Inescapable Economics

Just one quick point apropos of the last to posts on this site, one by me and one by Jim Manzi.

Suppose you started from the following goals:

- As a matter of fairness and equality of opportunity, I want to subsidize college (and graduate school) tuition for those who otherwise could not afford such an education.

- I want to minimize the economic and political distortions associated with my subsidy, and to minimize as well any costs to local and individual experimentation and choice.

Which would you prefer: providing grants to poor students to attend the university of their choice, or establishing a public university with free tuition?

I would assume that most people aiming to achieve the two goals articulated above would opt for the first option. Giving money to students who otherwise couldn’t afford tuition directly achieves goal #1. And it appears to achieve goal #2 because it doesn’t particularly interfere with individual consumer decisions.

But, per the argument of my post, this conclusion has embedded economic assumptions. Specifically, it assumed elastic supply of college spots. If, in an extreme example, there were a fixed number of spots for college, then the entire subsidy to poor students would be captured by incumbent institutions in the form of price increases. This, in turn, would make college unaffordable for a larger set of students, who would require a subsidy, and so on, rising tuition driving ever-higher levels of subsidy.

That doesn’t sound like it meets goal #2 at all. Whereas, in the same circumstances – completely inelastic supply – establishing a new, free tuition public university would have fewer negative collateral effects, either establishing a market where one did not previously exist (perhaps because no private institution could profitably serve this student population) or driving lower costs across the system. Poor students would have more choice, and the existing student population would not suffer.

On the other hand, if supply is completely elastic, the opposite would be true. A subsidy that went directly to students would increase supply without substantial negative externalities, and would preserve choice, while the creation of a public university with free tuition would drive marginal private universities out of business, resulting in reduced consumer choice. This drop in supply would, in turn, drive demand for more slots at the free university, which in turn would drive more private institutions out of business, resulting in ever-greater domination by the single, public provider.

So what should a public policymaker do?

One answer is to say: the public policymaker should not try to subsidize education for indigent students. It is impossible to know with sufficient certainty whether policy A (grants to poor students) or policy B (free tuition at a public university) would be optimal, and either one could have negative side effects that outweigh any benefit. Therefore, we should be epistemically humble, and abandon our goal of equal educational opportunity.

Abandoning goal #1, though, may not be political feasible – or we may simply consider abandoning it to be immoral. In either case, we cannot avoid trying to understand the possible implications of our policy choices, and evaluating them in light of the effects we observe.

My point is: we can’t tell from a description of the policy whether a policy will be “epistemically humble” or not. The policy that looks like the less-distorting one may only look less-distorting because you don’t investigate the economics of the situation. Once investigated, a credible argument may emerge that this policy is, in fact, the more distorting of the two available options. You cannot pick the more “humble” option without knowing the economics. It may appear that you can, but that’s an illusion. And if you’re convinced that the economics are also an illusion, then you simply cannot pick.

Eliminating economic arguments from the conversation because they are insufficiently scientific doesn’t make economics go away. Policies with still have economic effects, effects that may be very different from what folk wisdom would assume. We just won’t have any idea what they will be. That being the case, the epistemically humble thing to do is, in most cases, not to have a policy – that is to say, not to make any decisions that implicate large numbers of people in any way.

But if you have a scope to make policy, then everything – even the policy of doing nothing – is a policy. A policy of issuing currency backed by gold is a monetary policy, one that outsources questions of money supply to the gold mining industry. If you disclaim value to economic argument, you have no way of knowing whether it’s a good policy or a bad policy, but it’s still a policy, as is issuing a currency backed by nothing but a promise to manage the supply of currency “well” without any clear definition of what is meant by “well” (which is the policy of every industrialized country today). The only way not to make policy is to disclaim authority. If you don’t issue currency at all, then you have no monetary policy.

So the only way to avoid making decisions that implicate large numbers of people is not to have authority over large numbers of people – that is to say, not to have large states with the power to tax, spend, issue currency, etc.

Once you accept – or endorse – the existence of large states with significant involvement in the economy, you can’t avoid making policy with economic implications, and therefore you can’t avoid resort to economics as an input to decisionmaking. Epistemic humility, in such circumstances, has to involve some way of evaluating arguments within economics to assess relative levels of certainty, and some way of assigning weight to higher or lower levels of uncertainty within a policy calculus (very high variance in possible outcome might be worth it if the expected outcome is very positive, whereas very low variance might not be worth it if the expected outcome is very negative; this is a question of risk-tolerance).

That’s the way I see it, anyway.