Cash transfers are not necessarily wealth transfers

Link post

Here’s a com­mon ar­gu­ment:

The prob­lem with the poor is that they haven’t got enough money. There’s am­ple em­piri­cal ev­i­dence back­ing this up. There­fore, the ob­vi­ously-cor­rect poverty in­ter­ven­tion is to sim­ply give the poor cash. You might be able to do bet­ter than this, but it’s a solid baseline and you should of­ten ex­pect to find that in­ter­ven­tions are worse than cash.

There are tech­ni­cal rea­sons to be skep­ti­cal of cash trans­fers—which is why it is so im­por­tant that the cash trans­fer char­ity GiveDirectly is care­fully re­search­ing what ac­tu­ally hap­pens when they give peo­ple cash—but un­til fairly re­cently, these ob­jec­tions seemed to me like ab­struse nit­picks about an in­ter­ven­tion that was al­most an­a­lyt­i­cally cer­tain to be strongly benefi­cial.

But they’re not just nit­picks. Cash isn’t ac­tu­ally the same thing as hu­man well-be­ing, and the as­sump­tion that it can be sim­ply ex­changed into pretty much any­thing else is not ob­vi­ously true.

Of course, say­ing “X is pos­si­bly wrong” isn’t very helpful un­less we have a sense of how it’s likely to be wrong, un­der what cir­cum­stances. It’s no good to treat cash trans­fers just the same as be­fore, but be more gloomy about it.

I’m go­ing to try to com­mu­ni­cate an un­der­ly­ing model that gen­er­ates the ap­pro­pri­ate kind of skep­ti­cism about in­ter­ven­tions like cash trans­fers, in a way that’s in­tu­itive and not nar­rowly tech­ni­cal. I’ll be­gin with a parable, and then talk about how it re­lates to real-world cases.

Two cities, two stadiums

In a world where the only thing peo­ple en­joy is base­ball, there is a wealthy city. In this city, there is an ex­cel­lently de­signed base­ball sta­dium. The seats are am­ply sized and com­fortable. Even the worst seats have a good view of the field. There are plen­tiful ameni­ties, though the price of food and drink is high. There are awnings to block the rain. There are lights in case the game goes late.

Else­where, in a poor city, there is an­other base­ball sta­dium. This one is shod­dily built. The up­per seats are tiny to cram in as many peo­ple as pos­si­ble. The worst seats can hardly see the field, or are ex­posed to in­clement weather. The aisles are too nar­row. The mood is chaotic, and in the poorer ar­eas of the sta­dium – but not in the ex­pen­sive ar­eas with the best views – peo­ple of­ten get into fights. A smaller va­ri­ety of cheaper, lower-qual­ity food and drink is available. There is always a line for the bath­room, un­less you paid for a box with a pri­vate one.

You hap­pen to live in the rich city, and at­tend the rich sta­dium. You know two more things, from study­ing data within and be­tween many cities in this world:

  1. Within a city, the self-re­ported en­joy­ment of base­ball scales log­a­r­ith­mi­cally with in­di­vi­d­ual wealth. In other words, if you ask peo­ple to rate their hap­piness on a scale from 1 to 10, then no mat­ter how rich or poor some­one is, they’ll be the same dis­tance on the scale from some­one with twice their wealth. (For in­stance, if peo­ple with an an­nual in­come of $10,000 re­port an en­joy­ment level of 5.2 out of 10 on av­er­age, and peo­ple with an an­nual in­come of $20,000 re­port 5.5 out of 10, then peo­ple with an an­nual in­come of $40,000 re­port 5.8 out of 10.)

  2. Between cities, self-re­ported en­joy­ment of base­ball also scales log­a­r­ith­mi­cally with (av­er­age) wealth.

Clearly, en­joy­ment is a sim­ple pos­i­tive func­tion of money; the prob­lem with the peo­ple in the poor sta­dium is that they haven’t got enough of it. What’s more, since the func­tion is log­a­r­ith­mic, a dol­lar goes much farther for the poor than for the rich. So you send some to the poor­est peo­ple in the poor sta­dium, hop­ing that it will do then some large mul­ti­ple of the good it can do you.

Be­cause you know that good-sound­ing char­i­ta­ble in­ter­ven­tions of­ten fail for sur­pris­ing rea­sons, you de­cide to test your as­sump­tions, by fol­low­ing up with the re­cip­i­ents of the cash trans­fers. What do you find?

It turns out peo­ple care about two things: the qual­ity of their seat, and food.

Some re­cip­i­ents buy more, or bet­ter food. Be­cause the prices are lower over there, the differ­ence in qual­ity is large for them, even though the money would only make a small differ­ence to them. Other re­cip­i­ents buy their way into bet­ter seats. Again, since the seats in the poor sta­dium cost less than the seats in the rich sta­dium, they gain a lot more than you lose. Over­all, it looks like ev­ery­one who re­ceives money en­joys the game a lot more, so your be­lief in the mer­its of cash trans­fers has been con­firmed.

Then you learn about a third statis­ti­cal reg­u­lar­ity that flies in the face of ev­ery­thing you’ve learned so far:

  1. As in­di­vi­d­ual cities get richer, av­er­age en­joy­ment of base­ball games does not in­crease.

What’s go­ing on?

When some­one pays for on a nicer seat, their ex­pe­rience of the game is im­proved. But if they’ve out­bid some­one else for that seat, that other per­son is now stuck in a worse seat. Buy­ing some­one a nicer seat in the same sta­dium does not im­prove the av­er­age game en­joy­ment, be­cause – as­sum­ing fully booked sta­di­ums – it makes some­one else’s ex­pe­rience worse, be­cause they’re now stuck in the bad seat. So it mat­ters quite a lot how much of the vari­a­tion in peo­ple’s well-be­ing comes from things that can eas­ily be got more of, like food, and how much comes from lo­cally scarce goods, like choice seats.

But richer cities have nicer sta­di­ums! Doesn’t that mean that if you trans­fer enough money to peo­ple in poor cities, the poor city sta­di­ums will get bet­ter? Maybe not! Sta­di­ums are pretty hard to change sub­stan­tially once built. And maybe it was never the money that made the sta­di­ums bet­ter; maybe cities that have their act to­gether tend to be bet­ter both at mak­ing money, and at sta­dium-build­ing. You don’t have em­piri­cal rea­son to doubt this.

How should your strat­egy to im­prove peo­ple’s base­ball ex­pe­riences take this into ac­count?

First, cash trans­fers can still be of some value. For in­stance, the very poor­est spec­ta­tors may go hun­gry. If you send them money, and they use the money to buy food, they might en­joy the game a lot more with­out harm­ing any­one else. But if you keep giv­ing them money, even­tu­ally they’ll have enough food. The only thing left to buy is a po­si­tional good: bet­ter seats.

Se­cond, if there are high-wage cities with spare sta­dium seat­ing, you might want to help peo­ple move there. They’ll en­joy base­ball games more, since even the com­par­a­tively bad seats will be bet­ter, with­out harm­ing any­one else. Some might turn down the op­por­tu­nity out of loy­alty to their home­town team, but oth­ers might take you up on it.

Third, if there is a way you can im­prove your home sta­dium, this is an im­por­tant good. If you’re will­ing to learn for­eign cul­tural norms and be gen­uinely cu­ri­ous about why poor peo­ple have worse sta­di­ums, you might even be able to help them re­form their in­sti­tu­tions to get bet­ter sta­di­ums built, which could make a big differ­ence in the qual­ity of their lives.

The sec­ond and third points go to­gether well. If your town’s sta­dium is at ca­pac­ity, im­mi­gra­tion doesn’t help any­one en­joy a nicer game – but per­suad­ing the owner to add more seats can change that.

How this ap­plies to the real world

In The Price of Glee in China, Scott Alexan­der points out a few key facts about the hap­piness liter­a­ture:

  1. Within coun­tries, self-re­ported well-be­ing seems to scale log­a­r­ith­mi­cally of av­er­age in­come. (Again, that means that each dou­bling of in­come cor­re­sponds to the same, con­stant in­crease in re­ported hap­piness points.)

  2. Between coun­tries, a similar re­la­tion­ship seems to hold.

  3. Within coun­tries, per cap­ita GDP growth does not ap­pear to lead to cor­re­spond­ing in­creases in well-be­ing.

Th­ese are the same three statis­ti­cal reg­u­lar­i­ties I gave in the base­ball hy­po­thet­i­cal, and we should draw similar con­clu­sions. GiveDirectly is an ex­cel­lent cash-trans­fer char­ity. It’s on the GiveWell top char­i­ties list in large part be­cause it is tak­ing such care to col­lect ev­i­dence about what ac­tu­ally hap­pens when the global poor are sim­ply given money.

To that end, I found this Vox ar­ti­cle about GiveDirectly’s ba­sic in­come ex­per­i­ment in­ter­est­ing. In par­tic­u­lar, it’s in­ter­est­ing that the head­line case is some­one who some­times went hun­gry, but is not go­ing to spend the money on food. In­stead, she’s go­ing to spend it on what’s plau­si­bly a po­si­tional good in­stead:

She is ex­pect­ing her third child very soon. […] I asked Jack­lin if she’s ever gone the whole day with­out eat­ing; she has. I asked when the last time this hap­pened was. She told me, “Last week.”

But when the non­profit GiveDirectly told her that it would give her, and ev­ery other adult in her village, a ba­sic in­come pay­ment of 2,280 Kenyan shillings (about $22) a month for the next 12 years, she knew im­me­di­ately that she would not spend the money on food.

Her plan is to save the money and then use it to pay her chil­dren’s school fees.

She is starv­ing her­self, while preg­nant, in or­der to save for her kids’ for­mal school­ing. This looks like a re­ally bad out­come.

A brief di­gres­sion on education

But ed­u­ca­tion! She’s in­vest­ing in her kids! Isn’t that good?

The ed­u­ca­tion in­dus­try is already eat­ing the de­vel­oped world al­ive, with lit­tle ap­par­ent benefit. You might ar­gue that there are diminish­ing re­turns – in­tu­itively, ed­u­ca­tion seems im­por­tant. But, there’s am­ple ev­i­dence that de­vel­op­ing-world ed­u­ca­tion of­ten doesn’t re­ally im­prove peo­ple’s pro­duc­tive ca­pac­ity; it of­ten seems no bet­ter than obe­di­ence school, when it’s not be­ing used purely as a cer­tifi­ca­tion of the abil­ity to ob­tain a de­gree (and thus that you’re in the right so­cial class for cer­tain po­si­tions).

For a par­tic­u­larly dra­matic ex­am­ple, con­sider Scott’s re­port on his ex­pe­rience in Haiti:

Even if you’re one of the lucky ones who can af­ford to go to school, your first prob­lem is that the schools can’t af­ford pa­per: one of our hosts told sto­ries of Haitian high school­ers who were at the level of Western 5th graders be­cause they kept for­get­ting ev­ery­thing: they couldn’t af­ford the pa­per to take notes on!

The other prob­lem is more sys­temic: schools teach ev­ery­thing by un­in­spired lec­ture even when it’s com­pletely in­ap­pro­pri­ate: a worker at our camp took a “com­puter skills” course where no one ever touched a com­puter: it was just a teacher stand­ing in front of the class say­ing “And then you would click the word FILE on top of the screen, and then you’d scroll down to where it said SAVE, and then you’d type in a name for the file...” and so ob­vi­ously peo­ple come out of the class with no clue how to use an ac­tual com­puter. There’s the money is­sue—they couldn’t af­ford a com­puter for ev­ery stu­dent—and a cul­tural is­sue where ac­tu­ally go­ing to school is con­sid­ered noth­ing more than an an­noy­ing and rit­u­al­is­tic in­ter­me­di­ate step be­tween hav­ing enough money to go to school and get­ting a cushy job that re­quires ed­u­ca­tion.

Th­ese are both en­tirely con­sis­tent with a story where ed­u­ca­tion im­proves in­di­vi­d­ual out­comes by in­duct­ing peo­ple into the “ed­u­cated class”. And Scott con­tinues:

We heard hor­ror sto­ries of peo­ple grad­u­at­ing from nurs­ing school with­out even know­ing how to take a blood pres­sure—a nurse who used to work at the clinic would just make her blood pres­sure read­ings up, and give com­pletely non­sen­si­cal num­bers like “2/​19”. [W]hen cor­nered this nurse ab­solutely in­sisted that the blood pres­sure had been 219 and made a big fuss out of it.

Like­wise, bu­reau­crats also do not seem to be able to do the sorts of tasks we would ex­pect for­mal school­ing to qual­ify you for. One such task is alpha­bet­i­za­tion:

Gail, our pro­gram di­rec­tor, ex­plained that she has a lot of trou­ble with her Haitian office staff be­cause they don’t un­der­stand the con­cept of sort­ing nu­mer­i­cally. Not just “they don’t want to do it” or “it never oc­curred to them”, but af­ter months and months of at­tempted ex­pla­na­tion they don’t un­der­stand that sort­ing alpha­bet­i­cally or nu­mer­i­cally is even a thing. Not only has this messed up her office work, but it makes deal­ing with the Haitian bu­reau­cracy—har­row­ing at the best of times—pos­i­tively un­bear­able.

Gail told the story of the time she asked a city office for some pa­per­work re­gard­ing Doc­tors Without Borders. The lo­cal offi­cial took out a drawer full of pa­per­work and looked through ev­ery sin­gle pa­per in­di­vi­d­u­ally to see if it was the one she wanted. Then he started look­ing for the next drawer. After five hours, the offi­cial fi­nally said that the pa­per wasn’t in his office.

While in­vest­ing in ed­u­ca­tion to get a higher fu­ture salary is a net good in a sim­plified eco­nomic model, in prac­tice it’s of­ten just buy­ing what economists call rents. Even if those rents some­times take the form of a job, it’s clear from Scott’s ex­am­ple that school­ing is not en­abling gov­ern­ment bu­reau­crats to cre­ate more value for tax­pay­ers than less-schooled peo­ple would be able to do – they’re just out­com­pet­ing the un­schooled for a fixed pool of jobs where they don’t do much.

I sus­pect Haiti is es­pe­cially bad for a bunch of rea­sons, but I don’t know how ex­cep­tional it is. And it would be weird for marginal ed­u­ca­tion to be bad in ex­cep­tional bas­ket cases like Haiti, bad in well-to-do coun­tries like the US, but good in the mid­dle.

Here’s an­other anec­dote from a friend:

I had the op­por­tu­nity to ob­serve one poor school in In­dia and in­deed, school liter­ally didn’t hap­pen most days of the year for one rea­son or an­other. (Wait­ing for text­books was one rea­son I re­mem­ber them giv­ing.) Also, the teach­ers were steal­ing the food the ro­tary club pro­vided for free lunch. (Some­one no­ticed and then they stopped, which I sup­pose is nice.)

And of­ten when they showed up to school they were of­ten grad­ing pa­pers from other schools for sep­a­rate pay in­stead of teach­ing the kids.

At least they could get the teach­ers to show up to steal the lunches some­times.

Ban­er­jee and Du­flo’s Poor Eco­nomics is con­sis­tent with the view that this is a com­mon prob­lem in de­vel­op­ing coun­tries. In Chap­ter 4, they write:

In 2002 and 2003, the World Ab­sen­teeism Sur­vey, led by the World Bank, sent unan­nounced sur­vey­ors to a na­tion­ally rep­re­sen­ta­tive sam­ple of schools in six coun­tries. Their ba­sic con­clu­sion was that teach­ers in Bangladesh, Ecuador, In­dia, In­done­sia, Peru, and Uganda miss one day of work out of five on av­er­age, and the ra­tio is even higher in In­dia and Uganda. More­over, the ev­i­dence from In­dia sug­gests that even when teach­ers are in school and are sup­posed to be in class, they are of­ten found drink­ing tea, read­ing the news­pa­per, or talk­ing to a col­league. Over­all, 50 per­cent of teach­ers in In­dian pub­lic schools are not in front of a class at a time they should be. How are the chil­dren sup­posed to learn?
In 2005, Pratham, an In­dian NGO fo­cused on ed­u­ca­tion, de­cided to go one step fur­ther and find out what chil­dren were re­ally learn­ing. […] Close to 35 per­cent of chil­dren in the seven-to-four­teen age group could not read a sim­ple para­graph (first-grade level) and al­most 60 per-cent of chil­dren could not read a sim­ple story (sec­ond-grade level). Only 30 per­cent could do sec­ond-grade math­e­mat­ics (ba­sic di­vi­sion). […]
Un­for­tu­nately, In­dia is not unique: Very similar re­sults have been found in neigh­bor­ing Pak­istan, in dis­tan­tKenya, and in sev­eral other coun­tries. In Kenya, the Uwezo Sur­vey, mod­eled on ASER, found that 27 per­cent of chil­dren in fifth grade could not read a sim­ple para­graph in English, and 23 per­cent could not read in Kiswahili (the two lan­guages of in­struc­tion in pri­mary school). Thirty per­cent could not do ba­sic di­vi­sion. In Pak­istan, 80 per­cent of chil­dren in third grade could no tread a first-grade-level para­graph.

In ad­di­tion, par­ents seem to view ed­u­ca­tion as a way to buy a cre­den­tial, as a ticket into an “ed­u­cated class” job, not a way for their chil­dren to pick up valuable skills con­tin­u­ously:

Par­ents seem to see ed­u­ca­tion pri­mar­ily as away for their chil­dren to ac­quire (con­sid­er­able) wealth. The an­ti­ci­pated route to those riches is, for most par­ents, a gov­ern­ment job (as a teacher, for ex­am­ple), or failing that, some kind of office job.

Par­ents thus would rather pay for the com­plete ed­u­ca­tion of their high­est-po­ten­tial chil­dren, then pay for a mostly-com­plete ed­u­ca­tion for all their chil­dren. (Poor Eco­nomics does claim that the all-or-noth­ing cre­den­tial­ist view is un­re­al­is­tic and there are sub­stan­tial gains for each year of ed­u­ca­tion.)

I be­lieve very strongly that learn­ing is an im­por­tant form of real cap­i­tal, and it’s en­tirely pos­si­ble that for­mal school­ing is more like this in the de­vel­op­ing than in the de­vel­oped world, but I haven’t see the ev­i­dence, and at this point I think that if you try to jus­tify the benefits of a so­cial pro­gram by point­ing to the bare fact that peo­ple are us­ing it to buy more for­mal school­ing, I think this is mostly ad­verse rather than pos­i­tive ev­i­dence. (I’m more op­ti­mistic about ed­u­ca­tional pro­grams like SOLE that try to route around the hi­er­ar­chy, and fo­cus ex­clu­sively on learn­ing rather than cre­den­tials.)

Some in­vest­ment is real

While some in­vest­ments peo­ple make to bet­ter their or their chil­dren’s cir­cum­stances are po­si­tional, oth­ers seem much more like the sort of real in­vest­ment we might hope for, un­der the usual eco­nomic frame­work. For in­stance, one com­mon use of GiveDirectly’s cash trans­fers is to buy a metal roof, which lasts much longer than the more com­monly used and cheaper thatch roofs. The most con­ser­va­tive es­ti­mates GiveWell cites sug­gest that the an­nual re­turn on in­vest­ment for metal roofs is 7-14%, though the other es­ti­mates they re­port are sub­stan­tially higher.

Im­por­tantly, the re­turn on in­vest­ment for a tin roof is not plau­si­bly ex­trac­tive. Even though pur­chas­ing the roof im­poses a cost on the world (by in­creas­ing de­mand for the rele­vant in­puts), it also re­duces a cost (by re­duc­ing de­mand for thatch roof in­puts), and the re­duc­tion seems to be sub­stan­tially greater than the in­crease.

At least some re­cip­i­ents of cash trans­fers try to start busi­nesses, which is maybe the paradig­matic case of an in­vest­ment for which we should ex­pect a re­turn. But while lo­cals know their lo­cal econ­omy much bet­ter than I do, I am skep­ti­cal of a busi­ness plan where the ba­sic in­come is be­ing used to sub­si­dize vari­able rather than fixed costs. As Vox re­ports:

Sam­son [...] ex­plained his plan to go into cage fish­ing at the lake. He’d already bought the fish and just needed to buy feed, and the feed — per a cat­a­log he showed us — is ex­pen­sive. So he’s go­ing to use the GiveDirectly money for that part of the op­er­a­tion.

Busi­ness acu­men, like any­thing else, is a skill that can take time to de­velop—while there are se­ri­ous dis­ad­van­tages to mak­ing de­ci­sions for peo­ple from afar, we should at least not ex­pect that cash trans­fer re­cip­i­ents will im­me­di­ately make rea­son­able busi­ness de­ci­sions.

How should we think about cash trans­fers now?

I hope that most read­ers will be see that the point here is not that cash trans­fers are bad. The point is that when you look at the ev­i­dence about what hap­pens when rich peo­ple send poor peo­ple money, good news won’t look like a te­dious con­fir­ma­tion of an ob­vi­ous truth, but rather like an en­courag­ing out­come where one should have been ac­tu­ally un­cer­tain be­fore­hand. And you should be able to rec­og­nize am­bigu­ous or bad news as such, and not as­sume that it’s good by defi­ni­tion.

I hope that it will also mo­ti­vate some amount of ad­di­tional jus­tified skep­ti­cism of nom­i­nal rates of re­turn, as es­ti­mates of the so­cial value of an in­vest­ment. You, as some­one liv­ing in a rich coun­try, might not think that your ac­cess to higher-wage jobs is a re­li­able mea­sure of how much value you’re able to provide to the world. (If you did think this, the most benev­olent thing you could do would be to max­i­mize your nom­i­nal wealth.) If you are in fact skep­ti­cal of the mean­ingful­ness of your in­come as a met­ric, you should be similarly skep­ti­cal of the mean­ingful­ness of vari­a­tions in in­come of peo­ple in poor coun­tries.

Fi­nally, while ab­strac­tions like in­come lev­els and av­er­age self-re­ported hap­piness can be good start­ing points for gen­er­at­ing hy­pothe­ses, I hope I’ve been some­what per­sua­sive that they are not ad­e­quate met­rics for mak­ing philan­thropic de­ci­sions—one has to en­gage with what’s con­cretely hap­pen­ing in the world.

(Cross-posted on my per­sonal blog and the Effec­tive Altru­ism fo­rum.)