Political science and economics

by Henry on October 15, 2007

Tyler Cowen on the Nobel prize going to the mechanism design crowd.

In other words, no incentive scheme, no matter how clever, can get people to tell the truth. Grove, Clarke, Tideman, and Tullock lurk in the hallways. Note that a second price auction (let everyone bid and the winner pays the price of the next highest bid) fails in terms of Paretian optimality. The government takes the second price bid from the winner, but what should it do with the money? Either the government wastes resources by destroying wealth, or it redistributes that wealth in some way but then the resulting redistribution in turn feeds back into bids and we can no longer derive truth-telling as optimal (but is this really a practical problem?; my fear is that the entire incentive-compatibility literature has never gotten at the real reason why we don’t run the entire economy as a second-price auction.)

One of my favourite papers of all time, Gary Miller and Thomas Hammond’s “Why Politics is More Fundamental Than Economics: Incentive Compatible Mechanisms are Not Credible,” makes exactly this argument (link to abstract: full paper is behind paywall for non-academics, unfortunately) using clear language and simple mathematics. It also makes clear (a) that the problem doesn’t vanish if the surplus goes to a private actor rather than government, hence suggesting that many private sector schemes to elicit information etc are similarly problematic, (b) that one plausible historical solution has been to elicit the creation of bureaucratic norms of professionalism that encourage administrators not to behave as selfish rational actors and (c ) that the surplus problem is, properly considered, where politics enters into the argument, and a way of getting at the real reasons why we don’t run the economy using these mechanisms.

Also notable is another paper that Tyler links to on whether the Indian caste system was efficient or not. As Tyler notes politely, and Chris Hayes more pungently, this is a weirdly functionalist paper in the way that many economic analyses of institutions are weirdly functionalist. The professional deformity of the institutional economist is to seek explanations of institutional origins and change grounded in efficiency. In fairness, I should acknowledge that the professional deformation of the political scientist (and of many economic sociologists) is to seek explanations of institutions grounded in power and distributional questions, but it seems to me that this professional deformity gets things right a lot more often (institutions that are genuinely grounded in the desire to promote collective efficiency are relatively rare, and the Indian caste system is rather obviously not one of these rare exceptions).

Update: see further Jim Johnson for a more detailed account of how mechanism design “unintentionally …establishes the fundamental importance of politics“. On distribution v. efficiency, see also this very interesting new AJS article (via OrgTheory ) which seeks to assess the merits of distributional and efficiency theories in explaining the origins of transnational private regulation. Finally, those looking for some (mathematically pretty hairy) intro materials on mechanism design theory should go to Michael Greinecker.

{ 3 trackbacks }

Depth First Search » Blog Archive » Today’s Misc.
10.15.07 at 3:42 pm
Nobel Prize in Economics 2007 « Entertaining Research
10.15.07 at 4:27 pm
Crooked Timber: Why economics (and its Nobel) are political « Identity Unknown
10.15.07 at 4:37 pm

{ 27 comments }

1

Neel Krishnaswami 10.15.07 at 3:05 pm

Oddly enough, economics did help me figure out one of Chris Hayes’s rhetorical questions. He suggests the funny title “Extraordinary Rendition and Comparative Advantage: Does Outsourcing Torture Make Economic Sense?”

So, consider what the punishment for some crime (say, robbery) ought to be. The state might choose to fine the alleged robber or imprison him. In either case, the alleged robber loses something valuable (money or his liberty), which presumably serves as a deterrent for other would-be robbers.

Which ought we to support? The naive answer is that prison is expensive, so it’s more efficient to fine than to imprison. The government can use the money from a fine for useful purposes (e.g., restitution for the victims), whereas imprisonment is expensive, and pure deadweight loss.

However, when you do the public choice thing, we see that the word “alleged” becomes important. It’s also important that the government not have a big incentive to punish. If punishment is costly, then the government is less motivated to do it(*). But if it is low-cost or worse still profitable for the government, then you’re going to get a lot more of it, and many of those alleged robbers aren’t going to be actual robbers.

So the answer to Hayes’s question is: “No. Torture is cheap, so it’s a bad idea, and making it cheaper makes it a worse idea.”

(*) Tyler Cowen’s point about distribution is still important though, as the most casual glance at the political activity of prison guard unions will attest.

2

Matt Kuzma 10.15.07 at 3:05 pm

Reading this post was a lot like walking in on a fascinating conversation; this is exactly the sort of topic in economics that interests me, but I don’t know anything about it, really.

Could you point me to a good primer on the subject of mechanism design?

3

Michael Greinecker 10.15.07 at 3:20 pm

That one cannot settle redistributive questions in using the VCG-mechanism for financing public goods is nothing new. This was already shown in Green & Laffont 1979, who give an example where all possibe cost shares would make for an mechanism.

“Could you point to a good primer on the subject of mechanism design?”

Here

4

Jim Johnson 10.15.07 at 4:05 pm

On Henry’s point about insititutional economics, ROberto Unger rececntly refered to the bulk of that literature as “right-wing Hegelianism” to the extent that it posits the rational as real and the reverse. Unfortunately too many political sceintists follow suit and devote insufficient attention to distribution and power.

The Miller/Hammond essay gets at the basic point quite well ~ technocratic dreams of bypassing politics are hollow. The references Michael G. offers simply take an unnecessary detour through math-land. For my two-cents worth:
“http://politicstheoryphotography.blogspot.com/2007/10/2007-nobel-prize-economics-mechanism.html”:http://politicstheoryphotography.blogspot.com/2007/10/2007-nobel-prize-economics-mechanism.html

5

Sebastian Holsclaw 10.15.07 at 4:39 pm

“that one plausible historical solution has been to elicit the creation of bureaucratic norms of professionalism that encourage administrators not to behave as selfish rational actors and”

Now please realize that I didn’t reach behind the paywall, so the answer may be obvious if you have access to the whole paper: is this so much a solution or more of an often-failing strategy? The three societies which immediately leap to mind using that strategy on a large basis are China, India and the Caliphate. All three quickly saw the bureaucratic systems become deeply corrupt. Also the lengths taken to separate the bureaucrats from regular people aren’t really dooable today–the heavy reliance on eunuchs or people otherwise completely divorced from their families is noticeable. (Come to think of it, that is true of the bureaucracies of the Roman Catholic church for many of its middle years.)

The question is a matter of degree. The bureaucracies were pretty good at maintaining certain norms against assault by the random tyrannies of warring kings or clan leaders which have been so common throughout history. In that sense they were wildly successful in China and India, and somewhat successful in the Caliphate (very mixed in the Catholic Church). But compared to the more market oriented solutions which came later, they aren’t particularly successful. For the most part, bureaucracies are very conservative institutions. When information management/truth telling is about conserving it against destruction, they do a pretty good job. When you want to take the signals/truth telling and do something new with the information, they aren’t so good.

6

Slocum 10.15.07 at 4:48 pm

institutions that are genuinely grounded in the desire to promote collective efficiency are relatively rare, and the Indian caste system is rather obviously not one of these rare exceptions.

But is ‘desire’ necessary in such a case? That is, if the caste system were efficient that might explain its stability even if efficiency formed no part of the thinking of the system’s designers.

I’m definitely not claiming that the efficiency argument is right for the caste system, but only that, in general, an explicit desire for efficiency is certainly not required for efficiency to be an important consideration.

7

notsneaky 10.15.07 at 4:54 pm

“Could you point me to a good primer on the subject of mechanism design?”

Michael Greinecker of Yet Another Sheep, who called the prize (well, at least Hurwicz) has a list of some primers:

http://yetanothersheep.blogspot.com/

But beware, it’s mathy stuff.

8

Henry 10.15.07 at 5:04 pm

Sebastian – the paper – which I am happy to send to you if you are interested – bases its arguments on the experience of Progressive reformers in the US. The extent to which these lessons travel to other contexts is an open question (and all bureaucracies are prone to devolve into rent seeking over time – but the point is, as Weber also argues, that bureaucratic norms of impartiality etc do have consequences for behaviour apart from any sanctioning mechanisms that may be associated with them).

slocum – that an institution is efficient is certainly a plausible mechanism that may aid the institution’s survival. But it isn’t the only one – highly inefficient institutions have survived for very long periods of time. This is, in a sense, the problem that Doug North confronts in his work (although he is of two minds about efficiency-based explanations of change, which mean that his microfoundations are a bit hand-wavy in places). Jack Knight’s book _Institutions and Social Conflict_ is the classic explanation of why efficiency-based arguments are likely to apply only under highly limited circumstances. As Knight argues, power based considerations are more likely to be more important most of the time – institutions will persist when they are in the distributional interests of actors powerful enough to ensure that they persist. I note for honesty’s sake that Jack is a very good friend of mine, that we have collaborated on work applying these arguments, and that my own take on these issues is basically his with a few bells and whistles.

9

Drake 10.15.07 at 5:09 pm

Why can’t the solution require both a floor wax (incetives) and a dessert topping (political or professional norms)?

10

Jim Johnson 10.15.07 at 6:23 pm

Prompted by the exchange between Sebastian & Henry, and siding mostly with Henry, it is important to keep in mind a few things.

First, the causal story for the emergence of some insititution (market, norm, hierarchy etc.) typically will not invoke efficiency (or any other nice collective concern) as anything other than a fortuitous by-product. (Hence, Jack is right more often than “the right-wing Hegelians” who peddle institutionalism in ecoon-world.)

Second, markets are ‘efficient’ only under restrictive conditions. Exchange absent those conditions need not be effient or otherwise admirable. So simply invoking ‘markets’ or even ‘market-oriented’ solutions as against ‘bureaucracies’ is simply hand-waving. One needs to offer some reason to think the relevant initial conditions obtain or might be brought into being.

Finally, we need not make a decision to use any of the available insititutional mechanisms to coordinate all domains. we need to make much more fine-grained decisions.

11

notsneaky 10.15.07 at 6:34 pm

Hmm, I should probably read the comments/posts more carefully next time. Or have some coffee first.

Anyway. Inefficient institutions are also bad for those in power since they leave money on the table which the powerful could otherwise grab. There’s nothing in efficiency that says it’s good or just or whatever. And if you’re trying to explain an institution that has lasted for 3000 years, as the caste system has, through various social upheavals and changes, I don’t see how you can ignore efficiency considerations. While the cited article may be too formalistic, it seems like the caste system has had to achieve some goal and facilitate some purpose in order to have been so resilient and long lasting.

Compare that with the institution of slavery, another inefficient institution, which a lot of the time – with the very obvious exception of the US – pretty much ended on its own. If you were a Greek slave you pretty much just paid a fee to your owner and went about your business. The slave population of the Roman Empire fell from a peak at the time of Augustus to pretty much zero at the time of Valerian. In central and eastern Europe it pretty much disappeared by the 13th century (serfdom appeared later) though that has been explained by rise of Christianity. Etc.

12

notsneaky 10.15.07 at 6:46 pm

While we’re on the topic, Acemoglu, Johnson and Robinson in “Colonial Origins of Comparative Development” give three reasons why “bad” institutions may persist, the first two being essentially independent of power considerations;
1. Large costs of creating new institutions which are sunk once these are created. So even when better institutions are available it may not be worthy to change, and
2. Investments made by individuals in a society which are complimentary to a given set of institutions. This is basically the “you learn to learn with it” or the “better the devil you know” argument. Bad institutions may suck but you know how to live with them whereas you don’t know how to function with these new promised good institutions. A good part of disenchantment with the post-communist govs in Eastern Europe probably has something to do with this reason.

AJR’s third reason is power related – they basically argue that if the size of the elite who gets to set up/control the institutions is small then the benefits to each member of the elites from the bad-for-society/good-for-elite institutions is much larger. But bad institutions mean that if you improved them the size of the pie would grow. So if you could guarantee the elites their old incomes (which would be affordable) you could make everyone else better off too. But this brings up the “commitment problem”, which is subject of Acemoglu’s and Robinson’s “Economic Origins of Dictatorship and Democracy” which I highly recommend.

I’m not sure if this falls within the realm of Hegelian Rightwingism or not. Or New Institutionalism for that matter. Another guy who’s written a lot on it but is not a HRer is Avner Grief.

13

Jim Johnson 10.15.07 at 6:56 pm

If an institutional arrangement (such as slavery or universal suffrage) is an equilibrium then efficiency considerations need not have anything to do with its persistence. There are, after all, all sorts of inefficient equilibrium outcomes. And there is no reason to think we arrive at any equilibrium in ‘nice’ ways.

An institutional arrangement typically comes into being as a by-product of distributional conflict among differentially endowed actors. And it might persist due to, say, the inability of those disadvantaged by the way it operates to coordinate their opposition to it (due to simple collective action problems). Likewise, whether those who are advantaged by the operation of some institution might obtain even greater benefits by some other equilibrium institutions is insufficient to move away from an existing to a (for them) ‘superior’ arrangement for similar reasons. The problem with most economics of institutions thinking is the tacit assumption that there is some underlying mechanism driving institutional emergence and change in efficient ways. Smells Hegelian to me.

14

mq 10.15.07 at 7:00 pm

I’ve always thought Gary Miller was one of the more underrated social scientists out there. But his work is almost totally inaccessible and obscure. His papers aren’t even posted on his faculty web site, and his book “Managerial Dilemmas” is not seen around very much. POST YOUR PAPERS, GARY MILLER!

Also, anyone who thinks that inefficient institutions cannot survive knows nothing, and I mean nothing, about human history. Or human nature. It is highly unlikely that people who do well under the current system will risk what they currently have for some hypothetical future efficiency gain.

15

Jim Johnson 10.15.07 at 7:00 pm

PS: The fact that much of the institutional literature smacks of right wing Hegelianism (or as Henry, being much more polite than I, puts it – “functionalist”) may not be due to the conscious commitments of the theorists. My own view is that it stems from an inability to keep three theoretical tasks – analytical, explanatory and normative – sorted out. Jack Knight and I address this in our paper in APSR (March 07) should you be interested.

16

notsneaky 10.15.07 at 7:21 pm

“If an institutional arrangement (such as slavery or universal suffrage) is an equilibrium then efficiency considerations need not have anything to do with its persistence”

It depends whether it’s a stable or unstable and unique equilibrium. The thinking on this goes that when a shock occurs which opens up an opportunity of institutional change then better institutions might come about. Given a long enough time and a lot of those “mights” eventually, they will come about. So inefficient institutions might persist for a hundred, two hundred, maybe even five hundred years. But three thousand + is a really really long time.

But as I said over at MR, if you a Greg Clark style Malthusian then the only institutions that matter are those related to fertility and mortality. Everything else – including anything which affects productivity – gets eaten up by Malthusian forces. So, with the two caveats above, there’s no really anything like “inefficient” and “efficient” institutions. Standard of living is just proportional to mortality divided by fertility.

17

Walt 10.15.07 at 8:01 pm

I’m confused by the use of “efficient” on this thread. Do people mean “Pareto efficient”? Pareto efficiency is not at all incompatible with the importance of power or distribution (though it is a terriby misleading term).

18

Jim Johnson 10.15.07 at 8:02 pm

I’d say that for most political and economic agents a century or five is a really long time. Even if they think about future generations, how far out will they go? And, of course, the fact that some ineffcieint equilibrium crumbles due to a shock implies nothing about the chances that efficiency will drive the search for a successor. My bet is that bargaining among differentially endowed agents will be the underlyig causl mechanism that will generate a replacement.

19

notsneaky 10.15.07 at 8:08 pm

Yes, it’s generally meant to be Pareto. And yes, that’s the point. Pareto gives you a set. Power considerations or distributional issues should pick a point from this set. If this whole efficiency of institutions in very long run story has plausibility.

20

Henry 10.15.07 at 8:40 pm

notsneaky – I don’t think the argument here is about Pareto efficiency. The kinds of claim that Jim is making and that Jack Knight makes are compatible with Pareto efficiency – that is, they are working from arguments about bargaining power as modeled by mixed motive coordination games, in which the equilibrium outcomes are all Pareto efficient (i.e. for interested bystanders who haven’t done economics, they are outcomes in which no actor can be made better off without making another actor worse off). The point that they are making is more subtle. There is often a rhetorical slippage from Pareto efficiency to more general notions of social efficiency, and to the claim that institutions can be justified in terms of the collective efficiency benefits that they provide.

And that is surely what we arguing about here. The castes piece isn’t really about PE – in it, ‘efficiency’ seems to mean something like ‘maximizing trade and productive activity.’ More efficient arrangements have more exchange and production, less efficient ones have less. There is _no good theoretical reason_ to suppose that arrangements that are not efficient in this broad sense of the word won’t persist for very long periods of time, if they are in the distributional interests of actors powerful enough to keep them in place. To put it another way – powerful actors aren’t interested in maximizing collective benefits except when that maximizes their private gains. They will, if they are self interested rational actors, prefer institutional arrangements that give them a larger overall slice of a smaller pie, to arrangements that give them a smaller overall slice of a larger pie. In particular, they will disfavor proposed new institutional arrangements that might increase overall productivity but challenge their own privileged position. And shocks aren’t likely to change this – unless the shocks in question are ones that reduce their bargaining power (which gets back to what Jim and Jack and I have argued and are arguing).

21

Barry 10.15.07 at 10:18 pm

notsneaky, re the caste system in India: “So inefficient institutions might persist for a hundred, two hundred, maybe even five hundred years. But three thousand + is a really really long time.”

The first obvious question is – did ‘it’ persist for such a length of time. Or were there varying systems, with much of the history looking at them through the viewpoint of the present and near-past?

22

notsneaky 10.15.07 at 11:22 pm

“The castes piece isn’t really about PE – in it, ‘efficiency’ seems to mean something like ‘maximizing trade and productive activity.’ “

I haven’t read through the whole thing so perhaps I’m mistaken, but it seems to model the caste system as an institution which permits coordination on the “good outcome” in a Prisoner’s Dilemma. Since the “bad outcome” in a PD is not Pareto Efficient, this paper at least does seem to be about PE. You’re probably right about the larger HR NI literature.

Barry, you’re perfectly right.

I gotta go cook dinner.

23

Henry 10.16.07 at 1:18 am

notsneaky – there’s a fair amount of language suggesting otherwise – e.g. on p.4 when it talks about how increasing the number of occupations that are the monopoly of one cast increases efficiency. Obviously, you can’t ‘increase’ PE – a situation is either PE or it isn’t and that is about all you can say. But there is, as Tyler notes, some disconnect between the formal model and the empirical account, which generates some of these ambiguities (in addition to the criticisms outlined by Jim above, there often seems to be a lot of confusion in articles like this in translating the model into theoretical predictions that have empirical ramifications).

24

Sebastian Holsclaw 10.16.07 at 3:49 am

Yes, please feel free to email me. It sounds interesting

25

Tracy W 10.16.07 at 10:35 am

institutions that are genuinely grounded in the desire to promote collective efficiency are relatively rare

However:
a) institutions that are extremely inefficient will not survive long (because they won’t produce enough food to keep the population alive).
b) institutions that are less efficient than their neighbours will have less resources to devote to warfare, increasing their odds of being militarily defeated.
c) Institutions that make people better off will tend to attract people towards them, to the extent that people are capable of moving.

One of the things that baffles me about the Indian caste system paper, apart from the lack of empirical testing, is that the author never seemed to address why castes rather than guilds.

The paper’s maths imply that every country, or every country without an effective legal system, can only be efficient by using castes and a purity hierarchy, but we know that isn’t so. Clans and guilds can provide bargaining advantages as well.

Note, I am using efficiency here as something approximating the physics concept of output/input. I am not using it in the sense of pareto-efficient. I am well aware that calculating economic output/inputs is far harder than calculating say the heat efficiency of a gas furnance.

26

Martin Wisse 10.16.07 at 10:38 am

Nitpick: there’s no Nobel Prize in Economics, only a prize instituted in the memory of Alfred Nobel. Bad enough every bloody newspaper gets it wrong.

27

notsneaky 10.16.07 at 2:55 pm

Re 29

Actually in a long lost version of Alfred Nobel’s will it says:

Three Prizes for the Science-kings under the sky,
One for the Word-lords in their halls of stone,
Two for Peaceful and Healthy Men doomed to die,
One for the Dark Lords on their dark throne
In the Land of Ikan where the Shadows lie.
One Prize to rule them all, One Prize to find them,
One Prize to bring them all and in the darkness bind them
In the Land of Ikan where the Shadows lie.

Comments on this entry are closed.