State and Local Pensions: How Big Is the Problem?

I was recently assigned the task of starting off a discussion on state and local pension plans by describing the size of “the problem.”  My sense is that people like to be assured that the problem is really, really big – so big that SOMETHING HAS TO BE DONE.  I can play in that arena; I can produce big numbers.  But such an exercise does not seem very helpful.  The challenge is not to produce a snapshot of where plans are today, but to understand the key drivers for the future.

A snapshot measure of “the problem” is the funded ratio – the ratio of assets to liabilities.  The data show that state and local plans were more than 100 percent funded in 2000 before the bursting of the dot.com bubble, which sharply reduced assets and the funded ratio.  The funded ratio was starting to turn around when the 2008 financial crisis hit.  In 2012 – the latest data available – the aggregate funded ratio was 75 percent.  That is, assets amounted to 75 percent of the present discounted value of promised benefits.

Aha, say critics, the liability numbers reported by the actuaries are calculated using the expected long-run return on pension assets rather than by a rate that reflects the riskiness of the liabilities.  Since the benefits are guaranteed in many states, the appropriate discount rate should be closer to, say, the riskless rate.  This point is technically correct for reporting purposes.  So, discounting by 5 percent instead of 8 percent shows that state-local plans in the aggregate are only 50 percent funded.  That seems really bad.

Another measure of the problem is the unfunded liability – the difference between assets and the present discounted value of promised benefits.  Again, the magnitude depends on how the liability is calculated.  Using the actuaries’ assumption of 8 percent produces an unfunded liability of $0.9 trillion.  A discount rate of  5 percent $2.7 trillion.  And for those who like really big numbers a discount rate of 1 percent $9 trillion!!

But in real life, the burden of pensions on states and localities has nothing to do with the discount rate.  Rather it depends on the returns that plans can earn on their assets and the generosity of the benefit package.  Right now, pension expense is about 4.6 percent of state and local budgets.  If sponsors really do earn the assumed 8 percent return, costs will stabilize around that level going forward.  If they earn only 6 percent, costs will roughly double.   

How crazy is the 8 percent assumption?  Most experts will tell you it’s outlandish.  But my colleague Josh Hurwitz points out that the 8 percent consists of an assumed real (inflation adjusted return) of 4.7 percent and an inflation assumption of 3.3 percent.  An assumed real return of 4.7 percent for a portfolio with two thirds invested in stocks and other risky assets is not unreasonable.  The nominal rate looks high because the inflation assumption is too high.  But the inflation assumption also inflates wages and post-retirement cost-of-living adjustments (COLAs), so it also makes costs look too high as well.  In fact, in calculating pension costs, inflation more or less washes out.  So you can decide whether the assumption is reasonable.

In terms of benefit promises, plan sponsors have made an enormous number of changes in the last few years.   Our study of 32 plans in 15 states shows that COLA cuts, increased employee contributions and sharp reductions in benefits for new employees will markedly reduce pension costs.  Since the major savings are for new employees, the savings will not be felt for a while.  As a result, plans may face challenges for the next ten years.  But thereafter, if the cuts stick, costs should start to come down.

The real message is that we should stop debating what interest rate should be used to discount benefits for reporting purposes.  Let’s focus on the two drivers of the future impact of pensions on state and local budgets – earnings on assets and the generosity of benefits.  Plan sponsors have already made major changes.  Let’s see if they stick.  Their real return assumptions may not be much out of line.  Let’s see what happens to the markets.