Do Americans Really Favor a Revenue-Only Approach to Social Security Reform?

The National Academy of Social Insurance (NASI) released a survey last week (see video of CRFB President Maya MacGuineas at the release event) showing that a majority of Americans are willing to raise taxes – including their own – to preserve Social Security. Given the program's funding gap, it is an encouraging sign that Americans are willing to pay higher taxes to improve Social Security's solvency. Trade-off analysis, which NASI used, is an excellent way to get Americans to consider complex issues with pros and cons to all sides. Unfortunately, the design of the survey prevents the answers from being as useful as they could have been.

NASI's survey set out to determine how Americans would prefer to improve Social Security. Most respondents favored raising taxes (both the payroll tax rate and the taxable maximum) and increasing benefits. Respondents opposed raising the retirement age or means-testing benefits. This result differs from other similar surveys that found support for balanced packages of tax increases and decreased benefits for high earners.

Although it is impossible to know for sure, this survey's divergent results may have been driven by its framing of certain choices.

Most significantly, the results were likely influenced by the lack of sufficient benefit options to improve solvency, the absence of one of the most prominent options, and the language used to describe various changes. For example:

• Respondents did not have the opportunity to choose a similar amount of benefit and revenue changes. The largest potential revenue package would close 164 percent of Social Security's 75-year shortfall, whereas the largest potential benefit-side package could close only 65 percent, and the largest package of benefit increases would increase the shortfall by 55 percent. A person who wanted to restore solvency with benefit changes would therefore be forced to choose at least one revenue option, while a person who wanted to restore solvency with revenue changes could also substantially increase benefits.

Source: NASI
* Note: This total is more than the sum of its parts due to interaction effects.

• The options excluded the most prominent way to slow the growth of benefits: changing the initial benefit formula. Direct changes to the so-called PIA factor, or indirect changes through progressive indexing, are absent even though they have been a mainstay of Social Security reform plans. These changes are in nearly every solvency plan that slows benefit growth, including plans from Simpson-Bowles and President Bush, which rely on it for the plurality of their savings. Indeed, this option is so prominent that the Social Security Chief Actuary has modeled more than 20 ways to modify the PIA factors. Many of these changes would affect only benefits for high-earners, which might have been attractive to survey-takers who prefer a more progressive system.
• The means-testing option is a straw man policy that no one supports. Specifically, NASI describes a phase-out of benefits for high earners (already a relatively obscure policy proposal) as "requir[ing] people to provide proof of eligibility, based on their income, in order to receive Social Security benefits," making it sound like welfare. We are not aware of a person or organization that supports such an onerous method of means-testing, and there is no reason to think it would be administered in this way.
• The survey provides context for an increase in the Cost-of-Living Adjustment (COLA) but not for a decrease. It describes adopting the CPI-E as "basing it specifically on the inflation experienced by older people, who spend more on medical costs, which generally rise faster than other inflation" despite that fact that the Congressional Budget Office (CBO) questions whether seniors' cost-of-living is actually any higher. But when discussing the chained CPI, the survey describes the option as "a new measure of inflation that generally rises more slowly," without mentioning that many economists believe it is a more accurate measure of inflation.
• The survey minimizes the perceived impact of a two point payroll tax increase phased in over 20 years by describing it in small units: the weekly incremental employee-side effect of increasing taxes, or "50 cents per week each year, matched by the employer," when a fully phased-in policy would raise taxes by $1,000 per year. This same method -- describing the weekly impact of the incremental year-to-year change -- is not used for other policies. For example, it describes adopting the chained CPI, which has a similar incremental effect, as a benefit cut of "6.5% by the time a retiree reaches age 85" and a three-year retirement age increase as "an additional benefit cut of about 21%."

CRFB has often written about Social Security's financial shortfall (see our most recent paper or blogs), arguing that lawmakers should address it long before it causes a 23 percent benefit cut in 2033. Lawmakers will need to either raise revenue, reduce spending, or some combination. We tend to believe that packages with both revenue and spending changes are the most politically feasible. Our Social Security Reformer lets you design your own plan to restore Social Security solvency. Regardless of the plan lawmakers choose, they should act sooner rather than later, which allows changes to be smaller and more gradual.

Surveys, particularly this type of trade-off analysis, can illuminate the changes that Americans are willing to accept to restore solvency to Social Security and prevent a future cut in benefits if nothing is done by 2033. However, the survey's design prevents it from being as informative about Americans' preferences as it could have been. We're happy to work with NASI to improve this survey in future years: everybody benefits from getting the most accurate information possible.

table.alternate { border:1px solid black; border-collapse:separate; border-spacing:1px; } table.alternate tr:nth-child(odd) td{ background-color:rgb(227,235,244); } table.alternate tr:nth-child(even) td{ background-color:white; } table.alternate tr:last-child { border-top-style:double; border-top-color:black; } table.alternate a {text-decoration:none;}