Post #3222 by knighttof3 on the Berkshire Hathaway board

Stocks A to Z / Stocks B / Berkshire Hathaway (BRK.A) ❤

Unthreaded | Threaded | Whole Thread (66)

Post New | Post Reply | Report Post | Recommend It!

No. of Recommendations: 5

A simple "thank you" is grossly inadequate for this gem of a reply to my somewhat impudent question, but thank you!

Since there is no economic reason why high ROE firms should do better than low ROE firms (*) in these days of differing buyback policies, I would like to know why I should repose any faith that this screen will continue working.
(*) Meaning the low RoE firms are not reducing their equity - the denominator - as fast as the high ones, but their actual business and earnings and earnings growth (the numerator) could be better than the high ROE firms.

In fact there *is* a very good reason that high ROE companies should be above-average long run performers.

ROE is hard to measure reliably...there are lots of things that can give you a somewhat misleading result. The effect of buybacks is one of them, but that effect is so tiny that it can be ignored.
(if a firm buys back 20% of all of its shares, paying an average of 2 tiles book, both big numbers, it drops book per share "artificially" by only 11%--we have bigger flaws to worry about).

The reason ROE works, though, is that on book value per share is on average a "good enough" proxy for the replacement value of the firm's assets.
Not good enough to be entirely reliable in any specific case, but good enough to be a statistically useful predictor on average.

Imagine a widget factory with a $10m factory making $2m/year in net profit, 20% ROE. Assume $10m is both book value and the cost of duplicating the factory.
A competitor will be highly motivated to raise money at less than 20% cost of capital, build a duplicate factory, and undercut the first factory: they'll still make an above-average return.
This keeps going until it is no longer profitable for a new competitor to enter the fray, which is the point that the return on capital for the next would-be competitor is equal to the cost of capital for them all.
(axiomatically, the global average return on capital is equal to the global average cost of capital)

Let us continue to assume that the book value of our firm is a reasonable proxy for the cost to duplicate its assets.
But let's say the firm is making 20% ROE year in and year out for a long time.
What does that tell us? That, for whatever reason, competitors can't (or won't) set up in competition with them successfully: the company has an economic moat of some type.
This is, as we know, a very good business to be in. Sustainably high profitability that isn't being eaten away by competitors.
Some subset of those firms will also have the ability to deploy new capital into their business at the same high rate of return on assets, the very best of the best in terms of business economics.
If you're not buying equal amounts of all the companies out there, you at least want to make sure you are owning these great ones.

So, though ROE is not 100% reliable in the case of any specific firm, mainly because of quirks in the way assets are booked, it is almost always true that a great long term business (a business with moat giving them lasting high profitability) is going to have a high ROE.

So, why does the screen work? The businesses it picks are higher than average quality.
It will pick some duds that have undeservedly high numerical ROE, and presumably does badly on those. Most obviously, companies that are over-leveraged.
But it catches a very large fraction of the very best companies (it's hard to be one of the best without a high ROE), and eliminates almost all the firms with the ho-hum poor business economics: poor returns on capital employed.
In short, not all firms with high ostensible ROE numbers are good investments by a long shot, but almost all good long run investments have good ROE figures over time.
This filtering, though certainly flawed in any single instance, gives a statistical edge that seems to be big enough to be worth exploiting.

The temptation is to make the screen more and more sophisticated--skip those that are overleveraged, allow those that have negative book value, adjust for dud goodwill, and so on.
But that way lies the slippery slope. As the saying goes, if you torture a data set long enough, it will confess to anything.
But if you can buy 50 or 100 companies that will beat a dartboard on average over time by a percent or two, that's probably good enough, and you aren't in danger of fooling yourself too much.
The real world results will still be worse than the backtest--that's the nature of things. But it should still give you a good chance of outperformance, and if it's all just statistical noise you shouldn't do meaningfully *worse* than average over the long run.

You have posted an enormous number of such back-tested screens. Do you have any data for, say, how the screens you posted 8 years ago (or any such longish period of time) have done since posting them?

Good question.
I'll discuss only the prominent failures, since those may have the most useful insights to offer. They're certainly the ones prominent in my mind : )

The most interesting failure is a screen called YLDEARNYEAR, invented in 2003. (not by me)
It outperformed the market through thick and thin for a dozen years after it was published, the sort of thing that gives a lot of confidence that there is something "real" going on.
Then it promptly turned around and started underperforming the market for the next 5-6 years.
So, what's going on?
That screen seeks firms with both high earnings yields and high dividend yields.
That population of stocks itself seems to go in and out of fashion, not necessarily on a cycle matching bull and bear markets.
With hindsight, it appears that the screen magnifies (very successfully) the degree to which that dividend population is in vogue.
Some numbers:
In the first 11 years after the screen was published, the set of all moderately profitable firms with dividends outperformed the broad market by about 4.4%/year.
(for that I looked at all stocks in the Value Line 1700 database with P/E < 33 and any dividend yield, equally weighted)
Consequently, that was a very fertile hunting ground for any stock screen:
The YLDEARNYEAR screen outperformed the market by about 14.1%/year in this stretch. (5 fresh stocks bought each month, each held two months, for a 10 stock portfolio).
But in the next 6.1 years, the set of all profitable dividend earners underperformed the S&P by -5.6%/year, and the screen underperformed by -17.1%/year.
Dividend payers have come back into fashion again, outperforming the S&P by +5.4%/year in the last 3.25 years.
And, as you might now expect, the screen has come back into its own: it has outperformed the S&P by 17.7%/year in that same period.
During the stretch that dividend payers were out of fashion, almost all humans using this screen would have thrown in the towel.
Yet, had one stuck with it, it has still outperformed overall since its invention 20 years ago.
S&P returned 10.1%/year, the set of passably profitable dividend payers returned 11.5%/year, and the screen returned 14.3%/year.
One could potentially check to see if dividend earners are in or out of fashion lately, and use the screen only when the omens are good--tie situation doesn't change often. But it's hard to know in advance how that attempt would do long term.

The biggest total failure screen was one I created and suggested that simply looked at very large firms with great balance sheets and low P/E ratios, nicknamed "BlueCheaps".
This is one that I would definitely sweep under the carpet if I were still in the money management business : )
The original post that suggested it http://www.datahelper.com/mi/search.phtml?nofool=y...
In the backtest 1986-2009 inclusive it indicated performance 9.4%/year better than the S&P 500, with 5 stocks each month after trading costs.
But then, alas, among the large and financially sound firms, the ones without many profits became the ones that were the better performers. Much, much better.
The same screen 2010 to date underperformed the S&P 500 by -6.9%/year.
You were still holding big profitable stable firms making a lot of money, so individual picks didn't tank, and the returns of ~6%/year were relatively steady...you just didn't do nearly as well as a monkey with a dartboard would have. A portfolio of genteel mediocrity.
Lessons? One very big flaw of this and many other screens is that its results were based on a portfolio of only 5 stocks, sometimes 10.
There is so much randomness in having such a concentrated portfolio that the odds of a good backtest being good merely by chance become very high.
These days, I'm much more interested in a screen that indicates outperformance of (say) 2-4%/year with 30-60 stocks than one that teases you with seeming outperformance of 10%/year on only 4-10 stocks then falls flat.

Jim

Post New | Post Reply | Report Post | Recommend It!

Print the post

Unthreaded | Threaded | Whole Thread (66)

Prev | Next

Announcements

Berkshire Hathaway FAQ

Contact Shrewd'm
Contact the developer of these message boards.