Around the world, tens of millions of innocent people are enslaved in sweatshops, brick kilns & brothels, on farms and fishing vessels, inside homes, and in other dire situations. The largest categories of modern-day slavery are forced labour, bonded labour (a.k.a. debt bondage), sexual slavery, and coerced marriage.
In 2016, Giving What We Can (a part of the Centre for Effective Altruism) rated this cause at 3 stars out of a possible 5 in terms of its importance, neglectedness, and tractability. Also, it rated “the best [anti-slavery] charity we can find” (The Freedom Fund) -- focussing on its work in India in particular—as follows:
cost-effectiveness -- 3 stars (out of 5)
robustness of evidence − 3 stars
quality of implementation − 4 stars
But last September, The Freedom Fund released new data about its recent work in 3 states of India that showed an approximate 125,000-person decline in bonded labour. This decline—relative to the (untouted) cost of it—leads to my rough estimate that the cost of each free person who otherwise would have been in bonded labour was $80 or 71 euros.
This indicates that—contrary to GWWC’s rating in 2016 -- The Freedom Fund’s more recent efforts to reduce bonded labour in parts of India have been highly cost-effective.
In talking about modern-day slavery in terms of EA, I’d also discuss the range of disability factors that Giving What We Can used in 2016 (for the value of sparing someone from enslavement relative to the value of saving someone’s life). The mid-point of the range was 0.5 -- which would mean that sparing 2 people from enslavement would be equal in value to saving 1 person’s life. But the evaluation also said the value of freedom might be much lower or much higher. (Putting a higher value on freedom from slavery would have raised the 3-star rating that GWWC put on the importance of the cause.)
I’d also like to explain why optimal efforts against slavery in some settings (such as brick kilns) are much more cost-effective than optimal efforts against modern-day slavery in more diffuse situations (such as coerced marriages).
NOTE: My sources include reports by The Freedom Fund, books by Siddharth Kara (Modern Slavery, Bonded Labour), books by Kevin Bales (Ending Slavery, Disposable People), U.S. State Department Trafficking in Persons Reports, International Labour Organization reports, interviews with survivors, prosecutors, activists & care providers.
Here I cite 2 outside evaluations of The Freedom Fund’s anti-bonded labour work in N. India and in S. India. You will see that they measured the reduction in HOUSEHOLDS that included at least one bonded laborer—rather than estimating the reduction in the number of bonded laborers. (My estimate for the reduction in the number of bonded laborers came directly from The Freedom Fund.)
I literally do not know how to put links here (or anywhere). But if you Google the stuff I mentioned, those sources should be easy to find.
Here’s how I calculated the cost-effectiveness of The Freedom Fund:
$ 12.1 million Total spending in 2 states in Northern India (2014 thru 2019)
+ $6.8 million Total spending in 1 state in Southern India (2015 thru 2019)
_________
$18.9 million TOTAL SPENDING in the 3 states
Each year, FF’s cumulative spending in the 2 states in Northern India is stated in its Annual Impact Report. Ditto, FF’s cumulative spending in the 1 state in Southern India. So by subtraction, I computed FF’s annual spending in N. India and also in S. India.
For N. India, where the baseline for evaluating the work was early in 2016 and the endline was late in 2018, I subtracted FF’s spending there in 2014, 2015 & 2019. This gave me a figure for 2016 through 2018 that I was confident overstated the exact amount.
For S. India, where the baseline was mid-2018, I subtracted FF’s spending there in 2015 (that’s when the work started) and 2019. But because FF’s spending there in 2017 was much lower than in 2016 or in 2018, I was concerned that the 2017 figure might be some kind of anomaly, and so I decided NOT to estimate FF’s baseline-to-endline spending in S. India as: 2017 + half of 2016 + half of 2018. Instead, to be conservative, I added those 3 years of spending, and then multiplied the sum by 2⁄3. Estimating the baseline-to-endline cost this way gave me confidence that I was not underestimating it.
So:
$ 18.9 million TOTAL SPENDING in the 3 states
-$ 8.9 million FF SPENDING excluded due to being before the baseline or after the endline
__________
about $10 million in F.F. SPENDING in the 3 states between baseline and endline
$10 million divided by 125,000 (reduction in bonded labour) = $80 (cost per person spared bonded labour)
I’m going to post this & then respond soon re: assumptions.
This is such an important topic and I would love to hear more about it. I definitely want to hear more about the range of disability factors that Giving What We Can used in 2016 . I think my suggestion would be to hear more about calls to action, because now that we have identified this as a problem what are some things we can do? Really hope to continue discussing this.
EAs are Underrating the Anti-slavery Cause
Around the world, tens of millions of innocent people are enslaved in sweatshops, brick kilns & brothels, on farms and fishing vessels, inside homes, and in other dire situations. The largest categories of modern-day slavery are forced labour, bonded labour (a.k.a. debt bondage), sexual slavery, and coerced marriage.
In 2016, Giving What We Can (a part of the Centre for Effective Altruism) rated this cause at 3 stars out of a possible 5 in terms of its importance, neglectedness, and tractability. Also, it rated “the best [anti-slavery] charity we can find” (The Freedom Fund) -- focussing on its work in India in particular—as follows:
cost-effectiveness -- 3 stars (out of 5)
robustness of evidence − 3 stars
quality of implementation − 4 stars
But last September, The Freedom Fund released new data about its recent work in 3 states of India that showed an approximate 125,000-person decline in bonded labour. This decline—relative to the (untouted) cost of it—leads to my rough estimate that the cost of each free person who otherwise would have been in bonded labour was $80 or 71 euros.
This indicates that—contrary to GWWC’s rating in 2016 -- The Freedom Fund’s more recent efforts to reduce bonded labour in parts of India have been highly cost-effective.
In talking about modern-day slavery in terms of EA, I’d also discuss the range of disability factors that Giving What We Can used in 2016 (for the value of sparing someone from enslavement relative to the value of saving someone’s life). The mid-point of the range was 0.5 -- which would mean that sparing 2 people from enslavement would be equal in value to saving 1 person’s life. But the evaluation also said the value of freedom might be much lower or much higher. (Putting a higher value on freedom from slavery would have raised the 3-star rating that GWWC put on the importance of the cause.)
I’d also like to explain why optimal efforts against slavery in some settings (such as brick kilns) are much more cost-effective than optimal efforts against modern-day slavery in more diffuse situations (such as coerced marriages).
NOTE: My sources include reports by The Freedom Fund, books by Siddharth Kara (Modern Slavery, Bonded Labour), books by Kevin Bales (Ending Slavery, Disposable People), U.S. State Department Trafficking in Persons Reports, International Labour Organization reports, interviews with survivors, prosecutors, activists & care providers.
#####
Thanks for this writeup—I’d greatly appreciate any further information you could provide about anti-slavery.
What are the links for these sources?
Can you share your cost-effectiveness calculations for The Freedom Fund? What does it assume?
What are the best writeups of the problem? Could you link to them?
Here I cite 2 outside evaluations of The Freedom Fund’s anti-bonded labour work in N. India and in S. India. You will see that they measured the reduction in HOUSEHOLDS that included at least one bonded laborer—rather than estimating the reduction in the number of bonded laborers. (My estimate for the reduction in the number of bonded laborers came directly from The Freedom Fund.)
Institute of Development Studies & Praxis Institute for Participatory Practices, Northern India Hotspot Prevalence Study and Evaluation, 2019
https://opendocs.ids.ac.uk/opendocs/handle/123456789/14653
Institute of Development Studies & Praxis Institute for Participatory Practices, Southern India Hotspot Prevalence Study and Evaluation, 2019
https://opendocs.ids.ac.uk/opendocs/handle/123456789/14651
CORRRECTION in what I just posted re: how I calculated The Freedom Fund’s cost-effectiveness:
In Southern India, the baseline was mid-2016 and the end line was mid-2018.
I literally do not know how to put links here (or anywhere). But if you Google the stuff I mentioned, those sources should be easy to find.
Here’s how I calculated the cost-effectiveness of The Freedom Fund:
$ 12.1 million Total spending in 2 states in Northern India (2014 thru 2019)
+ $6.8 million Total spending in 1 state in Southern India (2015 thru 2019)
_________
$18.9 million TOTAL SPENDING in the 3 states
Each year, FF’s cumulative spending in the 2 states in Northern India is stated in its Annual Impact Report. Ditto, FF’s cumulative spending in the 1 state in Southern India. So by subtraction, I computed FF’s annual spending in N. India and also in S. India.
For N. India, where the baseline for evaluating the work was early in 2016 and the endline was late in 2018, I subtracted FF’s spending there in 2014, 2015 & 2019. This gave me a figure for 2016 through 2018 that I was confident overstated the exact amount.
For S. India, where the baseline was mid-2018, I subtracted FF’s spending there in 2015 (that’s when the work started) and 2019. But because FF’s spending there in 2017 was much lower than in 2016 or in 2018, I was concerned that the 2017 figure might be some kind of anomaly, and so I decided NOT to estimate FF’s baseline-to-endline spending in S. India as: 2017 + half of 2016 + half of 2018. Instead, to be conservative, I added those 3 years of spending, and then multiplied the sum by 2⁄3. Estimating the baseline-to-endline cost this way gave me confidence that I was not underestimating it.
So:
$ 18.9 million TOTAL SPENDING in the 3 states
-$ 8.9 million FF SPENDING excluded due to being before the baseline or after the endline
__________
about $10 million in F.F. SPENDING in the 3 states between baseline and endline
$10 million divided by 125,000 (reduction in bonded labour) = $80 (cost per person spared bonded labour)
I’m going to post this & then respond soon re: assumptions.
Clarification: I’m requesting an Early time slot.
This is a very important topic and I’d like to learn more about it. Hopefully, I’ll get to hear you discuss it at the Unconference.
Hi Douglas,
This is such an important topic and I would love to hear more about it. I definitely want to hear more about the range of disability factors that Giving What We Can used in 2016 . I think my suggestion would be to hear more about calls to action, because now that we have identified this as a problem what are some things we can do? Really hope to continue discussing this.