Friday, September 21, 2012

Regret Aversion by Proxy in a Coercive Environment

If you'll grant me the assumption that some of the new paternalism aims to relieve people of ex post regret, I'd like to reconsider the cell analysis I did a while back to see what happens when we add either private or state coercion. Regular readers will recall that I consider the two key considerations to regret over exchange to be expectations and search costs. A systematic bias in either one justifies concern. A systematic bias in both may justify legal intervention.

This post is quite long, so let's put a fold right about here:

In my earlier post on regret, I phrased things a little differently, looking more at the interaction between buyer and seller, so the axis that I called "information disadvantage", which was meant to get at asymmetries can be recast in a single-agent context as biased expectations, the source of which is perhaps less relevant for this difference analysis. With that in mind, the enhanced matrix is now:

                         Low         |        High
                   Information Cost  |  Information Cost  
                   |  Low  | High |  |  |  Low  | High |
Bias in     |  Low |   1   |   2  |  |  |   5   |   6  |
Expectations| High |   3   |   4  |  |  |   7   |   8  |

Again, I understand that it is possible to create tables using HTML, and there are probably handy-dandy web-based user interfaces that will write the code for me, but I judge the search costs to outweigh the probability-adjusted benefits of a marginally prettier table.

First, let's quickly review the low-coercion regret scenarios:

1) Unbiased expectations, cheap information. If things go wrong, the random bad luck is tempered by the knowledge that the person went in with open eyes.
2) Unbiased expectations, expensive information. Nothing to see here for behavioralists, but this could be a case of structural ignorance. Fixing this is part of the mission statement of many regulatory agencies (and the purpose of orgs like Underwriters Laboratories)
3) Biased expectations, cheap information. Intellectual lazyness if you asked an elitist. Rational irrationality if you ask a more tactful person. There is legitimate deontological disagreement over whether or not this cell merits paternalistic intervention.
4) Biased expectations, expensive information. If the information problems include behaviorial inconsistencies like (unwarranted) hyperbolic discounting or money-pump intransitivities and the law provides no relief from such activities, we may meet the necessary conditions for violating a presumption of liberty.

Let's for the moment assume that a concerned third party hasn't heard of my attempts to parse the determinants of regret and cares only about ex post expressions of rueful dolor. Smith smacks himself in the forehead after buying a bad taco from Jorge and Jones immediately cries, "something must be done!" If Jones is a fan of Thaler and Sunstein, he'll lean towards coercion by cognitive shortcut (at first, anyway. Evidence of the waning power of default settings for a great many things is pretty strong). If not, he will rely on coercion by edict, of which there are two varieties: private sanction or legislative dictum. The paternalized party must then recalculate decisions in the new environment.

Here are the stylized outcomes we can expect from this:

A) No change in behavior--both buyers and sellers remain in the market; third parties extract/dissipate rents.
B) Buyers exit the market, sellers remain. Something has to adjust here to move back in the direction of equilibrium.
C) Buyers remain in the market, sellers exit. Black markets emerge to fill customers' orders or low-cost producers obtain de facto monopoly power.
D) Both buyers and sellers exit the market.

I think the error in assumption during the prohibition era was that ex ante, temperance activists figured a Constitutional amendment would generate result (D), when what they got was mostly (C). I might make a similar argument for the poorly-named "wars" on drugs, poverty or terrorism.

If you, like me, are a fan of Douglass North, Barry Weingast and John Wallis (or heck, just about anyone who went through Wash U under North for that matter), you'll appreciate the claim that one of the defining characteristics of the state is its legitimized monopoly claim over the use of force. Outside of the scope of the state, black market patrons can claim no last-resort protection against fraudulent transactions. Cell (4) regret has no relief in the presence of banned markets.

Perhaps I'm getting ahead of myself here. We've actually got 16 cases to look at: the cross product of cells 5-8 and outcomes A-D. For the sake of brevity, and because the comparative statics are more than I really want to think about for a mere blog post, I'll just take a look at a few and urge you to consider some of the relevant tradeoffs and elasticities.

5D) Here, we've got things going bad once in a while by random bad luck.Some of the crazy scare stories you hear on Lenore Skenazy's blog from time to time will give you a decent grab-bag of examples of this (Free Range Kids is firmly in my top 2 of all blogs that contain the word "Kids"). Here, I find it hard to justify banning recess in response to some kid getting knocked over by a jelly ball and ending up in the ER with a broken arm, and I think the euvoluntaryist might offer some insights about the opportunity cost of protecting people from the fickle winds of fate.

7C) Banned activity where participants have biased beliefs and low search costs. It doesn't take a lot of rigorous research to discover that unprotected sex can lead to babies. In many places in the US, abstinence-only sex-education curricula actively raise search costs, pushing students from 7 into 8 (as much as is possible, anyway). Here again, I think concerned voters (and it sort of baffles me why teachers' lesson plans are subject to democratic wrangling, but that's a subject for another day) presume that a ban on condom education will mean that kids won't have sex. Perhaps that's right on some margins, but the empirical evidence doesn't seem to support this conclusion. Of course, if I were being Hansonian, I'd point out that sex ed bans aren't about preventing teen sex. Rather, they're a way to preserve farmer norms against forager sexual license, but I will direct interested readers to Robin's blog, Overcoming Bias for more thoughts on this specific topic.

So what's the upshot here? What's the point of writing all this? Well, when I did the old post on regret, the idea is to ask whether or not a proposed paternalistic policy might be morally justified ex ante on the basis of information asymmetries. I now add to that a plea for consideration of likely consequences. Even if the paternalist thinks she's got a better lock on averting regret on the behalf of others in her society, we'd be better off predicting what sort of outcomes we might expect. As much as I think the first part is overlooked, it's the second part that really wreaks most of the havoc.

No comments:

Post a Comment

Do you have suggestions on where we could find more examples of this phenomenon?