By Alicia H. Munnell
July 31, 2013

I have been thinking about risk-sharing in pension plans lately.  Theoretically, the way pensions are currently structured in the United States, sponsors bear all the risks in defined benefit plans and employees bear all the risks in defined contribution plans, such as 401(k)s.  Placing all of the risks on one party or another doesn’t seem like an optimal outcome.  And sponsors of defined benefit plans, at least, are looking for a better arrangement – a little in the United States, more in Canada, and a lot in The Netherlands. [EXPAND Read more]

In the public sector, the financial crisis and ensuing recession produced ex post risk-sharing.  A number of states eliminated or suspended their post-retirement cost-of-living adjustments (COLAs).  That these actions passed legal muster was surprising, at least to me, given that – in the vast majority of states – changing future benefits for current employees is extremely difficult.  The courts apparently determined that COLAs have a different status than “core” benefits.

Ex post risk-sharing, however, can seem arbitrary and unfair.  A more equitable arrangement would be one in which employees know in advance that the COLA is a contingent benefit.  The Wisconsin Retirement System was a pioneer in this area, directly linking the payment of a COLA to investment returns. Instead of being promised a COLA, retirees are provided performance-linked benefit increases if smoothed asset returns average at least 5 percent and all actuarial assumptions are met.  Retirees are guaranteed that their benefits will never fall below the initial level they receive at retirement.  The Wisconsin approach, however, leaves participants bearing risk only during their retirement years, not when they are working.  Most individuals would prefer to bear risk while they are young and working rather than older and retired.

The Canadians are taking a more comprehensive approach.  In May 2012, the government of New Brunswick introduced the Shared Risk Pension Plan, a promising innovation in the design of defined benefit plans that significantly reduces the employer’s risk.  The New Brunswick plan design divides pension benefits into “base” and “ancillary” components and provides a clear specification of how the plan would reduce benefits, increase contributions, and change its asset allocations in response to funding deficits; and how it would restore benefits, including previous benefit reductions, reduce contributions, and change its asset allocations as its financial condition improved.  All of these specifics are worked out on a plan-by-plan basis for the various plans adopting the new model.

The Canadian initiative was based on risk-sharing provisions in Dutch defined benefit plans, where benefits are based on indexed career average earnings.  In these plans, earnings are usually re-indexed each year to take account of inflation or wage growth, and after retirement benefits are generally indexed to the increase in prices or wages.  The level of indexation in any given year depends on the financial position of the pension fund.  Thus, if the fund is below its solvency target, the indexation rate for that year will be less than the growth rate of the relevant index.  For example, say that wages grow by 4 percent.  The indexation factor might be reduced to three-quarters of the wage growth rate.  In this case, retirees would see their benefits rise by 3 percent.  And active workers would see their wages adjusted by 3 percent for the purposes of calculating their earned pension rights for that year.  On the other hand, when pension plans are well funded, this process can work in reverse to catch up for prior years in which workers received less than full indexation.  (This summary is drawn from a 2007 paper by Eduard Ponds and Bart van Riel, who provide a helpful overview of the Dutch model.)

The whole risk-sharing business is a tough slog.  Employees don’t always understand what will happen to their benefits under alternative circumstances.  It is impossible to construct rules for all contingencies.  For example, the Dutch had not envisioned a financial collapse as dramatic as 2008.  And, finally, people do not like to have benefits taken away.

All that said, in the end, employers are not willing to bear all the risk, and, in the U.S. public sector, they didn’t in the wake of the financial crisis.  It seems much better to establish clear expectations about what will happen when things go wrong.  So let’s start working out the best approach to risk-sharing. [/EXPAND]