THE JOURNAL OF
ETF s , ETPs & Indexing
WINTER 2017 VOLUME 8 NUMBER 3 | www.IIJII.com
The Voices of Influence | iijournals.com
Leadership Matters: Crafting a Smart Beta Portfolio with a Founder-CEO Twist J OEL M. S HULMAN
Leadership Matters: Crafting a Smart Beta Portfolio with a Founder-CEO Twist
J OEL M. S HULMAN
I n recent years, considerable attention has f locked to smart beta strategies with investment managers applying reweighting schemes to peer bench- marks. 1 In essence, managers identify per- ceived f laws within an existing benchmark and craft a makeover solution offering a dif- ferent twist with hopes of capitalizing on embedded opportunities. Although billed as passive plays, these strategies comprise an active component cloaked in the inner workings of quant redesigns. 2 Savvy man- agers develop smart portfolios by selecting factors from a plethora of options, including market capitalization, volatility, and price- to-earnings ratios, (P/E) among others. Management attributes, however, despite being important criteria for active fund man- agers in selecting stocks, rarely appear as an exchange-traded fund (ETF) or smart beta option. 3 Although the absence or rarity of a management-based ETF does not, in itself, imply an investment dilemma demanding a solution, it does raise the broader, simpler question of why leadership does not matter when running a large, publicly traded com- pany. 4 In this article, we introduce a proxy factor for quality of management by cre- ating and rigorously testing a founder-CEO index relative to a comparative benchmark that includes many of the same holdings. 5 The spirit behind this selection stems from developing academic research, emerging
fund managers who specialize in this exper- tise, and the investment logic supporting the hypothesis that if leadership does indeed matter, it would most likely occur with a founder CEO, who has a greater likelihood of control, economic and personal incentives, ability to exert vision on governance issues (stewardship) and an extended (unimpeded) job tenure to see his or her vision through to completion. 6 Following the logic further, we note that if an index of founder CEOs provide superior risk–return benefits relative to a peer benchmark, a smart beta solution develops from the opportunity to overweight those companies within the benchmark. The overall result would then yield comparable risk characteristics (to those of the bench- mark) with superior risk-adjusted returns. 7 We conclude that although this approach does not work in all time periods or across all market conditions, it appears to be effective in economic environments favoring growth- oriented stocks and, in particular, specific investment sectors. 8
J OEL M. S HULMAN is a professor of
entrepreneurship at Babson College
and managing director at EntrepreneurShares, LLC in Boston, MA. shulman@babson.edu
MAKING A CASE FOR THE FOUNDER-CEO FACTOR
Smart beta approaches attempt to beat market capitalization-weighted bench- marks through the application of differing weighting methodologies that accentuate factors such as momentum, size, dividends,
W INTER 2017
T HE J OURNAL OF I NDEX I NVESTING
volatility, or value. Smart beta advocates often employ historical back tests demonstrating advantages over a specific period of time, promoting logical arguments as justification. 9 In this article, we utilize a founder-CEO factor as an enhancement to a U.S. large-cap growth index. We apply this variable coinciding with developing aca- demic and anecdotal evidence, along with emerging investment strategies, purporting benefits of investing alongside founder CEOs. 10 Legendary founder CEOs, such as Steve Jobs, Sam Walton, and Jeff Bezos, built extraordinary organizations that continue to generate exceptional personal and stakeholder wealth. 11 In their companies’ early formative years, these founder CEOs were known to eschew corporate bureaucracy, subscribe to a vision consisting of long-term leadership, cultivate organic company growth, and align executive compen- sation. Founder CEOs often assemble tight management teams with manageable debt levels and ensure that key expansion projects are held within reach. As stewards of the firm, founder CEOs appear driven by both eco- nomic and noneconomic incentives. Their entrepre- neurial culture keeps costs lean, enabling margins to expand, and retains key people. We presume that these unique governance attributes might contribute to better economic performance for the organization and superior stock results for shareholders. Academic arguments favoring a founder-CEO grouping attribute success to their long-term orienta- tion, longer tenure, higher ownership, younger firm age, higher relative expenditures on capital expenditures, and larger relative investments in research and development. 12 Founder CEOs may also be more likely to view the company as their life’s achievement, providing additional noneconomic motivation to help drive the organization to succeed. 13 Furthermore, founder-CEO firms might be more productive compared to professional coun- terparts due to reduced agency costs, continuity with leadership, greater reliance on founder reputation, and higher degree of firm-specific skills compared to non– founder CEOs. 14 Importantly, this research makes a distinction between the broader term “entrepreneur” and the spe- cific factor in this research, “founder CEO.” 15 There are likely many more factors involved in being labeled an entrepreneur than simply the term founder CEO. 16 Moreover, it would be an overstatement to presume that all founder CEOs provide an entrepreneurial outlook
or to assume that all entrepreneurs are founder CEOs. Many cases exist of well-recognized entrepreneurs who are not founder CEOs or of founder CEOs who are not entrepreneurs. 17 Founder CEOs who overcome early obstacles and persist after IPO often reward shareholders with strong stock returns. 18 We note the academic evidence that describes the benef its of investing in founder CEOs over non–founder CEOs and provide a carefully constructed methodology to implement this trading rule. 19 The model factor that we employ is based, in part, on these publications and the compelling evidence that suggests this factor may be beneficial to investors over an extended time period. 20 We incorporate our founder-CEO factor into a smart beta strategy using one of two basic approaches: We either 1) start with a U.S. large-cap growth index (benchmark index) and overweight the founder-CEO constituents or 2) simply buy a benchmark index or benchmark ETF and buy the desired founder-CEO index or individual securities within the basket of founder CEOs. 21 Given the risk–return exposure the investment manager desires, he or she can calibrate the overweighting levels in the benchmark index basket or relative amount of weighting in the benchmark index/ ETF versus the founder-CEO index. 22 Smart beta enthusiasts hope to offer better risk- adjusted returns than standard indexes by employing a passive investment strategy that targets rewarded risk premiums through alternative weighting schemes. 23 In our case, we attempt to enhance the returns of a U.S. large-cap growth index by targeting two com- bined management factors that we believe can enhance investment performance: company founders who are also CEOs. 24 The effectiveness of a smart beta index with a founder-CEO factor hinges on two basic criteria: 1) the ability to correctly identify founder-CEO compa- nies and 2) the likelihood that founder-CEO tenden- cies help improve shareholder performance. If the first criterion is difficult to detect on a consistent, reliable basis and/or the second factor becomes insignificant, then the exercise of creating a smart beta index with DOES SMART BETA WORK WITH FOUNDER CEO S ?
L EADERSHIP M ATTERS : C RAFTING A S MART B ETA P ORTFOLIO WITH A F OUNDER -CEO T WIST
W INTER 2017
E X H I B I T 1 Descriptive Statistics—Founder CEO versus Vanguard Growth
founder CEOs becomes meaningless. We believe in the possibility of both. 25 Consequently, our mission is to evaluate whether a U.S. large-cap growth ETF/index can generate enhanced performance with a modification to weights in company securities that are deemed to fall into the founder-CEO category. We propose an approach that we believe correctly identifies founder-CEO companies on a consistent basis and provides compelling results with promising poten- tial. The model appears to work well in many market conditions but may fail significantly in others. Overall, the trading rule seems to be successful much of the time and, we believe, becomes much more likely to hold true over an extended period of time. 26 Discovering founder CEOs requires considerably more effort than a simple word search due to database inconsistencies and inaccuracies. 27 We begin our quest for founder CEOs by using data from Bloomberg, Capital IQ, SP ExecuComp, and SEC company disclo- sures. 28 We start our process with the S&P 500 Index and then expand to the Russell 1000 Growth Index as our project develops. We observe that because the data sources do not all correspond with similar responses, we need to follow up each potential entry with a more detailed, company-specific examination. Many com- panies do not identify founders within their company biography or organizational title section, thus negating any opportunity to effectively screen or search based on this simple criterion. Furthermore, many companies have been delisted over the years because of mergers/ acquisitions, bankruptcies, management buyouts, or other reasons. Consequently, to correctly compile a list of the top 30 market capitalization firms (rebalanced quarterly) without survivorship bias, selection bias, or data omission bias, we need to initiate a search on every single publicly traded company prospectus from the date of index inception and identify each and every founder, year by year, with a consistent definition. 29 This is a very tedious process that requires extensive research time to ensure accuracy. Given our desire to create an index with an inception date of 2006, including the top 30 market capitalization firms (rebalanced quarterly), we ultimately examine 1,507 company prospectuses, including 106 delisted firms. 30 The descriptive statistics IDENTIFICATION OF FOUNDER CEO S
Source: Bloomberg.
and analytics of this grouping relative to a comparable U.S. large-cap growth benchmark are shown later.
Founder-CEO Descriptive Statistics
Exhibit 1 illustrates some of the descriptive statis- tics for the founder-CEO index along with comparisons for a U.S. large-cap growth index/ETF. 31 We select the Vanguard Growth Index (ticker: VUG) because it pro- vides the best fit to our founder-CEO index among U.S. large-cap growth indexes/ETFs based on overall characteristics (e.g., highest correlation, composition). 32 We note the founder-CEO index (relative to bench- mark) has (among other differences) a lower dividend yield (0.59% for founder-CEO versus 1.29% for the benchmark), higher P/E ratio (38.72% for founder- CEO versus 27.96%), significantly higher five-year sales growth rate (14.54% versus 10.20%), and higher R&D to sales (11.79% versus 8.15%). Moreover, the founder- CEO index represents companies that have a higher concentration in the Consumer Discretionary sector (28.40% versus 14.95% for benchmark), higher weight in Financials (16.39% versus 8.32%), lower weight in Health Care (9.78% versus 12.54%), lower weight in Information Technology (IT; 19.54% versus 29.72%), and higher weight in Real Estate (4.76% versus 2.59%). 33
W INTER 2017
T HE J OURNAL OF I NDEX I NVESTING
E X H I B I T 2 Summary of Annualized Returns, Year-End December 2006 through December 2015
Source: eVestment, all U.S. equity universe.
The founder-CEO index demonstrates a clear advantage over the VUG constituents on factors such as sales growth and R&D. Annualized five-year sales growth is 33% higher among founder CEOs, and the R&D-to-sales ratio is + 44% higher. By contrast, founder-CEOs hold a significantly lighter weight in sev- eral key sectors during our time period, including IT (19.54% versus 29.72%). 34 However, given the concen- trated nature of the founder-CEO index (30 holdings, rebalanced quarterly), we note the potential for signifi- cant sector variation on a year-to-year basis. 35 We rec- ognize that significant differences in sector weights can change the overall profile and total returns significantly. However, we find that regardless of sector weights in the index at any point in time, the performance dif- ferential (rewarding the founder-CEO index) tends to favor three sectors (i.e., IT, Health Care and Con- sumer Discretionary). 36 Moreover, the founder-CEO index tends to underweight Energy, Industrials, and Consumer Staples (where this index tends to underper- form or hold relatively light weights). By comparison, most U.S. large-cap growth indexes also overweight the same sectors and, in fact, may hold higher weights.
A few descriptive statistics are shown in Exhibit 1 (as of December 31, 2015).
HOW DID THEY PERFORM? RETURN SUMMARY
Exhibit 2 provides a summary of annual- ized returns comparing the founder-CEO index to the VUG for the period of December 2006 through December 2015. 37 Of the three distinct periods shown, the founder-CEO index outperforms the comparative benchmarks in all three one-year, three-year, and five- year periods. 38 Moreover, during the five-year period of January 2011 through December 2015, the annual- ized return of 16.58% for founder-CEOs ranks among the top one percentile on the eVestment database of 3,287 comparison funds within the All U.S. Equity Universe. 39 This return is well above the norm among all U.S. equity funds and 3.16% above the 13.42% annualized return for the benchmark fund. 40 During the three-year period (2013–2015), the founder-CEO index provides a 20.98% annualized return versus the U.S. large-cap benchmark return of 16.30%. This
L EADERSHIP M ATTERS : C RAFTING A S MART B ETA P ORTFOLIO WITH A F OUNDER -CEO T WIST
W INTER 2017
E X H I B I T 3 Summary of Annual Returns
Source: eVestment, all U.S. equity universe.
three-year return places the founder-CEOs in the top two percentile. During the one-year period (2015) the founder-CEO index ranks among the top five percentile of all U.S. equity funds. The calendar year returns (Exhibit 3) provide an overview of the annualized returns of the founder-CEO index relative to an eVestment all U.S. equity universe of 3,799 + funds as well as the U.S. large-cap fund bench- mark. 41 The 30-constituent founder-CEO index gen- erally performs in the top one-half of all funds in the database (seven out of nine years shown), although it fell below average in one year and well below average in another. The recessionary year of 2008, in partic- ular, represents a difficult period for the founder-CEO index. The index lost 45.27% (90th percentile among all U.S. equity funds) in 2008 and fell at least 5% to 10% below average for most U.S. equity funds and − 2.17% below the U.S. large-cap VUG benchmark ( − 45.27 versus − 43.1%). In 2012, the founder-CEO index gener- ates 13.23% (77th percentile among 3,991 all U.S. Equity funds in eVestment database), which lies approximately 1%–2% below average in performance for all U.S. equity
funds but 4.36% below the U.S. large-cap benchmark (VUG) of 17.59%. In contrast, the founder-CEO index provides relative strength against most other U.S. equity funds in years such as 2013, when it earns 48.77% (top five percentile among 3,925 U.S. equity funds); 2009, when it generates 48.74% (top 10 percentile among 4,331 all U.S. equity funds); 2010, when it produces 28.89% (top 15 percentile); and 2015, when it returns 7.42% (top five percentile among 3,799 all U.S. equity funds). 42 During the strong years for the founder-CEO index— 2015, 2013, 2011, 2010, and 2009—the index outper- forms the U.S. large-cap growth benchmark (VUG) by 4.16%, (7.42% − 3.26%), 15.45% (48.77% − 33.32%), 5.9% (7.39% − 1.49%), 7.94% (28.89% − 20.95%), and 11.89% (48.74% − 36.85%), respectively. The annual returns shown in Exhibit 3 demon- strate the outperformance of the founder-CEO index for most years of our study, but striking underperfor- mance during the critical recessionary year of 2008. Overall, the strong performance of the founder-CEO index, shown in Exhibit 2 on an absolute and relative basis, remains consistent with the year-by-year returns we observe in Exhibit 3.
W INTER 2017
T HE J OURNAL OF I NDEX I NVESTING
E X H I B I T 4 Summary of Return Information, December 2006 through December 2015
more for the comparative benchmark. At the other end, the founder-CEO index has 13 periods of − 6% or less (out of 108 months), which is greater than the 11 periods of − 6% or less for the VUG benchmark. Furthermore, extending the monthly distribution analysis to monthly returns exceeding + 4%, we see that the founder-CEO index accomplishes this feat 30.5% of the time (33 out of 108 periods) compared to the VUG, which does so 21.3% of the time (23 out of 108 periods). 45 The 30-stock founder-CEO index provides a greater likeli- hood (compared to a U.S. large-cap growth benchmark) of generating very strong monthly returns but may be more likely to generate strong negative returns. Overall, this more extreme behavior contributes to a higher stan- dard deviation of returns compared to the benchmark. 46 We observe from the distribution table of returns that situations in which monthly returns reside in the middle range ( + 4% to − 2%) occur more frequently with the VUG rather than with the founder-CEO index. 47 Exhibit 6 shows the cumulative returns and growth of a $1,000 investment from index inception. As the exhibit demonstrates, the founder-CEO index grows a $1,000 investment more than 142% in nine years to the level of $2,422. The VUG, on the other hand, appreci- ates the same investment by 81.5% to $1,815. The line representing the founder-CEO index provides a com- pelling pictorial that illustrates a clear advantage over the U.S. large-cap growth benchmark. Exhibit 7 provides peer comparisons with the founder-CEO index. The founder-CEO index provides an annual excess return of 3.48% over the VUG. Com- pared with 2,679 all U.S. equity funds in the eVestment database, the founder-CEO index ranks in the top six percentile. In terms of total returns, the founder-CEO generates 10.33% (top six percentile) during the period ranging from December 31, 2006 through December 31, 2015 (VUG total return for the same period was 6.85%) with a corresponding risk-adjusted alpha of 3.35% over the VUG (top 10 percentile). The IR is 0.45 (top five percentile) for the founder-CEO index, and the Sharpe ratio (SR) is 0.48 (top 15 percentile). 48 Exhibit 8 provides specific monthly returns for the founder-CEO index. A review of this exhibit illustrates the variability in returns for each month between year- end December 2006 through December 2015. Even in Cumulative Returns and Peer Analysis
Source: eVestment.
Exhibit 4 includes some additional summary return information for the founder-CEO index relative to the U.S. large-cap VUG benchmark. Consistent with the graph in Exhibit 2, the founder-CEO index beats the VUG benchmark for the one-year, three-year, and five-year periods. Furthermore, Exhibit 4 shows that the founder-CEO index dominates the comparative bench- mark for total performance since the inception of the index at the end of December 2006. The cumulative return for the founder-CEO index is 142.22% through December 2015, which is well ahead of the VUG return of 81.48%. Moreover, the average annualized returns over the entire period ranging from December 2006 through December 2015 is 3.48% higher (10.33% versus 6.85%, respectively). 43 Exhibit 5 shows how the founder-CEO index has a greater likelihood of monthly returns at both ends of the distribution spectrum. Both extremes (e.g., greater than 6% monthly return or less than − 6% monthly return) are more likely with the founder-CEO index than with the VUG. Out of the 108 monthly periods in the year-end December 2006 through December 2015 time period, the founder-CEO index provides 19 periods (17.6%) with monthly returns of 6% or more. 44 This compares with only 11 periods (10.2%) of monthly returns of 6% or Examining the Distribution of Returns
L EADERSHIP M ATTERS : C RAFTING A S MART B ETA P ORTFOLIO WITH A F OUNDER -CEO T WIST
W INTER 2017
E X H I B I T 5 Distribution of Monthly Returns, December 2006 through December 2015
Source: eVestment.
E X H I B I T 6 Cumulative Returns, December 2006 through December 2015
Source: eVestment.
W INTER 2017
T HE J OURNAL OF I NDEX I NVESTING
E X H I B I T 7 Peer Analysis, December 2006 through December 2015
Source: eVestment, all U.S. equity universe.
E X H I B I T 8 Monthly Returns
Source: eVestment.
very strong years in which the founder-CEO index pro- vides exceptional returns, such as 2009 and 2013, the index still experiences three months of negative returns. In contrast, calendar years with an extremely challenging economy, such as 2008, generate negative returns for 7 months out of 12. Investors should recognize that this
strategy, not unlike other concentrated equity strategies, may be subject to market volatility. Exhibit 9 provides a scatterplot of the risk–return of the founder-CEO index, VUG, and 2,679 other U.S. equity strategies in the eVestment database. The scatter- plot applies annualized returns on the y -axis given the
L EADERSHIP M ATTERS : C RAFTING A S MART B ETA P ORTFOLIO WITH A F OUNDER -CEO T WIST
W INTER 2017
E X H I B I T 9 Scatterplot Returns
Source: eVestment, all U.S. equity universe.
associated standard deviation of returns on the x -axis. The VUG is situated very close to the median inter- sects. Interestingly, the founder-CEO in the upper right quadrant suggests that, within the eVestment database of 2,679 U.S. equity investment strategies, relatively few strategies offer a similar risk–return pattern. Most of the strategies in the database contain less risk, and only a handful of the strategies offer more return. The founder-CEO index appears to offer an attractive risk– return trade-off for those investors willing to assume above-average risk. 49 Risk Analytics: How Did the Founder-CEO Index Perform? Exhibit 10 provides the risk analytics for the founder-CEO index relative to the VUG. The bench- mark provides a relatively high correlation (0.92) to the founder-CEO index. The beta for the founder-CEO index is above average risk (1.05), which corresponds with the higher standard deviation (19.72%). The above-average up capture (118.05%) and down capture (102.03%) ratios associated with the founder-CEO
index appear consistent with the return distribution chart that illustrates how returns for the founder-CEO index are more skewed to either a strong positive or negative distribution tail. We note that the higher up capture (e.g., 118.05%) over the downcapture (102.03%) provides an upward bias and appears consistent with a positive risk-adjusted alpha for the founder-CEO index (3.35%) over the VUG. 50 Clearly, the overall benefits of more frequent positive periods (corresponding with strong returns as shown in Exhibit 8) more than offset the above-average negative returns during this period of study. 51
PERFORMANCE ATTRIBUTION
Where Are Returns Generated? Asset Allocation or Security Selection?
The founder-CEO index provides performance of 142.27% during the time period of year-end 2006 through year-end 2015, compared to 81.48% for the U.S. large-cap (VUG) benchmark. Excess return is 60.79%. As we see in Exhibit 11, the performance
W INTER 2017
T HE J OURNAL OF I NDEX I NVESTING
E X H I B I T 1 0 Risk Analytics
Source: eVestment.
attribution of the founder-CEO index is concentrated in three primary sectors: IT, Health Care, and Con- sumer Discretionary. The IT sector generates 233.79% return (compared to 120.14% for the VUG bench- mark), the Health Care sector produces 587.37% return (versus 159.34% for the benchmark), and Consumer Discretionary provides 249.20% (versus 139.31% for VUG). Performance attribution represents a com- bination of security selection and asset allocation. As Exhibit 11 shows, of the 60.79% excess return, the vast majority (56.35% of 60.79%) corresponds to security selection. Only 4.43% corresponds to asset allocation. Notably, the founder-CEO index would have performed better if sector weights in the strong-performing Health Care and IT sectors matched the benchmark sector weights. 52 Continuing with the performance attribution analysis, Exhibit 11 illustrates how the founder-CEO index does not gain excess returns from sectors such as Industrials, Real Estate, Consumer Staples, Energy, Materials, and Utilities. We note that most of the weight of the index resides in three to four sectors (Consumer Discretionary, IT, Health Care, and Financials). When evaluating the performance of the founder- CEO index, we surmise that in addition to the market model, there are likely other factors that might be able to explain the 60% + excess return. Because our founder-CEO index has a strong bias toward three sec- tors (IT, Health Care, and Consumer Discretionary), a strong orientation toward growth, and a monthly return distribution that suggests momentum might be a consideration, we decide to first test for these factors. We first analyze our data against well-known factors Can Traditional Factor Models Explain Excess Returns?
to assess whether or not the founder-CEO factor might be redundant, or possibly better represented by some other factor. 53 Exhibit 12 shows the total active return of our founder-CEO index (blue line) versus the Fama– French three-factor model that examines market, value, and size. 54 As we can see from the Fama–French three-factor model, there is a significant gap between the blue line and all of the other lines. U.S. value represents − 6.53%, U.S. market provides − 0.36%, and among the three factors shown, U.S. size provides the dominant effect of 20.43%. This means out of the total active return (60.79%), the Fama–French model accounts for only 13.54%. Based on this analysis, it is unclear if any other factors, regardless of founder CEO, might be inf luential. We next decide to use another popular factor model (Carhart) that builds upon the Fama–French three-factor model by introducing the momentum factor. The Carhart model is shown in Exhibit 13. The Carhart model includes the same factors as before, but it still resides well below the blue line of total active return. U.S. value represents − 6.53%, U.S. market − 0.36%, U.S. size 20.43%, and the new momentum factor adds 2.17%. The momentum factor of 2.17% provides an improvement from the prior model, although it still leaves much of the total active return without explana- tion. Total active return of the founder-CEO index is 60.79%, and the Carhart model accounts for 15.71% of that amount, leaving a balance of 45.08% without explanation. 55 We next decide to examine a more exhaustive factor model (on Bloomberg) that incorporates virtually every known factor into the analysis. The results may be surprising to many academics and seasoned profes- sional investors and might very well be a major empirical finding in the academic literature. 56
L EADERSHIP M ATTERS : C RAFTING A S MART B ETA P ORTFOLIO WITH A F OUNDER -CEO T WIST
W INTER 2017
E X H I B I T 1 1 Performance Attribution
E X H I B I T 1 2 Fama–French Three-Factor Model
Source: Bloomberg.
The Founder-CEO Factor—Is It Real?
effect (attributed to founder-CEO model), equity return (summary), country, U.S. market, industry (summary), Consumer Discretionary, Consumer Staples, Energy, Financials, Health Care, Industrials, Materials, Utili- ties, Communication, Technology, style (summary), dividend yield, earnings variable, growth, leverage, momentum, profit, size, trade activity, value, volatility, currency, and time return. 57 Most of the factors have modest inf luence. We show the results in Exhibit 14. In our analysis, we evaluate factors contributing to the total active return during the period: December 29,
In an attempt to be exhaustive in the search for other explanatory factors in assessing the 60.79% excess returns, we utilize the Bloomberg factor model to include virtually every style, sector, country, currency, time-weighted, and other factors available. In addition to the Fama–French three factors and momentum (from the Carhart model), we encompass many others. In final form, the complete analysis we report includes 30 factors: total active return, factor return (summary), selection
W INTER 2017
T HE J OURNAL OF I NDEX I NVESTING
E X H I B I T 1 3 Factor Analysis: Carhart Model
Source: Bloomberg.
E X H I B I T 1 4 Complete Factor Analysis: 30 Factors Including Founder-CEO index
Source: Bloomberg.
2006 through December 31, 2015. The complete period demonstrates very strong capital market appreciation, although clearly parts of this economic cycle (November 2007–March 2009) experience strong market decline. As Exhibit 14 makes clear, during the nine-year period of December 29, 2006 through December 31, 2015, the founder-CEO index (shown by the green line)
outperforms the VUG benchmark by 60.79%. The selec- tion effect or founder-CEO index is, by far, the domi- nant factor. No factor among the other 29 comes close. Of the 60.79% total active return (shown by the blue line), the selection effect (founder-CEO index) helps explain 59.32%. The remaining balance of 1.47% aston- ishingly represents all of the remaining factors combined.
L EADERSHIP M ATTERS : C RAFTING A S MART B ETA P ORTFOLIO WITH A F OUNDER -CEO T WIST
W INTER 2017
E X H I B I T 1 5 Before and after Founder CEO Departs, Average Annualized Excess Return
Other significant factors include U.S. size (20.43%), Consumer Staples ( − 8.07%), Consumer Discretion (8.85%), U.S. value ( − 6.53%), volatility ( − 5.95%), and leverage ( − 5.49%). 58 These results are very compelling and, to many investors and academics, will likely be very surprising. We surmise from this analysis that the founder-CEO factor is not only a significant factor to consider for inclusion but, during our time period of study, is the most significant factor for evaluating the excess return. Although this analysis cannot address prior time periods, or necessarily predict the benefits for future time periods, we believe it would be prudent for academic scholars to, at minimum, include the founder- CEO factor for analysis in future studies. 59
(7.90% for the 5-year period and 8.26% for the 10-year period). By contrast, once the founder CEO leaves, the annualized excess return drops to 1.73% and 0.88% for the 5-year and 10-year periods, respectively. Notably, in the latter case, more than one-half of the five-year excess return (after CEO leaves) can be attributed to a single outlier situation. The differential between the period including the founder CEO and the period without the founder CEO is approximately 7% per year (excess return). 61 These results are striking and provide fur- ther support that a founder CEO makes an important difference. There are essentially two different paths that a fund manager might pursue in building a U.S. large- cap smart beta portfolio with the founder-CEO index: 1. Buy the founder-CEO index and replace part (or all) of a U.S. large-cap benchmark index or ETF. The fund manager can create a smart beta portfolio by increasing the weights of founder CEOs that are likely present in an existing index or ETF basket. 2. Build a portfolio starting with the founder-CEO index and add securities to help complete sectors that are underrepresented by the founder-CEO index. Building the Smart Beta Portfolio, One CEO-Founder Index Factor at a Time
Does a Founder-CEO Make a Difference? How Do Companies Perform after Departure?
A seemingly obvious question to this research is the effect had on a company after a founder-CEO departs from his or her firm. 60 Exhibit 15 provides an assess- ment for 5-year and 10-year periods before and after departure. In total, we have 129 companies with 5-year performance and 38 companies with 10-year perfor- mance. The results for both periods are very similar and significant. In the 5 or 10 years prior to departure, the founder-CEO company produces an annualized excess return (over market benchmark) of approximately 8%
W INTER 2017
T HE J OURNAL OF I NDEX I NVESTING
The first path is very simple. The fund manager simply needs to purchase one or two indexes. If the fund manager only buys the founder-CEO index, the portfolio will experience greater volatility relative to a more comprehensive set of holdings such as the S&P 500 Index, VUG, or Russell 1000 Growth Index (owing to 30 stocks versus 500 or 1000) but will likely generate greater returns over a longer time period relative to the benchmarks. The second approach in developing a smart beta portfolio requires the fund manager to initiate the portfolio with the founder-CEO index, assess the rela- tive sector weightings, and then fill the sector shortfall with other securities to match the benchmark index. If the investor actually holds all of the securities in the S&P 500 or Russell 1000 Growth Index, the fund manager can overweight the 30 securities in the founder-CEO index and reduce all of the other secu- rities in the index on a pro rata basis. 62 Because the founder-CEO index is an equal-weighted index, the active share for each security in the final portfolio will likely be higher than the original benchmark, although it will vary depending on the market cap of each secu- rity. 63 Given the 30-stock equal-weighted composition of the founder-CEO index, the composition of the index relative to the VUG or other more popular U.S. large growth indexes, such as the S&P 500 Growth Index or Russell 1000 Growth Index, may vary. 64 Notably, the similarity in the sector composition of the founder-CEO index relative to the VUG, S&P 500 Growth Index, or Russell 1000 Growth Index should not appreciably alter the sector exposure. The sector weights of the founder- CEO index relative to the S&P 500 Index (not S&P 500 Growth) will create a larger deviation owing to dif- ferences in growth orientation. 65 The final weights of each security in the portfolio will hinge on the level of risk–return the fund manager desires. 66 In situations in which a fund manager has sig- nificant AUM, including U.S. large-cap growth expo- sure, it may be prudent to dedicate a portion of a U.S. large-cap growth equity allocation to a passively man- aged founder-CEO index. The risk characteristics are clearly identified and can be monitored in an ongoing manner with relatively straightforward parameters. As this article demonstrates, when market conditions favor growth or market appreciation, the founder-CEO index PORTFOLIO RECOMMENDATIONS
tends to outperform its U.S. large-cap growth bench- mark. Moreover, the founder-CEO index outperforms peer benchmarks and other U.S. equity funds by a wide margin over the nine-year analysis, encompassing both very strong and negative market conditions. In some years of the analysis, the founder-CEO index per- forms in spectacular fashion. However, to be clear, the founder-CEO index also performs very poorly in some years. Fortunately, the ratio of bad-to-good perfor- mance is not symmetrical: Strong performance occurs more frequently relative to weak performance. Most key analytics, including excess returns, risk-adjusted returns, risk-adjusted alpha, IR, SR, and up capture, suggest that the founder-CEO index has compelling data to support a decision to include it in a portfolio basket. 67 The case for the smart beta portfolio implementing a founder-CEO index appears extremely compelling. Founder-CEO companies, during our study period, produce stronger performance than companies without founder CEOs. Factor analysis of our data provides further support. Results suggest that a U.S. large-cap growth fund that shifts larger weights to companies led by founder CEOs can enhance risk-adjusted portfolio performance over an extended period of time. The overall portfolio will likely see an increase in standard deviation of returns along with an increase in down capture. However, the corresponding increase in excess returns, risk-adjusted alpha, IR, SR, and up capture should more than compensate for the incremental risk exposure. As our analysis shows, there are periods of time when market factors go against the founder-CEO strategy and result in portfolio underperformance. If such an event occurs, then alpha generation will be lost, although typically only for a brief duration. The founder-CEO index has recently become available in the marketplace, although this is the first study to document the actual index. 68 Managers can simply craft a founder CEO SMA strategy around the founder-CEO index and complement the holdings with a separate U.S. large-cap index in the corresponding benchmark (e.g., S&P 500, S&P 500 Growth, Vanguard Large Cap Growth, or Russell 1000 Growth). Alterna- tively, fund managers who chose to research and create their own index can glean SEC disclosure documents SUMMARY
L EADERSHIP M ATTERS : C RAFTING A S MART B ETA P ORTFOLIO WITH A F OUNDER -CEO T WIST
W INTER 2017
and build their own founder-CEO set of holdings and invest alongside existing U.S. large-cap holdings. In either situation, the intent and implementation should be the same: Fund managers weight companies run by founder CEOs in a more concentrated manner (than currently applied) and assume risk exposures associated with this decision. 69 Incorporating a founder-CEO index overweight in a U.S. large-cap growth portfolio utilizes a logical argu- ment coupled with strong analytical support. We provide powerful evidence that leadership matters: U.S. large- cap growth companies run by founder CEOs (during our period of study) yield superior performance while the founder CEO is in control, and the excess perfor- mance cannot be easily justified by other well-known style or sector factors. But buyers beware. Investors who follow a smart beta portfolio employing higher weights to founder CEO–run companies should not expect better performance every year. Our analysis reveals that return patterns vary greatly and may be subject to peri- odic disappointment. However, for investors willing to embrace a modest increase in risk and hold their strategy for an extended period of time, our results demonstrate that the incremental risk may well be worth it. 1 Jacobs and Levy [2014] recognized that smart beta strategies are swiftly gaining market share and cited a Sep- tember 6, 2013 Financial Times article, “Smart Beta Band- wagon Triggers Alarm,” which notes that some industry experts believe smart beta might reach $6 trillion within the next few years (Marriage [2013]). They also identified experts who believe smart beta is a fad, with investors simply following a “label because it is fashionable.” “Smart Beta Bandwagon Gathers Pace,” a follow-up article by Financial Times a few months later, on April 29, 2014, references a Russell Investment survey that indicated as of that time, 32% of the 131 largest pension funds, endowments, and foun- dations had invested in smart beta. Whether smart beta is a f leeting fad or long-term industry trend is not yet clear, although the holdings continue to rise. 2 For example, Jacobs and Levy [2014] noted that the decision not to hold the capitalization-weighted market portfolio is an active decision in itself. Furthermore, they maintained that smart beta strategies require additional active decisions made at the outset, such as specific factor(s) to target, weighting method, and so on. 3 InvestmentexpertWarrenBuffett(amongothers)expends considerable energy in assessing the quality of management. ENDNOTES
He places great importance on the company management’s concerns for shareholders and seeks behaviors that align with stakeholder interests. Despite the popularity of a company management focus among active fund managers, few pas- sive ETFs focus on management criteria, and it appears even fewer, if any, address smart beta solutions. Solactiv and Entre- preneurShares are two firms that have a management-based, or founder-CEO, index. Global X has issued a founder-CEO ETF, and EntrepreneurShares has a series of entrepreneur mutual funds, separately managed accounts (SMAs), and ETFs. BlackRock (iShares), one of the market leaders in smart beta methodologies, has 43 smart beta strategies with approxi- mately $70 billion in assets under management (AUM) but, to date, no management-based ETF. 4 Leadership matters, or at least theoretically should matter. Logic aside, the notion that quality leadership traits can be appropriately measured and correctly valued for the proper time period may be a separate issue. We recognize that CEOs may not always receive the credit (or blame) for the company performance that is reported during their stay in office. Problems or solutions generated by a predecessor or changing market conditions beyond the control of the CEO often result in performance that is also beyond the control of the CEO. Moreover, the short duration of most CEOs of publicly traded companies (3–5 years) suggests that this exer- cise (allocating blame or credit) for performance while in office may be moot. We note that founder CEOs, unlike professional CEOs who are not the creator of the company, have (traditionally) a much longer duration in their position (7–10 years). The length of time differential for a founder CEO (compared to a professional CEO) is about double the norm. Moreover, we will later provide some evidence of market-adjusted returns before and after the founder CEO spends time in office. In short, we attempt to stake the argu- ment that 1) proper leadership matters and 2) founder CEOs may provide evidence of good leadership. 5 Fama and French [1993] created a framework to address factor-based smart beta strategies. In their article, they were able to demonstrate how certain factors, such as market capitalization and book to market equity, enabled investors to generate returns higher than predicted by the capital asset pricing model. They proposed a three-factor model that sug- gested market, size, and value help to explain much of the returns for a portfolio. Later, Fama and French expanded their three-factor model to a five-factor model that also included earnings and investment (increase in book equity). Carhart [1997] attempted to improve upon the Fama–French three-factor model by adding a momentum factor (creating a four-factor model). In this article, we explore the Fama– French factors as well as the incremental momentum factor by Carhart. For completeness, we also add at least 20 other factors provided by Bloomberg to examine potential factor
W INTER 2017
T HE J OURNAL OF I NDEX I NVESTING
7 We recognize that the outperformance could be attributed to other factors or market conditions and, later in the article, examine performance attribution across varying sectors and time periods. Moreover, in selecting a market benchmark, we first regress the return stream against a few market benchmarks holding comparable security populations (e.g., growth, large cap) and then select the benchmark with the highest correlation for comparative purposes. Moreover, in determining any potential advantage to our security selec- tion, we adjust for a number of factors, including size, style, momentum, liquidity, yield, quality, volatility, and profit- ability. We note that other methodologies that disentangle stock returns, such as by Jacobs and Levy [1988], might yield different results. We show later in Exhibits 4 and 10 that the risk associated with the founder-CEO index does not vary appreciably from a comparable benchmark. 8 By definition, a risk variable will not consistently provide returns in each and every period. As we show in the results section, results favor traditionally entrepreneurial growth sectors, such as Information Technology, Health Care, and Consumer Discretionary. Most stock holdings in this article represent these sectors. By contrast, relatively few companies reside in the Utility, Telecommunication, and Materials sectors. 9 Jacobs and Levy [2014] differentiated between smart beta and smart alpha and discussed the potential for over- crowding among smart beta strategies due to their simplicity and transparency. 10 Clearly, not all of the evidence is positive, although the preponderance of academic literature provides encour- aging results. Early research on founder CEOs by Johnson et al. [1985] showed a positive stock price reaction following the abrupt death of a corporate founder. Morck, Shleifer, and Vishny [1988] found a negative effect associated with founders and market valuation (principally among older firms). Research by Fahlenbrach [2009]; Palia, Abraham, and Chia-Jane [2008]; Villalonga and Amit [2006]; Adams, Almeida, and Ferreira [2009]; and Shulman [2010] has showed a positive effect of founder CEOs and investment performance. Evidence from the more recent, comprehensive studies provides the motivation for our smart beta analysis. 11 As we will see later, the returns to shareholders decline sharply after the founder CEO departs the organization. The effect is especially distinct in the period 5–10 years before the founder CEO retires/departs. 12 Fahlenbrach [2009] conducted a study of 2,327 large U.S. publicly traded companies over the period 1992–2002 and found significant differences between founder-CEO companies and successor-run companies. In particular, he identified differences in R&D, M&A, capital expenditure debt/assets, age, ownership, and stock performance. As we show later in the descriptive statistics section, R&D as a
exclusions or omissions. Our motivation has been to identify as many unique, relevant factors as possible. Earlier academic reviews surmised that what we deemed to be an entrepre- neur factor was likely already encompassed in factors such as growth, momentum, earnings, or some other well-explored category. We find that one entrepreneur factor that we proxy as a founder-CEO factor dominates the explanatory anal- ysis by far over the period from December 2006 through December 2015. We recognize that there may be other entre- preneur factors that can help explain excess returns, although we leave it to future research scholars to consider additional related characteristics. Notably, the founder-CEO factor seems to provide unique value, although a reason does not yet appear obvious. We surmise that characteristics embedded in this factor may help clarify this conundrum. For example, founder CEOs may assemble a unique governance or incen- tive structure that generates returns that exceed expectations (Leland and Pyle [1977]). Board composition; hiring prac- tices; growth financing/trajectory; employee compensation/ ownership; share classes/voting rights; selling, general, and administrative expenses; management attributes; and so on may all share some systematic commonalities with founder CEOs. Theoretically, any implied benefits or pricing anomaly should be priced away over time. However, it is possible that a series of complex factors underlie or are encompassed within founder CEOs that generate unusual or difficult-to-predict surprises (e.g., earnings, growth, productivity) that have not yet been discovered. 6 Citations later in this article discuss the merits of founder-CEO research. The evidence will make clear that founder CEOs have almost double the average duration in their jobs (as CEO), higher ownership levels, higher research and development (R&D) investments, better results with mergers and acquisitions (M&A), and stronger revenue growth while in office. Moreover, as our study shows, periods before and after the tenure of founder CEOs demonstrate a signifi- cant difference in relative risk-adjusted stock returns. While governance differences appear to be central to this research, the exact reasons for continued outperformance in the stock market are not clear at this time (although likely stemming from the list just given). Theoretically, investors in the mar- ketplace would observe this anomaly and price it away imme- diately. However, the results of our study show that returns are not symmetrical among constituents in the founder-CEO basket. Like any other portfolio, results can be skewed by significant winners and/or losers. Performance distribution results (shown later) indicate significant difference between founder CEOs and the benchmark. We surmise that founder CEOs have inherent governance traits that allow the possibility for exceeding market expectations year after year. Anecdotal cases such as Apple, Netf lix, Amazon, Google, and Facebook (among others) appear representative of this conclusion.
L EADERSHIP M ATTERS : C RAFTING A S MART B ETA P ORTFOLIO WITH A F OUNDER -CEO T WIST
W INTER 2017
percentage of sales is higher for founder CEOs than the peer benchmark. 13 Wasserman [2003] discussed how founder CEOs are different from professional CEOs. He noted that professional CEOs are older, have more years of prior work experience, are paid higher salaries, own significantly less of the compa- ny’s equity, and have less control. Dobrev and Barnett [2005] also described the identity of organizational founders as being closely aligned to their organization. In addition, O’Reilly and Chatman [1986] discussed the psychological bonds that link individuals to their organizations. 14 Gao and Jain [2011] provided an excellent overview of the theoretical development and hypotheses regarding why founder-CEO firms are likely to be more productive com- pared to non–founder-CEO firms. Fama and Jensen [1983], Nelson [2003], and Wasserman [2003] suggested that agency costs are lower in founder-run companies (implying founder- led companies are antiagency cost). Aldrich [1979] and Fischer et al. [2004] discussed the importance of a founder CEO during the transition to a public company. Basu, Dimitrova, and Paeglis [2009] noted that a newly public firm generally does not have its own reputation, so it needs to rely more heavily on its founder to gain investor attention. Finally, Gao and Jain [2011] argued that founder CEOs are characterized by a higher need for achievement, stronger psychological attachment to their company, tighter economic ties, larger ownership stakes, longer investment horizon, and higher degree of firm-specific skills. These attributes contribute to the founder CEO’s overall willingness and desire to pursue long-term strategies at the expense of short-term results, with corresponding improvement in post–initial public offering (IPO) performance. However, there is a counterbalance effect in play. Dobrev and Barnett [2005] discussed the increased likelihood of founder CEOs leaving an organization as it grows larger and matures and their comparative skill set diminishes in value as the firm matures. In contrast, they found the opposite effect with professional CEOs. 15 We define founder as the key individual or individuals who are/were with the company at inception or pre-revenue. In many cases, there are discrepancies among databases such as Bloomberg, Capital IQ, and company websites. We searched each company in the Russell 1000 database from the begin- ning of our time period to ensure that each founder held these characteristics. Individuals who transferred from an existing company with a spinoff are not labeled as founders. Notably, individuals who take over an existing enterprise and then recast history by labeling themselves as founders, such as the characterization of Ray Croc in the film “Founders” would not be labeled a founder in our study. We recognize individuals such as Ray Croc as worthy entrepreneurs, but we do not include these individuals in our founder-CEO study. Other research, such as that by Shulman [2010], has
provided an investment model of 15 entrepreneurial charac- teristics and built an index of this grouping. In the identifica- tion of entrepreneurs, the founder-CEO variable is included along with a broader set of individuals in addition to founder CEOs. This research might include individuals such as Ray Croc or people who display unique characteristics resulting in substantial growth and vision but who did not participate at company inception. 16 For example, entrepreneurs may have a different board composition, ownership structure, compensation arrangement, capital structure, growth orientation, R&D perspective, and long-term vision (among other differences) compared to non-entrepreneurs. 17 Shulman and Noyes [2012] discuss the differences between a founder and an entrepreneur. For example, Angelo Mozillo, the founder CEO of Countrywide, was widely viewed as a self-serving inside trader who helped contribute to the 2008 mortgage crisis (he paid a $67.5 million fine to the Securities and Exchange Commission [SEC]). Mr. Mozillo does not have the same business traits as well-known entrepreneurs such as Howard Schultz (Starbucks) and Elon Musk (Tesla). These two successful individuals came in to their organizations early, were visionaries who created wealth for themselves and others, but were not founders. Moreover, Warren Buffet (Berkshire Hathaway) is another unique example of an extremely successful entrepreneur (who was not a founder) who successfully grew an established organiza- tion into something more substantive. 18 We provide detailed analytics later in this article showing how a portfolio of the 30 largest market cap U.S. founder-CEO public companies (rebalanced quarterly), dating from June 2005 through September 2016, significantly outperform a peer group of 632 U.S. large-cap fund strategies provided on the eVestment data basis. The analytics show that the founder-CEO group are among the top 1% of all U.S. large-cap strategies during this time period, based on total returns and excess returns. As our analysis will reveal, the results are not consistent for each time period and vary considerably depending on the industry sector. 19 Gao and Jain [2011] and Fahlenbrach [2009], among others, provided evidence that shareholders of publicly traded founder CEOs perform better than professional CEOs. Gao and Jain examined the five-year post-IPO performance of 1,963 IPOs from 1997–2000. They found that high-tech companies run by founder CEOs were more likely to outper- form professional CEOs (especially when venture capitalists were not involved). Companies in low-technology areas did not outperform during this time period. Fahlenbrach showed that founder-CEO firms outperformed professional-CEO firms by 8% per year and suggested that long-term invest- ments in R&D, capital expenditure, and other initiatives were largely responsible.
W INTER 2017
T HE J OURNAL OF I NDEX I NVESTING
Page 1 Page 2 Page 3 Page 4 Page 5 Page 6 Page 7 Page 8 Page 9 Page 10 Page 11 Page 12 Page 13 Page 14 Page 15 Page 16 Page 17 Page 18 Page 19 Page 20 Page 21 Page 22 Page 23 Page 24 Page 25Made with FlippingBook - Online magazine maker