Three essays in empirical corporate finance

Hong, Songzhi (2024). Three essays in empirical corporate finance. University of Birmingham. Ph.D.

Preview

Hong2024PhD.pdf
Text - Accepted Version
Available under License All rights reserved.
Download (3MB) | Preview

Abstract

This thesis comprises three empirical essays in the field of corporate finance, all of which utilize the difference-in-differences methodology to mitigate potential endogeneity issues. The first essay within this thesis focuses on examining the influence of pay transparency on employee productivity. I introduce the staggered adoption of pay secrecy laws at the state level to measure the state pay transparency level. Pay secrecy laws were initially introduced to address gender and race pay gaps by promoting transparency and prohibiting pay secrecy practices (Kim, 2013, 2015). These laws aim to empower employees by providing information on salary and enabling them to address and potentially reduce pay disparities (Cullen and Perez-Truglia, 2018b; Cullen and Pakzad-Hurson, 2019).

Employers implement pay secrecy policies to prevent wage comparisons and employee dissatisfaction (Colella et al., 2007; Kim, 2015). However, some argue that pay secrecy contributes to pay discrimination (Kim, 2013, 2015; Baker et al., 2019). Therefore, in the United States, there is an increasing emphasis on pay transparency to narrow gender and ethnicity pay gaps (Trotter et al., 2017; Heisler, 2021). The impact of increased pay transparency on employee productivity is ambiguous. On one hand, it can improve morale by addressing issues of prejudice and favoritism (Cullen and Perez-Truglia, 2018), and reduce uncertainty regarding compensation employees expect for their efforts (Hsieh et al., 2019). It can also promote gender equality and lead to increased productivity and retention (Bennedsen et al., 2019). In summary, increased pay transparency can have a positive effect on employee productivity.

On the other hand, increased pay transparency may reduce job satisfaction and result in complaints or resignations due to wage comparisons. This is because individuals not only care about absolute income but also their relative income (Colella et al., 2007). Lower-paid employees may be dissatisfied, while higher-paid employees may perceive efforts to mitigate inequality as threats (Card et al., 2012; Breza et al., 2018). Dissatisfied employees may engage in destructive behaviors, and highly valued employees may leave for better opportunities elsewhere (Hitz and Werner, 2012; Kim and Marschke, 2005), which aligns with "Inequity Aversion Theory" (Adams, 1965; Cowherd and Levine, 1992) and "Relative Deprivation Theory" (Martin, 1981). Therefore, increased pay transparency could also have a negative effect on employee productivity.

To enhance precision and minimize confounding factors, this essay introduces a novel measure to proxy employee productivity. It divides EBITDA, excluding incomes and incorporating expenses unrelated to employees or dominated by managers, by the number of employees. This essay stands as the first to uncover a significant correlation between heightened pay transparency and diminished employee productivity across a substantial and diverse sample of U.S. firms. This finding lends support to the notion that wage comparisons can erode job satisfaction. On average, firms headquartered in states where pay secrecy laws have been implemented experience a 1.58% decline in employee productivity compared to the overall sample mean.

Furthermore, this analysis reveals that the impact of pay secrecy laws is more significant in companies based in states characterized by lower levels of social capital. This finding underscores the heterogeneous nature of the effects and provides valuable insights into the mechanisms through which these laws operate. Moreover, this essay uncovers a compelling relationship between increased pay transparency resulting from the implementation of pay secrecy laws and a decrease in employee salaries. This discovery offers a potential explanation for the observed decline in productivity.

This essay makes the following contributions. First, in response to concerns regarding pay secrecy practices and their potential for pay discrimination (Kim, 2013, 2015; Cullen and Perez-Truglia, 2018; Baker et al., 2019), my empirical research aims to investigate the causal effects of pay transparency on employee productivity. To address endogeneity concerns and isolate exogenous variation in peer group pay (Gao et al., 2021), I employ a difference-in-differences framework using the staggered adoption of pay secrecy laws and incorporate high-dimensional fixed effects. This study extends the existing literature on pay secrecy law adoption and salary transparency, contributing to a broader understanding of the topic. Previous research by Kim (2013, 2015), Baker et al., (2019), Bennedsen et al., (2019), Mas (2017), Duchini et al., (2020), and Gao et al. (2021) sheds light on pay inequality and gender pay gaps. However, the effects of pay secrecy laws on firm-level outcomes and generalized employee productivity have not been thoroughly explored.

Second, to accurately capture employee productivity, I draw inspiration from previous studies and utilize individual profits as a proxy, constructing a less noisy and more accurate measure of broader employee output. My findings are consistent with alternative measures used in previous studies (Kale et al., 2016; Lins et al., 2017; Gao et al., 2018), enhancing my confidence in this novel proxy.

Third, my analysis reveals that increased pay transparency is associated with lower employee productivity, providing empirical evidence supporting the notion that wage comparisons can reduce job satisfaction (Card et al., 2012). This finding contrasts with the positive influence of pay transparency on innovation found by Gao et al. (2021), which focused on patents and citations for minority inventors. This is because my study focuses on a broader measure encompassing all employees instead of only innovators, which enables a comprehensive evaluation of employee incentives and behaviors that plausibly have direct effects on productivity (Faleye et al., 2013).

The second essay in the thesis examines the effects of restricted executive mobility on institutional shareholding. It introduces the concept of the Inevitable Disclosure Doctrine (IDD), which imposes stricter restrictions on managerial mobility to protect trade secrets. Restricted executive mobility imposes higher costs on managers whose current positions are at risk. This situation leads to increased career concerns and a stronger inclination to engage in opportunistic behaviors that can improve their current employer's perception of their abilities (Kothari et al., 2009; Gao et al., 2018; Ali and Li, 2019). Additionally, limitations on external employment opportunities for managers reduce the pool of potential replacement CEOs, making it difficult for companies to identify and recruit qualified successors, thus necessitating the retention of the current CEO (Grande-Herrera, 2019). As a result, this disruption in the labor market's disciplinary mechanism enables the occurrence of executive opportunistic behaviors (Li et al., 2017; Kim et al., 2020; Ali and Li, 2019; Li et al., 2018; Gao et al., 2018; Islam et al., 2020; Na, 2020). Consequently, it can be argued that restricted executive mobility undermines the quality of corporate governance.

To address the challenges posed by restricted executive mobility and enhance corporate governance, companies may choose to increase institutional shareholding (Chung and Zhang, 2011). This is based on the recognition of the monitoring role played by institutional investors. Therefore, it can be inferred that following restricted executive mobility, institutional shareholding may increase. However, institutional investors need to carefully evaluate the costs and benefits associated with monitoring, and exercise prudence in their actions. Thus, the quality of corporate governance, which is influenced by opportunistic behaviors, becomes a significant factor in the decision-making process of institutional investors. This suggests that, following restricted executive mobility, institutional shareholding may decrease.

This study provides the first empirical evidence on the impact of recognizing the IDD on institutional investors' equity holdings. Firms in states acknowledging the IDD experience an average 2% decrease in institutional shareholding compared to the sample period mean. This decline primarily stems from reduced corporate governance indicated by agency costs. Notably, activist and long-term institutional investors exhibit sensitivity to executive mobility constraints, reinforcing their monitoring incentives. Additionally, the study demonstrates that IDD recognition's influence on institutions is more pronounced in knowledge-intensive industries, highlighting the impact of executive mobility on institutional shareholding due to trade secret protection. These findings shed light on institutional investors' motivations to target portfolio companies, aiming to mitigate monitoring costs and fulfill fiduciary responsibilities. The study contributes to the related literature by examining the outcomes associated with restricted executive mobility. It also expands understanding of the IDD's economic effects, particularly on ownership structure, and complements existing research on institutional investor engagement in corporate governance strategies.

The third essay of the thesis revisits the causal link between takeover threats and corporate default risk, incorporating a fresh perspective by examining the implementation of Second-Generation State-level Antitakeover Laws alongside traditional proxies for takeover threat. An active takeover market serves as an external disciplinary mechanism for managers, promoting managerial replacements in cases of poor performance (Manne, 1965; Fama and Jensen, 1983; Jensen and Ruback, 1983; Scharfstein, 1988; Lel and Miller, 2015). However, anti-takeover protections weaken this mechanism by shielding managers and increasing their managerial entrenchment, resulting in heightened agency costs of equity. Balachandran et al. (2022) argue that reduced takeover likelihood, caused by weakened disciplining mechanisms, exacerbates agency conflicts, leading to diminished cash flows for debt payments and increased default risk (Driss et al., 2021).

Conversely, takeover threats can have a negative impact on default risk. Garvey and Hanka (1999) find that companies influenced by second-generation state-level anti-takeover laws exhibit a reduction in reliance on debt. The decreased likelihood of hostile takeovers diminishes managerial career concerns and allows managers greater discretion in capital structure decisions, leading to a decrease in debt issuance (Grossman and Hart, 1980; Knoeber, 1986; Scherer, 1988; Stein, 1988; Jung et al., 1996). Furthermore, the financial leverage ratio can serve as an indicator of default risk (Traczynski, 2017; Cathcart et al., 2019), as default occurs when a company's asset value falls below its debt face value (Merton, 1974). Strengthened anti-takeover protection can potentially reduce default risk, benefiting debtholders, in line with the trade-off theory of capital structure (Kraus and Litzenberger, 1973).

This study is the first to show that firms incorporated in states influenced by Control Share Acquisition (CS) laws experience a decrease in default probability. Specifically, I find that these firms reduce their default risk by 18.2% on average compared to the mean default risk during the sample period. Additionally, I observe an increase in agency costs of equity, indicating that weakened external monitoring mechanisms prompt managers to exercise discretion in reducing debt usage. I suggest that agency costs of equity may serve as the possible link between takeover threats and default risk. This effect is particularly prominent in companies with high institutional shareholding, indicating that the decline in default risk for affected firms is a result of deteriorated corporate governance and heightened agency conflicts. Moreover, I report that the reduced default risk following the adoption of CS laws undermines shareholder interests, and managers' tendency to underinvest following the adoption of these laws suggests a preference for a "enjoy a quiet life" among executives.

This study contributes to the existing literature by expanding the research on determinants of corporate default risk. While previous studies have examined factors such as stock liquidity (Brogaard et al., 2017; Nadarajah et al., 2020), innovation performance (Hsu et al., 2015), and incentive structure (Bennett et al., 2015), I focus on the effects of anti-takeover protection on default risk. My findings differ from Balachandran et al. (2022) who argue that anti-takeover protection increases default likelihood derived from managerial opportunistic activities, which harms shareholder. In contrast, I conclude that decreased takeover threats are associated with a lower probability of default. Furthermore, I find that the reduced default risk resulting from the adoption of CS laws negatively affects shareholder benefits as well, contradicting Balachandran et al. (2022). I propose an alternative perspective that increased usage of anti-takeover provisions leads managers to prioritize their own well-being “enjoy a quiet life”, harming shareholders but benefiting debtholders (Bertrand and Mullainathan, 2003; Klock et al., 2005; Chava et al., 2009; Qiu and Yu, 2009; Gormley and Matsa, 2016). To support my hypothesis, I find evidence of managers' underinvestment following the adoption of CS laws. Therefore, my study contributes to understanding how managers' risk exposure influences their decisions and highlights that traditional agency conflicts related to "private benefits" may not be the primary driver of activities deviating from shareholders' interests.

Finally, this study extends the investigation into the effects of CS laws, which have received limited attention thus far. I demonstrate that CS laws, along with business combination laws, contribute to variations in corporate governance and have a negative impact on default risk.

In the three essays presented, I employ the difference-in-differences methodology, utilizing staggered treatments as natural experiments to facilitate causal investigations. By employing a difference-in-differences design, I take advantage of the occurrence of multiple shocks that affect different firms at various time points. Specifically, I compare the effects before and after the implementation of legislation changes in states where such changes were implemented (the treatment group), contrasting them with the effects observed in states where no such changes occurred (the control group) (Gormley and Matsa, 2016; Klasa et al., 2018; Ali et al., 2019). This methodology enables me to address potential biases related to the timing of the laws (Bertrand and Mullainathan, 2003) and overcome alternative explanations that may arise in settings with a single shock, where contemporaneous events could influence my findings (Roberts and Whited, 2013). Additionally, I introduce a one-year lag for the state-level laws and doctrines, allowing sufficient time for affected companies to adjust their firm-level outcomes and further mitigating concerns of reverse causality.

It is noteworthy that these laws were not primarily designed to impact my variable of interest, thus suggesting that this effect is likely an unintended consequence, rendering my utilization of these laws particularly valuable. My analysis incorporates state-by-year and industry-by-year fixed effects, which enhances the robustness of my findings. Specifically, since numerous companies are headquartered in states that differ from the ones in which they are incorporated, I have the opportunity to include the state-by-year fixed effect. In summary, by incorporating these high-dimensional fixed effects, I can alleviate concerns pertaining to unobserved heterogeneity associated with the industry, state, or observation year of the firms (Gormley and Matsa, 2016).

In order to ensure the validity of my analysis, I employ a comprehensive set of diagnostic or robustness tests. The foundational assumption of the difference-in-differences methodology is that, in the absence of the law or doctrine, the treatment group and control group would exhibit similar trends. I demonstrate that the trends in dependent variables prior to the treatment are indeed comparable between the two groups of firms. Moreover, I conduct a placebo test by randomly assigning states to adopt the law or doctrine, ensuring equal probabilities of adoption across all states. This guarantees that any observed differences between and within states are not systematic. My results reveal that the actual coefficient estimate from the baseline regression lies comfortably in the tail of the distribution of coefficient estimates derived from 1,000 simulated placebo tests. This finding eliminates the possibility that my results are merely due to chance.

To address potential self-selection bias arising from firm-specific characteristics that could influence my results, I perform a Propensity Score Matching (PSM) test. For each year, I match treatment firms with control firms based on the firm characteristics used as control variables in my baseline regression. I estimate the probability of being assigned to the treatment or control group using a logit regression that incorporates all control variables, year, state, and industry fixed effects. Utilizing the propensity scores obtained from this logit estimation, I conduct matching within a caliper of 0.01 without replacement. The results, after controlling for sample selection bias through the PSM method, support my baseline findings, emphasizing that my conclusions are not driven by systematic differences.

Cengiz et al. (2019) have highlighted potential econometric concerns related to the aggregation of discrete difference-in-differences (DiD) estimates using ordinary least squares (OLS). These concerns include heterogeneous treatment effects and the possibility of negative weights assigned to specific treatments. To ensure a more precise examination of the relationship, I employ stacked difference-in-differences (DID) estimates as an additional robustness check. The stacked DID method aims to transform the staggered adoption setting into a two-group, two-period design. In this transformed design, the difference-in-differences estimate captures the average effect of the treatment on the treated, while taking into account the relative sizes of the group-specific datasets and the variance of treatment status within those datasets. This approach uses more stringent criteria for admissible, clean control groups. And separate datasets are stacked in event-time, which is equivalent to a setting where the events happen contemporaneously. Finally, I find consistent results, which confirms that my difference-in-differences estimates are not sensitive to heterogeneous treatment effects.

The subsequent sections of this thesis are structured as follows. The first essay examines the impact of state-level pay secrecy laws on employee productivity and is titled "Pay Transparency and Employee Productivity: Evidence from State-level Pay Secrecy Laws." The second essay investigates the relationship between executive mobility and institutional ownership, drawing evidence from the Inevitable Disclosure Doctrine, and is titled "Executive Mobility and Institutional Ownership: Evidence from the Inevitable Disclosure Doctrine." The final essay critically evaluates the causal link between takeover threats and default risk and is titled "Takeover Threat and Default Risk: A Causal Re-evaluation."

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):