Summary

In this article, I estimate the cost-effectiveness of five Anima International programs in Poland: improving cage-free and broiler welfare, blocking new factory farms, banning fur farming, and encouraging retailers to sell more plant-based protein. I estimate that together, these programs help roughly 136 animals—or 32 years of farmed animal life—per dollar spent. Animal years affected per dollar spent was within an order of magnitude for all five evaluated interventions.

I also tried to estimate how much suffering each program alleviates. Using SADs (Suffering-Adjusted Days)—a metric developed by Ambitious Impact (AIM) that accounts for species differences and pain intensity—Anima’s programs appear highly cost-effective, even compared to charities recommended by Animal Charity Evaluators.

However, I also ran a small informal survey to understand how people intuitively weigh different categories of pain defined by the Welfare Footprint Institute. The results suggested that SADs may heavily underweight brief but intense suffering. Based on those findings, I created my own metric DCDE (Disabling Chicken Day Equivalent) with different weightings. Under this approach, interventions focused on humane slaughter look more promising, while cage-free campaigns appear less impactful. These results are highly uncertain but show how sensitive conclusions are to how we value different kinds of suffering.

My estimates are highly speculative, often relying on subjective judgments from Anima International staff regarding factors such as the likelihood of success for various interventions. This introduces potential bias. Another major source of uncertainty is how long the effects of reforms will last if achieved. To address this, I developed a methodology to estimate impact duration for chicken welfare campaigns. However, I’m essentially guessing when it comes to how long the impact of farm-blocking or fur bans might last—there’s just too much uncertainty.

Background

In 2024, Anima International contracted Nuño Sempere and I (Saulius Šimčikas) to assess their cost-effectiveness in Poland, partly to inform internal decisions. Both me and Nuño were compensated for the evaluation. However, I didn’t feel pressure to reach positive conclusions. Anima International communicated to us that they would like to publish the results no matter the results for transparency and accountability.

Results

You can see the estimates in this spreadsheet. I will not go through all the estimates in this article because they are presented in the spreadsheet. Here are the results:

In this article, K stands for one thousand (1,000), M stands for one million (1,000,000), and B stands for one billion (1,000,000,000). 

In the table above, we can see that all interventions help multiple animals per dollar spent. Also, the number of farmed animal years impacted per dollar spent in the future (the bolded column) is within one order of magnitude for all evaluated interventions. To me, this was very surprising. It weakly suggests that there are multiple cost-effective approaches in animal advocacy.

Note that the number of animals impacted by retailer plant-based commitments would be at least seven times higher if wild-caught fishes were included.^[1] However, reducing demand for wild-caught fish is morally ambiguous.

To account for factors in intensity of suffering and sentience, I translated these results into my own metric, DCDE (Disabling Chicken Day Equivalent). I also converted some results into the SAD (Suffering-Adjusted Days) metric introduced by AIM, which puts a lot less weight on short but more intense suffering. As you can see in the table below, the relative impact of cage-free and broiler campaigns depends a lot on which metric is used.

Later in this article, I show that Anima International programs in Poland seem cost-effective compared to those evaluated by ACE and AIM, when using SADs as the metric. For example, during AIM’s most recent research round, their cost-effectiveness threshold was 30 SADs per dollar—equivalent to their estimate for cage-free work in the Middle East, one of the programs they aim to launch.

However, if we measure cost-effectiveness in DCDEs, which puts much more weight on short but intense suffering, then things like humane slaughter reforms look a lot more cost-effective. For example, based on Grilo (2024), but using my own pain intensities, the cost-effectiveness of Shrimp Welfare Project’s Humane Slaughter Initiative would be 25,000 DCDE per USD - over forty times higher than the estimated future cost-effectiveness of evaluated programs of Anima International in Poland.

Explanations of the programs

• Cage-free commitments: This program works to eliminate cages for egg-laying hens by securing and enforcing corporate commitments to switch to cage-free systems. Current and future cage-free work is about ensuring that companies keep their commitments.
• Broiler commitments: This campaign seeks corporate pledges to improve the welfare of chickens farmed for meat (broilers). Only a few commitments have been secured in Poland so far, and none have been implemented yet. Almost all of the expected future impact in my estimates comes from an assumed 3% chance that slow-growing broilers will eventually become the industry standard in Poland.
• Stop the Farms: This program supports local communities in opposing the construction of new chicken and pig farms, highlighting negative effects on nearby residents. It is mostly run by one staff member. Poland is especially suited to this strategy due to the scarcity of locations far from residential areas.
• Fur farms: This campaign aims to ban fur farming in Poland, where around 3.4 million minks are currently farmed. Past impact comes from possibly preventing some mink farms from being built by making it a more risky investment. A small part of the work is also focused on the EU ban of fur farming and fur imports, but I didn’t include the impact of EU policy work in my estimates.
• Retailer plant-based commitments: Part of Anima International’s food transformation program, this campaign encourages retailers to commit to increase plant-based protein sales to at least 20% of their total protein sales. The campaign is only launching now. Hence my estimates depend on the intuitions of Anima International staff about how successful it will be. The campaign is expected to require less than one full-time equivalent.

For context, here is the 2024 budget for Anima International in Poland:

Notes about the graph above:

• Overhead costs (leadership, management, HR, admin, IT) are allocated to each program in proportion to its share of the overall budget. All estimates in this article are based on costs that already include a proportional share of these overheads.
• Food system transformation includes retailer plant-based commitments and advocating for the government, as well as the private sector, to do things like support alternative proteins.
• Expenses for chicken farm investigations are included under the broiler and cage-free program budgets, as they directly support those campaigns.

Why these estimates are very uncertain

I worry that people will interpret these estimates as much more certain than they are. My estimates largely rely on subjective opinions as there isn’t enough objective data to inform all parameters. Many of these judgments are based on the intuitions of Anima International staff members, especially those of Pawel Rawicki, the CEO of the Polish branch. As far as I could tell, Pawel and the other staff members I consulted made a genuine effort to remain objective, and the organisation has a culture that encourages open-minded discussion.

However, relying on staff opinions inevitably introduces some degree of bias. Those working on these issues are often the most optimistic about their impact, which can stem from several factors:

• Selection bias: People skeptical about an intervention’s effectiveness are less likely to work on it, leaving those with more favorable views to provide assessments.
• Motivated reasoning: Staff may naturally want to see their work as meaningful and impactful, both for personal satisfaction and external credibility.
• Echo chamber effects: Working in a like-minded team can reinforce shared optimism, making critical perspectives less prominent.

Avoiding such biases would be very difficult as it would be very difficult for a researcher to independently assess all the parameters involved. I still think it is useful to see if a campaign looks competitively cost-effective even if some of the inputs are based on intuitions of people working on the campaign. Many programs might look ineffective no matter what intuitions we enter.

I divided estimates into past and future. When estimating the impact of past work, I assumed that all the work stopped at the end of 2024.^[2] The estimates of past efforts cover all past years. For the cage-free program, this starts with 2015. If I were to redo the estimates, I would focus on more recent years, which would reduce the cost-effectiveness of past cage-free campaigns.^[3] 

Stop the Farms estimate is only based on the year 2021 because collecting data for all the years would’ve been very time-consuming for the staff. I was told that it was a typical year.^[4] There was disagreement between staff members of Anima International about whether the campaign will increase or decrease in cost-effectiveness in the future, so I assumed that the cost-effectiveness will remain the same.

Future estimates of all programs are very uncertain. They show the expected value of the number of animals helped. For example, all the future impact of the fur farm campaign comes from the possibility of a fur farming ban. People working on this ban predicted that there is a 35% chance that such a ban will be achieved by the end of 2027. I think that estimates for broiler and retailer plant-based campaigns are even more subjective and uncertain. These campaigns might end up having very little impact, or be very impactful - we just don’t know yet. Hence, donating to these campaigns could be considered as hits-based giving.

It’s also not even certain that some of these programs do more good than harm to animals.

• Stop The Farms campaign could be just moving animal production to countries where improving animal welfare could be even more challenging.  If a chicken farm is blocked in Poland, another farm might be built in, say, Romania to meet the same demand—potentially with even lower welfare for chickens. A similar argument can be made about efforts to ban fur farms in Poland, though I think it’s slightly less convincing because fur farming is steeply declining worldwide.^[5]

• Stop The Farms program might also be making it more difficult to build new high welfare farms, making welfare improvements more difficult, even though they don’t campaign against the building of farms that are known to be high-welfare.
• Retailer plant-based commitments campaign tries to reduce animal production, which could increase wild animal suffering, and hence overall animal suffering.
• I listed many ways chicken welfare reforms could have a negative impact here. 

My estimates focus only on direct, quantifiable effects, excluding potentially significant indirect impacts such as:

• Shifts in public attitudes
• Movement-building effects
• Media and policy influence
• Effects on wild animal suffering
• Impact on fundraising
• Impact on larger goals like moral circle expansion or the abolition of factory farming
• Unpredictable “Black Swan” effects

These could plausibly outweigh the direct impacts but are too uncertain to quantify.

I also don’t analyse room for more funding. I estimate the total impact per dollar spent. The value of a marginal dollar donated to the organisation could be very different.

Animal welfare metric

As mentioned, I created a metric called “Disabling Chicken Day Equivalent” (DCDE). One DCDE is as bad as a chicken spending one day in disabling pain. Disabling pain here means the same thing as it means when used by The Welfare Footprint Institute. It is defined as follows:

Pain at this level takes priority over most bids for behavioral execution and prevents all forms of enjoyment or positive welfare. Pain is continuously distressing. Individuals affected by harms in this category often change their activity levels drastically (the degree of disruption in the ability of an organism to function optimally should not be confused with the overt expression of pain behaviors, which is less likely in prey species). Inattention and unresponsiveness to milder forms of pain or other ongoing stimuli and surroundings is likely to be observed. Relief often requires higher drug dosages or more powerful drugs. The term Disabling refers to the disability caused by ‘pain’, not to any structural disability.

Where possible, my estimates used the conclusions from The Welfare Footprint Institute. They have four categories of pain: annoying, hurtful, disabling, and excruciating. They don’t provide how to weigh these different categories of pain against each other. To determine what these weights should be, I made three very similar small informal surveys, asking people to trade-off different types of pain against each other. In total, I got 43 responses that I deemed to be usable. You can see one of the survey designs here, and the survey results and analysis here.

Here are the results using the DCDE metric:

Converting results to DCDE required making some very speculative guesses about the suffering of animals that were not evaluated by The Welfare Footprint Institute. For example, I assumed that one day of mink farming is 1.5 times as bad as one day of fast-growing broiler farming. This is only based on intuitions of a few people I talked to about this. However, this is just a placeholder, because if we estimate the value of preventing a mink from being farmed in DCDEs, then it is very sensitive to how much intense pain minks experience, which is something I know nothing about.

I also used ChatGPT to guess consumption, lifespan, and welfare of all animals consumed for food in order to guesstimate DCDEs for retailer plant-based commitments. This guesstimate is also very uncertain.

Comparison to SADs

After inventing my own metric of DCDEs, I found that in 2024, AIM introduced an analogous metric they use for their charity evaluations called SADs. I converted my weightings into SADs below:

As you can see, the results look very different. Cage-free campaigns look relatively much more cost-effective. This is because AIM puts a lot less weight on more intense types of pain. Tables below show what amounts of time each metric assumes to be equivalent to one second of excruciating pain, and one day of disabling pain, respectively.

Even though my choices are probably based on less research than the ones by AIM, it’s noteworthy that different researchers came to such different conclusions about what these weights should be. There was a lot of disagreement about how to weigh different types of pain amongst the 43 respondents to my surveys as well. For example, one of the questions was “Imagine you have to choose between one second of excruciating pain and a period of annoying pain. How long would the annoying pain need to last for you to feel indifferent between the two options?” Answers included “like 10 minutes,” “10,000 years,” and “an indefinite period of time”.^[6] However, the SAD metric assumes that it’s 6 minutes, which is less than the answer of any respondent to my surveys.

I based the trade-offs used in the DCDE metric on the median values of survey responses. This approach minimized the influence of outlier responses from people who assigned extremely high priority to avoiding excruciating pain, like the ones mentioned above. Some other ways to interpret the survey would’ve led to excruciating pain being valued many times more than the DCDE metric does.^[7] But judging interventions solely on their ability to reduce extreme pain would misrepresent the effectiveness of programs designed with broader goals in mind.

I haven’t reviewed the literature on what weights are reasonable but Gómez-Emilsson (2019) argues that worse pains are many orders of magnitude worse than mild pains. I think that the question of how to weight different pain intensities used by the Welfare Footprint Institute deserves more research, and I imagine it could be researched in many different ways.

Comparison to other charities

Comparisons of SADs estimates

Converting my estimates into SADs also allows comparisons with interventions evaluated by ACE and AIM in 2024:

In the table above, we can see that Anima International compares favourably with other evaluated interventions. In this table, all the ACE estimates are from ACE-recommended charities, and both AIM estimates are for charities they aim to launch. Older ACE and AIM estimates don’t use SADs. I haven’t found other estimates I could compare to. Also, methodologies are different which make comparisons less fair.

Note that in the early years, cage-free campaigns in Poland relied on people working for very little or no money, which was likely unsustainable. This inflates the cost-effectiveness per dollar spent of past cage-free campaigns.^[8]

Comparisons of cage-free estimates

Both AIM’s and ACEs estimates are significantly more conservative. Hence, below I change some estimates to match my intuitions, so that they would be more comparable with my estimates of Anima International programs.

ACE’s cage-free estimate for Kafessiz Türkiye assumes that commitments were brought forward by 20.5 years, and also applies a yearly discount of 4%. This can be translated to years of impact estimate of 14.3. Based on a conversation with the director of Kafessiz Türkiye, I would’ve chosen a much higher value here. Perhaps 40 years because I see no way of cage-free changes happening in Turkey by themselves (you can see how I think about years of impact in the next section). This would increase the estimate for Kafessiz Türkiye to 150 SADs per USD. ACE doesn’t calculate the cost-effectiveness of implementation work. If Kafessiz Türkiye pursues such work, it would increase their cost-effectiveness further.

AIM estimate for cage-free campaigns in the Middle East doesn’t have a value equivalent to years of impact. But we can see that for Saudi Arabia, they expect 25,137,885 chickens to be impacted per year, and 220,991,111 chickens in total. Hence, we can calculate that they expect the impact to last for 220,991,111 / 25,137,885 = 8.8 years. I get the same number for other countries as well. I would choose a much higher number but I would also probably be much more pessimistic about the chances of success. Perhaps these two differences in opinion counteract each other.

If we use DCDEs instead of SADs, cage-free campaigns in the Middle East and Turkey appear much more cost-effective. These regions still use battery cages, which, according to the Welfare Footprint Institute, cause significantly more disabling pain—mainly due to nesting deprivation. In contrast, EU countries use furnished cages, which alleviate most of this suffering. Hence, the Welfare Footprint Institute’s estimates suggest that the switch from furnished cage to cage-free doesn’t reduce a lot of intense pain, which makes it look less effective when using to the DCDE metric.

For how many years do reforms have an impact?

One of the most uncertain parameter in all estimates is “years of impact”. We might have a solid estimate about how many animals are helped per year by a given reform. But it’s tricky to evaluate for how many years this effect might persist, especially now because the future seems so uncertain. Sometimes people choose a somewhat random (and usually very conservative) value like 5 years. Often, a very different value like 50 years might be just as reasonable of a choice, and that would make the estimated cost-effectiveness ten times as large.

For chicken welfare reforms, I attempted to develop a methodology for estimating years of impact but it’s still very much based on subjective intuitions. I am at a loss regarding how to think about years of impact for Stop The Farms and Fur Farming ban, and my choices here feel random.

Cage-free

I tried estimating years of impact using graphs like this:

A typical cage-free commitment in Poland might be won in 2018, with a deadline in 2025. Red and blue lines in the graph above represent cage-free rates of a given committed company under various scenarios. The blue line models cage-free rates of committed companies if they fulfil the commitment. The probability that companies might not fulfil their commitments is accounted for separately. The blue line goes down slowly to account for possibilities like:

• economic or sustainability pressures to go back to more efficient farming systems, especially during crises or wars
• companies dissolving or merging without maintaining the commitment,
• companies ignoring or forgetting commitments made decades ago

The red line models cage-free rates of committed companies in the hypothetical world where there was never any professional direct work on cage-free advocacy in Poland. In other words, it models the counterfactual scenario. In this case, the red line goes up mostly because of the possibility of a legislative EU cage ban some time in the future. It also goes up because of possibilities like:

• Campaigns run by unpaid volunteers
• Consumer demand for higher welfare products
• Influence of books, documentaries, videos, and social media on companies, governments, and voters.

Demand for higher animal welfare could rise significantly if transformative AI leads to much higher incomes. However, even in wealthy countries, there's probably no progress on the welfare of farmed species that lack focused advocacy — like ducks, sheep, and others.

The yellow line accounts for the possibility that commitments will stop being relevant due to things like x-risks, global catastrophic risks, societal collapse, cultured meat taking over, animals bred not to suffer, black swans, etc. I guess that yearly probabilities of one of these things happening are:

• 1.5% per year in 2025-2049
• 2.5% per year in 2050-2150
• 1% per year after 2150

These values are partly based on my own intuitions, and partly forecasts to this metaculus question. The question is about broilers but I imagine that predictions about egg-laying hens would be somewhat similar. The values of red and blue lines are based on my intuitions after talking about the question with multiple animal advocates. But they are also by no means authoritative. Please comment if you think these values should be different.

Finally, we estimate the expected proportion of hens used by companies that will be cage-free each year as follows: (blue - red) ✕ yellow. And then we add up the result for all years to calculate years of impact. In this case it’s 26 years.

Here is some reasoning that would lead to a similar conclusion. An average corporate commitment might be implemented in 2025. The EU cage ban for hens might be passed with a deadline of around 2055 (based on people's answers, that would be roughly the median guess). So about 30 years. A chance that commitments might no longer be relevant due to x-risks and other reasons would make it less than 30 years. The chance that an EU cage ban might never happen would make it more than 30 years. Perhaps these two effects cancel each other out, somewhat.

I also created an alternative estimate based on Pawel Rawicki’s (Anima International CEO) intuitions and a different counterfactual, which is better suited for estimating the years of impact for future work:

The result for future work was almost the same, but for the past work it was two times lower.

Broilers

Based on my conversations, most of the impact from broiler campaigns seem to come from a small probability that slow-growing broilers will become the default in the country, as you can see in the table below. If only a few companies implement the commitment, the changes might not persist for long.

Note that almost all the impact comes from just a 3% chance of these campaigns majorly succeeding. Pessimistic scenarios require fewer FTEs (Full Time Equivalents) because if campaigns don’t seem promising, they are more likely to be stopped.

I used the graph below to estimate what should be the years of impact value for the very optimistic scenario in the table above.

The graph suggests 28 years of impact. However, when extrapolating this far into the future, the methodology fails to account for the possibility that reforms might eventually happen with much less advocacy effort, due to a more favorable political or socio-economic climate. To reflect this, I subjectively reduce the years of impact in the optimistic scenario by 3 years, resulting in an estimate of 25 years.

Stop the farms

When local residents successfully protest against a planned chicken farm, production will likely increase elsewhere to meet demand—but how quickly? I found no clear methodology to estimate this and received no definitive answers when I asked on an economics forum. As far as I know, it could take anywhere from a week to 20 years, and the choice massively impacts cost-effectiveness.

I guess there’s a 45% chance that blocking a farm delays production by just one month, as the market might adjust quickly to meet demand elsewhere. I guess there’s a 50% chance of a one-year delay. And I guess there’s a 5% chance that stopping a farm leads to 10 fewer years of chicken farming—on the assumption that, in some cases, limiting supply can reduce long-term demand rather than simply shifting production. These assumptions yield an expected delay of one year, though this is highly uncertain.

The program also delays some farms, rather than blocks them. The average delay for that particular farm is about two years. But there is also the possibility that a different farm will be built (perhaps in another country) to fill the demand quicker. Hence, I guess the years of impact value of 0.6 years.

Note that it takes multiple years to get permits and to build a farm. Also, there are likely to be protests against new farms if they are built anywhere in Poland, so the increase in production would likely have to come from other countries.

This is the weakest part of my cost-effectiveness model. If anyone knows how to tackle this question, please comment.

Fur farms

Estimating the years of impact for a fur farming ban is similarly challenging due to uncertainty about how long the effects would last and the extent of production relocation. However, unlike broiler farming, the fur industry is in a steep long-term global decline. According to FIFUR, global mink pelt production fell from 52.2 million in 2019 to 12.3 million in 2023, and fox pelts from 13.7 million to 2.4 million. It's possible that in this might be a case where to some degree, supply creates demand: as long as farms exist, they’ll likely find markets, but if they're shut down, new ones are unlikely to open. Even if that is true, banning fur farms in Poland might still prolong the existence of fur farms in other countries as they might be able to find enough demand for longer. And it’s also very possible that new farms will be built elsewhere to meet the demand. In fact, it is my understanding that some fur production shifted to Poland after fur farming was banned in other countries. But even if that would happen, such shifts take time. Therefore, I guess the years of impact to be 6 years. It is very much a guess.

Thank you to

• Nuño Sempere for developing some of the estimates together. You can see his writeups here and here, and our last estimates together here. Perhaps 20% of work here has been done by Nuño or based on his ideas.
• Paweł Rawicki, Jakub Stencel, Bartosz Zając, Ilona Rabizo, and other staff members of Anima International for providing data and opinions about the programs.
• Vasco Grilo, Katarzyna Streich, Paweł Rawicki, Nuño Sempere, and Jakub Stencel for reviewing the article. All mistakes are my own.

1. ^{^}

    My estimates suggest that if the impact of wild-caught fish consumption was included, then the number of animals helped by retailer plant-based commitments might jump very roughly from 32 to 150 animals helped per dollar spent. I assume that these commitments will equally reduce the consumption of all animal products and even minor consumption of small fish like sprat and herring increases the numbers of animals consumed a lot. The number would significantly increase further if wild fish caught to feed farmed animals and bycatch were included (see this estimate as an example).

2. ^{^}

    For example, even though ECC commitments covering about 20 million broilers were achieved, it seems very unlikely that they would be implemented without further work on broiler welfare in Poland. Hence, my estimates assume that the past broiler work has no impact. That is, my estimates ignore the fact that past work makes achievements in the future easier. It assumes that all work on the program stops at the end of 2024. Past impact could be estimated more generously by assuming that winning the commitment is say 30% of the work. However, this would still fail to capture most of the impact of the past work, and estimates might be more difficult to interpret.

3. ^{^}

    This is partly because in the early days, cage-free work in Poland depended on volunteer and low-paid efforts, which were probably unsustainable.

4. ^{^}

    The lead on Stop The Farms program stated that 2021 was a typical year. He estimated that impact in 2023 and 2024 was likely somewhat higher and could remain elevated if larger farms continue to be built. However, we also need to account for atypical years like 2020 (COVID) and 2022 (invasion of Ukraine), when impact was lower. These factors seem to roughly balance out, making 2021 a reasonable baseline for estimating future cost-effectiveness.

5. ^{^}

    Some may argue that banning fur farming in Poland could simply shift production to other countries, limiting the net impact on animal suffering. While this concern is valid, I believe the risk is smaller in this case. Fur farming is already in steep decline—according to FIFUR, global mink pelt production fell from 52.2 million in 2019 to 12.3 million in 2023, and fox pelts from 13.7 million to 2.4 million. My intuition is that this is a case where supply creates demand: as long as farms exist, they’ll likely find markets, but if they're shut down, new ones are unlikely to open. That said, banning fur farms in Poland might still prolong operations elsewhere, like in China.

6. ^{^}

    Another example: one of the questions was “If you had to choose between one second of excruciating pain or a longer period of disabling pain, how long would the disabling pain need to last to feel equally bad overall?” Answers ranged from “5 seconds” to “more than 1 year.”

7. ^{^}

    I also considered other choices. I couldn’t use simple arithmetic mean because a few outlier answers like “more than 1 year” would end up inflating the weights a lot. Hence, I calculated a 30% Winsorized mean which replaces all values above the 85th percentile with the 85th percentile value, before calculating the arithmetic mean of the modified dataset. I thought that this would be the fairest way to interpret the survey results. However, the resulting conclusions were too extreme for mathematical reasons that are different to explain verbally.  For example, future broiler reforms ended up being about 3,700 times more cost-effective than future cage-free reforms, despite cage-free reforms affecting more years of chicken life per dollar spent. Also, 88% of the impact from broiler reforms came from an estimated 25 second reduction in excruciating pain per broiler affected.

8. ^{^}

    If we compare the cost-effectiveness of Anima International’s past and future cage-free campaigns in Poland per FTE (Full Time Equivalent) rather than per dollar spent, then my estimates suggest that past campaigns are only about 34% more cost effective than future campaigns.