Stocks, bonds and social security

by John Quiggin on March 5, 2004

Brad DeLong has had a string of posts referring to the possibility that some or all of the US Social Security fund should be invested in stocks rather than, as at present, in US Treasury bonds, of which the most pertinent is this one. This idea first came up in a major way in Clinton’s 1999 State of the Union speech, and has since had some play on the Republican side, especially now that privatization individual accounts seem to be off the agenda.

The key fact that makes the idea attractive is the equity premium, the fact that, historically the rate of return to investment in stocks has been well above that in bonds. This used to be explained by the fact that stocks were riskier than bonds. But ever since the work of Mehra and Prescott in the 1980s it’s been known that no simple and plausible model of the social cost of risk that would be generated by efficient capital markets can explain more than a small fraction of the observed premium. The immediate response, that of finding more complicated, but still plausible models hasn’t gone very far. The alternative explanation is that capital markets don’t do a very good job of spreading risk. For example it’s very hard to get insurance against recession-induced unemployment or business failure, even though standard models imply that this should be available.

Simon Grant and I have done a fair bit of work on this, with some specific attention to the Social Security issue. In this paper (large PDF file), published in the American Economic Review, we argued that substantial gains could be realized by investing Social Security funds in the stock market. We didn’t put a number on it, but I don’t find Brad’s half-embraced suggestion of $2.4 trillion in present value implausible.

An important point, though, is that investing in stocks will generally not be the best way to go, at least if the amount invested is large. A government agency holding, say 20 per cent of the shares in Ford and General Motors, would seem to have big problems. Leaving aside the specific institutional issues of the US Social Security fund, the obvious implication of the equity premium is that, unless there are large differences in operating efficiency between private and public enterprises, government ownership of large capital-intensive enterprises like utilities will be socially beneficial. The case is strengthened if monopoly or other problems mean that the enterprises have to be tightly regulated in any case. Again, Simon Grant and I have written this up, this time in Economica (PDF version available here)



Jack 03.05.04 at 12:53 pm

Is it generally accepted that Mehra and Scott proved the existence of the equity premium beyond challenge from selection bias? I thought that looking at equity returns across all markets active at the beginning of last century (Germany, Russia, Argentina etc.) the equity premium might not look so huge.

Secondly the argument seems very similar to the pension fund cult of equity which has not been serving them very well in recent years. It would certainly be subject to all of the same issues.

Thirdly I’m curious about how equivalent state ownership and equity market ownership actually are. Obviously the government holding 20 per cent of companies is not going to be simple but the leap to outright ownership brings a lot more baggage with it too. The UK Post Office is a good example. It has actually had some good managers and commited employees but has been burdened by a double whammy of high employment costs and underinvestment to the point of asset stripping entirely because it is publically owned.


theCoach 03.05.04 at 1:56 pm

Perhaps I am just viewing this on such a novice level that I do not understand, but wouldn’t the knowledge that stocks perform better than bonds, along with the knowledge that the risk is not as great as previously believed be available to all investors? If so, that would result in a relative increase in the price of stocks — someting that many people believe we are seeing right now. Daniel, can you inject some Hayek into this some how?


Joel B. 03.05.04 at 2:02 pm

Conservatives should have a general problem with the idea of government controlled funds invested in the market. In a perverse way, we can work our way back to state ownership of corporation something any capitalist should seek to avoid. State run pension funds seem to possess this problem as they exercise some sizable influence over private companies in an ownership, rather than regulatory capacity, a blurring of the tradition capitalist/socialist line. I do want to note, that the Recession Induced Unemployment Insurance actually tends to be quite available through banking institutions when you have a loan or line of credit. Not that I recommend the insurance, but I do think it has become quite widely available considering. The question become how much insurance, and I think companies largely come down against over-insuring individual against recession and job loss, too much risk, and the worst time to be paying out claims is when you(as the insurer) are hurting badly too(less investment income, if not loss).


Jeremy Osner 03.05.04 at 2:02 pm

I was led to believe by an article I read a while back, that moving the SS money from Treasury bonds to stocks would result in a large decrease in demand for Treasury bonds (because SS owns such a lot of them) and make US debt significantly more expensive. Is this plausible?


Joshua W. Burton 03.05.04 at 2:10 pm

Jack’s point about selection bias is a good one, and also occurred to me. There is a paper, “A Century of Global Stock Markets” (William Goetzmann and Philippe Jorion, NBER working paper 5901) that finds a real appreciation of about 1.5% — in other words, no equity premium. Treating the much higher US rate of return as typical amounts to a survivorship fallacy.

Another point often neglected by SS privatizers is that we don’t _have_ a trust fund for the whole retirement cohort; that’s not how SS works. What we have is a trust fund for the demographic bump, the diff between average retirement and boomer retirement. In 2000 dollars, it costs about $11T to retire a 30-year generation, but the 1940-1970 cohort is going to cost more like $14T. The baseline is pay-as-you-go, while the extra $3T (more than $7T by the time we pay it out) is trust fund.

The reason this quibble matters is that a pay-as-you-go system necessarily has lower returns than a trust-fund system, because _a whole generation of float is lost_. Imagine a totally stable world: no economic growth, no population growth, no unemployment, no inflation, everyone works from age 20 to age 70 and then lives to age 120, like Moses. There’s still a finite liquidity preference, so real interest rates are low but nonzero; let’s put them at 1.4%, so we get a safe-return doubling in 50 years.

Now the Blue Party wants a pay-as-you-go system, where my 50 years of work pays for my grandfather’s 50 years of superannuation; this involves no trust fund, no investment, and no return. The Red Party wants to put my 50 years of pay-in away, let it double, and then pay it back to me, which means twice as much retirement benefit for all. Where did the extra money come from?

Answer: look at the boundary condition. The Reds let some Depression-era coal miner starve to death, while the Blues fed the first generation of retirees, who never paid in, and are effectively carrying the resulting debt forever. Let’s say the Reds want to go over to a savings-plan system at a later date; the first generation who invests for themselves must _also_ borrow to pay their grandparents, and _the interest on that loan will eat up their investment return, to the end of time._

The only situation in which this looks attractive is a case like Chile’s, where the system defaults on current obligations, blames the outgoing thugs, and then starts clean with a Red Party system. The problem is that in such a country the accumulation of the initial trust-fund surplus does not guarantee its eventual payout; indeed, its very existence becomes an incentive to eventual looting. And so we’re back to the survivorship fallacy: it’s better to live in a country where they aren’t going to confiscate stuff, if only you can predict where that will be.


raj 03.05.04 at 2:40 pm

>This idea first came up in a major way in Clinton’s 1999 State of the Union speech, and has since had some play on the Republican side, especially now that…individual accounts seem to be off the agenda.

I realize that this is not the major thrust of the post, but I do want to point out that individual accounts already exist. They’re called Individual Retirement Accounts. They could be expanded and the allowable yearly contribution increased. And, if it is desired to integrate IRAs into social security, there could be a tax credit give on the SS tax for some portion of the IRA contribution. A special account would not be required for at least some portion of SS privatization.


Brad DeLong 03.05.04 at 2:54 pm

Re “We didn’t put a number on it”




cleek 03.05.04 at 2:55 pm

So, if the govt owns 20% of Ford (or any company), ins’t there going to be a huge incentive for the govt to prop up Ford in times of trouble?

And, wouldn’t partial ownership put the government in the position of of demanding that Ford do business a certain way? And isn’t that exactly what conservatives fear?


Jeremy Osner 03.05.04 at 3:09 pm

I see from the Brad Delong comments thread that Treasury bonds held by SS are different in some way from normal Treasury bonds. Does this undercut the argument about interest rates that I reference above?


limberwulf 03.05.04 at 4:16 pm

I think you are on to something. I would far prefer to see the money currently going to social security be allowed to be put into an IRA or similar program than to have the government do the investing. This gives the individual some options on where the money is invested, and takes away the truly scary option of the government actually owning portions of companies. The tendency of government to prop up corporations and vice-versa is already to great, no reason to strengthen that.


Barry 03.05.04 at 4:21 pm

Joshua W. Burton:

“The problem is that in such a country the accumulation of the initial trust-fund surplus does not guarantee its eventual payout; indeed, its very existence becomes an incentive to eventual looting. “

Current US situation: 1980’s, the Greenspan commission recommends accumulating surplus (with taxes mostly coming from middle and working class); 2000-01, Greenspan endorses the Bush tax cut (based on a rosy scenario, with cuts biased towards the rich as much as possible); 2004, Greenspan recommends cutting Social Security benefits.

Joshua, looks like you are correct.


terry 03.05.04 at 9:43 pm

Am I missing something? The current ss system is pay-as-you-go. The trust fund is a fiction. Social security taxes currently fund both ss payout and a sizable chunk of the federal budget. Investing some of those funds in the stock market makes no more sense to me than the government simply declaring that the T-bonds in the “trust fund” will begin earning 10%. All that matters, finally, for any given retiree are two things: 1) the amount of goods and services the economy is producing during his retirement; and 2) how they are distributed.


Joshua W. Burton 03.07.04 at 3:49 am

Terry: “Am I missing something? The current ss system is pay-as-you-go. The trust fund is a fiction.”

You’re partially right, Terry. The baseline of the system is pay-as-you-go, and if it weren’t for the demographic bump of the baby boom that would be the whole story. But the cohort born between 1945 and 1964 is about 30% bigger than it ought to be, against the overall exponential growth of US population. (That underlying growth, of course, is the _real_ Ponzi scheme, but so far we’re keeping ahead of Malthus with the help of technological progress and nonrenewable resources.)

Anyway, SS would have faced a serious crunch in the next twenty years, if Moynihan and company hadn’t created the trust fund. Since the mid-’80s, and continuing until about 2013, our payroll deductions overfund SS, and the surplus goes into a trust fund, which will be drawn down again to help fund the boomers’ retirement when Gen X reaches their peak earning years and still aren’t paying in quite enough.

But is the trust fund _real_? A vexed and politically charged question. If the US were investing in German debt instruments, I’d certainly call _that_ real, even though “it’s just going out again to fund other (German) programs.” Recall the scene in _It’s a Wonderful Life_, when Jimmy Stewart has to explain to the bank customers that their money isn’t in the vault; it’s in their neighbors’ homes.

OK, but if the US government (in its fiduciary role as SS trustee) invests in the US government (in its monetary role as the world’s leading debtor entity), is _that_ a real trust fund? My inclination is to say “yes”. Or, if someone gets argumentative about it and I’m feeling crabby, to reply along the lines of “Which part of ‘full faith and credit’ didn’t you understand?”

In the end, the SS surplus funds current US debt, in substitution for private investment that would otherwise fund the same debt. That lowers demand for US Treasury bills in the open market, which lowers interest rates, which makes the US economy more productive, which raises the future tax base, which creates the wealth that will fund the payback of the future SS deficit. When you’re talking about trillions of dollars, that’s as close to “real” as an investment can get, I think.


Joshua W. Burton 03.07.04 at 3:54 am

_That lowers demand for US Treasury bills in the open market, which….”_

Lowers _supply_ of US T-bills, of course. The rest of that causal sequence stands as written. Sorry, it’s Purim, and I’ve been drinking a bit.


terry 03.08.04 at 5:51 pm

Joshua, thanks for the explanation. What you seem to be arguing in the end is that higher levels of taxation (via FICA) applied to consumers (vs. investors) makes for a more productive economy and benefits all in the long run. The point that oftens eems neglected is that no amount of financial manipulation allows a government (as distinct from an individual) to “save” for the future. “Overfunding” the “trust fund” simply means shifting the tax burden.


Joshua W. Burton 03.08.04 at 7:00 pm

_”The point that often seems neglected is that no amount of financial manipulation allows a government (as distinct from an individual) to “save” for the future.”_

I think we’re in furious agreement about the substantive finance question. However, I am not persuaded that government savings are more (or less!) illusionary than private savings, excluding canned food in the basement. Both are promissory notes on future productivity. The SS trust fund is odd because (1) it’s a promise made by one agency of the US government to another, and (2) it’s such a big promise that macroeconomic effects are of the essence. But it’s a promise with a lot of heft and history behind it, just the same.


Joshua W. Burton 03.08.04 at 7:12 pm

_What you seem to be arguing in the end is that higher levels of taxation (via FICA) applied to consumers (vs. investors)…._

Oops, sorry: I didn’t read this carefully.

FICA “taxes” workers (consumers, if you like), but it only substitutes against investor taxation if you assume that we would be making up the difference through progressive taxation targeted at the investor class, were the SS surplus to stop pouring in. The more likely counterfactual, to my mind, is that we would hold (non-FICA) tax policy and discretionary spending fixed, and fund the missing piece by additional _borrowing_.

Thus, we are taking dollars out of workers’ pockets, not to leave dollars in (US taxpaying) investors’ pockets, but rather to avoid borrowing from (US _and_ foreign) investors on the open T-bond market. The benefit accrues to our own investors not through lower marginal taxes, but from lower real interest rates. Oh, and from the fact that we are paying the trust fund “interest” to ourselves, rather than to overseas T-bond holders.

Comments on this entry are closed.