Transition Costs and Public Employee Pension Reform

Chairman Metcalfe, Representative Cohen, and distinguished members of the committee, thank you for inviting me to testify on the subject of pension reform in the Commonwealth of Pennsylvania. I am a resident scholar at the American Enterprise Institute in Washington, DC. Previously I served as the deputy commissioner for policy and as the principal deputy commissioner of the Social Security Administration. In 2013–2014 I was the co-vice chair the Society of Actuaries Blue Ribbon Panel on Pension Funding. The opinions I express, however, are my own and not made on behalf of any organization with which I am or have been affiliated. 

I regret that I am unable to testify in person today owing to a travel conflict, but I hope that my written testimony will assist the committee in its consideration of public employee pension reforms. I would be happy to answer any written questions that members of the committee may have.

My testimony will focus on the idea that closing a traditional defined benefit (DB) pension to new hires and enrolling those newly hired employees in an alternate plan creates temporary or permanent “transition costs” that increase the cost of the now-closed DB plan. Claims of transition costs have been raised as an objection to public pension reforms in many states. One type of transition costs is based on the claim that accounting rules require that a closed plan pay off its unfunded liabilities more quickly. A second type of transition cost is based on the argument that a closed plan must fund its liabilities with a much lower-risk, and thus lower-returning, portfolio than a plan that is open to new entrants. I find that so-called transition costs are largely illusory and should not serve as an impediment to public employee pension reforms.

Introduction
Policymakers in cities and states around the nation are considering changes to address the rising costs and increasing budgetary risks of defined benefit public employee pensions. Some policymakers, including here in Pennsylvania, propose shifting newly-hired employees to defined contribution (DC) pensions similar to the 401(k)s used by most private sector workers. In a DC pension, employers and employees make a set (or “defined”) contribution to a retirement savings account. The account belongs to the employee, and he or she can choose how it is invested. The employee also bears the risk of the account’s investments. 

DC pensions pose less risk to taxpayers than DB plans and over the long term almost certainly would be cheaper. DC pensions are more transparent to policymakers, who have a clearer idea of how much pension provisions for employees will cost, today and in the future. And DC pensions can be more equitable to employees of different career-lengths. DB plans are very generous to full-career workers but pay little to short- and mid-term employees. DB plans’ uneven benefit accrual patterns endanger the retirement security of short- and mid-career employees. 

However, one objection to DC-based pensions for public employees is so-called transition costs. Transition costs are temporary cost increases associated with switching from a defined-benefit to a defined-contribution pension plan. Some advocates argue that these transition costs make switching to DC plans prohibitively expensive. The National Association of State Retirement Administrators states the logic succinctly: “The more underfunded a plan is, the more expensive it is to try to switch.”

I will address claims of transition costs in detail. However, common sense dictates that, if transition costs exist, they cannot be large. After all, there has been a massive shift in the US and throughout the world from DB to DC pensions. Not only did transition costs not prevent such a shift, but such costs have barely been mentioned in the several decades over which this DB-to-DC pension change has taken place.

I will here examine two types of transition costs. The first arises from an interpretation of accounting rules promulgated by the Government Accounting Standards Board (GASB) and claims that GASB rules would require a closed plan to more aggressively amortize its unfunded liabilities, raising costs in the short term. The second type of transition cost is believed to be generated by the need for a closed pension plan to shift to a less-risky, lower-returning investment portfolio. Neither of these claims should stand in the way of reforms to public employee pensions. 

Accounting-Based Transition Costs
Some claim that the GASB accounting rules require that a closed plan pay off its unfunded liabilities more aggressively. Doing so would causing a short-term increase in amortization costs, followed by lower costs thereafter. The National Institute for Retirement Security, the so-called “research and education” arm of the pension industry, claims that “Accounting rules can require pension costs to accelerate in the wake of a freeze.”

In fact, GASB standards don’t have the force of law, as is amply demonstrated by the nearly 60 percent of plan sponsors nationwide who last year failed to pay GASB’s supposedly “required” pension contributions. As economist Robert Costrell has clearly shown and as many pension actuaries now acknowledge, GASB standards are for disclosure purposes, not to guide funding. GASB itself declared that updated accounting rules issued last year “mark a definitive separation of accounting and financial reporting from funding.”

The Pennsylvania Public Employee Retirement Commission took a slightly different approach to the issue. It warned in 2013 that when a traditional pension plan is closed, “payroll will begin shrinking in the future and [amortization payments] will represent an increasingly larger percentage of payroll.” These payments “may become excessively burdensome for the remaining active member employers.”

But why? It is the government, not employees, who make amortization payments. What matters to the government is the actual dollar value of these payments, not the number of employees or employee payroll that amortization payments are “spread” over. Shifting newly hired employees to DC pensions will not magically make existing unfunded pension liabilities go away. These unfunded liabilities are effectively debts of the government, the vast majority of which must ultimately be paid. But nor does shifting newly hired employees to DC plans make these old unfunded liabilities larger.

Investment-Based Transition Costs
The second type of transition cost is based upon changing investment portfolios. The argument is that, once a DB pension is closed to new entrants, it must shift its investments toward much safer, more liquid assets that carry lower returns. Pennsylvania’s Public Employee Retirement Commission noted that such investment changes “would result in a lower valuation interest rate, which would result in higher actuarial accrued liabilities, requiring larger employer contributions as a percentage of payroll.”

Now, even if we assume that this claim is entirely correct, it does not imply that the government or taxpayers are made worse off by shifting to a less risky portfolio. Such a portfolio does carry lower expected returns but comes with a reward—lower risk, which means lower volatility of government pension contributions. The fact that the State Employees’ Retirement System (SERS) actuarially required contribution rose from about 4 percent of wages in 2007 to over 20 percent of wages today should warn policymakers about the dangers of taking excessive investment risk. The risk of pension investments passes through to the volatility of pension contributions. Lower-risk investments means more stable contributions for the government. Likewise, it can be shown using common financial calculations that the reduction in investment risk fully compensates the government for the reduction in expected investment returns. If holding bonds unequivocally made investors worse off, no one would hold bonds.

Nevertheless, there is good reason to treat these claims with skepticism. The logical reasoning is fairly straightforward. Most pensions do (or at least should) fund their near-term liabilities using less risky investments than their longer-term liabilities. This practice is often referred to as “asset-liability matching.” A closed plan no longer accumulates long-term liabilities, so over time the stock share of its portfolio declines, as does the expected return on its investment portfolio. But this cannot increase liabilities, precisely because it is the reduction in liabilities that causes the change in the investment portfolio. Simply put, eliminating liabilities cannot increase liabilities. 

It is difficult to estimate these effects precisely for SERS or other Pennsylvania plans because the plans release neither a year-by-year projection of their liabilities nor the mix of assets they use to fund short-, medium-, or long-term liabilities. In other words, the plans release a single figure representing their total liabilities discounted at a chosen interest rate. But plans do not release projections on the benefits they are obligated to pay in any given year. The Society of Actuaries Blue Ribbon Panel on Pension Funding recommended that plans do so, which would facilitate the type of calculations that are of interest here.

However, I can estimate, based on a stylized time-path of pension liabilities and the assumption that long-term liabilities are funded with 30 percentage points more stocks than near-term liabilities, that closing the plan to new entrants and thereby shortening the duration of plan liabilities would have a negligible effect on the expected investment return over the first several decades. Even over the longer term, the effects are small, perhaps 0.2 percentage points. The main reason the effects are so small is that pensions’ long-term liabilities also are small. In present value terms, most of a plan’s liabilities occur within the next 15–20 years. At the same time that SERS’s portfolio risk is declining, of course, the plan’s overall liabilities would be declining as well.

If this is the case, how is it that pension actuaries sometimes arrive at much larger estimates? The answer is unclear, given that they rarely specify their calculations. For instance, the Public Employee Retirement Commission cited an estimate from the Hay Group that, were SERS closed to new entrants, the risk of its portfolio should be reduced such that the expected return would decline by 1.5 percentage points over 40 years. However, as the Public Employee Retirement Commission noted, the “Hay Group did not provide an analysis of how they arrived at the interest rate structure.” A second actuarial firm, Buck Consultants, estimated that the expected return on SERS investments should fall by a larger 2.5 percentage points over 40 years. Yet a third actuarial firm, Milliman, recommended no change in investment practices. This is a case in which the actuarial firms should be asked, as we were back in grade-school, to “show your work.”

My best guess is that public plans assume that when a plan is open to new hires, it need not practice any significant asset-liability matching. That is, it might fund near-term liabilities with investments that are just as risky as those it uses to fund long-term liabilities. Indeed, it is almost mathematically impossible to assume otherwise. Given SERS currently places roughly 80 percent of its investments in risky assets, it would be difficult for SERS to make large distinctions between how it funds short- and long-term liabilities.

Indeed, there is evidence that supports this view. Over time, public plans have “matured,” meaning that more of their liabilities are for retirees and older workers and less are for younger employees. This process of maturation creates precisely the shortening of the duration of liabilities as would closing the plan, albeit on a more limited scale. And standard financial practice would be for a maturing pension plan to shift its portfolio toward safer investments. This is what US corporate pensions and public employee plans in other countries have done. 

Yet, US public sector plans have done precisely the opposite: even as their participant populations matured and the duration of their liabilities shortened, state and local pensions have opted to take more investment risk. This practice by public plans, the Society of Actuaries recently noted, goes “against basic risk management principles.” Thus, I would personally not be inclined to grant plans the benefit of the doubt on this issue.

Conclusion
Elected officials around the country are considering reforms to public employee retirement plans—and with good reason. The costs of these plans have risen significantly in recent years and the increasing risk of pension investments threaten to destabilize government budgets. Many reform options are available and policymakers should consider how much cost and risk taxpayers are willing and able to bear.

But one objection to shifting newly hired public employees to defined contribution pensions is the claim that doing so entails large “transition costs” owing to a closed pension plan taking less risk with its investments. If these claims were true, the massive world-wide transition from DB to DC pensions that we already have witnessed would not have taken place.