Executive compensation in listed companies has been a topic of research for many years because of the drastic increase in executive pay (Bebchuk and Grinstein 2005; Murphy 2013). Principal-agency theory suggests that efficient executive compensation plays an important role in aligning the incentives of executives with those of shareholders. Specifically, executive pay incentivizes executives to perform in the best interest of shareholders by relating their compensation to the firm’s achieved performance (Jensen and Meckling 1976; Olaniyi 2019). In particular, the use of deferred compensation packages is critically discussed in the literature (Bebchuk et al. 2002; Coles et al. 2006). On the one hand, such compensation packages are intended to reduce managerial myopia and align CEOs’ incentives with those of their firms’ shareholders. On the other hand, however, prior literature finds the use of large deferred compensation packages to be associated with agency problems, such as increased managerial risk-taking and rent extraction (Bebchuk et al. 2002; Coles et al. 2006). Because executive compensation has increased in many countries over the last decades, without a comparable increase in firm performance, dissatisfaction among shareholders and social concerns regarding income disparity have been rising (Wang and Wu 2021).

In an attempt to reduce excessive executive compensation, the European Parliament and the Council of the European Union passed the Shareholder Rights Directive 2017/828, which EU member countries had to incorporate into federal law. Article 9a and 9b of this directive provide shareholders with the rights to vote annually on the remuneration report, and at least every 4 years on the remuneration policy (“Say on Pay”) (Directive (EU) 2017/828, 2017). Thereby, SRD II aims to improve shareholders’ ability to voice concerns about the compensation policy and report (Directive (EU) 2017/828, 2017) and address potential agency problems (Bebchuk and Grinstein 2005).

Prior literature studying similar compensation-related legislation in the world, such as the Dodd-Frank in the US and the DRR in the UK finds mixed evidence on the effectiveness of Say on Pay on the level and the structure of executive compensation. In particular, findings regarding compensation levels are unclear. Whereas some studies find that executive compensation does not decrease following Dodd-Frank in the US (Guadalupe et al. 2013; Burns and Minnick 2013) or DRR in the UK (Conyon and Sadler 2010), other studies find that executive compensation levels decrease following the implementation of Say on Pay laws across countries (Correa and Lel 2016; Balsam et al. 2016). These studies are similarly opposite regarding the effect of such legislation on executive compensation structure, with some finding no significant change in incentive pay for CEOs (Guadalupe et al. 2013; Burns and Minnick 2013; Conyon and Sadler 2010) and other documenting increases in performance-based pay (Correa and Lel 2016; Balsam et al. 2016).

However, there is also literature suggesting that executive compensation levels increase following the introduction of Say on Pay laws (Kronlund et al. 2018; Iliev and Vitanovaa 2019). These studies suggest that Say on Pay laws do reduce apparently excessive fixed salaries of CEOs and at the same time increase CEO’s incentive compensation. Although this is consistent with the intended goals of such regulation, the increase in incentive pay often seems to outweigh the decrease in the reduction of fixed salaries, leading to increases in total compensation.

Article 9b of SRD II specifically mentions increased transparency of the remuneration report. This Article requires EU members to publicly disclose both the fixed and variable components of total compensation on an individual basis in the remuneration report, which could alleviate information asymmetries between shareholders and executives (Jensen and Meckling 1976; Olaniyi 2019). Consistent with this, prior literature finds that the implementation of compensation disclosure regulation is associated with decreases in compensation levels and increases in the pay-for-performance sensitivity (IY Wang et al. 2020; Perry et al. 2001; Vafeas and Afxentiou 1998; Craighead et al. 2004). However, more transparent disclosure is also associated with the phenomenon of ‘leapfrogging’, i.e., with compensation benchmarking leading to CEOs demanding higher compensation as their peers (DiPrete et al. 2010). In particular, the selection of (high compensation) peers to benchmark compensation against is associated with compensation levels ratcheting upwards (De Vaan et al. 2019; Faulkender and Yang 2013; Laschever 2013). Therefore, it is the transparency in the remuneration disclosure specifically that could incentivize CEOs to compare compensation with peers and attempt to beat others’ compensation leading to an increase in executive compensation levels that is not based on firm performance.

Because of the mixed evidence of the effectiveness of compensation (disclosure) regulation and Say on Pay laws, the hypotheses regarding the effect of SRD II on executive compensation levels and structure are presented in the null-form:

H1: The level of CEO compensation in German and Austrian firms does not change after the implementation of SRD II.

H2: The ratio of CEO deferred pay to total CEO compensation in German and Austrian firms does not change after the implementation of SRD II.

To examine the aforementioned hypotheses, this study employs a difference-in-differences (DiD) research design. The pre-period of the DiD relates to the years before the implementation of SRD II from 2016 to 2018, and the post-period related to the years after the SRD II from 2020 to 2022. The transition year 2019 is excluded. The treatment group of the DiD regression consists of listed companies in Germany and Austria that experience the implementation of SRD II; the control group consists of listed companies in Switzerland, which do not experience the implementation of SRD II. Switzerland constitutes a suitable control group country because it is comparable to Germany and Austria, but it is not a member of EU and EEA and, thus, is not under effect of SRD II (Fredriksson and Oliveira 2019). In addition, Switzerland does not experience changes in compensation-related regulation in 2019.

This study employs two DiD regression models corresponding to the two aforementioned hypotheses to investigate the impact of the implementation of SRD II on the level and structure of CEO compensation:

Total pay = β₀+ β₁ * Post_t+ β₂* Treatment_i + β₃ * Post_t * Treatment_i + β₄ * Firm_performance_i,t + β₅ * Firm_performance_i,t−1 + β₆ * Board_size_i,t + β₇ * Board_independence_i,t + β₈ * Institutional_ownership_i,t + β₉ * Firm_size_i,t + β₁₀ * Leverage_i,t + β₁₁ * Industry Fixed Effects_i * Year Fixed Effects_t + β₁₂ * Firms Fixed Effects_i + ε_i,t (1)

Deferred pay = β₀ + β₁ * Post_t + β₂ * Treatment_i + β₃ * Post_t * Treatment_i + β₄ * Firm_performance_i,t + β₅ * Firm_performance_i,t−1 +β₆ * Board_size_i,t + β₇ * Board_independence_i,t +β₈ * Institutional_ownership_i,t + β₉ * Firm_size_i,t + β₁₀ * Leverage_i,t + β₁₁ * Industry Fixed Effects_i * Year Fixed Effects_i + β₁₂ * Firms Fixed Effects_i + ε_i,t (2)

The dependent variable Total pay in equation (1) is the natural logarithm of the CEO’s total compensation in year t. The use of natural logarithms is consistent with prior literature (e.g., Correa and Lel 2016) and accounts for the skewness of the pay distribution. The dependent variable Deferred pay in equation (2) is the ratio of the sum of long-term incentive plans, restricted stock awards, and all other compensation to the CEO’s total compensation.

The indicator variable Post relates to whether a firm-year relates to the post-period from 2020 to 2022 or to the pre-period from 2016 to 2018. The indicator variable Treatment relates to whether company i is located in a country that adopts SRD II, i.e., Germany or Austria, or not, i.e., Switzerland. The interaction of these two indicator variables is the DiD estimator and the main variable of interest.

Several studies show that the level and structure of CEO compensation is associated with firm-specific and governance control variables such as firm performance (Bouteska and Mefteh-Wali 2021; Burns and Minnick 2013), firm size (Bryson et al. 2014; MJ Conyon 2014), firm leverage (Alves et al. 2016; Coles et al. 2006), board size, board independence, and institutional ownership (MJ Conyon and He 2012; MJ Conyon and Peck 1998; Core et al. 1999; Firth et al. 2007; Yermack 1996). Therefore, control variables relating to these characteristics are included in equation (1) and (2), as well as fixed effects relating to the firm and industry-year. Standard errors are clustered at the country-level and all variables are winsorized at the first and ninety-ninth percentile. Table 1 presents variable definitions for all variables.

CEO compensation data and firm financial information is obtained from LSEG Workspace. Because SRD II does not apply to listed companies outside of the EU, data for all other listed companies is obtained from LSEG Workspace, yielding 5,684 firm-year observations relating to 1,031 firms. Further sampling criteria require existing data for all variables and exclude financial services firms as well as small firms with less than USD 20 million in total assets. Because data availability differs regarding compensation structure, the sample sizes differ across estimations of equations (1) and (2). Table 2 presents the sample selection process. The sample for estimation of equation (1) comprises 1,822 firm-year observations relating to 477 firms; the sample for estimation of equation (2) comprises 1,524 firm-year observations relating to 396 firms.

Table 3 presents descriptive statistics for both samples. Panel A does not reveal noticeable differences in average compensation levels before and after the implementation of SRD II in the treatment or control group. In contrast, Panel B reveals an increase in the unconditional mean of the deferred compensation ratio of 8% for treatment firms after SRD II, whereas the control group did not change noticeably.