Monday, September 27, 2010

How big is the Pension Time Bomb?

Most of my more astute readers already understand that defined benefit pension funds is a time bomb waiting to explode. Now a new study that suggests that we don't even known how big the problem is.

To set the stage, the Wall Street Journal recently reported that public plans haven't changed their return expectations since 2001 and still have return expectations of 8%:
The median expected investment return for more than 100 U.S. public pension plans surveyed by the National Association of State Retirement Administrators remains 8%, the same level as in 2001, the association says.
The country's 15 biggest public pension systems have an average expected return of 7.8%, and only a handful recently have changed or are reconsidering those return assumptions, according to a survey of those funds by The Wall Street Journal.
 Corporate plans are not much better:
Corporate pension plans in many cases have been cutting expectations more quickly than public plans, but often they were starting from more-optimistic assumptions. Pension plans at companies in the Standard & Poor's 500 stock index have trimmed expected returns by one-half of a percentage point over the past five years, but their average return assumption is also 8%, according to the Analyst's Accounting Observer, a research firm.
To understand how realistic an 8% assumption is. Let's assume a 60% stock/40% bond asset mix. Let's be generous and say that the bond market can return 3%. That comes to a long-term equity return assumption of 11.3%, or an equity risk premium of 8.3%!

The actuaries add to the problem
David Merkel at Aleph Blog pointed out a bigger problem, i.e. the practice of the actuarial profession [emphasis added]:
When I was a young actuary, I was preparing to take the old Society of Actuaries test eight, which was the Investments exam. An older British actuary made a comment in one of the study notes that I had to think about several times before I understood it: “Risk premiums must be taken as earned, and never capitalized.
Sadly, the pension profession never got the memo on that idea. The setting of investment assumptions accepts as a rule that risk margins will be earned without fail. Therefore, when looking at a portfolio of common stocks in a pension trust, the actuary will assume that the equity premium will be earned over the long haul and build that into his discount rate assumptions and earned rate assumptions.
In other words, if an actuary is told that the equity risk premium is 8.3%. He will build it into his models and assume, come hell or high water, that stocks will earn 8.3% over bonds over the long run.

Unstable risk premiums
So far, this is all old hat for those who have been following the pension fund time bomb story. As interest rates have fallen, the net present value of pension liabilities rise, but return assumptions have fallen enough and a big yawning pension gap is the result. Moreover, the historical experience shows that an equity risk premium of 8.3% is, shall we say, a tad high.

Those estimates of equity risk premiums based on historical experience may not be valid. I was further shaken by the publication of an important academic paper entitled  New 'Risky' World Order: Unstable Risk Premiums: Implications for Practice by Aswath Damodaran of NYU. Here is the abstract:

Investors have to be offered risk premiums to invest in risky assets. These risk premiums take different forms in different asset markets: equity risk premiums (ERP) in stock markets, default spreads in bond markets and real asset premiums in other asset markets. These premiums have their roots in fundamentals and will vary as a function of uncertainty about the economy, the risk aversion of investors, information uncertainty and fear of catastrophe, among other factors. In practice, analysts in developed markets have generally looked backwards to estimate risk premiums, using historical data to arrive at their estimates. Implicitly, they assume that historical averages are not only precise but also that risk premiums are stable and revert back quickly to historical norms. In this paper, we present evidence that risk premiums in equity, bond and real asset markets are not only imprecise, but are also unstable and linked across markets. We present estimation approaches that are more in line with dynamic, shifting risk premiums. We argue that the resulting estimates can help use make more informed asset allocation and asset valuation judgments in portfolio management and better investment, financing and dividend decisions in corporate finance.
For newbies, let me explain the importance of this paper. Millions of business school and actuarial students have been taught Modern Portfolio Theory, or MPT. It is an investment theory that tries to maximize expected return for a given level of risk by carefully choosing the proportions of various assets in a portfolio. Harry Markowitz, who earned a Nobel Prize in Economics for the theory, modern portfolio theory introduced the idea of diversification as a tool to lower the risk of the entire portfolio without giving up high returns.

Modern Portfolio Theory: A choice between risk and return

Using MPT depends on knowing, or estimating, two numbers for each asset class – risk and return. In practice, most investors estimate returns for stocks using an equity risk premium using the following procedure: Historically, stocks have returned X% over bonds. Bond yields are currently Y%, therefore we can expect a return of X + Y for stocks.

The Damodaran paper calls that estimation procedure into serious question. He concluded that historical data studies indicate that estimates of asset class risk premiums don't really mean revert to long-term averages and they are unstable in the short and long run.

If an investor doesn’t have a good estimate for equity returns, then how can he build a portfolio? Damodaran suggests that investors should look to market based indicators. Consider the bond default spread as a signal of equity risk appetite and realized vs. estimated risk as another indicator, e.g. realized vs. implied volatility embedded in equity option markets around the world.

How high the pension deficits?
In the past year alone, we have seen studies indicating $1T deficits in public pension plans and other similar stories of pending disaster. Are US public plans "only" facing a $1T deficit?

Given the results indicated by the Damodaran paper, the scary part is we don't even know.


Andy Dong said...

It appears that a defined benefit fund deficit is out of the question. The real question is how big the deficit is.

The size of the deficit depends on relativities i.e. investment return relative to salary increases; actual investment returns relative to expected;

Some other considerations includes the increase in liability due to the increase in years of service (i.e. service cost).

It appears that none of which has gone in the favour of the pension providers.

Say a -2.5% p.a. investment performance in pension assets will create a 'hole' of 10% p.a. relative to an assumed 7.5%.

While salary increase has declines somewhat, it is not to the same extent as the underperformance of assets.

Adding on the power of compounding, $1 trillion deficit is not out of the question.

Unknown said...

Your post makes it sound like actuaries are unaware of the problem with their discount assumption. I assure you they are not (not that I am a pension actuary). Rather, they are often constrained to value pension obligations at the rate clients or the goverment or law forces them to.

Cam Hui, CFA said...


I am sure that many actuaries are aware of the problem. But if you do nothing or say nothing about it, then it becomes a I-was-just-following-orders Nuremberg defense when it all blows up.