forced me to make a choice – I would agree to cut short my human life in order to avoid that year of chickenhood (and let's ignore family ties here). Perhaps I wouldn't sacrifice 20 years of human life to avoid the 1 year of torture-adjacent chickenhood, but I'd quite possibly sacrifice 3 years – once you really think about how bad broiler stress, disease and death might be. This might inform calculations, e.g. if we mostly eliminate chicken farming and ending their existence takes them from -3 to 0 on the QALY scale, my earlier calculations could rise further to $9 quadrillion of prevented suffering per year.  

Conclusion

It may seem silly to put a $ number on the potential worth of reducing chicken suffering worldwide. That calculation itself doesn't directly achieve anything. But given the cultural and economic dominance of animal agriculture, it's worth exploring every possible way of demonstrating and presenting the broad idea that "the treatment of domesticated animals in industrial farms is perhaps the worst crime in history". Maybe some people used to thinking in monetary terms will be swayed by the numbers on how many non-human-but-sentient beings are under our control, how awful their lives are, and how (to a much greater degree than with human welfare) we could change that at almost the stroke of a pen. Farm animal welfare is not a minor consideration to come back to after we've nailed all other problems in the world: compared to at least most other issues, it is a giant, yet one that can be tackled relatively quickly and cost-effectively.

And, to get back to the more practical use of these calculations, hopefully this post highlights the possibility of including non-human animal welfare in official regulatory analyses and perhaps any similar corporate accounting, just as the costs of emissions are increasingly valued. Bearing in mind that cost-benefit analyses will always include fairly-ballpark estimates, this blog has shown that it's possible to have a go. If government departments did this, it wouldn't be surprising if they chose input values that (by design) gave outputs 100x or 1000x smaller, but that may still be enough to sway some analyses. More importantly it ensures civil servants working on animal agriculture don't spend all of their time thinking about the impacts on humans and none on the animals; and any transparent methodology would at least show people what assumptions and trade-offs have been made.

Thanks for reading! Comments welcome! In particular: if you did want to value animal welfare changes within cost-benefit calculations, how could the approach or numbers used here be improved?