Steve Reads

There should be a category on bookstore shelves for “little, dense, incisive, ingenious books.” [book: Exit, Voice, And Loyalty] would be one; Herbert Simon’s [book: The Sciences of the Artificial] would be right next to it.

Hirschman here examines three methods for addressing an organization’s deterioration: “exit,” the option of leaving the organization, not buying its products anymore, etc.; “voice,” the option of sticking with the organization and protesting in the hopes of improving it; and “loyalty,” which is really more a property that encourages you to stay with the organization longer (thereby probably delaying exit and giving voice more of a chance). Normally we think of exit and voice as separate powers: exit is what you use in a market economy, while voice is what you use when dealing with a government (which is typically hard or even impossible to exit). Hirschman asks a simple question: how do these powers interact?

The interaction turns out to be fascinating. Consider parents dissatisfied with the performance of their children’s public school. The parents who are most focused on school quality are likely to leave first and put their kids in private schools, leaving behind only the less quality-focused parents who are less likely to speak up. So the exit option (parents abandoning the school) diminishes the use of the voice option (speaking up). This is likely to accelerate the school’s decline.

Or consider a product (maybe a brand of automobile is a canonical example here) whose quality has diminished over time. Again, exit is likely to be used before voice: the customers most concerned about quality will bail first. If too many customers bail too fast, the company won’t have enough time to fix the product before it’s gone out of business. Whereas if too many customers don’t bail out quickly enough, the company will never learn that its products need to improve. So one can postulate a certain optimal level of quality-sensitivity in customers: not too high to kill the company before it fixes things, but not so low that the company stagnates. Modeling this formally would naturally use an analogue of price elasticity; whereas price elasticity measures the percentage decrease in sales that results from small percentage increase in price, quality elasticity measures the percentage change in sales for a small percentage change in quality.

Indeed, one of the neat little accomplishments in this neat little book is that it uses formal economic models to study the interaction of exit, voice, and loyalty even when the context is, say, the decline of a government rather than that of a company. The indifference curves are mostly confined to appendices, and Hirschman’s writing is clear enough that normally you can get the gist just as easily — if less rigorously — from his words as from his charts.

A government would be a classic case where exit is basically unused: yes, you could leave the United States in protest — and some large fraction of our countrymen promise to move to Canada at every presidential election — but would you? So the main option available to you in a democracy is voice. Similarly, you’re likely to raise your voice in a political party before you’d exit it, particularly if the alternative party is far away from you ideologically.

Here Hirschman consults a famous model by Harold Hotelling, metaphorically applied to ice-cream stands lined up along a beach. Imagine that beachgoers are distributed uniformly along the beach, and two ice-cream stands are trying to decide where to place themselves to capture the most customers. Imagine that the ice-cream stands initially start on opposite ends of the beach, but are free to relocate. The leftmost ice-cream stand realizes that if it moves a little bit toward the center, it can continue to pick up customers to its left (because those customers are still closer to the leftmost ice-cream stand than to the rightmost) while picking up those customers that lie less than halfway between the leftmost and rightmost stands. The rightmost stand realizes, similarly, that it can move to the center and continue picking up those customers to its right while picking up a few more in the middle. This process continues until both ice-cream stands are almost exactly at the middle, separated by a tiny sliver of distance.

Replacing the beach with some measure of political ideology, the conclusion is that it’s always in political parties’ interests to move to the center. There are initial obvious reasons why the analogy isn’t quite right. It’s not obvious that voters are uniformly distributed across the ideological spectrum, for one. For another, the Hotelling model assumes that customers don’t care about the cost of walking down the beach: they’ll pick the nearest stand, regardless of how far they have to walk to get there. In a political context, this would imply that voters don’t especially care which beliefs their parties hold; they just want the party to be “closer to me than the other party is.” Which isn’t obvious: perhaps people on the right or left ends of the political spectrum really want parties on the extremes, and will opt out of party politics altogether if they’re forced to pick among centrist alternatives.

Hirschman ends up rejecting the Hotelling model as it applies to political parties, for the reasons laid out above (more to the point: because it fails to work empirically). Here he’s rejecting what is apparently known as Hotelling’s Law, not to mention the median-voter theorem.

For such a small book and so few atoms (three, in fact: exit, voice, and loyalty), Hirschman’s is remarkably dense with interesting ideas. Highly recommended.