I empathize with Akash Kulgod's efforts to estimate the owned and street dog population of India. There are no good surveys of the national dog population. The government does publish estimates of the dog population but they are hopelessly wrong. However, there is an interesting method that could produce relatively accurate estimates of India's dog population. It turns out that the relative population of dogs in India (that is the number of dogs per 100 or 1,000 people) varies inversely with human density and one can capture that variation by plotting dogs per 100 (or 1,000) people against log human density in a particular community. For example, a survey in the state of Haryana reported that there were approximately 25 million dogs in the state. The trendline equation for the State of Haryana is y (dogs per 1,000 people) = -100 X (log human density per sqkm) + 440. A survey of dogs in Punjab produced data that gave a trendline equation of y = -32.5 X + 158.4. In Bangalore, the equation is y = - 49 X + 250. In Jamshedpur, the equation is y = -43.6 X + 204. Therefore, it is likely that one could develop a reasonably accurate estimate of the number of dogs in India using a range of human density values and estimates of human populations in different landscapes across India.
However, the Indian pet dog population is currently growing very rapidly and so it may be necessary to undertake a few careful surveys of pet and street dog populations to update the above equations.
This is fascinating. It strikes me that you can't avoid the population ethics question here. It's not obvious to me that the lives of stray dogs is net negative, so focusing just on the WALYs lost from pup mortality ignores the WALYs gained from pups being born and living at all. If stray dogs have net positive lives, then the WALY costs from pup mortality actually become negative (compared to if you sterilized the dogs) and the conclusion changes dramatically.
FRD dogs often lead lives filled with neglect, abuse and suffering, with the only solace being that they often die young.
I don't want to generalize from my experience, but since you appealed to the experience of anyone who has lived in India, I will say that this is not my experience! Stray dogs in my neighborhood basically coexisted with humans who mostly ignored them, and they never bit anyone in my knowledge.
Analytically, the relationship between stray dogs and humans seems super path dependent. If there was one incident in the past of a dog biting a person, that could spiral into humans antagonizing the dogs and dogs antagonizing the humans, but if there hasn't been such an experience, stray dogs are basically a benign feature of the environment that you can occasionally pet, and they won't bite because they get fed. It also surely depends on whether the dogs are being fed meat regularly (which anecdotally makes them more aggressive). I would love to see some more systematic evidence on stray dog living conditions, or anything you know of that suggests a wide-ranging experience of suffering.
Also, I care a lot about animal welfare and I love dogs, but a 1/30 conversion rate between WALYs and DALYs seems insanely high to me. I would save one human over 30 dogs and I suspect almost everyone would as well. Given that most of the DALYs come from WALYs in your estimate, a more conservative conversion rate like 1/100 would bring down the importance of this cause area a lot from 8 million DALYs to 3.3 million DALYs.
I really appreciate your comment Karthik as I was wrestling with the same feelings myself - the majority of streeties I've met at home and during travels have seemed to live happy, agentic lives. I'd go so far as to say some of them lived better lives than many companion dogs — they had complete freedom (to many housed dogs live the majority of their lives tied in the same place), had rich social lives and were allowed to live out their natural impulses. I'm hesitant to know how much to trust this experience, the happy/friendly dogs are far more likely to be the ones I've interacted with, and it seems undeniable that the high early mortality, short lifespans, and general insecurity of basic needs suggests a less than desirable state of being. But, I take back how strongly I worded the sentence you quote.
I also realized while writing the tension between highlighting WALYs lost by premature pup deaths and suggesting birth control as an intervention. It seems to boil down to - does the average streetie live a happy life? It's a seemingly impossible question and I don't know yet what my answer is. For what it's worth, on an IG poll I just ran 16 ppl voted No and 4 voted Yes.
About the DALY-WALY conversion, honestly I don't find comparing cortical neurons to assess degrees of suffering meaningful but I'll admit to not knowing a better way. I think this is where EA's forcing apples to be oranges so we can crunch them together breaks down. However, note that DALY-WALY is not the same as dog life - human life, because humans live 8x longer lives. So the 30:1 ratio is implying you should save a human baby over saving 240 puppies. Curious to know if you think that's too high based off your 1/100th, which would imply the 30:1 ratio is actually too conservative.
The lives vs life years thing shouldn't change our answer much. I would also not extend the lives of 30 dogs by 1 year compared to extending a human life by 1 year, and honestly the 1/100 conversion rate I mentioned is too high for me as well, I just used it as an example of how the comparison changes with a different conversion rate.
This seems to fall under the general confusion and difficulty of evaluating wild animal suffering, and I don't envy anyone who has to do that.
Thank you for writing up a well-researched article. Although I am skeptical that this would meet the effectiveness threshold for top funds, this might be of interest to local funds. I can imagine that local governments are willing to spend significant amounts on such a problem. Yet only if they are confident in alleviating the problem. However, the problem gets increasing attention. Stray dogs seem to be an issue in other countries such as Romania as well.
I find your conclusion " [...] an economic cost of 3 Billion USD every year. The economic cost of Animal Birth Control to contain a FRD population in a city of 1 million people is around 1 million USD [...]" a bit misleading as you compare costs in different population sizes. I would update this as either cost per million or per India.
[This was written for the Cause Exploration Prize]
Epistemic Status (how confident I am)
Slightly confident. I make a lot of educated guesses/extrapolations on numbers due to the limited data but they seem to be fairly consistent with the literature. I'm less confident on the validity of the DogPopDy model used to assess the tractability section as I couldn't play around with the model myself and it may have contain a mistake.
Main Takeaways
I estimated the free-ranging dog population in India to be 46 million, of which 12 million are adult females who give birth to 24 million pups every year.
Using Welfare-Adjusted Life Years (WALYs) and an early-stage mortality of 81%, 180 million WALYs are incurred from pups alone. Using the number of cortical neurons as a comparative ratio, one WALY is equal to 1/30 DALYs.
The WHO reports that 36% of all rabies cases occur in India. This translates to 1.3 million DALYS every year and 3 billion USD of economic losses every year from rabies and rabies prevention programs.
Taken together, an uncontrolled free-ranging dog population leads to 8 million DALYs every year in India.
Using the Dog Population Dynamic model (DogPopDy), I estimate keeping a 35k FRD population in-check in the 'best-case' scenario requires 1.2mil USD, and this is unlikely to work in real-world conditions.
Summary of the Problem
Free-ranging dogs (FRD) are dogs not confined to any physical space. Global estimates of the FRD population range from 500-700 million [1]. FRD dogs often lead lives filled with neglect, abuse and suffering, with the only solace being that they often die young.
Anyone living in India can attest to the ubiquity of FRD dogs and their often pitiful existence is a flaring source for heated debate between animal activists and residents who wish for their removal from neighbourhoods, with the former motivated by human-induced canine suffering and the latter by canine-induced human suffering. The sheer magnitude of both types of suffering provides strong motivation for intervention.
We can break FRD associated suffering/costs into three categories -
FRD Suffering
Human suffering from dog bites
Human suffering from rabies + Economic cost of rabies
Free-Ranging Dog population in India
The exact FRD population size in India is unavailable due to the feasibility of conducting such a national census. A 2012 government report estimates a size of ~17 million, however, this is quite unreliable as it was based on a crude methodology relying on passive surveillance by veterinary personnel. [3] A more reliable estimate can be found by extrapolating a large, rigorous study done in Punjab done between 2016-17 that estimated a mean population of "944,000 (extrapolated using human population data)—1,569,000 (extrapolated from km2)" using the SuperDuplicates method. The SuperDuplicates method was found to be the most accurate to real populations in a separate 2018 comparative study of different survey methods conducted in a village in India. [3] Using these numbers and collapsing over the rural vs urban difference in population, we arrive at a 20:1 human:dog ratio.[1]Since FRD's are primarily dependent on humans for food (handouts or scavenging) [4], it seems fair to extrapolate numbers based on human population and not land area. To be conservative and consistent with a ratio used in another paper [5], I will use a ratio of 1:30 (one FRD dog for every human). Given India's population of 1.38 billion, we arrive at a FRD population of 46 million.
Two aspects of the FRD population are important w.r.t any population intervention - the male:female ratio and the young:adult ratio. Using the same estimates from the punjab study, the FRD population was skewed male and adult — sex ratio of 1.23 males for every female and age ratio of 0.32 pups/juveniles for every adult dog. This leads to an breakdown of -
31 million adults and 15 million juveniles/pups
19 million male adults and 12 million females
A 2008 study in Jaipur found that 50% of adult females got pregnant in a year [6] and the median litter size was found to be 4 in a separate study by the Dog Lab in Kolkata. [7] Hence, an estimate for the number of pups born every year is 24 million. This is potentially a conservative estimate as other studies have found a significantly greater rate of pregnancy, larger litter sizes, and more than one breeding cycle per year. [8] [9]
Importance
Welfare-Adjusted Life Years (WALYs)
Researchers in 2018 proposed adapting the Disability-Adjusted Life Year approach for animals resulting in Welfare-Adjusted Life Years (WALYs) [10]. The WALY is the sum of a) the years lived with impaired welfare due to a particular cause and b) the years of life lost due to the premature death from the same cause. Since no data exists on the average healthy lifespan of FRD's, I will assume a healthy free-ranging dog lives on average for 10 years (a conservative estimate given the average dog lifespan seems to be 11.2yrs and the majority of FRD's are mixed breeds that tend to live longer).
Equating Human and Dog Suffering
It is not straightforward to compare the suffering of humans and dogs. A crude method to compare human and non-human suffering is to use the number of cortical neurons to assess how many animals equals one human [Scott Alexander does precisely this in a 2019 post [11]] A 2017 study estimated the number of cortical neurons in a golden retriever and a dog of unspecified breed, finding 627 million neurons and 429 million neurons respectively. [12] Taking the approximate middle ground leaves the average FRD with 500 million cortical neurons. The estimate for the number of human cortical neurons is around 15 billion [13], which leaves us with the same (!) 30:1 ratio, i.e. one DALY equals 30 WALYs. [2]
Indian FRD WALYs
Free-ranging dogs face an overwhelming mortality rate, especially when young. Researchers at the Dog Lab followed 364 pups in a five-year period and found only 19% surviving till the reproductive age (7 months) [7]. Given the previous estimate of 24 million pups born per year, this implies around 20 million pups die every year. Assuming 9 WALYs for each pup death and omitting the loss due to their impaired welfare, this results in 180 million WALYs.
High quality data on the average lifespan of a FRD does not seem available. The same 2008 Jaipur study estimated a female FRD living on average 3.8 yrs if she survived the first year. I'll assume the average FRD lifespan to be 5 years. If we take the simple picture that the FRD population remains stable every year at holding capacity, 4 million adult dogs die every year leading to 20 million WALYS.
Hence, the loss from premature deaths is 200 million WALYs, or, using our 30:1 ratio, 6.67 million DALYs every year.
This does not include the WALYs lost from disease which seem difficult to calculate given the lack of data.
Human DALYs and Economic Costs
The Integrated Disease Surveillance Programme (IDSP) reported 7.4 million dog bites occurring in 2018 in India. According to a 2021 systematic review, FRD-induced bites accounted for ~60% of all animal bites, i.e 4.5 million, and were mainly reported in male individuals belonging to lower socio-economic status. [14]
A 2015 paper estimating the global burden of endemic rabies found that [15]
canine rabies causes approximately 59,000 (95% Confidence Intervals: 25-159,000) human deaths, over 3.7 million (95% CIs: 1.6-10.4 million) disability-adjusted life years (DALYs) and 8.6 billion USD (95% CIs: 2.9-21.5 billion) economic losses annually. The largest component of the economic burden is due to premature death (55%), followed by direct costs of post-exposure prophylaxis (PEP, 20%) and lost income whilst seeking PEP (15.5%), with only limited costs to the veterinary sector due to dog vaccination (1.5%), and additional costs to communities from livestock losses (6%).
The WHO reports that 36% of all rabies cases occur in India. This translates to 1.3 million DALYS every year and 3 billion USD of economic losses every year from rabies and rabies prevention programs. These figures do not include the costs of dog bites that do not translate into rabies.
Hence, attempting to take consistently take conservative estimates and only considering FRD mortality and human rabies mortality/costs, the total DALYs incurred by FRD's every year is around 8 million DALYs.
Tractability
Animal Birth Control and Vaccination (ABCV)
Animal Birth Control and Vaccination (ABCV) is the most common intervention proposed for mitigating welfare/economics costs associated with FRDs. It involves catching adult FRDs, spaying/neutering, vaccinating and marking each dog, and releasing them back to the street. Cities across India launch ABCV campaigns, such as Nashik (human popl 1.5mil, estimated dog popl 50k) allotting $141,000 [16] and Belagavi (human popl 750k, estimated dog popl 25k) planning to allot $75,000 [17]. Previous attempts appear to be somewhat successful at reducing FRD population, with a study in Jodhpur seeing a "decrease in our dog population size ranging from 27.7% (Area 6) to 51.2% (Area 3) between 2005 and 2007." [18]
A useful paper published in 2020 - Modelling the challenges of managing free-ranging dog populations - outlines a "agent-based dog population dynamics model to examine the time, effort, financial resources, and conditions needed to successfully control FRD in a typical urban setting." The authors highlight that only in the “best-case” and high intensity scenario, ABCV resulted in a significant and lasting reduction in FRD population sizes. [5]
Specifically, for a city with 35k dogs, it requires 42k surgeries and $500k over a 5yr period to reduce the population to 1.5k dogs after 20 years. Here, best-case means that all dogs are equally catchable (i.e no dogs go uncaught) and no dog immigration occurs. Assuming a real-world scenario with a 5% uncatchable rate and 1% immigration, the models do not show population control.
I noted a factor that may enable a more optimistic interpretation of the model's results. The model caps monthly surgeries at each ABC center at 250, stating these were regulations set out by the Animal Welfare Board of India. However, looking at the guidelines [20], the number 250 only crops up as an example for how many kennels should be built (pg.59). If such regulations do not exist (and it's difficult to see why they would), the number of surgeries at each center are constrained mainly by operating theatres and veterinary staff.
A more pessimistic update is that the model uses a surgery cost of 700 INR (8.8USD). The AWBI has mandated a minimum rate of 1650 INR (21 USD), a 2.4x increase.
Accounting for this price increase, keeping a 35k FRD population in-check in the 'best-case' scenario requires 1.2mil USD. Even then, DogPopDy indicates this would not work in real-world conditions.
What then?
At this point, I'll admit to being left pretty stumped as I'd expected ABCV interventions to be more effective and tractable than they are. Alternate approaches include straight-up culling - which is evidently bad from a moral standpoint, not effective unless thoroughly implemented, and not tractable given cultural/social norms (thankfully!) and mass shelter-building and adoption - which doesn't seem to be tractable given the massive population and the fact that all dogs need to be sheltered/adopted for this to work.
Interventions that reduce the holding capacity of neighbourhoods, specifically reducing garbage spillovers and direct feeding, may be somewhat more effective and tractable. However, this essentially entails starving millions of dogs and barring people from feeding starving dogs as well as increasing the likelihood of dog bites (since aggressiveness is linked to hunger), all while having no tangible way of knowing if this is actually working.
A final important point - controlling FRD populations can only occur if there is dedicated multi-decade monetary support, widespread implementation, and continuous monitoring coupled with sophisticated research, modelling and target-setting. These are things governments tend to be woefully unable to deliver and programs that have been going on for years can be abruptly cut short by funding pauses, wiping out any benefits within a few years (e.g. Guwahati). These are also things EA is much better positioned to being able to carry out.
Conclusion
Free-ranging dog populations in India are associated with at least 8 million DALYS (6.67 from dog WALYS, 1.3 from human rabies DALYs) and an economic cost of 3 Billion USD every year. The economic cost of Animal Birth Control to contain a FRD population in a city of 1 million people is around 1 million USD spread over 5 years of intervention, assumes a closed and perfectly catchable dog scenario, and requires continuous monitoring and intervention over decades. Alternative interventions that are effective and tractable are not obvious.
The main actionable intervention at this point is to hit pause on the millions of dollars being spent/planning to be spent on ABC measures across cities in India before further research/modelling is done as it's likely they're a complete sink of money and human resources with little to no benefit.
The authors of the paper note that "We estimated a human to stray dog ratio of 28:1 and 32:1 in rural and urban areas, respectively." But, the numbers in table 3 don't map onto these ratios. Not sure of what's happening here which is why I play it safe and use a 30:1 ratio to estimate population size.
This post presents the executive summary from Giving What We Can’s impact evaluation for 2025. At the end of this post we share links to more information, including the full report and...
I used AI to fix transcription errors, rerrarange the ideas, and suggest tweaks to the title and some sentences.
Three of the most exciting projects to come out of EA in recent years are, in a vague sense, CEA spinouts:
* Kairos is directly a spinout of CEA and now handles most support for university AI safety groups. Basically everyone I've found who knows them is really excited about what they do
* NEST is an opinionated ideas-fi...
I empathize with Akash Kulgod's efforts to estimate the owned and street dog population of India. There are no good surveys of the national dog population. The government does publish estimates of the dog population but they are hopelessly wrong. However, there is an interesting method that could produce relatively accurate estimates of India's dog population. It turns out that the relative population of dogs in India (that is the number of dogs per 100 or 1,000 people) varies inversely with human density and one can capture that variation by plotting dogs per 100 (or 1,000) people against log human density in a particular community. For example, a survey in the state of Haryana reported that there were approximately 25 million dogs in the state. The trendline equation for the State of Haryana is y (dogs per 1,000 people) = -100 X (log human density per sqkm) + 440. A survey of dogs in Punjab produced data that gave a trendline equation of y = -32.5 X + 158.4. In Bangalore, the equation is y = - 49 X + 250. In Jamshedpur, the equation is y = -43.6 X + 204. Therefore, it is likely that one could develop a reasonably accurate estimate of the number of dogs in India using a range of human density values and estimates of human populations in different landscapes across India.
However, the Indian pet dog population is currently growing very rapidly and so it may be necessary to undertake a few careful surveys of pet and street dog populations to update the above equations.