Wednesday 21 October 2009

Illiberal Anarchy

Brad Taylor and I (mostly Brad) have been working on a paper arguing that a market-based anarchy could well be rather illiberal. Assume that we're in Caplan's "sweet spot" for a feasible and desirable system of market chosen law: the network coordinating the protection agencies and allowing for dispute resolution amongst them is strong enough to keep out rogue agents but not strong enough to form a cartel. Further, specify that most people have weakly liberal preferences: they're happy for folks to do their own thing so long as there's no cost imposed on others, but it's only a weak preference. Next, specify that there exists a group with strong meddlesome preferences with a real willingness to pay to have those preferences imposed on others. Finally, the group with preferences deplored by the meddlesome minority have a smaller aggregate effective willingness to pay to enjoy their preferred lifestyle than the meddlesome group has to stop them (though their willingness to be paid to stop exceeds the meddlesome group's willingness to pay).

What happens? The meddlesome folks offer to subsidize the purchase of protection agreements that include their preferred meddlesome clauses banning the sanctioned behaviour and, if the meddlesome willingness to pay is high enough and the sanctioned group small enough, banning those protected by non-compliant agencies from entering their properties. The weakly liberal majority take the contracts if the discount is sufficiently large and if the loss in utility from not being able to transact with the sanctioned group is sufficiently low. Nothing in the process violates rights, but the outcome is hardly libertarian.

In states of the world where most folks have weakly liberal preferences but a few folks have strongly meddlesome preferences, majoritarian democracy produces more liberal policies than does market-based anarchy. Markets are better at satisfying dollar-weighted preferences than is politics. Where most folks have weakly meddlesome preferences but a few have strongly libertine preferences, market-based anarchy produces more liberal outcomes than democracy. We then worry that democracy has a tendency to produce broad-based weakly meddlesome preferences (which I think typically runs through fiscal externalities) while a market-based anarchy would promote the creation of public or club groups through churches or sects able equally well to coordinate for the production of public bads like meddlesome activity.

John Humphries Humphreys (sorry!!) has had a go at our paper - excellent!

His first critique: we're using a teleological rather than a deontological norm for judging how free a society is. Guilty! But imagine, as a libertarian with some libertine preferences, weighing up where to move. If you move to Country A, all the laws are perfectly libertarian, but if you engage in some activity you like but they don't, you lose all opportunities for transacting with others. In Country B, there are a lot of stupid laws but there's still more space for you to live your life as you like. Country A is the libertarian monastery where you can do what you want, but if you're not up for vespers at 5 you're shunned; B is the world we're in.

Note that we didn't say that the meddlesome folks would pay the drug users to stop using drugs -- that would be the trivial solution and, in that case, he'd certainly be right: utility rules. Rather, if you're willing to pay $10,000 to defend your right to use drugs but if you'd only be willing to accept $100,000 in exchange for giving up your right to use drugs (ie income effects can matter), then somebody willing to spend $50,000 to make sure your neighbours will have nothing to do with you if you do use drugs makes you worse off and your neighbours better off. Yes, we're invoking a somewhat thicker description of liberty than a pure deontological standard. But would you really move to the libertarian monastery?

Brad notes as well that we need not invoke thicker concepts of liberty in cases where the network's members have sufficient consumer-driven willingness to pay for meddling, in which case libertine agencies simply are declared rogue and their members are treated as criminals. This is a case that Cowen worries about in his initial article, and it is not a case of the network simply becoming the state, though this latter argument may be somewhat semantic. In the Caplan-Cowen debate framework, the network becomes the state when it is strong enough to become a cartel: strong enough to declare any protection agency rogue and strong enough to punish any member that transacts with the rogue agent. In this case, it's still a coordination equilibrium as no member agent wants to deal with the rogue agency because of side payments from the meddlesome group.

Humphries Humphreys is right that folks can conjure up all kinds of scare stories, including the rich jerk buying up all the land around your house and forbidding your exit. That seems a pretty implausible fear. But in the real world, there are lots of folks who seem perfectly willing to expend real resources to make sure that you don't do things in your own house that have no effect on them. I don't think it's crazy to worry that these folks might get more influence under market chosen law. We're not conjuring up completely imaginary boogey-men here. Just hit the "paternalism" keyword on the right hand side of the blog...

Second, Humphries Humphreys worries about realism and how likely this 'worst case' might be. He's certainly right that there's no cause for concern if the minority affected is relatively large. The costs of losing transactions opportunities with a large group are very large indeed. But not so for a relatively small group. In terms of our Figure 3, Humphries Humphreys would be arguing that the region in which democracy dominates anarchy must be smaller. But the logic of the argument is such that there must be a region where democracy dominates. In the part of world-space where we currently live, with lots of folks having low intensity meddlesome preferences, anarchy produces more liberal outcomes. But there are other parts of the space.

Finally, Humphries Humphreys notes Taylor's argument that voluntarism encourages tolerance. Of course, this is true among the majority who currently have weakly meddlesome preferences. In the current paper, we're worried about the high preference intensity illiberal folks and the disproportionate influence they can have under anarchy.

Now, would all this have me refrain from pushing the button that would cause government to disappear after a 5-year delay? Well, I was only about 45% likely to push the button to start with; the worries here push that down to about 40%. [All stated probabilities are purely notional as no such button exists and, if one did, I surely would not be allowed access to it.] My biggest worry is that Caplan's sweet spot seems a pretty narrow space and the historical record isn't exactly replete with stable desirable anarchies. Heck, even outfits that ought to have been able to protect themselves outside the state fared rather poorly absent state protection.


  1. Excellent indeed. But you spelt my name wrong.

    I agree with your point that an anarchist society might be no fun. But I think it makes sense to use different words for different concepts and I'm precious about trying to defend the deontological meaning of "freedom".

    And I agree that paying people not to associate with drug dealers is different from paying you not to use drugs... though ultimately they are both acts of persuasion and not coercion.

    If you have an agency that has a geographical monopoly on coercion (ie the prudes are making and enforcing rules on people in an area who didn't agree) then I think that should be called a government. Though perhaps I'm not clear on the scenario you're hypothesising. Are you saying that the prudes will hire guns to invade the libertine community and enforce a no-drug rule? Or are you saying that the prudes will hire guns to protect their own land (and the land of their paid lackies) and enforce a no-drug rule?

    If nobody is willing to defend the libertines, then the hired guns become their government. If somebody is willing to defend the libertines, then you have the makings of a war. The outcome of the conflict will either be the success of the prudes (and the imposition of their government) or a successful defence (in which case the anarchist system sustains).

    I agree people will spend a lot of money trying to influence our behaviour. My boss spends money influencing me to show up. Busy-bodies spend a lot of money telling me to drink less. The KKK spends money promoting their view of a white homeland. I'm spending resources right now trying to convince your readers that anarchy is good. :)

    And I agree that these groups will often succeed in introducing rules in areas that are meddlesome. For example, many of my friends have a "no smoking in the house" rule, and gold clubs have dress codes. Bastards.

    But I find it highly implausible that all communities would sign up to the exact same set of rules. Even Muslim Spain managed to get by with three different legal systems... I'm sure anarchy would end up with more. To impose one set of rules on all communities would either involve a war, or an impossibly large amount of money. If we assume that one group has control of 99% of the resources, then the scare becomes real... but in that scenario I think drugs are the last of our worries.

    I also don't think you showed that anarchy would lead to a higher preference to impose prudism (TM) on the unwilling. You showed that there would be increased membership of controlling groups (I agree) but I don't think it logically follows that they will expend more resources trying to convert the world. If anything, their need to convert decreases once they are allowed to set up their own galt's gulch... not to mention the liberty-tolerance link explained by Brad in his Independent Institute paper. The relevance of this paper is *not* regarding the "strong preference -> bad anarchy" link, but the "anarchy -> stronger preferences" link.

    As for the "sweet spot" argument... I agree this is a worry. There are two responses. First, no system is perfectly stable, so all we need is reasonable stability. Second, I suggest that anarchist stability increases as we get richer, because the net benefits from violence decrease.

    I'm writing a paper on this now. The reason is that wealth leads to:

    1) higher opportunity cost of war (peaceful trading with rich people is good for you).

    2) higher value of life, meaning a higher perceived cost of war.

    3) lower benefits from war as it's harder to steal human capital (the main driver of rich society) rather than physical capital & land.

    I think anarchy is fairly unstable in poor countries (though possible for a while: medieval iceland) but is becoming more possible every year, insh'allah.

  2. Apologies on spelling. Yikes!

    I'm also generally predisposed to the deontological meaning of freedom. But if a deontologically free anarchy could wind up having less room in which to exercise freedom, wouldn't that be a worry?

    Staying in Caplan's "sweet spot", no agency has a geographical monopoly on coersion. Multiple agencies would provide protection services and would adjudicate disputes amongst themselves through one or several dispute resolution networks.

    In the more likely case, the prudes subsidise protection services for folks willing to exclude libertines from their properties. This works where strong preference libertines are relatively small in number (even some libertines wind up accepting the subsidy; other libertines are isolated). It isn't just keeping libertines from doing drugs on other folks' property; it's excluding folks who fail to sign on with companies preventing drug use even on the libertines' property. What does that mean? If a libertine steps off his property, no other protection agency will fairly adjudicate any dispute in which the libertine is involved and they will not recognize the libertine's protection agency. And the libertines' agencies are not strong enough to enforce the libertines' rights. Yes, deontological liberty fine (libertines agree to be predated on by stepping off their property), but not much fun.

    In the less likely / more expensive case, the agencies go after the libertine agencies as though they were aggressive rogue agencies (Mafia, Inc.) They have to have the ability to do this, otherwise Mafia, Inc can operate. Again, this doesn't make them the government: folks can still contract with whichever agency they like, but within bounds. Semantics whether you want to call this government or not.

    I also agree that folks would not all sign onto the same sets of rules. All we require is agreement on a narrow set of prohibitions, not a full set of rules. If there's strong meddlesome preferences by some against behaviour X, and most folks don't much care about X, then X gets prohibited with side payments coming from the high demanders to get the weak preference folks on side. Massive heterogeneity in all other rule structures, possibly. Except about X.

    On preferences, the argument there was more that the prudes' would be getting together to solve other collective action problems already and that their coordination costs for subsidising prudism more generally would then be much lower than currently.

  3. Yes. If deontological freedom led to shitty outcomes, I wouldn't like it. My preference for liberty isn't absolute. I prefer voluntary action to involuntary action... but I don't think liberty is the only moral "good" so if the outcome was too crap I'd sacrifice some freedom. As it happens, I think the outcomes in anarchy would be better. I'm a utilitarian anarchist in the David Friedman mould.

    But still, so many discussions are de-railed at the beginning because of people using different language... so I think it important to avoid confusion.

    I accept your example, and that is what I understood your point to be. As I said in my post, an uber-rich and sadistic "anarchist bully" could make life pretty crap, leaving a libertine community impoverished and isolated. Though I think it likely many would still join... especially us hippies. :)

    I think you underestimate how expensive that would be, how likely it is to fail... and more importantly you don't consider the dynamic of social evolution. People who spend absurd amounts of money trying to do something that won't work will tend to lose in the grand game of life. Consider Gary Becker's work on the economic costs of illogical bigotry. Even if we start the thought experiment with the prudes having 99% of resources (in which cases, drugs would be the last of our problems), it would not take long for the economic power to shift towards less bigoted people.

    Your thought experiment might work with examples of actions that are hated by most people, such as bestiality. But even then, I highly doubt you'll find a collective of prudes who want to spend sufficient money creating a world-wide anti-bestiality bribery campaign. There is little evidence of that sort of behaviour in the current system, and less incentive for it under anarchy. (And even if they did, the bestialitist would probably find a way around the law.)

    In the "war" case, I disagree on two levels. First, I don't think it would be more expensive. If you had a big enough force, an "invasion" of the libertine territory would probably be the cheapest and easiest way to prevent drugs (though it would still fail, as the current war on drugs has failed). And second, if an external power is setting the rules for acceptable behaviour in a fixed area, and then enforcing those rules (even if they out-source the security)... I think that's pretty much the definition of a government. The prudes may still live in anarchy, and they may only enforce a "friendly" minarchy over the libertines, but it is a government.

    The liklihood of this goes back to the old debate about the stability of anarchy.

    On "increasing prudism", I agree prudes (and other groups with diverse and sometimes strange preferences) would be more able to coordinate under anarchy. That's one of the prime virtues, leading to *lower* need to be evangelical. The costs of coordination may have decreased among the prudes, but the transaction costs of dealing with a range of communities would have increased (appropriate theme given our recent economics nobel laureates). Further, there is the "Taylor-tolerance" point, where social evolution would chose both people and groups that are relatively more tolerant of outsiders.

    It is this point -- the dynamics of social evolution under anarchy -- that I think is the major missing piece of your draft paper. You have shown there would be more "sects", but I think you missed the point that these groups would likely be more benign.

  4. I think it takes a lot more than an uber-rich and sadistic "anarchist bully" -- it takes very weak preferences among everyone else as well. If there are proportionately few libertines, such that each person loses very little in terms of expected trading possibilities, then it doesn't take a lot of subsidy to convince everyone else to accept the slightly cheaper defense provision contract that excludes libertines.

    I totally get the Becker point, but the Becker point as well is that if consumers have a taste for discrimination, and are willing to pay for it, then markets provide discriminatory outcomes. In general, such outcomes are much more cheaply achieved through governments (See Roback's fantastic work on Southern Streetcars, for example), but that's what we expect when most folks have weakly meddlesome preferences; when most folks have weakly libertarian preferences, democracy potentially produces decent results.

    I don't think we need actions that are hated by most people. In those cases, democracy also provides bad outcomes (median voter). And, it doesn't take some big world wide campaign to get folks on-side: it just takes that they get a $5 discount on their protection services if they accept the rider on their contract. Yeah, costs of that add up across a lot of people, but think about how much gets donated to various churches or causes already.

    We're down to definitional questions on whether the post "war" case counts as a state. We could call it a government that the libertines are prevented from choosing defense contracts that allow for X, but does that mean it's also government if misanthropes are prevented from choosing defense contracts that allow them to hit other folks? You'll rightly reply that one involves the use of force against others who haven't agreed to it, but where do we draw the line between that kind of external effect and external effects around "moral corruption" or other such psychic harms? My moral intuitions are as yours: we shouldn't be able to harm folks who haven't agreed to be harmed. But specifying what gets to count as harm gets tricky, and it's not something that economists have ever adequately addressed (as far as I've seen). Yes, we have Buchanan Stubblebine on externalities limiting things to Pareto-relevant ones, but that doesn't quite get us there.

    We could well have many benign sects, but you only need a couple of well-resourced ones to make the difference. Imagine the Scientologists adding a rider to most folks' protection services contracts that gives folks $10 off their bill in exchange for excluding some notable anti-Scientologist advocates. If most folks have only weak preferences for free speech and don't really know any anti-Scientologists (or Scientologists), then most folks take the deal.

    You're right that the paper does need more work though. Sabbatical soon!

    When next you're in NZ, we'll have beer and argue this stuff.

  5. Rather, if you're willing to pay $10,000 to defend your right to use drugs but if you'd only be willing to accept $100,000 in exchange for giving up your right to use drugs (ie income effects can matter)

    Isn't this an implausibly large income effect, though? If someone makes $30K/year, and would pay $10K/yr to defend their right to use drugs, I find it rather hard to believe that it would take $100K/yr to make them give up their right to use drugs.

    Also, this seems to be assuming a unified global rule set, and ignoring geographic effects. The more of the world consists of shitty states and libertarian monasteries, the more you increase the market demand for Super Sin City. With geographic diversity and freedom of association (ie, the world of seasteading), we get a beautiful feedback system: the less well the rest of the world caters to some set of locally-satisfiable preferences, the stronger the pressure to found a community for those preferences.

  6. Patri: The income effect could be smaller than that. You would expect pretty large income effects on some things, though. If I'm gay and only make a little over $10k per year, I can easily imagine the difference being much larger.

    Geographic difference decreases the scope for bigots to have their way, but there's still some scope. Seasteading would improve the situation further by increasing the costs of inter-enclave monitoring and enforcement.