Let's forget monetary policy because that adds a lot of macro in that's really not important for the point here. Just think about a hyperbolic discounting case and saving for retirement, because that has all the essential elements.
At a given point in time, you are a perfectly rational utility maximizer to blow your paycheck on frivolities. That decision is optimal as economists use the term "optimal". But at a future period, the path you've set yourself on no longer looks optimal. That's my phrasing - but is it fair enough? Does everyone familiar with the concept consider that accurate and does everyone not familiar with the concept get the point?
So what you conclude from this is that sometimes limiting your options or constraining yourself to a particular action can actually be welfare enhancing even if it doesn't feel that way at a given point in time. The traditional example is default contributions to a retirement plan. Out of sight, out of mind - and no need every two weeks to make the decision to save which you are doomed to mess up if you have that discretion.
Now, what I've said a couple places now (search for them yourself), is that anybody who understands this and commits to a default contribution plan practically by definition is doing what they "want" to do. As a commenter on Bob's blog pointe out, it's practically a tautology. That's true which is why I'm surprised I've gotten so much grief over it.
Staying within the neoclassical framework, would you ever commit to default contributions given a standard hyperbolic discounting function? No, you'd never ever ever commit to it. We would see none of thes plans out there.
And yet these plans exist!
How do we explain such a thing? Well I think the first step is to recognize the parsimony with which economists have chosen to model preferences. Clearly the hyperbolic discounting preference scheme won't allow you to sign up for any default contributions. So we have one of two choices:
- Assume irrational departures from otherwise rational behavior, or
- Assume that while hyperbolic discounting is an important way to characterize the observed actions of the human animal, there are a lot of times when our preferences actuallly do incorporate the preferences of our future selves as well (that's why we commit, after all!)
OK, with that in mind, back to monetary policy.
I know complete discretion is dangerous. This is why we like rules. This is also why we like constitutions, which are just rules that (1.) are harder to change and (2.) against which other rules are judged.
If you say "I think we should have a policy rule" or "I think we should have a constitutional rule", or if you're like me and say "I see some danger to making it a constitutional rule so I'm on the fence about that distinction but we should definitely have a credible rule whether that credibility comes from a constitution or from somewhere else", then practically by definition you are a person that likes what the rule tells you to do. You are a person that is weighting the preferences of your future self along with the preferences of your current self. Otherwise there is very little reason for you to go for the rule.
In the same way that default retirement contributions exist in the real world, constitutional and policy rules exist in the real world.
They exist because there are people out there who like them.
And they like them because the parsimonious hyperbolic discounting model - while very insightful for understanding the human animal - is not a perfectly accurate rendition of actual preferences.
One more point to confuse the matter further. The time inconsistency in the hyperbolic discounting function comes from an additional coefficient on utility in future periods besides the standard discount factor. You're basically assuming the conclusion. You're saying "I'm just tacking on this coefficient to make these preferences time-inconsistent".
That's a little different from how time-inconsistency emerges in the policy models. One of the things that has always irked me about the way these are set up is that in a way you only get the result if you initially assume some ignorance (that under reasonable circumstances may not be warranted) on the part of the policymaker. There's no crazy discounting. The problem in the model is that output and inflation don't actually end up responding in the way you might expect them to given the simple task of maximizing the objective function. Once inflation expectations are taken into account, simply maximizing the objective function doesn't give you the right result. So you figure out a rule that does give you the right result (it turns out that rule is setting λ = 0), and you're set.
But this is a "rule" in the sense that you presumably have the same objective function (this is the sense in which you are doing something "you don't like").
But λ is actually in your objective function! (this is the sense in which accepting the rule implies you do like it).
So the monetary policy models are a lot goofier in that sense, but the point is the same. You only make rules or constitutions to constrain your behavior if you see some value to constraining your behavior: if you think that constraining your behavior is desirable or preferable or... dare I say it... utility maximizing.