Recommendations: 6
Jim has mentioned several times about how a good company performance
will show up in the sustainable ROE numbers.
I decided to do a test. It is not very scientific as I just use the 30
DOW constituents. I used these because of the easy availability of the
ROE numbers for the last 10 years (2002  2011) from Valueline.
ROE10 is the average ROE over the last 10 years where provided by
Valueline. Note that numbers less than 0 and > 100% are reported by
them as NMF. I manually calculated these. I think there were only 2
companies where I didn't have the full 10 year history of ROE but I
just averaged what I had.
R10, R15 and R20 are the annualized total returns over the last 10, 15
and 20 years.
I sorted descending by ROE10 and then grouped them into 5 quintiles of
6 companies each.
ROE10 R10 R15 R20
37.5% 10.7% 7.0% 12.2%
27.1% 7.6% 6.1% 10.8%
21.6% 11.3% 10.9% 13.1%
17.6% 5.0% 3.4% 10.7%
10.9% 7.9% 2.4% 7.9%
Make of it what you will but I done think it actually shows a very strong
correlation. For example, if I remove the bottom company (AA) from the
bottom quintile, the 5 remaining companies average:
ROE10 R10 R15 R20
11.6% 10.6% 3.6% 9.1%
Obviously this takes no account of the type of business or the valuation
levels at the start and end of the periods.
This probably needs a much broader study to be of any use but I thought I'd
post it anyway.
StevnFool

Recommendations: 19
Returns Jan 1997 to date for all 1700 companies covered by Value Line. Using current ROE, not 10 year average ROE, so not as good a metric as yours. Though these numbers change pretty slowly, this is based on checking to ensure that your firm in still in its bucket every month, no trading costs. Figures include dividends.
top 3% by ROE = 14.65%/year (about 50 stocks) next 7% by ROE = 12.77%/year next 10% by ROE = 10.70%/year next 20% by ROE = 11.52%/year next 20% by ROE = 11.07%/year next 20% by ROE = 10.53%/year next 10% by ROE = 10.25%/year next 5% by ROE = 5.33%/year Bottom 5% by ROE = 2.33%/year
If you take just the best 3% bucket above and buy the 5 stocks with the highest 5 year growth in book value per share the average return is 22%/year. (though at that point trading costs become an issue) You can see how it's easy to become a quant.
Jim

Recommendations: 1
If you take just the best 3% bucket above and buy the 5 stocks with the highest 5 year growth in book value per share the average return is 22%/year.(though at that point trading costs become an issue) You can see how it's easy to become a quant.
No need to do all that!
1. Limit universe to NYSE 2. Buy top 5 by ROA TTM, hold for one year.
From 1/2/99  10/5/12 I get 21.65% annualized ... for an annual hold. Turnover is also low, maybe 20% average.
Anyone who's dying to rush in, the picks are: BP Prudhoe Bay Royalty Trust, Permian Basin Royalty Trust, Sabine Royalty Trust, North European Oil Royalty Trust, San Juan Basin Royalty Trust. I sense a pattern, lol ...

Recommendations: 0
"If you take just the best 3% bucket above and buy the 5 stocks with the highest 5 year growth in book value per share the average return is 22%/year."
Jim
Are these the current five?
GT, ARLP, ESRX, HLF and CL
John

Recommendations: 2
No need to do all that! 1. Limit universe to NYSE 2. Buy top 5 by ROA TTM, hold for one year. From 1/2/99  10/5/12 I get 21.65% annualized ... for an annual hold.
FWIW Though I can't do that with NYSE stocks, for the 1700 covered by Value Line that strategy returns around 9.5%/year, assuming .4% cost per trade (obviously not more than once per year). Average of all 252 possible sets of trade dates Jan 1986 to date. The return for that whole set of stocks in the same interval is 10.6%/year.
Jim

Recommendations: 0
Though I can't do that with NYSE stocks, for the 1700 covered by Value Line that strategy returns around 9.5%/year, assuming .4% cost per trade (obviously not more than once per year). Average of all 252 possible sets of trade dates Jan 1986 to date. The return for that whole set of stocks in the same interval is 10.6%/year.
I used stockscreen123 for the NYSE stocks.
Using backtest.org, I get 12% per year versus 2% for the S&P using "peta top 5" for 19992011. It's possible it was a disaster before 1999, don't know. Sounds like it from your 1986 test.

Recommendations: 0
Though I can't do that with NYSE stocks, for the 1700 covered by Value Line that strategy returns around 9.5%/year, assuming .4% cost per trade (obviously not more than once per year). Average of all 252 possible sets of trade dates Jan 1986 to date. The return for that whole set of stocks in the same interval is 10.6%/year.
Okay, using backtest.org:
1. dpc < 50 (debt as % of capital less than 50%) 2. enw top 5 (top 5 by "earned on net worth")
I get 21% for an annual hold vs. 10% for the S&P going back to 1986. That does include a rather insane +557% in 2009 though!
Too bad one can't see the actual picks. Because this screen looks useless prior to 1999 and then takes off like a rocket the last 13 years. Looks pretty unstable I must say ...

Recommendations: 0
If you take just the best 3% bucket above and buy the 5 stocks with the highest 5 year growth in book value per share the average return is 22%/year. (though at that point trading costs become an issue) You can see how it's easy to become a quant.
I tried:
1. enw top 50 2. bg5 top 5
which I think is what you suggested. Results are indeed killer for the one month hold from 1997 onward, but something bad was happening prior to that ... underperforms the S&P from 1986 to 1996 .... 8% vs. 15%.

Recommendations: 0
Jim,
My original results:
ROE10 R10 R15 R20
37.5% 10.7% 7.0% 12.2%
27.1% 7.6% 6.1% 10.8%
21.6% 11.3% 10.9% 13.1%
17.6% 5.0% 3.4% 10.7%
10.9% 7.9% 2.4% 7.9%
Your results:
top 3% by ROE = 14.65%/year (about 50 stocks)
next 7% by ROE = 12.77%/year
next 10% by ROE = 10.70%/year
next 20% by ROE = 11.52%/year
next 20% by ROE = 11.07%/year
next 20% by ROE = 10.53%/year
next 10% by ROE = 10.25%/year
next 5% by ROE = 5.33%/year
Bottom 5% by ROE = 2.33%/year
I think what is notable about both sets of results is poor performance
at the bottom of the pile. This makes me thing that if you were to use
ROE as one factor in a set of Value Investing stock selection rules, it
may make sense to just uliminate the poor ROE companies.
I'm thinking somthing like:
1. ROE average > X
2. No strong evidence of significant downward trend in ROE.
3. Leverage check (maybe something like average earnings for last 3 years > 20% debt)
4. Low P/E
5. etc.
So if one were to look at your data, what would the value of X be?
Perhaps the value that eliminates the bottom 10% by ROE? Do you know
what this value was on average in your test?
If I look at the 15 year CAGR of the test I did (closest period to your test):
Bottom 10% by ROE had a CAGR of 1.5%
Top 90% by ROE had a CAGR of 6.5%
The ROE cutoff was between 10.1% and 10.6% average over the last 10 years.
Perhaps the 20 year numbers are better:
Bottom 10% by ROE had a CAGR of 7.1%
Top 90% by ROE had a CAGR of 11.4%
The ROE cutoff is the same.
If I do some overtuning and set the ROE cutoff to between 16.1% and
16.4%, it eliminates the bottom 7 of the 30 DOW companies and I get the
following:
15 year CAGR:
Bottom 7 by ROE had a CAGR of 1.3%
Top 23 by ROE had a CAGR of 7.4%
20 year CAGR:
Bottom 7 by ROE had a CAGR of 7.1%
Top 23 by ROE had a CAGR of 12.1%
Perhaps a nice round number would be as good as any. ROE on average > 15% and no strong indication of a significant downward trend of ROE.
StevnFool

Recommendations: 4
Results are indeed killer for the one month hold from 1997 onward, but something bad was happening prior to that ... underperforms the S&P from 1986 to 1996 .... 8% vs. 15%.
The "enw" field at backtest.org doesn't mean the same thing 19861996 as it has since then. Value Line changed their lineup of fields in December 1996 and that was one of the big changes. "enw" stands for "earnings on net worth". It is derived from the field called "% Return Net Worth" prior to 1997 and from the field called "% Earned Net Worth" thereafter, which sounds plausible (which is why it was done) but it may in fact match the 1997andlater field called "% Earned Common Equity".
Maybe the discontinuity in results is a result of the known discontinuity of the data going into it, maybe not.
The main lesson is that high ROE (preferably without undo leverage) makes good sense from the point of understanding how a good business makes its money from a moat, and is generally confirmed by quantitative studies across all firms. Though it isn't enough to make you an investing genius and it might let through some duds, on average a nice steady high ROE is a very good sign. A monkey with a dartboard full of high ROE firms will almost certainly outperform a monkey armed with a dartboard of all firms.
As an aside, the more a firm is growing the more important the ROE. A growing firm with an uninteresting ROE is worth nothing extra due to its growth rate, and even a slowly growing high ROE firm is potentially worth a lot. The PEG ratio is misleading, as it conflates growth with worthwhile growth. Rather than growth/PE it should be something more like (1+ growth rate above inflation times amount by which ROE exceeds the long run median) times (the amount by which the cyclically adjusted earnings yield exceeds Siegel's constant of 6.5%). As a fun exercise, try this calculation for both Amazon and Walmart.
Jim

Recommendations: 0
So if one were to look at your data, what would the value of X be? Perhaps the value that eliminates the bottom 10% by ROE? Do you know what this value was on average in your test?
Looks like it's generally the negative numbers, which are more common that I suspected. Fraction of population of 1700 Value Line stocks by ROE since Jan 1997.
10.2% negative 4.6% zero or unpopulated 1.1% from 0 to 1 1.4% from 1 to 2 1.6% from 2 to 3 1.9% from 3 to 4 2.3% from 4 to 5 2.6% from 5 to 6 3.3% from 6 to 7 3.5% from 7 to 8 4.1% from 8 to 9 4.6% from 9 to 10 4.8% from 10 to 11 4.9% from 11 to 12 5.1% from 12 to 13 4.6% from 13 to 14 4.2% from 14 to 15 4.0% from 15 to 16 3.4% from 16 to 17 3.0% from 17 to 18 2.9% from 18 to 19 2.5% from 19 to 20 2.0% from 20 to 21 1.8% from 21 to 22 1.7% from 22 to 23 1.5% from 23 to 24 1.2% from 24 to 25 1.0% from 25 to 26 0.9% from 26 to 27 0.8% from 27 to 28 0.7% from 28 to 29 0.6% from 29 to 30 0.6% from 30 to 31 0.4% from 31 to 32 0.4% from 32 to 33 0.4% from 33 to 34 0.4% from 34 to 35 0.3% from 35 to 36 0.3% from 36 to 37 0.2% from 37 to 38 0.2% from 38 to 39 0.2% from 39 to 40 3.9% over 40 To recap the returns Negative ROE = 6.11%/year Zero or unknown ROE = 10.56%/year Lower half of positive ROE = 11.52%/year Top half of positive ROE = 12.77%/year
Bad negative numbers aren't worse than small negative numbers. The lowest returns are simply the subset "<0". If you wanted the highest return 30 stock portfolio given only the ROE field you'd choose the 30 stocks with ROE closest to 32%, which gives 15.6%/year before trading costs, perhaps a bit of a lucky outlier. I guess there are a few absurdly high ROE values that are a bad omen? A monthlyconstituted portfolio of all stocks 35% ROE or better returned 12.6%.
Jim

Recommendations: 0
To recap the returns Negative ROE = 6.11%/year Zero or unknown ROE = 10.56%/year Lower half of positive ROE = 11.52%/year Top half of positive ROE = 12.77%/year
Just for reference, I believe the cutoff between the upper and lower have of positive ROE is an ROE somewhere between 13% and 14%.
Requiring an ROE of 15% would then seem to suggest that you will in the better half of companies making money and if you require a sustainable ROE > 15% I think you would capture an even better subset.
Going back to another point you have made in the past about requiring ROE to be greater than the cost of capital. Is there a generally accepted figure for the weighted average cost of capital over time?
StevnFool

Recommendations: 2
Going back to another point you have made in the past about requiring ROE to be greater than the cost of capital. Is there a generally accepted figure for the weighted average cost of capital over time?
This is a very contentious subject. Translation: I think everybody else is an idiot, meaning I'm probably the one that is off base. There are standard formulae for calculating this but they seem utterly absurd to me. Berkshire's cost of capital has nothing to do with its share price or volatility thereof.
The starting point is fairly simple. If your company valuation is low you use debt and the cost is the interest rate you pay, and if your valuation is high then you use equity and the cost is some estimate of the forward+cyclically adjusted earnings yield.
I look at what capital the firm really is deploying (if any) or might soon deploy, and where it's really coming from. If they aren't deploying any new capital, then the answer is "N/A". If they aren't raising any equity then the equity isn't an input. If they use debt, that's the only thing I look at. It's only if they are using internally generated capital that it gets tricky; if you have a productive cash cow division bought many years ago at a basis that is a longago sunk cost throwing off cash, what's the cost basis on that money really? Berkshire's WACC might be well under 2% on this view, the only notable exception being the equity issued for part of BNSF which was quite pricey. Equity was issued at about 1.18x book (yes, that's all). If you assume that the valuation stays the same then both book and price are rising on trend at around 9%/year so you could think of that as the cost. Maybe bump that up a couple/few points based on normalized valuation levels. Actual borrowings are in the 13% range these days, and float has a negative cost.
About the only way in which I agree with standard practice is that WACC is a very company specific thing, so there isn't any good rule of thumb unless you're talking about the marketwide average cost. Globally speaking across all equities the average cost of capital is axiomatically equal to the average return on capital in equities averaged over time. That should provide a good sanity check, since the capweighted average of the WACCs you estimate should come close to Siegel's constant for the US. That's again contrary to theory, as most MBAs will come up with much higher WACCs than 6.5% above inflation.
Jim

Recommendations: 0
That should provide a good sanity check, since the capweighted average of the WACCs you estimate should come close to Siegel's constant for the US. That's again contrary to theory, as most MBAs will come up with much higher WACCs than 6.5% above inflation.
Jim
US inflation in the last 100 years has averaged a little under 3.5%
6.5% + 3.5% = 10%.
This would suggest then that a company deploying capital is destroying value on average if their ROE is less than 10%.
Most companies deploy capital as it is rare to see a company pay out 100% of earnings.
To give a margin to cover for accounting quirks and a general margin of safety, it would seem that requiring a sustainable ROE > 15% would be a sensible target.
StevnFool

Recommendations: 2
US inflation in the last 100 years has averaged a little under 3.5% 6.5% + 3.5% = 10%. This would suggest then that a company deploying capital is destroying value on average if their ROE is less than 10%.
No. You don't have to do an inflation adjustment on an instantaneous earningsrelated rate. A firm with 10% ROE is likely to have 10% ROE the next year with smaller dollars. With a shrunken yardstick you get bigger R and proportionally bigger E as well: the same ratio.
For example, an earnings yield doesn't have to be inflation adjusted. Other things being equal the earnings yield won't change with inflation. Most firms will on average be able to increase prices to compensate for their increased costs. This is only an average—some firms will be hurt by inflation and others will benefit, but it's the general rule. Also, occasional very high inflation will break the economy which is a separate problem. But in all but exceptional circumstances inflation can be ignored when looking at earnings yields (and ROE). It's borne out empirically too. The best fit model of rollingdecade historical US nominal earnings growth includes the term 0.94 times inflation.
ROE over 15% might be a sensible target because it will limit you to the better businesses, but it's not needed due to an inflation expectation. On average since 1997 cutoff that would correspond to 38% of the average 1584 companies covered by Value Line with the ROE field populated. You could probably even crank it up a bit and not miss many great firms. 20% would give you 20% of firms. Let's call it the 20:20 rule!
Jim

Recommendations: 1
You don't have to do an inflation adjustment on an instantaneous earningsrelated rate. A firm with 10% ROE is likely to have 10% ROE the next year with smaller dollars. With a shrunken yardstick you get bigger R and proportionally bigger E as well: the same ratio.
I don't think this is necessarily true. E could increase much more slowly (proportionally) due to inflation than R. Let's take an example. Beginning assets = $100 monetary, $200 nonmonetary (book value with no depreciation). Liabilities = 0. Beginning equity = $300. No inflation. Earnings = $100. Ending equity = $400 (all earnings go to retained earnings.) If we calculate ROE on the average equity, it comes to 2/7 ($100/$350) = 28.57%. On the beginning equity, it is $100/$300 = 1/3.
Now assume 10% inflation (not hyperinflation which will trigger asset reevaluation and change in equity.) Beginning equity = $300. Earnings = $110. Ending equity = $410. Average equity = $355. ROE = 110/355 = 30.985% ~ 31%. On beginning equity, ROE = $110/$300 > 1/3.
If you have to restate assets to $100 (monetary) + $220 (nonmonetary) = $320 then your ending equity will be $320 + $110 = $430. Average = $365. ROE = 30.1369%.
Conclusion: IF all net assets (assets  liabilities) are nonmonetary AND you calculate ROE on beginning equity AND there is no depreciation, THEN your statement will hold true.
In our case, assume $300 assets = $300 equity at the beginning, and no cash, and no depreciation charges; then at the end both would be valued at $330, not counting the recently added $110 in earnings, and ROE will remain at 1/3 = $110/$330.
Otherwise, ROE will increase with inflation.

Recommendations: 0
I don't think this is necessarily true....
I agree, it's not necessarily true for any given company. A company with lots of debt will do better with inflation, one with net cash will do worse, and so on. But on average across the economy inflation is a passthrough, except at times that it's so high that the economy just plain breaks. It's true for the "average" common or garden company.
Certainly to a first (and probably second) approximation you definitely don't need to add inflation rate to your desired ROE. Same as earnings yield, or as a rule of thumb anything that's an instantaneous ratio between two things rather than the ratio of prices of one thing at different times.
This is one of the main reasons it's not sensible to compare the earnings yield on equities (which is almost completely a real rate) to the long bond rate (which is definitely a nominal rate). When the 10 or 30 year long bond rate minus your expectation of inflation in the next 10 or 30 years exceeds the cyclically adjusted earnings yield of broad equities as was the case in the late 1990s, then bonds make sense. I bought a lot of long German bunds yielding around 6.5% back then.
Jim

Recommendations: 1
No. You don't have to do an inflation adjustment on an instantaneous earningsrelated rate
Jim,
The thinking behind my inflation adjustment was to related to WACC.
The ideal being.
Target ROE = WACC + some margin.
For WACC, I was reading from your post # 194745 where you said.
That should provide a good sanity check, since the capweighted average of the WACCs you estimate should come close to Siegel's constant for the US. That's again contrary to theory, as most MBAs will come up with much higher WACCs than 6.5% above inflation.
I inferred from the above that: * when you suggested Siegel's constant as a sanity check * and Siegel's constant happens to be 6.5% * and you then made reference to WACC's 6.5% above inflation
that you meant a sanity check on WACC would be Siegels constant + inflation.
This led me to the WACC of 10% and a target ROE of 15% with a 5% margin.
Perhaps I misunderstood.
StevnFool

Recommendations: 1
I inferred from the above that: * when you suggested Siegel's constant as a sanity check * and Siegel's constant happens to be 6.5% * and you then made reference to WACC's 6.5% above inflation
that you meant a sanity check on WACC would be Siegels constant + inflation.
Ahh, interesting viewpoint. I am hoist by my own petard.
The thing about WACC is that, like intrinsic value, it's really important and hard to calculate. Some people have come up with very specific ways of calculating it but I don't buy those much. So, though I know it's important, I don't really know how to estimate it well. I have never spent a lot of time estimating it and thus don't know my own mind on this. In short, I've never really thought about the issue of whether WACC is best thought of as a nominal or a real rate.
On general principles the only WACC that is useful is both forward looking and cylclically adjusted. The rate this year isn't really the one that counts.
The main starting place for me is debt, which is obvious. When inflation is high, corporates will be paying higher nominal interest rates. Without a doubt, the cost of debt for a company is best thought of as a margin above inflation. So the first linen of thinking is this: in an inflationary year the nominal cost of capital will be higher therefore the nominal return on capital had better be higher to match, so a constant ROA above a rising WACC will be a bad thing. But maybe this isn't an issue: capital (debt, anyway) is constantly renewed, so the cost of capital rises and falls in nominal terms over time with inflation rates, though with a little lag. So the gap between them (which drives profit) will stay relatively constant in the natural course of things.
I dunno. I'm going to have to think about this some more. It seems to me the only cost of capital that matters is the real cost per year averaged over the next 515 individual years. (not a rate of change between two times, but the time weighted average of many individual rates, each of which is a real rate in a oneyear interval). But perversely it seems to make sense to me to compare this to the current nominal ROA because the ROA will generally not change with inflation. To a first approximation, nominal revenues rise with nominal costs and the gap stays the same percentage. So I've led myself to the odd position of thinking that it actually makes sense to compare a nominal rate to a real rate.
Jim

Recommendations: 0
The thing about WACC is that, like intrinsic value, it's really important and hard to calculate.
I think that this has been an interesting thread. While we may not be able to calculate some of these numbers accurately, I think the posts in this thread have led me to the following conclusion.
If you are using historical ROE (over say the last 10 years) as one (of several) measures to assess a company, then looking for an average ROE greater than some cutoff in the region of 15% to 20% seems like a sensible number.
StevnFool


