The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
September 1, 1991 eCONOMIG COMMeNTORY Federal Reserve Bank of Cleveland i Increasing National Saving: Are IRAs the Answer? by David Altig and Katherine A. Samolyk k3aving, so advised the proverbial ant to the spendthrift grasshopper, allows an individual to "prepare today for the wants of tomorrow." Likewise for an economy, national saving provides resources for capital accumulation, expanding the economy's future productive capacity and hence its future potential output. The precipitous decline in average saving rates during the 1980s, as measured by the National Income and Product Accounts (NIPA), has convinced many that we have become a nation of grasshoppers in need of policies that will turn us back into a nation of ants. Among the currently popular policy options aimed at increasing national saving are proposals that would establish personal saving incentives by expanding federal tax deductions for contributions to individual retirement accounts (IRAs). One version of this type of policy was recently introduced by Senators Lloyd Bentsen (D-Texas) and William Roth (R-Delaware). At least some of the support for IRA tax subsidies is a result of recent U.S. experience. Preferential tax treatment of IRAs, which had been available to a limited number of people since the mid1970s, was made available to all wage earners by the Economic Recovery Tax Act of 1981 (ERTA). These deductions were substantially curtailed by the Tax Reform Act of 1986 (TRA86).3 Contributions to IRAs grew dramatically in the period between ERTA and TRA86, accounting for nearly 25 percent of ISSN 0428-1276 personal saving by 1986. Although their claims are controversial, several formal studies suggest that a considerable fraction of this growth represented net additions to personal saving.4 Even if this is true, evidence that IRA subsidies are effective in increasing personal saving does not necessarily support reinstatement of the subsidy provisions in place prior to TRA86. One well-known problem that policymakers must consider involves the contribution of IRA tax breaks, which reduce federal revenues, to growth of the federal deficit. Even if IRA subsidies do increase personal saving, it is far from clear that they have a positive effect on national saving. In the current debate, this tension between IRA policy and deficit reduction efforts is manifested in budget rules requiring that all lost revenues be replaced through expenditure reductions or alternative taxes. A further, and less discussed, problem is the potential link between proposed IRA subsidies and household debt. Allowing tax deductions for IRA contributions can — and will, we argue — create incentives for increased household borrowing. At a time of growing concern about the level of private-sector debt, it would indeed be ironic if policymakers adopted a tax policy that had the consequence, unintended as it might be, of increasing household leverage. In contrast to policies that subsidize particular forms of personal saving are In response to falling U.S. saving rates, Congress is considering legislation that would once again expand federal tax deductions for IRA contributions. But evidence shows that attempting to stimulate national saving by subsidizing personal saving has serious drawbacks, such as incompatibility with recent budget reforms and increased incentives for household borrowing. The authors argue that curtailing deductions for personal interest expense would be a more effective strategy for increasing national saving. those that would eliminate subsidies to particular types of personal borrowing. An example of the latter is legislation introduced by Congressman Frank Guarini (D-New Jersey) that would restrict interest deductions on certain purchases made from home-equity loans. The problems that mitigate the usefulness of IRA subsidy plans as policy tools for increasing U.S. saving rates do not arise under the type of policy contemplated in the Guarini proposal. Our principal message in this Economic Commentary is that, given the goal of increasing national saving by inducing greater personal saving, IRA subsidy schemes are inferior to policies that would eliminate incentives for household debt accumulation. • IRAs, Consumption, and the Federal Deficit A subsidy is essentially a gift—a gift with strings attached, perhaps, but a gift nonetheless. And so it is when the government grants tax deductibility for IRA contributions: If you agree to the rules governing these contributions, you get the gift. Suppose, then, that your IRA contribution yields a gift of $ 1,000. Would your saving increase? Quite likely it would. Under what most economists consider normal circumstances, some of the subsidy-provided windfall would be put away "for the wants of tomorrow." But — and here's the essence of the problem — some of the subsidy would immediately contribute to increased consumption. If the hypothesized IRA subsidy is not matched by a decrease in government consumption (or an increase in other taxes), every dollar of subsidy becomes a dollar of public-sector dissaving, or a dollar increase in the government sector's deficit. Thus, even if private saving expands by some fraction of the subsidy/gift, national saving (the sum of private and public saving) actually falls and national consumption rises. Where would the additional resources needed to satisfy this increase in consumption come from? They must arise from one of two sources: domestic firms or foreigners. If current consumption comes at the expense of resources available to domestic firms, then investment falls, and with it the capital stock available for future production. If current consumption is provided by foreigners, then they must be repaid from future production. In either event, the increase in current consumption will reduce the amount of domestic consumption available in the future. The implication of these observations for interpreting data from the 1982— 1986 IRA experiment is obvious. It is simply not sufficient to suggest that preferential tax treatment of IRAs caused personal saving to rise, as several studies have implied. The relevant issue, and the one captured in the con- cept of national saving, concerns the degree to which the policy encouraged or discouraged aggregate consumption. On this score, both casual observation of the path of national saving and formal econometric analysis suggest a healthy skepticism about the proposition that IRA subsidies over this period contributed in any substantial way to expanding the U.S. capital stock.' • IRA Subsidies in the Brave New Budget World The political environment into which new tax breaks must now be born differs considerably from the one that saw passage of ERTA. The "pay-as-you-go" provisions of last year's federal budget agreement preclude financing IRA tax deductions by expanding the deficit. Under the provisions of that agreement, any projected revenue loss must be counterbalanced by other taxes, reductions in federal expenditures, or both. Although the pay-as-you-go restriction is not necessarily an argument against adoption of IRA subsidies, it does introduce elements that increase the uncertainty about such a policy's net effects. If, for instance, projected losses from IRA subsidies are compensated with alternative taxes, then any saving disincentives associated with these new taxes must be factored against the presumed benefits of the IRA policy. Alternatively, if IRA subsidies are "paid for" by government spending cuts, the shift in resources from the government to households could increase capital accumulation if, as we have argued is likely, households respond by saving at least a fraction of the IRA tax break. But Congress could get a "dollar-fordollar" increase in public saving by reducing federal expenditures directly. In other words, the value of a reduction in government expenditures is not logically related to IRA policy. A broader point is that the new federal budget rules inevitably require the proposed IRA subsidies to be bundled with other, unrelated fiscal policies. Consequently, political obstacles and economic uncertainties confronting the proponents of new IRA policies are magnified. • IRAs and Household Debt Accumulation Concerns about the downward trend in U.S. saving rates have been paralleled by reservations about the recent increase in personal debt accumulation and by the perception that relatively high personal debt levels make the U.S. economy especially vulnerable to macroeconomic shocks. Indeed, concerns about low saving rates and high personal debt burdens are often raised by the same voices that sing the praises of tax-induced saving incentives. Ironically, a policy designed to increase the personal saving rate by reintroducing tax breaks for IRA contributions can have the unintended side effect of increasing household leverage. The potential for IRA subsidies to induce more borrowing is widely appreciated. As noted by the Treasury Department's Donald Lubick in testimony before the House of Representatives' 1980 hearings on saving incentives, taxpayers "have the possibility of gaming the system through borrowing and taking interest deductions and putting the borrowed money into the tax deferred account. ,,6 The opportunity to engage in this type of tax arbitrage introduces powerful incentives for debt accumulation, even in cases where IRA subsidies do induce a household to increase its net saving. In fact, it appears that IRA subsidies simultaneously increase borrowing and net personal saving: Taxpayers who purchased IRAs between 1980 and 1984 tended to have higher growth in debtrelated interest deductions than did otherwise similar taxpayers who did not purchase IRAs.7 What is even more pertinent to the issue of household financial fragility is that this behavior was particularly pronounced among taxpayers with lower levels of taxable wealth. Although the existence of tax loopholes makes the interpretation of taxable wealth levels tricky, low-wealth households are exactly the group that we might expect to be FIGURE 1 HOUSEHOLD DEBT, 1970 -1989 Percentage of GNP 451 40 35 Home mortgages 30 25 20 15 10 5 Other consumer credit I I 1970 1975 1980 1985 SOURCE: Board of Governors of the Federal Reserve System. most susceptible to payment problems in the event of a sudden drop in income. Of course, opportunities to leverage IRA-type investments not only increase household debt burdens, but also diminish the incentive to reduce consumption. In the final analysis, all policies designed to induce more saving must be judged by the standard of their effects on aggregate consumption. • An Alternative None of the problems discussed so far — the possibility of larger federal deficits, the problematic nature of avoiding larger deficits with other fiscal policies, and increased incentives for household borrowing — imply in and of themselves that IRA subsidy plans should be avoided. As with all policy options, the relevant calculations require weighing the costs and benefits. If there were a reasonable expectation that the net effect of such subsidies would be an increase in national saving, then the real or potential costs might be well worth bearing. One should ask, however, whether other policies could attain the objective without encountering these complications. We contend that there is a straightforward alternative: Make borrowing less attractive by restricting the deductibility of household interest expenses. A policy that reduced household dissaving would obviously serve the ultimate end of increasing personal saving. More to the point of our argument, such a policy would not suffer the same deficiencies as one that attempted to increase personal saving via IRA subsidies. Because eliminating or reducing interest deductions would raise revenues directly, the policy could be pursued without expanding the federal deficit or worrying about additional policies to maintain budget balance at the margin. Furthermore, a policy that removed the incentive to borrow would simultaneously discourage debt accumulation and encourage saving rather than consumption. Thus, the household leverage issues that appear in the IRA subsidy case simply would not arise under an approach that limits interest deductions. But would such deductions truly have the desired effect of reducing household borrowing and increasing personal saving? We claim that evidence from the 1980s, while ambiguous with respect to the effectiveness of IRA policies, strongly supports the proposition that tax incentives have a profound effect on household borrowing behavior. • Personal Interest Deductions and Saving: Some Evidence Evidence that personal interest deductions can exert a powerful influence on household borrowing can be found in the pattern of household debt accumulation subsequent to TRA86. The key element of TRA86 in this regard was the elimination of deductibility provisions for interest payments on debt not secured by real estate. Thus, while mortgage interest retained its status as a deductible expense, interest on debt explicitly associated with consumption expenditures — primarily consumer installment credit — lost its preferential tax treatment. Figure 1 illustrates the behavior of different categories of household debt from 1970 to 1989. It is apparent that the growth in installment credit (as a percentage of GNP) characterizing the period from 1983 to 1986 came to a halt after 1986. The "other consumer credit" category, which reflects nonmortgage-related, nonrevolving debt (and so includes obligations that are less likely to be interest sensitive), was relatively unaffected by TRA86, continuing the slight downward drift relative to GNP that has characterized this series since 1970. Home mortgage debt, on the other hand, exploded after 1986, rising 26 percent relative to GNP between 1986 and 1989. This surge in mortgage-related debt cannot be accounted for by rapid growth in residential real estate acquisition. The ratio of the value of residential structures to GNP has been fairly stable since the end of the 1981-1982 recession. It is more likely that most of the growth in mortgage-related debt since 1986 reflects a shift in borrowing behavior from installment and other nonmortgage debt, for which interest payments are no longer tax deductible, to debt that is secured by residential real estate, which is still tax preferred. Indeed, the Federal Reserve's 1988 Survey of Consumer Attitudes indicates that 64 percent of existing home-equity lines of credit (and about 20 percent of traditional homeequity loans) had been originated since 1986. Furthermore, 35 to 40 percent of these loans had been specifically used in repayment of other debt. Additional evidence on the relative effectiveness of IRA subsidies compared with limitations on personal borrowing deductions can be found by examining cross-country saving behavior. For instance, a significant part of the difference in U.S. and Canadian saving rates between 1961 and 1985 can be explained by the fact that interest payments on consumer debt were fully deductible here, but not in Canada. In fact, evidence indicates that the tax treatment of interest expense may be a more important determinant of saving behavior than the tax treatment of long-term saving instruments like IRAs. • Conclusion Many well-informed and thoughtful people believe that policies designed to increase national saving rates are imperative for the long-term health of the U.S. economy. In light of this, it is easy to understand the appeal of tax policies that would provide incentives for higher levels of long-term saving. Unfortunately, proposals to subsidize IRAs suffer from complications ranging from ultimate ineffectiveness to political infeasibility to undesirable side effects. Past experience shows that an alternative tax policy that makes borrowing less attractive would be effective and would avoid many of the complications associated with IRA subsidies. If we are to look toward incentives in the personal tax code to stimulate capital accumulation, our efforts should focus on policies that discourage dissaving rather than on those that resurrect saving incentives with ambiguous track records. • Footnotes 1. "The Ant and the Grasshopper," Aesop's Fables. 2. There are very good reasons to believe that the NIPA measure of aggregate saving is a poor indicator of the true rate at which the U.S. capital stock is growing (see, for example, Robert Eisner, "The Real Rate of U.S. National Saving," Review of Income and Wealth, vol. 37 [March 1991], pp. 15-32). But regardless of how adjustments to the NIPA calculations may affect measured levels of national saving, the implication that saving rates were, on average, historically low in the 1980s remains true. In any event, we take the perceived need for higher saving levels as a given and discuss the efficacy of alternative tax policies in that context. 3. TRA86 allows full tax deducibility of IRA contributions for taxpayers with adjusted gross income (AGI) of less than $40,000 and for taxpayers not covered by private pension plans. Partial deductibility is available to taxpayers whose AGI is between $40,000 and $50,000. No deductions are allowable for taxpayers whose AGI exceeds $50,000. By 1985, almost 38 percent of all IRA contributions were made by individuals whose income exceeded $50,000, while more than 50 percent were made by persons with incomes greater than $40,000. TRA86 contains no provisions to adjust these cutoff income levels for inflation. In practice, this means that the real income level at which IRA deductions are phased out declines every year. 4. An excellent critical overview of these studies can be found in Jane G. Gravelle, "Do Individual Retirement Accounts Increase Savings?" Journal of Economic Perspectives, vol. 5 (Spring 1991), pp. 133^*8. Federal Reserve Bank of Cleveland Research Department P.O. Box 6387 Cleveland, OH 44101 Address Correction Requested: Please send corrected mailing label to the above address. Material may be reprinted provided that the source is credited. Please send copies of reprinted materials to the editor. 5. On the former point, see Laurence J. Kotlikoff, "The Crisis in U.S. Saving and Proposals to Address the Crisis," National Tax Journal, vol. 43 ( September 1990), pp. 233-46. On the latter, see William G. Gale and John Karl Scholz, "IRAs and Household Saving," unpublished manuscript, University of California, Los Angeles, July 1990. 6. This testimony is recounted by Daniel Feenberg and Jonathan Skinner, "Sources of IRA Saving," Tax Policy and the Economy, Cambridge, Mass.: National Bureau of Economic Research, 1989, pp. 25-46. 7. See Feenberg and Skinner, op. cit. 8. See Glenn B. Canner and Charles A. Luckett, "Home Equity Lending," Federal Reserve Bulletin, vol. 75, no. 5 (May 1989), pp. 333^4. 9. See David Altig, "The Case of the Missing Interest Deductions: Will Tax Reform Increase U.S. Saving Rates?" Federal Reserve Bank of Cleveland, Economic Review, vol. 26, no. 4 (1990 Quarter 4), pp. 22-34. David Altig and Katherine A. Samolyk are economists at the Federal Reserve Bank of Cleveland. The authors are grateful to Michael Bryan, Jeffrey Hallman, and Sharon Parrottfor useful comments. The views stated herein are those of the authors and not necessarily those of the Federal Reserve Bank of Cleveland or of the Board of Governors of the Federal Reserve System. BULK RATE U.S. Postage Paid Cleveland, OH Permit No. 385
@FedFRASER