Wednesday, April 03, 2013

The pension "default" issue, and income vs. consumption taxation

A discussion at NYU Law School of a draft paper by colleagues on behavioral economics, plus the fact that Brian Galle's paper at the colloquium next week will be on the topic of "nudges," pushed me to an I hope moderately interesting thought about retirement policy through the tax code that I thought I would mention here.
Recent empirical findings, from a number of papers by leading researchers, suggest that employees' participation rates in employers' pension plans are enormously sensitive to the default setting.  That is, if they are automatically enrolled and would need to opt out, they participate at far higher rates than if they need to enroll by specifically approving in advance regular paycheck deductions to fund program participation.  This arguably undercuts viewing their decisions as rationally responding to consistent preferences about present versus future consumption, the tax planning issues, etcetera, since opting in or out is fairly easy to do, and thus seems verging on trivial relative to the significance of the choice on how one ends up.

Now, there are possibly "rational" expectations within conventional economics for this, at least under bounded rationality with limited information and time.  E.g., perhaps employees who are enormously responsive to the default setting regard it as a statement, by someone who is knowledgeable and has their best interests at heart, regarding what they most likely should do, etc.  I don't find this line of rationalization very persuasive, but suppose we conclude that greater participation is a good thing, whether people voluntarily choose it or not from either setting.  This could either be based on the notion that they will otherwise mistakenly under-save, or it could even reflect an externalities problem (they rationally under-save on this view that this will cause them to be treated more generously than others when they retire).

But for now, let's just suppose we accept the following propositions.  First, making pension contributions the default setting, and thus requiring employees to opt out if they'd rather get cash today, has much bigger behavioral effects than giving tax benefits to retirement setting.  Second, this makes both them and everyone else better-off.  Third, we could change the default, repeal the tax benefits for pension contributions, and get more retirement saving while also saving $100 billion per year, from the budgetary effects of repealing the tax benefits.

Deliberately overstating the claim in order to make it clearer, one could argue the following.  People are so inattentive that the tax benefits for retirement saving are like a lump sum subsidy (i.e., a lump sum tax with the sign reversed).  So there is simply no need for the subsidy if we change the default rule instead.

This admittedly is overstated unless one has quite a radical critique of the extent to which people focus on the actual long-term after-tax consequences of their labor supply decisions.  But there is an element of this claim even in more modest versions of it, which admittedly appear to receive significant empirical support.

OK, onto the punchline I want to add here, which I believe is more novel than anything said here so far.  Suppose initially that one sufficiently agreed with the above claim to support repealing pension tax benefits and relying on the replacement of opt-in with opt-out to produce high participation levels.  Now suppose that one also favored switching from an income tax to a consumption tax, based in part on the view that, outside the employee pension setting, people generally ARE deciding how much to save based in part on its tax treatment.  (Saving for future consumption is disfavored by an income tax relative to immediate consumption, whereas a consumption tax is neutral if the same tax rate applies in all years.)

Then one arguably should at least consider supporting a consumption tax in which employee pensions, unlike all other saving, actually ARE taxed under an income tax-like approach.  After all, it's still (by hypothesis) a lump sum tax, and the fact that you've changed the tax system's baseline approach to saving outside the employer retirement plan context arguably doesn't matter.

I don't quite mean this (i.e., enactment of a consumption tax is accompanied by distinctively applying unfavorable income tax treatment to retirement saving through work) as a serious proposition.  But it does help illustrate how conventional modes of tax policy thinking become less certain once one starts incorporating behavioral economics evidence and reasoning.


Werner Haslehner said...

Would behavioural economics not also suggest that the two cases are not really the same, as a 'tax penalty' is not the same as a 'tax subsidy', i.e. people (irrationally) do not care about a tax break they could obtain if (any) action is required, but they may nonetheless care about a tax penalty (as compared to the changed base-line of taxation) they face without any action?
I thought that may follow from the idea of loss aversion, which one may well translate into 'tax aversion'.

Daniel Shaviro said...

Yes, that is another point that could affect the analysis.