Photographer: Bradley C. Bower/Bloomberg
Photographer: Bradley C. Bower/Bloomberg

In the comments section on my recent post on Detroit’s pension woes, longtime reader Bronxcobra asks: Why couldn’t we just have had companies buy their pensions from insurance companies?

You can, after all, pay an insurance company to sell you a deferred annuity. When we were reforming busted pensions in the 1970s, why didn’t we just mandate that companies and governments do this, rather than construct their own pension funds?

The first answer to that question is that such annuities are expensive. Warren Buffett wrote an extraordinary letter to Katharine Graham in 1975 about the Washington Post’s pension system, in which he vividly makes this point:

“If you promise to pay me $500 a month for life, you have just expended -- actuarially, but nevertheless, in a totally tangible sense -- about $65,000. If you are financially good for such a promise, you would be better off (if I have an average expectancy regarding longevity for one my age) handing me a check for $50,000. But it wouldn’t feel the same.”

If you want to adjust those numbers for inflation, he goes on to point out, it will cost even more. A lot more, because the company making the promise has to bear considerable inflation risk.

Defined-benefit pensions would probably never have been established if the costs had to be borne this way. Arguably, defined-benefit pensions were done in even by the relatively easy accrual requirements mandated by the Employee Retirement Income Security Act. Pensions managed to limp along for about a decade, because the high inflation of the 1970s quickly eroded the real value of what they were paying their retirees. But when inflation fell in the early 1980s, the number of defined-benefit pensions went into steep decline.

Of course, if the politicians of yesteryear had to pay for their pension promises, they might not have made them in the first place. But these are the forces arrayed against such a move: It would likely have ended defined-benefit pensions rather than make them safer. And no one wanted to end defined-benefit pensions.

The second problem is that insurers can fail, just as companies do. It’s relatively rare, because they’re heavily regulated. But it’s not impossible.

A third problem is transaction costs: Many large companies and multi-employer plans actually self-insure for health insurance, because once they’re large enough to have an actuarially sound pool, it’s cheaper. They may have a health insurer provide the administration, but they provide the funds for payments. Such a reform might well make pension contributions even more expensive than they have to be.

And the fourth problem is that replicating a traditional defined-benefit pension would actually be a pretty complex product, requiring guesses at your final earnings and whether you’ll leave before you fully vest. We could alter the product so that you’re paid out a certain number of dollars for every year of service at a given salary, and that’s that. But this would probably result in a much smaller pension for many people, and it would have encountered pushback from unions and civil-service organizations, for whom seniority’s perks are sacred.

And here’s the thing: This doesn’t necessarily even solve the problem we want to solve (employers who don’t put enough of a contribution into the fund to guarantee the promised payout). Presumably, corporations will put their pension funds out to bid. And who will be the lowest bidder? That’s right: the fellow with the most optimistic assumptions.

Ultimately, we’re relying on state insurance regulators to ensure that companies don’t make wild-and-crazy bets on their ability to provide a deferred annuity of a certain amount. But we were relying on pension regulators to make sure that pension funds stayed sound, and look how well they did. If we can’t get pension regulation right, then why do we think we’ll get deferred-annuity regulation right?

One way is simply to be much more conservative. Pensions would be invested in the sort of things that insurers get to invest in, which is to say blue chips, highly rated bonds and no exotic securities, thankyouverymuch. They would need sizable cash reserves, as well as very conservative assumptions about the growth rate of the assets they were managing and the expected payouts to the people whose retirements they were ensuring.

All of which brings us back to my earlier point: A really safe defined-benefit pension plan, one with a less than 1 percent chance of failure, would not have been a popular product. In some sense, the very existence of defined-benefit plans depended on the unrealistic assumptions that were made about the future. If we had done something sensible, such as buying deferred annuities from insurers at prevailing rates, then we would not now be fretting about the destruction of the defined-benefit pension plan, because people would probably never have had them in the first place.