Neutral Institutional Monism strikes again

Will Wilkinson continues to be on fire with regard to all things Happiness Policy related. In this specific instance, in response to a suggestion that since individuals may be systematically bad at estimating their own happiness paternalism is justified, he illustrates once more the point of neutral institutional monism,[1] which in plainer terms I'd summarize as "sauce for the goose is sauce for the gander," and that institutions are made up of the same systematically mistaken people that are to be paternalized:[2]

Even if some experts know better than we do what will make us happy, it is very unlikely that those experts will be the ones determining paternalistic policy. Consider that the very same people who make systematic mistakes about their future feelings are the voters in democratic elections. And, as Bryan Caplan’s work shows [.doc], voters are likely to be even more profoundly mistaken about politics than about their own affairs.

Successful politicians are likely to reflect the biases of the voters. Take the War on Terror, for example. Gilbert’s work implies that whatever the harm of another terrorist attack may be, it probably will not be as bad as we imagine, and we would get over the wound rather more quickly than we think (indeed, more quickly than it may be comfortable to acknowledge). But that doesn’t keep voters or politicians (who share the same psychology, after all) from being extremely anxious about another terrorist attack.

There’s a good chance that lots of people mispredict how bad things would be if drugs were legalized, or if same-sex marriages were legally recognized. The consequence is that we get politicians who appeal to us because they make the same errors—or even because they convince us that things will be even worse than we thought (which was already way worse than it would really be) if they aren’t elected. And these are the people who determine paternalistic policies, not experts like Daniel Gilbert.

Even if you ask politicians to appoint experts, they will not consult experts on expertise to determine who the real experts are. Their beliefs about who is and isn’t an expert, like their beliefs about anything else, will reflect their biases. If I were President, Leon Kass might be the last person I would think of to head my Council on Bioethics. But if you look at his bio, you can certainly see why he looks like an expert to some people. The upshot is that happiness-based paternalistic policy may be more likely to be based on the work of Dr. Rick Warren than on the work of Dr. Daniel Gilbert.

So, a realistic account of human psychology shows that we can be pretty bad at predicting what is really going to make us happy. But a similarly realistic account of actually existing political institutions shows that they are likely to be even worse than individual decision-makers. If we make systematic errors, then democracy will simply aggregate our errors. And politicians, who make the same errors we do, will reflect our errors, and will often have an incentive to reinforce them to their political benefit. Even if expert knowledge exists—even if Daniel Gilbert knows better than you do about what will and won’t make you happy—democratic institutions will not be reliable at identifying it or applying it.


fn1. Defined by Will thusly:

According to neutral institutional monism (NIM), there are stable structures of interdependent interaction, that is, institutions. Different institutions have different properties. It is crucial to recognize that these properties are not determined some kind of metric of resemblance to an ideal type, but by messy details about the real rules of the game, which are generally a melange of assurances and threats (some paradigmaticlly coercive, some reputational, some internally generated by conscience) impossible to locate in a simplistic taxonomy.

Let’s see how quickly, and how far, we can travel from our original topic.

Economics textbooks provide a theory of ideal types. MARKETS are filled with strange omniscient creatures who have preferences over every conceivable combination of goods, who can trade instantaneously with zero transactions cost, etc. STATES are, more or less, that which can do anything MARKETS can’t—mainly, fix “suboptimal” patterns of interaction by changing relative prices by taxing, subsidising, regulating, credibly committing to punish defectors from contracts, and otherwise wielding coercive power with amazing laser precision.

Additionally, ADAMANTIUM is the hardest material in the world, which raises the profound question: Can Captain America’s shield shatter Wolverine’s bones?

However, once we step outside the Marvel Universe, the question is ill-formed, for ADAMANTIUM is, in fact, an empty category. Nothing fits the description. Likewise, once we step out of the enchanting Neo-classical Universe, it doesn’t even make sense to posit the STATE as the answer to MARKET failure, for STATE and MARKET, as laid out in the model, are empty categories.

NIM is interested in this market and this state. Don’t worry. We can draw lots of generalizations about institutions. For example! Markets need well-defined property rights, reliable mechanisms of enforcement, enough social trust, etc. In order for states to work, the state agents need a structure of incentives compatible with the goals of the paper, de jure, institution. Otherwise the effective, de facto, institutions will simply be something else. (”But it’s called the Ministry for Health and Welfare!”) There’s lots more we can say in general. But NIM demands that we resist thinking we’ve illuminated much of anything by tagging something a “state” or a “market.”

fn2. Paternalized? Sorry if neologism.

Share this

I'm glad of that second

I'm glad of that second footnote. On first reading I mistook paternalized for penalized, which made just enough sense that I didn't question it.