In general, I have urged states to abandon the defined-benefit ("DB") pension model and move workers to defined-contribution ("DC") plans such as 401(k). This shift has been commonplace in the private sector; there is also significant public-sector precedent for DC, including public workers in Michigan and Alaska and most state university systems.
There is one set of governments that require a modification of this prescription: those in the 15 states where many public workers are exempt from Social Security. State and local governments have never been required to participate in Social Security. Most agreed to opt in long ago--and once you're in, you can't go back--but in these states, some or most government employees are exempt, relying instead on their employer-sponsored pensions. Unsurprisingly, states where public workers don't get Social Security tend to pay larger pensions; of the eleven states recently listed in the New York Times as paying the highest pensions relative to public worker salaries, nine are states where many public workers are outside Social Security.
A shift to a pure-DC system in states that opt out of Social Security would expose public workers to unacceptable risk of destitution in old age. Reforms in these states should be designed to include a guaranteed antipoverty benefit, either by opting into Social Security or by retaining a reduced defined benefit. Additional benefits should be provided through a defined-contribution structure.
The first option, entering Social Security, is well-precedented: most government workers have been in the system for decades. Such a move would have the benefit of making part-career government employment more appealing; currently, Social Security rules are very unfavorable for workers who spend part of a career paying into Social Security and part working for a government that opts out of the plan. However, states have a very good reason to stay out of Social Security: participation requires both the employer and the employee to pay 6.2 percent Social Security tax.
Of course, workers receive valuable Social Security benefits in exchange for those taxes. But the trade is not usually a very good deal. In the case of a two-earner married couple born in 1973 and earning income in line with the national average wage index, the effective internal real rate of return on Social Security contributions is approximately 2.32 percent.
Assuming 2 percent inflation (for a nominal return of 4.32 percent), this compares unfavorably to the returns on defined-benefit pensions, whether you take pension plans' 8 percent discount rates at face value or substitute a market-value rate in the 5-6 percent range. If you predict that Social Security benefits will be reduced in a future fiscal adjustment, the comparison gets worse. (If you're wondering why most states took the plunge into Social Security despite this math, one reason is that the numbers looked a lot more favorable decades ago when most made that choice.)
The return on Social Security contributions varies with the taxpayer's circumstance--returns are best for married couples with only one earner, and for people with very low incomes. But the average worker can expect a less favorable return from Social Security than he or she could get by investing conservatively, or than a conservatively invested defined-benefit plan could sustainably offer. This isn't surprising; it reflects the fact that payroll tax is only partly a form of retirement saving, and is partly diverted to finance non-Social Security operations of government.
The economics of opting out of Social Security, therefore, are compelling both for state and local government employees and the taxpayers who pay their salaries. The opt-out states are going to be strongly motivated to find pension reforms that allow them to stay out of the Social Security system.
Without Social Security, these governments will need some way to ensure that their workers cannot be left destitute in retirement, which means some defined-benefit component of retirement benefits. However, this does not mean that reform should be limited to tweaks of the existing DB system, such as raising contribution rates or the retirement age. Instead, pension reforms in non-Social Security states should follow two key principles.
One is that the defined benefit should be sized to be an antipoverty benefit. Essentially, this means reforming the benefit to look more like Social Security. In most states, defined-benefit pensions replace more than 50 percent of pre-retirement income, and in many cases can be drawn once an employee reaches a retirement age in the late 50s or early 60s. Social Security will replace 31 percent of pre-retirement income for the typical Baby Boomer, and the full benefit cannot be drawn until age 66. States should emulate this, offering defined benefits that are smaller and available later in life--and therefore much less expensive.
Retirement benefits in excess of this antipoverty benefit should be structured as defined-contribution benefits. Governments should also design these benefits on a sliding scale similar to Social Security--the defined benefit should replace a larger share of pre-retirement income for the workers with the lowest incomes, with the defined contribution plan gaining more importance for higher-income workers.
Once states have designed a limited defined-benefit along these lines, they should take steps to ensure that it remains limited. The best way to do this is to require in the state constitution that any retroactive increases in pension benefits must be fully funded at the time of enactment. This will prevent lawmakers from sweetening pension benefits while sending the bill to future taxpayers, one of the key political risks inherent in DB systems. One non-Social Security state, Maine, already has such a provision in its state constitution.
The risks to taxpayers associated with defined benefit pensions--countercyclical pressures on taxes and spending, and temptations to overpay due to opaque accounting--are present even in the non-Social Security states. Yet these states also have compelling reasons not to abandon the defined-benefit model altogether. Good plan design can mitigate the risks associated with DB and make the tradeoff worthwhile, but the risks will never go away.
This prescription assumes that federal law remains unchanged. At the federal level, lawmakers should consider making Social Security mandatory for government workers, in the name of expanding the tax base. Keeping public employees in Maine out of Social Security is good for both Maine taxpayers and Maine public employees. But it's bad for taxpayers everywhere else, as we lose out on roughly $30 billion in annual payroll tax collections from the exempt states. I don't expect Mainers (or Californians, or what have you) to selflessly give up this sweet deal; but the rest of us would be well advised to take it away.
If public workers were forced into Social Security, then the reform prescription for the 15 non-Social Security states would become the same as everywhere else. That would be bad news for taxpayers and public workers in places like California and Maine. But it would make the tax burden more equitable across states, and would be good news for most Americans.
Josh Barro is the Walter B. Wriston Fellow at the Manhattan Institute.