Quantifying the Impact of Economic Growth on Meat Consumption

(up­date: see https://​​fo­rum.effec­tivealtru­ism.org/​​posts/​​EztxZhPQ8Qv8xWe3v/​​kbog-did-an-oop­sie-new-meat-eater-prob­lem-num­bers)

I de­cided to do some anal­y­sis to gain in­sight on the im­pact of eco­nomic growth on meat con­sump­tion in the de­vel­op­ing world, which has oc­ca­sion­ally been dis­cussed in the past.

First I took the find­ing of York and Gos­sard (2004) that for ev­ery $1,000 in­crease in per cap­ita PPP GDP, Afri­can coun­tries con­sume 1.66 kg more meat per per­son per year.1 For some per­spec­tive on the sig­nifi­cance of that differ­ence in GDP, see https://​​en.wikipe­dia.org/​​wiki/​​List_of_coun­tries_by_GDP_%28PPP%29_per_cap­ita.

Then I used FAOSTAT data on meat pro­duc­tion2 and con­sump­tion3 from var­i­ous sam­ples of low in­come Afri­can coun­tries which are tar­geted by Givewell-recom­mended top char­i­ties to de­ter­mine what pro­por­tion of that 1.66 kg is likely to come from var­i­ous types of meat.

I ex­cluded dairy and eggs be­cause I figured that all those an­i­mals are also slaugh­tered for meat at some point and there­fore are already taken into ac­count, given that these are de­vel­op­ing coun­tries. More­over, egg pro­duc­tion num­bers are not available whereas dairy farm­ing causes by far the low­est amounts of suffer­ing. I also ex­cluded game meat be­cause its im­pact is on wild an­i­mal suffer­ing, it isn’t farmed, and I doubt that marginal meat con­sump­tion in­cludes game meat. Fi­nally, I ex­cluded meat of “all other types” from the anal­y­sis.

I also ex­cluded fish be­cause there is no sig­nifi­cant cor­re­la­tion be­tween in­come and fish con­sump­tion in Afri­can coun­tries.1

Then I used Brian To­masik’s es­ti­mates of suffer­ing per kg to de­ter­mine days of suffer­ing caused to var­i­ous an­i­mals. For cat­tle, camels, horses, mules, and asses I used his es­ti­mate for suffer­ing per kg of beef; for pigs, sheep and goats I used his es­ti­mate for suffer­ing per kg of pork; for all poul­try and rab­bits I used his es­ti­mate of suffer­ing per kg of chicken.4 Even though To­masik is un­cer­tain about these num­bers them­selves and points out that they shouldn’t be taken at face value, they provide a de­cent gen­eral guide.

This gave me num­bers of days of suffer­ing for the var­i­ous kinds of an­i­mals farmed in sub-Sa­haran Africa. Note that To­masik ad­justed the num­ber of days of suffer­ing to ac­count for the differ­ences in av­er­age sub­jec­tive bad­ness based on how differ­ent kinds of an­i­mal farms op­er­ate in the US.

Here are some sig­nifi­cant con­sid­er­a­tions not taken into ac­count in the model that will af­fect the amount of suffer­ing caused to farm an­i­mals. (Plus sign means that this con­sid­er­a­tion will cause the amount of suffer­ing to be greater than the anal­y­sis in­di­cates. Minus sign means that this con­sid­er­a­tion will cause the amount of suffer­ing to be less than the anal­y­sis in­di­cates. Ques­tion mark in­di­cates that the di­rec­tion is un­clear.) To­gether, they give strong rea­son to be un­cer­tain about the re­sults of these calcu­la­tions.

· Do­mes­tic an­i­mals are smaller in de­vel­op­ing parts of Africa than they are in US fac­tory farms, mean­ing more an­i­mals are raised per quan­tity of meat. (+)

· Afri­can farms are sub­ject to gen­er­ally weaker stan­dards of reg­u­la­tion, slaugh­ter meth­ods, etc. (+)

· Most of the new meat con­sump­tion in Africa is poul­try and pork,5 so the num­bers in these calcu­la­tions give too much weight to less-suffer­ing-in­ten­sive cat­tle. (+)

· Marginal Afri­can farm pro­duc­tion is not always go­ing to be fac­tory farm­ing and may in­volve less in­ten­sive con­di­tions than farm­ing in the US. (-) (How­ever, much of the new pro­duc­tion is fac­tory farms,6 and tra­di­tional farm­ing may not be any more hu­mane.)

· There may be sup­ply side elas­tic­i­ties on a re­gional level. (-)

· Suffer­ing per kg could be differ­ent for goats and sheep than it is for pigs. (?)

· Pro­duc­tion ra­tios may differ from con­sump­tion ra­tios of differ­ent types of meat in the tar­get coun­tries. (?) (Not the case for the 4th calcu­la­tion.)

· Con­sump­tion ra­tios of differ­ent types of meat for poor Afri­cans may differ from that of Afri­cans in gen­eral. (?)

· The in­come effect on meat con­sump­tion for poor Afri­cans in the tar­get coun­tries may differ from that of Afri­cans in gen­eral. (?)

Due to these con­sid­er­a­tions, I would say that the num­bers here should be taken with a grain of salt, prob­a­bly al­low­ing for a fac­tor of +/​- 10 for rea­son­able but not great con­fi­dence.

Con­sid­er­a­tions en­tirely out­side the model: im­pact of de­vel­op­ment on wild an­i­mal suffer­ing, cli­mate change, tech­nolog­i­cal progress, global eco­nomic de­vel­op­ment, etc.

Another thing to un­der­line: this is the effect of eco­nomic de­vel­op­ment on meat con­sump­tion. It is not the effect of eco­nomic aid on meat con­sump­tion. Depend­ing on how much you think AMF/​SCI/​DWI con­tribute to changes in in­come per cap­ita, you will have to eval­u­ate them differ­ently. Even GiveDirectly’s effects aren’t straight­for­ward to calcu­late per donor be­cause GDP here is be­ing mea­sured in terms of pur­chas­ing power par­ity, not a nom­i­nal cur­rency con­ver­sion.

I’ve left the ac­tual num­bers for the end be­cause I didn’t want any­one to take them and run with them with­out first hav­ing to read all the dis­claimers about un­cer­tainty and as­sump­tions. So here is my quick sum­mary: de­pend­ing on how we are mea­sur­ing meat con­sump­tion in aid tar­get coun­tries, and in­clud­ing my +/​-10 guessti­mate for cer­tainty as well as tak­ing the high­est and low­est val­ues of the 4 differ­ent calcu­la­tions I performed, an in­crease in $1000 to a per­son’s in­come will gen­er­ate:

Between 0.04 and 13 days of large an­i­mal suffer­ing (per per­son per year at $1000 ad­di­tional in­come); me­dian about 1

Between 0.16 and 41 days of pig, goat, and sheep suffer­ing; me­dian about 2

Between 0.65 and 138 days of chicken and rab­bit suffer­ing; me­dian about 10

Clearly if the an­swer is at the low end of the range then eco­nomic de­vel­op­ment is un­equiv­o­cally good in di­rect well-be­ing re­spects. If the an­swer is near the high end then it could be difficult to claim that the welfare benefits of eco­nomic de­vel­op­ment are more im­por­tant than the an­i­mal suffer­ing, de­pend­ing on how much you care about an­i­mals.

Now as Carl Shul­man points out in the com­ments, if this pos­si­bil­ity is trou­bling to you, then you should re­con­sider why you aren’t pri­ori­tiz­ing an­i­mal-ad­vo­cacy char­i­ties over poverty char­i­ties in the first place.

Fi­nally, here are the Ex­cel sheets so that you can view them:

https://​​docs.google.com/​​spread­sheets/​​d/​​192J3lQgi7nlsksSLrx9B_NUqqnpfKC1dNa5oF2Fnr9w/​​edit?usp=sharing

And here they are in ed­itable form:

https://​​docs.google.com/​​spread­sheets/​​d/​​1DUEDkKgsJCEg0vADSm1UPVm­sTOMxJe1BpI3D232VDYU/​​edit?usp=sharing

Sources:

1. http://​​smas.chemeng.ntua.gr/​​miram/​​files/​​publ_141_10_2_2005.pdf York and Gos­sard study on in­comes and meat consumption

2. http://​​fao­stat3.fao.org/​​down­load/​​Q/​​QL/​​E FAOSTAT data on production

3. http://​​fao­stat3.fao.org/​​down­load/​​FB/​​FBS/​​E FAOSTAT data on consumption

4. http://​​re­duc­ing-suffer­ing.org/​​how-much-di­rect-suffer­ing-is-caused-by-var­i­ous-an­i­mal-foods/​​ Brian To­masik’s num­bers on farm an­i­mal suffering

5. http://​​www.slide­share.net/​​guy­col­len­der/​​trends-in-live­stock-pro­duc­tion-and-con­sump­tion-cees-de-haan-world-bank World Bank slideshow, shows that most new meat con­sump­tion is poul­try and pork (slide 4)

6. http://​​www.wor­ld­watch.org/​​node/​​1826 Fac­tory farm­ing is the fastest grow­ing type of food pro­duc­tion, es­pe­cially in the de­vel­op­ing world.