Tax-preferred retirement accounts are hugely popular among participants but widely panned by academics who study saving behavior. The two principal criticisms are (1) that tax incentives such as 401(k)s and IRAs aren’t effective at encouraging individuals to save more, and (2) that these incentives amount to “upside-down” subsidies that deliver the largest benefits to wealthy Americans in the highest tax brackets. In an excellent new article, David Kamin carefully considers these criticisms and concludes that the case against tax-preferred retirement accounts is overstated. “[B]ased on the available evidence,” Kamin writes, “tax incentives probably should be used to help correct failures in people’s saving decisions—correcting the ‘internality’ that people impose on themselves because of their biases.”
Kamin begins by canvassing the empirical evidence suggesting that a substantial fraction of American workers are “under-saving” for retirement. He goes on to evaluate the evidence that tax incentives can encourage individuals to save more. On this point, research results vary widely, with some studies finding that tax incentives have little or no effect and others finding that tax incentives play an important role in boosting rates of saving.
Kamin then takes a close look at a much-publicized recent study by economists Raj Chetty (now at Stanford), John Friedman (now at Brown), and Copenhagen-based collaborators who studied the effect of tax incentives on retirement saving in Denmark. Chetty et al. arrive at the dramatic result that each $1 of government expenditure on tax incentives for retirement saving raises total saving by only 1 cent. As Kamin notes, the Denmark study “has helped form the basis of an increasingly accepted conventional wisdom—despite some findings of the earlier literature—that tax incentives do not increase private saving to any significant degree.” Kamin’s critique of the Denmark study is nuanced—I won’t delve into the details here, except to say that pages 51 to 66 of Kamin’s draft provide an impressively clear explanation. Kamin’s re-analysis suggests that even in the Denmark case—which is Exhibit 1A for the “retirement tax incentives don’t work” point of view—tax incentives did have a positive effect on retirement security.
Showing that tax incentives can actually increase retirement security is a necessary element of any defense of such incentives, but it does not close the case. More broadly, Kamin stresses that individuals have heterogeneous preferences for saving and heterogeneous income requirements later in life. He argues that tax incentives—along with nudges such as automatic enrollment and a mandatory floor set by Social Security—play a useful role in light of the diversity of individuals’ wants and needs.
Yet even if we accept that (a) tax incentives for saving actually do result in individuals having more resources available for consumption in retirement, and (b) individuals have heterogeneous savings preferences and retirement income requirements, that does not necessarily mean that tax incentives help. The key question is whether tax incentives increase saving among the subset of the population that under-saves (and, beyond that, whether the response among under-savers is sufficiently large to justify the tax expenditure). Kamin acknowledges as much in the article: the case for tax incentives is, he says, a “tentative one.” But insofar as the case against such incentives is based on the premise that these incentives accomplish nothing, Kamin’s article provides a powerful defense.
As noted at the outset, the case against our current system of tax incentives for retirement saving is two-fold: the first critique is based on effectiveness (or lack thereof), and the second critique is based on distribution (the “upside-down subsidy” point). Kamin largely accepts the latter charge and argues that existing tax incentives should be reformed to remedy this flaw. At the end of the article, though, I found myself thinking that just as the ineffectiveness critique is overstated, the “upside-down” critique might be as well.
Kamin notes that traditional IRAs and 401(k)s provide two benefits: (1) the normal return on savings is exempt from taxation, and (2) for some participants, their tax rate at the time of withdrawal is less than their rate when they contribute. The second benefit is significant for individuals who find themselves in middle or high brackets at the time of contribution but fall into a low bracket by the time of withdrawal. It is of no benefit for individuals who remain in the same tax bracket (high or low) throughout their lives. This is arguably a feature, not a bug: saving increases welfare because it allows individuals with declining income to smooth consumption across their lifetime, but if an individual’s income does not decline in retirement, then it’s difficult to argue that they should save more.
Meanwhile, the first benefit—exemption of the normal return on savings—is already available to financially sophisticated, high-net-worth individuals even outside of tax-preferred retirement accounts. They can invest in zero- or low-dividend stocks, avoid realization, and if they want liquidity later in life, they can borrow against securities at attractive rates (more attractive the higher their net worth). At death, their heirs get stepped-up basis. In effect, financially sophisticated, high-net-worth individuals can create their own unlimited Roth IRAs. (If their preference is for fixed income assets rather than equities, they can pursue the same strategy through municipal bonds.)
One might argue, then, that existing tax incentives for retirement saving are in fact “right-side up.” They deliver the largest benefits to individuals who (a) experience a sharp drop in income at retirement, and (b) lack the financial sophistication (or the net worth) to pursue Roth-like investment strategies on their own. High-income individuals might on first glance look like they are benefitting the most, because the value of excluding 401(k) contributions from taxable income appears to be the largest for them. But these individuals might be about as well off in after-tax terms (and the federal government might collect about the same from them over the long haul) even if they didn’t have access to tax-preferred accounts.
This defense of traditional retirement tax incentives on distributive grounds is, like Kamin’s argument, a tentative one. And note that the defense is much weaker with respect to Roth accounts: Roths deliver the largest benefits to individuals who find themselves in a higher tax bracket at retirement; it’s very difficult to see why, from a consumption-smoothing perspective, we should subsidize saving among individuals whose income is increasing. Moreover, additional evidence would be needed to show that high-net-worth individuals can indeed arrange their affairs so as to obtain the same exemption benefit that existing tax preferences provide. At the very least, though, Kamin’s article should inspire us to think more rigorously—and perhaps more sympathetically—about tax-preferred retirement accounts.`