When 3% May Be a Bum Number

Times Staff Writer

Three percent is a big number in President Bush’s plan for private Social Security investment accounts.

Too big, say some people on both sides of the privatization debate.

The administration’s long-term assumption is that money paid into the Social Security trust fund will earn an average of 3% a year, after inflation, on the Treasury bonds the fund owns.

That return also is the “neutral” or “hurdle” rate -- the return investors would have to beat in their private accounts to end up better off than if they simply stuck with regular Social Security benefits, according to the administration’s proposal.


The problem is that the current after-inflation, or real, annualized return on Treasury bonds is well below 3%. And some people who spend considerable time thinking about markets say it would be more realistic to assume a return below 3% in the longer term as well.

“Three percent is way too high,” said Jeremy Siegel, a finance professor at the University of Pennsylvania and the author of “Stocks for the Long Run,” the now-famous 1990s book on investment return expectations.

Bill Gross, one of the world’s top bond market authorities and manager of the Newport Beach-based Pimco Total Return fund, says that to earn a 3% real return on a government bond today, “you would have to invest in Mexico or Russia to get that -- not in the United States.”

So what if real Treasury returns continue to fall short of 3%? If that’s also the hurdle rate for private Social Security accounts, many Americans might be unwilling to take a chance on them, because poor Treasury returns might raise doubts about future returns on stocks and other assets as well.

If you fear that good investment returns in general will be harder to come by in the next decade or two, you might figure it’s best to just stick with traditional Social Security benefits, whatever they turn out to be.

Under Bush’s plan, workers would be allowed to divert part of their Social Security payroll taxes to private accounts that could invest in broadly diversified stock or bond portfolios.


The trade-off is that workers who opted for private accounts would agree to accept less in traditional Social Security benefits in retirement.

Three percent would be the dividing line: If you earned a real annualized return higher than 3% on your private account, that money plus your reduced traditional benefit would total more than what you’d have if you stayed solely with the traditional benefit program.

If you earned less than a 3% real return on your private account, your total combined retirement benefits would be less than under the traditional program. So private accounts would be a gamble: Fall short of that 3% hurdle, and you lose.

A 3% real return may not seem like much of a hurdle, if you’re looking back over the last 25 years.

From 1980 through 2004, long-term government bonds produced an average real return of 6.7% a year, according to data tracker Ibbotson Associates in Chicago.

Stocks did better than that: The blue-chip Standard & Poor’s 500 index scored an average real return of 9.5% a year in that period.


But as everyone knows, the 1980s and 1990s were spectacular decades for financial assets. The stock market soared as the economy boomed. And even as inflation declined, long-term interest rates were extraordinarily high for much of the period, providing a hefty payoff for bond owners.

Measured over longer periods, however, returns haven’t been nearly as generous on government bonds. From 1946 through last year, the average annualized real return on long-term Treasuries was slightly less than 1.6%, according to Ibbotson.

The current return also is well below the administration’s 3% long-term expectation, if you use the benchmark long-term government bond -- the 10-year Treasury note -- and subtract inflation as measured by the U.S. consumer price index.

The yield on the 10-year T-note was 4.27% as of Friday. The CPI was up 3% in the 12 months through January. That means the real annual return is about 1.27% at the moment (we’re simplifying here, to make the point).

Current government bond returns are low because long-term interest rates have come down dramatically since the mid- 1980s. And even though the Federal Reserve has been raising short-term rates since June, long-term bond yields have continued to fall for much of that period (although they have rebounded somewhat in recent weeks).

Fed Chairman Alan Greenspan referred to the relatively low yields on long-term bonds as a “conundrum” when asked about them during congressional testimony in mid-February.


The question is: Are there good reasons to believe that real returns on long-term government bonds could rise to at least the 3% level in the rest of this decade and beyond?

One way to get there would be for interest rates to rebound substantially, of course. But that would be bad for the economy and, most likely, bad for returns on competing investments like stocks held in private Social Security accounts or elsewhere.

Another way to boost the real return on bonds would be for inflation to fall sharply, while bond yields held steady or rose. But if inflation declined because of a weaker economy, that too might be a bad omen for the stock market.

The very long-term trend has been one of shrinking real returns on U.S. bonds.

By Professor Siegel’s calculations, long-term government bonds produced a 4.8% average annual real rate of return from 1802 to 1870, a 3.7% return from 1871 through 1925 and a 2.2% return from 1926 through 2001.

As noted above, the number shrank further in the 1946-2004 period, to about 1.6%.

Implicit in those declining real returns is that Treasury bonds have become much more popular over the decades. They just about had to become more popular with investors worldwide, given how much the U.S. has needed to borrow.

There is plenty of debate today about America’s ability to continue attracting foreign capital to fund our huge trade and budget deficits. If that money dries up, it’s conceivable that interest rates could rocket in response.


Absent the doomsday scenario, however, many experts see real long-term government bond returns staying close to where they are now.

“In every case we’re guessing, but if I were to guess where real rates would be 10 years from now I’d guess 2%, not 3%,” said John Shoven, an economics professor at Stanford University.

Shoven is a fan of private Social Security investment accounts and has written extensively on the idea. But he said the government’s assumption of a 3% real rate of return on Treasury bonds “is not realistic today.”

Why does the administration use that number?

White House spokeswoman Claire Buchan said the 3% figure “is the number the Social Security Administration actuary uses, and all of the estimates in the president’s proposal comport with” the actuary’s data.

Stephen Goss, the chief actuary at the Social Security Administration in Washington, said he considered 3% “a reasonable place to be” given that the real return on government bonds since 1980 has been more than twice that number.

“It’s our best judgment based on the moment,” Goss said. It wouldn’t make sense, he said, for the Social Security system’s assumptions to change too frequently, given that the system is supposed to take a long-term view of its income and outgo.


What if that 3% bond assumption in fact turns out to be way too high? Some supporters of private accounts say that if the Social Security trust fund can’t earn at least 3% on its Treasury IOUs, then the money won’t be there for the promised traditional benefits anyway.

In other words, if Treasury bond real returns remain relatively poor over time, future retirees who rely solely on the traditional system would be more likely to see their benefits reduced.

That would be another reason for investors to take a chance in private accounts, supporters say; it isn’t far-fetched, they say, that even if government bonds earn a real return of 2% or less over time, stocks might earn, say, 4% -- very low by historic standards but perhaps still better than the alternatives, and better than whatever traditional Social Security benefits might provide decades from now.

It is, ultimately, all guesswork. But what we know is that a 3% real rate of return on government bonds can’t be had today.

At a time when many investors worry that decent financial asset returns will be tougher to come by in the next two decades than the last two, a private- account proposal with a 3% hurdle rate runs the risk of looking like too much stick and not enough carrot.



Dwindled returns

Taking a very long view, the real, or after-inflation, total return on long-term U.S. government bonds has fallen sharply over 200 years.


Average annualized real return on U.S. bonds in four periods.

1802-1870: 4.8%

1871-1925: 3.7%

1926-1945: 2.2%

1946-2004: 1.6%


Source: Prof. Jeremy Siegel; Ibbotsun Associates


Tom Petruno can be reached at For recent columns on the web, visit