Roemer's ridiculous Kantian equilibrium.

Prof. John Roemer has introduced the notion of Kantian- as opposed to Nash- equilibria such that, in a symmetric game, where everybody is in the same boat, the Kantian strategy is to do what everyone would think were the best thing to do for everybody. This is Pareto efficient.

Roemer says- 'I do not claim that Kantian optimization is rational in one-shot games. What I argue is that in many situations, the same players counteract with each other all the time, and a morality of solidarity may develop that induces players to optimize in the Kantian manner, if they come to feel a sense of solidarity. It is not my task here to show how this emerges; rather, I want to show that if Kantian optimization is a moral protocol in a population, then many inefficiencies in market economies can be resolved. I suggest readers think of Kantian optimization as a moral code.'

Roemer gives a 'tragedy of the commons' type of example of fishermen on a lake. They each voluntarily restrict their catch by reducing Effort by the same proportion. Roemer says this is Pareto efficient both in the game and in the economy.

Is he right? No. Some fishermen should find another job where there is no negative externality. That's good for the game- the remaining fisher men can expend more effort for a greater reward- and it is good for the economy because more labor and enterprise becomes available for other activities.

If fishermen won't quit for a Kantian reason then it would make sense to license only those whose transfer earning are very low to continue fishing. Those with high transfer earnings, which may rise over time thanks to 'acquired advantage', should not get licenses. Further, the underlying problem could be tackled directly- e.g. by removing invasive species, developing sustainable fish farming practices etc.

'Tragedy of the commons' can be very good things. The reason the Western Europeans and their descendants are so much better off than Bengalis is that they over-fished their inland and coastal waters and thus had to develop trans-oceanic navigational technology. They could then accumulate surpluses through trade and conquest which they could use for capital intensive, technological, industrialization. No doubt, they suffered greatly from pollution and social injustice and so forth. London had 'pea-soup' fogs when my father first visited. By the time I got to England in the Seventies, such events were a distant memory. Fish had returned to the River Thames. Dickensian slums had been replaced by modern flats with indoor lavatories and central heating. This had nothing to do with 'Kantian' morality or 'solidarity' or any such thing. Rather, necessity had been the mother of invention. Grave crises occurred and this caused Governments to develop wholly novel methods of tackling problems at their root.

What is Roemer's motivation? He has a bone to pick with behavioural economics. He says 'An experimenter observes in the lab that subjects do not play what the experimenter believes is the Nash equilibrium of the game (think of trust games, public-good games, the ultimatum and dictator games). So the experimenter looks for ‘exotic’ preferences, which would, if held by the subjects, produce the observed outcome as a Nash equilibrium. This is done by inserting arguments like altruism, a concern for fairness, a concern for equality, etc., as arguments of preferences. In contrast, my approach is to keep preferences classical, but to alter the way that agents optimize. I believe my approach is superior, because it decentralizes the cooperative solution even when it is not obvious what that solution is.

In other words, instead of saying people have 'altruistic' preferences- which is a truth we all know about ourselves- Roemer is saying we use an optimization rule of an abstract type- which we all know we don't use at all because we are not trained in mathematical economics.

It is a common occurrence that we make mistakes when we try to choose the most economical solution. We misread the price-sticker or fail to check we have enough money in the bank. On the other hand we are certain we like pizza and hate sushi. Our preferences are things we know and are seldom mistaken about. Rules of optimization, however, are a mystery to most of us and we very often make mistakes when calculating our best course of action.

 Contrast this with the lab games that I listed above: in all those cases, we can immediately see what the cooperative solution is, and so it is not so hard to design preferences that will make that solution a Nash equilibrium of the game so defined. 

Nor is it hard to pretend people are using an optimization rule which they don't in fact know and can't use because they are crap at math.

But in many cases, the cooperative solution is not obvious. Take the fishing example above: it is far from obvious what the Pareto efficient allocation in which each keeps his catch is. (Under general conditions of convexity, such an allocation exists, and is locally unique: see Roemer and Silvestre [1993].) 

Yes, but in general, it is not effectively computable. For a large enough lake, it is very unlikely that fishermen know exactly how many other fishermen have taken out their boats. Furthermore, the would look blankly at you if you asked them what 'general conditions of convexity' means in plain English.

But the multiplicative Kantian equilibrium locates it without resorting to inventing new preferences for the players. 

So does the topological Nietzchean equilibrium and the Tensor Calculus Minskowskian equilibrium and the motivic cohomology Rahul Gandhian equilibrium.

Consider the price of blow jobs at the truck stop. It is blindingly obvious that Zizek and his ilk, who- it is my fond belief- exclusively supply the commodity in question, do so using a neo-Voevodskian model because Zizek pays his clients in a cryptocurrency denominated in complex numbers with a real argument of zero.

How would you go about assigning new preferences to the fishers so that the Nash equilibrium of the new game is the Pareto efficient allocation in which each keeps his catch? In other work, I have shown there is no natural way to do this (Roemer [in press]).

Nor is it natural to start gassing on about 'multiplicative Kantian equilibria'. The natural thing is to say 'Academic Economists are stupid cunts.' and find some other object for one's derision.

Roemer solves the problem of Carbon trading by introducing a notion of a 'Walras-Kant' equilibria. He says 'there are no ex ante limits on emissions, and no ex ante allocation of emissions credits to countries. So these two contentious problems in the discussion of global emissions’ control are solved by the use of Kantian optimization – that is to say, by cooperation in the choice of emission ‘supplies.’ The citizens supply the permission to the countries in toto to emit. The agreement among countries specifies that firms may not emit until it is verified that total emissions will be no greater (in fact equal to) the citizenry-determined total supply of emissions.
How the fuck are citizens- people like you and me- supposed to hit upon the total 'supply' of emissions? Scientists try to estimate what current emissions are but those estimations vary greatly. There is no consensus on this highly contentious matter. Moreover, if we can't stop people killing each other in distant countries- or indeed on the streets of our own cities- how are we supposed to ensure enterprises all over the world will abide by what is in fact an ex ante limit? Suppose a particular country decides to enforce the compact. How does it know others will also do so? If some cheat, perhaps this country should prohibit all emissions so the target is met. But that would spark a 'Yellow Vest' type revolt. Laws are costly to enforce. The world does not possess the resources to police a global Carbon trading regime.

Mathematical Economists, notoriously, understand human psychology about as well as a potted plant.

Roemer concludes his paper in a hilarious fashion-

 The psychology of Kantian optimization The differentia specifica of the models here proposed is Kantian optimization in the labor-supply decision (or in the emissions decision). Having a formal definition of cooperation is, obviously, a pre-condition to embedding cooperation in equilibrium models. It will likely be the case that skepticism regarding my proposal will focus upon the realism of supposing that a large population of producers can learn to optimize their labor-supply (or emissions) decisions in the Kantian manner.

Nonsense! Everybody knows that people wake up in the morning saying to themselves- 'I must reduce my emissions by 2.76 percent and raise my work effort by 1.53 percent today'.  Suppose you have a hangover. Then wifey reminds you at the breakfast table 'Darling do remember to reduce your emissions by 2.76 percent. The way you were farting in bed means that you have already emitted 87.95 percent of your methane quota. Also you must increase your work effort by 1.53 percent. Otherwise the Kantian equilibrium will be disrupted.'

Who in their right mind would be skeptical about Roemer's path-breaking work? Even little babies are able to keep meticulous accounts of their emissions.

There are, I think, three necessary conditions for the psychological accessibility of such behavior: desire, understanding, and trust.

Very true. Wifey desires, understands and trusts us. That is why she very helpfully reminds us about how much we have to reduce our emissions or raise our work effort before kissing us and sending us off to work singing 'my baby takes the morning train so as to contribute to the Kantian equilibrium relating to emissions and work effort. Then, the fucker comes home drunk and blackens my eyes. I'm gonna Bobbit him but good one of these fine days coz his emissions smell godawful.'

Citizens/workers must desire to cooperate with each other, they must view themselves as part of a solidaristic society, whose members believe that cooperation in economic decisions is the modus operandi.

Citizen/ workers must desire to cooperate with each other, they must view themselves as part of solidaristic society or else we will Gulag their sorry ass.

But why should the Kantian optimization protocol appeal to people as the preferred mode of cooperation? I think the motivation must be in the conception of fairness or solidarity embodied in the statement, “I should only reduce (increase) my labor supply if I would like all others to reduce (increase) their labor supplies in like manner.”

Very true! The reason I won't work is coz there are coma patients who can't work. Not till every last coma patient is able to increase her labor supply will I do likewise.

Our brains love symmetry, and fairness always, I believe, involves a conception of symmetrical treatment.

The reason I won't get out of bed is because it would be unfair to coma patients.

Secondly, people must understand that cooperation in the labor-supply decision internalizes the externalities that are improperly treated with Nash optimization. That’s what Benjamin Franklin was appealing to when he uttered his famous phrase about hanging together, which later became a motivational slogan in the American labor movement. For instance, in the market-socialist model, each must understand that if all increase their labor supply by a small increment, each person’s income increases by the wage times that increment, because what a worker loses in the tax on her wage, she gains back in the increased value of the demogrant.

Demogrant? Sign me up. If coma patients get one, I too must receive equal treatment.

Thirdly, individuals must trust that others will behave cooperatively as well, and will not take advantage of their own cooperative behavior, by optimizing in the Nash manner.

One way of establishing trust is by killing or otherwise excluding people we don't trust coz maybe they look a bit different or follow a different faith.

If these three conditions are met, then the method of implementing cooperative behavior is not difficult:

Nor is the method of implementing the Thousand Year Reich under our Great Leader!

for instance, in the market-socialist model, each worker should choose his labor supply to equalize his marginal rates of substitution between commodities and labor to his gross real wage, rather than his after-tax wage.

Why stop there? Why not get him to equalize his marginal rate of substitution to whatever the Great Leader tells him to? After all, the tax taken out of his gross real wage may be being used for nice Gulags and Concentration Camps where 'Arbeit macht frei'

Rather than thinking “Is the disutility of an extra day’s work worth to me the after-tax wage increment?” the worker should ask whether it is worth the gross wage increment.

That way he won't resent his tax money being used for cool stuff like Gas Chambers.

If we believe people are capable of optimizing in the Nash manner, optimizing in the Kantian manner is no more cognitively demanding, if the necessary conditions are met.

The necessary conditions are precisely the ones which would make Nazism or Stalinism the best of political systems.

To return to my earlier comment, these results suggest that the market (conjoined with price-taking behavior) is an even more powerful allocation mechanism than standard theory suggests. In many cases -- I do not have a complete characterization of them – inefficiencies of market equilibrium are due not to ‘the market,’ but to the behavioral protocol of Nash optimization. This is a mathematical claim, based on the efficiency theorems above, which is true regardless of the realism of Kantian optimization.

Mathematical claims are only true mathematically. They are not true about anything which actually exists.

 In games played by members of a society with a common culture, where members have learned to trust each other, due to a history of repeated interactions characterized by cooperative behavior, Kantian optimization may, however, become a moral norm.

Thus we should get rid of immigrants and people of other Faiths because they don't belong to our common culture.

This has been tried. It has failed. Innovation grinds to a halt. There is high trust but transactions are low value. Smart people emigrate.

How that comes to be is another story. I believe there are many examples in real life of simple Kantian equilibrium: many people recycle their garbage, even if penalties for failing to do so do not exist; voting can be viewed as a simple Kantian equilibrium; the British ‘doing their bit’ in the two world wars is a simple Kantian equilibrium or perhaps an additive Kantian equilibrium; the degree of tax compliance in most advanced democracies is far greater than can be rationalized by reasonable risk preferences and existing penalties, and is perhaps better understood as due to Kantian optimization (I pay taxes because that’s the action I’d like all to take); participation in labor strikes and demonstrations may be more convincingly explained as Kantian behavior than Nash behavior ( à la Olson [1965]). I have discussed these examples in more detail elsewhere (Roemer [in press]).

Patriotism- such as that exhibited by the Brits 'doing their bit'- is not Kantian. We wanted to kill as many Germans as possible. That is why women flocked to the munitions factories and men rushed to enlist.

Participating in labor strikes is not Kantian. It is an attempt to capture a rent. Voluntary Recycling is 'virtue ethics', not Kantian, because it makes one feel better about oneself in a way which may not obtain if everybody does it. Voting for mandatory recycling is Kantian.

I do not claim to have airtight proof that people are, indeed, optimizing according to a Kantian protocol in the above examples (and many others), but observation is consistent with this explanation.

Because Roemer's significant other is constantly reminding him to reduce his emissions by 2.56 percent while increasing his work effort by 3.72 percent. He says 'shut the fuck up, you old harridan! I know perfectly well what my contribution to the Walras-Kant equilibrium should be today. Everybody knows these things the moment they wake up in the morning. So stop your nagging or I will punch your lights out.'

And let us not belittle the suggestive role of theory: once we have a precise model of a behavior, we may be stimulated to look for it in history and in the laboratory, and be surprised at how often it turns up.

Why would be surprised? Is it not  a fact that at every moment in the day we are aware of how much we must increase our work effort or reduce our emissions so as to comply with the Kantian equilibrium?

OMG, just farted very loudly. My emissions have exceeded my allotment for today. Whatever shall I do?

Oh. Right. I need to reduce my work-effort because that has a knock on effect of reducing other people's emissions. Thus, I must quit my job and go on disability. Come to think of it, Roemer's Kantian equilibrium isn't as silly as it sounds.