How to Reform Social Security, Part II

The last time I wrote for an academic audience about Social Security reform was this paper, which I presented in Sweden two years ago at a conference sponsored by the Foundation for International Studies on Social Security. In that paper, I argued that the most sensible way to restore solvency was by raising the age at which the system pays out full benefits, also called the normal retirement age, or NRA. (The discussion below is similar to the section on "Moving Forward," which starts on page 9 of that paper.)

Arnold Kling recently wrote about the need to reform Social Security on his blog and at TechCentralStation and concluded that, "The retirement age is the most logical and effective policy lever." I think he is exactly right, primarily because it squarely addresses one of the two main reasons why the system is projected to run into cash flow problems in the near future.

The dramatic increase in Social Security's cost rate over the next several decades is mainly due to the increase in the number of beneficiaries relative to covered workers. In 2004, there are 30 beneficiaries per 100 workers. By 2080, there will be 54, an increase of about 80 percent (as shown in this table). Over the same period, the program's cost rate--the ratio of benefits paid to taxable payroll--will increase from 11.07 to 19.39 percent, an increase of about 75 percent (as shown in this table). That the increases are so close shows that the projected deterioration in Social Security's finances is almost entirely the product of demographic shifts.

There are two reasons why there will be about 80 percent more beneficiaries relative to workers in the future: lower fertility rates and lower mortality rates (see this table for the history and this one for the projections). There is probably no easy way to link a reform of the system to changes in fertility rates, but it seems straightforward to base reform on changes in mortality rates.

Failing to index the retirement age to life expectancy implies an increasing portion of adulthood spent collecting benefits. Policy makers could craft a sensible justification for raising these ages. Consider how much easier it is to explain why retirement ages have to increase than it is to make sense of, say, an effective switch from wage indexing to price indexing in the calculation of the benefits (the centerpiece of Commission Model 2). Other reforms could be introduced to compensate for the impact of lower fertility rates on the system's finances, and I would not be against the ones that work by reducing future benefits rather than raising future payroll taxes. I would prefer to just stick with one policy lever to keep things simple.

A rough calculation suggests that the normal retirement age would have to increase to 73 by 2080 in order to restore solvency using only this policy lever. Though it seems to be a large change, it can be phased-in at a rate of about a year of age per decade (if we start now). When fully phased in, a worker who wished to retire at the currently legislated normal retirement age of 67 would face actuarial reductions of about 40 percent of benefits (e.g., the current early retirement reduction factor of 6 2/3 percent per year for 6 years). It would be perfectly reasonable to help people retire earlier than the NRA by facilitating a system of personal accounts, and, as I noted earlier, if any new money is coming into the system, it should be done through personal accounts.

Raising the NRA by this much would create a few complications, which could be easily addressed. Here is a list of the ones I think are most important:

1. The Early Eligibility Age. Workers can first claim benefits at the early eligiblity age (EEA), which is currently 62 and not scheduled to increase. Benefits are reduced at an actuarially fair rate for each year that benefits are taken before the NRA. When the NRA is 67, the total reduction for those 5 years will be 30 percent of benefits. If the NRA goes up, then the EEA should also go up. Otherwise, some people who live for a long time after claiming benefits at age 62 would be at risk for poverty later on in old age.

2. Pressures on the Disability Insurance Program. Workers who become disabled before their normal retirement age are eligible for disability benefits. These benefits are more generous than benefits taken early in the absence of disability. With higher ages for full and early benefits, there is more of an opportunity (and possibly a greater incentive) for workers to claim disability. This will increase program costs and can be addressed by having the normal retirement age increase a bit faster than a year per decade, if necessary.

3. The Bigger Hit to Low-Earners. People with higher incomes over their lifetime tend to live longer. Raising the Normal Retirement Age therefore disadvantages lower income workers more than higher income workers. This impact can be offset by changes in the benefit formula to give higher replacement rates for lower earnings, paid for by lower replacement rates at higher earnings levels.

With the retirement ages linked to projected improvements in life expectancy, personal accounts can serve two purposes. First, their accumulations can facilitate retirement at ages earlier than the legislated early and normal retirement ages. There is nothing wrong with people retiring when they choose--they just should not do so at the public's expense or at the risk of their own poverty later in old age. Second, when confronted with the magnitude of the increase in the NRA required to restore solvency, policy makers may decide that they prefer to bring new revenues into the system rather than rely solely on higher retirement ages. Investing those revenues in personal accounts rather than the Trust Fund helps to ensure that the additional revenues are actually saved for retirement rather than becoming a source of funds for the government to finance the rest of its expenditures.

The reason I like using higher retirement ages to restore solvency is that it simply brings our definition of "Old-Age" in Old-Age, Survivors, and Disability Insurance in line with contemporary (and evolving) notions of life expectancy. It gets Social Security to live within its means and allows the system to continue alleviating poverty among the elderly.