The Financial Review 44 (2009) 179 212 One Man Two Hats: What’s All the Commotion! Jay Dahya∗ Baruch College, CUNY Laura Galguera Garcia University of Oviedo Jos van Bommel University of Oxford Abstract We examine performance in publicly listed U.K companies over a period that encompasses the issuance of the Cadbury Committee’s Code of Best Practice, which calls for the abolition of the combined CEO/COB position We find that companies splitting the combined CEO/COB position to conform to the Code’s requirement did not exhibit any absolute or relative improvement in performance when compared to various peer-group benchmarks We not necessarily scoff at mandated board structures, but the evidence suggests that this particular legislature coerced the abandonment of the combined CEO/COB position and appears to be wide of the mark Keywords: Cadbury, CEO, directors, governance, UK JEL Classifications: G28, G30, G34 ∗ Corresponding author: Baruch College, City University of New York, One Bernard Baruch Way, Box B10-225, New York, NY 10010-5585; Phone: (646) 312-3511; Fax: (646) 312-3451; E-mail: jay.dahya@baruch.cuny.edu We appreciate assistance in compiling the data from Companies House U.K., Lloyds Share Registration Services, Hemmington Scott, and the Subotnick Financial Services Center This paper has benefited from the helpful comments of Linda Allen, David Denis, Diane Denis, Jack Francis, Joanne Li, Colin Mayer, John McConnell, and seminar participants at Baruch College, Queen’s University, Belfast, and FMA European Meetings 2005 Jay Dahya acknowledges financial support from the Baruch College Fund, INQUIRE-UK, and the Eugene M Lang Junior Faculty Fellowship C 2009, The Eastern Finance Association 179 180 J Dahya et al./The Financial Review 44 (2009) 179–212 Traditionally, the top executive position on the boards of directors of publicly traded corporations around the world has been held by a single individual holding the title of joint chief executive officer (CEO) and chairman of the board (COB) For example, in 1990, the positions of CEO and COB were combined for 372 of the Fortune 500 firms (i.e., 74.4% of the firms) Similarly, in 1990, of the largest 25% of firms (ranked by market capitalization) listed on the stock exchanges of the United Kingdom, France, Australia, Belgium, Canada, Japan, Italy, India, and Brazil, the titles of CEO and COB were combined in more than 60% of the listed firms in each country During the 1990s and beyond, publicly traded corporations worldwide have experienced increased pressure to separate the combined position of CEO and COB so two different individuals would maintain the titles of CEO and COB Between 1992 and 2004, at least 16 countries have witnessed the publication of reports sponsored by their governments, or sometimes by their major stock exchanges, advocating that two different individuals hold the positions of CEO and COB.1 The new standard typically requires a major overhaul in the leadership structure on corporate boards As far as the United States is concerned, the recommendation has shown up in the Sarbanes-Oxley Act and has garnered much publicity For all the aforementioned countries, the proposal has been implemented with a sense of urgency, with little evidence to indicate that its adoption would actually prove beneficial Arguably, the global movement toward the separation of the combined chairman and CEO position can be traced back to the Committee on the Financial Aspects of Corporate Governance, commonly referred to as the Cadbury Report, in the United Kingdom in 1992 The Cadbury Committee Report recommended that the positions of CEO and COB in U.K.-listed companies be held by two different individuals Compliance of this key recommendation required U.K companies to make noteworthy changes In 1988, for example, 63.6% of the Financial Times (FT) 500 and 57.6% of all London Stock Exchange (LSE) companies had the positions of CEO and COB held by the same individual By 2000, 9.8% of the FT 500 and 22.8% of all LSE-listed companies had the posts of CEO and COB held by a single person An underlying presumption of this movement is that boards with different individuals holding the positions of CEO and COB will significantly improve the quality of board monitoring and as a result lead to better corporate performance However, in large measure, this presumption lacks empirical support Various studies show that combining the positions neither improves nor destroys corporate performance (Berg and Smith, 1978; Chaganti, Mahajan and Sharma, 1985; Baliga, Moyer and Rao, 1996; Brickley, Coles and Jarrell, 1997; Daly and Dalton, 1997; Palmon and Wald, 2002) There are several explanations as to why prior studies might fail to find a convincing relationship, if one exists First, firms tend to The countries include Australia, Belgium, Brazil, Canada, Cyprus, Czech Republic, France, Greece, India, Japan, Kenya, Kyrgyz Republic, Malaysia, Singapore, South Africa, and the United Kingdom J Dahya et al./The Financial Review 44 (2009) 179–212 181 combine the positions of CEO and COB during top executive succession events Disentangling the effects of executive succession from the effects of combining the positions of CEO and COB has proved to be a significant challenge (Brickley, Coles and Jarrell, 1997) Second, the variables themselves may be endogenous That is, if combining the positions of CEO and COB does affect corporate performance, then when every top executive position is at its optimum configuration, there will be no relation to observe in the cross-section between the combined position of CEO/COB and performance.2 Finally, prior studies have focused primarily on U.S companies, and most U.S companies continue to combine the positions Thus, it is difficult to find corporate boards that not combine the positions to serve as a control group In this study, we investigate further the relation between combining the positions of CEO and COB and corporate performance over the years surrounding the issuance of the Cadbury Report (in December 1992) in the United Kingdom By using U.K firms in our analysis, we hope to alleviate some of the shortcomings attributed to studies conducted with U.S data First, because corporate boards were pressured (through a government-sponsored mandate) to separate the combined positions of CEO and COB, the effects of CEO succession can be disentangled from the effects of combining the positions Second, it can be argued that the Cadbury recommendation represented an exogenous event that radically altered top executive leadership structures in the United Kingdom And, third, the aforementioned changes, as shown in Figure 1, were concentrated over a relatively short time and the large sample of top executive positions with altered structures provides a clean before- and after-event analysis.3 Additionally, exploring the adoption of this one key Cadbury recommendation is interesting in its own right because many other countries appeared to have modeled their corporate governance codes on the Cadbury Report Because the Cadbury Report was one of the first of such national mandates, it has now been in effect long enough that the impact on corporate performance, if any, can be observed Thus, this study can be thought of as a precursor of what might occur in other countries that have adopted (or are contemplating adopting) similar guidelines The key question that we ask in this study is whether U.K companies that complied with the Cadbury recommendation, by separating the combined position of CEO and COB, experienced an improvement in corporate performance We find that compliance is not associated with any (statistically or economically The rationale for this explanation is drawn from studies by Bhagat and Black (2002) and Hermalin and Weisbach (1991), among others, who report evidence consistent with endogeneity between board composition and corporate performance Figure presents the proportion of all publicly traded industrial companies that combined the positions of CEO and COB on the LSE against those on the New York Stock Exchange (NYSE) from 1986 through 2000 As observed in Figure 1, the aftermath of the Cadbury Report witnessed widespread reduction in the number of firms combing the positions of CEO and COB (from 72% to 20%) In comparison, the proportion of firms combining the titles of CEO and COB for firms listed on the NYSE is essentially unchanged (around 78% throughout) CEO and COB 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 1986 UK - Combined CEO and COB 1987 1988 1989 US - Combined CEO and COB 1990 1991 Year 1992 1993 1994 1995 1996 1997 U.K and U.S industrial firms listed on the LSE and NYSE, respectively, combining the positions of CEO and COB over 1986–1997 The solid line is the set of firms on the NYSE, and the dashed line is the set of firms on the LSE that after December 1992 were coerced to separate the combined CEO and COB position in order to be in compliance with the Cadbury Committee recommendation (the adopted-Cadbury firms) To determine which set to classify a firm, we tracked the names of the persons holding the titles of CEO and COB each year for each British firm from the Stock Exchange Yearbook and from Proxy Statements for each U.S firm Figure Fraction of All Industrial Firms with a Combined 0.90 182 J Dahya et al./The Financial Review 44 (2009) 179–212 J Dahya et al./The Financial Review 44 (2009) 179–212 183 significant) improvement in operating or stock price performance relative to benchmark companies Nonetheless, we recognize that our measure of operating performance may be biased since all sample firms experience a drop in earnings over 1989–1992 and what we observe is merely a natural mean reversion in earnings for a given benchmark over this period To address this concern, we recompute our performance measure using a performance-matching method (Barber and Lyon, 1996) Against this more robust benchmark, companies that adopted the recommendation experienced either deterioration or no change in operating performance relative to benchmark companies over the period of our analysis As a further check on our performance measure, we conduct our analysis using stock price in place of operating performance The results on stock returns show that separating the combined CEO and COB post is not associated with any significant improvement in stock price performance relative to benchmark companies In sum, our analysis on the adoption of this one key item of the Cadbury Report dashes hope that this feel-good factor will lead to global corporate governance rapture As with any study of this nature, numerous caveats are in order First, we not necessarily scoff at mandated top executive structures, but the evidence in this study shows that such a mandate was not associated with any improvement in the performance of U.K companies We also acknowledge the experience may be unique to the United Kingdom or to the time period studied Studies of other countries in time will provide insight regarding the extent to which the findings from the U.K setting can be generalized Background 1.1 A first mover: The Cadbury Committee Report Arguably, the global movement toward the separation of the combined CEO and COB position can be traced back to the Cadbury Committee’s report on the financial aspects of corporate governance issued in December 1992 The Cadbury Committee was appointed by the Conservative Party government in the United Kingdom with a broad directive to examine the financial aspects of corporate governance The key proposals of the Committee were contained in the Code of Best Practice, which presents the committee’s recommendations on the structure and responsibilities of corporate boards of directors in U.K listed companies Among other things, the Code of Best Practice recommended the separation of the combined CEO and COB position so two different individuals hold the positions of CEO and COB As of 2007, the Code has not been enacted into securities law, and compliance with the Code is entirely voluntary Nonetheless, the Code was heavily backed by the LSE, which as of June 1993 has required a statement from each listed company on whether the company is in compliance with the recommendations contained in the Code To gauge the significance of the Cadbury Report, it is imperative to appreciate the circumstances surrounding its publication The Cadbury Committee was appointed 184 J Dahya et al./The Financial Review 44 (2009) 179–212 following several corporate collapses of prominent FTSE companies over the late 1980s and early 1990s, including Maxwell Communications PLC and Colorol PLC, among others At its publication, those on both sides of the debate received the Cadbury Code with a great degree of cynicism On the one hand, there were those who felt that legislation would be inevitable to enforce the recommendations contained in the Code On the other hand, there were those who felt that the balance between stockholders and senior management is best left to capital market forces 1.2 Global movement toward separating the combined CEO and COB position The Cadbury Committee Report sparked global corporate governance frenzy Between the publication of the Cadbury Report in December 1992 and December 2002, at least 16 other countries witnessed publication of similar guidelines calling for the separation of the combined CEO and COB position In the spirit of the Cadbury Report, in 1994, the Dey Report established guidelines for boards of publicly traded Canadian corporations and the King Report purported similar rules for South African corporations In 1995, the Bosch Committee’s Report on Corporate Practices and Conduct in Australia prescribed that two different individuals should hold the titles of CEO and COB in publicly traded firms, and the Conseil National du Patronat Francais in the Vienot Report, published in France, put forth similar recommendations From 1996 to 1999, five additional countries issued guidelines coercing corporations in Japan (Report of the Corporate Governance Forum of Japan, 1997), Belgium (Cardon Report, 1998), Brazil (Report of the Instituto Braziliero de Governanca Corporativa, 1999), Greece (Principles of Corporate Governance issued by the Greek Capital Markets Commission, 1999), and India (Kumar Mangalam Report issued by Securities and Exchange Board of India, 1999) to separate the combined position Since 2000, at least six more countries have promulgated guidelines in line with the Cadbury recommendation These include Cyprus (Cyprus Stock Exchange, 2003), Czech Republic (Czech Securities Commission, 2001), Kenya (Private Sector Corporate Governance Initiative sponsored by the Stock Exchange Committee, 2003), Kyrzyg Republic (Securities Commission, 2002), Malaysia (Securities Commission Malaysia, 2000), and Singapore (Council on Corporate Disclosure and Governance established by the Securities Commission, 2001) The general presumption underlying the global movement toward the eradication of the combined CEO and COB position is that boards with two separate individuals holding the post would facilitate an independent check on the behavior of the CEO and more importantly improve the overall quality of corporate monitoring and performance 1.3 Prior studies Financial economists have devoted considerable effort on analyzing the link between board composition and corporate performance Studies on the topic fall J Dahya et al./The Financial Review 44 (2009) 179–212 185 into one of two categories: (1) those that examine the link between the fraction of outside directors on the corporate board and corporate performance, and (2) those that examine combining the CEO and COB positions and corporate performance The first category can be further subdivided into those that affect the way in which boards accomplish discrete tasks and those that examine cross-sectional relations between board composition and corporate performance.4 Our review on the more pertinent second category will be brief since Brickley, Coles, and Jarrell (1997), Dahya and Travlos (2002), and Palmon and Wald (2002) provide a comprehensive review of prior literature on the connection between combining the CEO and COB positions and corporate performance Studies on the relation between combined titles and corporate performance, generally correlate various measures of corporate performance on whether the CEO and COB positions are combined or not For example, Brickley, Coles and Jarrell (1997) study a sample of 661 large U.S firms contained in the Forbes survey of executive compensation in 1988 They perform a battery of tests including cross-sectional regressions of performance, as measured by return on capital and stock returns, against an indicator variable for a combined CEO and COB and various control variables With various regression specifications, they fail to report any statistically significant relation between combined titles and corporate performance The apparent conclusion is that combining the CEO and COB positions in large U.S corporations is an efficient set-up that preserves shareholder value Of course, Brickley, Coles and Jarrell (1997) is not the only study to analyze the relation between combining the CEO and COB and corporate performance Baliga, Moyer and Rao (1996) compare operating performance among 181 industrial U.S firms that have changed the structure of the combined CEO and COB position This study fails to locate any operating performance changes surrounding changes in the status of the combined CEO and COB position Similarly, Berg and Smith (1978) and Daly and Dalton (1997) analyze differences in the financial performance of those companies combining the titles of CEO and COB versus those that not Both studies fail to show any differences in performance for the two sets of firms and conclude that combining the CEO and COB positions is indeed not a suboptimal board configuration These findings appear to contrast with observations advanced by Jensen (1993) and Hermalin and Weisbach (1991) who suggest that the combined position of CEO Studies that explore how board composition affects the way in which boards accomplish discrete tasks, such as responding to hostile takeovers, hiring and firing the CEO, and setting CEO compensation, include Brickley, Coles and Terry (1994), Brickley and James (1987), Byrd and Hickman (1992), Core, Holthausen and Larcker (1999), Cotter, Shivdasani and Zenner (1997), Franks and Mayer (1996), Shivdasani (1993), among others Whereas, Agrawal and Knoeber (1996), Bhagat and Black (2002), Dahya and McConnell (2007), Dahya, Dimitrov and McConnell (2008), Coles, Daniel and Naveen (2008), Boone, Field, Karpoff and Raheja (2007), Denis and Sarin (1999), Hermalin and Weisbach (1991), Kaplan and Reishus (1990), Mehran (1995), and Yermack (1996) explore the cross-sectional relation between board composition and corporate performance 186 J Dahya et al./The Financial Review 44 (2009) 179–212 and COB, among other things, lessens the monitoring ability of the corporate board Consistent with these observations, Palmon and Wald (2002) argue that the widely reported absence of a relation between leadership structure and firm performance may be due to the neglect of firm size as an important explanatory variable Specifically, they hypothesize that small firms benefit more from the clarity and decisiveness of decision making under a single sure-footed executive, while large firms benefit more from the checks and balances associated with a separation of the two functions In their empirical analysis, Palmon and Wald (2002) report a negative announcement period stock return when small firms announce the switch from a combined CEO/COB to separate functions on the Dow Jones Interactive Newswire and a positive announcement effect when large firms undertake the same maneuver Consistent with the results reported by Palmon and Wald (2002) on large U.S companies, Dahya, Lonie and Power (1996), Dedman (2000), and Carapeto, Lasfer and Machera (2005) report positive announcement returns to the separation of the unitary leadership structure for various samples of large U.K publicly traded companies from 1989 through 2003 One obvious shortcoming with the studies on leadership structure and corporate performance is that the two variables may be endogenously determined (Hermalin and Weisbach, 1998) If so, it could be that firms that are inclined to combine the CEO and COB positions may also be more likely to display inferior (superior) performance If that is the case, the results observed in prior U.S studies (a combined CEO/COB position neither improves nor destroys corporate performance) and in contemporaneous U.K studies (a positive stock price effect to leadership structure changes in large firms) might indeed be spurious Because the Cadbury Report spurred dramatic changes in the board configuration of U.K companies, we are presented with a large sample of changes in the CEO/COB position over a relatively short time, which provides an opportunity to alleviate endogeneity, if any, and also limits the effects of top executive succession on corporate performance Against this backdrop and as the first mover in the global movement toward the eradication of the combined CEO and COB position, this study explores the separation of the combined CEO and COB position on corporate performance in U.K companies, in relation to the issuance of the Cadbury Report Sample and data sources Our sample includes industrial companies listed on the LSE from 1989 through 1996 In the aggregate, 1,124 firms enter the analysis with at least one year of data on top executive structure and certain financial information To begin, we record the name(s) of the CEO and COB from the board roster in the Stock Exchange Yearbook in each year for each firm from 1988 through 1996 This enables us to record CEO/COB succession in each year from 1989 through 1996 for our sample firms Since we are only concerned with changes in the combined position of CEO and COB that are not associated with CEO/COB succession, we screen all CEO/COB successions unrelated 187 J Dahya et al./The Financial Review 44 (2009) 179–212 Table Descriptive statistics for U.K industrial firms listed on the LSE, 1989–1996 Descriptive statistics on publicly traded U.K industrial firms (with data on board composition and financial information) from 1989 through 1996 In the aggregate, the sample includes 1,124 firms that enter the analysis with at least one year of data Accounting information and share prices are taken from Datastream Board and equity ownership characteristics are taken from annual filings at Companies House and supplemented with data from the Corporate Register and Stock Exchange Yearbooks from 1988 through 1996 Year Variable Sample size Joint CEO/COB (%) 1989 700 1990 803 1991 814 1992 881 1993 882 1994 937 1995 907 1996 902 56.29 54.05 51.84 50.40 43.20 39.70 37.60 33.48 Board size Percentage of outside directors 6.90 0.30 7.10 0.32 7.30 0.33 7.30 0.34 7.90 0.41 8.00 0.44 8.10 0.47 8.20 0.48 CEO equity ownership (%) 2.52 2.64 3.13 4.80 5.29 5.15 5.50 5.66 Book value of assets (£m) Market value of equity (£m) 242.60 240.80 253.90 269.40 278.00 291.40 315.60 359.10 246.00 239.50 259.40 278.50 317.10 314.20 336.90 361.50 to Cadbury compliance and remove them from further analysis.5 In this way, we limit our analysis to all changes in the combined CEO and COB position that not stem from CEO/COB succession For each firm in the sample, for each year 1989–1996, we also used the Stock Exchange Yearbook, the Corporate Register, and annual filings maintained at Companies House to determine the size of the corporate board, the number of outside directors, the total number of shares held by the CEO, the total number of shares held by board members, the total number of shares held by institutions, and the number of block shareholders; a block shareholder is defined as any institutional shareholder owning greater than three percent of the company’s common stock Summary statistics on the combined position of CEO and COB, outside directors, book value of total assets, market value of equity and CEO, board and institutional share ownership, along with the number of firms in the sample from 1989 through 1996 are presented in Table The primary statistic of interest to us is the fraction of firms that combined the positions of CEO and COB (Joint CEO/COB) Table shows a marked decrease in the fraction of firms that combined the positions of CEO and COB (Joint CEO/COB) from 56% in 1989 to 33% in 1996 The decrease in the number of firms combining the titles of CEO and COB are concentrated in the years We also require that a statement containing details of the (nonsuccession related) change in the top executive leadership structure be filed with Companies House In an effort to corroborate the Companies House filings, we examine news articles in the Financial Times, Extel Weekly News Summaries and Macarthy’s New Information Service If data on the announcement of a separation in the combined CEO and COB position is not available for a given firm, we directly liaise with the company to ascertain that information, whenever possible 188 J Dahya et al./The Financial Review 44 (2009) 179–212 following the issuance of the Cadbury Report.6 A similar marked increase in the percentage of outside directors (from 30% to 48%) and board size (from 6.9 to 8.2) is observed over the same time period Table also shows that the fraction of shares owned by the CEO increased, on average, by 125% and the median market value of equity and median book value of assets grew, on average, by 50% over the eight-year interval For our analysis, we split the sample firms into four mutually exclusive groups: (1) the set of firms for which two individuals maintain the titles of CEO and COB every year in which they were listed on the LSE over the period 1989–1996 (we call this the Always separated CEO and COB set, 318 firms); (2) the set of firms that always combined the positions of CEO and COB (we call this the Always joined CEO and COB set, 247 firms); (3) the set of firms that recently created the combined position of CEO and COB (we call this the Created joint CEO and COB set, 169 firms); and (4) the remaining set of firms that comprises those that split the combined position of CEO and COB, such that two individuals would hold the titles of CEO and COB any year during the interval 1989–1996 (we call this the Adopted Cadbury & separated joint CEO and COB set, 390 firms) The summary statistics contained in Table are expanded on in Table for each of the four sets of firms Panel A is the Adopted Cadbury & separated joint CEO and COB set, Panel B is the Always separated CEO and COB set, Panel C is the Joined CEO and COB set, and Panel D is the Created joint CEO and COB set The average firm in the “Adopted Cadbury & separated joint CEO and COB” set shows a marked increase in board size from 6.6 members in 1989 to 8.3 members in 1996 A similar increase in board membership is observed in Panels B and C for firms that always separated the CEO/COB position (from 7.6 to 8.5 members) and for firms that always joined the CEO/COB positions (from 5.5 to 6.8 members) The “Created joint CEO and COB,” set in Panel D, exhibits a mild decrease from 6.4 to 5.7 members All four panels in Table show an average increase of 5–15% in the fraction of outside directors on the corporate board Consistent with Dahya and McConnell (2007), this increase can be ascribed to the widespread adoption of the other key recommendation of the Cadbury Report that established a minimum representation of outsiders on the board This implies that in our multivariate analysis we should control for the fraction of outside directors on the corporate board By definition, of course, the “Adopted Cadbury & separated joint CEO and COB” set will begin with a large proportion of companies combining the posts of CEO and COB and will end up with zero by 1996 As might be expected, this statistic shows its We acknowledge that the Cadbury recommendation cannot be held entirely accountable for the whole decline However, we have been meticulous to exclude all succession-related changes in the CEO/COB position To alleviate this concern further, we have plotted a line-of-best-fit using data on the number of firms combining the positions of CEO and COB over 1989–1991 to estimate the proportion of firms with a unitary leadership structure in the absence of a Cadbury-type recommendation by 1996 Our estimate suggests that 50% of the firms, at most would combine the two posts The actual number of firms combining the CEO and COB positions by 1996 was 33%—a staggering 17% lower than the linear estimate Low ROA (in %) (B) Moderate ROA (in %) (C) High ROA (in %) (D) −1.39 −0.83 −0.11 0.14 0.08 9.02 8.76 8.93 8.99 8.65 a 1.28b 1.53b 1.47b −0.09 −0.03 −0.37 0.42 0.47 0.43 1.00 1.72b 2.30a and b indicate significance at the 0.01 and 0.05 level y − to y + y − to y + y − to y + −0.68 −0.85 0.32 1.04 1.62 Low ROA (in %) (E) 0.36 0.94 0.41 8.94 9.21 9.30 9.88 9.35 Moderate ROA (in %) (F) ROA in percent) 18.88 19.21 19.30 19.35 19.31 Panel B: Change in return on assets (mean y−1 y y+1 y+2 y+3 Panel A: Return on assets (mean ROA in percent) Event year (A) Split CEO/COB titles over 1989–1996 firms 0.25 −0.06 0.38 19.95 20.21 20.20 19.89 20.33 High ROA (in %) (G) Industry- & performancematched always-separate CEO/COB titles firms 0.28 −0.19 −0.83 Low ROA (in %) (col B minus col E) (H) −0.45 −0.97 −0.78 Moderate ROA (in %) (col C minus col F) (I) Difference in means 0.17 0.53 0.05 High ROA (in %) (col D minus col G) (J) 1.15 2.51a 1.57b 7.74 8.21 8.89 10.25 9.31 Pre 1992 (in %) (K) −0.33 −0.37 −0.49 8.92 8.54 8.59 8.55 8.43 Post 1993 (in %) (L) Split CEO/COB titles over 1989–1996 firms 0.43 1.29b 1.15 7.92 8.20 8.35 9.21 9.07 Pre 1992 (in %) (M) 0.63 0.92 0.87 8.99 9.09 9.62 9.91 9.86 Post 1993 (in %) (N) Industry- & performancematched always-separate CEO/COB titles firms 0.72 1.22b 0.42 Pre 1992 (in %) (col K minus col M) (O) −0.96 −1.29b −1.36b Post 1993 (in %) (col L minus col N) (P) Difference in means The sample includes 1,124 firms that enter the analysis with at least one year of data We split the sample into four mutually exclusive groups: (1) the set of firms that separated the titles of CEO and COB so as to come into compliance with the Cadbury Committee recommendation listed on the LSE any year over the period 1989–1996 (the split CEO/COB titles set), (2) the set of firms that maintained separate titles in all years in which they were listed over the period 1989–1996 (the always-separate CEO/COB titles set), (3) the set of firms that never split the titles of CEO and COB any year in which they were listed over the period 1989–1996 (the joint CEO/COB titles set), and (4) the remaining set of firms which comprise those that joined separate CEO and COB titles to form a single post any year in which they were listed over 1989–1996 (the create joint CEO/COB titles set) To determine into which set to classify a firm, we identified the CEO and COB each year for each firm from the Stock Exchange Yearbook Accounting information to compute ROA is taken from Datastream ROA is calculated as earnings before interest, taxes and depreciation divided by the total book value of assets Industry matching is based on FTIC Performance matching is based on ROA in year y − ROA and change in ROA for firms that split the joint CEO/COB titles and their industry- and performance-matched firms that always maintained separate titles for the CEO and COB on the LSE over the period 1989–1996 Table 198 J Dahya et al./The Financial Review 44 (2009) 179–212 Low ROA (in %) (B) Moderate ROA (in %) (C) High ROA (in %) (D) −1.39 −0.83 −0.11 0.14 0.08 9.02 8.76 8.93 8.99 8.65 a 1.28 1.53b 1.47b −0.09 −0.03 −0.37 0.42 0.47 0.43 0.79 1.15 1.92b and b indicate significance at the 0.01 and 0.05 level y − to y + y − to y + y − to y + b −1.47 −0.97 −0.68 −0.32 0.45 Low ROA (in %) (E) 0.64 0.93 1.11 8.21 8.60 8.85 9.14 9.32 Moderate ROA (in %) (F) ROA in percent) 18.88 19.21 19.30 19.35 19.31 Panel B: Change in return on assets (mean y−1 y y+1 y+2 y+3 Panel A: Return on assets (mean ROA in percent) Event year (A) Split CEO/COB titles over 1989–1996 firms Industry- & performancematched joint CEO/COB titles firms −1.56b 0.23 0.54 20.25 19.44 18.69 20.48 20.79 High ROA (in %) (G) 0.49 0.38 −0.45 Low ROA (in %) (col B minus col E) (H) −0.73 −0.96 −1.48b Moderate ROA (in %) (col C minus col F) (I) Difference in means 1.98a 0.24 −0.11 High ROA (in %) (col D minus col G) (J) 1.15 2.51a 1.57b 7.74 8.21 8.89 10.25 9.31 Pre 1992 (in %) (K) −0.33 −0.37 −0.49 8.92 8.54 8.59 8.55 8.43 Post 1993 (in %) (L) Split CEO/COB titles over 1989–1996 firms 1.23 1.16 1.34b 6.88 7.25 8.11 8.04 8.22 Pre 1992 (in %) (M) 0.02 0.69 0.84 8.96 8.94 8.98 9.65 9.80 Post 1993 (in %) (N) Industry- & performancematched joint CEO/COB titles firms −0.08 1.35b 0.23 Pre 1992 (in %) (col K minus col M) (O) −0.35 −1.06 −1.33b Post 1993 (in %) (col L minus col N) (P) Difference in means The sample includes 1,124 firms that enter the analysis with at least one year of data We split the sample into four mutually exclusive groups: (1) the set of firms that separated the titles of CEO and COB so as to come into compliance with the Cadbury Committee recommendation listed on the LSE any year over the period 1989–1996 (the split CEO/COB titles set), (2) the set of firms that maintained separate titles in all years in which they were listed over the period 1989–1996 (the always-separate CEO/COB titles set), (3) the set of firms that never split the titles of CEO and COB any year in which they were listed over the period 1989–1996 (the joint CEO/COB titles set), and (4) the remaining set of firms which comprise those that joined separate CEO and COB titles to form a single post any year in which they were listed over 1989–1996 (the create joint CEO/COB titles set) To determine into which set to classify a firm, we identified the CEO and COB each year for each firm from the Stock Exchange Yearbook Accounting information to compute ROA is taken from Datastream ROA is calculated as earnings before interest, taxes and depreciation divided by the total book value of assets Industry matching is based on FTIC Performance matching is based on ROA in year y − ROA and change in ROA for firms that split the joint CEO/COB titles and their industry- and performance-matched firms that maintained joint titles for the CEO and COB on the LSE over the period 1989–1996 Table J Dahya et al./The Financial Review 44 (2009) 179–212 199 Low ROA (in %) (B) Moderate ROA (in %) (C) High ROA (in %) (D) −1.39 −0.83 −0.11 0.14 0.08 9.02 8.76 8.93 8.99 8.65 a 1.28b 1.53b 1.47b −0.09 −0.03 −0.37 0.42 0.47 0.43 0.20 0.88 1.42b and b indicate significance at the 0.01 and 0.05 level y − to y + y − to y + y − to y + −1.52 −1.49 −1.32 −0.64 −0.10 Low ROA (in %) (E) −0.95 0.22 0.00 6.99 6.24 6.04 7.21 6.99 Moderate ROA (in %) (F) ROA in percent) 18.88 19.21 19.30 19.35 19.31 Panel B: Change in return on assets (mean y−1 y y+1 y+2 y+3 Panel A: Return on assets (mean ROA in percent) Event year (A) Split CEO/COB titles over 1989–1996 firms Industry- & performancematched created CEO/COB titles firms −2.58a −2.41a −2.24a 21.23 19.40 18.65 18.82 18.99 High ROA (in %) (G) 1.08 0.65 0.05 Low ROA (in %) (col B minus col E) (H) 0.86 −0.25 −0.37 Moderate ROA (in %) (col C minus col F) (I) Difference in means 3.00a 2.88a 2.67a 1.15 2.51b 1.57b 7.74 8.21 8.89 10.25 9.31 −0.33 −0.37 −0.49 8.92 8.54 8.59 8.55 8.43 1.41b 1.97b 1.70b 7.24 8.00 8.65 9.21 8.94 Split CEO/COB High titles over ROA 1989–1996 firms (in %) (col D Pre Post Pre minus 1992 1993 1992 col G) (in %) (in %) (in %) (J) (K) (L) (M) −1.94b −2.25a −2.17a 8.89 6.89 6.95 6.64 6.72 Post 1993 (in %) (N) Industry- & performancematched created CEO/COB titles firms −0.36 0.54 −0.13 Pre 1992 (in %) (col K minus col M) (O) 1.61b 1.88b 1.68b Post 1993 (in %) (col L minus col N) (P) Difference in means The sample includes 1,124 firms that enter the analysis with at least one year of data We split the sample into four mutually exclusive groups: (1) the set of firms that separated the titles of CEO and COB so as to come into compliance with the Cadbury Committee recommendation listed on the LSE any year over the period 1989–1996 (the split CEO/COB titles set), (2) the set of firms that maintained separate titles in all years in which they were listed over the period 1989–1996 (the always-separate CEO/COB titles set), (3) the set of firms that never split the titles of CEO and COB any year in they were listed over the period 1989–1996 (the joint CEO/COB titles set), and (4) the remaining set of firms which comprise those that joined separate CEO and COB titles to form a single post any year in which they were listed over 1989–1996 (the create joint CEO/COB titles set) To determine into which set to classify a firm, we identified the CEO and COB each year for each firm from the Stock Exchange Yearbook Accounting information to compute ROA is taken from Datastream ROA is calculated as earnings before interest, taxes and depreciation divided by the total book value of assets Industry matching is based on FTIC Performance matching is based on ROA in year y − ROA and change in ROA for firms that split the joint CEO/COB titles and their industry- and performance-matched firms that created joint titles for the CEO and COB on the LSE over the period 1989–1996 Table 200 J Dahya et al./The Financial Review 44 (2009) 179–212 J Dahya et al./The Financial Review 44 (2009) 179–212 201 operating performance when the benchmark comprises firms that always separated the combined CEO and COB position It is worth reiterating that firms with high operating performance prior to the adoption of the Cadbury recommendation do, in general, outperform all peer firms; however, the results are not always statistically significant 3.4 Changes in ROA performance from before to after December 1992 One additional question that may arise from our analysis so far is that we have not distinguished between firms that split the combined position of CEO and COB following the publication of the Cadbury Report and those that voluntarily chose this undertaking pre-December 1992 In fact, 390 firms in the adopted-Cadbury set split the combined post of CEO and COB between January 1989 and December 1992 We therefore dichotomize the sample further into those firms that came into compliance before and after December 1992 and repeat our analysis of ROA using the industry- and performance-matched firms as benchmarks for each subsample, pre-1992 and post-1993 The results are given in Table 4, columns K–P, when the benchmark comprised of the always separated In Table 5, columns K–P, we mimic the analysis for the always joined benchmark firms In Table 6, columns K–P, we repeat the analysis with the created joint CEO and COB matched firms The results on this analysis in Tables 4–6 are strikingly different and contingent on when the firms adopted the Cadbury recommendation As shown in columns K and L in Tables 4–6, firms that separated the combined position prior to the publication of the Cadbury Report, on average, report a change in ROA that is 2.51% and 1.57%, respectively, over y − to y + and y − to y + Both increases in ROA are statistically significant at the 0.05% level or less In contrast, the change in ROA for firms splitting the combined position following the issuance of the Cadbury Report, on average, exhibited poorer performance For example, the change in ROA for those firms splitting following the publication of the Cadbury Report is −0.37% and −0.49%, respectively, over y − to y + and y − to y + However, neither decline in ROA is statistically significant One interpretation of this result is that adoption of the Cadbury recommendation following its issuance led to a modest deterioration in operating performance, albeit not significant Columns M–P in Tables 4–6 report results for industry- and performancebenchmarked change in ROA for firms adopting the Cadbury recommendation in the pre-1992 and post-1993 periods The message that surfaces is fourfold First, firms creating the combined CEO and COB position following the issuance of the Cadbury Report appear to be correlated with a sudden decline in operating performance Second, firms adopting the recommendation following the issuance of the Cadbury Report appear to also be correlated with a decline in operating performance, albeit, to a lesser degree Third, firms adopting the recommendation after December 1992 perform worse than firms in the always separated and always joined sets but significantly better than firms that created the combined CEO and COB position Fourth, 202 J Dahya et al./The Financial Review 44 (2009) 179–212 voluntary separation of the combined CEO and COB position pre-publication of the Cadbury Report is rewarded with a minor uptick in corporate performance that exceeds the always separated and always combined sets (in Tables and 5, respectively) and mixed results for the created joint CEO and COB set Taken at face value, these results appear to fly in the face of the Cadbury Committee with respect to this one key recommendation The empirical evidence suggests that the selection of an optimal leadership structure appears to differ across firms and that firms seem to be able to choose their own optimal leadership structure Pressure toward the separation of the combined CEO/COB position in the United Kingdom appears to be misguided Our findings are consistent with the conjecture advanced by Dahya and Travlos (2002) and the empirical findings of Schmid and Zimmermann (2007) who fail to show any relation between leadership structure and corporate performance in a sample of 152 industrial companies traded on the Swiss Exchange in 2002 At least with respect to the combined CEO/COB position, the univariate statistics show that the recommendation of the Cadbury Committee was ill-advised 3.5 Multivariate tests of statistical significance By construction, the univariate tests control for industry factors that might affect ROA and also for the presence of mean reversion in ROA One further factor that has been shown to influence ROA is size of the firm To control for the influence that size may have on ROA and changes in ROA, we estimate regressions using the change in industry-matched and industry- and performance-matched ROA across event time as the dependent variable In addition to firm size, we also include the fractional equity ownership of the CEO and the percentage of outside directors on the corporate board as additional control variables since our descriptive statistics in Table hint a possible relation between these variables and CEO/COB configuration In the first set of regressions, we include the adopted-Cadbury firms and their matched always separated firms The independent variables are an indicator for “Adopted Cadbury & separated joint CEO and COB” firms (1), or “Always separated CEO and COB” firms (0), the log of book assets, and an indicator for pre (0)- and post (1)-December 1992 adoption The regression is estimated separately for intervals y − through y + 1, y − through y + 2, and y − through y + The regressions are estimated separately with the always separated industry-matched firms and with the always separated industry- and performance-matched firms Thus, there are six regressions, the results of which are presented in Panel A of Table The coefficients on the percentage of outside directors, firm size and the post1992 indicator are never statistically significant, and the coefficient on top executive equity ownership is negative and significant in about half the regressions in Panel A of Table We are most interested in the split CEO/COB indicator variables The coefficient on this variable is negative and insignificant in all but one regression Thus, in comparison with firms that always separated after controlling for firm size, y − to y + y − to y + Adj R2 0.0033 0.025 (0.41) −0.117 (0.04) −0.000 (0.96) −0.001 (0.97) Percentage of outside directors (%) Top executive equity ownership (%) Log assets Split CEO/COB titles Post 1993 indicator 0.0062 0.028 (0.38) −0.038 (0.26) −0.001 (0.49) −0.011 (0.33) 0.000 (0.99) −0.005 (0.65) 0.027 (0.12) −0.002 (0.86) Split CEO/COB titles over 1989–1996 (1) Intercept Panel A: Split CEO/COB titles firms matched with always-separate CEO/COB titles firms Independent variable 0.0130 0.043 (0.15) −0.031 (0.42) −0.001 (0.42) −0.001 (0.95) −0.038 (0.06) 0.023 (0.38) y − to y + Split CEO/COB titles over 1989–1996 & industry-matching firms’ changes in ROA (dependent variable = ROA) 0.0029 0.020 (0.45) −0.109 (0.05) −0.000 (0.87) −0.003 (0.84) −0.002 (0.98) 0.026 (0.18) y − to y + 0.0082 0.031 (0.34) −0.110 (0.05) −0.001 (0.64) −0.014 (0.25) −0.007 (0.57) 0.022 (0.25) y − to y + (continued) 0.0075 0.045 (0.14) −0.045 (0.33) −0.001 (0.45) −0.025 (0.28) −0.013 (0.55) 0.010 (0.71) y − to y + Split CEO/COB titles over 1989–1996 & industry- and performance-matching firms’ changes in ROA (dependent variable = ROA) The dependent variable in each regression is the change in ROA over the interval indicated in the column heading The independent variables are an indicator variable to identify whether a firm split the titles of CEO/COB (1) or not (0), the percentage of outside directors, the equity ownership of the most senior executive, the log of book value of assets and an indicator identify whether the firm split the CEO/COB titles after 1992 The sample includes 1,124 firms that enter the analysis with at least one year of data We split the sample into four mutually exclusive groups: (1) the set of firms that separated the titles of CEO and COB so as to come into compliance with the Cadbury Committee recommendation listed on the LSE any year over the period 1989–1996 (the split CEO/COB titles set), (2) the set of firms that maintained separate titles in all years in which they were listed over the period 1989–1996 (the always-separate CEO/COB titles set), (3) the set of firms that never split the titles of CEO and COB any year in which they were listed over the period 1989–1996 (the joint CEO/COB titles set), and (4) the remaining set of firms which comprise those that joined separate CEO and COB titles to form a single post any year in which they were listed over 1989–1996 (the create joint CEO/COB titles set) To determine into which set to classify a firm, we identified the CEO and COB each year for each firm from the Stock Exchange Yearbook Accounting information for ROA is taken from Datastream ROA is calculated as earnings before interest, tax and depreciation divided by beginning of year assets Log assets are the logarithm of book assets as of the beginning of the year Industry matching is based on FTIC Performance matching is based on ROA in year y − The ordinary least squares regressions are reported with robust standard errors using Huber-White-Sandwich estimates of variance p-values are reported in parentheses Regressions of the change in ROA on indicator variable for split CEO/COB titles (1) or not (0), percentage of outside directors, top executive equity ownership, log of firm assets, and an indicator variable for the split of CEO/COB titles before (0) or after (1) December 1992 for U.K industrial firms listed on the LSE over the period 1989–1996 Table J Dahya et al./The Financial Review 44 (2009) 179–212 203 y − to y + y − to y + 0.0059 0.0085 0.0143 0.011 (0.55) −0.044 (0.07) 0.000 (0.77) 0.007 (0.21) Percentage of outside directors (%) Top executive equity ownership (%) Log assets Split CEO/COB titles Post 1993 indicator Adj R2 0.001 (0.94) −0.007 (0.33) Split CEO/COB titles over 1989–1996 (1) Intercept 0.0193 0.047 (0.08) −0.038 (0.11) 0.000 (0.94) 0.022 (0.06) 0.010 (0.42) −0.007 (0.42) Panel C: Split CEO/COB titles firms matched with created joint CEO/COB titles firms Adj R2 0.039 (0.15) −0.036 (0.23) 0.002 (0.30) −0.006 (0.51) Percentage of outside directors (%) Top executive equity ownership (%) Log assets Split CEO/COB titles post 1993 indicator 0.030 (0.22) −0.042 (0.22) 0.002 (0.22) −0.014 (0.22) 0.017 (0.28) −0.027 (0.05) 0.024 (0.12) −0.012 (0.31) Split CEO/COB titles over 1989–1996 (1) Intercept Panel B: Split CEO/COB titles firms matched with joint CEO/COB titles firms Independent variable 0.0071 0.038 (0.29) −0.029 (0.18) 0.001 (0.65) 0.024 (0.06) 0.033 (0.07) −0.006 (0.50) 0.0256 0.043 (0.13) −0.050 (0.15) 0.002 (0.30) −0.025 (0.06) −0.029 (0.04) 0.016 (0.29) y − to y + Split CEO/COB titles over 1989–1996 & industry-matching firms’ changes in ROA (dependent variable = ROA) 0.0060 0.009 (0.48) −0.040 (0.09) 0.000 (0.90) 0.005 (0.38) 0.022 (0.11) −0.001 (0.97) 0.0036 0.048 (0.10) −0.013 (0.73) 0.002 (0.34) −0.013 (0.34) −0.017 (0.26) 0.008 (0.66) y − to y + 0.0055 0.050 (0.05) −0.034 (0.12) 0.000 (0.79) 0.021 (0.07) 0.039 (0.05) −0.009 (0.19) 0.0211 0.039 (0.15) −0.021 (0.65) 0.001 (0.65) −0.028 (0.02) −0.022 (0.14) 0.010 (0.58) y − to y + 0.0090 0.011 (0.55) −0.019 (0.29) 0.001 (0.54) 0.019 (0.08) 0.048 (0.02) −0.002 (0.78) 0.0175 0.050 (0.09) −0.049 (0.16) 0.001 (0.68) −0.030 (0.01) −0.028 (0.05) 0.010 (0.59) y − to y + Split CEO/COB titles over 1989–1996 & industry- and performance-matching firms’ changes in ROA (dependent variable = ROA) Regressions of the change in ROA on indicator variable for split CEO/COB titles (1) or not (0), percentage of outside directors, top executive equity ownership, log of firm assets, and an indicator variable for the split of CEO/COB titles before (0) or after (1) December 1992 for U.K industrial firms listed on the LSE over the period 1989–1996 Table (continued) 204 J Dahya et al./The Financial Review 44 (2009) 179–212 J Dahya et al./The Financial Review 44 (2009) 179–212 205 industry, prior performance, and pre-/post-1992, the change in operating performance for companies that changed the CEO and COB position so two different individuals would maintain the titles of CEO and COB was not greater than the benchmark firms The set-up of Panel B of Table mimics Panel A except that the regressions reported in Panel B include an indicator for “Adopted Cadbury & separated joint CEO and COB” firms (1) or “Always joined CEO and COB” firms (0) In Panel B, the percentage of outside directors is typically not significant, and the coefficient on book assets and top executive ownership is never significant Importantly, however, the coefficient on the split CEO/COB indicator is negative in all regressions but only significant at the 0.10 confidence level in three of the six regressions These results suggest that firms combining the CEO/COB position perform better than firms that adopted the Cadbury recommendation and subsequently split the joint CEO/COB position This casts further doubt on the motivations of the global movement toward dispensing of the combined CEO and COB position The setup of Panel C of Table mimics Panel A except that the regressions reported in Panel C include an indicator for “Adopted Cadbury & separated joint CEO and COB” firms (1) or “Created joint CEO and COB” firms (0) In Panel C, the percentage of outside directors and top executive equity ownership is typically not significant and log assets are never significant The coefficient of primary concern, the split CEO/COB indicator is positive in all regressions and is significant at the 0.10 confidence level or better in three out of six regressions and in a fourth at the 0.11 confidence level The multivariate analysis (after controlling for industry, prior performance, asset size, percentage of outside directors, top executive equity ownership, and whether the CEO/COB split occurred post 1992) of the adoption of the Cadbury recommendation does not foreshadow a consistent improvement in corporate earnings relative to peergroup benchmarks The results show that firms combining the CEO and COB titles throughout the period of our analysis perform marginally better than firms heeding to the call of the Cadbury Committee However, in firms that created the combined title of CEO/COB following issuance of the Cadbury mandate (i.e., from 1993 through 1996), there is a modest deterioration in operating performance relative to the adoption Further analysis on this rather small set of firms that created the joint CEO/COB title shows that in the majority of cases, the fusing of the positions occurs following CEO or COB departure Based on our evidence, hopes that the feel-good provision contained in the Cadbury Report to separate the combined CEO/COB position would lead to greater board oversight and corporate performance improvements are probably misplaced 3.6 Endogeneity The multivariate analysis in Table confirms that firms combining the CEO/COB positions not perform any worse than firms that adopted the Cadbury recommendation and subsequently separated the combined position However, 206 J Dahya et al./The Financial Review 44 (2009) 179–212 as we noted at the onset, one concern with studies on unitary leadership structure and corporate performance is that the two phenomena may be endogenously determined outcomes of a simultaneous process We have taken several steps to reduce that concern First, we use issuance of the Cadbury Report as the date for centering our analysis As observed in Figure 1, the period following the publication of the Cadbury Report was associated with a dramatic reduction in the number of firms combining the posts of CEO and COB (i.e., 1993–1997) Furthermore, these firms had long histories of operating with a combined CEO and COB These steps should moderate concerns about endogeneity being the cause of our results All the same, we conduct a Durbin-Wu-Hausman test (augmented regression test) for endogeneity, which determines whether the estimates from the ordinary least squares regressions in Tables are consistent and not endogenous.13 The p-value on the Durbin-Wu-Hausman test statistic rejects exogeneity at the 0.06 level The rather small and inconclusive p-value on the Durbin-Wu-Hausman test does cause some concern, thus, we take one further step We begin by estimating a linear system of two equations with change in ROA (over y − through y + 1) and an indicator for firms combining the posts of CEO and COB as the possible endogenously determined variables We employ a two-stage least squares instrumental variable regression (2SIV) In line with Boone, Field, Karpoff and Raheja (2007), Coles, Daniel and Naveen (2008), and Dahya, Dimitrov and McConnell (2008), we use the firm’s age in years and the firm’s stock return variance as instruments for the firm’s combining of the CEO and COB posts Data on the age of the firm is ascertained from the Corporate Register in each year, 1989–1996 Stock returns from 1987 through 1996 are extracted from Datastream, and the variance in returns is computed over a 24-month period prior to each year, 1989–1996, for each sample firm In line with Dahya and McConnell (2007), we use the prior year’s variables as instruments for change in ROA Specifically, we include lag change in ROA from y − to y − 1, firm size in y − 1, the percentage of outside directors on the corporate board in y − 1, and the fractional equity ownership of the CEO in y − We also include industry indicators as suitable instruments because they affect change in ROA but not affect governance (Lins, 2003; Durnev and Kim, 2005) In the first stage of the 2SIV procedure, we regress the change in ROA (over y − through y + 1) for firms that separated the titles of CEO and COB during 1989–1996 matched on industry with firms that always separated the titles of CEO and COB In these regressions, the independent variables are identical to those in the first column of Panel A in Table plus instrumental variables (and industry indicator variables), and then we regress the indicator variable for the firm’s combining the CEO and COB against the fraction of outside directors on the board and fractional CEO equity 13 Durbin-Wu-Hausman is a modified form of the Hausman test (Hausman, 1978) Green (2008) provides information on the use and specification of the Durbin-Hausman-Wu test J Dahya et al./The Financial Review 44 (2009) 179–212 207 ownership from the first column of Panel A in Table along with two instrumental variables (firm age and variance in stock returns) We then take the predicted values from the first stage as regressors in the second stage along with the same independent variables used in the first stage excluding the instrumental variables Though not shown in a table, the results on change in ROA from the second stage of the 2SIV are similar to those presented in the second column of Panel A of Table From this analysis, we are most interested in the coefficient on the split CEO/COB titles indicator from 1989 through 1996, in the second stage of the 2SIV, which is negative and statistically insignificant over all intervals with p-values that range between 0.29 and 0.64 To be thorough, we repeat the 2SIV analysis above matching changes in ROA against our two other benchmark firms that always combined the CEO/COB titles and created the joint CEO/COB title Again, the results of the 2SIV regression are close to those presented in the second column of Panels A and B of Table 7, respectively We then repeat the aforementioned 2SIV analysis, replacing change in ROA over y − to y + with change in ROA over y − to y + and then again with change in ROA over y − to y + for all three benchmark firms In short, the results on change in ROA over y − to y + and y − to y + are undifferentiated to those presented in the third and fourth columns, respectively, in Panels A, B, and C from Table As a final check on the instrumental variables regressions, we estimate a linear system of three equations with change in ROA (from y − through y + 1) and the indicator for firms combining the posts of CEO and COB as the possible endogenously determined variables We employ a three-stage least squares instrumental variable regression (3SIV) since it is more efficient than 2SIV The key difference between the 2SIV and 3SIV is that the latter takes account of any differences in the error terms of the reduced form equations being correlated by employing an additional regression stage that includes all the endogenous variables, their predicted values, and all independent variables in a generalized least squares framework (Zellner and Theil, 1962) In our analysis, the coefficients on change in ROA in the third stage of the 3SIV regressions are almost identical to those reported in the second stage of the 2SIV regressions, which would imply that our coefficients on the 2SIV are as efficient as those in the 3SIV The results on the 2SIV and 3SIV regressions are available from the authors in tabular form Sensitivity analysis and robustness tests In this section, we discuss some of the sensitivity analysis and robustness tests 4.1 Stock prices to measure corporate performance We also employed the market-to-book assets ratio to measure corporate performance The market-to-book assets ratio is a forward looking measure of corporate 208 J Dahya et al./The Financial Review 44 (2009) 179–212 performance, which presumably takes into account any future benefits associated with a change in leadership structure.14 For each firm in the sample, data on the book value of assets, the book value of equity, the market value of equity, and the book value of assets are taken from Datastream each year for which data are available 1986 through 1999 The market-to-book asset ratio is measured as the book value of assets minus the book value of equity plus the market value of equity scaled by the book value of total assets We then replace ROA with the market-to-book assets ratio in the univariate and multivariate analysis In the univariate analysis, the market-to-book asset ratio is higher in the set that separated the joined position of CEO/COB in conformance with Cadbury relative to the three benchmark sets These results on the market-to-book asset ratio appear to contradict those using ROA and stock prices as performance measures However, in the multivariate setting, the indicator variable on firms that split the combined position of CEO/COB is positive with an average p-value of 0.14 when we replace the market-to-book asset ratio for ROA Thus, it would seem that replacement of ROA with a forward looking performance measure does appear to influence the relation between Cadbury compliance and performance but not enough to overturn our key findings We also rerun the analysis using market-adjusted stock returns to measure corporate performance For each firm in the sample, market-adjusted stock returns are computed using data from Datastream each year for which data are available 1986– 1999 We then repeat our regression analysis substituting market-adjusted stock returns for ROA (over y − to y + 1, y − to y + 2, and y − to y + 3) In general, the results on stock returns are similar in sign and magnitude to those for ROA reported in Table 4.2 Stock price response to separation of the joined CEO and COB position We also run an event study on the announcement when firms split the combined CEO and COB title Announcement dates for separations of the combined CEO and COB position are taken from Extel Weekly News Summaries and annual filings maintained at Companies House For the 324 separations of the combined CEO and COB position, the two-day announcement period excess stock return is 0.21% with a p-value of 0.13 Thus, the stock price response to adoption of this key Cadbury recommendation is positive but not statistically significant at conventional levels This result does not change when we split the sample into the changes occurring before December 1992 and those after (i.e., the excess return is 0.23% in the preCadbury period and 0.19% over the post-Cadbury period) 14 Studies by Kang and Stulz (1996) and Loderer and Peyer (2002) in corporate finance employ a similar approach to ours by taking the market-to-book asset ratio or a proxy of Tobin’s Q as suggested by Chung and Pruitt (1994) and Perfect and Wiles (1994) as a forward looking measure of corporate performance J Dahya et al./The Financial Review 44 (2009) 179–212 209 4.3 Firm size effect The data in Table shows that firms that created the joint title of CEO/COB are substantially smaller in terms of size, as measured by the book value of assets and the market value of equity As noted earlier, Palmon and Wald (2002) attributed the absence of a relation between leadership structure and firm performance to firm size As they noted, one might envisage the situation where small firms benefit more from the clarity of leadership under a single CEO, while large firms might benefit more from the checks and balances associated with a separation of the two functions To examine the effect of firm size on our data we run two separate tests In the first test, we rank all the 390 firms that separated the titles of CEO and COB from 1989 through 1996 by the book value of assets and then split the set into two equal groups: 195 firms with book value of assets above the firm at the median (i.e., large firms) and 195 firms with book value of assets at the median or below (i.e., small firms) We then compare ROA over y − through y + 1, y − through y + 2, and y − through y + between the large and small subsets We find that ROA over all three intervals is higher in the subset of larger firms, though the difference between the groups is never significant at the 0.05 level or less It would seem that firm size does affect the relation between leadership structure and firm performance In the second test, we recompute the six regressions in each panel of Table with an additional independent variable that is constructed by interacting log assets with the indicator variable on whether the firm split CEO/COB titles from 1989 through 1996 In all six regressions of Panel A, the sign on the interaction variable is positive but is only statistically significant, at the 0.10 level or less, over the y − through y + intervals The same pattern emerges when we examine the interaction term in the six regressions in Panel B Interestingly, the relationship disappears when we include the interaction term in the six regressions in Panel C The sign on the interaction variable in Panel C is positive in two regressions and negative in four regressions and none of the variables are statistically significant From the analysis, firm size does seem to affect the relation between corporate performance and a change in leadership structure in the multivariate framework though the relation is not clear cut Our results are now reconcilable with prior U.K studies on changes in leadership structure in large U.K firms by Dahya, Lonie and Power (1996), Dedman (2000), and Carapeto, Lasfer and Machera (2005) that report positive announcement effects to the separation of the joint CEO/COB position 4.4 One alternative metric of corporate performance—CEO turnover To gauge the relative significance (i.e., the degree of power and stature) of the combined CEO/COB position on the board of directors, we also undertake a first-pass analysis on whether combining the CEO and COB positions influences 210 J Dahya et al./The Financial Review 44 (2009) 179–212 a board’s decision to dismiss an ineffective CEO.15 For this purpose, we conduct a simple test to check whether CEO turnover is indeed less likely in firms with a combined CEO/COB The probability of turnover (and forced turnover) in firms that combined the CEO/COB position anytime from 1989 through 1996 is higher than in firms that always separated the positions of CEO and COB The difference in the rate of turnover between these two sets of firms is 2% (p-value = 0.13) and is consistent with U.S evidence on turnover in the combined CEO/COB position reported by Goyal and Park (2002) but is at odds with Dedman (2000) who finds some evidence of entrenchment in a sample of large U.K firms that announced turnover in the joint CEO/COB position Commentary and conclusion During the 1990s and beyond, pressure has come to bear on publicly traded corporations worldwide to separate the combined position of CEO and COB so two different individuals would maintain the titles of CEO and COB A presumption that appears to underlie this movement is that splitting the combined position would significantly improve the quality of corporate board oversight and translate to better performance Dahya, McConnell and Travlos (2002) examine the relation between corporate performance and CEO turnover before and after firms adopted the key recommendations of the Cadbury Report in the United Kingdom, and they find that CEO turnover is significantly more sensitive to performance after firms came into compliance with the report by moving to three outside directors However, the study failed to show any correlation between combining the titles of CEO and COB, corporate performance, and CEO turnover Also, they remained skeptical about any direct correlation between corporate performance and board configuration In this study, we examine the decision to combine the CEO and COB positions and performance directly We find that U.K firms that split the combined title of CEO/COB in conformance with the Cadbury recommendation not show any improvement in operating performance, relative to various peer-group benchmarks, from before to after adoption of this key recommendation We also fail to show a significant stock price response at announcement time of the decision More importantly, we find that firms who always joined the titles of CEO and COB display operating earnings that are at least at par with firms separating the combined titles of CEO and COB, if not higher The results strongly suggest that recent legislature coercing the abandonment of the combined CEO and COB position appears to be off beam A concern that might be raised about our analysis is whether the results from U.K firms surrounding the issuance of the Cadbury Report can be generalized to other countries that have adopted similar Codes 15 The data for this analysis are taken from Dahya, McConnell and Travlos (2002) who examine CEO turnover in a random sample of 460 exchange-listed U.K corporations over 1989–1996 J Dahya et al./The Financial Review 44 (2009) 179–212 211 of Best Governance Practice We not address this here, but expect future studies by others will explore this question References Agrawal, A and C.R Knoeber, 1996 Firm performance and mechanisms to control agency problems between managers and shareholders, Journal of Financial and Quantitative Analysis 31, 377–397 Baliga, B.R., R.C Moyer, and R.S Rao, 1996 CEO duality and firm performance: What’s the fuss? Strategic Management Journal 17, 41–53 Barber, B.M and J.D Lyon, 1996 Detecting abnormal operating performance: The empirical power and specification of test-statistics, Journal of Financial Economics 41, 359–399 Berg, S.V and S.K Smith, 1978 CEO and board chairman: A quantitative study of dual v unitary board leadership, Directors and Boards 3, 34–49 Bhagat, S and B Black, 2002 The non-correlation between board independence and long-term firm performance, Journal of Corporation Law 27, 232–273 Boone, A.L., L.C Field, J.M Karpoff, and C.G Raheja, 2007 The determinants of corporate board size and composition: An empirical analysis, Journal of Financial Economics 85, 66–101 Brickley, J.A and C.M James, 1987 The takeover market, corporate board composition, and ownership structure: The case of banking, Journal of Law and Economics 30, 161–180 Brickley, J.A., J.L Coles, and G Jarrell, 1997 Leadership structure: Separating the CEO and chairman of the board, Journal of Corporate Finance 3, 189–220 Brickley, J.A., J.L Coles, and R.L Terry, 1994 Outside directors and the adoption of poison pills, Journal of Financial Economics 35, 371–390 Byrd, J.W and K.A Hickman, 1992 Do outside directors monitor managers? Evidence for tender offers, Journal of Financial Economics 32, 195–221 Carapeto, M., M Lasfer, and K Machera, 2005 Does duality destroy value? Working paper, City University Business School, London Chaganti, R.S., V Mahajan, and S Sharma, 1985 Corporate board size, composition, and corporate failures in retailing industry, Journal of Management Studies 22, 400–417 Chung, K.H and S.W Pruitt, 1994 A simple approximation of Tobin’s Q, Financial Management 23, 70–74 Coles, J.L., N.D Daniel, and L Naveen, 2008 Board: Does one size fit all? Journal of Financial Economics 87, 329–356 Core, J.E., R.W Holthausen, and D.F Larcker, 1999 Corporate governance, chief executive officer compensation and firm performance, Journal of Financial Economics 51, 373–406 Cotter, J.F., A Shivdasani, and M Zenner, 1997 Do independent directors enhance target shareholder wealth during tender offers? Journal of Financial Economics 43, 195–218 Dahya, J and J.J McConnell, 2007 Board composition, corporate performance, and the Cadbury Committee Recommendation, Journal of Financial and Quantitative Analysis 42, 535–564 Dahya, J and N.G Travlos, 2002 Does the one man show pay? Theory and evidence on the dual CEO revisited, European Financial Management, 6, 85–98 Dahya, J., O Dimitrov, and J.J McConnell, 2008 Dominant shareholders, corporate boards and corporate value: A cross-country analysis, Journal of Financial Economics 87, 73–100 Dahya, J., A Lonie, and D Power, 1996 The case for separating the roles of CEO and COB: An analysis of stock market and accounting data, Corporate Governance: An International Review 4, 71–77 Dahya, J., J.J McConnell, and N.G Travlos, 2002 The Cadbury committee, corporate performance, and top management turnover, Journal of Finance 57, 461–483 Daly, C.M and D.R Dalton, 1997 CEO and board chairperson roles held jointly or separately: Much ado about nothing, Academy of Management Executive 11, 11–20 Dedman, E., 2000 Executive turnover in U.K firms: The impact of Cadbury, Accounting and Business Research 33, 33–50 212 J Dahya et al./The Financial Review 44 (2009) 179–212 Denis, D.J and A Sarin, 1999 Ownership and board structures in publicly traded firms, Journal of Financial Economics 52, 187–223 Durnev, A and E.H Kim, 2005 To steal or not to steal: Firm attributes, legal environment, and valuation, Journal of Finance 60, 1461–1493 Franks, J and C Mayer, 1996 Hostile takeovers and correction of managerial failure, Journal of Financial Economics 40, 163–181 Goyal, V.K and C.W Park, 2002 Board leadership structure and CEO turnover, Journal of Corporate Finance 8, 49–66 Green, W., 2008 Econometric Analysis, 6th ed (Pearson-Prentice Hall, Upper Saddle River, NJ) Hausman, J.A., 1978 Specification tests in econometrics, Econometrica 30, 54–78 Hermalin, B.E and M.S Weisbach, 1991 The determinants of board composition and direct incentives on firm performance, Financial Management 20, 101–112 Hermalin, B.E and M.S Weisbach, 1998 Endogenously chosen boards of directors and their monitoring of the CEO, American Economic Review 88, 96–118 Jensen, M., 1993 The modern industrial revolution, exit, and the failure of internal control systems, Journal of Finance 48, 831–880 Kang, J and R.M Stulz, 1996 How different is Japanese corporate finance? An investigation of the information content of new security issues, Review of Financial Studies 9, 109–139 Kaplan, S.N and D Reishus, 1990 Outside directorships and corporate performance, Journal of Financial Economics 27, 389–410 Lins, Karl V., 2003 Equity ownership and firm value in emerging markets, Journal of Financial and Quantitative Analysis 38, 159–184 Loderer, C and U Peyer, 2002 Board overlap, seat accumulation and share repurchases, European Financial Management 8, 165–192 Mehran, H., 1995 Executive compensation structure, ownership and firm performance, Journal of Financial Economics 38, 163–184 Palmon, O and J.K Wald, 2002 Are two heads better than one? The impact of changes in management structure on firm performance by firm size, Journal of Corporate Finance 8, 213–226 Perfect, S.B and K.W Wiles, 1994 Alternative construction of Tobin’s Q—An empirical comparison, Journal of Empirical Finance 1, 313–341 Schmid, M.M and H Zimmermann, 2007 Should Chairman and CEO be separated? Leadership structure and firm performance in Switzerland, Schmalenbach Business Review 60, 182–204 Shivdasani, A., 1993 Board composition, ownership structure, and hostile takeovers, Journal of Accounting and Economics 16, 167–198 Yermack, D., 1996 Higher market valuation of companies with a small board of directors, Journal of Financial Economics 40, 185–211 Zellner, A and H Theil, 1962 Three-stage least squares: Simultaneous estimation of simultaneous equations, Econometrica 30, 54–78 ... market-to-book asset ratio is measured as the book value of assets minus the book value of equity plus the market value of equity scaled by the book value of total assets We then replace ROA with the. .. performance We also employed the market-to-book assets ratio to measure corporate performance The market-to-book assets ratio is a forward looking measure of corporate 208 J Dahya et al. /The Financial... J Dahya et al. /The Financial Review 44 (2009) 179–212 J Dahya et al. /The Financial Review 44 (2009) 179–212 205 industry, prior performance, and pre-/post-1992, the change in operating performance