Skip to main content
SpringerLink
Log in

The effects of the mandated disclosure of CEO-to-employee pay ratios on CEO pay

  • Original Article
  • Published:
International Journal of Disclosure and Governance Aims and scope Submit manuscript

Abstract

This study examines whether the mandated disclosure of CEO-to-employee pay ratios motivated firms to curb CEO pay prior to their first pay ratio disclosures. In July 2010, the US Congress directed the SEC, via Section 953(b) of the Dodd–Frank Act, to enforce a rule that requires firms to disclose the ratio of the CEO’s pay to the median employee’s pay (the “rule”). The SEC proposed the rule in September 2013 and adopted it in August 2015. Though the SEC contends that the rule is intended to benefit shareholders, opponents claim that the rule is intended to shame firms into reducing CEO pay. The opponents’ claim is consistent with theory that posits, and evidence that suggests, that disclosure mandates can be used to motivate disclosers to adopt certain, desired behaviors. Based on this theory, as well as evidence indicating that firms likely expected to incur reputational losses from disclosing high pay ratios, this study hypothesizes that firms that are required to comply with the rule (relative to firms that are not) curbed CEO pay prior to their first pay ratio disclosures. This study further hypothesizes this relative curb on CEO pay was greater for firms that are more sensitive to the reputational effects of the rule—that is, firms that are more susceptible to public scrutiny of or adverse stakeholder reactions to pay ratios. These hypotheses are tested through a difference-in-differences research design that examines changes in residual total CEO pay (i.e., the portion of total CEO pay that is not predicted by economic determinants) from the periods before to the periods after the SEC’s proposal and adoption of the rule. Although there is no evidence of a curb on residual CEO pay in response to the SEC’s proposal (or adoption) at the average firm, there is evidence of a curb in response to the proposal (but not adoption) at firms that are more susceptible to public scrutiny of or adverse stakeholder reactions to pay ratios. Thus, regardless of the intended consequences of the rule, firms that are more sensitive to the reputational effects of pay ratios behaved as if the rule shamed them into curbing CEO pay. This is important because it contributes to the current understanding of how disclosure mandates can change firm behavior. The results of this study should therefore be of interest to legislators, regulators, special-interest groups, and the public.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Similar content being viewed by others

Notes

  1. Section 953(b) actually calls for an employee-to-CEO pay ratio. However, the rule that was proposed and later adopted by the SEC requires a CEO-to-employee pay ratio.

  2. Affected and unaffected firms are described in the background (second) and research design (fourth) sections.

  3. In this paper, curb and similar words (e.g., restrain) are used in a relative sense. That is, a curb on CEO pay means that one group of firms increases (decreases) CEO pay to a lesser (greater) extent than another group of firms.

  4. Unless explicitly noted, the estimates for the responses of the more sensitive firms to the SEC’s proposal (adoption) are based on the difference between affected and unaffected firms in the change in residual total CEO pay from the period before to the period after the proposal (adoption) for more sensitive relative to less sensitive firms.

  5. Since there are no unaffected firms in the S&P 500 over the period 2011–2016, the analysis for S&P 500 firms is limited to affected firms. Accordingly, the estimates for the responses of S&P 500 firms to the SEC’s proposal (adoption) are based on the change for affected firms in residual total CEO pay from the period before to the period after the proposal (adoption) for S&P 500 relative to non-S&P 500 firms.

  6. There are many other studies on the effects of mandated disclosure on discloser behavior. For a review of such studies, as well as other disclosure-related studies, see Dranove and Jin (2010) and Leuz and Wysocki (2016).

  7. Another study that could perhaps explain some of these mixed results is that of Breza et al. (2018). The findings of their field experiment, conducted in India, help us to understand when pay disparities might or might not negatively affect worker productivity and thus firm performance.

  8. See also Branco and Delgado (2017) for a study of communication strategies in the wake of pay ratio disclosure.

  9. See SEC (2015c, p. 43, fn. 90; p. 263, fn. 703), SEC (2015d, para. 5), and White (2015, para. 28).

  10. See “Appendix 2” for a description of the method by which the seven types of exempt firms are identified.

  11. See the SEC’s (2017) descriptions of FPIs and MJDS filers in Topics 6 and 16, respectively.

  12. For EGCs, see Section 102(a) of Congress’s Jumpstart Our Business Startups (JOBS) Act of 2012. For ABS issuers and RICs, see the answers to questions 19 and 20, respectively, of the SEC’s (2015b) Generally Applicable Questions on Title I of the JOBS Act.

  13. The Code of Federal Regulations (C.F.R.) is published by the Office of the Federal Register National Archives and Records Administration.

  14. It should be noted that there may be some instances in which this is not the case (e.g., SEC 2015c, pp. 132–133).

  15. See “Appendix 3” for a description of the method by which TotalComp is calculated.

  16. As stated previously, it is required that a firm’s status as affected or unaffected remain constant over the test period 2011–2016. It is assumed that a firm’s status in 2010 is the same as that over the test period.

  17. For affected firms, Institutional Shareholder Services (ISS) Incentive Lab Participant Data by Fiscal Year is used to compute CEO tenure as one plus the difference between the current fiscal year and the first fiscal year for which the CEO held that position at the firm. For unaffected firms, CEO tenure is provided by Equilar: http://www.equilar.com/, an executive compensation and corporate governance data firm. There are two differences between affected and unaffected firms with respect to measured CEO tenure. First, measured tenure could be lower than actual for affected firms because 1998 is the first year covered in the abovementioned ISS dataset. Thus, a CEO that began at a firm prior to 1998 is restricted to have begun in 1998. Second, measured tenure for affected firms is restricted to whole numbers, whereas measured tenure for unaffected firms is not. These differences should not impact the estimation of %Resid_Total for two reasons: First, Eq. (1) is estimated by affected status. Second, Eq. (1) uses the natural logarithm of tenure.

  18. As noted, ln_Sales, BTM, ROA, and Industry require the following Compustat FA variables: AT, CSHO, IB, LT, PRCC_F, SALE, and SIC. Deleted are observations with missing values for any of these seven variables. In addition, due to ln_Sales, deleted are observations if SALE is less than or equal to zero. Likewise, due to BTM, deleted are observations if LT or market value (the product of PRCC_F and CSHO) is less than or equal to zero. Finally, due to ROA, deleted are observations with missing values for lagged AT. Also deleted are observations if AT or lagged AT is less than or equal to zero. BH_Ret requires the CRSP Monthly Stock variable RET. Deleted are observations with missing or nonnumeric values for RET. Moreover, after merging the CRSP Monthly Stock data with the Compustat FA data (using CUSIP numbers as firm identifiers) and computing BH_Ret, deleted are observations where BH_Ret does not reflect 12 months of data.

  19. As noted, Compustat PA is used in determining whether firms have defined benefit plans. Missing values are set to zero for PBPRO, PBPRU, and PBARR. Compustat FA is used in determining whether firms are in the top decile of sales per employee. Observations where SALE or EMP is missing or less than or equal to zero are deleted. However, this action is taken solely for the purpose of determining whether firms are in the top decile of sales per employee. Observations are not required to have Compustat FA data in the merged Compustat PA/Compustat FA dataset, nor are they required to have Compustat PA data. In the merged dataset, missing values are set to zero for Compustat PA data but left as is for Compustat FA data.

  20. Industry membership is determined by Fama–French 48 industry codes using data downloaded from Kenneth R. French’s website: http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/det_48_ind_port.html. Left unclassified in the download, but included in the sample, are four SIC codes: 3990, 6797, 9995, and 9997. Based on review of Fama and French (1997, Appendix A), this study classifies SIC codes 3990, 9995, and 9997 as industry code 48 (other/almost nothing) and SIC code 6797 as industry code 47 (trading).

  21. HighCustomer is based on Chakravarthy et al.’s (2014) variable LTProduct. The industry codes represent the following industries: consumer goods; healthcare; medical equipment; pharmaceutical products; machinery; electrical equipment; automobiles and trucks; aircraft; shipbuilding and railroad equipment; communication; personal services; business services; computers; electronic equipment; banking; insurance; real estate; and trading.

  22. HighCommunity is based on Chakravarthy et al.’s (2014) variable Retail. The industry codes represent the following industries: agriculture; food products; candy and soda; entertainment; consumer goods; healthcare; pharmaceutical products; construction; automobiles and trucks; precious metals; non-metallic and industrial metal mining; coal; petroleum and natural gas; communication; personal services; business services; computers; wholesale; retail; restaurants, hotels, and motels; banking; insurance; and real estate.

  23. Since Proposal and Adoption overlap in 2016, the effect of that year is subsumed by Adoption. Accordingly, in Eq. (2) and other models that include both Proposal and Adoption as independent variables, the proposal and adoption periods are 2014–2015 and 2016, respectively.

  24. When SP500 is used to test the public-scrutiny hypothesis, a different model, detailed below, is used.

  25. Data sources used in constructing the samples are noted in the research design (fourth) section and the appendices. Wharton Research Data Services (WRDS) was used in preparing this inquiry. This service and the data available thereon constitute valuable intellectual property and trade secrets of WRDS and/or its third-party suppliers.

  26. As noted in the research design (fourth) section, unlike the affected firms, the unaffected firms (SRCs) are not required to include change in pension value in the total compensation reported in the SCT. This study does not speculate as to the impact that this difference has on the inferences drawn from descriptive statistics (which, unless noted, are not based on tests of statistical significance). This study does, however, control for this difference in its research design by estimating percentage residual total compensation by affected status.

  27. The general format of Panel C of Table 1 is based on Balsam et al. (2016, Panel C, Table 1). The two-digit SIC code ranges are based on the Occupational Safety and Health Administration’s (n.d.b) SIC Division Structure.

  28. Untabulated statistics show that the mean (median) TotalComp for affected firms (N = 5003) over the test period is $9.03 ($7.14) million. Panel A of Table 4 shows that the mean %Resid_Total over the test period is 2.0% greater for affected (N = 5003) relative to unaffected (N = 1000) firms. Accordingly, the mean residual total compensation is $181 thousand ($9.03 million × 2.0%) greater for affected firms, and the median residual total compensation is $143 thousand ($7.14 million × 2.0%) greater for affected firms.

  29. The R2s reported in tables (e.g., Table 5) that report the results of estimations that include firm fixed effects are lower than one would expect for estimations that include firm fixed effects. This is because the models that include firm fixed effects are estimated using Stata’s xtreg, fe procedure. William Gould, StataCorp, notes “In the xtreg, fe procedure the R2 reported is obtained by only fitting a mean deviated model where the effects of the groups (all of the dummy variables) are assumed to be fixed quantities. So, all of the effects for the groups are simply subtracted out of the model, and no attempt is made to quantify their overall effect on the fit of the model” (n.d., para. 1).

  30. The mean TotalComp for affected firms (N = 5003) over the test period is $9.03 million (untabulated). The mean %Resid_Total for affected firms (N = 5003) over the test period is 1.9 percent (Table 4, Panel A). Accordingly, the mean residual total compensation for affected firms is $172 thousand ($9.03 million × 1.9%). Therefore, the significant coefficient of − 0.103 on Affected × Proposal × HighExcessPayt−1 indicates a relative curb in mean residual total compensation of 10.3% or $18 thousand ($172 thousand × 10.3%). Note that the 10.3% curb is estimated from a sample of 4467 affected (and 808 unaffected) observations.

  31. The mean residual total compensation for affected firms (N = 5003) over the test period is $172 thousand (fn. 30). Thus, the significant coefficient of − 0.164 on Affected × Proposal × HighMV indicates a relative curb on mean residual total compensation of 16.4% or $28 thousand ($172 thousand × 16.4%). Note that the 16.4% curb is estimated from a sample of 5003 affected (and 1000 unaffected) observations.

  32. The mean TotalComp for S&P 500 firms (N = 2192) over the test period is $12.64 million (untabulated). The mean %Resid_Total for S&P 500 firms (N = 2192) over the test period is 14.2% (untabulated). Accordingly, the mean residual total compensation for S&P 500 firms is $1.80 million ($12.64 million × 14.2%). Therefore, the significant coefficient of − 0.122 on SP500 × Proposal indicates a relative curb on mean residual total compensation of 12.2% or $220 thousand ($1.80 million × 12.2%). Note that the 12.2% curb is estimated from a sample of 2192 S&P 500 (and 2811 non-S&P 500) firms.

  33. The mean residual total compensation for affected firms (N = 5003) over the test period is $172 thousand (fn. 30). Thus, the significant coefficient of − 0.152 on Affected × Proposal × HighEmployee indicates a relative curb on mean residual total compensation of 15.2% or $26 thousand ($172 thousand × 15.2%). Note that the 15.2% curb is estimated from a sample of 5003 affected (and 1000 unaffected) observations.

  34. The mean residual total compensation for affected firms (N = 5003) over the test period is $172 thousand (fn. 30). Thus, the significant coefficient of − 0.128 on Affected × Proposal × HighCustomer indicates a relative curb on mean residual total compensation of 12.8% or $22 thousand ($172 thousand × 12.8%). Note that the 12.8% curb is estimated from a sample of 5003 affected (and 1000 unaffected) observations.

References

Download references

Acknowledgements

This paper is based on the author’s dissertation, which was completed at Florida State University, College of Business, Department of Accounting, 821 Academic Way, Tallahassee, FL 32306, USA. This paper has benefited from the input of several scholars. The author thanks Ronald Strauss (editor) and two anonymous referees for helpful comments during the review process. The author also thanks his dissertation chair, Rick Morton. His guidance was instrumental in the design and execution of this work. The author also thanks the other members of his dissertation committee: Bruce Billings, Tim Zhang, and Yingmei Cheng. The author finally thanks the faculty and doctoral students of Florida State University as well as the faculty of Xavier University for comments on this study’s proposal.

Author information

Author notes

  1. Tristan B. Johnson

    Present address: Department of Accounting, Mitchell College of Business, University of South Alabama, 5811 USA Drive South, Mobile, AL, 36688, USA

Authors and Affiliations

    Authors
    1. Tristan B. Johnson
      View author publications

      You can also search for this author in PubMed Google Scholar

    Corresponding author

    Correspondence to Tristan B. Johnson.

    Ethics declarations

    Conflict of interest

    The author states that there is no conflict of interest.

    Additional information

    Publisher's Note

    Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

    Appendices

    Appendix 1

    Variable definitions

    Variable

    Definition

    CEO compensation expectation model variables

    TotalComp

    The total CEO compensation for the fiscal year disclosed by the firm in column (j) of its Summary Compensation Table. Data are obtained from Institutional Shareholder Services (via Wharton Research Data Services) for affected firms and Equilar for unaffected firms. See the research design (fourth) section of the paper and “Appendix 3”

    ln_TotalComp

    The natural logarithm of TotalComp

    ln_Tenure

    The natural logarithm of the CEO’s tenure in years for the fiscal year. Data are obtained from Institutional Shareholder Services (via Wharton Research Data Services) for affected firms and Equilar for unaffected firms. Due to differences between ISS and Equilar datasets, tenure is measured differently for affected and unaffected firms. See the research design (fourth) section of the paper for further details

    ln_Sales

    The natural logarithm of net sales (Compustat Fundamentals Annual SALE) for the fiscal year

    BTM

    Book-to-market ratio for the fiscal year, calculated as total assets (Compustat Fundamentals Annual (FA) AT) divided by the sum of total liabilities (Compustat FA LT) and market value (the product of Compustat FA PRCC_F and CSHO)

    BH_Ret

    The buy-and-hold return for the fiscal year (calculated using CRSP Monthly Stock RET)

    ROA

    Return on assets for the fiscal year, calculated as net income (Compustat Fundamentals Annual (FA) IB) divided by average total assets (Compustat FA AT)

    Dependent variable (for hypotheses tests)

    %Resid_Total

    Percentage residual total compensation for the fiscal year, calculated as the residual that results from regressing ln_TotalComp on its economic determinants over the period 2010–2016. See Eq. (1)

    Independent variables (for hypotheses tests)

    Affected

    An indicator variable that equals one for a select group of firms that are required to comply with the rule (affected firms) and zero for a select group of firms that are not required to comply with the rule (unaffected firms). See the research design (fourth) section of the paper and “Appendix 2”

    Proposal

    An indicator variable that equals one for observations with fiscal years that end in the period 2014–2016 and zero otherwise

    Adoption

    An indicator variable that equals one for observations with fiscal years that end in 2016 and zero otherwise

    HighExcessPay

    An indicator variable that equals one for affected (unaffected) firms for which %Resid_Total for the fiscal year is greater than the median of the pooled sample of affected (unaffected) firms and zero otherwise

    Variable

    Definition

    HighMV

    An indicator variable that equals one for affected (unaffected) firms for which market value, as defined in the BTM calculation, for the fiscal year is greater than the median of the pooled sample of affected (unaffected) firms and zero otherwise

    SP500

    An indicator variable that equals one for firms in the S&P 500 (Compustat Index Constituents TIC equals I0003) for the fiscal year and zero otherwise

    HighEmployee

    An indicator variable that equals one for firms with defined benefit plans or firms with high sales per employee for the fiscal year and zero otherwise. Firms with defined benefit plans are “identified as those with non-zero values for Compustat [Pension Annual] variables PBPRO [projected benefit obligation], PBPRU [projected benefit obligation (underfunded)], or PBARR [pension benefits discount rate]” (Chakravarthy et al. 2014, p. 1338) for the fiscal year. Firms with high sales per employee are identified “as those in the top decile of sales [Compustat Fundamentals Annual (FA) SALE] per employee [Compustat FA EMP] (calculated by year across all available Compustat firms)” (Chakravarthy et al. 2014, p. 1338) for the fiscal year

    HighCustomer

    An indicator variable that equals one for firms in the Fama–French 48 industry codes 9 (consumer goods), 11 (healthcare), 12 (medical equipment), 13 (pharmaceutical products), 21 (machinery), 22 (electrical equipment), 23 (automobiles and trucks), 24 (aircraft), 25 (shipbuilding and railroad equipment), 32 (communication), 33 (personal services), 34 (business services), 35 (computers), 36 (electronic equipment), 44 (banking), 45 (insurance), 46 (real estate), or 47 (trading) (Chakravarthy et al. 2014, p. 1338) for the fiscal year and zero otherwise

    HighCommunity

    An indicator variable that equals one for firms in the Fama–French 48 industry codes 1 (agriculture), 2 (food products), 3 (candy and soda), 7 (entertainment), 9 (consumer goods), 11 (healthcare), 13 (pharmaceutical products), 18 (construction), 23 (automobiles and trucks), 27 (precious metals), 28 (non-metallic and industrial metal mining), 29 (coal), 30 (petroleum and natural gas), 32 (communication), 33 (personal services), 34 (business services), 35 (computers), 41 (wholesale), 42 (retail), 43 (restaurants, hotels, and motels), 44 (banking), 45 (insurance), or 46 (real estate) (Chakravarthy et al. 2014, p. 1338) for the fiscal year and zero otherwise

    Appendix 2

    Identification of firms that are not required to comply with the rule

    This “Appendix” describes the method by which the types of firms that are not required to comply with the rule are identified. As noted in the research design (fourth) section of the paper, exempt from the rule are the following types of firms: foreign private issuers (FPIs), US-Canadian Multijurisdictional Disclosure System (MJDS) filers, emerging growth companies (EGCs), registered investment companies (RICs), asset-backed securities (ABS) issuers, wholly owned subsidiaries, and smaller reporting companies (SRCs). For several reasons, as discussed in the research design section, only SRCs are included in the control group of unaffected firms, and the other six types of exempt firms are therefore excluded from the sample. The method by which these types of firms are identified (and, if applicable, excluded) is discussed below.

    The identification begins with a sample of Compustat Fundamentals Annual (FA) firms over the test period 2011–2016. The identification requires the following Compustat FA variables: consolidation level (CONSOL), currency (CURCD), country in which the firm is headquartered (LOC), common shares outstanding (CSHO), closing stock price at the end of the fiscal year (PRCC_F), net sales (SALE), debt issuance (DLTIS), and Standard Industrial Classification Code (SIC). Missing values are set to zero for DLTIS but observations are deleted if missing values for CONSOL, CURCD, LOC, CSHO, PRCC_F, SALE, or SIC.

    The sample first excludes non-US firms, which should, as noted in the research design section, exclude most FPIs and MJDS filers. To exclude non-US firms, the sample excludes observations that do not have headquarters in the US [LOC does not equal USA (see Srinivasan et al. 2015)] or those that do not have data in US dollars (CURCD does not equal USD).

    Next, the sample excludes EGCs, RICs, and ABS issuers. The SEC’s (2017) Financial Reporting Manual is used to identify EGCs. According to the SEC (2017, p. 316), a firm is an EGC if it meets the following requirements:

    • “It had total annual gross revenues of less than $1 billion during its most recently completed fiscal year.”

    • “It has either (1) not yet had or (2) had after December 8, 2011, its first sale of common equity securities pursuant to an effective registration statement under the Securities Act of 1933.”

    • “It has not met any of the disqualifying provisions.”

    Regarding the disqualifying provisions, the SEC (2017, pp. 317–318) states that an EGC is no longer classified as such upon the earliest of any of the following events:

    • “The last day of the fiscal year in which its total annual gross revenues are $1 billion or more.”

    • “The last day of the fiscal year following the fifth anniversary of the date of the first sale of common equity securities of the issuer under an effective Securities Act registration statement as an EGC.”

    • “The date on which it has issued more than $1 billion in non-convertible debt in the previous three years.”

    • “The date on which it becomes a large accelerated filer.”

    Based on these requirements, classified as EGCs are observations that pass five tests: (1) the revenue test, (2) the public age test, (3) the debt test, (4) the large accelerated filer test, and (5) the historical status test. The revenue test is that net sales (SALE) for the current fiscal year are less than $1 billion. In implementing this test, deleted are observations with net sales (SALE) that are less than or equal to zero. The public age test is that the number of years since the firm’s securities have been traded, calculated as one plus the difference between the current fiscal year and the first fiscal year with a non-missing value for the closing stock price (PRCC_F), is less than five (see Impink 2015). The debt test is that the sum of debt issuances (DLTIS) over the past three fiscal years, including the current fiscal year, is less than $1 billion. To prevent a firm from having missing values for DLTIS in a year prior to its existence, DLTIS is set to zero in fiscal years t – 1 and t – 2 (fiscal year t – 2) if the firm’s age in years, calculated as one plus the difference between the current fiscal year and the first fiscal year for which the firm is covered in Compustat (Armstrong et al. 2013b), is equal to one (two). After this adjustment, all remaining observations with missing values for DLTIS in fiscal year t – 1 or t – 2 are deleted. Finally, observations are deleted if DLTIS is less than zero in fiscal year t, t – 1, or t – 2. The large accelerated filer test is that market value (used as a measure of public float), calculated as the product of the closing stock price (PRCC_F) and common shares outstanding (CSHO), for the current fiscal year is less than $700 million (SEC 2009). The historical status test is that the firm has qualified as an EGC (i.e., passed the revenue test, the public age test, the debt test, and the large accelerated filer test) in not only the current fiscal year but also all prior fiscal years. If an observation passes the five aforementioned tests, then it is deleted, thereby excluding EGCs.

    RICs can be classified as face-amount certificate companies, unit investment trusts, or management companies, and management companies can be classified as open-end companies or closed-end companies (Congress 2018, p. 23). Open-end companies fall under SIC code 6722, while closed-end companies, face-amount certificate companies, and unit investment trusts fall under SIC code 6726 (Occupational Safety and Health Administration n.d.a). Accordingly, RICs are excluded by deleting observations with the SIC codes 6722 or 6726. In a similar manner, ABS issuers are excluded by deleting observations that fall under SIC code 6189 (SEC 2015a).

    Considering that wholly owned subsidiaries have also been excluded (by requiring that CONSOL equal C), SRCs now remain as the only type of exempt firm in the sample. In other words, the sample now consists of SRCs and affected firms. SRCs are then identified according to the criteria per the SEC’s (2017, pp. 233–234) Financial Reporting Manual:

    • “An entity is a smaller reporting company if it has a public float … of less than $75 million and it is not an investment company, asset-backed issuer or majority-owned subsidiary of a parent that is not a smaller reporting company.”

    • “If the public float of the issuer is zero because the issuer had no public equity outstanding or no market for its equity existed, the issuer must have annual revenues of less than $50 million, as reported in its most recent fiscal year (12 months) for which audited financial statements are available.”

    Based on these requirements, as well as the fact that the sample has already excluded RICs, ABS issuers, and wholly owned subsidiaries, SRCs are identified as observations that meet either of two conditions. One is that market value (used as a measure of public float) for the current fiscal year is between zero and $75 million. The other is that market value for the current fiscal year is zero and net sales for the current fiscal year are less than $50 million. The end result of this process is therefore a Compustat FA dataset that is limited to affected firms and SRCs (which serve as the control group of unaffected firms).

    Appendix 3

    Calculation of total CEO compensation disclosed in the Summary Compensation Table

    This “Appendix” describes the method by which the total CEO compensation disclosed by a firm in its Summary Compensation Table (SCT), TotalComp, is calculated. The data required for TotalComp come from two sources. The first is Institutional Shareholder Services (ISS). ISS Incentive Lab Participant Data by Fiscal Year is used to obtain CEO-firm-year (PARTICIPANTID-CIK-FISCALYEAR, where CURRENTCEO = 1) data. Excluded are all observations for a firm in a year if it has more than one observation for that year. This screen should exclude observations for which CEO compensation does reflect the entire year.

    SCT data for these CEO-firm-years are then obtained from ISS Incentive Lab Summary Compensation Table. Missing values are set to zero for the eight SCT variables: salary (SALARY), bonus (BONUS), stock awards (STOCKAWARDS), option awards (OPTIONAWARDS), nonequity incentive plan compensation (NONEQUITYCOMP), change in pension value and nonqualified deferred compensation earnings (PENSIONNQDC), all other compensation (OTHERCOMP), and total compensation (TOTALCOMP). Observations for which TOTALCOMP does not equal the sum of its components are deleted. Also deleted are observations for which any of the eight SCT variables is less than zero. An additional restriction is imposed on total compensation by deleting observations for which TOTALCOMP is not greater than zero. Finally, observations are deleted if any of the pre-2006 SCT variables LEGACYOPTIONS, LEGACYLTIP, or LEGACYOTHERCOMP is not either missing or equal to zero. TotalComp is then set equal to the ISS variable TOTALCOMP.

    The CEO compensation (ISS) data is then merged with the affected/unaffected firm (Compustat Fundamentals Annual (FA)) data, as described in “Appendix 2,” by matching on CIK and fiscal year end date. This requires a third ISS Incentive Lab dataset, Company Data by Fiscal Year, because the other two abovementioned ISS datasets do not include fiscal year end date.

    Following the merge, TotalComp is missing for almost all of the unaffected firms. Accordingly, Equilar (http://www.equilar.com/), an executive compensation and corporate governance data firm, is contracted with to collect and provide CEO-firm-year data for a sample of unaffected firms. The sample of unaffected firms must have (1) at least one observation in both the pre-proposal period (2011–2013) and the post-proposal period (2014–2016), (2) at least one observation with no ISS data for both the current and prior year, (3) Compustat FA and CRSP Monthly Stock data to calculate the determinants of CEO pay, and (4) a non-missing CIK (the firm ID used to merge ISS and Compustat FA data). From the Equilar data, excluded are (1) all observations for a firm in a year if it has more than one observation for that year, (2) observations for which the CEO’s tenure, a variable provided by Equilar, is less than 1 year, and (3) observations for which the CEO does not serve through the end of the fiscal year, using a resign date variable provided by Equilar. These screens should exclude observations for which CEO compensation does reflect the entire fiscal year. Also excluded are all observations for a CEO if their start date at a firm is not consistent across fiscal years and observations for which the CEO’s tenure is missing.

    Regarding the Equilar SCT data, missing values are set to zero for the SCT variables: salary, bonus, stock awards, option awards, nonequity incentive plan compensation, change in pension value and nonqualified deferred compensation earnings, all other compensation, and total compensation. With respect to stock awards and option awards, Equilar provides two values of each. One shows awards in accordance with FASB ASC 718, while the other shows awards in accordance with FAS 123R. Deleted are observations with non-zero values for both FASB ASC 718 and FAS 123R stock awards and option awards. Equilar also breaks nonequity incentive plan compensation into its short- and long-term components. Deleted are observations for which nonequity incentive plan compensation is not the sum of its components. Likewise, deleted are observations for which change in pension value and nonqualified deferred compensation is not the sum of its pension and nonqualified deferred compensation components—both of which are provided by Equilar. Also deleted are observations for which total compensation is not the sum of its components. Further deleted are observations for which total compensation is not greater than zero and observations for which any of the seven other SCT variables is less than zero. TotalComp is then set equal to the total compensation provided by Equilar.

    The unaffected firm CEO compensation (Equilar) data is then merged with the affected/unaffected firm (Compustat FA) data by matching on CIK and fiscal year end. Since ticker symbols are the firm identifiers for Equilar, deleted are observations for which a ticker-CIK match cannot be made. Lastly, for firms that have data from both ISS and Equilar, a requirement is imposed that each observation have the same value for the ISS and Equilar SCT variables.

    Rights and permissions

    Reprints and permissions

    About this article

    Check for updates. Verify currency and authenticity via CrossMark

    Cite this article

    Johnson, T.B. The effects of the mandated disclosure of CEO-to-employee pay ratios on CEO pay. Int J Discl Gov 19, 67–92 (2022). https://doi.org/10.1057/s41310-021-00128-y

    Download citation

    • Received:

    • Accepted:

    • Published:

    • Issue Date:

    • DOI: https://doi.org/10.1057/s41310-021-00128-y

    Keywords

    Search

    Navigation