Unconscious Economies

Here’s an in­sight I had about how in­cen­tives work in prac­tice, that I’ve not seen ex­plained in an econ text­book/​course.

There are at least three ways in which in­cen­tives af­fect be­havi­our: 1) via con­sciously mo­ti­vat­ing agents, 2) via un­con­sciously re­in­forc­ing cer­tain be­havi­our, and 3) via se­lec­tion effects. I think per­haps 2) and prob­a­bly 3) are more im­por­tant, but much less talked about.

Ex­am­ples of 1) are the fol­low­ing:

  • When con­tent cre­ators get paid for the num­ber of views their videos have… they will de­liber­ately try to max­imise view-count, for ex­am­ple by craft­ing vague, click­baity ti­tles that many peo­ple will click on.

  • When sales­peo­ple get paid a com­mi­sion based on how many sales they do, but do not lose any salary due to poor cus­tomer re­views… they will se­lec­tively boast and ex­ag­ger­ate the good as­pects of a product and down­play or sneak­ily cir­cum­vent dis­cus­sion of the down­sides.

  • When col­lege ad­mis­sions are partly based on grades, stu­dents will work re­ally hard to find the teacher’s pass­word and get good grades, in­stead of do­ing things like be­ing in­de­pen­dently cu­ri­ous, ex­plo­ra­tory and try­ing to deeply un­der­stand the subject

One ob­jec­tion you might have to this is some­thing like:

Look at those peo­ple with­out in­tegrity, just try­ing so hard to op­ti­mise what­ever their in­cen­tives tell them to! I my­self, and in­deed most peo­ple, wouldn’t be­have that way.
On the one hand, I would make videos I think are good, and hon­estly sell prod­ucts the way I would sell some­thing to a friend, and make sure I un­der­stand my text­book in­stead of just mem­o­ris­ing things. I’m not some kind of microe­co­nomic robot!
And on the other hand, even if things were not like this… it’s just re­ally hard to cre­atively find ways of max­imis­ing a tar­get. I don’t know what ap­peals to ‘the kids’ on YouTube, and I don’t know how to find out ex­cept by pay­ing for some huge sur­vey or some­thing… hu­man brains aren’t re­ally de­signed for do­ing max­imis­ing like that. I couldn’t op­ti­mise in all these clever ways even if I wanted to.

One re­sponse to this is:

Without en­gag­ing with your par­tic­u­lar ar­gu­ments, we know em­piri­cally that the con­clu­sion is false. There’s a wealth of econo­met­rics and micro pa­pers show­ing how de­mand shifts in re­sponse to price changes. I could dig out plenty of refer­ences for you… but heck, just look around.
There’s a $10.000/​year day­care close to where I live, and when the moms there take their kids to the cin­ema, they’ll tell them to pre­tend they’re 6 and not 7 years old just to get a $3 dis­count on the tick­ets.
And I’m pretty con­fi­dent you’ve had per­sua­sive sales­peo­ple ped­dle you some­thing, and then went home with a lin­ger­ing sense of re­gret in your belly…
Or have you ever seen your friend in a queue some­where and ca­su­ally slid in right be­hind them, just to get into the venue 5 min­utes ear­lier?
All in all, if you give peo­ple an op­por­tu­nity to earn some money or time… they’ll tend to take it!

This might or might not be a good re­ply.

How­ever, by ap­peal­ing to 2) and 3), we don’t have to make this re­sponse at all. The effects of in­cen­tives on be­havi­our don’t have to be con­sciously me­di­ated. Rather...

  • When con­tent cre­ators get paid for the num­ber of views their videos have, those whose nat­u­ral way of writ­ing ti­tles is a bit more click­bait-y will tend to get more views, and so over time ac­cu­mu­late more in­fluence and so­cial cap­i­tal in the YouTube com­mu­nity, which makes it harder for less click­bait-y con­tent pro­duc­ers to com­pete. No one has to change their be­havi­our/​or their strate­gies that much—rather, when chang­ing in­cen­tives you’re chang­ing the rules of game, and so the win­ners will be differ­ent. Even for those less for­tu­nate pro­duc­ers, those of their videos which are on the click­bait end of things will tend to give them more views and money, and in­so­far as they just “try to make videos they like, see­ing what hap­pens, and then do­ing more of what worked”, they will be pushed in this direction

  • When sales­peo­ple get paid a com­mis­sion based on how many sales they do, but do not lose any salary due to poor cus­tomer re­views… em­ploy­ees of a more Machi­avel­lian char­ac­ter will tend to perform bet­ter, which will give them more money and so­cial cap­i­tal at work… and this will give Machi­avel­lian char­ac­ter­is­tics more in­fluence over that work­place (be­fore even tak­ing into ac­count re­turns to scale of cap­i­tal). They will then be in po­si­tions of power to de­cide on which new poli­cies get im­ple­mented, and might choose those that they gen­uinely think sound most rea­son­able and well-ev­i­denced. They cer­tainly don’t have to mer­cilessly op­ti­mise for a Machi­avel­lian cul­ture, yet be­cause they have all been pre-se­lected for such per­son­al­ity traits, they’ll tend to be bi­ased in the di­rec­tion of choos­ing such poli­cies. As for their more “no­ble” col­leagues, they’ll find that out of all the tac­tics they’re com­fortable with/​able to ex­e­cute, the more sales-y ones will lead them to get more hi-fives from the high-sta­tus peo­ple in the office, more room in the bud­get at the end of the month, and so forth

  • When col­lege ad­mis­sions are partly based on grades… the case is left as an ex­er­cise for the reader.

If this is true and im­por­tant, why doesn’t stan­dard econ text­books/​courses ex­plain this?

I have some hy­pothe­ses which seem plau­si­ble, but I don’t think they are ex­haus­tive.

1. Selec­tion pres­sure for ex­pla­na­tions re­quiring the fewest in­fer­en­tial steps

Microe­co­nomics is pretty coun­ter­in­tu­itive (for more on the im­por­tance of this, see e.g. this post by Scott Sum­ner). Writ­ing text­books that ex­plain it to hun­dreds of thou­sands of un­der­grads, even just us­ing con­sciously schem­ing agents, is hard. Now both “se­lec­tion effects” and “re­in­force­ment learn­ing” are in­de­pen­dently difficult con­cepts, which the ma­jor­ity of stu­dents will not have been ex­posed to, and which aren’t the ex­plana­tory path of least re­sis­tance (even if they might be re­ally im­por­tant to a small sub­set of peo­ple who want to use econ in­sights to build new or­gani­sa­tions that, for ex­am­ple, do bet­ter than the dire state of the at­ten­tion econ­omy. Such as LessWrong).

2. Fo­cus on math­e­mat­i­cal modelling

I did half an MSc de­gree in eco­nomics. The fo­cus was not on in­tu­ition, but rather on some­thing like “ac­quire math­e­mat­i­cal tools en­abling you to do a PhD”. There was a lot of fo­cus on not mess­ing up the mul­ti­vari­able calcu­lus when solv­ing strange op­ti­mi­sa­tion prob­lems with solu­tions at the bound­ary or in­volv­ing util­ity func­tions with awk­ward kinks.

The ex­tent of this math­e­ma­ti­sa­tion was some­times scary. In a fi­nance class I asked the tu­tor what prac­ti­cal uses there were of some ob­scure deriva­tive, which we had spend 45 mins and sev­eral pages of stochas­tic calcu­lus prov­ing the­o­rems about. “Oh” he said, “I guess a few years ago it was used to scheme Ital­ian grand­mas out of their pen­sions”.

In classes when I didn’t bother ask­ing, I mostly didn’t find out what things were used for.

3. Fo­cus on the prop­er­ties of equil­ibria, rather than the pro­cesses whereby sys­tems move to equilibria

Clas­sic econ joke:

There is a story that has been go­ing around about a physi­cist, a chemist, and an economist who were stranded on a desert is­land with no im­ple­ments and a can of food. The physi­cist and the chemist each de­vised an in­ge­nious mechanism for get­ting the can open; the economist merely said, “As­sume we have a can opener”!

Stan­dard micro deals with un­bounded ra­tio­nal agents, and its ar­se­nal of fixed point the­o­rems and what-not re­veals the state of af­fairs af­ter all max­i­mally ra­tio­nal ac­tions have already been taken. When asked how equil­ibria man­i­fest them­selves, and emerge, in prac­tice, one of my tu­tors hel­plessly threw her hands in the air and laughed “that’s for the macroe­conomists to work out!”

There seems to be lit­tle at­tempts to teach stu­dents how the solu­tions to the un­bounded the­o­rems are ap­prox­i­mated in prac­tice, whether via con­scious de­ci­sion-mak­ing, se­lec­tion effects, re­in­force­ment learn­ing, memet­ics, or some other mechanism.

Thanks to Niki Shams and Ben Pace for read­ing drafts of this.