UBI for President

Cross posted, as always, from Pu­tanu­monit.

Epistemic sta­tus: I am not an economist, and nei­ther is An­drew Yang. On the other hand, it’s not like the ex­perts are brim­ming with great solu­tions to the prob­lems dis­cussed herein, or are united in con­sen­sus on the main is­sues.


An­drew Yang is run­ning for pres­i­dent in 2020 on a plat­form of Univer­sal Ba­sic In­come. Last week I got a chance to hear from An­drew di­rectly and ask him a few ques­tions about it. I came away cau­tiously op­ti­mistic about try­ing UBI (up from cau­tiously neu­tral), and mas­sively im­pressed by An­drew Yang (up from not hav­ing heard of him). It’s time to talk about UBI, and it’s time to pay at­ten­tion to Yang 2020.

UBI and the La­bor Au­toma­tion Tsunami

Yang lays out the case for ur­gent UBI on Sam Har­ris’ pod­cast, in an in­ter­view with Quillette, and in great de­tail in his book, The War on Nor­mal Peo­ple. I’ll try do jus­tice to the idea with a brief sum­mary.

Who is “the me­dian Amer­i­can“? It is a per­son with­out a col­lege de­gree, with­out much of a sup­port net­work, and with­out $500 to spare in case of an emer­gency. She works in re­tail (4.3 mil­lion jobs) or a call cen­ter (2.5 mil­lion). He’s a fast food worker (3.7 mil­lion) or a truck driver (3.5 mil­lion). Th­ese mil­lions of jobs are be­ing au­to­mated to­day and will keep be­ing so in the im­me­di­ate fu­ture, along with ware­house work­ers, ac­coun­tants, and ra­diol­o­gists.

Millions of Amer­i­cans are about to meet the fate of the 2 mil­lion or so con­struc­tion work­ers who lost their jobs af­ter the fi­nan­cial crisis. What are those 2 mil­lion up to? In short: idle­ness, video games, Oxy­con­tin, loneli­ness, de­spair. As their num­bers swell, and McKinsey pre­dicts that 15% of all jobs [1]. will be dis­placed by au­toma­tion by 2030, this cheer­ful list may grow to in­clude crime and ri­ot­ing. What solu­tions are Amer­i­can poli­ti­ci­ans offer­ing to this loom­ing crisis?

Trump blames the cheat­ing for­eign­ers, even though the num­ber of Amer­i­cans who benefit eco­nom­i­cally from trade (roughly all of them) swamps the num­ber of Amer­i­can work­ers who lose their job as a re­sult of trade (a mere 90,000 a year). Bernie blames Ama­zon and Wal­mart, the two com­pa­nies that be­sides pro­vid­ing mas­sive con­sumer sur­plus are also the em­ployer-of-last-re­sort to more than 2 mil­lion low-skil­led Amer­i­can work­ers. Other poli­ti­ci­ans mum­ble about “re­train­ing”, but cur­rent gov­ern­ment re­train­ing pro­grams are both tiny and in­effi­cient. There is no vi­able plan to scale them to mil­lions of work­ers, and no rea­son to ex­pect that they will work.

By and large, the main­stream po­si­tion on la­bor force au­toma­tion among US poli­ti­ci­ans is to ig­nore it. The worse the prob­lem gets, the stronger the de­sire to ig­nore it grows.

An­drew Yang’s solu­tion is to provide ev­ery adult Amer­i­can cit­i­zen with $1,000 a month, no strings at­tached. This will be paid for by a com­bi­na­tion of a con­sump­tion tax (like a 10% VAT), re­duc­tion in other welfare pay­ment (one can re­ceive UBI xor ex­ist­ing welfare), and “ad­di­tional rev­enues from eco­nomic growth”. I per­son­ally think that the lat­ter is short­hand for “more taxes”, ideally, of the Ge­or­gist va­ri­ety.

Here’s what Yang says will hap­pen when ev­ery Amer­i­can gets $12k a year guaran­teed:

  • Peo­ple will be en­couraged to work as UBI elimi­nates welfare cliffs and gives peo­ple the slack to move to bet­ter jobs /​ bet­ter cities.

  • Peo­ple will start busi­nesses and do cre­ative, non­profit and care work.

  • Peo­ple will be hap­pier and healthier with a solid safety net be­neath them.

  • Bureau­cratic over­head will be re­duced be­cause the one thing our gov­ern­ment knows how to do effi­ciently is send­ing a lot of peo­ple checks each month.

Here a cou­ple of out­comes that I worry may hap­pen when ev­ery Amer­i­can gets $12k a year guaran­teed:

  • Ul­ti­mately, a UBI is a trans­fer of money. The net re­cip­i­ents of that money will be the broke and un­em­ployed (great), but also ren­tiers and land­lords (not great).

  • Women will spend more time on house­work and rais­ing kids (great), but men will spend more time play­ing video games (not great?).

I asked An­drew about both points, and the ex­tent of our dis­agree­ment seems to lie in differ­ent gen­eral mod­els of eco­nomics and hu­man be­hav­ior.

What UBI Buys

An­drew and I both no­ticed that Amer­i­cans spend a lot of their in­come on hous­ing, health­care, and ed­u­ca­tion – hence­forth, HH&E. But we have differ­ent an­swers to why this is the case.

From what I gath­ered, An­drew’s ba­sic model is that Amer­i­cans spend a lot of money on HH&E be­cause they (HH&E) are ex­pen­sive. Un­der this as­sump­tion, if Amer­i­cans had ex­tra money they could then af­ford to spend it on other things that will make them hap­pier. An­drew’s policy plat­form con­tains a lot of ideas to make HH&E cheaper. For ex­am­ple, reg­u­lat­ing the ra­tio of ad­minis­tra­tors to stu­dents at uni­ver­si­ties, or sin­gle-payer health­care with flat salaries rather than pay-per-ser­vice for doc­tors. With the ex­tra in­come, Amer­i­cans will get more and bet­ter ser­vices.

My model is that Amer­i­cans spend a lot of money on HH&E be­cause they (Amer­i­cans) are rich. Un­der this as­sump­tion, if Amer­i­cans had ex­tra money HH&E would in­crease in price to ab­sorb any ex­tra dis­pos­able in­come. This is a some­what coun­ter­in­tu­itive model, and it is built of the fol­low­ing com­po­nents:

  1. For mea­sur­ing how much money peo­ple have to spend, Ac­tual In­di­vi­d­ual Con­sump­tion(AIC) is vastly bet­ter than mea­sures like GDP per cap­ita.

  2. Go­ing by AIC, Amer­i­cans are way richer than the rest of the world – about 50% richer than most de­vel­oped ar­eas like Europe and Ja­pan.

  3. Most Amer­i­cans can af­ford a full suite of goods whose sup­ply is not con­strained: food, cloth­ing, elec­tron­ics, trans­porta­tion, etc.

  4. Not ev­ery Amer­i­can can af­ford the ex­pen­sive items in the above cat­e­gories: or­ganic quinoa, de­signer jeans, the lat­est iPhone, a new car. The differ­ence be­tween those and the ver­sions af­ford­able to ev­ery Amer­i­can: rice, H&M jeans, Nokia 2, a 2008 Toy­ota is al­most en­tirely a differ­ence of sig­nal­ing, not of qual­ity.

  5. Really, all Amer­i­cans should read my guide to buy­ing things smartly.

  6. HH&E all have their sup­ply con­strained in one way or an­other by the gov­ern­ment: the num­ber of houses built, the num­ber of hos­pi­tals in a city and num­ber of doc­tors cre­den­tialed, the num­ber of ac­cred­ited schools and teach­ers.

  7. HH&E is get­ting more ex­pen­sive with­out get­ting bet­ter. Apart­ments in San Fran­cisco, health­care out­comes, skills ac­quired in school – none of them have im­proved much in the last few decades but all have mul­ti­plied in price.

  8. HH&E in­volves a huge amount of zero-sum sig­nal­ing.

  9. Here’s a thor­ough anal­y­sis show­ing that US Health­care costs are ad­e­quately ex­plained by Amer­i­can’s ex­tra AIC, and we should ex­pect health­care to be less of a zero-sum sig­nal­ing com­pe­ti­tion than houses and schools. Espe­cially schools.

Bring­ing it all to­gether: the ma­jor­ity of Amer­i­cans have more money than it takes to cover one’s non-HH&E needs. What­ever money is left over is spent by rich Amer­i­cans on com­pet­ing for a limited sup­ply of nice neigh­bor­hoods, pres­ti­gious schools, im­pres­sive health­care. The cost dis­ease trick­les down: if Columbia uni­ver­sity can hire 1,000 ad­minis­tra­tors and charge $240,000 for a four-year de­gree that teaches few use­ful skills, lower ranked schools can also get away with more bloat, higher tu­ition, and less ed­u­ca­tion. Ris­ing rents in San Fran­cisco push hous­ing prices up 80 miles away in Gilroy.

So: a $1,000/​month UBI will prob­a­bly in­crease the dis­pos­able in­come of most Amer­i­cans, at the ex­pense of the top 10-20%% who will pay a lot more in ex­tra taxes. Giv­ing money to the very poor­est Amer­i­cans will en­able them to buy bet­ter food, clothes etc. But the ex­tra money given to ev­ery­one else, which will be the ma­jor­ity of UBI, will go to land­lords, school ad­minis­tra­tors, health­care providers, and lo­cal mo­nop­o­lies – all with­out in­creas­ing the qual­ity of HH&E ser­vices pro­vided one bit.

Un­doubt­edly, both An­drew’s model and mine are sim­plis­tic, in­com­plete, and im­pre­cise. Nei­ther of us is dog­mat­i­cally com­mit­ted to ei­ther view, the true an­swer cer­tainly lies some­where in the mid­dle, or even to the side. But on the mar­gin, my model makes UBI sound a lot less at­trac­tive than it would be oth­er­wise.

UBI and Work

An­drew Yang, cit­ing the Roo­sevelt In­sti­tute think tank, claims UBI will in­crease the num­ber of work­ing Amer­i­cans by 4 mil­lion. Un­for­tu­nately, that’s not what the Roo­sevelt In­sti­tute says: they ac­tu­ally put the num­ber at 1.1 mil­lion [2], and even that is based on a macro model rather than on ob­served ev­i­dence.

The ob­served ev­i­dence is mixed, and very lit­tle of it is ob­vi­ously rele­vant to ba­sic in­come that is per­ma­nent, uni­ver­sal, and im­ple­mented in a gi­ant and rich na­tion. Pu­tanu­monit raised thou­sands of dol­lars for a ba­sic in­come pro­ject in Kenya, but that was de­signed to help Kenyans. The fact that a Kenyan starts work­ing af­ter get­ting UBI be­cause he can af­ford a wheelbar­row to cart sup­plies doesn’t mean that an Amer­i­can truck driver who lost his sup­ply-cart­ing job to a robot will use UBI to find al­ter­na­tive em­ploy­ment.

An­drew’s story goes some­thing like this: af­ter los­ing his job, the truck driver is stuck. He doesn’t have spare money for train­ing or re­lo­ca­tion, and go­ing on welfare limits his abil­ity to move and try things, es­pe­cially if it’s dis­abil­ity welfare. With UBI he could af­ford to move to a big city, pay rent and vo­ca­tional school train­ing for a few months, and rein­vent him­self as a plumber or A/​C re­pair­man or strip­per.

My story goes some­thing like this: peo­ple work be­cause they need money. If they need money less, they will work less. I’ll have to see a lot of ev­i­dence to con­tra­dict this sim­ple story. An­drew him­self agreed that for most peo­ple who drop out of the la­bor force, UBI will not pull them back in.

But as I wrote a cou­ple of years ago: this is not a de­ci­sive ar­gu­ment against UBI. Work­ing hours are f=“http<strong>://​ben­jam­in­rosshoff­man.com/​costs-are-not-benefits/​“>not a benefit to be max­i­mized, they’re a cost. John May­nard Keynes fa­mously pre­dicted that we would all work at most 15 hours a week. He made that pre­dic­tion shortly be­fore work­ing him­self to death. But the rea­son we don’t work 15 hours a week is the weird equil­ibrium we’re in of what is val­ued by so­ciety.

Hu­mans don’t in­trin­si­cally value “hours worked”. We value things like sta­tus, sex, com­mu­nity, plea­sure. In mod­ern so­ciety, we learned to as­so­ci­ate a lot of this with work and con­sump­tion. This is es­pe­cially true of men, which is why men left out of the work-con­sump­tion cy­cle fall into greater de­spon­dency than women.

Here’s An­drew:

I will say that if you dig into the data, you find that men and women ex­pe­rience idle­ness differ­ently. […] The data shows that women who are out of work get in­volved in the com­mu­nity and go back to school and do things that are quite pro­duc­tive and pro-so­cial. Whereas, men who are out of work spend 75 per­cent of their time on the com­puter play­ing videogames and sur­fing porn—and then tend to de­volve into sub­stance abuse and self-de­struc­tive be­hav­iors. Men who are out of work vol­un­teer less than em­ployed men, even though they have more time.

Put­ting on the cyn­i­cal evolu­tion­ary psy­chol­ogy hat I bor­rowed from Ge­offrey Miller, I would guess that un­em­ployed men vol­un­teer less be­cause vol­un­teer­ing doesn’t get them re­spected or laid. That’s the ul­ti­mate rea­son men do things (don’t ask me how it goes with blog­ging).

But putting on my op­ti­mistic evo-psych hat I no­tice that 20,000 years ago Homo sapi­ens­males com­peted for sex and sta­tus by tel­ling sto­ries, or see­ing who can throw a rock farthest, or paint­ing hands on the walls of a cave. This can prob­a­bly be ap­prox­i­mated to­day by play­ing bas­ket­ball or backgam­mon with your bud­dies, or by com­pet­ing for karma on a Dun­bar-sized sub­red­dit. If we stop den­i­grat­ing peo­ple who do this, their lives may not be so mis­er­able. I think that a lot of the “sub­stance abuse and self-de­struc­tive be­hav­iors” fol­low not from play­ing video games, but from feel­ing guilty and shamed over play­ing video games. A safety net of so­cial re­spect can be as im­por­tant as a safety net of cash.

Of course, if ev­ery­one played The Witcher 3 all day there would be no one to de­velop The Witcher 4: Witcherer than Ever. So­ciety needs peo­ple to be pro­duc­tive to grow and pros­per. But not ev­ery­one can be equally pro­duc­tive.

The me­dian truck driver is 49, high school-ed­u­cated, and has few skills other than driv­ing a truck. It seems some­what ar­bi­trary to blame him (truck­ers are 94% male) for not guess­ing 25 years ago that truck­ing will get au­to­mated be­fore what­ever other jobs he may have cho­sen. The US has mas­sive re­serves of pro­duc­tivity and growth in the mil­lions of skil­led im­mi­grants who would come given the chance, and in prepar­ing the next gen­er­a­tion for a 21st-cen­tury econ­omy. As one of the former plan­ning to have some of the lat­ter, I can af­ford to pay for some trucker-wire­head­ing.

Bot­tom line: I’m not very op­ti­mistic about UBI as a panacea for those left be­hind by au­toma­tion, but I think it’s prob­a­bly worth the ex­per­i­ment. De­spite the po­ten­tial benefits, no other coun­try seems will­ing to take up the gaunt­let. Amer­i­cans can af­ford it, and it’s not cer­tain that we can af­ford to con­tinue ig­nor­ing the prob­lem of la­bor au­toma­tion.

Pu­tanu­monit En­dorses Yang 2020

Iron­i­cally, I found al­most ev­ery­thing else about An­drew Yang more im­pres­sive than his defense of UBI.

Yang struck me as thought­ful, cu­ri­ous, and hum­ble, and yet with enough charisma to not let those three traits en­tirely sub­marine his poli­ti­cal prospects. He’s a sec­ond-gen­er­a­tion im­mi­grant with no whiff of iden­tity poli­tics. His back­ground is in so­cial en­trepreneur­ship: he cre­ated and sold an ed­u­ca­tion com­pany, then founded a non-profit that cre­ates jobs in cities like Cleve­land and Pitts­burgh.

More im­por­tant than An­drew’s per­son­al­ity and street cred are his poli­tics. Yang is liberal but pro-busi­ness and skep­ti­cal of gov­ern­ment’s abil­ity to do many things well aside from cut­ting checks and pass­ing sim­ple reg­u­la­tions. The 71 non-UBI points on Yang’s policy plat­form can be de­scribed as “the most sen­si­ble thing that can squeeze in­side the Demo­crat Party Over­ton win­dow, erring on the side of cau­tion and in­cre­men­tal­ism”.

And Yang has a plan – to get into the na­tional spotlight by sneak­ing up on Iowa.

The long tor­tu­ous pro­cess of Amer­i­can pres­i­den­tial elec­tions starts with the Iowa cau­cuses. “Cau­cus” means that in­stead of just cast­ing a vote, Iowans have to gather some­where and yell at each other for a few hours; last time out only 170,000 peo­ple both­ered show­ing up. This means that 30,000 votes or so may be enough for a top 2 or top 3 finish, hope­fully spark­ing a pos­i­tive cas­cade of me­dia cov­er­age and pop­u­lar­ity.

In An­drew’s words: “Any time 50 Iowans gather in a room and speak the words ‘Ba­sic In­come’, I ap­pear in a puff of smoke to tell them about it.” The mes­sage is well re­ceived in a pur­ple state with an econ­omy based on man­u­fac­tur­ing and agri­cul­ture.

Will this strat­egy work? Most peo­ple you know will al­most cer­tainly not be­come the pres­i­dent, and that is prob­a­bly true of An­drew Yang as well. And yet, Yang got an avowed poli­ti­cal pas­sivist fired up about his can­di­dacy enough to write 3,000 words about it and en­courage all of you to spread the word. I do this for two rea­sons.

The first comes from Eliezer Yud­kowsky, who asked us to Stop Vot­ing for Nin­com­poops.

I se­ri­ously think the best thing you can do about the situ­a­tion, as a voter, is stop try­ing to be clever. Don’t try to vote for some­one you don’t re­ally like, be­cause you think your vote is more likely to make a differ­ence that way. Don’t fret about “electabil­ity”. Don’t try to pre­dict and out­wit other vot­ers. Don’t treat it as a horse race. Don’t worry about “wast­ing your vote” – it always sends a mes­sage, you may as well make it a true mes­sage.
Re­mem­ber that this is not the an­ces­tral en­vi­ron­ment, and that you won’t die if you aren’t on the win­ning side. Re­mem­ber that the threat that vot­ers as a class hold against poli­ti­ci­ans as a class is more im­por­tant to democ­racy than your fights with other vot­ers. For­get all the “game the­ory” that doesn’t take fu­ture in­cen­tives into ac­count; real game the­ory is fur­ther-sighted, and be­sides, if you’re go­ing to look at it that way, you might as well stay home. When you try to be clever, you usu­ally end up play­ing the Poli­ti­ci­ans’ game.
Clear your mind of dis­trac­tions…
And stop vot­ing for nin­com­poops.
If you vote for nin­com­poops, for what­ever clever-sound­ing rea­son, don’t be sur­prised that out of 300 mil­lion peo­ple you get nin­com­poops in office.
The ar­gu­ments are long, but the vot­ing strat­egy they im­ply is sim­ple: Stop try­ing to be clever, just don’t vote for nin­com­poops.
Oh – and if you’re go­ing to vote at all, vote in the pri­mary.

In my years in the US, I have seen sev­eral dozen pres­i­den­tial can­di­dates. I think that An­drew Yang is the first one that isn’t a nin­com­poop.

The sec­ond rea­son comes from Mil­ton Fried­man (who, by the way, made the case for a form of UBI 50 years ago).

Only a crisis – ac­tual or per­ceived – pro­duces real change. When that crisis oc­curs, the ac­tions that are taken de­pend on the ideas that are ly­ing around. That, I be­lieve, is our ba­sic func­tion: to de­velop al­ter­na­tives to ex­ist­ing poli­cies, to keep them al­ive and available un­til the poli­ti­cally im­pos­si­ble be­comes the poli­ti­cally in­evitable.

This I be­lieve is An­drew Yang’s ba­sic func­tion: to keep the ideas of UBI and sen­si­ble Demo­crat-tol­er­able eco­nomic policy al­ive and available for when the crisis comes. Even if Yang doesn’t make it all the way to the White House, there is prob­a­bly no bet­ter way to get his ideas out there. And with the crisis of au­toma­tion and un­em­ploy­ment com­ing sooner rather than later, we are go­ing to need those ideas.


[1] Yang’s web­site states that: “The smartest peo­ple in the world now pre­dict that a third of all work­ing Amer­i­cans will lose their job to au­toma­tion in the next 12 years.” Yang also men­tioned the num­ber of dis­placed jobs as 30% on Sam’s pod­cast, and said that he got this figure from the McKinsey re­port on the fu­ture of work.

How­ever, the re­port only es­ti­mates the num­ber of jobs that are po­ten­tially re­place­able, and gives a range of 0-30% with 15% be­ing the me­dian es­ti­mate.

[2] I found this re­port by the Roo­sevelt In­sti­tute mod­el­ing the macroe­co­nomic im­pact of UBI. Table 3 shows the po­ten­tial la­bor force in­crease un­der var­i­ous UBI sce­nar­ios. The rele­vant sce­nario is num­ber 12: $1,000 a month funded by in­creased taxes. The es­ti­mated im­pact of sce­nario num­ber 12 is an in­crease of 1.11 mil­lion jobs. 4 mil­lion ex­tra jobs will only be added if UBI comes en­tirely from deficit spend­ing.

I didn’t ex­pect that when writ­ing a column in praise of An­drew Yang I’ll end up call­ing him out for mis­re­port­ing num­bers, but the pri­ori­ties of Pu­tanu­monit are clear: truth in num­bers first, poli­tics fifty sev­enth.