The Problem With Pensions II - A Blast From the Past

I wrote about pension fund return assumption problems a few days ago.  Then, while discussing another matter, a friend forwarded me this letter from Warren Buffett from December of 2001.  While reading it, I came to a part near the end where Buffett addresses the very point I was trying to make about buying long duration assets below your portfolio return assumption (I put this part in italics below).  Without further ado, I think this entire excerpt is worth reading (starts on page 8 of the link). Remember, this is from nearly 10 years ago:

"Let me show you another point about the herd mentality among pension funds--a point perhaps accentuated by a little self-interest on the part of those who oversee the funds. In the table below are four well-known companies--typical of many others I could have selected--and the expected returns on their pension fund assets that they used in calculating what charge (or credit) they should make annually for pensions.

Now, the higher the expectation rate that a company uses for pensions, the higher its reported earnings will be. That's just the way that pension accounting works--and I hope, for the sake of relative brevity, that you'll just take my word for it.

As the table below shows, expectations in 1975 were modest: 7% for Exxon, 6% for GE and GM, and under 5% for IBM. The oddity of these assumptions is that investors could then buy long-term government noncallable bonds that paid 8%. In other words, these companies could have loaded up their entire portfolio with 8% no-risk bonds, but they nevertheless used lower assumptions. By 1982, as you can see, they had moved up their assumptions a little bit, most to around 7%. But now you could buy long-term governments at 10.4%. You could in fact have locked in that yield for decades by buying so-called strips that guaranteed you a 10.4% reinvestment rate. In effect, your idiot nephew could have managed the fund and achieved returns far higher than the investment assumptions corporations were using.

Why in the world would a company be assuming 7.5% when it could get nearly 10.5% on government bonds? The answer is that rear-view mirror again: Investors who'd been
through the collapse of the Nifty Fifty in the early 1970s were still feeling the pain of the period and were out of date in their thinking about returns. They couldn't make the necessary mental adjustment.

Now fast-forward to 2000, when we had long-term governments at 5.4%. And what were the four companies saying in their 2000 annual reports about expectations for their pension funds? They were using assumptions of 9.5% and even 10%.

I'm a sporting type, and I would love to make a large bet with the chief financial officer of any one of those four companies, or with their actuaries or auditors, that over the next 15 years they will not average the rates they've postulated. Just look at the math, for one thing. A fund's portfolio is very likely to be one-third bonds, on which--assuming a conservative mix of issues with an appropriate range of maturities--the fund cannot today expect to earn much more than 5%. It's simple to see then that the fund will need to average more than 11% on the two-thirds that's in stocks to earn about 9.5% overall. That's a pretty heroic assumption, particularly given the substantial investment expenses that a typical fund incurs.

Heroic assumptions do wonders, however, for the bottom line. By embracing those expectation rates shown in the far right column, these companies report much higher earnings--much higher--than if they were using lower rates. And that's certainly not lost on the people who set the rates. The actuaries who have roles in this game know nothing special about future investment returns. What they do know, however, is that their clients desire rates that are high. And a happy client is a continuing client.

Are we talking big numbers here? Let's take a look at General Electric, the country's most valuable and most admired company. I'm a huge admirer myself. GE has run its pension fund extraordinarily well for decades, and its assumptions about returns are typical of the crowd. I use the company as an example simply because of its prominence.

If we may retreat to 1982 again, GE recorded a pension charge of $570 million. That amount cost the company 20% of its pretax earnings. Last year GE recorded a $1.74 billion pension credit. That was 9% of the company's pretax earnings. And it was 2 1/2 times the appliance division's profit of $684 million. A $1.74 billion credit is simply a lot of money. Reduce that pension assumption enough and you wipe out most of the credit.

GE's pension credit, and that of many another corporation, owes its existence to a rule of the Financial Accounting Standards Board that went into effect in 1987. From that point on, companies equipped with the right assumptions and getting the fund performance they needed could start crediting pension income to their income statements. Last year, according to Goldman Sachs, 35 companies in the S&P 500 got more than 10% of their earnings from pension credits, even as, in many cases, the value of their pension investments shrank."

-KD