The lunacy of Dow 116k (or, why all our pension funds are screwed)

I'm going to try to make this one as quick as I can, because I could really rant about it for days. I've talked here before about how the only market analysts who get publicity are the ones making outrageous calls (Dow 3,000! Dow 25,000 by August!), and how this has everything to do with the severely skewed incentive structure surrounding punditry.

Now, with the market nearing multi-year highs, we're in prime territory for these kinds of silly calls, on both sides of the bear/bull debate. This one just happens to be the most absurd, and for those of you who follow me on Twitter, you may already have seen my teaser of this issue. From MarketWatch's Chuck Jaffe:

My favorite market forecast of all time came in the fall of 1995, when mutual fund pioneer Bill Berger came to Boston and predicted the Dow Jones Industrial Average would rise to 116,200.  

He didn’t think it would happen overnight. In fact, the 70-something founder of the Berger Funds figured the market would need until the year 2040 to reach it. (He wryly suggested that if he was proved wrong, people come find him to discuss it; sadly, he died a few years later.) 

Of course, Dow 116,200 would be a far more ridiculous thought than even, say, Dow 36,000, the title of a popular-but-wrong-headed book from the bubble days of the late 1990s, were it not for one thing: As the Dow touched 13,000 this week (and the Nasdaq Composite flirted with 3,000), it’s actually stunningly close to being on track toward making Berger’s prediction come true.

Stunningly close, eh? Let's unpack this one for a minute. Jaffe goes on to extrapolate the annual return over the last 17 years (a completely arbitrary time period that is only relevant based on the exact timing of the prediction he's attempting to analyze), assuming that it will persist at exactly this same rate (7.2% compounded--he actually uses 6.75% because he's starting in June and not March, whatever) over the next thirty years. Using that compounding, the Dow will be well above 100k and on its way to 200k by 2046.

The problem is, this compounded annual rate is absolutely an accident of the time period he's using. For the first five years following Berger's Dow 116k call, the market went on an unprecedented vertical run-up, fueled by an influx of Baby Boomer cash and the dot-com bubble frenzy. The compounded annual return between 1995 and 2000 was a staggering 21%, leading the stock market to more than double over those five years.

Unfortunately for Jaffe (and Berger), it's made basically zero headway since then. In the 12 years following that once-in-a-lifetime bull market, the market has simply lurched from one asset bubble to the next, treading water for a long enough time that just about any index with a continuous compounding assumption (like, um, a pension fund) now finds itself hopelessly behind the curve. The only reason that the Dow 116k call looks like anything resembling a reasonable prediction is the specific and arbitrary points that our author has selected for his analysis.

To show what I mean, let's use a couple of other equally arbitrary starting and ending points for our analysis, to see what kind of wildly different conclusions we can reach (we'll list the author's prediction first, for comparison's sake). You might want to click on the table to get a clearer view.

I'm a particular fan of the "Dow 26 Million" call that you could make if you extrapolated from the March 2009 bottom until today--that's a good one. And the "Dow Zero" call at the bottom of the table is a bit harsh--the Dow would actually hit an asymptotically small level of nine cents by 2046 under that projection, so you wouldn't be totally wiped out.

The point is, any time you try to extrapolate too far based on arbitrary data sets, you're completely a slave to your small sample. Yes, I recognize that the 1995 to 2012 time horizon is the widest range of the bunch, and therefore your biggest and hypothetically least-random sample. But is the 12 years since 2000, where the market has eked out only a 1.9% annual return, meant to be ignored simply because the 5 years prior were so good? If 12 years isn't a significant sample, then why is 17? And if those 5 years from 1995 to 2000 are so important, why aren't the 5 years from 2007 to 2012--with their meager 1.43% annual gain, even after this year's ramp job--equally important?

You get my point by now, so I'll back off. The problem is, every single pension fund in the country uses exactly this kind of infinite compounding analysis in order to claim its solvency. These funds generally assume a consistent return of 7% to 8% (sometimes more) into perpetuity, in order to have enough funds to pay out their accumulated liabilities.

In other words, if the Dow isn't at 116,200 by 2046, these funds are gonna have one hell of a problem on their hands. So, you can laugh at this prediction if you want... but in doing so, recognize that your pension (if you've got one) relies on just this kind of mathematical lunacy. Gooooooooo Dow!!

[MarketWatch]