Friday, February 19, 2010

Incentives and Lags

Means-tested mortgage modification punishes people for having high incomes, and rewards them for having low incomes. I have pointed out that this may cause home owners to reduce their incomes, and think twice before making efforts to increase them.

Technically, a mortgage modification offers (immediate) payment reductions on the basis of the prior year's income. A number of economists have told me that this kind of means test would not affect the supply of income, because at the time of the modification, it is too late to go back and change one's prior year income.

This argument is easy to disprove by contradiction. Recall that personal income tax returns are filed in April, and the tax owed (or refund received) with the return depends on income in the prior calendar year. So, according to the above logic, the personal income tax does not affect the supply of income because, when taxpayers have to write the check to (or cash the check from) the IRS, it is too late for them to go back and change the prior year's income that determined the amount of the check.

The food stamp program is means-tested -- the more income a family has, the less food stamps it gets. Again, there is a lag between earning and benefit reduction. One common scenario is that the family files a quarterly report indicating its income for (some part of) the quarter, and that report is used to determine food stamps to be allocated in the following quarter (states vary in the timing details). By the above logic, food stamps' means test would not affect the supply of income of program participants because it is too late to change their previous quarter's income.

For most of the U.S. history of social security (and all of the history of many other countries), social security benefits were determined as a function of the elderly beneficiary's earnings: the more he or she earns over the limit, the less benefits paid. Again, the benefit reduction occurs in the following year. By the above logic, the work decisions of elderly people would be unaffected by the social security earnings test because, by the time benefits are paid, it is too late for the beneficiary to adjust his or her prior year's income.

As a matter of economic reasoning, you probably see that incentives would likely matter even when the cash flow consequences follow the incented behavior with a lag. But my point here is not about economic reasoning -- it's about empirical findings. A large literature has found that the income tax, means-tested subsidies, and earnings-tested subsidies all reduce the supply of income, despite the fact that the cash flow consequences of that supply follow with a lag (yes, economists argue about the magnitude of these responses, but they all admit that some kind of response is there).

What is remarkable is that economists have debated and written about marginal tax rates for decades, yet this "lag" issue was never raised as an excuse for high marginal tax rates -- until now. What is it about my exposition of incentives that finally elicits this reaction? Am I guilty of explaining things too clearly?

[Note that the personal income tax, food stamps, socials security etc., are ongoing programs, whereas mortgage modification on a massive scale is new to this recession. I agree that a new program may be less well understood. But also note that, unlike income tax payers and various means-tested program participants, homeowners can choose to delay (or redo) their modification until they get properly informed, and otherwise prepare their situation for the modification application.]

1 comment:

dWj said...

It's not current tax rates (on past income), then, that people are responding to, but expectations for future tax rates (on current and future income). If people believe (with certainty) that this is a one-off, it won't affect incentives (insofar as they're rational and not solvency-constrained). I think that's unlikely, though, in part because of the breadth of what "one-off" means here; not just that this policy isn't indicative of a future means-tested mortgage modification program, but that it isn't indicative of a propensity for future means-tested anything, which is surely false. If a one-shot means-tested mortgage modification program is followed by a one-shot means-tested education credit, which is followed by a one-shot means-tested retirement savings credit, reasonably intelligent people will start connecting the dots. In this sense I don't think anything is truly a one-off.