If Liberals Hate Job Lock, Why Do They Like Defined Benefit Pensions So Much?

Last week, when the Congressional Budget Office released a report saying that Obamacare's insurance subsidies would reduce people's inclination to work, liberals celebrated this as a victory over "job lock." And I agree: tying health insurance to work makes people more inclined to work, but that's not necessarily a good thing.

We don't want an economy where job-based health insurance discourages people from starting businesses or staying home to raise kids. We don't want a health care system that prevents people who can otherwise afford retirement from quitting their jobs at age 63. And more broadly, we shouldn't structure benefits in a way that makes people beholden to their employers and penalizes them for changing jobs.

But! Given all that, why are liberals often so keen on defined-benefit pension plans, which are explicitly designed to create job lock?

Here's how the typical defined-benefit plan works. First, it has a significant vesting period, and if you quit during that period you get no benefits at all. In 2010 reforms, New York State raised this vesting requirement from five years to ten. This creates job lock: Workers are captive to their jobs, waiting around for their benefits to vest.

Even after the vesting date, pension benefits accrue non-linearly, with accrual typically backloaded into the third decade of work. This creates more job lock, as workers have to stick around with the same employer until late in their careers when pension benefit accrual is actually generous.

In 2010, Robert Costrell and Michael Podgursky provided a bracing example from the Missouri Public School Retirement System. A teacher who starts work at age 26 can be expected to accrue a pension benefit worth a modest $100,000 by age 46. (That's a net present value of benefits, not an annual benefit amount.) But by age 56, his or her accrued pension benefits would soar above $600,000.

And then the teacher would face reverse job lock, with accrued pension benefits actually falling with every year worked. If that Missouri teacher worked all the way to age 65, his or her accrued pension benefit would fall to around $500,000, because the guaranteed annual payments don't rise fast enough to offset the fact that the worker's age of retirement keeps getting closer to death.

This system is designed to cause the teacher to work until his or her mid-50s and then quit, whether or not that's good for the teacher or the school system. This is job lock, and it's bad for mid-career entrants, people who would like to work until they die, parents who want to take a few years out of the workforce to raise children, and even full-career teachers who split their careers between two states (and therefore two pension systems).

The usual justification is that job lock is good for the employer: It helps a firm or government agency retain experienced workers once it has expended significant money and energy on training. That may be true in some fields, especially the military. But it's not a general principle. Teachers, for example, get up to speed quickly: Half their gains from experience come in the first year of teaching, and a teacher with five years of experience is on average as effective as one with 25 years.

Similarly to health policy, we should look for a retirement policy that promotes adequate saving without creating job lock. What would that mean? Well, let's start by looking through the key features of defined-benefit pension plans, and keep only the ones that are good.

(1) Here's the best thing about defined-benefit pensions: They are supported by large, comingled investment funds, and therefore have the benefit of professional management and economies of scale. 401(k) is a disaster for a number of reasons, but one big problem is that investors pay high fees because small accounts are costly to manage, and most individuals have no idea how to manage their own investments.

Felix Salmon is right: comingled funds are great. But you can have a comingled fund without any of the other trappings of a defined-benefit plan: You can have smooth accrual of benefits and you can have some or all of the investment risk borne by the plan participants, rather than by its sponsor. Galveston, Texas has a retirement plan like this.

(2) Job lock is another key feature of traditional pension plans, and it's bad. But it's easy to design a defined-benefit plan with no job lock. It's called a cash balance plan, and it works like this: for every year worked, a percentage of the worker's salary is credited to an account on his or her behalf, which then grows annually at a prescribed rate. As you work longer, your benefit grows proportionately, and there's no penalty for retiring early or late. Nebraska has a plan like this for public workers. Wells Fargo also had one when I worked there, and I left with a cool $7,000 balance after three years of work, which I then rolled into my IRA.

(3) Traditional pensions shift investment risk away from retirees toward someone else. This is desirable, to an extent. People with low and moderate incomes are not in a good position to bet their retirement on the stock market not crashing in the five years before they retire.

But, as I'll discuss below, the assumption of risk is costly, so we want to be targeted about it. This is a good reason for what I like to call vertical-hybrid retirement plans: A basic benefit with little risk to the retiree, and more risk loaded on the retiree as his or her wealth rises. We already sort of have a system like this: Social Security is a fairly flat benefit with no investment risk for the beneficiary, and other sources of retirement income are riskier. Employers could adopt a similar split: defined benefit accrual up to an annual cap, and riskier add-on accounts for higher paid workers.

But the biggest issue is the last one...

(4) Defined-benefit plans often have a higher benefit value than defined-contribution plans. This isn't universal; many university workers have high-value DC plans through TIAA-CREF, and federal employees have a split system where about half the value is on the DC side. But the average employer 401(k) match amounts to just 2.7% of salary, which is a key reason that 401(k) has not proved a sufficient substitute for traditional pensions.

Because Americans tend to be undersaved, a higher benefit value is desirable - so long as its cost is properly accounted for. Unfortunately, public sector pension plans hide the ball on cost by discounting future benefit payments at a 7-8% annual rate, effectively assuming that high equity returns will pay much of the cost of future benefits.

The oft-cited figure that public pension gaps in 2011 amounted to just 0.2% of GDP over 30 years is based on this error. That's the gap in the mean case where pension investment funds hit their return targets; above-normal returns could reduce that gap to zero, while weak returns would make the gap much larger. Unfortunately, that's not a wash; weak pension fund performance would be correlated with weak economic and fiscal performance, hitting governments with rising bills for pension fund contributions at exactly the same time that tax receipts fall and demand for government services rises.

In effect, by investing pension funds in risky equities and then guaranteeing fixed payouts, governments are providing costly insurance to pensioners, agreeing to make large fiscal transfers toward pension funds in the event of weak economic performance. An insurance company would charge you a great deal for the service of converting a 7% average return on investments into a 7% guaranteed return, as you can see if you price out an annuity, but the cost of that insurance shows up nowhere in the public accounts, until a recession hits and pension fund managers come to state legislatures with the news that taxpayer contributions must rise at the same time the overall state budget is strained, as happened in 2009-11.

That insurance cost does show up on pension accounts in the private sector: Regardless of how they invest their pension funds, private pension sponsors must discount future benefits at a rate that is tied to interest rates on safe bonds, reflecting the certainty with which they will have to pay out. That means General Electric records a much higher (and more accurate) cost for promising a $10,000 pension payment to a worker in 2030 than New York City would record for promising the same payout.

Quite relatedly, governments are more keen on the defined-benefit model than private companies are, because only the governments get to say they are eating a free investment lunch. And this gets to the core problem, other than job lock, with the way defined benefit pensions operate in the U.S. today.

Despite their backers' ideological differences, traditional pensions and 401(k) are built on the same fallacy: That we can set aside x% of GDP today in order to finance future consumption by retirees exceeding x% of future GDP, by investing those retirement funds in assets that rise in value faster than the economy as a whole grows. In both cases, those excess returns can only be achieved through investment risk, and that risk has to be borne somewhere: either by workers themselves, or by firms and state and local governments.

And for that reason, fixing the problem of retirement undersaving is going to require doing two fairly unpleasant things. One is admitting that retirement is expensive, and can't be paid for simply by relying on supernormal returns somewhere. If we want x% of GDP to consist of consumption by retirees, we should find a way to set aside x% of GDP as retirement savings.

The other is having the federal government play an increased role in retirement, because nobody else is in a good position to do it. State and local governments are ill-suited to guarantee pension benefits because they are supposed to balance their budgets annually. Firms have been getting out of the defined-benefit business for decades, as workers change jobs more often and federal rules requiring adequate funding have made pension plans less favorable to comapnies. And 401(k) itself is not working at all, with high fees, mismanaged investments, and not enough money being deposited.

With health policy, as liberal readers of the CBO report noted, the guiding vision should be to ensure individuals are adequately covered while separating labor market decisions from health decisions. The same goal should apply to retirement policy: Adequate saving, without making workers unduly dependent on or wedded to their employers. Only the federal government can make that happen.