ACTIONABLE ADVICE FOR FINANCIAL ADVISORS: Newsletters and Commentaries Focused on Investment Strategy

Follow us on
 Facebook  Twitter  LinkedIn  RSS Feed

    Last 14 days

Most Popular Articles

Most Popular Commentaries

    Last 12 Months

Most Popular Articles

Most Popular Commentaries

More by the Same Author

The Persistence of Profits:

The Quality Conundrum, Part I
Knight Capital
By Matt Malgari
August 21, 2012

Display as PDF     Print    Email Article    Remind Me Later

Bookmark and Share

Kailash’s Key Concepts: Reversion, ROEs and Returns

·         Relentless ROEs?

·         Economic Gravity Reigns Supreme

·         ROEs Reconsidered

·         High & Low ROE Companies Experience Meaningful Reversion

Introduction: Relentless ROEs?

As avid consumers of a wide variety of investment material, we have to admit that certain firms have demonstrated such proficiency at lucid commentary they have become in-house favorites. In some cases, the writing has such a profound impact that we find ourselves entangled in visceral debates post publication, vexing our wonderful programmers with follow-up questions seeking further clarification or insight. In our opinion, Grantham Mayo (GMO) is one of these firms, kindly offering some of the most erudite market commentary in the industry. Their recent white paper, Profits for the Long Run: Affirming the Case for Quality by Chuck Joyce and Kimball Mayer, proved to be no exception.

The authors offer readers a wonderful roadmap that helps explain how fundamental constructs can effectively be brought to bear on the highly topical item of ―low-risk investing.‖  We strongly recommend the white paper as the work is loaded with exceptional insights into the intersection of volatility, fundamentals and returns. What created a collective brain stall in our group however, was the powerful analysis on the persistence of high and low ROE companies. Effectively GMO manages to show that companies with the highest (lowest) ROEs five years ago displayed persistently higher (lower) ROEs on a current basis.

GMO took its universe of stocks, broke it into quartiles based on ROEs and then plotted the five-year forward ROEs of the highest and lowest quartiles over time. Although our methodology differs slightly, we ran a similar exercise on our all but micro (ABM) universe of stocks (1) and found results that confirmed what GMO showed: those companies that were more (less) profitable five years ago based on ROE seem to remain in their respective economic stations into the future as shown in Fig. 1. Coming from avid believers in mean reversion, this chart carried an unsettling message (to us anyway) by questioning what we see as the laws of economic gravity.

Economic Gravity Reigns Supreme: A Brief Digression

In considering the work done by GMO, we realized that our beliefs about the resilience of profits may have been stilted by our obsessive interest in the trade-offs surrounding growth, a topic we discussed in our prior piece  The Siren Song of Growth. In an effort to reorient ourselves, we ran some data around the persistence of growth rates to at least ensure that our intellectual ―magnetic north‖ had not been destroyed by Messrs. Joyce and Mayer. Looking at our ABM universe of companies, we examined the percentage that maintained above-median growth over a ten-year period in sales, earnings and EBITDA and then compared this data to a coin toss. You can see in Fig. 2 that while firms do beat the coin-toss on sales (possibly through acquisitions or discounting), when it comes to profits, the level of persistence conforms almost perfectly to random probability.

To better illustrate the point, we show the pass rates for above-median earnings growth using the actual number of companies over time. Growth managers who promise clients whole portfolios of durable, high growth firms may want to study this chart carefully as the odds of identifying a firm that grows earnings above the median for two consecutive years is just ~13% (191 of 1,493) and only ~4% for four consecutive years (54 of 1,493).

We observed that even when we relax the criteria to allow firms to miss one or even two years of above-median growth within a given time series, the probability set still conforms to randomness, further supporting the argument for reversion due to economic gravity.

We understand that some investors may chafe at the concept of above median as a credible construct for identifying a ―true growth company, instead preferring firms that have a demonstrated track record of meaningfully superior outperformance. In order to address this, we created a more stringent universe, looking only at those firms that had managed to stay in the top quartile of growth for each of the prior three years and then at their ability to merely stay above the market’s median growth rate.  This would seem to offer growth aficionados the best of both worlds:  a high historical hurdle and a lower forward-looking hurdle. You can see from the data in Fig. 5 that using this criteria the companies continue to show modest outperformance at the sales and EBITDA levels but very closely approximate the coin-toss on an earnings basis.

What might interest readers more about the above chart is the actual number of companies that make up those statistics. In our view, the statistics prove misleading as they continue to speak the language of robust probabilities when, in fact, the sample size proves the exercise almost meaningless. Looking just at earnings in Fig. 6, we find that out of our universe of nearly 1,500 stocks, only 12 would theoretically make the top quartile growth requirement for three consecutive years based on historical average percentages. Looking forward at the pass rates for that small sample, we can see that trying to predict which companies will be able to generate above-median growth appears to be an absurd if not impossible task on an ex-ante basis.(2)

Very simply, we can see that when it comes to growth, the notion of sustainable and persistent above-median levels is not supported by the data.

ROEs Reconsidered

Having successfully re-established our footing in the world of growth, we promptly returned to the issue of persistence displayed in ROEs. Could the intersection of a simple balance sheet item with the income statement somehow point us to a group of uber-rational operators who, seeing flat-to-declining earnings, began reducing book value through repurchases or whittling it down on a relative basis through high dividend payouts? Or could the solution be something less complex? Looking again at our version of the GMO exhibit in Fig. 1, we realized the chart showed that the stocks with the highest (lowest) ROEs five years prior consistently had higher (lower) ROEs in the forward years, but it did not express any change in the two groups’ indigenous levels of profitability.

We mapped the rolling change in ROE over the five-year horizon which allowed us to see how the progression of book value and profitability was causing each group’s economic profile to develop. The results were dramatic, showing that despite displaying ROEs that were persistently higher (lower) five years in the future, the two groups were undergoing rapid mean reversion. You can see from Fig. 7 that companies in the top quintile went on to suffer nearly an 8% average absolute decline in their ROEs over the five years that followed, while those companies in the bottom quintile went on to enjoy nearly a 7% average absolute increase in their ROEs during the same span of time.

What seems equally remarkable about Fig. 7 is that the companies in the highest (lowest) quintile of ROEs at the outset always and without exceptionexperienced falling (rising) ROEs over the ensuing five years. In zero observed periods did the high ROE companies manage to see their ROEs actually rise— at best they barely stayed constant in one instance in the early 1970s. Similarly, the companies that were the lowest ROE franchises always experienced

increasing ROEs over the ensuing five years, with only one period of near-unchanged status in 2005.(3)

In an effort to better understand this phenomenon, we further broke the universe of ROEs into deciles and then mapped the median ROE of each cohort over the average five-year period.  What we found was a smooth rate of attrition in high performing companies and significant increases among those companies that started in the lowest decile of ROEs. Looking at the top decile for example, you can see that this group typically begins with a median ROE of nearly 30%, only to see that metric decay to less than an 18% ROE by the end of the period. At the other end of the spectrum, you can see that the worst starting decile of companies typically began with a median ROE of -9% and experienced over a 14% absolute swing to the positive by the end of the fifth year, resulting in a typical median ROE for the cohort of 5%.

Consistent with these findings, we found that, on average, only ~37% of the high ROE companies managed to remain in the top quintile of profitability by year five, while less than ~28% of the low ROE companies on average remained in their economic caste, with the rest moving up over the ensuing years. This, in our view, highlights the magnitude of economic mean reversion underway in both groups.

Realizing that on average, less than 40% of the quality companies managed to maintain their ROEs at a high enough level to qualify for top quintile status by year five, we decided to look at the starting and ending positions of the highest decile of ROE companies by ROE, margins, asset turnover and leverage to see what could be learned.  Below, Fig. 10 shows the typical distribution of the ROEs in our top decile of ROE companies after five years. By year five, just over 20% of the companies manage to stay in the top decile with the rest scattering across the various deciles.  Most notable about this chart, however, is the presence of a significant survivorship issue— nearly 30% of the companies have simply vanished from the sample by the end of a typical five-year period.(4)  We were unable to study this issue in more detail, but categorizing and quantifying the reason behind the disappearance would seem a critical next step as a preponderance of bankruptcies would cast a dark hue on these findings while chronic take-outs would obviously be quite positive. We believe the answer may lie somewhere in the middle with the results offering investors a push but admit we lack the numerical answer we would prefer.

We then decided to look at the evolution of margins for those companies that were in the top decile of ROEs in year zero and year five (Fig. 11). The dark blue bars below show the distribution by decile of margins for all the top decile ROE companies in year zero. We can see the results are highly monotonic with the majority of companies in the top decile of ROEs also having margins that put them in the top half of all margin deciles. That high ROEs and high margins would be correlated should surprise no one, but looking at how these companies fared by year five proves eye-opening to advocates of the concept of persistent quality. In year five, the representation is still very modestly tilted towards that top half of margin deciles, but you can clearly see that much of the asymmetry has evaporated with a significantly more linear representation across all margin deciles than in year zero. Quite simply, the profit structure of many of these top ROE firms has experienced meaningful deterioration by year five.

Looking at two of the other underpinnings of ROEs, we can see an additional explanation for the ROE attrition previously described by the declining asset turnover and falling leverage. Looking below at asset turnover, we find that in year zero, the top decile of ROE firms is heavily over-represented in the top half of the asset turnover deciles, with that representation becoming significantly more linear by year five.


Similarly, we find that these high ROE firms have a tendency to actually be over-represented in the category of leverage in year zero, and by year five they have undergone a meaningful reversal, with their representation switching to a bias of being over-represented in deciles with below-average leverage.


While we have not made a hard study of the underlying issue, we would also note that another facet that might merit further examination would be that the potential for significant industry bias may partially explain the observed persistence between the high and low groups. You can see from Fig. 14 that since 1967 the high ROE cohort has had a significant amount of over-representation in Industrials, Discretionary and Staples while the low ROE cohort has been overweight in areas typically associated with more volatile underpinnings such as Energy, Materials and IT.

Conclusion: High & Low ROE Companies Experience Meaningful Reversion

From this analysis we can make observations concerning the ability of firms to maintain high levels of profitability as defined by return on equity. Most importantly, while our findings  indicated that companies in the highest quintile or decile of ROEs had meaningfully higher ROEs five years later than those that started in the lowest quintile or decile, the intervening years extracted a significant economic toll with falling asset turnover, declining leverage and falling margins driving ROEs lower. Similarly, we found that companies in the lowest quintile or decile of ROEs had meaningfully lower ROEs five years later than those that started in the highest quintile or decile of ROEs but our analysis showed that they benefited from a significant improvement in their economic profile over a five-year horizon. In our next paper we will discuss the implications of these issues for investors with different time horizons and the potential for meaningful alpha discovery when intersected with robust value and technical constructs.


(1) All but micro (ABM) stocks include companies with a market cap greater than ~$500 million in today’s dollars, also any companies with a share price less than $3 per share and a market cap that is less than ~$5 billion in today’s dollars are excluded as well.

(2) Applying historic probabilities to today’s universe

(3) We recognize that some of this, particularly in the Low ROE sample, may be due to the survivorship issue which is discussed later in this paper

(4) In addition to bankruptcies and take-outs, a portion of the ―Exited ABM Universe‖ is due to insufficient data available to calculate the companies’ ROEs which could be due

to a multitude of additional factors


For more information on Kailash Concepts, details of Knight Capital’s products and services please contact Timothy Straus at


DISCLOSURE STATEMENT: This information and opinions in this report have been prepared by Knight Capital Americas LLC and may be distributed by Knight Capital Americas LLC or its affiliates (collectively, "Knight"). Any recommendation contained in this report may not be suitable for all investors. Moreover, although the information contained herein has been obtained from sources believed to be reliable, its accuracy and completeness cannot be guaranteed. Knight has no obligation to tell you when opinions or information in the report change.

Analyst Certification I, Matt Malgari, certify that (i) the views expressed in this report accurately reflect my personal views about the companies and securities that are the subject of this report and all other companies and securities mentioned in this research report that are covered by us, and (ii) no part of our compensation was, is, or will be, directly or indirectly, related to the specific recommendations or views expressed in this report.


This report is issued and approved for distribution in the United States by Knight Capital Americas LLC, member FINRA, NYSE and SIPC. This document is a product of Knight Capital Americas LLC. The parent company, Knight Capital Group, Inc. ("Knight") is comprised of several trading and related entities under common control. Such entities include Knight Capital Americas LLC, Knight Capital Europe Limited, Knight Execution & Clearing Services LLC and Knight Capital Asia Limited.


(c) Knight Capital



Display as PDF     Print    Email Article    Remind Me Later
Remember, if you have a question or comment, send it to .
Website by the Boston Web Company