Hide table of contents

Summary

  • Corporate campaigns for chicken welfare increase wellbeing way more cost-effectively than the best global health and development (GHD) interventions.
  • In addition, the effects on farmed animals of such interventions can at least influence prioritisation within GHD, and those on wild animals might determine whether they are beneficial or harmful.
  • I encourage Charity Entrepreneurship (CE), Founders Pledge (FP), GiveWell (GW), Open Philanthropy (OP) and Rethink Priorities (RP) to:

Corporate campaigns for chicken welfare increase nearterm wellbeing way more cost-effectively than GiveWell’s top charities

Corporate campaigns for chicken welfare are considered one of the most effective animal welfare interventions. A key supporter of these is The Humane League (THL), which is one of the 3 top charities of Animal Charity Evaluators.

I calculated the cost-effectiveness of corporate campaigns for broiler welfare in human-years per dollar from the product between:

  • Chicken-years affected per dollar, which I set to 15 as estimated here by Saulius Simcikas. Note Saulius estimates broiler and cage-free campaigns affect 41 chicken-years per dollar, 2.73 (= 41/15) times as much as the broiler campaigns on which I am relying.
  • Improvement in welfare as a fraction of the median welfare range when broilers go from a conventional to a reformed scenario[1], assuming:
    • The time broilers experience each level of pain defined here (search for “definitions”) in a conventional and reformed scenario is given by these data (search for “pain-tracks”) from the Welfare Footprint Project (WFP).
    • The welfare range is symmetric around the neutral point[2], and excruciating pain corresponds to the worst possible experience.
    • Excruciating pain is 1 k times as bad as disabling pain[3].
    • Disabling pain is 100 times as bad as hurtful pain.
    • Hurtful pain is 10 times as bad as annoying pain.
    • The lifespan of broilers is:
      • In a conventional scenario, 44.5 days (= (42 + 47)/2), which is the mean between the means for the European Union and the United States according to WFP.
      • In a reformed scenario, 56 days according to WFP.
  • For both broilers and a random human, 8 h each day is spent sleeping, i.e. 1/3 (= 8/24) of the time.
  • For broilers, the welfare from positive experiences per time awake is symmetric of that of hurtful pain[4].
  • Median welfare range of chickens, which I set to RP's median estimate of 0.332.
  • Reciprocal of the intensity of the mean human experience, which I obtained supposing humans:
    • Sleep 8 h each day, and have a neutral experience during that time.
    • Being awake is as good as hurtful pain is bad. This means being awake with hurtful pain is neutral, thus accounting for positive experiences.

I computed the cost-effectiveness in the same metric for the lowest cost to save a life among GW's top charities from the ratio between:

  • Life expectancy at birth in Africa in 2021, which was 61.7 years according to these data from OWID.
  • Lowest cost to save a life of 3.5 k$ (from Helen Keller International), as stated by GW here.

The results are in the tables below. The data and calculations are here (see tab “Cost-effectiveness”).

Intensity of the mean experience as a fraction of the median welfare range

Broiler in a conventional scenario

Broiler in a reformed scenario

Human

-2.58*10^-5

-4.76*10^-6

3.33*10^-6

Broiler in a conventional scenario relative to a human

Broiler in a reformed scenario relative to a human

Broiler in a conventional scenario relative to a reformed scenario

-7.75

-1.42

5.45

Intensity of the mean experience as a fraction of that of the mean human experience

Broiler in a conventional scenario

Broiler in a reformed scenario

-2.57

-0.472

Improvement in chicken welfare per time when broilers go from a conventional to a reformed scenario as a fraction of...

The median welfare range of chickens

The intensity of the mean human experience

2.11*10^-5

2.10

Cost-effectiveness (human-years per dollar)

Corporate campaigns for broiler welfare

31.5

Lowest cost to save a life among GW's top charities

0.0176

Corporate campaigns for broiler welfare relative to lowest cost to save a life among GW's top charities

1.79 k

According to my results, corporate campaigns for broiler welfare are 1.79 k times as effective as the lowest cost to save a life among GW's top charities. I am not surprised. Here I got a ratio 6.48 (= 11.6/1.79) times as high, essentially because I used a moral weight 7.26 (= 2.41/0.332) times as high as RP's median welfare range (which I used above). This was not available at the time, but I trust it much more than my previous estimate, so I think the lower ratio of 1.79 k is more accurate.

To get a ratio of 1:

  • Everything else equal, the median welfare range of chickens (relative to humans) would have to be 1.85*10^-4 (= 0.332/(1.79*10^3)), which is 92.5 % (= 1.85/2.00) the one I guessed here for nematodes. I do not see this being possible.
  • Assuming broiler welfare is worth zero outside hedonism, this would have to be given a weight of 0.0559 % (= 1/(1.79*10^3)). This is very much against what Bob Fischer says here. “Even if hedonic goods and bads (i.e., pleasures and pains) aren’t all of welfare, they’re a lot of it. So, probably, the choice of a theory of welfare will only have a modest (less than 10x [i.e. at least 10 % weight for hedonism]) impact on the differences we estimate between humans’ and nonhumans’ welfare ranges”.

So the takeaway to me is that corporate campaigns for chicken welfare increase nearterm wellbeing robustly more cost-effectively than GW’s top charities, which are plausibly among the best GHD interventions.

Effects of global health and development interventions on animals are neglected and unclear

GHD interventions decrease mortality or increase economic growth. These tend to increase the consumption of farmed animals (see meat-eater problem), or impact net forest area, thus changing the number of animals. To illustrate, I show in the next sections the effects on animals of GW's top charities may at least influence which countries they should target, or even determine whether they are beneficial or harmful. Nonetheless, these considerations have not been researched by GW.

Farmed animals

The table below contains the relative reduction in the cost-effectiveness of saving lives due to increased consumption of poultry caused by saving lives in each of the countries targeted by GW's top charities analysed here. I have focussed on poultry because I think there is especially good data from WFP on the conditions of chickens. I got the estimates from the product between:

  • Absolute value of the intensity of the mean experience of broilers in a reformed scenario as a fraction of the median welfare range of chickens relative to the intensity of the mean human experience, which I estimated to be -1.42.
  • Median welfare range of chickens, which I set to RP's median estimate of 0.332.
  • Production of poultry per capita in 2019 in each country as a fraction of the global one to the power of 1.5.
    • I computed the fraction from these and these data from Our World in Data (OWID).
    • 1.5 instead of 1 such that each doubling of poultry consumption per capita makes the conditions of farmed chickens 1.41 (= 2^0.5) times as bad. This is a very rough approximation, as I expect the lives of farmed chickens to be positive for low poultry consumption per capita, and eventually become negative as it increases, which will arguably happen. From these data from OWID, the population of chickens in Africa increased 2.98 % (= (1.81/1.20)^(1/(2014 - 2000)) - 1) per year between 2000 and 2014.

The data and calculations are here (see tab “Poultry”).

Country

Consumption of poultry per capita in 2020 as a fraction of the global one (%)

Relative reduction in the cost-effectiveness of saving lives due to poultry (%)

Mean of the countries below

13.1

2.67

Burkina Faso

12.4

2.06

Cameroon

18.9

3.89

Chad

2.34

0.169

Cote d'Ivoire

16.0

3.01

Democratic Republic of Congo

0.659

0.0253

Guinea

5.51

0.612

Kenya

7.83

1.03

Mali

16.0

3.03

Mozambique

22.4

4.99

Niger

4.86

0.506

Nigeria

6.72

0.82

South Sudan

30.6

7.99

Togo

30.3

7.86

Uganda

9.27

1.33

World

100

47.2

These results suggest accounting for poultry does not matter much for GHD interventions. Among the countries targeted by GW’s top charities, the relative reduction in the cost-effectiveness of saving lives ranges from 0.0253 % for the Democratic Republic of Congo to 7.99 % for South Sudan.

Nevertheless, I believe the results above underestimate the reduction in cost-effectiveness, because:

  • I have not accounted for other farmed animals. From my estimates here, the negative utility of farmed chickens is only 30.6 % (= 1.42/4.64) of that of all farmed animals globally. This suggests accounting for all farmed animals would lead to a reduction in cost-effectiveness for the mean country of 8.72 % (= 0.0267/0.306), which is not negligible. So accounting for the effects of GHD interventions on farmed animals may lead to targeting different countries.
  • I have used the current consumption of poultry per capita, but this, as well as that of other farmed animals, will tend to increase with economic growth. I estimated the badness of the experiences of all farmed animals alive is 4.64 times the goodness of the experiences of all humans alive, which suggests saving a random human life results in a nearterm increase in suffering.

On the other hand, greater economic growth may be associated with moral circle expansion, and lead to technological innovations that can increase the welfare of farmed animals, or make alternatives more convenient, cheaper and tastier[5]. An additional major uncertainty is the welfare range of chickens. I have used RP's median estimate, but the 5th and 95th percentile are 0.602 % (= 0.002/0.332) and 2.61 (= 0.869/0.332) times as large. Furthermore, as Julian Jamison noted, assuming disabling pain is 10 (instead of 100) times as bad as hurtful pain leads to broilers in a conventional scenario having positive lives[6].

Overall, I am quite uncertain about the magnitude of the effect on farmed animals, but think it may well lead to at least different prioritisation within GHD interventions. So I believe it should be integrated in cost-effectiveness analyses of GHD interventions. This will involve further research, for instance, on forecasting how prevalent will factory-farming become in low-income countries.

Wild animals

The table below contains the absolute value of the relative variation in the cost-effectiveness of saving lives due to changes in the population of wild terrestrial arthropods caused by increased deforestation. I do not know whether the variation corresponds to an increase or decrease, as I am quite uncertain about whether wild arthropods have good or bad lives (see this preprint from Heather Browning and Walter Weit). I got the estimates from the product between:

  • Decrease in forest area per capita in 2015, which I computed from these and these data from OWID. As a 1st approximation, I assume net change is forest area is directly proportional to population.
  • Decrease in density of terrestrial arthropods due to deforestation, which I estimated to be 280 M/ha following this.
  • Intensity of the mean experience of wild terrestrial arthropods as a fraction of that of humans, which I estimated to be 0.200 % here (see 4th column of table).

The data and calculations are here (see tab “Wild terrestrial arthropods”).

Country

Decrease in forest area per capita in 2015 (m^2)

Decrease in the number of wild terrestrial arthropods per capita in 2015

Absolute value of the relative variation in the cost-effectiveness of saving lives due to wild terrestrial arthropods

Mean of the countries below

20.5

574 k

1.15 k

Cameroon

24.3

681 k

1.36 k

Mali

0

0

0

Mozambique

89.1

2.50 M

4.99 k

Niger

6.17

173 k

346

Nigeria

8.88

249 k

497

Togo

3.96

111 k

222

Uganda

11.0

308 k

616

World

6.93

194 k

388

The results suggest the increase in human welfare from GW's top charities saving lives is much smaller than the increase/decrease in that of wild terrestrial arthropods, since the absolute values of the relative variation in cost-effectiveness are much higher than 1. Nonetheless, these are quite uncertain because they are (in my model) directly proportional to the welfare range of silkworms. I have used RP's median estimate, but the 5th and 95th percentile are 0 (= 0/0.002) and 36.5 (= 0.073/0.002) times as large.

All in all, I can see the impact on wild animals being anything from negligible to all that matters in the nearterm. So, as for farmed animals, I think more research is needed. For example, on forecasting net change in forest area in low-income countries.

Note the impact on wild animals may also be the major driver of the overall nearterm effect of interventions which aim to improve the welfare of farmed animals[7]. For example, corporate campaigns for chicken welfare will tend to make chicken and eggs more expensive, which can lead to an increase in the consumption of beef, and therefore more deforestation, thus decreasing the population of wild terrestrial arthropods. Nevertheless, I think the positive/negative impact on wild animals is much larger for interventions which focus on reducing the consumption of farmed animals (like ones around abolitionism), instead of improving their living conditions.

Regarding the impact of human diet on animal welfare (of both farmed and wild animals), Michael St. Jules suggested Matheny 2005, this and these posts from Brian Tomasik, this post from Carl Shulman, and Fischer 2018.

Miscellaneous thoughts on organisations aligned with effective altruism

As far as I can tell, organisations aligned with effective altruism do not consider the effects of GHD interventions on animals. Below is some brief additional discussion, by alphabetical order of organisation.

Charity Entrepreneurship

CE seemingly has strong reasons to account for effects on animals. According to CE’s weighted animal welfare index, the “total welfare score (with evidence)” of:

  • “FF [factory-farmed] broiler chicken” is -1.75 (= -56/32) times that of a “human in a low middle-income country”, which is 1.01 times the value of -1.73 I got for broilers in a reformed scenario in my early cost-effectiveness analysis (see 1st table).
  • “Wild bug[s]” is -1.31 (= -42/32) times that of a “human in a low middle-income country”, which is 656 times the value of 0.200 % I used in my early estimation of the effects on animals.

These suggest the impacts of GHD interventions will be similar to what I estimated for farmed animals, and 3 orders of magnitude as large for wild animals.

Founders Pledge

As part of FP’s prioritisation, Stephen Clare and Aidan Goth published 3 years ago this analysis[8] comparing the cost-effectiveness of THL and Against Malaria Foundation (AMF), which is one of GW’s top charities. According to its Guesstimate model, the cost-effectiveness of THL is 852 (= 23/0.027) times that of AMF, which (considering the uncertainty involved) is pretty close to the ratio of 1.79 k I got in my early cost-effectiveness analysis.

Stephen and Aidan highlighted the moral weight of chickens relative to humans as a major uncertainty. However, this has meanwhile been narrowed down thanks to RP’s (great!) moral weight project. Maybe FP has not focussed much on animal welfare[9] due to other considerations, such as not having a fund for it (see FP’s funds).

GiveWell

GW determines the value of consumption and saving lives as a function of age based on surveys of its team, donors and beneficiaries (see here). I think it would make some sense to include questions about the importance of animals in such surveys. Nonetheless, I think it would be much better to combine RP's median welfare ranges with empirical evidence about how further away from the neutral point (as a fraction of the median range) is the mean experience of animals. Something like what I did, but way more in-depth!

I believe it would be hard for people to come up with good estimates describing the importance of animals in surveys. As Bob Fischer commented here:

The upshot of Jason's post on what's wrong with the "holistic" approach to moral weight assignments, my post about theories of welfare, and my post about the appropriate response to animal-friendly results is something like this: you should basically ignore your priors re: animals' welfare ranges as they're probably (a) not really about welfare ranges, (b) uncalibrated, and (c) objectionably biased.

Welfare ranges are not the sole determinant of the importance of animals, but they are a key input. So trusting our priors regarding them will imply coming up with an inaccurate assessment of how much consideration we should give to animals. Moreover, I suppose GW's team, donors and beneficiaries would not naturally be open to the possibility of defining moral weights as a function of the country, but that arguably makes sense given consumption of animals and deforestation vary across countries (and so do effects on animals). Alas, the moral weight of saving a life can even be negative under some circumstances (although killing people is still bad!).

Additionally, for the sake of transparency, it would be good if GW described in their website how they think about effects on animals. 8 months ago, I asked GW for feedback on this post related to the meat-eater problem. I was told my message was passed to the research team, but I have not heard back.

Open Philanthropy

From OP’s global health and wellbeing cause prioritisation framework:

When it comes to other outcomes like farm animal welfare or the far future [not so far if you think existential risk in the next 100 years is around 1/6], we practice worldview diversification instead of trying to have a single unified framework for cost-effectiveness analysis.

I think diversification makes sense in general, but the details matter. There is a (somewhat remote) sense in which a fossil fuel company is practising worldview diversification if it is decreasing its own emissions while increasing extraction of fossil fuels such that it overall contributes to global warming. However, if the goal really is mitigating global warming, it makes sense to focus on the overall contribution of the company to it.

Saving lives increases the nearterm welfare of humans, but it decreases that of farmed animals, and has unclear effects on wild animals. I think effects on farmed animals are sufficiently clear to be integrated into cost-effectiveness analyses, and that we should invest more resources into understanding those on wild animals (relative to global health and wellbeing interventions, at the margin).

Related to learning more, I am glad OP has supported RP's moral weight project. At the same time, I wonder whether it should have happened before it directed hundreds of millions of dollars towards GHD interventions. Not only because of their effects on animals, but owing to animal welfare interventions increasing wellbeing way more cost-effectively, as I showed in my early cost-effectiveness analysis. This is in agreement with OP’s post on worldview diversification[10]:

  • If you value chicken life-years equally to human life-years, this implies that corporate campaigns do about 10,000x as much good per dollar as top charities. If you believe that chickens do not suffer in a morally relevant way, this implies that corporate campaigns do no good.[3]
  • One could, of course, value chickens while valuing humans more. If one values humans 10-100x as much, this still implies that corporate campaigns are a far better use of funds (100-1,000x). If one values humans astronomically more, this still implies that top charities are a far better use of funds. It seems unlikely that the ratio would be in the precise, narrow range needed for these two uses of funds to have similar cost-effectiveness.

The value of chickens depends on how much weight one gives to hedonism, about which Alexander Berger (OP’s co-CEO) writes[11]:

We think that most plausible arguments for hedonism end up being arguments for the dominance of farm animal welfare. We seem to put a lot of weight on those arguments relative to you, and farm animal welfare is OP GHW’s biggest area of giving after GiveWell recommendations. If we updated toward more weight on hedonism, we think the correct implication would be even more work on FAW, rather than work on human mental health.

In the same comment, Alexander mentions:

We [OP] think it is a mistake to collapse worldviews in the sense that we use them to popular debates in philosophy, and we definitely don’t aim to be exhaustive across worldviews that have many philosophical adherents. We see proliferation of worldviews as costly for the standard intellectual reason that they inhibit optimization, as well as carrying substantial practical costs, so we think the bar for putting money behind an additional worldview is significantly higher than you seem to think. But we haven’t done a good job articulating and exploring what we do mean and how that interacts with the case for worldview diversification (which itself remains undertheorized). We appreciate the push on this and are planning to do more thinking and writing on it in the future.

If OP's worldviews are not supposed to correspond to popular debates in philosophy, and having more is costly, should the ones of nearterm animal and human welfare be unified? I agree worldview diversification "remains undertheorized".

I asked Alexander and Lewis Bollard at the end of January whether they thought this analysis about the effects of terrestrial arthropods on the cost-effectiveness of GiveWell's top charities was any relevant, but I have not heard back.

Rethink Priorities

RP’s Worldview Investigations Team seems perfectly positioned to study how to account for the effects on animals of GHD interventions, and figure out what the greater cost-effectiveness of corporate campaigns to increase wellbeing implies.

I asked here whether RP’s GHD team was considering addressing effects on animals in their work, but I have not heard back (and was downvoted). I had also contacted RP about the post on terrestrial arthropods at the end of January, and was told my message was forwarded to the GHD team, but I have not heard back either.

Complex cluelessness should not be ignored

I do not think it is fair to ignore the effects on animals because they look like a crucial consideration. We are in a case of complex cluelessness, not one of simple cluelessness where very uncertain effects can be ignored based on evidential symmetry. Me looking now to the right might ultimately create a storm somewhere, but just as well prevent it, so we can ignore these considerations. In contrast, increasing population size will robustly lead to greater consumption of food, which has certain impacts on farmed and wild animals.

I agree that, mathematically, E(“overall effect”) > 0 if:

  • “Overall effect” = “nearterm effect on humans” + “nearterm effect on animals” + “longterm effect”.
  • E(“nearterm effect on humans”) > 0.
  • E(“nearterm effect on animals”) = k_1 E(“nearterm effect on humans”).
  • E(“longterm effect”) = k_2 E(“nearterm effect on humans”).
  • k_1 + k_2 = 0.

That being said, setting k_1 + k_2 to 0 seems unfair under complex cluelessness. One could just as well say k_1 + k_2 = -1, in which case E(“overall effect”) = 0. Since I am not confident |k_1 + k_2| << 1, I am not confident either about the sign of E(“overall effect”), nor about whether GW's top charities are beneficial or harmful.

Let me try to illustrate how I think about this with an example (originally commented here). Imagine the following:

  • Nearterm effects on humans are equal to 1 in expectation.
    • This estimate is very resilient, i.e. it will not change much in response to new evidence.
  • Other effects (on animals and in the longterm) are -1 k with 50 % likelihood, and 1 k with 50 % likelihood, so they are equal to 0 in expectation.
    • These estimates are not resilient, and, in response to new evidence, there is a 50 % chance the other effects will be negative in expectation, and 50 % chance they will be positive in expectation.
    • However, it is very unlikely that the other effects will in expectation be between -1 and 1, i.e. they will most likely dominate the expected nearterm effects.

What do you think is a better description of our situation?

  • The expected overall effect is 1 (= 1 + 0) in expectation. This is positive, so the intervention is robustly good.
  • The overall effect is -999 (= 1 - 1 k) with 50 % likelihood, and 1,001 (= 1 + 1 k) with 50 % likelihood. This means the expected value is positive. However, given the lack of resilience of the other effects, we have little idea whether it will continue to be positive, or turn out negative in response to new evidence. So we should not act as if the intervention is robustly good. Instead, it would be good to investigate the other effects further, especially because we have not even tried any hard to do that in the past.

Am I uncertain about the value of killing people too?

No, killing people is bad! Not saving lives has drastically different consequences from killing people, which is much more anti-cooperative. For what it is worth, I think I am much more against killing than the median citizen. For example, I suspect most people would be in favour of militarily supporting Ukraine even if it was known that it increased the number of people killed in the Russo-Ukrainian War, whereas I would tend to prefer whatever prevented the most war deaths.

However, for the same reasons I am not confident about whether saving lives is good or bad, I do not know whether a random person dying (without being killed) is beneficial or harmful.

I do not know whether saving lives is good longterm

One can argue saving lifes is robustly good longterm (k_2 >> k_1) based on the capability approach to human welfare, despite nearterm effects on humans plus animals being unclear. I am sympathetic to this argument, but think it is too general. There are obvious benefits of being able to live a long and healthy life, but I also worry about humans having the capability of factory-farming animals whose lives are pretty bad. Note the title of the post is “the capability approach to human welfare” (emphasis mine). Interestingly, I have recently listened to Martha Nussbaum on the Clearer Thinking podcast, and it looks like her book Justice for Animals: Our Collective Responsibility attempts to extend the capability approach to non-human animals.

In the same way it is better to focus on differential progress over economic growth, I would rather increase good capabilities over all capabilities, and it is unclear to me what is the net effect of increasing population at the margin. There are many indirect longterm effects. The answer may vary too, depending on factors like year, country and age.

I believe saving lives would more easily be good if there were much fewer humans, because in that case it would decrease the risk from extinction, which is good given my presumption that the expected value of the future is positive. I am open to the possibility that saving lives is a good proxy for longterm value for the current population too, but this is not obvious to me. I think it warrants empirical investigation, for example, into impacts on democracy levels. This in particular seems to be a neglected topic. From Kono 2009 (emphasis mine):

Although many people have argued that foreign aid props up dictators [and so might GHD interventions?], few have claimed that it props up democrats, and no one has systematically examined whether either assertion is empirically true. We argue, and find, that aid has both effects. Over the long run [what matters most?], sustained aid flows promote autocratic survival because autocrats can stockpile this aid for use in times of crisis. Each disbursement of aid, however, has a larger impact on democratic survival because democrats have fewer alternative resources to fall back on.

In addition, I tend to think it would be a surprising and suspicious convergence if saving lives as cost-effectively as possible was the best way to improve the longterm future. I would expect metrics more closely related to existential risk to be better. For example:

Additionally, it is worth keeping in mind longtermist interventions can save lives quite cost-effectively too. For example:

  • The cost-effectiveness of 3.95 bp/G$ I estimated here for longtermism and catastrophic risk prevention (for method 3 with truncation) naively corresponds to saving a life for 316 $ (= 1/(3.95*10^-4*8)), which is 11.1 (= 3500/316) times as cost-effective as the lowest cost among GW's top charities (from Helen Keller International).
  • Joel Tan estimated lobbying for arsenal limitation is 5 k times as cost-effective as GW’s top charities. “The headline cost-effectiveness will almost certainly fall if this cause area is subjected to deeper research”. “That said, results are robust, insofar as the low-confidence tractability estimates can drop by three whole magnitudes and still leave the intervention to be comfortably more cost-effective than GiveWell[‘s top charities]”.

Note these interventions would look even more cost-effective after accounting for their effect on the far future.

What would I like to see?

Thinking at the margin, I would say scope-sensitive ethics imply prioritising animal welfare over global health and development. I think the scale of the welfare of farmed animals and wild terrestrial arthropods is 4.64 and 253 k times as large as that of humans, so accounting for them seems crucial a priori.

So I encourage organisations, especially the ones I discussed above aligned with effective altruism, to:

Acknowledgements

Thanks to Jeff Kaufman, Michael St. Jules, and Sanjay Joshi for feedback on the draft.

  1. ^

     From this page of WFP, broilers in reformed scenarios have an average daily gain of 45 to 46 g/d, whereas ones in conventional scenarios have 60 and 62 g/d.

  2. ^

     This assumption influences the improvement in welfare as a fraction of the median welfare range, but not the cost-effectiveness of corporate campaigns for broiler welfare in human-years per dollar. For example, if welfare could range from something as good as disabling pain is bad to excruciating pain, the welfare range would become 50.05 % (= (1 + 1 k)/(2 k)) as large. Consequently, the improvement in welfare as a fraction of the median welfare range would become 1.998 (= 1/0.5005) times as large, but so would the intensity of the mean human experience. As a result, the cost-effectiveness in human-years per dollar would remain the same, since it is directly proportional to the improvement in welfare as a fraction of the median welfare range, and to the reciprocal of the intensity of the mean human experience.

  3. ^

     I encourage you to check this post from algekalipso, and this from Ren Springlea to get a sense of why I think the intensity can vary so much.

  4. ^

     This assumption affects the (signed) intensity of the mean experience of broilers, but not the improvement in their welfare when they go from a conventional to a reformed scenario, because the lifespan of broilers and value of them being alive is the same in both scenarios. As a consequence, the assumption does not impact the cost-effectiveness of corporate campaigns for broiler welfare.

  5. ^

     Thanks to Sanjay Joshi for noting this point.

  6. ^

     The intensity of the mean experience as a fraction of the median welfare range would be 8.24 %, instead of -777 %.

  7. ^

     Thanks to Michael St. Jules for noting this point. I had thought about it, but had not written it down, possibly due to motivated reasoning.

  8. ^

     If I recall correctly, the one which got me thinking about comparisons between animal welfare and GHD interventions!

  9. ^

     Their only report on animal welfare was published in November 2020.

  10. ^

     Thanks to Michael for noting these points.

  11. ^

     Thanks to Michael for letting me know about Alexander’s comment.

  12. ^

     See section “Climate damage is increasing non-linearly” in this report from FP.

Comments50
Sorted by Click to highlight new comments since:

Hey Vasco,

Love the post; I think it is super valuable to have these sorts of important conversations, directly thinking about cross-cause comparison. It’s worth noting that CE does consider cross-cause effects in all the interventions we consider/recommend, including possible animal effects and WAS effects. Despite this, CE does not come to the same conclusion as this post; here are a couple of notes on why:

Strength of evidence discounting: CEAs are not all equal when they are based on very different strengths of evidence, and I think we weight this factor a lot heavier. It's quite common for the impact of any given intervention to regress fairly heavily as more research/work is put into it. We have found this in CE’s, GW’s and other EAs’ research. This can be seen in even more depth in the GiveWell and EA forum writings on deworming and how to deal with speculative effects that possibly have very high upsides. For example, I would expect a five-hour CEA to be constantly off (almost always in a positive direction) compared to a 50-hour CEA. A calculation made at two different levels of rigor should not be directly compared. (This does not mean shorter-form CEAs are not worth doing, but I think we have to take their cons and likely regressions a lot more seriously than this post currently does.) This discounting should be even more heavily applied to flow-through effects, as the evidence for them is way lighter than the direct effects. We tend to use something akin to the weighted quantitative modeling used here.

Marginal funding and reliability in effects: Here’s a good example of how a CEA can regress really quickly; GiveWell typically does CEAs on marginal donations made, whereas many other CEAs - including the one you use from Saulius - do not consider marginal funding. I currently think that the marginal dollar to corporate campaigns is way less impactful when compared to the average dollar of spending pre-2018. This can affect a CEA quite drastically. Another example is the funding of numerous animal interventions through corporate campaigns, which have become the “hit” of the animal movement. However, these campaigns often are seen as cost-effectiveness without clear before hand knowledge of the impact an additional dollar of funding would have accomplished. It is a bit like measuring CE’s cost effectiveness by looking at the top charity we incubated and assuming future charities will be equal to that. Variance is a real pain, and it’s not even clear if other corporate campaigns will be equally cost-effective to cage-free. On the other hand, top GW charities have this built in; they are not estimating the average EV of AMF’s top three historical campaigns, they are estimating the impact of marginal average future funding.

Variable animal effects dependent on intervention: You touch on this, but I think there is an important point missed. The effects on animals vary quite a lot, depending on the intervention. Interventions that primarily affect mortality in Africa, for instance, end up looking like how you describe. But morbidity-focused interventions, mental health focused interventions, and family planning interventions are all significantly less affected by this consideration. Same goes for any intervention that operates in contexts where there is lower meat consumption (such as in India). I think if you remodeled this for an organization like Fortify Health (Iron fortification in India), it would result in rather different outcomes.

If you combine these factors and look at a marginal dollar to FH vs a marginal dollar to THL (both of them with similarly rigorous CEAs and flow-through effects that are discounted based on certainty), I think the outcomes would be different enough to change your endline conclusion.
The non-epistemic difference I have is to do with ecosystem limitations, and is more specific to CE itself vs. general EA organizations. When we launch a charity, we need 1) founders 2) ideas, and 3) funding. Each of these are fairly cause area limited (and I think limiting factors are often more important than total scale). For example, if we aimed to found 10 animal charities a year (vs 10 charities across all the cause areas we currently focus on) I do not think the weakest two would be anywhere near as impactful as the top two, and only a small minority of them would get long-term funding. In fact, with animal charities making up around a third of those we have launched, I think we already run close to those limitations. This means that even if we thought that animal charities were more impactful than human ones on average, the difference would have to be pretty large for us to think that adding a 9th or 10th animal charity into the animal ecosystem would be more impactful than adding the first or second human-focused charity. I expect a version of this consideration can apply to other actors too. In general, I believe that given the current ecosystem, more than ~three-five charities founded per year within a given area would start to result in cannibalization between charities.

Thanks again for the consideration of this; I do think people should do a lot more cross-cause thinking, and I expect there are some really neglected areas that have significant intercausal impact.

Hi Joey,

It’s worth noting that CE does consider cross-cause effects in all the interventions we consider/recommend, including possible animal effects and WAS effects.

I have searched for "animal" in all the 16 reports of CE's global health and development recommendations, and I did not find any discussion that extending human lives would increase the consumption of animals. In contrast, decreasing the birth rate is highlighted as being a positive externality in terms of animal welfare in 3 of the 16 reports:

  • "A lower birth rate is also associated with fewer CO2 emissions and a gain of welfare
    points due to averted consumption of animal products" (here).
  • "Finally we believe this intervention could have important positive externalities on animal welfare. Increasing uptake for contraception and preventing unintended births would reduce family sizes and their overall consumption in animal products.
    A lifetime of consumption of these products leads to an considerable amount of suffering for animals raised in factory farms. Preventing unintended births therefore indirectly decreases demand for these products, thereby decreasing the number of animals raised for food. We have modeled these effects using CE̗s welfare points system in our CEA" (here).
  • "We found that this intervention has two kinds of externalities. It positively affects climate change, with three tonnes of CO2 emissions per dollar spent. It also positively affects animal welfare, with 377 welfare points gained per dollar spent" (here).

I think CE's reports should mention the negative externalities on farmed animals due to extending human lives, considering CE's reports on family planning discuss the positive externalities on farmed animals due to decreasing fertility.

The last bullet above also illustrates your global health and development recommendations could be net harmful based on your own numbers. I think 100 welfare points (WPs) are roughly as good as averting 1 DALY (because 100 WPs is the maximum total welfare possible), so 377 WP/$ of positive externalities respect around 3.77 DALY/$. This is 379 (= 3.77/0.00994) times the cost-effectiveness of GiveWell's top charities of 0.00994 DALY/$, i.e. the effects on animals are way larger than those on humans according to CE's report. This claims the effects on animals are positive due to decreasing population size, so it directly follows that saving lives (increasing population size) has negative effects on animals, and the negative effects on animals would be much larger than those on humans (trusting the numbers of the report).

Hey Vasco,

Quick response below as I am limiting my time on the EA forum nowadays. I am far less convinced that life saving interventions are net population creating than I am that family planning decreases it. Written about 10 years ago, but still one of the better pieces on this IMO is David Roodman's report commissioned by GiveWell.
In addition, our welfare points are far less certain estimates when compared to our global health estimates. This matters a lot, e.g., I would regress weaker CEAs by over 1 order of magnitude even from the same organization using similar methods, and it could be 3+ orders of magnitude across different orgs and methods. AIM in general is pretty confident e.g. that our best animal charities are not 379x better than a top GiveWell charity even if a first pass CEA might suggest that.

I think for externalities you can get yourself pretty lost down a rabbit hole based on pretty speculative assumptions if you are not careful. We try to think of it a bit like the weight quantitative modeling described here and only include effects that we think are major (e.g. 10%+ effect after uncertainty adjustments on the total impact). We also try to take into account what effects we expect founders considering these ideas would most likely consider to be decision relevant for them.

In general I think we aim to be more modest about moral estimates (particularly when they are uncertain or hotly debated) and try to recommend the peak intervention across different cause areas without making a final verdict. I also think this call in our case does not reduce our impact as there are pretty natural caps to every cause area, e.g., I do not think the animal movement could effectively absorbed 10 new charities a year anyways.

I hope this is helpful!

Best,
Joey

Thanks, Joey!

I am far less convinced that life saving interventions are net population creating than I am that family planning decreases it. Written about 10 years ago, but still one of the better pieces on this IMO is David Roodman's report commissioned by GiveWell.

Fair! From Wilde et. al (2020), whose abstract is below (emphasis mine), bednets increase fertility 1 to 3 years after their distribution, but decrease it afterwards, so population initially increases (because bednets also decrease nearterm mortality), but may decrease soon after the distribution.

We examine the extent to which recent declines in child mortality and fertility in SubSaharan Africa can be attributed to insecticide-treated bed nets (ITNs). Exploiting the rapid increase in ITNs since the mid-2000s, we employ a difference-in-differences estimation strategy to identify the causal effect of ITNs on mortality and fertility. We show that the ITN distribution campaigns reduced all-cause child mortality, but surprisingly increased total fertility rates in the short run in spite of reduced desire for children and increased contraceptive use. We explain this paradox in two ways. First, we show evidence for an unexpected increase in fecundity and sexual activity due to the better health environment after the ITN distribution. Second, we show evidence that the effect on fertility is positive only temporarily – lasting only 1-3 years after the beginning of the ITN distribution programs – and then becomes negative. Taken together, these results suggest the ITN distribution campaigns may have caused fertility to increase unexpectedly and temporarily, or that these increases may just be a tempo effect – changes in fertility timing which do not lead to increased completed fertility.

I guess the above partly generalises to other interventions. If saving lives decreases population, it may well decrease welfare (if the increase in welfare per capita is not sufficiently large), thus being harmful under many moral views. Likewise for family planning interventions. CE's theories of change for the family planning interventions of the 3 reports I mentioned above have as outcome decreasing unwanted pregnancies. Are you assuming this is intrinsically valuable, or are you super confident that it leads to higher human welfare (because the increase in human welfare per capita exceeds the decrease in population)? I think the outcome should at least be increasing human welfare (and, ideally, increasing welfare accounting for both humans and animals).

I think for externalities you can get yourself pretty lost down a rabbit hole based on pretty speculative assumptions if you are not careful. We try to think of it a bit like the weight quantitative modeling described here and only include effects that we think are major (e.g. 10%+ effect after uncertainty adjustments on the total impact).

I agree, but I think it is at least worth mentioning the potential negative externalities on animals (without getting lost into rabbit holes). I also think it would be good to justify that the regression of the potential negative externalities is so large that they become negligible, especially if a direct interpretation leads one to conclude they overwhelm the direct effects (as in the report I discussed in the previous comment).

I am far less convinced that life saving interventions are net population creating than I am that family planning decreases it. Written about 10 years ago, but still one of the better pieces on this IMO is David Roodman's report commissioned by GiveWell.

From the abstract of David Roodman's paper on The Impact of Life-Saving Interventions on Fertility (written in 2014):

In places where lifetime births/woman has been converging to 2 or lower, saving one child’s life should lead parents to avert a birth they would otherwise have. The impact of mortality drops on fertility will be nearly 1:1, so population growth will hardly change. In the increasingly exceptional locales where couples appear not to limit fertility much, such as Niger and Mali, the impact of saving a life on total births will be smaller, and may come about mainly through the biological channel of lactational amenorrhea. Here, mortality-drop-fertility-drop ratios of 1:0.5 and 1:0.33 appear more plausible.

So it looks like saving lives in low income countries decreases fertility, but still increases population size.

Hi Joey,

Thank you so much for taking the time explain your reasons in great detail! I broadly agree with all the points you make. 

It’s worth noting that CE does consider cross-cause effects in all the interventions we consider/recommend, including possible animal effects and WAS effects.

Could you elaborate on how CE does this? Among the 9 CE's health reports of 2023, I only found 3 instances of the word "animal". Here (emphasis mine):

A lower birth rate is also associated with fewer CO2 emissions and a gain of welfare points due to averted consumption of animal products.

Here (emphasis mine):

There are reasons to believe that this situation may change in the near future, as poorer countries undergo the so-called “nutrition transition” toward diets high in sugars, fat, and animal foods (Reardon et al., 2021).

Here (emphasis mine):

Animal studies also suggest that improving oxygen access may reduce mortality rates.

Only the 1st of these refers to animal welfare, and has very little detail.

Marginal funding and reliability in effects: Here’s a good example of how a CEA can regress really quickly; GiveWell typically does CEAs on marginal donations made, whereas many other CEAs - including the one you use from Saulius - do not consider marginal funding. I currently think that the marginal dollar to corporate campaigns is way less impactful when compared to the average dollar of spending pre-2018. This can affect a CEA quite drastically.

Saulius commented that (emphasis mine):

Hey, I am the author of the corporate campaigns cost-effectiveness estimate you mention. In case it's relevant, I recently spent 3 months doing another (much more detailed) cost-effectiveness estimate of chicken welfare reforms (corporate and legislative) that I unfortunately can not make public.  According to this new estimate, in 2019-2020 chicken welfare reforms affected 65 years of chicken life per dollar spent. According to the same new estimate, the cost-effectiveness in 2016-2018 was about 2.5 times higher. So while it's true that lately campaigns were not as cost-effective as they were some years ago, I think that they are still very cost-effective. In fact, even more  cost-effective than my linked report [which I used in my post] suggests because in that report I think I underestimated the cost-effectiveness. Also, because of the research of the Welfare Footprint Project, I now think that these reforms are more important to chickens than I  thought previously (although I haven't yet examined the broiler book in detail).

So cost-effectiveness used to be higher, but Saulius' updated estimate of 65 years of chicken life per dollar is 4.33 (= 65/15) times as high as the one I used in my BOTEC. If the 2019-2020 average cost-effectiveness is also about 4.33 times as high as the current marginal cost-effectiveness, my BOTEC will not be too off. I did not easily find estimates for the marginal cost-effectiveness. Kieran Greig (from RP) surveyed groups working on corporate campaigns globally, and told me roughly 1 year ago that:

These campaigns have some pretty significant room for more funding. Easily in the millions of dollars per year.[1]

Are there any quantitative analyses of the marginal cost-effectiveness?

The effects on animals vary quite a lot, depending on the intervention. Interventions that primarily affect mortality in Africa, for instance, end up looking like how you describe. But morbidity-focused interventions, mental health focused interventions, and family planning interventions are all significantly less affected by this consideration.

Great point! It crossed my mind, but I ended up not including it.

Strength of evidence discounting: CEAs are not all equal when they are based on very different strengths of evidence, and I think we weight this factor a lot heavier. It's quite common for the impact of any given intervention to regress fairly heavily as more research/work is put into it.

I agree this tends to be the case, but I am not sure how much. For example, I have the impression RP's median welfare ranges are higher than what most people expected a priori. In general, it seems hard to know how much to adjust estimates, and I guess it would be better to invest more resources (at the margin) into decreasing our incertainty.

  1. ^

    Further details are confidential:

    - "I apologize that I can't share too much specifically as I promised organizations that those results would be confidential".

This seems to rest heavily on Rethink Priorities' Welfare Estimates. While their expected value for the "welfare range" of chickens is 0.332 that of humans, their 90% confidence for that number spans 0.002 to 0.869, which is so wide that we can't make much use of it.

Seems to be a tendency in EA to try to use expected values when just admitting "I have no idea" is more honest and truthful.

I mean to be fair to OP (edit: I meant original poster) they make their uncertainty really clear throughout and the conditionals it entails. I don't think it's fair to say they're not being honest and truthful.

Hi zchuang,

I agree OP's writings have high reasoning transparency (certainly much more than my posts). In the very 1st bullet of their post on worldview diversification, they write:

  • Some people think that animals such as chickens have essentially no moral significance compared to that of humans; others think that they should be considered comparably important, or at least 1-10% as important. If you accept the latter view, farm animal welfare looks like an extraordinarily outstanding cause, potentially to the point of dominating other options: billions of chickens are treated incredibly cruelly each year on factory farms, and we estimate that corporate campaigns can spare over 200 hens from cage confinement for each dollar spent. But if you accept the former view, this work is arguably a poor use of money.

However, after RP's moral weight project, I do not think it is reasonable to assume (in expectation) that "chickens have essentially no moral significance compared to that of humans". In general, OP's decision-making around how much should be allocated to each worldview remains unclear to me.

Sorry I meant OP as in original poster not OpenPhil. But nice response nonetheless!

I'd suggest editing your top-level post (with brackets, like this: [the original poster, originally wrote "OP" which was ambiguous])

Hi Henry,

Thanks for engaging!

This seems to rest heavily on Rethink Priorities' Welfare Estimates. While their expected value for the "welfare range" of chickens is 0.332 that of humans, their 90% confidence for that number spans 0.002 to 0.869, which is so wide that we can't make much use of it.

Note that:

According to my results, corporate campaigns for broiler welfare are 1.71 k times as effective as the lowest cost to save a life among GW's top charities.

So, using RP's 5th percentile welfare range instead of the median one, corporate campaigns for broiler welfare are still 10.3 (= 1.71*10^3*0.002/0.332) times as effective. However, there is also large uncertainty in how bad are the lives of broilers and human relative to their median welfare ranges. This means the true 5th percentile will tend to be lower than the 10.3 I just calculated. I guess the uncertainty stemming from the median welfare range is similar to that from the mean experience relative to the median welfare range, so I think there is less than 10 % chance that corporate campaings for broiler welfare are less effective than the lowest cost to save a life among GW's top charities. I suppose RP will look into building on their moral weight project.

Seems to be a tendency in EA to try to use expected values when just admitting "I have no idea" is more honest and truthful.

I am also concerned about acting as if expect values are resilient, i.e. assuming they will not easily change in the future in response to new information. On the other hand, large uncertainty in the welfare range of chickens does not necessarily imply the median welfare range lacks resilience. My understanding is that RP's research tried to integrate most of the available evidence, which means narrowing the interval of possible values may be difficult.

Hi Vasco - 

It's great that you're so passionate about this, but I find it extremely surprising that you're willing to draw such strong conclusions based on such weak evidence and ad hoc assumptions. For instance if I change your assumption that debilitating pain is 100x as bad as hurtful pain, and instead assume that it is only 10x as bad (and don't change anything else), your calculations imply that even under the conventional scenario broiler chickens have net positive lives (and hence presumably that we should be eating as many of them as possible and donating to advocacy groups that promote chicken consumption, at least given total utilitarianism). 

Are you so certain that it is 100x as bad, even within an order of magnitude? If so why? I did read both of the links in your fn 2 but found them unconvincing for your claims, e.g. the first one only discusses the logarithmic nature of the scale but nothing about specific magnitudes. From the second one we learn that multiple voluntary human activities (tattoos, 'natural' childbirth, perhaps eating hot peppers) fall into the disabling category... which suggests maybe it's not necessarily so horrific after all.

That's just one assumption amongst many. And yet your "takeaway is that corporate campaigns for chicken welfare increase nearterm wellbeing robustly [emphasis mine] more cost-effectively than GW’s top charities". 

Julian

I don't think this assumption Vasco made is reasonable, and it substantially overestimates the pleasure conventional broilers are likely to experience:

Broilers being awake is as good as hurtful pain is bad. This means being awake with hurtful pain is neutral, thus accounting for positive experiences.

There are a few issues with this:

  1. While in disabling pain, they shouldn't be experiencing any pleasure, by WFP's definition, so you should subtract the time spent in disabling pain.
  2. While suffering from behavioural deprivation, they probably shouldn't experience any significant pleasure, either, so you should subtract that time, too. They're suffering from behavioural deprivation because they're prevented from being active. They probably don't find just sitting around very pleasurable, although it could be mildly pleasurable (annoying pain intensity). (WFP's estimates assume they don't suffer from behavioural deprivation while eating.)
  3. It would be surprising if all of their leftover time was pleasurable of intensity similar to hurtful pain, rather than just annoying pain. Per day on average, they spend around 3.2 hours eating, 0.25 to 3 hours foraging/exploring (in the 3rd week of life and after, but more in the first two weeks) and at most 0.25 hours dustbathing, based on WFP's estimates. Some of this could be pleasurable of intensity similar to hurtful pain and generously all of it could be. They don't generally have enough space to play, and even if they did, they'd probably do it during their foraging hours already accounted for. The rest of the time is presumably basically inactive, e.g. resting, which could be pleasurable, but it seems unlikely to have intensity (much) greater than annoying pain, so should probably roughly match annoying pain. This inactive time could also easily be unpleasant instead, too, because of the high stocking densities (from social stress, heat, feces, air quality).

I'd expect that Vasco overestimated the amount of pleasure they experience at least 2 times with this assumption. We get at least around 2x too much just from point 3, assuming the inactive hours aren't very pleasurable.

Note that some pains WFP estimated overlap in time, so you don't want to double or triple subtract times spent in pain, and this makes actually calculating the time left for pleasure trickier. Even WFP's pain estimates attributed to lameness is made up of 3 types of pains that may overlap in time for the most severe cases of lameness: the direct pain of the condition in the legs, hunger from not eating enough because it's painful to get food, and thirst for the same reason. Also, pains are probably subadditive, but WFP treated them as additive, so may have overestimated pain this way.

The math is easier for egg-laying hens in conventional cages because the only particularly pleasurable activities they might engage in are eating, around 2-4 hours/day. They don't get to dustbathe, forage, explore, walk around or even stretch their wings in conventional cages. Even in furnished cages, they quickly run out of litter to forage.

Thanks for the input Michael - your estimates seem reasonable / defensible to me. On the other hand, it also seems reasonable / defensible to argue that time spent just sitting around is fairly highly pleasurable for chickens (relative to their maximum): many humans prefer doing nothing to active foraging (NB I'm being serious), and chickens (like all prey) are evolved to be wary of predators and at risk of dying at any moment. My sense is that the default welfare state for all living beings is nontrivially positive (we see this in human survey data, and it makes sense evolutionarily), so a chicken that is both alive and not at risk of being eaten or starving might be in very good shape in chicken terms. I simply don't know, which leads to...

However the broader point, which all three of us seem to agree on, is that all of these estimates are wildly uncertain and should be taken with many large grains of salt and (imo) not used to draw any firm conclusions about what should happen (except that we can agree less pain is better than more pain). Reasonable people can and do disagree about what it's like to be a chicken in captivity.

I appreciate you pointing out these possibilities. You might indeed be right, and I think it's a position new evidence could end up supporting. However, I don't think you or really anyone would be warranted in believing the average broiler welfare overall to be positive in expectation if they were well-informed about their conditions and the current state of evidence. Maybe we should just withhold judgement. However, using Welfare Footprint Project's analysis, and being, like them, careful in the attribution of welfare states and more careful the more intense, there would be more expected pain than expected pleasure.

I do think it's plausible the default (e.g. most common) welfare state for wild red jungle fowls, i.e. the chicken's wild progenitor and counterpart, is positive, or at least that positive is more common than negative. I might even lean somewhat towards that, but it depends on how common the threat of predators is and how long-lasting the negative effects of predator exposure are. But this and comparisons to humans (which ones?) are quite weak priors from which to conclude broilers frequently experience pleasure of intensity similar to hurtful pain just from sitting/resting, and there are multiple reasons to be skeptical or even expect negative welfare instead. Conventionally farmed chickens are in very unnatural, monotonous and limiting environments, often have painful and limiting health conditions and face multiple chronic stressors their wild counterparts don't face. Their environments are especially not conducive to high baseline moods or much good to attend to when they're not active, and they also contain substantial bad.[1]

On comparisons to wild animals, as foragers, I don't imagine red jungle fowls would often be at risk of starvation in the wild, and in fact a decent share of broilers (or hours of broiler life) suffer from hunger and thirst, according to WFP: broiler breeders in particular are chronically hungry and food-deprived, and other severely lame broilers also seem to suffer significantly from hunger. I agree that the absence of predators should make a difference (although I'm very uncertain about how much). A condition-informed survey of expert opinion found the welfare of conventional broilers below the cutoff for "acceptable welfare" (although this doesn't imply net negative in particular) and far below the welfare for nature, which was the second highest rated after only backyard flocks. Ratings of nature had relatively high variance, with 3 of the 27 experts even putting it below the acceptability cutoff.

 

Also, if sitting around is pleasurable at hurtful intensity, and disabling pain is only 10x as intense as hurtful pain, things like fresh large bone breaks (e.g. leg in humans, keel in chickens), the pain of birth without painkillers or anaesthesia, panic attacks, the part of a tattoo experience where it felt "Like someone slicing into my leg with a hot, sharp live wire"  (assuming they are disabling and not excruciating) would only be about 10x as bad as just sitting around is good per minute. I personally find that counterintuitive. Maybe you don't, but it's worth pointing out what the conjunction of views you're defending implies.

  1. ^

    Maybe just sitting is comfortable, but it could be uncomfortable due to poor litter quality (e.g. ammonia buildup) and contact burns/dermatitis, leg pain or heat (although I think the most intense of these are largely already accounted for by WFP). Maybe watching other chickens is interesting, but it could be stressful, given high stocking densities and social dominance. Maybe their inactive non-highly pained moods are based on some kind of mean of their active and pained welfares, like if you have fun often and don’t suffer often, you’ll still be in a good mood when you’re not having fun, and if you’re in pain often, but don’t have much fun, you’ll be in a bad mood even when you’re not in much pain.

Thanks for commenting, Michael!

Broilers being awake is as good as hurtful pain is bad. This means being awake with hurtful pain is neutral, thus accounting for positive experiences.

I agree there are a few issues with this. However (I have added what follows in a footnote):

This [the above] assumption affects the (signed) intensity of the mean experience of broilers, but not the improvement in their welfare when they go from a conventional to a reformed scenario, because the lifespan of broilers and value of them being alive is the same in both scenarios. As a consequence, the assumption does not impact the cost-effectiveness of corporate campaigns for broiler welfare.

If the WFP is capturing most of the painful experiences (weighted by intensity), and pleasurable experiences are negligible, then my assumption will not influence the cost-effectiveness of corporate campaigns. It can potentially change whether chickens have good or bad lives, and therefore impact whether consuming less animals is good/bad, but I think this is pretty unclear anyway for other reasons (e.g. effects on wild animals).

I'd expect that Vasco overestimated the amount of pleasure they experience at least 2 times with this assumption. We get at least around 2x too much just from point 3, assuming the inactive hours aren't very pleasurable.

If I assume all the time not classified by the WFP is neutral, I get the lives of broilers in a conventional and reformed scenario are, per unit time, 3.08 and 1.07 times as bad as human lives are good. So the lives of broilers in a conventional and reformed scenario would become worse by a factor of 1.19 (= 3.08/2.58) and 1.86 (= 1.07/0.574).

Great points, Julian!

For instance if I change your assumption that debilitating pain is 100x as bad as hurtful pain, and instead assume that it is only 10x as bad (and don't change anything else), your calculations imply that even under the conventional scenario broiler chickens have net positive lives (and hence presumably that we should be eating as many of them as possible and donating to advocacy groups that promote chicken consumption, at least given total utilitarianism).

I assumed disabling (not debilitating) pain is 100 times as bad as hurtful pain, but my 90 % confidence interval would be something like 10 to 1 k. As a result, I would not be too surprised if broilers in conventional scenarions had positive lives.

That's just one assumption amongst many.

Because of this, I am decently open to the possibility that we should be eating more/less factory-farmed animals (including chickens). 

And yet your "takeaway is that corporate campaigns for chicken welfare increase nearterm wellbeing robustly [emphasis mine] more cost-effectively than GW’s top charities".

Supposing hurtful, disabling and excruciating pain are each as bad as annoying pain (instead of 10, 1 k and 1 M times as bad, as I guessed), the cost-effectiveness of corporate campaigns for broiler welfare would still be 40.5 times that of the lowest cost to save a human life. In other words, for corporate campaigns for broiler welfare to be as effective as GW's top charities, one would have to assume, for example:

  • Hurtful pain 1 order of magnitude (OOM) less bad.
  • Disabling pain 3 OOMs less bad.
  • Excruciating pain 6 OOMs less bad.
  • A median welfare range of chickens 2.5 % (= 1/40.5) as high as RP's best guess.

Combinations like this seem sufficiently unlikely for one to say "corporate campaigns for chicken welfare increase nearterm [emphasis mine] wellbeing robustly more cost-effectively than GW’s top charities". If we include indirect longterm effects, I still guess corporate campaigns to be more effective, but not robustly so.

Thanks for the quick and constructive reply! 

(and yes apologies for the typo: I meant "disabling" not "debilitating")

I admit I'm still unconvinced by several of the assumptions and still believe that they require a bit more discussion / support / defense; e.g. in addition to the ones above, the claim that welfare is symmetric around the neutral point or (as discussed elsewhere in the comments) that their welfare range is 0.33 that of humans. I'm also sympathetic to the comment that was somewhat skeptical regarding the expected marginal impact of best-guess future advocacy. 

However I agree you may well be right that for a broad range of values, improving animal welfare (even if already positive, which I was too focused on) is more cost-effective than GW top charities, and this is an important point. I personally would find it even more informative and convincing to see some illustrative sets of parameter values along the lines of your "to get a ratio of 1" exercise (which I thought was a nice touch). What is the most plausible combination of values leading to a ratio of 1? However I realize it's not fair to ask you to do that - you've already put in a lot of useful work here.

As you acknowledge, an extremely broad set of parameters will lead to the conclusion that we should be eating more chickens rather than fewer. Of course the Humane League doesn't see it that way, and in general I would find animal-welfare advocates much more compelling if they didn't seem to always also push for veganism; imho it makes them sound ideological rather than evidence-driven. [You won't be surprised to hear that I'm not vegan - however I would happily vote for more humane animal farming regulations (if there were a sufficiently high probability of being pivotal...), and I'm open to being convinced to donate to charities that focus solely on improving conditions.] 

Likewise you say you yourself are open to the outcome that we should be eating more factory-farmed animals rather than fewer, which I appreciate. [although I note that in your post you refer without caveats to the "negative utility of farmed chickens"]  Given that as we've seen many plausible assumptions in your model would lead to such a conclusion, would you suggest that your framework implies that anyone believing something like those values (as I do) should in fact eat chickens and actively encourage all their friends to do so? I ask this not simply to play devil's advocate (esp since I sincerely believe in that position myself: everyone please eat more chickens!) but to continue to stress-test a bit on how seriously the model is meant to be taken with respect to any concrete conclusions.

sincerely believe in that position myself: everyone please eat more chickens!

What is the case for eating more chickens?

If farmed chickens plausibly have overall net positive lives (per the discussion above), and if you're something like a total utilitarian, then you should want more of them to exist; hence eat more in order to at least weakly increase demand / production.

Alternately, if you think it's very difficult to know for sure whether chickens have net positive lives or not, and you enjoy the taste of chicken, then that's another case for eating more of them.

Thanks for the question, mlovic, and welcome to the EA forum! Thanks for clarifying, Julian.

If farmed chickens plausibly have overall net positive lives (per the discussion above), and if you're something like a total utilitarian, then you should want more of them to exist

This might be true, but not necessarily so. I strongly endorse expected total hedonistic utilitarianism (classical utilitarianism), but would not be confident about increasing the consumption of factory-farmed animals even if they had good lives (although my best guess is that chickens do not):

  • Conditional on factory-farmed animals having good lives, one should arguably guess wild animals also have good lives. Consequently, since the scale of the welfare of wild animals is much larger than that of factory-farmed animals, one should do what increases the welfare of wild animals. So, because animal foods require much more land than plant-based foods, one would still want to decrease the consumption of factory-farmed animals.
  • Even if one thinks wild animals have bad lives, or is mostly agnostic about it, the increase in the welfare of factory-farmed animals may be outweighted by other negative effects. For example, I think enslaving people with good lives would be bad today[1] under classical utilitarianism, as it would erode impartiality by implicitly attributing a lower welfare range than justified to the enslaved. Likewise for factory-farmed animals, although less so.

If one is not confident about factory-farmed and wild animals having good/bad lives (I am not), and thinks this is a crucial consideration due to not giving major weight to the 2nd bullet above, one should focus on learning more about that question (e.g. Welfare Footprint Project's research), or improving their lives (e.g. corporate campaigns for chicken welfare).

  1. ^

    And also in the past, but less bad, holding the quality of life of the enslaved constant.

Thanks for the constructive reply too!

claim that welfare is symmetric around the neutral point

I assumed the welfare range is symmetric around the neutral point, but this does not impact the cost-effectiveness of corporate campaigns in human-years per dollar. To illustrate, if I had supposed the welfare range goes from excruciating pain to something as good as hurtful pain is bad, the welfare range would become about 0.5 times as wide (in reality, a little over 0.5 times as wide). Consequently:

  • The improvement in chicken welfare (when broilers go from a conventional to a reformed scenario) as a fraction of the median welfare range of chickens would become 2 (= 1/0.5) times as large.
  • The intensity of the mean human experience as a fraction of the median welfare range of humans would become 2 times as large.

The cost-effectiveness of corporate campaigns in human-years per dollar is directly proportional to the ratio between the above, so it would not change.

or (as discussed elsewhere in the comments) that their welfare range is 0.33 that of humans

I agree:

An additional major uncertainty is the welfare range of chickens. I have used RP's median estimate, but the 5th and 95th percentile are 0.602 % (= 0.002/0.332) and 2.61 (= 0.869/0.332) times as large.

I still think this is not a major issue, but I can see there is some margin for reasonable disagreement. I might do a Monte Carlo simulation modelling everything as distributions one of these days.

FWIW, in a previous analysis, I estimated corporate campaigns can be anything between 4.36 % to 34.1 k times as effective as GW's top charities (5th to 95th percentiles). The reason for my 95th percentile being roughly 1 M times my 5th percentile is me having used a moral weight distribution with 95th percentile about 1 M times as large as the 5th percentile. In contrast, RP's 95th percentile is only 434 (= 0.869/0.002) times RP's 5th percentile, and my uncertainty about the intensity of the various types of pains is similar, which means the 5th and 95th percentile of the cost-effectiveness ratio can be guesstimated from 0.602 % and 2.61 times the median ratio. So, if I were to do a Monte Carlo simulation, I guess I would conclude corporate campaigns are something between 10 to 4 k times as effective as GW's top charities (around 2.5 OOMs of uncertainty, as the median welfare range). Maybe this goes down to 1 to 400 times if one is quite pessimistic about marginal cost-effectiveness.

Of course the Humane League doesn't see it that way, and in general I would find animal-welfare advocates much more compelling if they didn't seem to always also push for veganism; imho it makes them sound ideological rather than evidence-driven.

I sympathise with animal-welfare advocates pushing for veganism because the most common objections to it are pretty weak (e.g. animals do not feel pain, animals are less intelligent/powerful, and the lives of factory-farmed animals are roughly as good as those of free range animals).

Likewise you say you yourself are open to the outcome that we should be eating more factory-farmed animals rather than fewer, which I appreciate. [although I note that in your post you refer without caveats to the "negative utility of farmed chickens"]

To be honest, I had not realised it was so easy to get positive lives for chickens (I seem to remember that I played with the numbers in the Sheet, but I think I was focussing on the cost-effectiveness ratio). I have added to the post the following:

[I have used RP's median estimate, but the 5th and 95th percentile are 0.602 % (= 0.002/0.332) and 2.61 (= 0.869/0.332) times as large.] Furthermore, as Julian Jamison noted, assuming disabling pain is 10 (instead of 100) times as bad as hurtful pain leads to broilers in a conventional scenario having positive lives[4].

I think 100 is significantly more reasonable than 10, but thanks for noting this!

Given that as we've seen many plausible assumptions in your model would lead to such a conclusion, would you suggest that your framework implies that anyone believing something like those values (as I do) should in fact eat chickens and actively encourage all their friends to do so?

Since I am quite uncertain about whether consuming more animals is good/bad[1], I would probably focus on:

  • Informing people about what is involved (most people overestimate the welfare of factory-farmed animals, and might not want to continue eating them if they have low welfare, even if it is positive, and maybe it is good to have norms against creating beings with low positive welfare, even if the total view is right).
  • Pushing people towards eating factory-farmed animals with better lives (e.g. broilers in reformed scenarios instead of conventional ones), as opposed to promoting veganism.
  1. ^

    Actually, I have a draft about this. If you like, comments are welcome!

Thanks again - all very constructive / helpful. I've updated some of my beliefs (partly toward the scale of this issue, as you intended, but also toward current factory farming not being as bad as I would have guessed... although I admit most people probably know less about conditions than I did), and I hope you have as well.

The only place I wanted to specifically respond is to your comment that you "sympathise with animal-welfare advocates pushing for veganism because the most common objections to it are pretty weak" - this doesn't make sense to me. We should only advocate for positions where the strongest objections are weak, not where the most common objections (which might be terrible ones) are weak. Again, tbh, it sounds more ideological than evidence- or logic-based.

I took a quick look at your linked doc and it looks good (to me): there is truly a lot of uncertainty about both basic direct outcomes (do conventional factory chickens have net positive or negative lives?) and indirect ones (what is the impact on wild animals and what is their welfare? how does it affect [human] economic and moral growth?). I would also add that while we can be sure that there is some positive elasticity between "one person stops eating chickens" to "future chicken production is lower in expectation" we don't currently have any idea what that number is (I've looked into this somewhat carefully), so that's another huge level of uncertainty. Anyone claiming that they know that the 'right' answer is not to eat animals, including many EAs and animal charities, is stepping way beyond the actual state of knowledge.

"one person stops eating chickens" to "future chicken production is lower in expectation" we don't currently have any idea what that number is (I've looked into this somewhat carefully)

Can't we make informed estimates, even if they have wide ranges? We multiply the demand shift by  (based on equilibrium displacement models, or this), with long-run elasticity estimates from the literature.

(FWIW, I'm also sympathetic to diet change being net negative in the near term, mostly because of the impacts on wild invertebrates and maybe fish. So I mostly focus on welfare.)

I'm a professor of economics, but thanks for the link explaining elasticity :) 

The answer is no, we can't just do that, since those approaches assume nontrivial changes (and/or they assume everything is continuous, which the real world isn't). One plausible simple model of supermarket (or restaurant) purchasing behavior is that when observed demand goes above/below a certain threshold relative to predicted demand, they buy more/less of the input next cycle. From an individual point of view, the expected aggregate demand of other agents in any time period will be a Gaussian distribution (by the law of large numbers), and the threshold will be away from the mean (doesn't make sense to update every time), which implies that one's probability of being the marginal buyer at the threshold declines exponentially (not linearly, as it would be for macro-level shifts and as you are implicitly assuming). From the ACE link: "we can approximate the supply and demand curves in this region by straight lines" - no, you can't do that (for individual behavior) without substantive additional assumptions or a lot of legwork into how these decisions actually get made.

In any case I have no idea if that's the right model, because I haven't studied supermarket supply chain management. As far as I can tell (but I'd love to see this somewhere), nobody in either the econ lit or animal welfare lit has tried to do this at the level required to make what I would consider an informed estimate; we're not just talking about a factor of 2 or 3 here. That knowledge gap doesn't seem to stop the latter group from making very strong claims; they mostly don't even seem to understand or acknowledge the high uncertainty and strong assumptions.

This sounds like Budolfson's buffer model. Have you seen the response by McMullen and Halteman? They describe supply chains and management practices in the section "Efficient Responsive Supply Chains and Causal Efficacy".

Also, this short first-hand account for grocery stores in an older article from the EA community on the issue, quoted from a comment on a post in an EA Facebook group on the issue.

I agree that it's probably true most people don't know the right reasons to believe that their individual purchase decisions make much difference on average, because most people know basically nothing about supply chain management.

I had seen some of this, but not the specific paper (ungated) by McMullen & Halteman - thanks!

First of all note that the two sources you cite directly contradict one another: the first-hand anecdotal account says there is essentially no meat waste even in very small groceries, while M&H (p.12) say there is a modest constant unavoidable waste that is in fact higher in smaller / local stores than for big outfits. Indeed M&H are internally inconsistent: they say that the market is highly competitive (although they only give a very incomplete reference for this on p.14, which I couldn't find any trace of; my googling found this source suggesting a net profit margin for farming/agriculture of 5.7%, which is middling - better than aerospace/defense or healthcare), but then they also state (p.23) that larger firms have up to 60% lower costs than smaller ones -- so how do the latter survive if the industry is so competitive? All of these are bad signs right off the bat.

Second note that none of these sources actually do any data analysis or try to examine original data about the markets or supply chains; they are armchair papers. My whole point is that depending on which of several reasonable assumptions one makes, different conclusions will be drawn. The only way to adjudicate this is to actually figure out what's going on in the real world, and neither of these sources attempts to do that. Hint: neither of them gives an empirically-derived concrete estimate for individual-level elasticity.

Third (to finally answer your question!), no my hypothetical model is not the same as the way they are using the term "buffer" (which seems to be more about maintaining a minimum level of excess in the system; mine is simply about the optimal tradeoff between stockouts vs excess/waste). For instance M&H say (p.25) "if there is some probability (1/n) that any given purchase will occur on a threshold, then the threshold action will trigger a reduction in production of around n units, yielding an expected impact equal to 1" (and from the reducing suffering page: "The probability that any given chicken is the chicken that causes two cases instead of three to be purchased is 1/25"). Well yes - if it's linear then the expected effect is the same order of magnitude as the input. My model was precisely one where the probability is plausibly not linear: in any given cycle, total sales are much more likely to be near the mean than near the threshold, so every individual would correctly believe that their own actions are very unlikely to change anything, which is not inconsistent with the (obviously correct) claim that large changes in demand are roughly linear and do influence things according to whatever macro-level elasticity has been estimated for chickens.

Or my 30-second model might be wrong - I'm not claiming it's correct. I'm claiming that we don't know, and the fact that none of these sources seems to have even considered it (or any other ones), and don't even realize the nature of the assumptions they're making, and nevertheless draw such strong conclusions, is again a bad sign.

Second note that none of these sources actually do any data analysis or try to examine original data about the markets or supply chains; they are armchair papers. My whole point is that depending on which of several reasonable assumptions one makes, different conclusions will be drawn. The only way to adjudicate this is to actually figure out what's going on in the real world, and neither of these sources attempts to do that. Hint: neither of them gives an empirically-derived concrete estimate for individual-level elasticity.

This seems fair and seems like the strongest argument here. Even M&H only say they "briefly sketch the contours of a positive argument for consumer efficacy".

 

While I think this doesn't undermine your point that people could come to reasonable differing conclusions about this case, it's worth pointing out the same is true about counterfactuals for basically all charity and altruistic work based on similar arguments, so this case doesn't seem categorically special. Some level of guesswork is basically always involved, although to different degrees, and levels of "robustness" can differ:

  1. GiveWell has estimates for the value of counterfactual spending by other actors, but it mostly only reflects government actors, plus the Global Fund. What about Open Phil and smaller donors? (Maybe they can be ignored based on Open Phil's own statements, and assuming smaller donors don't pay attention to these funding levels, and so don't respond.) Some of the numbers they do use are also just guesses. They go further than basically anyone else, but is it far enough? How much less cost-effective could they be?
  2. For individuals doing altruistic work, if they didn't do it (e.g. they didn't take the job), what would others have done differently, and with what value? ("Replaceability".)
  3. There are other effects on things we don't or can't measure. Does the charity undermine governance and do harm in the long run as a result? What about the effects on nonhuman animals, farmed and wild? What about the potential impacts much further into the future, through economic growth, climate change or space colonization? This gets into cluelessness and the McNamara fallacy.

Yes all fair, and I'd say it goes beyond counterfactuals. I'm not sure people fully realize how sensitive many conclusions are to all sorts of assumptions, which are often implicit in standard models. I am on record disagreeing strongly with John Halstead about the likely cost-effectiveness of advocating for economic growth, and I feel similarly about much of the longtermist agenda, so this isn't specific to animals. My personal sense is that if you can save an existing human life for a few thousand dollars (for which the evidence is very clear, although point taken that the marginal impact isn't definitively pinned down - however I'd guess within a factor of two,), that's an extremely high bar to overcome.

First of all note that the two sources you cite directly contradict one another: the first-hand anecdotal account says there is essentially no meat waste even in very small groceries, while M&H (p.12) say there is a modest constant unavoidable waste that is in fact higher in smaller / local stores than for big outfits.

Fair. I think the anecdotal account is a limiting case of M&H where the waste is very close to 0, though, so the arguments in M&H would apply to the anecdote. M&H's argument doesn't depend on there being modest constant unavoidable waste rather than essentially none.

 

Indeed M&H are internally inconsistent: they say that the market is highly competitive (although they only give a very incomplete reference for this on p.14, which I couldn't find any trace of; my googling found this source suggesting a net profit margin for farming/agriculture of 5.7%, which is middling - better than aerospace/defense or healthcare), but then they also state (p.23) that larger firms have up to 60% lower costs than smaller ones -- so how do the latter survive if the industry is so competitive? All of these are bad signs right off the bat.

This doesn't show they're internally inconsistent.

They probably meant the market is highly competitive in absolute terms, not among the very most competitive markets in the US. The argument they make isn't meant to depend on the relative competitiveness of the industry among industries, and it wouldn't be valid if it did.

Small farms can survive by product differentiation and competing in different submarkets. They can sell niche, specially labelled/described products, like organic, free range or locally raised, and they can charge premiums this way. They can sell in different places, like farmers markets, to small local grocers or to restaurants trying to appear more responsible/ethical, and charge more this way. Broiler farms producing fewer than 100,000 broilers/year only made up around 5% of the market in 2001 (Fig 2), so it's pretty plausible and I'd guess it's the case that small broiler farms with much higher production costs sell differentiated products.

I wasn't gesturing toward the relative competitiveness because it's important per se (you're right that it isn't) but rather as a way to gauge absolute competitiveness for those who don't already know that a net profit margin of 5.7% isn't bad at all. My intuition is that people realize that both defense and healthcare firms make decent profits (as they do) and hence that this fact would help convey that farmers (whether large or small; and if your point is that they can differentiate themselves and do some monopolistic competition then you're already on my side vs M&H) are not typically right on the edge of survival.

However I don't personally think the level of competition is crucial to anything here. M&H believe that it's necessary for their argument (in the abstract they say their case rests on it), so I was pointing out that (a) it's actually not that competitive; and (b) if they do think it's truly competitive (i.e. not differentiated) then that is indeed inconsistent with their own claim on p.23, which is a bad sign for their analysis.

My main point (which you don't seem to have responded to) remains that these are all conceptual arguments making various particular assumptions rather than actually trying to estimate an individual-level impact with a combination of a concrete well-defined model and empirics.

The edge of survival is not the only relevant threshold here. Chicken farmers don't own the birds they raise and only raise them when given a contract, so it's not entirely their choice whether or not and when they raise any chickens. From M&H:

Instead, the threshold-triggered event is a particular grower’s failure to get a contract to raise birds at all, or a delay in the next shipment of birds, a switch to a different type of agriculture, or a rancher’s choice to sell her land to a developer.

And even if their net profit margins were 5.7% on average, many farms could still be on the edge of survival. Also from M&H:

Even in industries that are vertically integrated, like the market for chickens, “growers” often operate with heavy debt, barely above poverty, and parent firms give them only short-term contracts (J. MacDonald 2008).

From MacDonald, 2008:

Net farm income is the difference between gross farm income and operating expenses, and it amounts to 25-27 percent of gross farm income in each size class. Net farm income, however, varies widely among broiler operations, where a quarter of farms experience losses—negative net farm income. Poor productive performance may be one source of negative net income since, on average, operations with negative net farm income receive fees of 4.8 cents per pound, compared with 5.1 cent per pound for those with positive net income. Depreciation is a more important factor explaining differences in net income. On farms with negative net farm income, depreciation expenses account for 39 percent of gross income, on average, compared with 13 percent for other operations. Farms with recent major capital expenditures will usually record substantial depreciation expenses, often large enough to generate negative net farm incomes. Correspondingly, older operations with fully depreciated assets rarely report negative net incomes.

Furthermore, the 20th percentile of household income[1] across broiler farmers was $18,782 in 2011, according to the USDA, and so close to the poverty line at the time. However, the household income for chicken farmers is relatively high recently, in 2020 (USDA).

 

Also, about differentiation, I don't see what the existence of some small high-cost farms selling to small niche submarkets tells you about the behaviour or competitiveness of the conventional large farms, which account for almost all of the combined market. I don't think it's a case of monopolistic competition; these are just a few separate submarkets, like free range and organic. Maybe those selling locally are acting nearly monopolistically, with the "local" label or by selling to farmers markets, but it also doesn't really matter, because they're selling to a tiny submarket and their supply is very limited. If a kid sets up a lemonade stand in their neighbourhood and sells lemonade above grocery store prices, you wouldn't conclude from this that an individual lemonade company can set higher prices for grocery stores (or distributors?), where almost all of the lemonade is bought, without being pushed out of the market.

  1. ^

    The USDA's definition:

    Household income measures the cash income flowing to a household and available for expenditures during a year. For farmers, household income combines the income that the household receives from off-farm activities with the income that the household receives from the farm business, net of expenses and payments to other stakeholders in the business.

Third (to finally answer your question!), no my hypothetical model is not the same as the way they are using the term "buffer" (which seems to be more about maintaining a minimum level of excess in the system; mine is simply about the optimal tradeoff between stockouts vs excess/waste). For instance M&H say (p.25) "if there is some probability (1/n) that any given purchase will occur on a threshold, then the threshold action will trigger a reduction in production of around n units, yielding an expected impact equal to 1" (and from the reducing suffering page: "The probability that any given chicken is the chicken that causes two cases instead of three to be purchased is 1/25").

Sorry, I could have been more explicit in my comment. I wasn't referring to the rest of the Reducing Suffering article, and I didn't mean that any of that article referred to your model. M&H refer to a model similar to yours (Budolfson's buffer model), but not in the section that I referred to (and from which you quote). What I meant is that both propose more plausible models of markets (more plausible based on observations of how grocery stores behave), and I was pointing to those alternative proposals.

M&H summarizes the main takeaway from Budolfson's buffer model:

If a person is facing a decision with this kind of uncertainty, and they have good information about the probability of being near a threshold, this can dramatically alter the expected impact calculation. (...) Similarly, if a person knew that their purchase of a chicken was not near the threshold, they could, he argues, purchase the chicken without worry about consequences for animals.

Budolfson is correct in claiming that expected impact calculations cannot always assume that an action, on the margin, would be the same as the average effect of many such actions. The standard expected utility response given by Singer and Kagan can depend crucially on the kind of information that a person has about the location of thresholds.

This is an illustration of Budolfson's buffer model, directly from Budolfson, 2018:

Richard makes paper T-shirts in his basement that say ‘HOORAY FOR CONSEQUENTIALISM!’, which he then sells online. The T-shirts are incredibly cheap to produce and very profitable to sell and Richard doesn’t care about waste per se, and so he produces far more T-shirts than he is likely to need each month, and then sells the excess at a nearly break-even amount at the end of each month to his hippie neighbor, who burns them in his wood-burning stove.Footnote10 For many years Richard has always sold between 14,000 and 16,000 T-shirts each month, and he’s always printed 20,000 T-shirts at the beginning of each month. Nonetheless, there is a conceivable increase in sales that would cause him to produce more T-shirts—in particular, if he sells over 18,000 this month, he’ll produce 25,000 T-shirts at the beginning of next month; otherwise he’ll produce 20,000 like he always does. So, the system is genuinely sensitive to a precise tipping point—in particular, the difference between 18,000 purchases and the ‘magic number’ of 18,001.

Presumably there could also be a conceivable decrease in sales that would cause Richard to produce fewer T-shirts, too. Richard has a historical monthly demand range that serves essentially the same purpose as your predicted demand, with thresholds for setting alternative future procurement/production decisions far enough away from the centre of the historical range, or in your case, predicted demand.

EDIT: so your last paragraph seems wrong:

I'm claiming that we don't know, and the fact that none of these sources seems to have even considered it (or any other ones), and don't even realize the nature of the assumptions they're making, and nevertheless draw such strong conclusions, is again a bad sign.

Interesting - thanks for the extra info re Budolfson. I did in fact read all of M&H, and they give two interpretations of the buffer model, neither of which is related to my model, so that's what I was referring to. [That's also what I was referring to in my final paragraph: none of the sources you cited on that side of the causal efficacy argument seems to have considered anything like my model, which remains true given my current knowledge.]  In fact if Budolfson was saying something more like my model, which does seem to be the case, then that's an even worse sign for M&H because they must not have understood it.

The paragraph you quote from Budolfson is indeed more similar to my model, except that in my case the result follows from profit-maximizing behavior (in a competitive industry if you like!) rather than ad hoc and unusual assumptions. 

Suppose that I consider a threshold (for increasing or decreasing production next cycle) right at the mean of expected sales (15,000 in the example): half the time I'll stockout and have disappointed customers; half the time I'll have extra stock and have to sell it on a secondary market, or give it away, or waste it. Which is worse for business? Plausibly stocking out is worse. So my threshold will be higher than the mean, reducing the probability of stocking out and increasing the prob of excess. The optimal level will be set just so that at the margin, the badness of stocking out (larger) multiplied by the prob of stocking out (smaller) will exactly offset the badness of excess times the prob of excess. Because it is above the mean, which is in fact the true best-guess state of the world (ignoring any individual consumer), and because the distribution around the mean will plausibly be Gaussian (normal), which declines exponentially from the mean - not linearly! - every individual consumer should rationally believe that their decision is less than 1/n likely to be taking place at the threshold. QED.

I'm not sure what you mean by M&H not understanding Budolfson. They give a brief overview of the model, but the section from M&H I referred to ("Efficient Responsive Supply Chains and Causal Efficacy") describes the market as they understand it, in a way that's not consistent with Budolfson. The implicit reply is that Budolfson's model does not match their observations of how the market actually works.

I think how they'd respond to your model is:

  1. stores do use explicit demand predictions to decide procurement,
  2. they are constantly making new predictions, 
  3. these predictions are in fact very sensitive to recent individual purchase decisions, and actually directly so.

Suppose the store makes stocking decisions weekly. If demand is lower one week than it would have otherwise been, their predictions for the next week will be lower than they would have otherwise been. Of course, there's still a question of how sensitive: maybe they give little weight to their actual recent recorded purchases[1] relative to other things, like others' market forecasts or sales the same time in past years.[2] But M&H would contend that actually they are very sensitive to recent purchases, and I would guess that's the case, too, because it probably is one of the most predictive pieces of information they can use, and plausibly the most predictive. They don't provide direct estimates of the sensitivity based on empirical data and maybe they don't back these claims with strong enough evidence at all (i.e. maybe stores don't actually usually work this way), and it's fair to point out these kinds of holes in their arguments if someone wants to use their paper to make a strong case.


Here are relevant quotes:

For example, modern grocery stores have check-out procedures that track the sale of each product and automatically order replacements from the parent companies. Even in industries that are not vertically integrated, standard information technology allows firms to track sales in great detail, down to individual transactions (Salin 2000). In addition, these companies track the rates of orders to optimize shipping and refrigeration times and to minimize waste. (...) In this kind of system, the large distributors that contract with farms actually do know the rate at which chickens are being purchased throughout their network.

(...)

Given this description of the way these markets function, we can now describe the causal chain that connects an individual’s purchase to a farmer’s production decision. When a person decides to stop purchasing chickens, the result is that their local grocery store automatically starts ordering chickens more slowly, to reflect the decreased rate of sale. The distributor (perhaps Chickens R Us) will automatically adjust their shipments of chickens to that store. Since some shipments will require preset bundles of chickens, there will be a threshold at which a delivery of meat comes a day later, to reflect the slower demand. This “threshold” does not mean, however that the information going down the supply chain is less precise. As Chickens R Us is managing their supply of chickens in the distribution network, they are also managing the rate at which they send contracts of birds to their “growers” and the number of growers that get contracts.

I would correct the one sentence to "When a person decides to stop purchasing chickens, the result is that their local grocery store automatically starts ordering chickens more slowly than they otherwise would have, to reflect the lower than otherwise rate of sale."

  1. ^

    Or, indirectly, through leftover stocks or stockouts.

  2. ^

    Although eventually that should get picked up.

I still haven't read Budolfson, so I'm not claiming that M&H misinterpret him. As I said, I did read their entire paper, and in the section specifically about him they describe two interpretations of "buffer", neither of which matches my model. So if his model is similar to mine, they got it wrong. If his model is different than mine, then they don't seem to have ever considered a model like mine. Either way a bad sign.

Everything you write about how you think they might respond to me (i.e. your three bullet points and the subsequent paragraph) is 100% consistent with my model and doesn't change any of its implications. In my model stores use predicted demand and can update it as often as they want. The point is that purchasing is in bulk (at least at some level in the supply chain); therefore there is a threshold; and the optimal threshold (every single time) will be chosen to be away from the mean prediction. This can still be extremely sensitive, and may well be. [Apologies if my brief descriptions were unclear, but please do take another look at it before responding if you don't see why all this is the case.]

To the final point, yes of course if someone decides to stop purchasing then the store [probabilistically] starts ordering fewer chickens [than otherwise]; I didn't disagree with that sentence of theirs, and it is also 100% consistent with my model. The question is the magnitude of that change and whether it is linear or not, crucial points to which they have nothing to contribute.

EDIT: I did misunderstand at this point, as you pointed out in your reply.

 

Ok, I think I get your model, but I don't really see why a grocery store in particular would follow it, and it seems like a generally worse way to make order decisions for one. I think it's more plausible for earlier parts of the supply chain, where businesses may prefer to produce consistent volumes, because there are relevant thresholds (in revenue) for shutting down, downsizing, expanding and entering the market, and it's costly to make such a decision (selling/buying capital, hiring/firing staff) only to regret it later or even flip-flop.[1] It takes work to hire someone, so hiring and firing (in either order) is costly. Capital assets lose value once you purchase or use them, so buying and selling (in either order) is costly. If changes in a business' production levels often require such a decision, that business has reason to try to keep production more consistent or stick with their plans to avoid accumulating such costs. But not all changes to production levels require such decisions.

(I don't mean to imply you don't understand all of the above; this is just me thinking through it, checking my understanding and showing others interested.)

I don't think a grocery store has to adjust its capital or staff to order more or less, or at least not for the vast majority of marginal changes in order size. Same for distributors/wholesalers.

I'm not sure about broiler farms. They'd sometimes just have to wait longer for a contract (or never get one again), or maybe they'd get a smaller contract and raise fewer broilers (the market is contract-based in the US, and the farms don't own the broilers[2]), so it often just wouldn't be their decision. But on something like your model, if a farm was planning to enter the market or expand, and contracts or revenues (or market reports) come only slightly worse than expected (still above the threshold in your model, and which is far more likely than coming below the threshold), they'd enter/expand anyway. For farms not planning to expand/enter the market, maybe they'd even take on a contract they don't expect to pay for its variable costs, just to get more favour from the companies contracting them in the future or to push out competitors. Or, just generally, the contracts would very disproportionately be above their thresholds for shutdown, as they expect them to be. Also, many individual farmers are probably subject to the sunk cost fallacy.

Then there are the integrator/processor companies like Tyson that contract the farms. A small number of companies control a large shares of this part of the supply chain, and they've been caught price-fixing (see here and here), which undermines the efficiency (and of course competitiveness) of the market. Below their predictions, maybe they'd want to keep giving farms contracts in order to keep them from shutting down or to keep them from switching to competitors, because it'll be harder/slower to replace them if demand recovers, or just to hurt competitors. Or, if they were already planning to expand production, but sales come in below expectation, they'd do it anyway for similar reasons.


Here's an example for a grocery store:

Suppose, to avoid stockouts (like you propose they should), as a rule, they order 7 more units than (the expected value of) their predicted sales.

Suppose they would have predicted 123 sales for the next period had you not abstained. Because you abstained, they instead predict 122. So, as a result of your abstention, they order 129 instead of 130, and you make a difference, at least at this level.

Now, maybe they need to order in specific multiples of units. Say they need to order in multiples of 10, and they order the minimum multiple of 10 that's at least 7 over what they predict.

In the above case, your abstention makes no difference, and they would order 130 either way, but that's just one case. The threshold to order 10 fewer is when the prediction modulo 10 would have been 4 and your abstention drops it below that.[3] If you look at a randomly sampled period where they need to order, there's not really any reason to believe that their prediction modulo 10 will be especially unlikely to be 4 compared to any of the other digits.[4]

 

  1. ^
  2. ^

    Also:

    Broiler production contracts add another risk aside from the way in which compensation is determined. Traditionally, broiler contracts have not required strong commitments by integrators. In 2006, about half of broiler contracts were “flock to flock”; that is, the integrator made no specific commitment to provide birds beyond the current flock’s placement. Those contracts that specified a longer duration (usually 1 to 5 years) rarely committed the integrator to a specified number of birds or flocks in a year.

  3. ^

    For their prediction x, if x mod10=4, then they order x+16. If x mod10=3, then they order x+7.

  4. ^

    I guess one way would be if they have sufficiently consistent purchases and choose a supplier based on the multiple to get their prediction modulo the multiple away from the threshold. I think it's very unlikely they'd switch suppliers just to get their predictions in a better spot with respect to multiples.

Hi - thanks again for taking more time with this, but I don't think you do understand my model. It has nothing to do with capital assets, hiring/firing workers, or switching suppliers. All that it requires is that some decisions are made in bulk, i.e. at a level of granularity larger than the impact of any one individual consumer. I agree this is less likely for retail stores (possibly some of them order in units of 1? wouldn't it be nice if someone actually cared enough to look into this rather than us all arguing hypothetically...), but it will clearly happen somewhere back up the supply chain, which is all that my model requires.

Your mistake is when you write "Say they need to order in multiples of 10, and they order the minimum multiple of 10 that's at least 7 over what they predict." That's not what my model predicts (I think it's closer to M&H's first interpretation of buffers?), nor does it make economic sense, and it builds in linearity. What a profit-maximizing store will do is to balance the marginal benefit and marginal cost. Thus if they would ideally order 7 extra, but they have to order in multiples of 10 and x=4 mod10, they'll order x+6 not x+16 (small chance of one extra stock-out vs large chance of 10 wasted items). They may not always pick the multiple-of-10 closest to 7 extra, but they will balance the expected gains and losses rather than using a minimum. From there everything that I'm suggesting (namely the exponential decline in probability, which is the key point where this differs from all the others) follows.

And a quick reminder: I'm not claiming that my model is the right one or the best one, however it is literally the first one that I thought of and yet no one else in this literature seems to have considered it. Hence my conclusion that they're making far stronger claims than are possibly warranted.

(I've edited this comment, but the main argument about grocery stores hasn't changed, only some small additions/corrections to it, and changes to the rest of my response.)

Thanks for clarifying again. You're right that I misunderstood. The point as I now understand is that they expect the purchases (or whatever they'd ideally order, if they could order by individual units) to fall disproportionately in one order size and away from each threshold for lower and higher order sizes, i.e. much more towards the middle, and they've arranged for their order sizes to ensure this.

I’ll abandon the specific procedure I suggested for the store, and make my argument more general. For large grocery stores, I think my argument at the end of my last comment is still basically right, though, and so you should expect sensitivity, as I will elaborate further here. In particular, this would rule out your model applying to large grocery stores, even if they have to order in large multiples, assuming a fixed order frequency.

Let’s consider a grocery store. Suppose they make purchase predictions  (point estimates or probability distributions), and they have to order in multiples of , but I’ll relax this assumption later. We can represent this with a function  from predictions to order sizes so that , where  is an integer-valued function.  can be the solution to an optimization problem, like yours. I’m ignoring any remaining stock they could carry forward for simplicity, but they could just subtract it from  and put that stock out first. I’m also assuming a fixed order frequency, but M&H mention the possibility of "a threshold at which a delivery of meat comes a day later". I think your model is a special case of this, ignoring what I'm ignoring and with the appropriate relaxations below.

I claim the following:

  1. Assuming the store is not horrible at optimizing,  should be nondecreasing and scale roughly linearly with . What I mean by “roughly linearly with ” is that for (the vast majority of possible values of) , we can assume that , and that values of  where , i.e. the thresholds, are spaced roughly  apart. Even if different order sizes didn't differ in multiples of some fixed number, something similar should hold, with spacing between thresholds roughly reflecting order size differences.
  2. A specific store might have reason to believe their predictions are on a threshold much less than  of the time across order decisions, but only for one of a few pretty specific reasons:
    1. They were able to choose  the first time to ensure this, intentionally or not, and stick with it and  regardless of how demand shifts.
    2. The same supplier for the store offers different values of  (or the store gets the supplier to offer another value of ), and the store switches  or uses multiple values of  simultaneously in a way that avoids the thresholds. (So  defined above isn’t general enough.)
    3. They switch suppliers or products as necessary to choose  in a way that avoids the thresholds. Maybe they don’t stop offering a product or stop ordering from the same supplier altogether, but optimize the order(s) for it and a close substitute (or multiple substitutes) or multiple suppliers in such a way that the thresholds are avoided for each. (So  defined above isn’t general enough.)

If none of these specific reasons hold, then you shouldn’t expect to be on the threshold much less than  of the time,[1] and you should believe , where the expectation is taken over your probability distribution for the store’s prediction .

How likely are any of these reasons to hold, and what difference should they make to your expectations even if they did?

The first reason wouldn’t give you far less than  if the interquartile range of their predictions across orders over time isn’t much smaller than , but they prefer or have to keep offering the product anyway. This is because the thresholds are spaced roughly  apart,  will have to cross thresholds often with such a large interquartile range, and if  has to cross thresholds often, it can’t very disproportionately avoid them.[2]

Most importantly, however, if  is chosen (roughly) independently of , your probability distribution for  for a given order should be (roughly) uniform over 0,..., ,[3] so  should hit the threshold with probability (roughly) . It seems to me that  is generally chosen (roughly) independently of . In deciding between suppliers, the specific value of  seems less important than the cost per unit, shipping time, reliability and a lower value of .[4] In some cases, especially likely for stores belonging to large store chains, there isn’t a choice, e.g. Walmart stores order from Walmart-owned distributors, or chain stores will have agreements with the same supplier company across stores. Then, having chosen a supplier, a store could try to arrange for a different value of  to avoid thresholds, but I doubt they’d actually try this, and even if they did try, suppliers seem likely to refuse without a significant increase in the cost per unit for the store, because suppliers have multiple stores to ship to and don’t want to adjust  by the store.

Stores similarly probably wouldn’t follow the strategies in the second and third reasons because they wouldn’t be allowed to, or even if they could, other considerations like cost per unit, shipping time, reliability and stocking the same product would be more important. Also, if the order quantities vary often enough between orders based on such strategies, you’d actually be more likely to make a difference, although smaller when you do.

 

So, I maintain that for large stores, you should believe .


 

And a quick reminder: I'm not claiming that my model is the right one or the best one, however it is literally the first one that I thought of and yet no one else in this literature seems to have considered it. Hence my conclusion that they're making far stronger claims than are possibly warranted.

Fair. I don't think they should necessarily have considered it, though, in case observations they make would have ruled it out, but it seems like they didn't make such observations.

but it will clearly happen somewhere back up the supply chain, which is all that my model requires.

I don't think this is obvious either way. This seems to be a stronger claim than you've been making elsewhere about your model. I think you'd need to show that it's possible and worth it for those at one step of the supply chain to choose  or suppliers like in a way I ruled out for grocery stores and without making order sizes too sensitive to predictions. Or something where my model wasn't general enough, e.g. I assumed a fixed order frequency.

  1. ^

    It could be more than , because we’ve ruled out being disproportionately away from the threshold by assumption, but still allowed the possibility of disproportionately hitting it.

  2. ^

    For realistic distributions of  across orders over time.

  3. ^

    I would in fact expect lower numbers within 0, ...,  to be slightly more likely, all else equal. Basically Benford's law and generalizations to different digit positions. Since these are predictions and people like round numbers, if  is even or a multiple of 5, I wouldn't be surprised if even numbers and multiples of 5 were more likely, respectively.

  4. ^

    Except maybe if the minimum  across suppliers is only a few times less than , closer to  or even greater, and they can’t carry stock forward past the next time they would otherwise receive a new shipment.

We should only advocate for positions where the strongest objections are weak, not where the most common objections (which might be terrible ones) are weak.

I agree. Sorry for not being clear in my previous reply. By "I sympathise with animal-welfare advocates pushing for veganism", I meant that I can see from where they are coming, not that I rationally endorse veganism.

Hey Vasco, you make lots of good points here that are worth considering at length. These are topics we've discussed on and off in a fairly unstructured way on the research team at FP, and I'm afraid I'm not sure what's next when it comes to tackling them. We don't currently have a researcher dedicated to animal welfare, and our recommendations in that space have historically come from partner orgs.

Just as context, the reason for this is that FP has historically separated our recommendations into three "worldviews" (longtermism, current generations, and animal welfare). The idea is that it's a lot easier to shift member grantmaking across causes within a worldview (e.g. from rare diseases to malaria, for instance) than across worldviews (e.g. to get people to care much more about chickens). The upshot of this, for better or for worse, is that we end up spending a lot of time prioritizing causes within worldviews, and avoiding the question of how to prioritize across worldviews.

This is also part of the reason we don't have a dedicated animal welfare researcher — we haven't historically moved as much money within that worldview as within our others. But it's actually not sure which way the causality flows in that case, so your post is a good nudge to think more seriously about this, as well as the ways we might be able to incorporate animal welfare considerations into our GHD calculations, worldview separations notwithstanding.

Thanks for sharing your thought, Matt!

For balance and completeness… Would it make sense to add something (or another piece) considering the impact of chicken welfare improvement funding on wild animal welfare?

Hi David,

Thanks for the comment. I think that would make sense (in another piece)! Somewhat relatedly, I wrote that:

Regarding the impact of human diet on animal welfare (of both farmed and wild animals), Michael St. Jules suggested Matheny 2005, this and these posts from Brian Tomasik, this post from Carl Shulman, and Fischer 2018.

I liked this post. It was thought provoking.

I just wanted to note that you are correct in highlighting the “human” part in my post on the capability approach. To me, capabilities are the best way to think about human welfare but some variant of utilitarianism is the best way to think about the welfare of (most?) animals, but I’ve no good way to exchange between those and I find that unsatisfying.

Thanks, Ryan!

Interestingly, I have recently listened to Martha Nussbaum on the Clearer Thinking podcast, and it looks like her book Justice for Animals: Our Collective Responsibility attempts to extend the capability approach to non-human animals.