Key Executive Turnover and Operational Performance in Poorly Performing Firms

Abstract

Prior research on key executive-Chief Executive Officer (CEO) has yielded mixed results as to whether CEO turnover is important for a company. This study focuses on CEO turnover in poorly performing firms. My findings suggest that financial performance is improved following CEO turnover in poorly performing firms, which is consistent with the theory that a new CEO can make a difference and CEO turnover improves post-turnover operational performance.

Introduction

As today's shareholders and boards of directors increase their expectations for key executiveChief Executive Officer (CEO) performance standards, the incidence of CEO turnover is accelerating in the U.S. and around the world. The rise and the fall of these powerful CEOs are "the new normal phenomena," according to a 2002 survey on 2500 largest publicly traded corporations conducted by Booz Alien Hamilton, a well-known consulting firm. The report of Lucier, Spiegel and Schuyt (2002) demonstrates that the turnover rate of CEOs at major corporations increased by 53% between 1995 and 2001. During the same period, the number of CEO turnover because of poor financial performance increased by 130%. In addition, the average tenure of CEOs declined from 9.5 years in 1995 to 7.3 years in 2001. In 2000, the peak year for CEO turnover, 11.2 percent of CEOs lost their positions, compared to 6 percent in 1995. In December 2002, 68 US CEOs left their positions in only one month time period, according to Challenger, Gray & Christmas, an executive-headhunting firm.

There has been considerable literature in corporate finance examining the financial performance after CEO turnover. Most studies in this field focus on event studies of the stock market reaction to CEO turnover. Financial event studies have generally used market indicators to examine immediate investor reaction to CEO turnover announcement rather than the actual cash flows generated by the firms over the first two or three years of new CEOs' tenures (Finkelstein and Hambrick, 1996). Stock price reactions around the time of CEO turnover reflect investors' expectations regarding these outcomes but do not reveal the outcomes themselves (Huson, Malatesta and Parrino, 2004). Shen and Cannella (2002) suggest that examining the impact of CEO turnover on a firm's actual operating performance, not investor reactions or market valuation, would be more accurate and reliable than market based studies.

A growing body of research addresses the issue from the perspective of accounting-based operational performance (e.g., Denis and Denis, 1995; hotchkiss, 1995; Huson, Malatesta and Parrino, 2004; Khorana, 2001; Wiersema, 2002). Daily, Certo, and Dalton (2000) argue that accounting measures reflect the current operating performance of a firm, while market measures indicate investors' perceptions of the firm's future performance potential. By examining post-CEO-turnover financial performance in poorly performing firms, I measure the relation between CEO turnover and post-CEO-turnover financial performance. I employ both pre- and post-CEO-turnover accounting and market-based financial performance. Although a substantial body of work has been done on CEO turnover, only a very few papers pay special attention to financially distressed firms (e.g., hotchkiss, 1995). My study contributes to CEO turnover literature by further investigating whether poorly performing firms benefit from CEO turnovers.

The remainder of the article is organized as follows. The next section reviews the development of literature associated with CEO turnover and hypothesis. section III describes the methodology, data collection procedures and the formation of final sample. section IV discusses the empirical results. section V presents the summary and conclusions focusing on the implications and ideas for further research.

Literature Review and Hypothesis Development

There has been considerable research on determinants of CEO turnover. Results have shown that the likelihood of CEO turnover is negatively related to firm pre-turnover financial performance measured by both accounting and market-based indicators. Hermalin and Weisbach (2002) report that a large number of papers have documented that there is a positive relation between CEO turnover and poor performance in large corporations, as well as other types of organizations. Hatfield, Worrell, Davidson, and Bland (1999) find that the firings of key executives were preceded by poor prior financial performance as measured by earnings per share. Engel, Hayes, and Wang (2003) find that accounting-based measures are more important and precise in CEO turnover decision. In summary, empirical results have concluded that deteriorating accounting performance before CEO turnover may be a cause for the top management turnover (Huson, Malatesta, and Parrino, 2004). In addition, Farrell and Whidbee (2003) argue that a CEO turnover may happen when forecasted 5-year EPS growth is low and there is greater uncertainty among analysts about the firm's long-term forecasts.

One theory on why CEO turnover matters is that boards tend to choose new CEO successors to break the existing status quo and send a positive reform message to the internal (employees and other executives) and external stakeholders (investors and stock market). Weisbach (1995) suggests that management changes are important events for corporations because they lead to reversals of poor prior decisions. He shows evidence that new CEOs often reverse the investment decisions of their predecessors. This finding is independent of the cause of the predecessor's departure. It shows that the new CEOs will adjust the corporate strategic directions and make a positive difference to firm performance. Thus, under this theory, a new leader can redirect the firm operation and improve shareholder wealth (Rowe and Davidson, 2000).

An alternative theory is that a single executive is relatively unimportant and CEO turnover does relatively little to have any effect on financial performance (e.g., Khanna and Poulsen, 1995; Warner, Watts and Wruck, 1988; Weisbach, 1988; Wiersema, 2002). According to the agency models developed by Holmström (1979) and Shavell (1979), poor financial performance is due to the bad fortune in certain time period, not bad management by individual managers. The departing CEO is the scapegoat for bad financial performance. Therefore, the change of CEO does not improve post-turnover financial performance.

Some recent empirical research seems to suggest the theory that a new CEO can make a difference and CEO turnover improves post-turnover operational performance (e.g., Denis and Denis, 1995; hotchkiss, 1995; Huson, Malatesta and Parrino, 2004; Khorana, 2001). These results are consistent with the notion that CEO turnover reflects the determination of board to change trembling financial performance. hotchkiss (1995) indicates that management turnover improves post-turnover firm financial performance. Denis and Denis (1995) find that average and median industry-adjusted return on total assets (ROA) increase over periods starting one year before and ending two or three years after CEO turnover. Khorana (2001) finds that the dismissal of poorly performing managers leads to substantial improvements in post-turnover financial performance relative to the past performance of the fund. Huson, Malatesta, and Parrino (2004) find that CEO turnover announcement abnormal stock returns are significantly positive related to subsequent improvements in operational performance after CEO turnover.

There is also some empirical evidence supporting the second theory (e.g., Khanna and Poulsen, 1995; Warner, Watts and Wruck, 1988; Weisbach, 1988; Wiersema, 2002). For example, Wiersema (2002) analyzes company performance two years before and two years after CEO turnover and finds that companies with CEO turnovers experienced no significant improvement in their operating earnings or their stock performance. In her research, operating earnings (earnings before interest and taxes) as a percentage of total assets averaged 11.2 percent two years before, and 11.8 percent two years after CEO turnover. Return on assets (ROA) averaged 2.6 percent and 2.4 percent before and after CEO turnover, respectively. In other words, operational performance did not improve after CEO turnover.

Although considerable research has been done on CEO turnover, only a very few papers pay special attention to poorly performing firms. Studying CEO turnovers in poorly performing firms is important for at least two reasons. First, CEO turnovers in poorly performing firms may be more important because improvement is more critical in poorly performing firms, compared with other firms. Will CEO turnover work under this circumstance? second, most prior turnover studies examine the turnover phenomenon in firms with a wide range of previous financial conditions (e.g., Denis and Denis, 1995; hotchkiss, 1995; Huson, Malatesta and Parrino, 2004; Khorana, 2001; Shen and Cannella, 2002; Wiersema, 2002). Focusing on poorly performing firms where change is needed can measure the power of tests that would be diluted by turnover in which no change are needed and the new CEO is hired to keep the status quo . Further, little CEO turnover research has directly focused on poorly performing firms. In CEO literature, only hotchkiss (1995) analyzes 197 firms that emerge from bankrupt protections. Although her evidence indicates that management turnover improves future performance, her sample includes only extremely financially distressed firms. As a whole, it appears that the CEO turnover literature on poorly performing firms is rather limited. I argue that it is a good direction to examine the relation between CEO turnover and post-CEO-turnover financial performance in poorly performing firms.

The continuous poor performance shows that incumbent CEOs do not have capabilities to manage firms and the boards totally lose the confidence in the incumbent CEOs. I argue that the risk of adverse selection is relatively low because boards in poorly performing firms know what kind of CEO will fit the current environment much better. Finally, CEO successors in poorly performing firms may have more mutual support from the boards, other senior executives, and the investors to initiate strategic change, thus improving post-turnover operational performance. Haddadj (1999) shows that firms undergoing CEO turnover are more likely to introduce strategic orientation than firms not undergoing turnover. I expect a positive relation between CEO turnovers and post-turnover operational performance in poorly performing firms. The above arguments lead to the following hypothesis:

Hypothesis. CEO turnovers in poorly performing firms will be positively associated with the post-turnover operational performance.

Data and Methodology

Sample and Data Collection

In most CEO turnover studies, researchers examine CEO turnover impact on financial performance without using pre-CEO-turnover financial performance as one important criterion to define sample selection, which may overstate the impact of CEO turnovers in some well-performing firms. In this study, I collect data on CEO turnover for each year from 1984 through 1999. A primary data source for CEO turnover and senior executive turnover was the senior executive list provided in the Forbes annual survey of compensation from 1983 to 2002. In addition, I use Lexis/Nexis, proxy statements, Wall Street Journal Index, and New York Times Index as a second source to identify CEO turnovers are not directly related to a takeover, spin-off or divesture. I obtain 1053 CEO turnovers between January 1, 1984 and December 31, 1999. second, I use Standard and Poor's COMPUSTAT database to find out whether these 1053 observations have low Z-scores (

An interesting feature of the Z-score model is its ability to withstand certain types of accounting irregularities and earning management. Consider the recent high-profile bankruptcy of WorldCom, in which management "cooked the book and improperly recorded billions of dollars as capital expenditures instead of as operating expenses" (Chuvakhin and Gertmenian, 2003). Such a treatment would overestimate earnings and overestimate assets. But one year before its filing for Chapter 11 bankruptcy, WorldCom's Z score is around "0.798 in 2001, much below the safe standard" (Chuvakhin and Gertmenian, 2003).

A deteriorating Z-Score can accurately signal financial trouble ahead and provide a simple conclusion than other ratios. Given its reliability, Z score is probably better used as an excellent indicator of relative financial health. This is the reason why I use low Z-scores in the one year prior to CEO turnovers as a main indicator to identify poorly performing firms. If the CEO turnover announces event taken place before July 1, the previous year's Z-score of troubled firm is used considering that the departing CEO should be blamed for poor financial performance. If CEO turnover is announced in the second half of the fiscal year, the current fiscal year's Z-score of troubled firm is used because it is the same year in which the CEO turnover event takes place. Finally, a total 101 CEO turnovers at 93 sample firms meet the low Z score requirements before CEO turnover in 1053 CEO observations between 1984 and 1999.

Table 1 presents description of the CEO turnover samples in this study. 101 of 1,053 observations that are related to the CEO turnovers have low Z scores before CEO turnovers. Consistent with the survey on CEO turnover conducted by Booz Allen Hamilton, the increasing pattern of CEO turnovers can be observed after 1993. At the same time, the poor-performance-related CEO turnovers increase, too.

The original sample consists of 1053 CEO turnovers between January 1, 1984 and December 31, 1999. To be included in my sample, I use Standard and Poor's COMPUSTAT database to find out whether these 1053 observations have low Z-scores (

Panel A of Table 2 reports the sample characteristics of original sample, which consists of 101 observations of 93 sample firms that had CEO turnovers between January 1, 1984 and December 31, 1999. The distribution of the sample is reported in Panel A by two-digit Standard Industrial Classification (SIC) code. CEO turnovers in poorly performing firms are heavily concentrated in regulated industries, such as electric and gas Services. Transport, computer hardware and software have higher percentage of CEO turnovers in poorly performing firms in this study.

Panel B of Table 2 contains summary statistics for all 101 observations of 93 sample firms that had CEO turnovers between January 1, 1984 and December 31, 1999. Panel B of Table 2 also shows that the sample firms are typically very large publicly traded firms which average $ 4.186 billion in annual sales, 21,930 employees, and $ 8.948 billion in total assets at one year before CEO turnovers.

Methodology

In the CEO turnover literature, return on assets (ROA) is often selected as the financial measurement because it is a well-understood and widely used accounting measure of operating performance (Zajac, 1990). To identify the sources of any changes in the operating performance after CEO turnover, I use four primary financial performance measurements in this study: operating income return on assets (OROA); operating income return on sales (OROS); Z score, which is a much broader financial ratio to reflect the financial health of a firm and finally; the market-based price (market) to book ratio. OROA is obtained by dividing operating income by the average of beginning- and ending period book value of total assets.

I define the CEO turnover year as Year O in this study. The sample time period is from one year before CEO turnover until 2 years after CEO turnover (Year -1 to Year +2). I identify each CEO turnover year by examining the exact CEO turnover news event at the different time periods of a certain fiscal year. To help ensure that improvements in performance are not industry-driven, I adjust all above financial ratios by using the performance-based control group matching method described in Barber and Lyon (1996). The industry comparison group will isolate the industry effects of different sample firms, which can provide a more reliable financial analysis of pre- and post-CEO-turnover financial performance. I follow the Barber and Lyon (1996) method using two characteristics to establish my industry comparison group for each sample. First, I identify all firms in the same four-digit or two-digit SIC code. second, I apply 90% to 110% filter of preCEO-turnover Z score at Year -1 to identify the industry comparison group. This -10% to +10% range yields more well-specified statistical results than other alternative filters in these sampling situations (Barber and Lyon, 1996). Each sample firm is matched to comparison firm with the same COMPUSTAT SIC code. Comparison firms' Z score performance measures over one year before CEO turnover are in the range of -10% to +10% of the sample firm's Z score performance. Eventually, I am able to identify an industry comparison group for each sample firm. Finally, out of 101 observations, I construct 83 industry comparison groups for 83 sample firms.

For all of my tests, I use the same industry comparison firms over time. Thus, the matched industry comparison firms selected for the sample firms in year prior to CEO turnover remain the same from Year -1 to Year +2. Following the industry matching methods described in Barber and Lyon (1996), I use each median performance of the industry comparison group to avoid extreme outliners. Each sample firms' performance is adjusted by subtracting the median performance of its industry control group as the matched performance adjusted measure. A comparison of the post-CEO-turnover values with the pre-turnover benchmark measures the impact of a CEO turnover on firm performance. Changes in pre- and post-CEO-turnover financial performance are examined on both unadjusted basis and matched industry-adjusted basis. Matched industry-adjusted comparisons allow me to examine performance changes in CEO-turnover firms irrespective of any industry-wide factors that may be affecting operating performance.

Empirical Results

Financial Performance Change

Table 3 contains the comparisons of the financial performance across the 3 years around CEO turnover, t and t+1 and t+2. Table 3 compares the financial performance of sample firms with CEO turnovers with similar financial measurements of the matched industry control groups. I make four sets of comparisons from Year O to Year +2: operating income return on assets (OROA); operating income return on sales (OROS); Z score, and finally; the market-based price (market) to book ratio. The year of CEO turnover refers to Year 0. The following year of CEO turnover is Year +1, etc. The first set of columns in the table includes unadjusted financial performance. The second set of columns in the table shows the matched industry control group performance. The last set of columns in the table reports the industry-adjusted financial performance.

Panel A of Table 3 reports the change in OROA from Year 0 to Year +2, which measures the effectiveness of the firm's to create operating income following CEO turnover. The changes in unadjusted OROA increase from 11.57% in Year O to 11.85% in Year +1, which is significant at 5 percent level. I use a nonparametric procedure to compare OROA in Year +1 or in Year +2 with OROA in Year O individually. I perform a two-tail Wilcoxon sign-rank test to determine the significance level of differences between OROA in Year +1 or in Year +2 and OROA in Year O in Table 3.

Table 3 presents the sample firms' financial performance from Year O to Year +2 between January 1, 1984 and December 31, 1999. The year of CEO turnover refers to year O. The following year of CEO turnover year is year +1. To make sure that improvements in performance are not industry-driven, I adjust 4 major financial ratios by using the performance-based control group matching method described by Barber and Lyon (1996). The industry comparison group will isolate the industry effects of different sample firms, which can provide a more reliable financial analysis of pre- and post-CEO-turnover financial performance.

I follow the Barber and Lyon (1996) method using two characteristics to establish my industry comparison group for each sample. The first is the four-digit Standard Industrial Classification (SIC) code and the second is the pre-CEO-turnover Z score in Year -1. First, a matched industry comparison group is constructed for each sample firm on the basis of four-digit Standard Industrial Classification (SIC) code. second, 90% to 110% filter of pre-CEO-turnover Z score at Year -1 is employed to identify the portfolio firms in industry comparison group. If there is no individual firm with Z score in Year -1 in with the same four-digit SIC code, I match pre-CEO-turnover Z score performance in Year -1 within the 90% to 110% filter by using all firms with the same two-digit SIC code. Finally, out of 101 observations, only 83 industry comparison groups were constructed for 83 sample firms. The final remaining sample is composed of the set of 83 sample firms from 1984 to 1999.

Four financial ratios are used as primary financial performance measurements in this paper: operating income return on assets (OROA); operating income return on sales (OROS); Z score, which is a much broader financial ratio to reflect the financial health of a firm and finally; the market-based price (market) to book ratio. OROA is obtained by dividing operating income by the average of beginning- and ending period book value of total assets. P-values for significance of differences between sample firms in Year +1 or in Year +2 and sample firms in Year O are calculated using a Wilcoxon sign-rank test for medians. Significance tests are based on two-tail Wilcoxon sign-rank test with significance level indicated by 0.01, 0.05, or 0.10. All accounting and financial data are from Compustat.

I measure the impact of CEO turnover on financial performance in the poorly performing firms by applying the more powerful matched industry comparison group method suggested in Barber and Lyon (1996). I detect abnormal improvement of operational return in this study. For example, the median matched adjusted OROA of sample firms improves from the level below industry benchmark (-0.18%) in Year 0 to the level above industry benchmark (0.3%) in Year +1. The Wilcoxon signed-rank test between industry-adjusted OROA in Year +1 and industry-adjusted OROA in Year 0 is significant at 5 percent, which is consistent with the result of Wilcoxon signed-rank test between unadjusted OROA in Year +1 and unadjusted OROA in Year 0. Here, my results of Wilcoxon signed-rank test indicate that both unadjusted OROA and industry-adjusted OROA have a positive improvement one year following CEO turnover. This evidence supports hypothesis that CEO turnovers are positively related to post-CEO-turnover operational performance in poorly performing firms.

Panel B of Table 3 shows the change in OROS from Year 0 to Year +2, which also measures the efficiency of the firm's operations following CEO turnover. The unadjusted OROS of sample firms increases from 30.19% in Year 0 to 31.46% in Year +1. The result of Wilcoxon signed-rank test between unadjusted OROS in Year +1 and unadjusted OROS in Year 0 is significant at 10 percent, which provides some evidence to indicate that unadjusted OROS improves one year following CEO turnover. Therefore, the result of Wilcoxon signed-rank test between unadjusted OROS in Year +1 and unadjusted OROS in Year 0 provides some evidence consistent with hypothesis.

Panel C of Table 3 contains the results of the change in Z score from Year 0 to Year +2, which measures a much broader financial performance change following CEO turnover. The change in unadjusted Z score increases from 1.2397 in Year 0 to 1.2631 in Year +1. The result of Wilcoxon signed-rank test between unadjusted Z score in Year +1 and unadjusted Z score in Year 0 is significant at 10 percent. In Year +2, unadjusted Z score is 1.3795, an impressive improvement from 1.2397 in Year 0. The result of Wilcoxon signed-rank test between unadjusted Z score in Year +2 and unadjusted Z score in Year 0 is significant at 1 percent. The continuous improvement in Z score shows that new CEOs make positive efforts to change financially distressed conditions, which are created by the unprofitable investment projects developed under the leadership of departing CEOs. The overall financial performance measure (Z score) in sample firms outperforms the industry benchmark in one year following CEO turnovers. Beginning from Year 0, troubled firms move to the right direction due to the CEO turnover and related strategic changes, which brings mean reversion trends to the meltdown of financial performance. In summary, these results provide some evidence to hypothesis.

As an alternative to three accounting-based financial performance measurements, Panel D of Table 3 presents the change in market to book ratio from Year 0 to Year +2. Market to book ratio is a market determined measure and unrelated to accounting-based operational performance. But market to book ratio measures how investors think about the change in financial performance surrounding CEO turnover. The unadjusted market to book ratio of sample firms improves from 1.5473 in Year 0 to 1.6545 in Year +2. The result of Wilcoxon signed-rank test between unadjusted market to book ratio in Year +2 and unadjusted market to book ratio in Year 0 is significant at 1 percent level. In addition, the median matched adjusted market-to-book ratio of sample firms improves from 0.0695 in Year 0 to 0.1073 in Year +1. The result of Wilcoxon signed-rank test between industry-adjusted market-to-book ratio in Year +2 and industry-adjusted market-to-book ratio in Year 0 is significant at 10 percent. The changes in industry-adjusted market to book ratio are positive, the result of Wilcoxon signed-rank test between industry-adjusted market-to-book ratio in Year +2 and industry-adjusted market-to-book ratio in Year 0 provides some evidence to show the improvement in market sentiment and optimism to poorly performing firms.

In summary, the empirical evidence in Table 3 is consistent with some previous studies that CEO turnover improves post-turnover operational performance (e.g., Denis and Denis, 1995; hotchkiss, 1995; Huson, Malatesta and Parrino, 2004; Khorana, 2001). As mentioned earlier, hypothesis predicts a positive association between CEO turnover in poorly performing firms and post-turnover operational performance. My results suggest that both unadjusted OROA and industry-adjusted OROA experience a positive improvement one year following CEO turnover. Further, unadjusted OROS and Z score tend to improve subsequently one year following CEO turnover. Thus, hypothesis is supported.

Conclusion

My findings suggest that financial performance is improved following CEO turnover in poorly performing firms, which is consistent with the theory that a new CEO can make a difference and CEO turnover improves post-turnover operational performance (e.g., Denis and Denis 1995; hotchkiss 1995; Huson, Malatesta, and Parrino 2004; Khorana 2001). I find that both unadjusted OROA and industry-adjusted OROA experience a positive improvement one year following CEO turnover. Further, the changes in unadjusted OROS and Z score tend to improve subsequently two years following CEO turnover. These results, overall, support the view that top management changes are important events for corporations because they lead to reversals of poor prior decisions (Weisbach, 1995). Clearly, new CEO successors may break the existing status quo and send a positive reform message.

This study advances understanding of CEO turnover in poorly performing firms, but it has limitations. First, due to the definition of poorly performing firms in this study, the sample size is relatively small. Second, one regulated industry (electric and gas services) has a higher weight in total sample. Third, although I use a long sample time period and performance-based industryadjusted matching method suggested in Barber and Lyon (1996) to measure post-turnover operational performance, which also may be influenced by mean-reversion of the accountingbased performance time series rather than CEO turnover. So the findings and implications of this study must be considered in light of its limitations. Future research should also be continued at CEO succession planning area in poorly performing firms.

* I am grateful to Wallace Davidson for his valuable guidance, encouragement and kindness. I also thank Lyle Wallis and other anonymous reviewers for their help.

By: Lei Wen, Ph.D.

Lei Wen, Ph.D. is an Assistant Professor of Finance at the H. W. Siebens School of Business at Buena Vista University, 610 W. Fourth Street, Storm Lake, Iowa 50588

Copyright Credit Research Foundation Second Quarter 2005
Provided by ProQuest Information and Learning Company. All rights Reserved