Monday, December 15, 2008

Issues with Ghilarducci GRA proposal: The cost of guaranteed returns

As noted here, Teresa Ghilarducci's Guaranteed Retirement Account plan would guarantee contributors a return of 3 percent above inflation. This sounds attractive, but is in fact a potentially very expensive proposition. Here's why.

The Social Security Trustees project that the real return on non-inflation protected government bonds will be 2.9%; inflation-indexed bonds generally return around 0.5% less than nominal bonds since they're safer. So the GRA plan is promising returns higher than could be gained on similar market-based investments, which implies a subsidy.

Moreover, returns on GRA accounts would be paid out from a fund invested in risky assets that would be managed by Social Security. Supposedly riskless returns backed by risky assets is a recipe for trouble – think about Fannie/Freddie, DB pensions and the Pension Benefit Guarantee Corporations, etc. I wrote about the costs of guaranteeing Social Security personal accounts here, and the underlying logic is the same here.

A guarantee of a given rate of return is like a "put option," which allows you to sell an asset at a given price even if the market price of the asset has declined. You can put a price on this put option using the Black-Scholes formula. For instance, let's assume that you invest $1000 in a GRA account and have 20 years until retirement. At a 3% real return, that means the GRAs must repay you $1806. Let's also assume that the riskless interest rate is 2.5% and that investment fund backing the GRAs holds half stocks, half bonds, and has a volatility (meaning standard deviation of annual returns) of 12%. Given this, the cost of guaranteeing a 3% real annual return on a $1000 contribution is $272. (Here's a spreadsheet if you want to play with the numbers.)

In other words, there's an implicit subsidy of around 27% on each GRA contribution. Given that total annual contributions could top $350 billion, a 27% implicit subsidy is a pretty big deal if it's not budgeted for. Probably the biggest lesson we should draw from the current financial crisis is to be very careful about guaranteeing market investments against market risk.

7 comments:

Anonymous said...

Good analysis.

I'm not fond of this idea (having done just fine with my 401k, thank you very much). I believe that Ghilarducci claims that the funding for this scheme could come from current tax breaks (which are deferrals, NOT exemptions, by the way).

What's not explicitly stated is that the 27% subsidy you describe would have to come from (somebody's) taxes, to the tune of an additional 1.35% (combined employee and employer "contributions).

Of course, another way to address the shortfall is for the government to define a lower "official" inflation rate than the "real" inflation rate, as is current Social Security practice (ref. to http://www.shadowstats.com/alternate_data).

Try getting by on 75% (GRA + SS) of your pre-retirement income with growing medical costs more than supplanting your work-related expenses (and subsidized medical coverage).

I for one am for tax simplification . Get rid of "special" taxes and lump them all together, and then dole them out as social engineering dictates.

As for Social Security fund "managment", would this be different than the current Ponzi scheme for social security?

"I'm from the Govermnent, and I'm here to help you".... (and get re-elected).

Anonymous said...
This comment has been removed by a blog administrator.
Anonymous said...

Andrew,
The Trustees Report table to which
you point readers, "Table V.B2:
Additional Economic Factors" states
clearly that the interest rate
assumptions are 2.9% real and
5.7% nominal. So, your starting
premise that "the long-term return
on nominal (meaning, non-inflation protected) government bonds will be 2.9%" is incorrect. There may
be problems with the GRA proposal, but it is not fair to start your
critique with false assumptions.

Andrew G. Biggs said...

Anon,

You caught an error by me, but it's not really the error you think. What I should have said is, the projected real return on nominal bonds is 2.9%, which means that the bonds are denominated in nominal terms (meaning there's no inflation adjustment) but the expected real return is 2.9%. I wasn't meaning to say that the expected nominal return was 2.9%. The point was that if the expected real return on a riskier bond is 2.9%, then the expected real return on a CPI-indexed bond should be below 2.9%. I think that point still holds. But thanks for pointing it out -- I'll fix in the text.

Andrew G. Biggs said...

M.C.,
One thing that makes this worse is that the 27% subsidy isn't paid in each year; it's paid as needed, and it would be needed more in bad times (like today) than in good times. It's likely that equity returns in the medium term will be higher than normal, since prices have dropped so much in the past year. So right now, offering a guarantee like that isn't likely to be too expensive. But in the next downturn the costs could really kick in.

WilliamLarsen said...

Another proposal to save Social Security or to transition to private accounts. I hate to break the bad news to you, but the cost of saving SS is the exact identical cost as transitioning to private accounts which is the identical cost of keeping Social Security the way it is. Every plan and I mean every plan out there assumes you can create something from nothing.

Take this current plan that proposes garenteeing 3% above inflation. First off, who is going to pay this? Second how does this change the current situation where Social Security has continued to earn very close to this amount and still cannot pay full benefits in the future.

Raising the payroll tax from 10.6% to 12.5% could eliminate the shortfall between revenues and expenses over the next 75 years. However, in year 2079 Social Security would face the age-old problem it now faces in 2041. How will it pay promised benefits? The tax rate needed then would be 19% and the trust fund will have been exhausted. This just puts off the inevitable.

“...the 75-year time horizon is arbitrary since it ignores what happens to system finances in years outside the valuation period. For example, we could eliminate the actuarial deficit by immediately raising the payroll tax by 1.86 percent of payroll. However, as we move one year into the future, the valuation window is shifted by one year, and we will find ourselves in an actuarial deficit once more. This deficit would continue to worsen as we put our near term surplus years behind us and add large deficit years into the valuation window. This is sometimes called the "cliff effect" because the measure can hide the fact that in year 76, system finances immediately "fall off the cliff" into large and ongoing deficits."[1]

[1] Strengthening Social Security and Creating Personal Wealth For All Americans, Final Report December 21, 2001, Page 70,
http://www.csss.gov/reports/Final_report.pdf

Then there is the problem with investing outside of Treasury Notes. If SS stops investing in U.S. Treasuries will anyone else do it? If the demand drops for U.S. Treasuries, does the cost of borrowing go up requiring ever more in Federal Income Taxes, which offsets any net gain you made on these other investments?

One more thing, if you are making 3% above inflation of 2%, your real compounded rate of return is not 3%, the effective rate of return is 2.94118%. Subtracting different exponential rates of growth from each other cannot be done linearly.

All these plans fall short because there is no new capital identified or they do not include the new capital costs when comparing different proposals to the current problem.

If SS is earning such a bad return now, why would anyone want to contribute another penny towards the same lousy investment. You cut your losses, you do not add to them.

Anonymous said...
This comment has been removed by a blog administrator.