The Denver Post is reporting that negotiators are nearing a deal on PERA, the generous defined-benefit plan that most state workers have benefited from over the years:

The major changes to the Public Employees’ Retirement Association include increasing employee and employer contributions by 2 percent and reducing cost-of-living increases for current retirees from 3.5 percent this year, capping them at 2 percent….

Several issues remain to be resolved, most revolving around age of retirement and years of service needed to get full benefits, but both men said those issues could be resolved by the time lawmakers convene for their 120-day session next week….

So let’s assume that accounting for the government worked the same as accounting for a private pension.  In fact, in this case, there’s no good reason why it shouldn’t.   Basically, the plan has assets and obligations, but both of those change over time.   So the inputs to the model are 1) Actuarial Assessments, and 2) Interest Rate Assessments.

Actuarial assessments include things like Years of Service, Age of Retirement, Years of Benefits, Salary Increases (due to seniority), Benefit Increases (due to age).   Interest rate assessments include benefit inflation, health care inflation, discount rate, and return on plan assets.

The things that can be adjusted generally fall into Actuarial Assessments, and that’s where the article focuses.  Retirement age and years of service all fall into this category.  What’s critical is the stuff that’s left out.  We have no idea what the plan’s assumed rate of inflation, discount rate, rate of benefit inflation or health care inflation are, or what the assumed return on investment is.  We don’t know what they’ve assumed them to be in the past.  If those numbers are unrealistic, or even aggressive, we’ll likely find ourselves right back in the same place a few years from now.

Consider a simple scenario, where the plan assumes a constant 8% real return on plan assets.  Historically, this might be reasonable.  But if the bulk of the return is in the out years, the plan will have depleted its assets before those returns can catch up, and will run out of money.  (Cool graphs on this topic here.)  If you could forecast how returns would change over time, you’d have a more accurate model, but the fact is, as we’ve seen time and again, it’s impossible to make those sorts of predicts 5 years out, never mind 25 years out.  Which means that the solvency of any defined-benefit plan is mostly guesswork.  Promises of long-term solvency are simply mirages.

Maintaining a defined-benefit for incoming and even current employees  is not realistic (promises made to those already retired must be honored).  The only fair way to move forward is to transition to a defined-contribution plan, which has only assets, and by definitions, no liabilities.  Unfortunately, the political will for this move doesn’t seem to exist.

UPDATE: According to the actuarial projections accompanying PERA’s legislative recommendations, they are indeed projecting a constant 8.0% return for the next 30 years.  This strikes me as aggressive.  But they key point to remember is that these returns are never constant, and that the shape of that returns curve strongly affects the ending balance.  There is simply no way for even the best prognosticators to get that right, and worse, no acknowledgment in the docs that it even matters.