A recent analysis from the Penn Wharton Budget Model (PWBM) claims that expanding Social Security benefits along the lines of Rep. John Larson’s (D-Conn.) Social Security Act of 2100 (“the Act”) would slow economic growth. The model warrants a closer look, not just because it casts doubt on Social Security expansion, but because some of its dubious assumptions can be used against almost any policy that raises progressive taxes to pay for programs tilted in favor of low- and moderate-income Americans.

The Act, which has over 200 cosponsors, would increase payroll tax revenues to pay for expanded benefits while eliminating or greatly reducing Social Security’s long-term deficit. Among other things, the act would subject earnings above $400,000 to the Social Security payroll tax (earnings above $132,900 are not currently taxed); gradually raise the payroll tax rate; increase benefits in a progressive fashion;1 and change the consumer price index used for the cost-of-living adjustment to better match the higher inflation faced by seniors.

The PWBM projects that the Act would reduce GDP by 2 percent in 2049. According to the analysis:

“The reason for the poorer economic performance [relative to alternative reforms suggested by the PWBM that combine revenue increases with benefit cuts] is that the Act does not reduce benefits and, in fact, increases benefits by 0.61 percent of future taxable payroll by 2049. Taxes that distort economic activity are then used to reduce the actuarial balance over time, including these new benefits. The Act, therefore, actually decreases the need for higher-income households to save more for their own retirement, whereas combined reforms generally increase the need.”

In other words, increasing payroll taxes and expanding benefits reduces after-tax pay and the need for private saving, shrinking the labor supply and the funds available for investment. Since the model ignores the positive effect of progressive redistribution on aggregate demand, these negative supply-side effects slow economic growth.

A positive impact

Other economists, however, notably former Council of Economic Advisors Chair Jason Furman and former Treasury Secretary Lawrence H. Summers, view the Act as having a positive impact on economic growth, despite the need to raise taxes:

“Social Security benefits should in fact be raised for the majority of recipients. Despite the fact that Social Security is an efficient, defined benefit program, its benefits are not generous by global standards and leave too many seniors in or close to poverty. Expanding benefits only works, however, if people are willing to pay more during their working years—meaning that revenue, and not just from high-income households, has to be an important part of the solution, as in the plan advanced by Congressman John Larson. The advantage of this approach is that it would expand aggregate demand, as the additional spending by the elderly would outweigh any reductions in spending associated with higher payroll taxes, in part because the plan could lead households to save less.”

In this Keynesian view, progressive redistribution increases aggregate demand and economic growth. Though Furman and Summers approve of increasing payroll tax revenues to pay for expanded benefits, they view the tax increase as dampening the expansionary impact of redistribution from working-age households to older beneficiaries who tend to spend the money more quickly. Though Furman and Summers don’t elaborate on their reasoning, the dampening effect doesn’t primarily stem from labor market distortions, but more likely from fiscal contraction (since eliminating Social Security’s projected shortfall shrinks the unified budget deficit).

Though the negative impact of taxes on work effort looms large in the PWBM, taxes have an ambiguous effect on work effort in standard microeconomic models. Payroll and income taxes make work less appealing by reducing the opportunity cost of leisure (the “substitution effect”), but also reduce take-home pay and therefore cause the demand for leisure and other normal goods to fall (the “income effect”). While most economists estimate that the substitution effect is dominant, the net effect on the labor supply may be quite small because of the offsetting income effect.

The PWBM, however, assumes that labor supply is very sensitive to tax increases. Thus, while a Congressional Budget Office (CBO) review of the literature found that the Frisch labor supply elasticity, which measures the responsiveness of labor supply to tax increases, ranged from 0.27 to 0.53, the PWBM assumes a Frisch elasticity of 0.5, close to CBO’s upper-bound estimate.

This appears to be a pattern with the PWBM, where important parameters fall within the range of respectable opinion but are tilted in the direction of finding negative effects of tax increases and budget deficits. In addition to a high labor supply elasticity, the PWBM assumes that the saving rate is highly sensitive to after-tax investment returns, and that investment, in turn, is highly dependent on domestic saving despite international capital flows. With assumptions that exaggerate the negative effects of taxes and deficits, spending cuts are left as the surefire way to boost economic growth, in line with the apparent preferences of some of PWBM’s funders, such as former Microsoft CEO Steve Ballmer.

Generally speaking, policies that increase saving will, all else equal, tend to boost growth in supply-side models, while those that tend to reduce saving—including expanding Social Security benefits—will have the opposite effect. Social Security reduces private saving because people would save more in its absence, though not enough to offset the loss of benefits. Social Security itself functions mostly as a pay-as-you-go program, with payroll taxes from workers going directly to retirees and other beneficiaries.2 The net effect of expanding a program that has little direct effect on government saving but indirectly reduces private saving is a reduction in national saving. However, since the Act raises taxes not only to expand benefits but also to close the projected shortfall, it is not at all clear that it reduces national saving relative to CBO baseline projections.

In addition, since wealthy people tend to save a greater share of their incomes and savings is assumed to drive investment in supply-side models, upward redistribution generally translates into greater investment and faster growth in supply-side models. The PWBM, for example, projects faster growth in the near term from the Trump tax cuts due to increased private saving and investment.3 In Keynesian models, on the other hand, regressive redistribution tends to decrease aggregate demand because high-income households tend to save more. Conversely, progressive redistribution, as in the Act, boosts economic growth except when the economy is already operating at full employment, which is likely to be the exception, not the rule.

Whether you believe the PWBM analysis or Furman and Summers depends on whether you think economic growth is, as a rule, demand- or supply-constrained. Many economists split the difference, viewing Keynesian models as “short-term” and supply-side models as “long-term.” This gives the latter, which are often based on theory rather than grounded in reality, more weight in policy discussions than they deserve, especially since the full employment assumption is rarely spelled out.4

Whereas Keynesians often take into account some supply-side effects, the reverse is generally not true of supply-side models, which preemptively rule out demand-side effects on employment and economic growth by assuming that the economy is operating at capacity in the long-run. In these models, the growth rate depends only on the size and productivity of the workforce, the level of investment, the return on capital, and perhaps technological change. Thus, the only effect on employment derives from more people choosing to enter the labor force or work more hours, often in response to tax cuts. But as John Maynard Keynes famously quipped about assuming that unemployment and underemployment can only be temporary deviations from the full-employment norm, “In the long run, we’re all dead.”

In recent years, many economists have pushed back against the idea that boosting aggregate demand only affects economic growth after temporary cyclical downturns. Summers in particular has revived the theory of “secular stagnation,” whereby mature economies can operate below capacity indefinitely (or, see this productive capacity itself eroded) due to a chronic shortfall of aggregate demand stemming from households’ increasing propensity to save and firms’ decreasing propensity to invest, a combination that can drive real interest rates toward zero and prevent the economy from self-correcting by generating ever-lower interest rates. Other economists, including former Fed Chairman Ben Bernanke, UC Berkeley economist Brad DeLong, Nobel Prize winner Paul Krugman and former International Monetary Fund chief economist Olivier Blanchard, have espoused similar “secular stagnation,” “savings glut,” or “liquidity trap” arguments that call into question whether what the United States needs right now is greater savings. Even if one thought that advanced country growth has been supply-constrained in the past, recent developments (particularly the rise of inequality that has transferred income from high-spending to low-spending households) may well have pushed these advanced economies into becoming demand-constrained.

In a recent paper, Summers and Bank of England economist Łukasz Rachel estimate how expansionary social insurance programs have helped to prop up the global economy. Summers and Rachel estimate that expanding programs such as Social Security and Medicare pushed the “neutral” (a.k.a. “equilibrium” or “natural”) real interest rate up by 3.2 percentage points over the past 40 years in the U.S. and other advanced economies. Since Rachel, Summers and others estimate that the real neutral rate—the highest rate consistent with maintaining full employment—is now barely above zero, this suggests that past these expansions, rather than being a drag on growth, have buoyed what would otherwise be an economy requiring large budget deficits to stay afloat.

To recap, both Keynesian and supply-side models are in agreement that the Act would have the intended effect of increasing retirement income, smoothing lifetime consumption, and redistributing from working-age and higher-income households to retirement-age and lower-income households. They also agree that expanding benefits will tend to reduce private saving, though supply-side models such as the PWBM emphasize this aspect more (and the effect of the Act on national saving is ambiguous because it also closes Social Security’s projected shortfall). However, the two types of models differ as to whether more saving or more spending is needed to boost economic growth.

If an otherwise useful policy slows growth, this is not by itself reason to dismiss it out of hand. Growth and distribution both matter if the goal is improving the lives of the greatest number of people, and Social Security is modestly redistributive toward people with lower incomes.5 Moreover, policies that have other benefits—in the case of Social Security, smoothing consumption over the life-cycle and insuring against death and disability—can improve wellbeing regardless of their effects on distribution or economic growth.

In this case, however, there is reason to doubt that the benefits of Social Security expansion come at the expense of economic growth. Rather, Social Security expansion is a win-win policy, except maybe for wealthy people trying to avoid paying payroll taxes on earnings above the current cap.

Note: The description of the Rachel and Summers paper contained an error that was corrected on October 18, 2019.

End Notes

1. The Act would increase the lowest benefit multiplier from 90 percent to 93 percent. The higher multiplier would apply to average indexed monthly earnings (AIME) up to $926. Therefore, all workers with AIMEs equal to or higher than $926 would see a $28 per month increase in benefits (those with lower earnings would see a smaller dollar increase but larger percentage increase than those with earnings above that threshold).

2. Social Security is an off-budget program with dedicated funding and a mostly pay-as-you-go structure, in contrast to advance-funded employer and individual retirement plans such as 401(k)s, IRAs and traditional pensions. However, Social Security can have a modest effect on the unified budget deficit, which includes off-budget programs, as funds are contributed to and later withdrawn from the Social Security trust fund to accommodate demographic bulges such as the Baby Boomers. The Social Security trust fund reduced the unified budget deficit when the Boomers were working and the trust fund was growing, and will increase the deficit as the Boomers retire and the trust fund shrinks. However, in contrast to other government programs, Social Security cannot increase the federal debt over the long run because funds can only be withdrawn that were saved in the first place.

3. In the out-years, the model projects that growth may be slower than it would have been without the Tax Cuts and Jobs Act due to higher interest rates caused by a rising debt level.

4. For example, advocates of the U.S.—Mexico–Canada Trade Agreement, the Trans-Pacific Partnership and other trade deals have touted supply-side employment gains from models that, by design, ignore potential demand-side job losses.

5. Social Security retirement and spousal benefits replace a larger share of pre-retirement earnings for low earners. Low-income workers and dependents are also more likely to receive disability and survivor benefits. However, this progressivity has been partially eroded by growing gaps in life expectancy between high- and low-income beneficiaries.