Site Meter

Tuesday, February 22, 2011

Felix Salmon has a proposal

Pfau makes a very basic calculation that for someone on a constant real wage, saving for 30 years and then living for another 30 years on 50% of their final salary, saving about 16% of your salary each year into a portfolio of 60% stocks and 40% bonds will put you into safe territory.

Of course, real wages aren’t constant over time, and all the other figures are highly variable too. But the bigger message certainly resonates with me: spend less effort on trying to boost your annual returns, when you have very little reason to believe in your alpha-generation abilities, and spend more effort on maximizing your savings every year.


Matthew Yglesias comments

"They Could Call It “Social Security”"

I comment on Yglesias.





You ignored part of the quoted passage "a portfolio of 60% stocks and 40% bonds ". It wouldn't work so well if one invested only in bonds. Salmon's point is that people would be much better off buying and holding a diversified porfolio than trying to pick winners.

It would be like social security if the social security administration bought 60% stock 40% bonds not 100% treasuries. I think that would work out rather well frankly. Such an approach would elminate all future budget deficits in expected value. Of course the cost ist aht there would be surpluses in good times and deficits in recessions. We can't have that can we ? If the Federal government automatically ran a huge deficit during a recession,then we won't have recessions at all.

If there is a problem with this proposal, I haven't heard of it. It has been debated on the web for years.

Back to the quoted passage to explain more.

If an individual followed the rest of the proposal but bought only treasury securities, as the social security administration does, then that individual would be poor when old. This isn't like social security, because, constant population and wage growth, the SSA can pay proportional to the increase in total wages paid (due to both increased wages per worker and increased employment). This is a better return than the return on treasuries.

It also shows why a little issue like the baby boom creates such trouble. It is costly ex ante to the SSA, because for us boomers it has to build up a surplus and only gets the treasury rate on that surplus. Also the Republicans stole the money for tax cuts and wars but that's another issue.

3 comments:

dm said...

Have you read Geanakoplos and Zeldes?

http://www1.gsb.columbia.edu/mygsb/faculty/research/pubfiles/3352/GZ%20MVASSB%202010%20final.pdf

"The US Social Security system is “wage- indexed”; that is, future benefits are tied directly to the economywide average wage index around the year of the worker’s statutory retirement age. [...]
We argue that wages and stock prices are linked in the long run, effectively linking Social Security benefits to the performance of the stock market."

Robert said...

I haven't read Geanakoplos and Zeldes and trust your summary. I note that "linked" does not mean "the same." If one variable is consitently 5 more than another or 5 times another then they are perfectly correlated and very very linked.

The real wage bill grows about 3% per year. The average real return on stock over most very long periods is around 7% per year. The numbers are correlated but they are not the same.

Also, of course, the year to year variance in the return on stock is huge compared to the increase in the wage bill or the return on treasuries (the Social Security Adminstrations current portfolio).

Thus the Social security administration got a slightly better return than if it had bought stock in 1999 and sold March 2009. This was an extraordinary event, since it hadn't happened before in the history of the SSA.

On the other hand, if it held out till June 2009 to sell, then it would have had a larger trust fund than it currently has.

The historical effect of the proposal which I discuss would have been a larger trust fund every day more than 10 years after the program started except for a few around March 2009 and those only if the program started in 1999.

dm said...

"I haven't read Geanakoplos and Zeldes and trust your summary. I note that "linked" does not mean "the same." If one variable is consitently 5 more than another or 5 times another then they are perfectly correlated and very very linked. "

Well, think of the risk in wages as having the risk of a delevered stock portfolio (e.g. a 70/30 bond stock mix). The G-Z claim is that stocks and wages are more strongly correlated over longish horizons (retirement like horizons). A lot of the short-term noise in stocks reverts away at lower frequencies, leaving a less risky long run price path that tracks wages more closely. This is the sense that Social Security is like a risky defined contribution plan.

Your point about how to invest the trust fund is a separate issue. The solvency of the social security as a stand alone entity is largely a political concern. The federal obligations are still there no matter how it is funded. So, if you believe the equity premium is not a risk premium, but a free lunch, then it doesn't matter whether it is the SSA or some other agency that does the investing.

If the G-Z claim is valid, and you think the solvency of SSA matters, then it actually supports your proposal. An "easy" way to immunize the risk of SSA benefits is to hold a mix of stocks and bonds.

"The historical effect of the proposal which I discuss would have been a larger trust fund every day more than 10 years after the program started except for a few around March 2009 and those only if the program started in 1999."

But, did anyone lose a dollar investing with Bernie Madoff if they got out before 2008? :)