Pay-for-performance relationship:
Difference between REITs and other financial
companies: pre- and post-crisis
Alexander Tholenaar
Supervisor: Erasmo Giambona
University of Amsterdam
Faculty of Economics and Business
Master’s thesis Finance: Asset management & Real estate
Abstract: This paper investigates the difference in CEO compensation between REITs and other financial companies pre- and post crisis. Panel data regressions are performed for stock returns on four incentive measures. Results show that REITs got higher stock returns when paying CEOs more compensation relative to other financial companies pre-crisis, however this premium disappears post-crisis. After several robustness tests, it was concluded that the results are robust to changes in the dependent variable, the inclusion and exclusion of control variables, and the different treatment of outliers. The conclusion is made that there exists a relationship between CEO compensation and shareholder interests, however not always positive.
2
Statement of Originality
This document is written by Student [Alexander Tholenaar] who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.3
1. Introduction
Executive compensation is a well-discussed subject in both the media and academic literature. This was especially so during and after the financial crisis of 2008. Chief executive officers (CEOs) continued to receive high compensation while others lost their jobs during the financial crisis. There is an argument, however, that executives did not receive proper incentives, which led to the crisis in the first place. According to those in favor of this argument, executive compensation was too short-term. This led the Obama administration to discuss how it could change compensation schemes across the financial service industry and align pay more closely to long-term performance. Murphy (1999) says that stock ownership provides the most direct link between shareholder and CEO wealth. Yet the literature claims that financial companies with higher equity compensation perform worse than those with lower equity compensation (Fahlenbrach and Stulz, 2011). The question then is how executives can be best incentivized through compensation. This paper investigates how closely the interests of CEOs in the financial service industry were aligned with the interests of shareholders before and after the crisis. An interesting sub-industry is that of the real estate investment trusts (REITs). These are highly regulated and must meet several conditions. For instance, they must distribute 90% of their earnings as dividend to be granted corporate tax exemption. This limits the amount of free cash flow to REIT managers, and reduces the principal-agent problem. Moreover, when most earnings are distributed to shareholders, shareholders may want to reward managers for their efforts toward cost reduction to increase earnings, and hence increase dividends. These features make it interesting to examine REITs. Vemala et al. (2014) and Sonenshine et al. (2016) found a change in the composition of executive compensation following the financial crisis relative to those before the financial crisis. They found that the composition changed so that cash compensation decreased and equity-based pay increased. Based on these findings, one can assume that the compensation scheme became more in accord with shareholder interests. The aim of this study is to discover whether REIT executive compensation schemes better align with the interests of shareholders compared with non-REIT financial companies, and if this premium, if it existed pre-crisis, disappeared post-crisis.4 To do so, this paper uses panel data regressions on buy-and-hold returns (BAHR) and various incentive measures for the period 2000-2016. Pre-crisis results indicate that two of the four measures for REITs are significantly positively related to returns, and the other two measures are insignificant. There is evidence that REITs that paid higher compensation received higher returns relative to other financial companies, and there is no evidence that they received lower returns pre-crisis. Post-crisis results indicate that two of the measures are significantly negatively related to returns; however, these are only significant at the 10% level; the other two measures are insignificant. There is little evidence that REITs that paid higher compensation received lower returns relative to other financial companies, and there is no evidence that they received higher returns post-crisis. The results indicate that there exists a premium, pre-crisis, in-stock return for REITs that paid their managers more compensation relative to other financial companies. However, post-crisis, this premium disappears. The paper proceeds as follows. In the next chapter, I discuss previous papers related to the pay-for-performance discussion. In Chapter 3, the data collection is explained, and some summary statistics are presented. Chapter 4 explains the methodology used in this paper. Results of the various regressions are discussed in Chapter 5. Finally, conclusions are drawn in Chapter 6.
2. Literature review
The problem of executive pay in relation to shareholder interest is a classic example of the principal-agent problem. This problem occurs when a person (the agent) or entity makes decisions on behalf of (or that has impact on) another person or entity (the principal), but the agent is motivated to operate for their own interest, which contrasts with the interests of the principal. In the context of executive pay, the CEO is the agent and the shareholders are the principals. Surprisingly, little research has been done about the relationship between executive pay and shareholder interest. Most papers related to this topic examine the reverse relationship: how stock return influences executive pay. However, this paper investigates how executive pay influences stock return, and whether higher executive pay leads to higher stock returns. Most of the studies that investigate the influence of executive pay on firm performance use various accounting measures, such as earnings5 or return on equity. Although these measures are of interest to shareholders, these are not directly related to stock return, which is the ultimate interest of a shareholder. One of the first to discuss the relationship between executive pay and stock return was Masson (1971). Using a sample of 39 firms, Masson (1971) found that firms with CEOs whose compensation better paralleled stockholders’ interests performed better in the stock market in the post-war period. He concludes that his work indicates that a primary goal of managers is the stock performance of their firm. Abowd (1990) investigated, using a sample of 250 large firms, whether the sensitivity of manager compensation with respect to corporate performance is related to corporate performance in the following year. He found that a 10% increase in payment is associated with a 400 to 1,200 basis point increase in shareholder return. Lewellen, Loderer, Martin and Blum (1992) examined the compensation of the three highest paid executives from a sample of 49 Fortune 500 industrial firms, stating that higher pay leads to higher market returns. They conclude that compensation packages are designed to reduce agency costs. Mehran (1995) was one of the first who focused more on the structure of executive pay rather than the level of compensation. He found that equity-based pay was positively related to firm performance. He concludes that the form, rather than the level, of compensation is what motivates managers to increase shareholder wealth. A 1996 study by McConaughy and Mishra found that increasing pay-performance sensitivity increases the excess return of firms with poor prior performance. Cooper et al. (2014) examined the link between incentive pay (in which they define incentive pay as the payment of restricted stock), options and other long-term compensation, and future stock return. After sorting firms annually into deciles based on CEO compensation, they found a strong, negative relationship between annual pay and future returns. Their results suggest that incentives that are meant to align executive interests with shareholder interests do not translate into higher returns. They also found that this effect is stronger at firms with weaker corporate governance, and at firms in which the CEOs are overconfident. Their explanation for this is that CEOs who are overconfident and who receive high excess pay undertake activities such as overinvestment, which lead to shareholder wealth losses. Malmendier and Tate (2009) investigated the relationship between so-called “superstar,” award-winning CEOs and firm performance. They found that firms with award-winning CEOs subsequently underperform in stock performance as well as in operating performance, and, at the
6 same time, award-winning CEOs receive higher compensation. Malmendier and Tate (2009) conclude that the superstar system has negative consequences for shareholders. Fahlenbrach and Stulz (2011) examined the relationship between bank performance during the recent credit crisis of 2008 and executive compensation immediately before the crisis. They found some evidence that firms with a more equity-based compensation performed worse during the crisis. The REIT industry, in contrast to other industries, is highly regulated. For a company to receive REIT status, and to have corporate tax exemption, several conditions must be met. They must have at least 100 shareholders and no more than 50% of the shares held by 5 or fewer shareholders. They must have at least 75% of their total assets invested in real estate, cash or government securities, and 75% of their gross income needs to be derived from rents or mortgage interest. Finally, and perhaps the most important rule in the principal-agent context, REITs must pay at least 90% of their taxable income as dividends. This last rule changes the principal-agent problem. The effect that this 90% pay-out rule has is that the amount of free cash flow is relatively small for REITs. This reduces the free-cash-flow problem introduced by Jensen (1986). Jensen found that free cash flow allows managers to finance projects that generate returns below the cost of capital, which reduces value. Alongside this argument that makes it interesting to examine REITs, Chopin, Dickens and Shelor (1994) argue that examination of REITs is interesting because, under the previously mentioned condition, owners who want to maximize distributed profit for any given level of sales should reward managers for their efforts toward cost reduction. Using data from 1998 to 2005, Lewis and Terrel investigated whether CEO compensation affects profit efficiency in the case of REITs. The writers argue that much research has been done on this topic, but, for REIT CEOs, there is a gap in the academic literature. Previous research indicates that incentive-based compensation improves firm performance, as measured by sales revenue, economic value added and the ratio of net income to share price. Although academicians have found this positive link between performance measures and compensation, Lewis and Terrel found no difference in profit efficiency between different CEO pay levels. Moreover, they found no evidence that high, incentive-based compensation packages create higher profits. Pennathur and Shelor (2002) investigated the relationship between increases in CEO compensation and stock return from 1993 to 1999, using a sample of REITs. In contrast to Lewis and
7 Terrel, they found evidence that compensation increases are related to stock return. Chopin et al. (1994) found some evidence that is positively related to revenue. Noguera and Highfield (2005) tested the effect of incentive-based compensation on REIT performance using the alignment of interest hypothesis. In contrast to what they expected, they did not find a positive relationship between compensation and performance. Instead, they found a negative relationship. Bauer, Eichholtz and Kok (2009) used a unique and leading corporate governance database to test whether corporate governance influences the performance of U.S. REITs. Using a sample of 220 REITs, they found a significant relationship between firm level governance and REIT firm value, but only for REITs with low pay-out ratios. They performed the same analysis on a sample of 5,000 firms, finding a strong, significantly positive relationship, indicating that the weak relationship between governance and performance in the real estate industry is explained by what they call a REIT effect. Vemala et al. (2014) and Sonenshine et al. (2016) found a change in the composition of executive compensation following the financial crisis relative to those before the financial crisis. They found that the composition changed so that cash compensation decreased and equity-based pay increased. Based on these findings, one can assume that the compensation scheme moved toward greater accordance with shareholder interests. The contribution of this paper to the existing literature is twofold: First, this paper makes the distinction between REITs and other financial companies. Although REITs are also financial companies, they must follow distinct rules that make the group more homogeneous. Hence, REITs can be considered a special industry within the financial company industry. In the pay-for-performance context literature, REITs received some attention in the 2000s. However, many of these papers found weak evidence, and, moreover, many contradicted each other. Besides, these studies only investigated the pay-performance relationship for REITs. It is much more interesting, however, to compare the results of this special (REIT) industry with those of the financial industry in general. This way, we can investigate whether these rules (that REITs face) can help to solve or reduce the problems that occur when owners pay managers based on performance. The second contribution is that this paper compares the pre-crisis pay-for-performance relationship with the pay-for-performance relationship post-crisis.
8 3. Data Data on executive compensation were collected from the Compustat Execucomp database. Data on company performance were found in the CRSP database. Accounting data can be found in the Compustat database. Companies can be selected based on their standard industrial classification code. For financial companies, these are between 6000 and 7000; and for REITs, it is 6798. Because REITs are in both classification ranges, they are excluded from the sample of financial companies. For a comparison before and after the financial crisis, data were collected for the years 2000 to 2016, so that there are two equal sub-samples, as the crisis occurred in 2008. Following Vemala et al. (2014), the data between 2000 and 2007 illustrate the pre-crisis data, and 2009–2016 is considered to be the post-crisis data. I use four incentive measures: two short-term and two long-term. The short-term incentive measures are percentage of equity pay (calculated as all yearly stock and option awards divided by the total compensation multiplied by 100) and the yearly change in total compensation. Following previous papers, the first long-term incentive measure is the dollar change in the value of a CEO’s stock and options holdings that would result from a one percentage point increase in the company stock price, calculated as follows: 𝐷𝑜𝑙𝑙𝑎𝑟 𝑔𝑎𝑖𝑛 𝑓𝑟𝑜𝑚 + 1% = 0.01 ∗ 𝑃𝑟𝑖𝑐𝑒!"∗ 𝑆ℎ𝑎𝑟𝑒𝑠!"+ 𝑂𝑝𝑡𝑖𝑜𝑛𝑠!" (1) The other long-term incentive measure, also following previous papers, is the CEO’s percentage firm ownership, calculated as the sum of all shares (restricted and unrestricted) and options (exercisable and unexercisable) held by the CEO, divided by the total shares outstanding and then multiplied by 100. Table 1a provides a summary of statistics of accounting variables for a 2001 sample of financial firms for the fiscal years 2000-2007. Table 1b shows the same statistics for a sample of 2,230 financial firms for the fiscal years 2009-2016. Both samples cover large financial firms. This is not surprising, because Cadman, Klasa and Matsunaga (2010) prove that Execucomp is biased toward large firms. The average total asset value is $62.3 billion for 2000-2007, and $68.7 billion for 2009-2016. Although the
9 Table 1a Sample summary statistics for fiscal years 2000-2007. The table provides a summary of statistics for key variables from a 2001 sample of financial companies for the fiscal years 2000-2007. The data are from the Compustat Bank Fundamentals annual databases. All accounting variables are measured in millions of dollars. Table 1b Sample summary of statistics for the fiscal years 2009-2016. The table provides a summary of statistics for key variables from a sample of 2230 financial companies for the fiscal years 2009-2016. The data are from the Compustat Bank Fundamentals annual databases. All accounting variables are measured in millions of dollars. average book value of equity increased, average market capitalization decreased from $10.6 billion for 2000-2007 to $9.8 billion in 2009-2016. In both periods, the average book-to-market ratio is below 1, indicating that, on average, the firms in both samples are overvalued. In the later period, firms are slightly less overvalued compared with the first period. Examining the return on equity and return on assets, there are few investment grade firms in the sample. While most investors seek a minimum return on equity of between 12% and 15%, and a minimum return on assets of 5%, in the first sample period, average return on assets was 2.8% and the average return on equity was 13%. In the second sample period, average return on assets and the average return on equity decreased to 2.2% and 4% respectively. Table 2 provides a summary of statistics of CEOs’ annual compensation, annual bonus, equity portfolio and four compensation incentive measures. The first two
Lower Upper Standard
Variable Number Minimum Quartile Median Quartile Maximum Mean Deviation Total liabilities 2001 23.97 2476.9 7040.6 29,838.6 2,074,033 56,934.0 170,325 Total assets 2001 140.6 3549.3 8938.0 34,114.4 2,187,631 62,311.1 182,059 Market capitalization 2001 4.595 1162.5 3032.3 8674.3 273,598.1 10,620.3 25,550.8 Return on assets 2001 -41.86% 0.98% 1.49% 3.51% 41.89% 2.77% 4.70% Return on equity 1992 -496.61% 9.42% 13.46% 17.32% 190.10% 12.99% 19.67% Cash/total assets 1930 -0.504% 9.46% 2.30% 4.09% 49.30% 4.14% 6.12% Book-to-market ratio 1992 -34.02 0.38 0.52 0.69 10.98 0.56 0.98
Lower Upper Standard
Variable Number Minimum Quartile Median Quartile Maximum Mean Deviation Total liabilities 2230 30.47 2348.4 6350.2 19,688.6 2,341,061 60,253.6 233,224 Total assets 2230 140.4 3652.0 8725.4 24,074.6 2,573,126 68,749.4 258,633 Market capitalization 2230 10.22 1084.2 2694.1 7350.9 307,295.1 9801.4 24,840.7 Return on assets 2230 -50.8% 0.69% 1.21% 3.26% 50.1% 2.22% 4.20% Return on equity 2229 -4139.3% 4.82% 8.47% 11.9% 548.3% 3.97% 103.8% Cash/total assets 2148 0% 0.97% 1.82% 4.28% 54.4% 4.22% 6.79% Book-to-market ratio 2229 -14.33 0.49 0.72 0.99 6.90 0.79 0.74
10 columns provide statistics for the fiscal years 2000-2007, and the last two columns for the fiscal years 2009-2016. The average total compensation decreased from $6.5 million in the first period to $5.6 million in the second period. This is mainly the result of a large decrease in option grants. Average (median) option grants decreased from 2,307.5 (499.9) to 564.3 (0), this is a percentage decrease of 75% (100%). Interestingly, while option grants decreased, average (median) stock grants increased from 1,372 (0) to 2,534.9 (1,196.2), representing a percentage increase of 85% (100%). Initially, most equity compensation came from options, but now the larger part of equity compensation comes from shares. Most of the compensation comes from performance-based pay, while the base salary was only 11% in the first period and 14% in the second period. Table 2 Executive compensation. The table provides a summary of statistics for key CEO compensation variables from two samples of financial companies for the fiscal years 2000-2007 and 2009-2016. The data are from the Compustat ExecuComp database. Values are reported in thousands of dollars. Most of the variables in the table are taken directly from ExecuComp. ‘‘Cash bonus’’ is defined as the sum of bonus and non-equity incentive award pay-outs; ‘‘Dollar gain from +1%’’ is equal to the dollar change in the executive’s stock and option portfolio value for a 1% change in the stock price; “Percentage equity pay” is the sum of stock and option grants divided by total compensation times 100; and “Annual change in compensation” is defined as the annual percentage change in total compensation. 2000-2007 2009-2016 Median
Mean Median Mean Annual compensation Total compensation 3285.2 6589.2 3658.5 5675.0 Salary 700 718.1 789.8 830.2 Cash bonus 760 1823.5 940.5 1516.4 Stock grant ($) 0 1372.0 1196.2 2534.9 Option grant ($) 499.9 2307.5 0 564.3 Other compensation 81.6 354.8 78.1 226.8 Cash bonus/salary Bonus 1.2 2.5 1.2 1.9 Total bonus 2361.2 5503.0 2618.9 4615.6 Cash bonus 760 1823.5 940.5 1516.4 Equity bonus 1291.7 3679.5 1500 3099.2 Total bonus/salary 3.4 7.1 3.5 521.2 Cash bonus/total bonus Equity portfolio 0.4 0.4 0.4 0.4 Value of total equity portfolio 29,102.8 134,962 18,612.5 54,634.9 Value of shares 18,683.9 105,700.2 13,596.8 44,077.5 Value of exercisable options 3142.8 16,559.7 112.4 5600.5 Value of unexercisable options 630 12,194.9 0 1239.9 Value of unvested stock 63.8 4710.4 1729.8 4493.6 Value of total equity portfolio/total compensation 7.4 49.2 4.7 5214.6 Value of shares/salary Incentive measures 26.6 167.9 17.4 35,389.0 Percentage equity pay 42.4 40.2 43.5 41.5 Percentage ownership 1.5 3.1 0.9 1.8 Dollar gain from +1% 518.4 1612.9 267.5 717.9 Annual change in compensation 5.0 46.7 7.3 34.9
11 To align CEO incentives and shareholder interests, CEOs are rewarded for achieving accounting-based goals in cash and equity bonuses. An analysis of these bonuses can be seen in the second section of the table. Again, in this section of the table, most of the compensation comes from performance-based pay. In 2000-2007, however, the average (median) of the total bonus was 7.1 (3.4) times larger than the base salary, and in 2009-2016 it was 521.2 (3.5) times larger. In both periods, in the mean and in the median, equity bonus is the larger part of total bonus; 40% of the annual bonus was paid in cash. The average value of a CEO’s total equity share is $134 million in the first period. In the second period, there is a decline of more than 50% of this value. This is probably because these companies suffered huge losses during the financial crisis, which resulted in falling stock prices. In both periods, about 80% of the total equity portfolio held by CEOs consisted of shares. The fact that in the second sample period option-based compensation declined can also be seen by the value of exercisable and unexercisable options held by CEOs in their total equity portfolio; most options have an expiration date. Overall, the total equity portfolio is large compared with the annual compensation, as the median ratio of the value of total equity portfolio to annual compensation is 7.4 in the first sample and 4.7 in the second sample period. Therefore, one could assume that CEO incentives mostly come from the accumulated equity portfolio, and not from annual compensation, as Hall and Liebman (1998) and Core and Guya (1999) already pointed out. So, initially, it looks like “Percentage ownership” and “Dollar gain from +1%” are better incentive measures. All four incentive measures are presented in the final section of Table 2. Percentage equity pay is around 40% in both sample periods. Average CEO ownership was 3.1% in the first period and declined to 1.8% in the second period. This means that for every $1,000 increase in common equity, the average CEO’s wealth increased by $31 in the first period and $18 in the second. The average dollar gain for a percentage increase in stock price is $1.6 million in the first period and $0.7 million in the second. In Table 3, the average stock return for both periods is illustrated. In the first period, the average return is 3.7%, and in the second period it is 2.1%. This means that, on average, CEOs gained $5.9 million in the first period, and $1.5 million in the second, in wealth through the increase in stock price. The final measure measures the yearly percentage change in total compensation, which was, on average, 46.7% in the first period and 34.9% in the second.
12 Table 3 Return statistics. The table provides a summary of statistics for key variables from a sample of 2,230 financial companies for the fiscal years 2009-2016. The data are from the Compustat Bank Fundamentals annual databases. All accounting variables are measured in millions of dollars.
4. Methodology
This thesis investigates the relationship between CEO compensation and shareholder return. To do so, it is necessary to explain some of the incentive measures of compensation that could influence shareholder return. As shareholder return is generated through stockholders’ equity, and equity pay is part of a CEO’s bonus system, it would be logical to assume that incentive measures related to equity compensation are good incentive measures. One logical measure is the percentage of equity pay relative to total annual compensation; this is a short-term measure as it only considers the equity compensation in that specific year. However, most of the equity bonus will only be vested after some years, so it might be better to use a measure that also takes into account the equity bonuses from previous years, in other words, a long-term measure. I use two of these measures, which have also been used in previous literature. The first is the percentage ownership of the CEO in the company, calculated as the sum of all shares (restricted and unrestricted) and options (exercisable and unexercisable) held by the CEO, divided by the total shares outstanding and multiplied by 100. The other measure is the dollar change in the value of a CEO’s stock and options holdings that would result from a one-percentage point increase in the company stock price. There is a fourth and final measure, which represents the annual change in total compensation. So, there are two long- and two short-term measures. The incentive measures are calculated as follows: 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑒𝑞𝑢𝑖𝑡𝑦 𝑝𝑎𝑦!" = !!"#$ !"#$%& !"#!" !"!#$ !"#$%&'()*"&!"∗ 100 (1) 𝐴𝑛𝑛𝑢𝑎𝑙 𝑐𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 𝑐ℎ𝑎𝑛𝑔𝑒!" = !"!#$ !"#$%&'()*"&!"!!"!#$ !"#$%&'()!%!"!! !"!#$ !"#$%&'()*"&!"!! ∗ 100 (2) Standard
Sample period Minimum Median Maximum Mean Deviation
2000-2007 -30% 3.1% 91% 3.7% 8.1%
13 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝!" = #!!!"#$!"!#!!"#$%&!" #!"##"$ !!!"#$ !"#$#%&'(&)!"∗ 100 (3) 𝐷𝑜𝑙𝑙𝑎𝑟 𝑔𝑎𝑖𝑛 𝑓𝑟𝑜𝑚 + 1%!" = 0.01 ∗ 𝑃𝑟𝑖𝑐𝑒!"∗ 𝑆ℎ𝑎𝑟𝑒𝑠!" + 𝑂𝑝𝑡𝑖𝑜𝑛𝑠!" (4) A log transformation is applied to both the “percentage ownership” and “dollar gain from +1%” measures. This transformation reduces the influence of extreme values and makes the distribution more normal. Following Fahlenbrach and Stulz (2011), this paper uses BAHR as long-term performance. Another approach widely used in academic papers is abnormal performance. However, Fahlenbrach and Stulz (2011) argue that abnormal performance is a better approach when evaluating performance in a portfolio context, and BAHR are better when explaining the cross-sectional variation in performance. To control for determinants of stock performance other than executive compensation, this study controls for previous stock return, book-to-market ratio and the log of the firm’s market value. These variables were found in previous studies and are known to be related to stock returns. Besides controlling for these three variables, I also include the ratio of cash bonus to salary in regressions with long-term incentive measures, following Fahlenbrach and Stulz (2011). They argue that, “When executives receive high cash bonuses for success but when bonuses cannot go below zero for failure, executives potentially have incentives to take risks that are not in the interests of the shareholders or of the safety and soundness of their institutions because that part of their compensation is not affected by the size of the loss that results from their actions.” After all the data are prepared, four regressions, one for each incentive measure, were run: 𝐵𝐴𝐻𝑅!" = 𝛼 + 𝛾 ∗ 𝐷!+ 𝛽!∗ 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑒𝑞𝑢𝑖𝑡𝑦 𝑝𝑎𝑦!"+ 𝛽!∗ 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑒𝑞𝑢𝑖𝑡𝑦 𝑝𝑎𝑦!" ∗ 𝐷! + 𝛽!∗ 𝐹𝑖𝑟𝑚𝑠𝑖𝑧𝑒!"+ 𝛽!∗ 𝑅𝑒𝑡𝑢𝑟𝑛!"!!+ 𝛽! ∗ 𝐵𝑜𝑜𝑘 − 𝑡𝑜 − 𝑚𝑎𝑟𝑘𝑒𝑡 − 𝑟𝑎𝑡𝑖𝑜!" + 𝜀!" 𝐵𝐴𝐻𝑅!" = 𝛼 + 𝛾 ∗ 𝐷!+ 𝛽!∗ 𝐴𝑛𝑛𝑢𝑎𝑙 𝑐𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 𝑐ℎ𝑎𝑛𝑔𝑒!" + 𝛽! ∗ 𝐴𝑛𝑛𝑢𝑎𝑙 𝑐𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 𝑐ℎ𝑎𝑛𝑔𝑒!"∗ 𝐷! + 𝛽!∗ 𝐹𝑖𝑟𝑚𝑠𝑖𝑧𝑒!"+ 𝛽! ∗ 𝑅𝑒𝑡𝑢𝑟𝑛!"!!+ 𝛽!∗ 𝐵𝑜𝑜𝑘 − 𝑡𝑜 − 𝑚𝑎𝑟𝑘𝑒𝑡 − 𝑟𝑎𝑡𝑖𝑜!" + 𝜀!"
14 𝐵𝐴𝐻𝑅!" = 𝛼 + 𝛾 ∗ 𝐷! + 𝛽!∗ 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝!"+ 𝛽!∗ 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝!" ∗ 𝐷! + 𝛽!∗ 𝐹𝑖𝑟𝑚𝑠𝑖𝑧𝑒!"+ 𝛽!∗ 𝑅𝑒𝑡𝑢𝑟𝑛!"!!+ 𝛽! ∗ 𝐵𝑜𝑜𝑘 − 𝑡𝑜 − 𝑚𝑎𝑟𝑘𝑒𝑡 − 𝑟𝑎𝑡𝑖𝑜!"+ 𝛽!∗ 𝑐𝑎𝑠ℎ 𝑏𝑜𝑛𝑢𝑠/𝑠𝑎𝑙𝑎𝑟𝑦!"+ 𝜀!" 𝐵𝐴𝐻𝑅!! = 𝛼 + 𝛾 ∗ 𝐷! + 𝛽!∗ 𝐷𝑜𝑙𝑙𝑎𝑟 𝑔𝑎𝑖𝑛 𝑓𝑟𝑜𝑚 + 1%!"+ 𝛽!∗ 𝐷𝑜𝑙𝑙𝑎𝑟 𝑔𝑎𝑖𝑛 𝑓𝑟𝑜𝑚 + 1%!" ∗ 𝐷! + 𝛽!∗ 𝐹𝑖𝑟𝑚𝑠𝑖𝑧𝑒!"+ 𝛽!∗ 𝑅𝑒𝑡𝑢𝑟𝑛!"!!+ 𝛽! ∗ 𝐵𝑜𝑜𝑘 − 𝑡𝑜 − 𝑚𝑎𝑟𝑘𝑒𝑡 − 𝑟𝑎𝑡𝑖𝑜!"+ 𝛽!∗ 𝑐𝑎𝑠ℎ 𝑏𝑜𝑛𝑢𝑠/𝑠𝑎𝑙𝑎𝑟𝑦!"+ 𝜀!" After these regressions are run, two hypotheses were tested. The first hypothesis tests whether pre-crisis compensation schemes for REIT executives are better aligned with shareholder interest than executive compensation for other financial companies. The second hypothesis investigates if in the post-crisis period, compensation schemes for other financial companies are just as well aligned with shareholder interest as executive compensation for REITs.
5. Results
In this section, the main hypotheses are tested. First, the main results for the years 2000-2007 are discussed; next, the results for the years 2009-2016 are discussed. Finally, I compare the results from 2000-2007 with those of 2009-2016. Table 4a presents the main regression results for the first sample period. The first two columns display the results for the short-term measures, and the last two columns display the results for the long- term measures. In the first regression, equity pay has a significantly negative coefficient and a positive coefficient for REITs, although not significant. Regression 2 shows the results for the second short-term measure. Again, the coefficient for REITs is positive, although now statistically significant. Both coefficients are also economically significant, because the standard deviations are 458.8 for non-REITs and 51.1 for REITs. So, an increase of one standard deviation in the change of annual compensation results in higher returns of 1.56% for non-REITs and 3.34% for REITs. Although both REIT coefficients for the short-term are positive and economically significant, the results are ambiguous because only one of the two measures proved statistically significant. Percentage ownership for REITs is significantly positively related to BAHR. This effect is economically significant. The standard deviation of the logarithm of percentage15 ownership is 0.533. One standard deviation increase in percentage ownership of REIT executives leads to higher returns of 3.72%. In the final regression, dollar gain from +1% has a positive insignificant coefficient. The control variables are similar between the short-term measures and between the long-term measures, but also between the short-term and long-term measures. In all the regressions, the previous returns are negatively related to the following year’s BAHR, that is, companies with high returns performed worse the following year. The book-to-market ratio is also negatively related to BAHR. Growth firms outperformed value firms in this sample, which contrasts with the findings of previous researchers. Table 4a Buy-and-hold returns and chief executive officer incentive measures. The table displays the results from panel regressions of BAHR from a sample of financial companies and REITs from 2000-2007 regarding incentive measures and firm characteristics. “Equity pay” is the percentage equity paid relative to total compensation. “Change in total compensation” is the annual change in total compensation. “Ownership” is the percentage of shares held by the CEO relative to the total shares outstanding. “Dollar gain from +1%” is the logarithm of the dollar change in wealth of the CEO for a 1% change in share price. P-values are reported in parentheses. Statistical significance at the 1%, 5% and 10% level is indicated by ***, **, and *, respectively.
(1) (2) (3) (4)
VARIABLES BAHR BAHR BAHR BAHR
Equity pay (%) -0.00123*** (0.001) Equity pay REIT (%) 0.00132 (0.340) Change in total compensation 0.000034*** (0.007) Change in total compensation REIT 0.000315*** (0.000) Ownership (%) -0.0187 (0.190) Ownership REIT (%) 0.0697** (0.031) Dollar gain from +1% -0.0102 (0.484) Dollar gain from +1% REIT 0.0258 (0.407) Cash bonus/salary 0.0103*** 0.0103*** (0.000) (0.000) Lag return -0.747*** -0.744*** -0.771*** -0.772*** (0.000) (0.000) (0.000) (0.000) Market value -0.00225 0.000479 -0.0694** -0.0622* (0.939) (0.987) (0.028) (0.0509) Book-to-market -0.539*** -0.531*** -0.588*** -0.588*** (0.000) (0.000) (0.000) (0.000) Constant 0.483* 0.410 1.004*** 0.984*** (0.052) (0.105) (0.000) (0.000) Observations 1,572 1,572 1,730 1,730 R-squared 0.250 0.250 0.245 0.245 Number of firms 346 346 401 401
16 Table 4b presents the regression results of the second sample period. They appear similar to Table 4a. Where, in Table 4a, the change in total compensation and percentage ownership for REITs were significant, and percentage equity pay and dollar gain from +1% insignificant, these are now significant, although only at the 10% level, and the change in total compensation and percentage ownership is now insignificant. Both statistically significant measures are also economically significant. For every standard deviation increase in percentage equity pay and dollar gain from+1%, BAHR are respectively 5.52% and 6.53% lower for REITs than for non-REITs. The significant incentive measures for non-REITs are change in compensation, percentage ownership and dollar gain from+1%. These measures are significant in a statistical and economical manner. The control variables did not change much. Market value is now positively related to BAHR. One possible explanation for this is that, following the crisis, investors were more confident about larger companies than smaller companies. The first thing I notice when comparing the results from 2000-2007 with the results from 2009-2016 is that the significant incentive measures changed from the change in annual compensation and percentage ownership to percentage equity pay and dollar gain from +1%. From 2000-2007, REIT executives cared more about how much their total compensation changed and how much ownership they had; but from 2009-2016, REIT executives cared more about how much equity they received and how much they earned for a percentage increase in shareholder value. Besides the change in the significance of the incentive measures, the change in sign is also noticeable. The significant incentive measures for REITs went from a positive sign to a negative sign. Where higher compensation incentives for REIT executives initially led to higher BAHR, in the later sample it led to lower BAHR. One possible explanation for this change is that REIT executives who earn higher compensation toke more risk, which the market rewarded pre-crisis but punished post-crisis. From 2000 to 2007, non-REIT executives attached value to short-term incentives only; but from 2009 to 2016, these executives attached more value to long-term incentives, whereas REIT executives attached value to both short- and long-term incentives in both periods. In the first sample period, two of the four incentive measures for REIT executives are statistically and economically significantly positively related to BAHR. This supports the first hypothesis, which states that REITs whose managers received higher compensation also earned higher stock returns relative to non-REIT managers. Although the two incentive measures that were
17 significant for REITs in the first sample period are now insignificant, the other two incentive measures became significantly negatively related to BAHR. To answer the second hypothesis, about whether this return premium, which REIT managers received relative to non-REIT managers, disappeared after the financial crisis, I need to determine if the REIT coefficients in the second sample period became zero. The incentive measures that were initally significantly positive became insignificant, and the two measures that were insignificant became significantly negative. To answer the second hypothesis, I need to compare the magnitude of the two significant positive measures in the first sample period to the magnitude of the two significant negative measures in the second period. Although the sign changed from positive to negative, the impact this has on the BAHR can be smaller or larger. As in the first sample period, the magnitude of one standard deviation increase was 3.34% and 3.72%; and in the second sample period, this magnitude of one standard deviation was 5.52% and 6.53%. The magnitude of standard deviation increase is, in the second sample period, clearly larger relative to the first sample period. So, the second hypothesis needs to be rejected at the 10% significance level; however, at the 5% significance level, none of the incentive measures are significant and the second hypothesis cannot be rejected.
18 Table 4b Buy-and-hold returns and chief executive officer incentives measures. The table displays the results from panel regressions of BAHR from a sample of financial companies and REITs for 2009-2016 regarding incentive measures and firm characteristics. “Equity pay” is the percentage equity paid relative to the total compensation. “Change in total compensation” is the annual change in total compensation. “Ownership” is the percentage shares held by the CEO relative to the total shares outstanding. “Dollar gain from +1%” is the logarithm of the dollar change in wealth of the CEO for a 1% change in share price. P-values are reported in parentheses. Statistical significance at the 1%, 5%, and 10% level is indicated by *** , ** , and * , respectively. (1) (2) (3) (4)
VARIABLES BAHR BAHR BAHR BAHR
Equity pay (%) -0.00028 (0.589) Equity pay REIT (%) -0.0022* (0.070) Change in total compensation 0.0000988* (0.063) Change in total compensation REIT 0.000253 (0.112) Ownership (%) -0.0318* (0.067) Ownership REIT (%) 0.0548 (0.134) Dollar gain from +1% 0.0578*** (0.002) Dollar gain from +1% REIT -0.0608* (0.060) Cash bonus/salary 0.0183*** 0.0171*** (0.001) (0.002) Lag return -0.677*** -0.679*** -0.698*** -0.689*** (0.000) (0.000) (0.000) (0.000) Market value 0.123*** 0.118*** 0.119*** 0.0710** (0.000) (0.000) (0.000) (0.012) Book-to-market -0.314*** -0.313*** -0.314*** -0.308*** (0.000) (0.000) (0.000) (0.000) Constant -0.519* -0.519* -0.549 -0.413 (0.065) (0.059) (0.062) (0.136) Observations 1,910 1,910 1,887 1,887 R-squared 0.276 0.276 0.288 0.288 Number of firms 308 308 304 304
19
6. Robustness check
Table 5a Tobin’s Q and chief executive officer incentive measures. The table displays the results from panel regressions of Tobin’s Q from a sample of financial companies and REITs from 2000-2007 regarding incentive measures and firm characteristics. “Equity pay” is the percentage of equity paid relative to the total compensation. “Change in total compensation” is the annual change in total compensation. “Ownership” is the percentage of shares held by the CEO relative to the total shares outstanding. “Dollar gain from +1%” is the logarithm of the dollar change in wealth of the CEO for a 1% change in share price. P-values are reported in parentheses. Statistical significance at the 1%, 5% and 10% level is indicated by *** , ** , and * , respectively. (1) (2) (3) (4)VARIABLES Tobin’s Q Tobin’s Q Tobin’s Q Tobin’s Q Equity pay (%) 0.00068 (0.229) Equity pay REIT (%) 0.000112 (0.901) Change in total compensation 8.87e-06 (0.051) Change in total compensation REIT 0.0001*** (0.0308) Ownership (%) 0.0383* (0.0875) Ownership REIT (%) -0.0677** (0.0458) Dollar gain from +1% 0.0288 (0.191) Dollar gain from +1% REIT -0.0047 (0.906) Cash bonus/salary 0.00176 0.00178 (0.492) (0.487) Lag return -0.110 0.113 -0.0384 -0.0386 (0.570) (0.562) (0.553) (0.551) Market value 0.192*** 0.193*** 0.166*** 0.193** (0.000) (0.000) (0.000) (0.012) Book-to-market -0.0988** -0.0981** -0.156*** -0.157*** (0.0296) (0.0310) (0.000) (0.000) Constant -0.414 -0.140 -0.0799 -0.399 (0.365) (0.736) (0.842) (0.383) Observations 1,572 1,572 1,950 1,950 R-squared 0.117 0.115 0.108 0.106 Number of firms 346 346 416 416 According to Lu and White (2014), a robustness check is a common part of empirical studies. A robustness check examines how core regression coefficients behave when the regression specification is modified by adding, removing or replacing the dependent- or independent variable. When the coefficients are robust, this is commonly interpreted as evidence for structural validity. In this paper, a robustness check was performed by replacing the dependent variable with Tobin’s Q, winsorizing the original regressions, leaving out the cash bonus to salary variable, and adding a long-term control variable in the short-term compensation regressions. It is evident, if you compare Table 4a (b) with Table 5a (b), that the coefficient of
20 interest, the REIT interaction term, does not change much overall. In the first period, the only significant difference is that the REIT interaction term with percentage ownership changes from a significant positive coefficient to a significant negative coefficient. The control variables behave in a similar way to the regressions with BAHR. As for the second period, the only significant difference is that the REIT interaction term with percentage equity pay changes from a significant negative coefficient to an insignificant positive coefficient. Again, the control variables behave in a similar manner to the regressions with BAHR. The same results hold when adding and leaving control variables and if the data is being winsorized. It is safe to conclude that the results of tables 4a and 4b are robust to changes in the dependent variable, inclusion and exclusion of control variables, and different treatment of outliers. Table 5b Tobin’s Q and chief executive officer incentive measures. The table displays the results from panel regressions of Tobin’s Q from a sample of financial companies and REITs from 2009-2016 regarding incentive measures and firm characteristics. “Equity pay” is the percentage of equity paid relative to the total compensation. “Change in total compensation” is the annual change in total compensation. “Ownership” is the percentage of shares held by the CEO relative to the total shares outstanding. “Dollar gain from +1%” is the logarithm of the dollar change in wealth of the CEO for a 1% change in share price. P-values are reported in parentheses. Statistical significance at the 1%, 5% and 10% level is indicated by *** , ** , and * , respectively. (1) (2) (3) (4)
VARIABLES Tobin’s Q Tobin’s Q Tobin’s Q Tobin’s Q Equity pay (%) -0.000316 (0.171) Equity pay REIT (%) 0.000293 (0.545) Change in total compensation 0*** (0.000) Change in total compensation REIT 8.65e-5 (0.166) Ownership (%) 0.0125 (0.222) Ownership REIT (%) 0.000635 (0.981) Dollar gain from +1% -0.00326 (0.781) Dollar gain from +1% REIT -0.0681** (0.0433) Cash bonus/salary -0.00291 -0.00218 (0.171) (0.299) Lag return 0.617*** 0.619*** 0.262* 0.265* (0.000) (0.0648) (0.0684) (0.000) Market value 0.102*** 0.100*** 0.158*** 0.169*** (0.000) (0.000) (0.000) (0.000) Book-to-market -0.0444** -0.0447** -0.0205 -0.261 (0.0353) (0.297) (0.194) (0.000) Constant -0.290 -0.430 -0.413 -0.288 (0.186) (0.223) (0.226) (0.182) Observations 1,910 1,910 2,187 2,187 R-squared 0.454 0.454 0.283 0.287 Number of firms 308 308 311 311
21
7. Conclusion
In this study, I investigated the pay-for-performance relationship in the REIT and non-REIT financial service industry, pre- and post-crisis. Real estate investment trusts receive a tax exemption when they distribute 90% of the corporate earnings as dividend. This reduces the principal-agent problem, and the CEOs of these companies are more focused on increasing the profit of the company. It was expected that REITs who awarded their CEOs more compensation received a stock return premium in return relative to non-REIT financial companies before the 2008 financial crisis. Also, it was expected that this premium disappeared following the crisis. In Chapter 5, the results of several panel regressions of incentive measures and firm characteristics on BAHR were discussed. The results indicate that two of the four incentive measures used in this paper had a positive coefficient pre-crisis. This supports the first hypothesis. The results of the regressions in the period following the financial crisis indicate that REITs with a 5% significance level had the same coefficients as non-REITs; and with a 10% significance level had negative coefficients (i.e. with a 5% significance level it supports the second hypothesis, but with a 10% significance level it does not). In Chapter 6, several robustness checks were performed to check the structural validity of the regressions. It was concluded that the results are robust to changes in the dependent variable, the inclusion and exclusion of control variables, and the different treatment of outliers. The conclusion can be made that there exists a relationship between executive compensation and shareholder interest; however, this is not always positive. Bearing previous research in mind, it seems that executive compensation is not the best way to solve the principal-agent problem that occurs when ownership and control are separated.22
8. Bibliography
Abowd, J. M. (1990). Does performance-based managerial compensation affect corporate performance?. ILR Review, 43(3), 52-S.Bauer, R., Eichholtz, P., & Kok, N. (2010). Corporate governance and performance: the REIT effect. Real Estate Economics, 38(1), 1-29.
Boschen, J. F., & Smith, K. J. (1995). You can pay me now and you can pay me later: The dynamic response of executive compensation to firm performance. Journal of Business, 577-608.
Chopin, M., Dickens, R., & Shelor, R. (1995). An empirical examination of compensation of REIT managers. Journal of Real Estate Research, 10(3), 263-277.
Cooper, M. J., Gulen, H., & Rau, P. R. (2014). Performance for pay? The relation between CEO incentive compensation and future stock price performance.
Davis, B., & Shelor, R. (1995). Executive compensation and financial performance in the real estate industry. Journal of Real Estate Research, 10(2), 141-151.
Ely, K. M. (1991). Interindustry differences in the relation between compensation and firm performance variables. Journal of Accounting Research, 37-58.
Fahlenbrach, R., & Stulz, R. M. (2011). Bank CEO incentives and the credit crisis. Journal of Financial Economics, 99(1), 11-26.
Ghosh, C., & Sirmans, C. F. (2003). Board independence, ownership structure and performance: evidence from real estate investment trusts. The Journal of Real Estate Finance and Economics, 26(2), 287-318.
Ghosh, C., & Sirmans, C. F. (2005). On REIT CEO compensation: does board structure matter?. The Journal of Real Estate Finance and Economics, 30(4), 397-428.
Hallman, G., Hartzell, J. C., & Parsons, C. A. (2011). Incentive compensation and the likelihood of termination: theory and evidence from real estate organizations. Real Estate Economics, 39(3), 507-546.
Hartzell, J. C., & Starks, L. T. (2003). Institutional investors and executive compensation. The journal of finance, 58(6), 2351-2374.
Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance, and takeovers. The American economic review, 76(2), 323-329.
23
Lewellen, W., Loderer, C., Martin, K., & Blum, G. (1992). Executive Compensation and the Performance of the Firm. Managerial and Decision Economics, 65-74.
Lewis, D., & Terrell, D. Do Incentives for REIT Executives Improve Efficiency?.
Malmendier, U., & Tate, G. (2009). Superstar ceos. The Quarterly Journal of Economics, 124(4), 1593-1638.
Masson, R. T. (1971). Executive motivations, earnings, and consequent equity performance. Journal of Political Economy, 79(6), 1278-1292.
Mehran, H. (1995). Executive compensation structure, ownership, and firm performance. Journal of financial economics, 38(2), 163-184.
Mishra, C. S., McConaughy, D. L., & Gobeli, D. H. (2000). Effectiveness of CEO pay-for-performance. Review of Financial Economics, 9(1), 1-13.
Noguera, M. C., Highfield, M., & Nagel, G. L. (2007). CEO incentive-based compensation and REIT performance. Mississippi State University. Nyberg, A. J., Fulmer, I. S., Gerhart, B., & Carpenter, M. A. (2010). Agency theory revisited: CEO return and shareholder interest alignment. Academy of Management Journal, 53(5), 1029-1049. Pennathur, A. K., & Shelor, R. M. (2002). The determinants of REIT CEO compensation. The Journal of Real Estate Finance and Economics, 25(1), 99-113.
Sonenshine, R., Larson, N., & Cauvel, M. (2016). Determinants of CEO Compensation before and after the Financial Crisis. Modern Economy, 7(12), 1455.
Vemala, P., Nguyen, L., Nguyen, D., & Kommasani, A. (2014). CEO Compensation: Does Financial Crisis Matter?. International Business Research, 7(4), 125.
Lu, X., & White, H. (2014). Robustness checks and robustness tests in applied economics. Journal of econometrics, 178, 194-206.
