Monday, November 8, 2010

Early retirement accounts as a fix for Social Security?

National Review's Reihan Salam has some nice things to say about my recent Enterprise Blog post on Social Security personal accounts, which shows that raw comparisons of the benefits paid by accounts to those paid by Social Security can be misleading. Given that, deciding whether to fight for personal accounts as part of Social Security reform—versus, say, focusing on holding the line on taxes while encouraging more outside retirement saving—is a choice conservatives need to wrestle with. I appreciate Reihan's comments (and return the favor—loved your book!).

Reihan then raises the suggestion of "early retirement accounts," an option I hadn't considered in a while. This idea, proposed a few years back by Phil Longman of the New America Foundation, is similar to ordinary personal account plans, but with a twist. In most account plans, the account would pay for part of your total Social Security benefit throughout your retirement. For instance, you might retire at 67 and each year thereafter half your total benefit would come from your account while the other half would come from the traditional program. The twist with early retirement accounts is that the account would pay your full benefit for the first few years of retirement—say, from 67 through 72—and then you'd receive your full benefit from the current program.

The advantage of this approach is that it would tend to encourage later retirement, since people with defined contribution accounts generally work longer than those with traditional defined benefit pensions. It would also ensure a safety net of government-provided benefits later in life, addressing a concern raised by opponents of personal accounts.

The problem with the idea, I think, is that the numbers don't really add up. According to Longman, "The Social Security Administration's Office of Policy estimates that the savings from this extension in the retirement age would not only cover the full cost of the early retirement accounts, but would also make the system solvent in the long term." I was at the SSA at the time and remember the plan (one person inside the agency was a strong supporter of it) but I don't think those estimates are right. And the fact that they're not right shows how difficult it is to solve the Social Security problem.

Using the Policy Simulation Group's Social Security models, I simulated a shift in the normal retirement age to 72 and the early retirement age to 68, taking full effect around 40 years from now. The idea is that individuals who held early retirement accounts their full lives would be subject to the full retirement age increase, while for older folks it would be phased in. Without including the accounts, the plan fixes around three-quarters of Social Security's long-term deficit. If you include the accounts, then most likely the plan would increase the deficit relative to current law.

Now, you can always tweak the parameters to make things fit, so the fact that the numbers don't add up isn't a big deal by itself. But it's the fact that you'd have to really tweak the parameters—say, shift the retirement age to 80 or so—that shows the scale of the Social Security shortfall. It might be smaller than Medicare, but it's bigger than pretty much any other problem the government needs to fix. I don't think early retirement accounts would be politically salable post-tweaking.

An alternate solution would be to build more savings outside of Social Security, say, by automatically enrolling all Americans in 401(k) plans. If everyone saved enough to cover their needs for a few extra years of early retirement, we could shift the retirement age back without making enormous cuts in monthly benefits. My recent AEI paper on increasing the early retirement age of 62 hits on some similar points.

All that said, both Reihan and Phil Longman deserve credit for thinking outside the box on Social Security reform. That's probably the only way we're going to get the program fixed.

5 comments:

Thomas said...

Your article on increasing the Social Security EEA is a valuable contribution to a difficult policy issue. I do have some questions, though:

First, you seem to make some significant assumptions regarding how people will react to raising the EEA to 65. As you would be the first to acknowledge, raising the EEA to 65 doesn't mean that people cannot continue to retire at 62 and collect social security later. What percentage of future retirees did you assume would change behavior by delaying retirement as a result of raising the EEA?

You wrote:

"Social Security's finances would be largely unchanged by an increase in the EEA. The long-term deficit, which is 1.92 percent of taxable payroll under the 2010 Social Security Trustees Report, would rise to about 1.93 percent of payrollThe life of the Social Security trust fund would be extended by around five years, from 2037 to 2042."

I have a hard time reconciling these two statements. It appears that the SS finances would actually worsen as a percentage of payroll, but solvency would be extended by five years. How can both be true? Is the five year extension due solely to the fact that many would delay retirement in 2037 thus pushing back solvency at that time (by three years)??

You also wrote:

"...the Social Security payroll tax should be reduced or eliminated for individuals over age 62, giving older Americans the incentive to work and employers the incentive to hire them".

It appears the effect of this suggestion is not included in your calculations above regarding solvency. Wouldn't this suggestion, if implemented, reduce solvency of SS (but increase general revenues)?

Andrew G. Biggs said...

In the article itself, I don't myself make any assumptions regarding how people will react to a higher EEA. The numbers I get are derived from a model, which obviously itself depends on assumptions, but the point here is simply that they're not mine (so any bias in the numbers isn't mine). The model does project that most people who are working as of age 62 will continue to work at least part of the way to the new EEA, while others who were already out of the workforce would remain out and still others would go on disability.

The two statements regarding social security's financing actually are compatible because they don't really measure the same thing. Raising the EEA implies higher tax revenues in the short term, since people will work longer, but higher benefits in the longer term. So the near-term increase in revenues can slightly prolong the life of the trust fund while the overall effect would be to slightly worsen system financing. The long-term effect on the actuarial balance is the better number, but I included the trust fund exhaustion date just because so many people reference it.

You're right that I didn't include the effects of lowering the payroll tax rate for older workers, though I could do so depending on when/how much I decided to lower it. This would reduce solvency, which would have to be made up as part of a larger solvency package of taxes, benefits, etc.

Thomas said...

Thanks for your response.

As to the first issue, I had thought the assumptions were built into a model. As with all models, the outcome is only is as good as the model. It's one thing for a model to perform merely arithmetic functions, but here we're talking about predicting human behavior. As we've seen with respect to other models, such as the multipliers used for government spending and taxing, these models seem to be pretty useless. That's not your fault, of course, but should we be putting such heavy emphasis on this type of social security fix when the outcome is not very predictable? At the very least, I would want to see what assumptions the model is using.

Thanks for your explanation on the second point, which I anticipated as well. Personally, I think you should highlight the second point regarding the slight long-term fiscal decline of the social security trust fund over the short-term effect. We've had too many "quick fixes" to our financial system already.

That said, I do acknowledge that your proposal might benefit the overall financial condition of the country by increasing overall general revenues, but again that is subject to the big caveat that a significant number of persons would actually work longer if the EEA were increased. I also worry about the effect of people working longer on the job opportunities for younger workers, an issue that deserves more attention. It seems to me that the main benefit of persons working longer is that they would have higher retirement benefits (public and private) that would support them in fewer years of retirement than the current system.

Thanks again for your thoughtful response.

Thomas said...

Thanks for your response.

As to the first issue, I had thought the assumptions were built into a model. As with all models, the outcome is only is as good as the model. It's one thing for a model to perform merely arithmetic functions, but here we're talking about predicting human behavior. As we've seen with respect to other models, such as the multipliers used for government spending and taxing, these models seem to be pretty useless. That's not your fault, of course, but should we be putting such heavy emphasis on this type of social security fix when the outcome is not very predictable? At the very least, I would want to see what assumptions the model is using.

Thanks for your explanation on the second point, which I anticipated as well. Personally, I think you should highlight the second point regarding the slight long-term fiscal decline of the social security trust fund over the short-term effect. We've had too many "quick fixes" to our financial system already.

That said, I do acknowledge that your proposal might benefit the overall financial condition of the country by increasing overall general revenues, but again that is subject to the big caveat that a significant number of persons would actually work longer if the EEA were increased. I also worry about the effect of people working longer on the job opportunities for younger workers, an issue that deserves more attention. It seems to me that the main benefit of persons working longer is that they would have higher retirement benefits (public and private) that would support them in fewer years of retirement than the current system.

Thanks again for your thoughtful response.

Andrew G. Biggs said...

The outcomes from the model I've used seem to be consistent with those from SSA and CBO; this doesn't mean it's right -- they all could be wrong -- but it does seem to support the results.

I think the conclusion that people would work longer with a higher EEA is supported by the fact that a lower EEA seems to have encouraged people to retire earlier. If the EEA weren't a factor then you wouldn't see those results going in either direction. I'm not saying it's the only factor, but I think most people agree it is one. But you're right that this isn't exact science; because Social Security is a national program there aren't many differences between states that we could exploit to get a better feel for the effects of different policies.