STUMP » Articles » Geeking Out (and Illinois Pensions): Fixing a Graph and Assigning Blame for Underfundedness » 29 October 2017, 19:51

I used a graph last week that I knew was wrong (well, part of it was wrong).

It’s my graph of the unfunded liabilities of the 5 Illinois state-funded pension plans, and how it stacks up over time. I know I’ve mentioned this issue before — it doesn’t work well when there are negative components (i.e., items that reduced the unfunded liability). This is a well-known problem with stacked column graphs, and there aren’t a lot of good fixes… but I’ve come up with something.

Also, I’ve updated the data. The prior data went up to FY 2014.

When a component, such as investment performance, has a negative aspect, I just reduce all the positive amounts in a pro-rata sense because the point is to assign blame in these graphs… it’s not like there’s much to assign good behavior for. In general, most of the elements only add to the liability (over contributions don’t often come about, for example). I would probably use a waterfall chart if there had been enough good-behaving items.

I don’t have that issue with Illinois pensions. Only every so often the investments do well enough to really cut into the unfunded liability.

Also, I’ve had an upgrade to the current version of Excel on one of my computers, which has different default graph palettes to choose among. I chose the colorways called Marquee. Woo. Excitement.

Downloadable spreadsheet here.

THE RESULT: ILLINOIS PENSIONS LACK DUE TO UNDERFUNDING

Here is my fixed graph — I start the unfunded liability in 1984, because it was a mere $7.3 billion in that year.

For fiscal year, it ended at over $125 billion. Yeah, I have the graph exceed the vertical scale I give on the axis… but not by a lot.

This is a big graph. I started it in 1990, so you can really see it grow.

The 5 state-funded pensions are included. I’m trying to build up similar graphs for other Illinois pensions, specifically IMRF. Because IMRF is a (close to) 100% ARC payer.

That is not the case for the 5 state funded plans. The largest single component driving the unfunded liability for these is that the state chose to underfund these pensions. Out of $126 billion in unfunded pension liabilities in fiscal year 2016, $52 billion is due to decades of deliberate underfunding.

That was a choice.

A few other items were also a result of choice: benefit increases, and salary amount changes (due to contract changes, etc.)

Investment results are less a matter of direct choice, but we can see that they are small compared to the deliberate underfunding.

MEASURING IN A RELATIVE WAY

Here we go – let’s break it down to percentage:

41% of the unfunded liability in 2016 was due to underfunding. 9% was due to investment underperformance.

But I want to point out the 24% due to changes in actuarial assumptions. There were three major changes that I could see over the years — valuation rate, mortality assumption, and payroll increase assumption.

So let us think about these.

ACTUARIAL ASSUMPTION CHANGES: DO NOT ACTUALLY MAKE PENSIONS MORE EXPENSIVE

Actuaries do not determine when you die. We look at stats and say “this looks like the pattern”. Sometimes actuaries project trends… and in public pensions, many plans do not project that people will live longer into the future.

The pensions cost whatever they’re going to cost. The cost is ultimately determined by the cash flow amounts and their timing.

Actuaries try to figure out what amount of money one needs right now to cover future cash flows. To do that, we have to make a lot of assumptions.

So, it turns out that the Illinois pension funds used to have a much higher assumption for return on assets/discount rate. It used to be 8%. Then, starting in 2010, the plans started to notch these down.

At of now, some of the plans use: 7.25%, 7%, and 6.75%. That is one of the large movers.

Mortality assumptions are a bit slower to come to the fore, but those also now reflect longer lives in retirement.

So here’s my comment: let us assume the new valuation assumptions are closer to what is actually going on with the amount & timing of future cash flows. And that the funding levels would stay what they are (in absolute terms).

Then the unfunded liability due to these valuation changes would really be unfunded liabilities due to too-small employer contributions.

This is starting to get into philosophy. It is very clear that at least 47% of the unfunded liability is due to deliberate choices (underfunding + benefit increases).

I would argue, from a philsophical viewpoint, that all of the shortfall is due to underfunding.

But that’s a bit abstract. I will come back to it with an analysis of the IMRF unfunded liability.

Compilation of Illinois posts