On Liberty vs. Efficiency

To "win" a debate you aren't supposed to tip your opponent to the arguments you'll use.  But to promote a more productive conversation, that is exactly what you might do.  So in this post I'll lay out my basic (rather technical) argument for tonight's debate.  I've said:

The topic, as I see it, is the relative value/importance for economists of pushing "liberty," i.e., a policy of minimal government interference, and "efficiency," a standard policy evaluation metric that attempts to neutrally weigh policy consequences for different people.

Humans often find themselves in conflicts where they might make (and enforce) "deals" instead of "fighting" (or doing "nothing").  Such conflicts are often complex enough for many parties to be uncertain how they would fare, relative to fighting, under various possible deals.  In such situations, I see a noble and important role for expert arbiters who are "neutral," i.e., who develop deserved reputations for suggesting "win-win" deals where most or all parties should expect to benefit, relative to fighting.  Given access to such neutral expert advisers, conflicting parties can make better deals, to their mutual benefit. 

One reason I'm proud to be an economist is that we often fill this neutral expert arbiter role to varying degrees, and could do so even more if we tried.  And "efficiency," also known as "cost-benefit analysis," helps make this possible.  To estimate the efficiency of a deal, relative to a status quo, one adds up estimates for each person of the dollar value that person would place on this deal. 

Each person's dollar value of a deal is (minus) the compensation he or she would need to become indifferent to (i.e, not care to choose) this deal plus compensation package, relative to the status quo.  This says how much he wants the outcomes produced by this deal (and may or may not agree with what he says he wants).  The sum of such dollar values says how efficient, or good, is the deal; if it is positive, the deal is good; else it is bad.  

Of course since estimates can err, one can be mistaken about who wants a deal how much.  So when cash compensations can be cheaply implemented, economists recommend using individual value estimates to design cash transfers that maximize the chance everyone benefits from a deal.  We also recommend the meta-deal of consistently implementing efficient-enough deals (plus cash transfers when cheap) across many disputed topics.  The law of large numbers ensures that unless our estimate errors are strongly correlated, the more topics we add to a meta-deal, the more confident we can be that everyone will benefit, in the sense of getting more of what they want. 

If you believed that economists fairly and consistently estimated individual deal values, then there is only one way you could expect not to benefit from the meta-deal of consistently following their advice across many topics.  This would be if you had private information showing you were consistently hurt more than they expect, in topic after topic.  This would have to be info economists cannot see, and that you cannot credibly communicate.  Such a scenario is possible, but extreme.

The real problem is not such extreme possibilities, but the assumption that economists fairly and consistently estimate individual deal values.  If economists are "partisan," i.e., if they consciously or unconsciously try to favor some folks over others when estimating efficiency, and if the social institutions that combine individual contributions do not sufficiently balance or correct for such partisan bias, then economists' net advice could also be partisan.  This would make it more likely, though hardly imply, that many folks might on net lose from the meta-deal of consistently following economists' deal advice.

Economists have many social conventions that reduce partisan influences:

  1. We prefer efficiency as a neutral evaluation criterion. 
  2. We prefer clear simple math models with explicit assumptions, and prefer particular standard assumptions; these allow models to be more easily compared, and discourage searching for favorable assumptions. 
  3. We similarly prefer explicit statistical data analysis, and prefer particular standard statistical models and techniques, again facilitating comparison and discouraging search.  
  4. Peer reviewers watch for partisan bias, and interest groups fund economists to flag partisan contributions by opposing interests. 

Even so, I must admit that partisan influences pervade professional economists, often due to shared partisan sympathies among the most influential economists.  Yes, this limits the usefulness of following economists' advice, especially on politically charged and high profile topics.  And I am saddened to see little interest in institutions, such as prediction markets, that could substantially reduce such partisan bias. 

Yet, do not let the best be the enemy of the good; economists have much relevant expert knowledge, and their advice does on average help folks find win-win deals.  Economic efficiency is tarnished, but still powerful.  No other large community of expert advisers has anywhere near the economists' deserved reputation for consistently suggesting win-win deals. The more that economists agree, and the further from politically charged topics, the more your can trust their advice.

What else can we rely on, besides economists' expert advice, to suggest good deals to conflicting parties?  A variety of cheap heuristics are available, of varying quality.  The height heuristic, i.e., prefer the outcome suggested by the tallest person, has little to recommend it.  The majority rule heuristic, i.e., prefer the outcome a majority says they prefer, is better but can give multiple conflicting answers, can depend on how issues are bundled, can suffer from saying diverging from wanting, and can poorly reflect an unequal status quo.

"Liberty" is the heuristic of always preferring outcomes that minimize government involvement.   As cheap and easy-to-apply heuristics go, this one is pretty good.  Given two random deals the freer one is probably more efficient.  Given two actually proposed deals, instead of two random ones, my guess is that the freer deal tends to be more efficient.  But this is not a claim that economists as a whole endorse; this is just my personal guess.  And freer deals are clearly not always more efficient; governments can solve real and important coordination problems. 

Some propose liberty not as a heuristic for finding agreeable deals, but as a rallying call to arms. They want "good" folks to join in fighting "bad" folks who oppose liberty, not because this will help good folks get what they want, but just because such a war is the "right" thing to do.  But note that if everyone thought that a certain act was right, and they all wanted enough to act to do that right, then that right act would also be efficient.  Conflicts between "right" and efficiency only occur because some people do not want what is "right."

In my role as one person among many, who can join fights or support deals, I will choose in order to get what I want.  This may sometimes include joining a fight, and sometimes that fight might be to achieve more liberty.  But in my role as economic adviser, a role I admire and embrace, I will try to fairly and consistently suggest the most efficient deals, as the availability of such advisers offers a great opportunity for everyone to get more of what they want. 

I accept that such deals may not always contain the most liberty possible, because while people do usually want liberty, all else equal, they often want other things that conflict with liberty.  In my role as a neutral adviser, it is not my place to tell people they should want something other than they do want; my job is just to get them more of what they want.  Since this is exactly what I would want an expert adviser to do for me, it is what I will do for them.  If loving them in this way is wrong, I don't want to be right.

Added: Note the odd position you put yourself in if you embrace liberty as an efficiency hueristic because of economists' advice, yet refuse to accept other economists' deal advice.

GD Star Rating
Tagged as: , ,
Trackback URL:
  • Grant

    I’m happy to see someone refer to “liberty” or “non-aggression” as a heuristic. I think this is the best description of it.

    Robin seems to have much less skepticism than I (and probably Bryan) have on the ability of economists to estimate the costs and benefits of a deal. As I see economics today, it has produced a very small number of very important, true insights, and a very large number of arguments which we can’t identify as being true or false (though we know most of them are probably false). Such insights would include the benefits of free trade, the inefficiencies caused by externalities, the problems with wage and price controls, etc.

    In the past, political leaders have produces vast and complex arrays of propaganda (often religious) to legitimize their rule. It isn’t clear to me that quite a bit of economics isn’t a modern version of this. Economists are just people, and are self-interested and non-neutral. As a layman, the non-aggression heuristic seems a better rule-of-thumb than trusting in the benevolence of economists and political elites. The benefit of non-aggression is that it is simple enough for anyone to understand (at least in most situations), while economics is definitely not.

    I’m going to guess at the things Bryan will object to in this argument:

    -You seem to place a lot of faith in the ability of economists to estimate the costs and benefits of a deal. Efficiency is not objectively measurable (and may be entirely subjective). It seems to me it is so rarely measurable that the liberty heuristic almost always useful.

    -Economists have little training in estimating the costs of government solutions. They use detailed models to estimate how much a political action will enhance welfare, but they often do not even consider the costs of political action and political power (e.g., regulatory capture, what the next congress will do with it, etc). Of course, this is a problem with current economists, not preferring efficiency in general.

    -You don’t seem to have suggested an alternative process. The problem here seems to be that we cannot observe the result of the free market process without getting politics out of the way; the presence of political power makes a market no longer free, so it cannot produce free market results. Since we often can’t compare a free market result to a political result, we have to instead weigh the expected results of political processes vs. market processes. Saying there exist Pareto-efficient deals which a market won’t create seems to be besides the point to me; you need to show that political processes will create these deals. Unfortunately the current crop of economists don’t seem to know very much about political processes. If alternative processes which produce more efficiency than liberty are not found, any arguments of liberty vs. efficiency are moot.

    -I assume you will bring up Futarchy, or other means to hold political actors accountable for the success or failure of their actions. Some actors certainly have the knowledge and incentives to create Pareto-efficient deals which may not arise in traditional markets. I suspect Bryan will take the stance that politics would not create such a system (or if it did, it would be perverted in some way).

  • http://profile.typepad.com/robinhanson Robin Hanson

    Grant, of course economists have errors in estimating costs and benefits of deals, but you set a very low bar for us if the standard is whether we can predict better than the liberty heuristic, which as you admit is very crude.

  • http://williambswift.blogspot.com/ billswift

    Anyone who respects the expertise of economists should read this – http://www.jerrypournelle.com/science/voodoo.html .

  • http://profile.typepad.com/michaeljameswebster michael webster

    Robin, do you have a paper on Sen’s Paretian Paradox? Recall that Sen, and many others after him, produced a simple two agent problem in which although coordination was efficient, it would violate the liberties of each agent to achieve it. I would like to better understand your view in the light of what you might say or have said about the Paretian Paradox.

  • Grant

    Robin, that depends who makes up “we”. If you could perfectly select all the truths out of the current body of economic thought and do away with all the untruths, you would of course have something great. If “we” meant economists selected at random I’d be much more skeptical. If your selection process for good and bad economic knowledge was political in origin, I’d be even more skeptical, though much less so if it was a politically-created prediction market.

    Isn’t it common for scientists to vastly over-estimate the truthfulness of their own thought and discipline? Why should economics be any different? This engineer who’s read a lot of economics sees a few extremely important, simple insights, surrounded by a whole lot of hand waving.

    Though to be fair to the liberty heuristic, the Internet and new financial and legal mechanisms may allow for deal-making on scales previously unprofitable for market actors.

    I look forward to the debate.

  • http://profile.typepad.com/robinhanson Robin Hanson

    Michael, Sen’s example is a familiar example of how liberty need not be efficient.

  • TGGP

    I’m with John Holbo regarding Sen.

  • Phil

    What is the reaction to Nassim Taleb’s take on the dismal science and its practitioners?

  • http://atheorist.livejournal.com Johnicholas

    2. We prefer clear simple math models with explicit assumptions, and prefer particular standard assumptions; these allow models to be more easily compared, and discourage searching for favorable assumptions.

    I think that pursuing simplicity and using standard mathematical models can sometimes INTRODUCE bias.

    For example, suppose a company is choosing between two policies, one which is slightly more ethical than the other. An economist performs a cost-benefit analysis, and comes up with a clear, and high, dollar value associated with the less-ethical policy. The economist has difficulty performing the cost-benefit analysis of the possible goodwill associated with the more-ethical policy, but overall, it looks like the less-ethical policy is on the whole better.

    If there are mistakes in the analysis, the wrong policy might be pursued. It’s possible that the difficulty of the analysis makes the rate of mistakes even higher than the simple heuristic “just be ethical”. However, if the entire discipline is about doing cost-benefit analyses using simple mathematical models, it is unlikely that the economist will suggest NOT doing the analysis, or NOT following the results of the analysis.

    Many people hate economists’ apparent focus on money. I’m familiar with the idea that it’s really about utility, but I still think they have a point, even if they don’t express it well. Presumably the people comprising the corporation are pro-social, cooperative, nice people, with a great deal of interest in future generations. However, something about the structure of decision making leads the aggregate to behave short-sightedly and greedily. Possibly this is partly due to too much attention to simple cost-benefit analyses, which will systematically treat simple-to-model things like money differently than hard-to-model things like goodwill.

  • http://profile.typepad.com/michaeljameswebster michael webster

    Robin, the Sen problem is designed to show that there is no coherent ranking of outcomes, a social choice function, in which both liberty and efficiency are accepted as constraints.

    So, your view is then, efficiency should trump libertarian concerns in the Sen type problems?

    Or is are these class of problems not part of your overall concern?

  • http://silasx.blogspot.com Silas

    @michael webster and Robin Hanson:

    I confess the Sen Liberal Paradox has always confused me. Under Sen’s (rather stringent) definition of what counts as efficient, if something really were efficient, then anyone holding a right that got in the way of the deal would simply — wait for it — waive that right. Supporters of liberty believe you should be allowed to waive rights, so there is no conflict.

    So the Sen Paradox amounts to a criticism of a position no one takes — a libertarianism where rights are replaced with obligations. His definition calls any a trade a “violation of rights” because it infringes on the owner’s property rights in whatever he trades away.

    That said, under the less stringent definition of efficiency that Robin Hanson uses, there is a conflict between liberty and effiency. And one reason to prefer liberty (or at least err heavily in favor of it) is that economists’ partisan-ness will increase in proportion to how often you take their advance. Allow them to dictate policy, and it quickly degrades into “give us stuff”. Note the similarly to the justification Eliezer Yudkowsky gave for deontological ethics: you run on corrupted hardware, so you should rule out the use of certain means, even if they work for the greater good.

  • Unnamed

    Arguing for win-win deals where all parties benefit is one thing, but it’s harder to be “neutral” in the typical economic analysis which argues that some parties’ benefits outweigh other parties’ costs. Your standard for efficiency, which gives equal weight to a $1 gain to any individual, “gives each individual a weight inversely proportional to his or her marginal utility of wealth” (to quote Brad DeLong’s timely post). If wealth has diminishing marginal value, then your calculation of efficiency will be systematically biased in favor of the wealthy and against the poor.

    It might not have come across clearly then, but in the comments to this post of yours I was trying to make a similar argument: the choice of a method to compare value for different individuals is a morally-laden question. Outside of the rare case of Pareto improvements, an economist isn’t as neutral an adviser as you suggest.

  • Douglas Knight

    But in my role as economic adviser

    Advice is Not about Policy. This whole argument makes the big assumption that your advice is actually desired and followed, rather than selectively attended. The appeal of the liberty heuristic to me is the public choice argument that its simplicity makes it easy to monitor. It is also hard to corrupt, as Silas says, although, in practice, economists are never made dictators, so they are never corrupted in a self-serving way, but in a patron-serving way.

  • http://profile.typepad.com/michaeljameswebster michael webster


    Hanson is arguing a side of debate, that is is socially preferable to use efficiency rather than liberty as a measure of policy considerations.

    Sen constructed a little tinker toy example to show that the concept “socially preferable” was incoherent if it had to allow both optimality and respect liberty. It is not simply an example which shows that the valuation of outcomes based on optimality or liberty is different.

    There are a number of responses to Sen’s proof, one being that it has nothing to do with the “real” issue that Hanson is debating. I am simply trying to locate Hanson’s concerns within social choice theory.

  • http://silasx.blogspot.com Silas

    @michael webster: Sen constructed a little tinker toy example to show that the concept “socially preferable” was incoherent if it had to allow both optimality and respect liberty.

    Yes, and I have shown this to be in error because “liberty” is input into the model as “you have this right and you may not waive it”, rather than how normal people would interpret liberty, “you have this right and you may waive it if you wish.”

    It is not simply an example which shows that the valuation of outcomes based on optimality or liberty is different.

    Well good, because it doesn’t even do that, for the reason I gave above.

    I am simply trying to locate Hanson’s concerns within social choice theory.

    And I’m simply trying to point out that the premise of your question, “Sen, … produced a simple two agent problem in which although coordination was efficient, it would violate the liberties of each agent to achieve it.” is at best misleading, because it considers the waiver of a right by its holder to be a violation of liberty, which neither Hanson nor Caplin believe.

  • http://profile.typepad.com/johnleemk johnleemk

    Mine is a simplistic (and thus probably incorrect) understanding, but I would think that the Prisoner’s Dilemma is a classic example of a case where both parties would be better off if they were less free. (Unless, that is, the utility of liberty exceeds that of the gain from moving to a higher payoff outcome.)

  • http://macroethics.blogspot.com nazgulnarsil

    governments can solve real and important coordination problems.

    I don’t understand how the efficiency of a government solution can be measured. prices emerge from competing uses for resources.

  • George Weinberg

    I think Silas is right. I’m looking at the paradox as discussed at the Wikipedia page.

    As written, Bob will not spontaneously go to the chick flick, and Alice will not go unless Bob goes also, but each prefers both going to neither. Alice can guilt-trip Bob into going by saying “I’ll only go if you go also” without applying actual coercion.

    Incidentally, with MYOB preferences this sort of thing seldom if ever happens. Bob doesn’t want to go to the chick flick himself, but he wants Alice to go so badly that he would rather both go than neither. WTF? Alice wants to go, but she wants Bob to go with her so badly she would rather neither go than both go alone. Seriously, WTFF?

  • Grant

    That wikipedia page is on Sen’s paradox is excellent. I’d thought the paradox arose in a world where transactions could not occur (i.e., the opposite of a “Coasian” world), but I suppose a narrower definition of liberty like Silas points out has the same effect in this scenario. Either way it doesn’t seem to be a real critique of the liberty heuristic (any more than a world without transaction costs is a critique of mixed markets).

    nazgul, that was my thought as well. However, this could just mean that liberty is the most efficient policy.

  • http://profile.typepad.com/robinhanson Robin Hanson

    All, Sen’s example as he stated it didn’t make clear why the parties could not make a mutually beneficial deal; so it is only a good example if you assume a barrier to such deals. But I don’t understand the obsession with it here; economists have a long list of standard “market failure” examples where liberty and efficiency conflict.

    Johnicholas, leaving something messy out does increase value estimate errors, but doesn’t obviously allow particular people to identify the sign of the error about them. So they can still expect to benefit from following such advice.

    Unnamed, the point is to approach a Pareto improvement; only dollar weights and transfers can do that.

    Douglas, how do you monitor the arguments for and against liberty as a heuristic? Economists should continue to offer better advice than a simple heuristic can offer, and eventually prove ourselves worthy of being followed.

    nazgulnarsil, I am arguing that expertise that takes decades to acquire has value; you should not be expected to understand exactly how such experts calculate everything.

  • Grant

    I think there are two seperate arguments here: One is which ideas in economics have value? The second is can and will politics use the value of economics to increase welfare?

    Economics influences market processes (perhaps more than it influences political ones?). Trusting a liberty heuristic or trusting some economists doesn’t seem to be mutually exclusive choice to me (especially since economics is a smorgasbord of ideas).

  • http://diogenes42.blogspot.com diogenes

    No other large community of expert advisers has anywhere near the economists’ deserved reputation for consistently suggesting win-win deals.

    Heh — exactly who says this? Maybe I read the wrong sites or interact with the wrong people, but I rarely find anyone outside of econ giving considering economists “highly” (not that the converse is true, just that where is the “data” for this presumed reputation, cause I don’t see it).

    The more that economists agree, and the further from politically charged topics, the more your can trust their advice.

    HEH, to write this sentence in the current financial crisis is very amusing. Perhaps you ought to ask your colleagues on the religion of deregulation and what happened with that. It didn’t seem that there was much disagreement on this topic until things blew up.

    As Phil mentioned — Taleb and many others seems to think very lowly of economics in general — so I find it puzzling you think economists have a good rap — cause I’ve never see anything to suggest that.

  • http://silasx.blogspot.com Silas

    @Robin Hanson: All, Sen’s example as he stated it didn’t make clear why the parties could not make a mutually beneficial deal; so it is only a good example if you assume a barrier to such deals. But I don’t understand the obsession with it here;

    The obsession isn’t with Sen’s example; the obsession is with why people give it any attention in the first place, since it’s so useless and so misleading as to be wrong. It amounts to saying:

    1) Two arbitrary rankings can be different.
    2) If you call one of them “liberty”, despite this being the opposite of what everyone means by the term, then liberty is bad.

    And Sen isn’t saying that there’s a barrier to these potential deals; he’s saying the deals violate liberty!

    Hint to Sen: If your position requires that waiving rights is the same as violating those rights, you made a mistake somewhere.

  • Douglas Knight

    how do you monitor the arguments for and against liberty as a heuristic?

    I meant that it is easier monitor whether whether a simple heuristic is implemented than to monitor whether complicated arguments are implemented. Also, assessing the arguments for a heuristic is a one-time cost, while checking that new arguments are not trojan horses is an on-going cost.

    Economists should continue to offer better advice than a simple heuristic can offer, and eventually prove ourselves worthy of being followed.

    That completely ignores the public choice issues: better for whom?

  • http://timtyler.org/ Tim Tyler

    Re: better for whom? – Robin’s definition of “efficiency” answers that with: “everyone, equally weighted”.

  • http://macroethics.blogspot.com nazgulnarsil

    what grounds are there for weighing everyone equally? People don’t like to compromise with groups who are too dissimilar from themselves. Combined with robin’s comment that people shouldnt be expected to understand the decisions of the experts, this would seem to create a situation where people are being told to do something they don’t want to do and not being given an intelligible explanation. The resulting friction should figure into calculation no?

  • http://arare-litus.blogspot.com Arare Litus

    Does efficiency measures not assume that valuations are meaningfully comparible? Is this true for any “interesting” issue? Also, couldn’t one say that if you really (and strongly) believed this (efficiency over liberty) Robin, that you would not enter into a debate format, but mearly take a poll at the begining, sum the results, and say “that is the winning idea here?”, “I win” or (more paradoxically) “I lose” (repeat in case of tie until summed indifference is broken?). Doesn’t debate assume a liberty perspective?


  • http://timtyler.org/ Tim Tyler

    Re: what grounds are there for weighing everyone equally?

    If you are dealing with a democratic organisation, or one where the members like the American Declaration of Independence, or go in for political correctness, that assumption might be useful.

    It seems to me that there are other cases where it is not such a useful assumption. Ignoring the views of the old is common: they will be dead soon anyway – so their opinions don’t matter. Ignoring the views of the young is common: they are too inexperienced to know better – and so on.

    In some circles, idea that “everyone is equal” is considered to be politically-correct nonsense – and if you are offering economic advice to groups of such folks, “efficiency” seems unlikely to be a good selling point.

  • Pingback: Overcoming Bias : Minimal Morals

  • Pingback: Overcoming Bias : Reply to Caplan

  • Pingback: TheMoneyIllusion » Does Bryan Caplan believe in free trade?

  • Pingback: Dealism | Bayesian Investor Blog