STUMP » Articles » Pension Return Assumptions and Discount Rates: An Overview » 18 December 2016, 08:26

A few news stories have gone by recently, with respect to discount rates (and asset return assumptions) made by public plans.

The big one is that Calpers is considering reducing its discount rate/return on assets assumption:

Opinion: Every taxpayer should watch what Calpers decides about its investment-returns forecast

The investment team at Calpers has been signaling that the current assumption of 7.5% long-term investment returns may be too high, and it would probably peg it around 6% instead.

The 7.5% rate is hardly unusual among government pension funds. But as the $300 billion gorilla in the pension world, any change — likely to be discussed at its Dec. 19 board meeting — will be watched by other pension plans that could then have a harder time justifying their own targeted returns.

This chart of expected-return assumptions of some of the larger pension plans shows that while about 40% of the largest corporate plans assume returns 7% and below, only 9% of public pensions assume returns 7% and below.

There is an element of subjectivity in estimating future returns, and the decision comes loaded with behavioral and political implications. The number determines how well-funded a pension plan is, which in turn determines the amount a state or municipality needs to pay into their pension system. A higher number can make the pension plan appear healthier, requiring lower contributions today.

If the funded status appears artificially high, one risk is that officials can promise a level of benefits that they may be unable to maintain. In the future that can lead to higher taxes, fewer services and even a cut in actual pension benefits.

The flip side is that reducing the expected return to 6% may double the immediate contributions of some municipalities, according to Pension & Investments, adding to local budget stresses.
……
This case highlights two negative behavioral implications of using artificially high future investment assumptions to value pension benefits. First, the impression of being well funded can lead to making promises you cannot keep. In 1993 Dallas provided generous benefits without appropriately accounting for how it might be able to meet them. Second, expecting higher returns can drive money into riskier investments with the promise of higher returns. In the case of Dallas, the pension system invested in real-estate deals that didn’t deliver as expected.

…..

Critics of moving to a more financially accurate method of pension accounting rightly point to the immediate pain that would cause. If municipalities in California have to double their contributions by reducing discount rates from 7.5% to 6%, then it would be a disaster for them if they used a discount rate similar to that of corporate pensions.

However, there is something flawed in this logic. Instead of acknowledging the problem and then finding a solution, the critics would rather pretend the problem doesn’t exist. It’s the difference between “you owe $6 trillion, let’s work out a way for you to pay it off over time” and “you only owe $1 trillion, no big deal”.

Unless the true extent of underfunding is brought to light, we may be providing beneficiaries with a false hope that their promised benefits are secure. If taxpayers reach the limit of what they are willing to contribute, and state and federal governments are unable or unwilling to step in, then a cut in benefits may end up on the cards.

So I will make the note that public pension funds do tend to set their discount rate on the liability cash flows at the same level as their assumption of return on assets.

I’m not going to argue the theory behind this, but there is no reason that these should be the same. The asset cash flows and liability cash flows have different risk profiles, and should be discounted differently.

I will cover the Calpers-specific debate in a later post, but I want to set up this week’s posts below.

BACKGROUND

I’ve done a few posts on this earlier:

• Public Pensions Primer: Time-Weighted vs. Dollar-Weighted Returns
• Primer: Average Investment Returns – Which Average?
• Public Pensions Primer: Choice of Discount Rate Influences Results
• Public Pensions Primer: The Choice of Discount Rate and Return Volatility
• Public Pensions Primer: How Discount Rates Work

From the first post in that list, I had put together this diagram:

Having fixed up a spreadsheet that had been giving me absurd results for some plans (if you ever use XIRR in Excel, ask me about this…it’s not a bug — it’s a feature!), I can now share the results of the time-weighted average versus dollar-weighted average for almost all the plans in the Public Pensions Database.

Spreadsheet can be downloaded here.

TIME-WEIGHTED V. DOLLAR-WEIGHTED RESULTS

Here’s the graph:

The line on the graph is the dividing line between where the two average returns are the same. The points above the line are troublesome: that’s where the time-weighted returns (which are the ones reported in media) are higher than the internal rate of return on the system.

The ones below the line are interesting, as the internal rate of return is higher than the time-weighted.

I have labeled a few of the points, just to give an idea. I will be following up on specific plans – notably, Dallas Police & Fire, later this week.

NERDY DETAILS

Simplifications made:
- start with earliest market value of assets, as on the first of that year (earliest possible year: 2001)
- end with the market value of assets for either fy 2014, or fy 2015, depending on latest available
- net cash flows (contributions – assets) occur midyear
- removed the following four pension plans:
Arizona State Corrections Officers
Colorado State
Portland Fire and Police Disability Retirement Fund
Seattle Employees Retirement System

I had to remove them due to wonky numbers – Colorado state is the only sizeable one of these four, and there is some fishy stuff in that I think its database entry got corrupted. But that’s an issue for later.

I have done testing of moving the cash flow point around, but it doesn’t move the result too much, which is what I expected.

Compilation of Dallas posts