The Second Best

In eco­nomics, the ideal, or first best, out­come for an econ­omy is a Pareto-effi­cient one, mean­ing one in which no mar­ket par­ti­ci­pant can be made bet­ter off with­out some­one else made worse off. But it can only oc­cur un­der the con­di­tions of “Perfect Com­pe­ti­tion” in all mar­kets, which never oc­curs in re­al­ity. And when it is im­pos­si­ble to achieve Perfect Com­pe­ti­tion due to some un­avoid­able mar­ket failures, to ob­tain the sec­ond best (i.e., best given the con­straints) out­come may in­volve fur­ther dis­tort­ing mar­kets away from Perfect Com­pe­ti­tion.

To me, per­haps be­cause it was the first such re­sult that I learned, “sec­ond best” has come to stand gen­er­ally for the yawn­ing gap be­tween in­di­vi­d­ual ra­tio­nal­ity and group ra­tio­nal­ity. But similar re­sults abound. For ex­am­ple, in So­cial Choice The­ory, Ar­row’s Im­pos­si­bil­ity The­o­rem states that there is no vot­ing method that satis­fies a cer­tain set of ax­ioms, which are usu­ally called fair­ness ax­ioms, but can per­haps be bet­ter viewed as group ra­tio­nal­ity ax­ioms. In In­dus­trial Or­ga­ni­za­tion, a duopoly can best max­i­mize prof­its by col­lud­ing to raise prices. In Con­tract The­ory, ra­tio­nal in­di­vi­d­u­als use up re­sources to send sig­nals that do not con­tribute to so­cial welfare. In Public Choice The­ory, spe­cial in­ter­est groups suc­cess­fully lobby the gov­ern­ment to im­ple­ment in­effi­cient poli­cies that benefit them at the ex­pense of the gen­eral pub­lic (and each other).

On an in­di­vi­d­ual level, the fact that in­di­vi­d­ual and group ra­tio­nal­ity rarely co­in­cide means that of­ten, to pur­sue one is to give up the other. For ex­am­ple, if you’ve never cheated on your taxes, or slacked off at work, or lost a mu­tu­ally benefi­cial deal be­cause you bar­gained too hard, or failed to in­form your­self about a poli­ti­cal can­di­date be­fore you voted, or tried to mo­nop­o­lize a mar­ket, or an­noyed your spouse, or an­noyed your neigh­bor, or gos­siped mal­i­ciously about a ri­val, or sounded more con­fi­dent about an ar­gu­ment than you were, or took offense to a truth, or [in­sert your own here], then you prob­a­bly haven’t been in­di­vi­d­u­ally ra­tio­nal.

“But, I’m an al­tru­ist,” you might claim, “my only goal is so­cietal well-be­ing.” Well, un­less ev­ery­one you deal with is also an al­tru­ist, and with the ex­act same util­ity func­tion, the above still ap­plies, al­though per­haps to a lesser ex­tent. You should still cheat on your taxes be­cause the gov­ern­ment won’t spend your money as effec­tively as you can. You should still bar­gain hard enough to risk los­ing deals oc­ca­sion­ally be­cause the money you save will do more good for so­ciety (by your val­ues) if left in your own hands.

What is the point of all this? It’s that group ra­tio­nal­ity is damn hard, and we should have re­al­is­tic ex­pec­ta­tions about what’s pos­si­ble. (Maybe then we won’t be so eas­ily dis­ap­pointed.) I don’t know if you no­ticed, but Pareto effi­ciency, that so called op­ti­mal­ity crite­rion, is ac­tu­ally in­cred­ibly weak. It says noth­ing about how con­flicts be­tween in­di­vi­d­ual val­ues must be ad­ju­di­cated, just that if there is a way to get a bet­ter re­sult for some with oth­ers no worse off, we’ll do that. In in­di­vi­d­ual ra­tio­nal­ity, its ana­log would be some­thing like, “given two choices where the first bet­ter satis­fies ev­ery value you have, you won’t choose the sec­ond,” which is so triv­ial that we never bother to state it ex­plic­itly. But we don’t know how to achieve even this weak form of group ra­tio­nal­ity in most set­tings.

In a way, the difficulty of group ra­tio­nal­ity makes sense. After all, ra­tio­nal­ity (or the po­ten­tial for it) is al­most a defin­ing char­ac­ter­is­tic of in­di­vi­d­u­al­ity. If in­di­vi­d­u­als from a cer­tain group always acted for the good of the group, then what makes them in­di­vi­d­u­als, rather than in­ter­change­able parts of a sin­gle en­tity? For ex­am­ple, don’t we see a Borg cube as one in­di­vi­d­ual pre­cisely be­cause it is too ra­tio­nal as a group? Since achiev­ing perfect Borg-like group ra­tio­nal­ity pre­sum­ably isn’t what we want any­way, maybe set­tling for sec­ond best isn’t so bad.