Here's another tax-induced pecuniary externality that concerns me. The list price on college costs has been rising faster than inflation. This prompted politicians to act, and they created 529 plans to allow families to save for college in a tax-advantaged way. Austan Goolsbee has strongly and correctly criticized some of the design issues of these plans, particularly the administrative links to states and the (I think resulting) high management fees. For now, I just want to focus on how the tax advantage is distributed and its impact on future prices.
Contributions are made with after-tax dollars, sometimes with a state income tax deduction, but the returns on the portfolio compound tax-free and withdrawals are tax-free as long as they are used for college expenses. It's like a Roth IRA that way. (You can learn more here.) The benefits of the 529 plan accrue in proportion to a family's tax rate and desired amount of education expenses. Let's leave aside the almost surely positive correlation between (family) income and desired education expenses, which will reinforce the following points, and focus just on the simple fact that the tax rate increases with family income. It is more financially advantageous for a high-income family to invest a dollar in a 529 plan than it is for a low-income family to do so.
But you might say, "Okay, but doesn't the low-income family still get a benefit?" The answer depends on whether you think the supply curve for education is flat or upward sloping. Do you believe that colleges, when faced with dedicated accounts like 529 plans that will pay a penalty if they are not used on college costs, will raise their list prices? About 20 years ago, the conjecture that they would was named the "Bennett Hypothesis," after then-Secretary of Education William J. Bennett, who decried the tendency for colleges to raise tuition prices when the federal government stepped up its financial aid programs. I have little doubt that it is true.
The pecuniary externality comes in when we think about how much the tuition will go up. What drives that? I'd argue that it will be the average size of a 529 plan, as would be the case in any market responding to an increase in consumers' willingness to pay for a good. Since the tax advantage is positively related to income, even if all of the money going into 529 plans were new saving, it would be the higher income families that would have the larger-than-average 529 balances and the lower income families that would have the smaller-than-average 529 balances. (If the higher income families are simply shifting money from other accounts to 529 plans, then this strengthens the argument.)
Putting this all together, we can infer that the list price increases in college costs could outstrip the capacity of low-income families to pay them from their 529 plans. Depending on how much colleges raise their list prices and how the details of financial aid programs work out, lower income families may be worse off by the presence of 529 plans, even if they are saving through them. It is not the low-income families' own 529 plans that make them worse off--it is the high-income families' 529 plans and their greater benefits to using them. The impact of the latter on the price is the tax-induced pecuniary externality.
It's lousy public policy. But as much as I don't like these plans as a policy instrument, I have one for each of my two children. It doesn't make sense financially to leave the money on the table, given what's going to happen to list prices.