The U.S. economy may well be sitting on a trillion-dollar time bomb, in the form of unrealistic pension return expectations. It affects defined benefit, defined contribution and cash balance plans, alike.
Is this an alarmist proposition? Alarming, yes. Alarmist, no. The average U.S. pension sponsor assumes nearly a 9% return for plan assets, in gauging the earnings impact and pension expense. This figure is actually higher than 9% if we focus only on corporate defined benefit plans. The average defined contribution investor (our employees!) likely expects even more. Let's assume the average plan actually earns 6% or 7%, not just next year, but as a long-term internal rate of return. The difference is two to three percentage points. Multiplied by $4.8 trillion, the aggregate assets for the 1,000 largest plans, this means return expectations are too high by around $100 billion to $150 billion. Given that corporate defined benefit plans are roughly one-third of the total, reported earnings might be too high by $30 billion to $50 billion.
But, 6% to 7% returns are ridiculous, no? No, but humor me on this for a moment; I'll come back to the return assumptions shortly.
Actuarial smoothing will take care of the $100 billion to $150 billion, yes? No. It's not a one-year problem, if long-term returns are two or three percentage points below expectations. It's $100 billion, per year. Smoothing can amortize last year's $100 billion, this year's $100 billion, and so forth; but, in time, all of these amortized slices grow to become a full $100 billion per year increase in the annual cost of pensions and, if we look only at the corporate side, as much as $50 billion more in annual pension expense. The magnitude of the problem is actually the net present value of $100 billion per annum.
Warren Buffett, in an interview in Fortune magazine last December, testified to this problem of using too high an assumption for return on assets. Mr. Buffett warned companies with ROA of more than 6.5% "that anyone choosing not to lower assumptions - CEOs, auditors and actuaries all - is risking litigation for misleading investors. And directors who don't question the optimism thus displayed simply won't be doing their job."
How does a $50 billion earnings impact compare with corporate earnings? Well, in 2000 and 2001, earnings for the S&P 1,500 totaled $524 billion and $252 billion, respectively, with price-earnings ratios of 25 and 46 times earnings. If return assumptions had been dropped by three percentage points, then reported earnings would have been $474 billion and $202 billion, respectively, and the year-end p/e ratios would have been 28 and 58 times earnings. Yes, $50 billion a year is significant.
Wait a minute. Am I not aggregating DB and DC? Yes, I am. Well, then, no problem. We've got a DC plan: No asset-liability problem to worry about. Wrong. Our employees have an asset-liability problem to worry about, one they don't even begin to understand. They've seen literature showing how well they'll do in retirement based on "conservative" 8% returns. Suppose they earn 6% (or less, given the higher expenses of DC plans and the less-than-professional asset allocation decisions made by the participants themselves). Suppose inflation erodes those returns to two to four percentage points above inflation. They'll just have to defer their retirement plans by enough years to make up the difference. And, they'll be darn grumpy during those extra years.
But, we've earned double-digit returns over the last five, 10, 15, 20 and 25 years. It's preposterous to expect 7% or less, no? Not really. Assuming an asset allocation mix of 70% equities and 30% bonds (yielding 6%), stocks would have to earn 10.7% to attain a 9% composite return. Stock returns have only four constituent parts:
* 1.5% dividend yield
* 2.5% consensus for future inflation
* 0% p/e expansion (dare we assume more?)
* X% real growth in dividends and earnings
This arithmetic suggests that, to get to a 7% return estimate, we need a mere 3% real growth in dividends and earnings. We can do far better than that, no? No. Historical real growth in dividends and earnings has been 1% to 1.5%.
How did we get into this bind? We have seen the largest bull market in history, by most measures. In the rise from mid-1982 to early 2000, the dividend yield on stocks fell from 6% to 1%, while the earnings yield (the reciprocal of the p/e ratio, or the dollars of earnings for each $100 invested) tumbled from 13% to 3%. This five-percentage-point drop in the dividend yield and 10-percentage-point drop in earnings yield must reduce our return expectations for stocks by no less than several percentage points.
So, do I suggest institutional investors eliminate their stock holdings or go to their CEOs with a recommendation to cut actuarial return assumptions to 6%? Do I suggest that actuaries and consultants counsel their clients to take these steps? No; I'm not in the business of recommending career suicide.
I do not recommend slashing the equity allocation: stocks still offer about the same long-term returns as bonds at today's yields. A move into bonds won't boost returns.
But, I do suggest:
* institutional investors hold a bit less stocks than their peers, so they aren't outliers on the aggressive (and dangerous) end of the spectrum.
* interesting returns are still available in many out-of-mainstream asset classes.
* institutional investors should err on the side of caution on actuarial assumptions, insofar as their top management permits them to do so.
* the actuarial and consulting community should alert their clients that they have a bridge to cross, to better align their return assumptions with what is realistic in the wake of an immense bull market. Acting sooner defuses the time bomb; acting later may not.
* investment managers should take account of pension assumptions in gauging the earnings quality of their holdings.
* dumping the problem on our employees, by switching to defined contribution plans, takes a headache away from current management and presents a severe migraine to the next generation of management.
But $100 billion is not $1 trillion. So, where did that wacky, hyperbolic number come from? The average duration of pension liabilities is roughly 15 years. So, right now, pensions are relying upon roughly a $100 billion overestimate of pension returns over an average of 15 years. That's $1.5 trillion. And that includes the fact that our employees are blissfully unaware of their own DC asset-liability problem, which is a huge part of the problem.
Oh, one last request. Please don't shoot the messenger!
Robert D. Arnott is managing partner of First Quadrant LP, Pasadena, Calif.