After introducing the goal of the Scrrener / Score with the last post, let’s jump directly into the calculations:
Building the model – Step 1: Total Return on equity
Before starting to build the model, one has to define how to identify a stable, value adding business. One could use many variables, past earnings, free cashflows etc.
I personally view the long term developement of shareholders equity (including dividends and buybacks) as the best indicator if true “value” was created. This is of course not my original idea but essentially how Warren Buffet is reporting his yearly results at Berkshire.
In the times of modern IFRS accounting, one should mention that EPS (adjusted for dividends and buy backs) does not have to be equal the developement of equity from year to year. The real number usually is hidden on the second page of the P&L under “comprehensive income” which includes all kind of effects which get booked directly into equity (I.e. defined benefit pension plan valuation changes etc.)-
As comprehensive income is not very well maintained in the bloomberg historic database, I calculate “Total return on equity” with the following “basic” equation:
Total Return for (t) = Equity per share (t) – Equity per share (t-1) + Dividend paid (t)
A little bit trickier are capital increases and stock repurchases. As long as they are executed at book value, nothing changes on a per share basis. However if they are executed belw or above one will have direct impact on the per share equity and should be adjusted.
For the time being, I will stick with the “basic” formula and adjust on a case by case basis. In general I would assume that shares with significant capital increases or massive share repurchase programs don’t fit my “boring but sexy” category anyway.
The final computation than is relatively easy:
Total Return on Equity % (t) = Total return on Equity (t) / Equity (t-1)
If I do this for 10 years for example, I get TROE’s for those 10 years. With those 10 Yields, I can then calculate the average TROE in %
So the result of Step 1 is: Average Total return on Equity for period (t; t-x) where X can be 10 or 5 or any other period.
In theory the following applies: The higher the average TROE the better.
Step 2: Quantifying business volatility
The TROE average itself does not mean a lot. If we have for instance a company which has 20% TROE in one year and 0% in the next year, the average TROE will be the same as a company which shows 10% in both years.
As I am looking for “boring” companies, I have a high preference for stable TROEs. In contrast to classical CAPM, I will completely ignore stock market volatility (“Mr. Market”) and focus only on fundamental volatility.
So in order to account for fundamental volatility, I will use the standard deviation of the single period TROEs calculated above.
If we calculate the standard deviation of the TROEs, we can then in a next step calculate something similar to a Sharpe ratio which I would call the “fundamental sharpe ratio” for a company which is:
Fundamental Sharpe Ratio = Average TROE (t;t-x) / standard deviation TROE (t,t-x)
As an example, I have calculated those numbers for 4 random German DAX companies:
|
Avg 10 Y TROE |
10 Y std. dev |
Fundamental sharpe |
BASF SE |
12.6% |
8.1% |
1.5 |
ADIDAS AG |
17.7% |
12.1% |
1.5 |
BAYER AG-REG |
5.4% |
13.2% |
0.4 |
BAYERISCHE MOTOREN WERKE AG |
11.5% |
9.7% |
1.2 |
We see some significant differences here. On a first look, BASF and Adidas look very attractive, whereas Bayer looks rather bad. However, the fundamental Sharpe Ratio says nothing about how the volatility of the results is related to the valuation of the stock.
Step 3: Valuation model
In order to determine if a stock is attractively valued compared to the volatility of its business model, we have to build a simplified valuation model.
For the Boss Score I assume the following:
– the company will earn constantly its average past TROE going forward (without any growth) based on its current book value
So for Adidas for example: Adidas had a book value of 26.33 EUR and a average TROE of 17.7%, I will assume (26.33*17.7%)= 4.66 EUR per share for every year in the future, howver on a constant basis without any growth. This is somehow a similar perspective to the “EPV” from Greenwald.
In order to come up with a real value, I have to discount the future cashflows with somthing. And here comes the trick:
I use the CAPM formula to determine the discount rate, but intead of the market based “Beta” factor, I use the “fundamental Sharpe ratio” to determine the final equity ratio. I am not sure if this is mathematically correct in any way, but the intuitive idea is that the better the relationship between TROE and volatility of the TROE, the lower the discount rate.
As we all (hopefully) now, the CAPM calculates teh discount rates in the following way:
Discount rate (x) = risk free rate + Beta x (Equity market premium)
For the “Boss Score”, this changes into
Discount rate (x) = risk free rate + (1/fundamental sharpe ratio) x (Equity market premium)
With the discoutn rate we can then easily calculate the “intrinsic value” of any share under our model which is:
Intrinsic value = Assumed constant Profit / Model discount rate
As a final step, in order to have a direct relation to the current market Price, the “Boss Score” is then calculated the following way:
Boss Score = (Intrinsic Value / Market price) – 1
The final Boss Score can be interpreted the following way:
Boss Score > 0: Stock is based on the model undervalued
Boss Score < 0: Stock is based on the model overvalued, does not earn its cost of capital
The higher the Boss Score the better.
Before actually applying the screen to “live” data, it might make sense to define expectations in order to be able to “test” the outcomes of the “Boss Sore”. I would expect that (among others)
– companies with high leverage and / or cyclical business models will score relatively badly
– banks should score badly
– companies which a lot of one off write offs should score badly
– companies which earn below their cost of capital or which are in terminal decline should score badly
– “boring” comanies with stable returns and low P/B should score well
– the screen should bring up companies which might not show up in other screens
In the next post I will test the “Boss Score” first with the German DAX companies to see if the results make sense.