16 December 2008


Sorry if I mis-interpreted your positions as Austrian. They seemed to extoll a situation with classic market failures as able to succeed without government intervention. Yeah, if you believe that government intervention is needed for market failures then that would be neo-classicism. But, private info in insurance markets is one of the classic market failures, so a neoclassicist to me, would endorse government intervention here. But, I agree this is not justification of government intervention by any means. It's not the point of the post. The point of the post is, if the government is intervened as it is, how can it reduce costs.

The argument in rational ignorance is the following. You claimed that poor US policies could be understood if economists abandoned a notion that voters are self-interested outside of voting, but altruistic when voting. Yes, this is possible, but I'm not sure if I agree that economists adopt that model. Altogether differently, I prefer to understand poor US policies as the result of rational ignorance, or non-representative political system. Yes, satisfying these two policies are not sufficient conditions for good policies. But, not satisfying them is an intuitive way for me to understand poor policies.

My arguments are not arguments for government intervention. They are arguments for, if the government will intervene to provide government insurance such as Medicaid and Medicare, how can it do so for the lowest cost.

True, there aren't formal models of costs and benefits of specification of use of government transfers. Yes, at first glance it appears that is improving to eliminate constraints, but just because there aren't models of the benefits of these constraints, doesn't mean that in reality there aren't benefits of these constraints. For example, it could be modeled that the search and process costs of converting baby food stamps is a deterrent to converting food to money with which alcohol or drugs might be bought.

Again, these models are not complete by any means. Academic models are a work in progress, and in desperate need of improvement, that 's the point of grad school econ. The point of this blog is not to simply apply leading models to policies, (that is policy analysis), it is also to include pieces of reality that are not modeled. This is true for unconstrained redistributive transfer models, and for public choice models.

1 comment:

Ryan said...

I really hate to be "that guy", and I'm beating a dead horse while I'm at it. But I think I might be misunderstanding what you're saying about rational ignorance and positive theories of government. Rational ignorance is specifically about people being rational and self-interested. The entire public choice project is to assume that voters and politicians are people too (well, voters are anyway) and since we assume that people are rational, we should assume voters and politicians are rational. And this is very consistent w/ neoclassical economics. Neoclassical economics isn't a philosophy of "when markets fail, you should have the government intervene." It's a set of axioms coming out of individual rationality. Even "market failures" are about people being rational -- the only difference is that here self-interest leads to bad results.

So let's be good neoclassical economists. A good neoclassicist says public goods are under-provided, with provision going to zero as N grows large. What is good government? Non-rival and nonexcludable, for one thing. And voting? Voting for the public good? Them too. The US's N is very large indeed, so a consistent neoclassical economist says we should not expect these things. Let's suppose that people are informed and there's no problem with representation. Does this mean we expect socially optimal policy? Not at all! We're good neoclassical economists, and we know that even under the assumption of single-peaked preferences, democratic gov't doesn't imply optimal gov't, because there's no reason why the median voter should prefer the best policy. (Simple example: I propose we take $1 from everyone, throw away 1% of them, and redistribute them to everyone born on an odd-numbered day. There are more odd-numbered days than even, so this might very well win a majority [it certainly wouldn't violate rationality] but it's also obviously a bad policy. This is basically a dumbed-down version of Meltzer-Richards, right?) Again, this is basic neoclassical economics: people are self-interested, and there is by definition no reason why self-interested people should act with the goal of helping others. The whole idea of using gov't to correct market failures is to act for social interests ... but if we're talking about a market failure, we're already talking about a situation where rational actors wouldn't act in the social interest, which suggests we should be really careful about how we think the mechanism is supposed to work here. When people hand-wave about this, it basically amounts to them saying that they think gov't actors aren't rational (which is especially remarkable when talking about democracy -- particularly since none of the standard models of market failure make a lick of sense if people aren't rational).

I understand what you're saying about constraints on redistribution, but two points are in order. (1) It's not altruism; it's paternalism. I'm just saying that in a positive sense -- just that you can't model this as "your utility function is an argument in mine" (that is, it's not "I'm happy when you're happy" but "I'm happy when you do this, whether that makes you better off or not"). But this is important, and it'd be nice if more people would admit this (with one exception, everyone seems to take this as a normative criticism of intervention, which might suggest something). (2) "Good constraints" don't scale up well at all. It's one thing if some individual or group of individuals is crazy or irresponsible or can't be trusted to look out for their interests. Okay, fine, maybe that's true, and maybe we should constrain them. But how does this work for the whole society? Is everyone irresponsible? And suppose that's true: it's a little hard to see how that problem can be solved. If I'm a bit irresponsible, how can it help anything for me to be the one telling you what to do? If you're crazy, why are you constraining me? A model of "everyone is irresponsible" sounds like an argument for less government, not more. If no one can be trusted to manage themselves, then they really can't be trusted to manage others (e.g., vote).

As an addendum: if the goal here is redistribution, then why do we all tax-deduction? That's useless for the poor, and extremely useful for the rich. If the reason why people wanted the gov't to provide/subsidize health care was for redistribution, we would just have Medicaid. If we also wanted to redistribute to the very ill, we would just be talking about catastrophic insurance. (I really want to stress this. Catastrophic insurance is insurance. Everything else is not. We're specifically subsidizing chronic, regular care. I mean, c'mon: Medicare? Is uncertainty a big feature in higher health expenditures by the elderly? Not so much. The only uncertainty is whether you'll live that long -- which of course cuts against redistributing to the old. And this is a general thing. To an extremely large extent, "health insurance" is not insurance in the economic sense of the term, which, again, is why adverse selection is sort of irrelevant here, even it does swamp positive selection.)

I understand that your main goal isn't to justify gov't policy, but to improve it. I agree enthusiastically. That's kind of the point of all this -- it's why I first commented. I'm saying, hey, is there anyone out there who can give a consistent, economically-literate defense of the basic outlines of US health care policy? And specifically, of the parts that very much aren't insurance (i.e., the vast majority of it)? If there isn't, then we can make a very simple, very easy improvement: get rid of it. I mean, this is basically intro-econ stuff, right? Demand curves slope down. If you subsidize something, you get more. Subsidize a lot, get a lot more. So what's the reason to suppose that we already get way, way too little spending on regular health care? I don't mean to put you on the spot, and obviously you shouldn't be expected to know everything. Let's relax a bit here. Suppose there were massive chocolate subsidies, such that chocolate were basically free. If you had no theory of saying people buy too little chocolate, and some evidence that they don't want any would you say "my arguments are for, if the government will intervene to provide chocolate, how can it do so for the lowest cost ..." and point out that economics is a work in progress, and after all, how much can we trust those studies saying more chocolate doesn't make people better off? Or would you say, hey, if no one can give a good reason for this, maybe we should stop subsidizing chocolate? It's obvious: absent argument for, you'd be against. So why is the threshold of evidence for health care set so high?

I hope I don't sound like I'm being antagonistic here. I'm just very interested in these topics, and I really appreciate your patience in talking through them.