Center Sponsors Briefing on Economic Impact of Climate Change Policies

Issue Brief: Assessing the Effectiveness of Saving Incentives

ACCF Center for Policy Research: 1996 Research Projects

Senator Frank Murkowski (R-AK), Chairman of the Committee on Energy and Natural Resources, joined top experts on climate change issues for an ACCF Center for Policy Research-sponsored briefing for Hill staff and the press on March 28.

The briefing presented new research on the economic impact of climate change policies now being debated in Europe, the United States, and the United Nations, and explained the effects of proposed climate change policies on U.S. economic growth and competitiveness, the world economy, and concentrations of carbon dioxide in the atmosphere. Senator Murkowski and the scholars also shed light on current international discussions aimed at amending the 1992 Rio Treaty (Framework Convention on Climate Change). Understanding the impact of these international discussions is critical because a decision on post-2000 greenhouse gas emissions, which could have important economic ramifications for the United States, is expected shortly.

Senator Murkowski supported the need to back climate change negotiations with rigorous economic analysis to insure that the consequences of mandated greenhouse gas reductions for U.S. growth, jobs, and the economy as a whole are known. "Near-term CO₂ reduction measures may impose steep economic costs without providing environmental benefits," the senator stated. "Our goal is to ensure that the climate change debate is a balanced debate, mindful of science and economics. If human-induced climate change is a problem that demands action, we want them to be the right actions, taken at the right times, in the right places," Murkowski concluded.

Climate change experts participating in the briefing included Dr. W. David Montgomery, vice president of Charles River Associates; Dr. Jae Edmonds, technical leader of economic programs, The Battelle, Pacific Northwest Laboratories; and Dr. Thomas Schelling, Distinguished University Professor, University of Maryland.

Dr. Montgomery advocated an economically rational approach to climate change policies that addresses three related questions: (1) What actions can be supported on the basis of current understanding of climate science and economics? (2) How can the best use be made of the new information that current scientific research and technology development will provide? and (3) When should emission reductions begin? Dr. Montgomery argued that the Berlin Mandate prejudges all these issues by concentrating on near-term targets for emissions from industrial countries. He also suggested, "Specific policies and measures must be evaluated, not general goals. No goal can be assessed without specifying the policies likely to be used to achieve it."

Dr. Edmonds analyzed the cost of various CO₂ emission reduction options. He stated that "by allowing for flexibility in where and when emission reductions are made, costs could be substantially reduced through international cooperation and the optimal timing of emission reductions." Such flexibility can reduce costs by more than 80 percent, potentially saving the international community trillions of dollars in mitigation costs, observed Dr. Edmonds. "Reliance on more flexible alternatives reduces costs more effectively than adopting weaker, but still inflexible, commitments," he concluded.

The Berlin Mandate does not require the developing countries to participate in any efficient greenhouse gas abatement program, but the costs of decarbonizing their economies, at least for the first 50 years, will have to be borne by the wealthy countries, according to Dr. Schelling. In this way, greenhouse gas abatement becomes a foreign aid program, since the benefits will almost exclusively accrue to the developing countries.

A nagging question remains that usually does not get attended to at meetings on greenhouse climate issues, Dr. Schelling added. The question is whether three or four generations hence, we get more benefit for our money by investing directly in public health, education, water resources, infrastructure, industry, and family planning in developing countries, for example, than we obtain from reduced climate change. An added concern is whether the benefits accrue earlier, to people who more desperately need them, when we apply the resources directly to development. "We must stop treating the international greenhouse gas problem in isolation from the developing world's economic needs, Dr. Schelling concluded.

An Economic Perspective on Climate Change Policies, the new book just released by the ACCF Center for Policy Research, contains studies and commentary by these scholars and others. To order, contact the Center at 202/293-5811; 202/785-8165 fax. (ISBN: 1-884032-05-2 / February 1996 / 236 pages / paperback / U.S. $25.00.)

The question of whether saving incentives such as Individual Retirement Accounts (IRAs) and 401(k) plans stimulate new saving or merely reshuffle existing assets has long been of interest to policymakers and academic experts. Professors R. Glenn Hubbard of Columbia University and Jonathan S. Skinner of Dartmouth College have recently completed a review of the scholarly research on this question and find that most studies suggest that IRAs and 401(k)s do increase net saving.1 The Hubbard/Skinner (H/S) analysis also provides a new cost-benefit approach for analyzing the "success" of saving incentives where success is measured by comparing the ratio of new saving to the government tax revenue lost because of the deductibility of IRAs and 401(k)s.

Background on Saving Incentives

During the last twenty years, a number of public policies have been introduced in the United States to provide incentives for personal saving, H/S note. IRAs were developed in the 1970s to provide a tax-deductible saving program for employees who lacked pension plans. Despite the fact that half of all workers were eligible because they had no pension plan at the time, fewer than one percent of taxpayers contributed.

IRA eligibility was expanded by the Economic Recovery Tax Act of 1981 so that most working taxpayers could contribute, and IRA limits were increased. From 1981 to 1982, IRA contributions increased from $5 billion to $28 billion. During the 1982-1986 period, households contributed more than $170 billion to IRA accounts; by 1986, IRA contributions were about one-fifth of aggregate personal saving.

The Tax Reform Act of 1986 curtailed IRAs so that IRA contributions were not fully deductible except for workers without employer pensions, families with annual incomes under $40,000, and individuals with incomes under $25,000. Tax-deductible contributions to IRAs fell by 62 percent in 1987, and have remained low since then.

In recent years, another type of targeted saving program, the 401(k) plan, has become prominent. Although 401(k)s have been available since 1978, only since the Treasury Department clarified rules for their use in 1981 have they attracted interest from companies who had previously maintained taxable saving plans, or who were looking for alternatives or supplements to defined-benefit pension plans. In many ways, 401(k)s are similar to traditional IRAs: they involve tax-deductible contributions, tax-free inside buildup, limits on annual contributions, and restrictions on early withdrawals. However, there are differences. The plans are available only to employees of organizations that elect to sponsor such plans, and employee contributions occur through regular payroll deductions, whereas IRA contributions may be made at the employee's discretion. In addition, employers are permitted to supplement employee contribution rates to 401(k) plans.

Do IRAs and 401(k)s Increase Saving?

H/S survey nearly all of the scholarly studies on the effectiveness of saving incentives. Their report shows that most of the studies over approximately the last twelve years find that IRAs and 401(k)s do increase net saving, and do not merely result in taxpayers moving existing assets from one account to another. Among the scholars whose research supports the effectiveness of saving incentives are Steven Venti of Dartmouth College, David Wise of Harvard University, Daniel Feenberg of the National Bureau of Economic Research, Douglas Joines of the University of Southern California, Jonathan Skinner of Dartmouth College, James Manegold of the University of Southern California School of Accounting, Martin Feldstein of Harvard University, James Poterba of the Massachusetts Institute of Technology, and Hubbard himself.

For example, Poterba, Venti, and Wise (1994) analyze the saving behavior of workers eligible and not eligible for 401(k)s. Their study shows that after controlling for incomes of workers, levels of wealth among those eligible for 401(k)s are substantially higher than among those not eligible. Figure 1 focuses on a given annual income range, $40-50,000, to correct for differences in income between the two groups. In 1984, median financial assets, excluding 401(k) and IRA balances, were roughly the same between the two groups, as shown in the first two columns on the left. Those assets remained generally unchanged between 1984 and 1991. (Many eligible workers do not, in fact, contribute anything to a 401(k) plan; Poterba, Venti, and Wise include them in the sample to avoid the criticism that individuals who choose to contribute to 401(k)s were eager savers anyway.) From 1987 to 1991, however, Poterba, Venti, and Wise found that the median financial wealth of those eligible for 401(k) plans rose sharply, largely because of 401(k) contributions.

On the other hand, H/S survey and critique the research of scholars such as Eric Engen and William Gale of the Brookings Institution, John Karl Scholz of the University of Wisconsin-Madison, Thomas De Leire of Stanford University, and Orazio Attanasio of the Universitas di Bologna in Italy. These authors conclude that saving incentives generate little or no new saving, at least in the short run.

After careful review of the literature on saving incentives, H/S conclude that the actual impact of IRAs and 401(k)s likely lies between the divergent positions sometimes espoused in the literature, that is, "saving incentives do nothing for national saving" versus "saving incentives are nearly all new saving." Their analysis suggests that each dollar contributed to an IRA results in $0.20 to $0.40 in new saving.

Measuring the Effectiveness of Saving Incentives

Taking the middle ground on the effectiveness of saving incentives raises an entirely new set of questions, according to H/S. Suppose, for example, one takes the view that only 26 cents of every dollar in IRA contributions represents new saving, as found in a 1995 study by Joines and Manegold. Does this imply that IRAs are ineffective and just a drain on the U.S. Treasury? Or should 26 cents of new saving per dollar of IRA contribution be taken as a mark of success for saving incentive programs? To answer this question, H/S develop a cost-benefit analysis of saving incentive programs.

Suppose, H/S postulate, that a particular saving incentive generates only 4 cents of new saving per dollar of contribution. Determining whether this saving program is successful depends on its cost in terms of foregone federal tax revenue. If the program loses only 1 cent of revenue per dollar of contribution, then the answer might well be yes-after all, one gets 4 dollars in new saving per dollar in revenue cost. This benefit-cost analysis for the IRA program is captured by defining the following ratio:



The numerator and denominator are stocks rather than flows, and are defined for a particular time tafter the initial IRA contribution. For example, suppose that the taxpayer is in the 36 percent tax bracket, and that 26 cents of the IRA contribution is drawn from new saving. The ratio in the equation above would therefore be 62 cents (36 cents saved through reduced tax liability plus 26 cents of new saving) divided by the revenue loss of 36 cents. The benefit-cost ratio for the first year after the IRA contribution is 62/36, or 1.72. Thus, there is an increase in private saving of $1.72 per $1 loss in revenue.

H/S state that the benefit-cost ratio in the first year alone from IRAs is misleading, because there is a greater revenue loss over time from income from funds that would have been saved in taxable form. In addition, IRAs generate revenue when they are cashed out. In calculating the benefit-cost ratio, H/S focus on the change in the stock of private wealth accumulated over the period for which the IRA is held, divided by the accumulated revenue loss over the same period.

H/S's benefit-cost calculations require assumptions about interest rates, tax rates, the length of time for which the IRA is held, and the tax treatment of the saving had it been saved in a taxable form. H/S use the tax regime in effect from 1982 to 1986 because most of the estimates from the academic studies are based on data from that period. They assume a holding period of 22 years, which corresponds to buying the IRA at age 50 and cashing it out at age 72, an initial marginal tax rate of 36 percent (1995), a final retirement tax rate of 28 percent, an average tax rate on interest and dividend income of 32 percent, and a 60 percent exclusion for capital gains. The representative portfolio, whether invested in an IRA or taxable assets, is assumed to be 29 percent in equity initially, with the remainder in a combination of long-term and short-term bonds, an aggregate portfolio consistent with 1985 data. During the period from 1900 to 1990, the geometric mean of the nominal return in the stock market was 9.35 percent, and the geometric mean of a portfolio with one-half short-term bonds and one-half long-term bonds was 4.0 percent.

H/S's calculations of the additional private capital accumulation per dollar of foregone revenue for a range of estimates of new saving per dollar of IRA contribution are shown in Table l. The first row in Table 1 shows how the marginal impact of IRAs on capital accumulation depends on assumptions about the fraction of IRA contributions that are new saving. If there is no new saving from the IRA contribution (as in Gale and Scholz [1994]), a debt-financed IRA program leads to an increase in private saving of $0.22. Under the assumption that the IRA is debt-financed, the net national capital stock would fall by $0.78 (a $0.22 increase in private saving, and $1 reduction in government saving). At the compromise estimate of 26 cents in new saving, as suggested by Joines and Manegold (1995), the implied increase in private capital accumulation is $2.21 per dollar devoted to the IRA program. Again, for a deficit-financed IRA, the net capital stock increases by $1.21. A relatively modest saving effect of IRAs can translate into a quite substantial "bang for the buck" in terms of capital growth per dollar of foregone tax revenue. The estimated effects are even larger when the marginal saving effect is 40 cents per $1 IRA contribution ($4.31 increase in the private capital stock) or $0.60 ($12.01 increase in the private capital stock).

H/S note that the tax code changed substantially in 1986 due to the Tax Reform Act. In order to measure the impact of IRAs on saving using more current parameters, H/S assume a marginal tax rate of 28 percent for contributors, a 24 percent marginal tax rate at retirement, no exclusion for capital gains, and an average 26 percent tax rate on dividends and interest. To reflect the increasing share of equities in IRA and Keogh plans, H/S also assume that plan assets are divided equally between stocks and bonds. The third row in Table 1 shows the incremental effects on capital accumulation under this scenario. While the negative impact of the IRA on capital accumulation is more pronounced when there are no saving effects at all from IRAs (first column), the incremental impact is quite similar ($2.09) to the pre-1986 tax rules for their assumed midpoint estimate of 26 cents of new saving.

H/S find that IRAs need not stimulate very substantial amounts of new saving per lost dollar of revenue to generate favorable marginal increases in the capital stock per dollar of initial revenue loss. In fact, even if the aggregate effects of a given IRA program are not large in terms of overall increases in new net saving, the revenue costs, once properly accounted for, are even smaller, the authors observe.
The effectiveness of 401(k) plans also depends on how they affect capital accumulation per dollar of foregone revenue, just as for IRAs, state H/S. There are fewer scholarly studies of how 401(k)s affect saving behavior, however, so developing a benchmark estimate is more difficult than for IRAs. If, on the one hand, 401(k)s were entirely effective at generating new saving, the incremental private capital accumulation per dollar of revenue cost would be $94, which would make them extremely effective. If, on the other hand, sorting or reshuffling accounts for half of the observed increase in wealth accumulation, the 401(k) program generates a net increase of $3.60 in private saving per dollar of foregone revenue.

Conclusions

H/S find, after a review of the literature on saving incentives, that even using conservative assumptions about the extent to which contributions to saving incentives represent new saving, saving incentives generate substantial net capital accumulation over time per dollar of foregone revenue.

Notes

1. R. Glenn Hubbard and Jonathan S. Skinner, "Assessing the Effectiveness of Saving Incentives," unpublished, December 21, 1995.

References

Gale, William G., and John Karl Scholz. 1994. IRAs and Household Saving. American Economic Review 84 (December):1233-1260.

Hubbard, R. Glenn and Jonathan S. Skinner. 1995. Assessing the Effectiveness of Saving Incentives. Unpublished.

Joines, Douglas H., and James G. Manegold. 1995 (February). IRAs and Saving: Evidence from a Panel of Taxpayers. Mimeograph. University of Southern California.

Poterba, James M., Steven F. Venti, and David A. Wise. 1994. 401(k) Plans and Tax-Deferred Saving. In David A. Wise, ed., Studies in the Economics of Aging. Chicago: University of Chicago Press.

• The impact of carbon dioxide reduction policies on lifestyles and real wages; and either
• The use of risk prioritization for setting regulatory priorities; or
• The effect of increasingly stringent reporting of toxic material use on U.S. investment, economic growth, and competitiveness.

An additional climate change topic, yet to be determined, will also be included.

Tax Policy for the 21st Century. This day-long symposium, which will take place at the National Press Club on December 5, 1996, will feature provocative new research by renowned experts. Presentations will include:

• An examination of the impact of social security privatization on U.S. economic growth;
• An international comparison of the cost of capital as affected by tax policy;
• An assessment of the role of investment in plants and equipment in increasing productivity, real wages, and economic growth; and
• An analysis of how a consumption tax would affect saving, investment, and income distribution.

To receive further information or to register for these symposia, call the ACCF at 202/293-5811.

Papers presented will be available from the Center in book form in mid-1997.