1 Introduction

A crucial component of effective corporate governance is an executive compensation scheme that incentivizes appropriate value-enhancing actions (Kay, 2017; Ritz, 2022). In particular, boards should include performance measures in executive pay plans that provide specific direction and motivation towards value-creating goals. Both short-term operational objectives and long-term strategic priorities are essential to overall organizational success (Kaplan, 1994; Kaplan & Norton, 2008). Long-term strategies need to be operationalized through short-term goals and targets. As such, well-governed executive pay plans should be designed to incentivize both short-term operational objectives and long-term strategic priorities. For this reason, CEO performance-based incentives typically comprise both annual cash bonuses and longer-term equity incentives. The latter primarily serve to align CEO pay with long-term shareholder return and, therefore, predominantly include Total Shareholder Return (TSR) and/or earnings per share (EPS) performance measures (Conyon et al., 2000; Abernethy et al., 2015). In contrast, annual bonus plans are intended to incentivize short-term performance on specific operational objectives that will ultimately increase shareholder value in the long run (Murphy, 1999; Murphy & Jensen, 2011; Larcker & Tayan, 2015). For example, First Solar Inc. (2015: 24) states that its bonus plan “provides at-risk variable compensation linked to short-term corporate, organizational, and strategic goals”.

Many authors highlight the value of linking performance measures to organizational goals and strategies (see, for example, Kaplan & Norton, 1993, 2001; Dossi & Patelli, 2010; Franco-Santos et al., 2012; Abernethy et al., 2021; Bellisario et al., 2021). As such, the literature has long been interested in the question of whether performance measures in CEO pay plans are fit for purpose. For long-term equity incentives, this is relatively easy to investigate by examining the overall correlation between CEO compensation and shareholder returns (e.g., TSR) (Murphy, 2012). In contrast, boards choose from a wider array of performance measures for bonus plans, in comparison to equity incentives (Bloomfield et al., 2021). Since short-term objectives vary across firms depending on firm-specific circumstances, boards need to clearly communicate specific and targeted objectives to provide meaningful incentives. Therefore, prior literature exploring the effectiveness of CEO annual bonus plans has examined the link between firm-specific characteristics and the choice of bonus plan performance measures (Lambert & Larcker, 1987; Bushman et al., 1996; Ittner et al., 1997; Murphy & Oyer, 2001). These studies provide modest evidence to suggest that boards choose CEO bonus performance measures to direct CEO effort towards value-creating actions.

However, the previous research concentrates on broad classifications of bonus plan performance measures—accounting versus market measures (Lambert & Larcker, 1987), individual versus corporate measures (Bushman et al., 1996), and financial versus non-financial measures (Ittner et al., 1997). CEO bonus plans have evolved over time to include multiple performance measures. We lack insights on boards’ aptitude to link short-term operational objectives to precise performance measures. Virtually every US public firm incentivizes the enhancement of short-term firm performance with annual executive cash bonus payments. They are an important element of overall CEO performance-based pay, comprising approximately 40% of CEO performance-based incentives in fiscal 2014 (the period of this study). Clearly then, boards view annual cash bonuses as an important vehicle for directing CEO effort at specific short-term actions. Yet, the business press constantly criticize CEO bonus payments for being excessively high and overly complex (Rapoport, 2014; Francis & Fuhrmans, 2019). This tension surrounding CEO bonus payments motivates us to examine the alignment between specific bonus plan performance measures and short-term operational firm objectives.Footnote 1

This is a pertinent governance question for two reasons. First, the execution of relevant, short-term, measurable objectives should generate long-term shareholder value. Depending on firm-specific circumstances, different short-term actions, such as increasing revenues, controlling costs, or managing cashflows, may be more appropriate at different times. It is essential for boards to include in bonus plans the relevant performance measures that will direct and motivate CEO effort towards desirable actions. Second, although smaller in magnitude than equity pay, bonuses can have stronger incentive effects (Murphy, 2012). For instance, TSR is a volatile market measure that is affected by many external factors outside management’s control (Kay, 2017). Research suggests that, for performance-related pay schemes to be successful, employees need to believe they can influence the evaluated performance outcomes (Kauhanen & Piekkola, 2006). CEOs are unlikely to be highly incentivized if they cannot identify a link between their current actions and future TSR. In contrast, CEOs are likely to be aware of actions they can take now to meet short-term operational objectives such as growing market share or maintaining dividend payments. Furthermore, given that firms typically settle bonuses in tangible cash in a short timeframe, the immediacy and certainty of cash bonuses can provide for stronger incentives than risky equity compensation (Murphy, 2012).

We hand-collect data on pre-established quantitative performance measures in CEO annual bonus plans from S&P 500 firm Compensation Discussion and Analysis (CD&A) disclosures. We capture boards’ reliance on each performance measure through the weight assigned to the measure. Such detailed information is not available in large-scale databases such as Execucomp and ISS Incentive Lab. While ISS Incentive Lab contains data on executive performance-measure types and weights, we believe our dataset is superior for several reasons.Footnote 2 In designing the study, we faced a trade-off between higher-quality hand-collected cross-sectional data and lower-quality panel data available from proprietary sources. We opted to hand-collect higher-quality cross-sectional data for fiscal year 2014. We acknowledge the limitation of our dataset to one fiscal year and we address endogeneity concerns by hand-collecting additional performance-measure data on a subsample of firms in fiscal year 2021.

Our paper provides two contributions: one to practice and one to theory. First, we demonstrate that boards fail to incentivize value-enhancing CEO effort in some circumstances, consistent with Tosi’s (2008: 160) observation that explanations for incentive alignment and incentive monitoring are incomplete. Specifically, CEOs of firms with relatively higher operating expenses should be motivated to seek operational efficiencies and cost savings. However, we find that performance measures to capture such desirable cost-cutting actions are not appropriately emphasized in CEO bonus plans. The practical implication of this finding is that boards need to dedicate more scrutiny to the weighting of performance metrics in CEO bonus plans. CEO performance metrics should be carefully reevaluated on an annual basis to ensure strong alignment with firm-specific circumstances and objectives. Where rising cost structures put downward pressure on profit margins, managers must be incentivized to immediately bring costs under control. Otherwise, the inability to reduce costs fast enough may result in business failure (Amankwah-Amoah & Zhang, 2015).

Second, our fiscal 2021 subsample results highlight the increasing importance of social-and-environmental performance to CEO bonus payouts. Specifically, social-and-environmental measures account for 8.2% of CEO bonus in fiscal 2021, in comparison to 3.6% in fiscal 2014. Our findings suggest a growing belief across boards that linking executive pay to ESG (Environmental, Social, Governance) measures will help to alleviate rising stakeholder concerns about firms’ sustainability practices. Yet, optimal-contracting theory tells us that performance measures should only be included in executive pay if they are relevant and informative (Hölmstrom, 1979; Bushman et al., 1996). We observe a greater reliance on sustainability measures in CEO bonus plans, yet the literature lacks a clear understanding of this governance approach. For this reason, we call on academic scholars to explore whether ESG-based incentives are indeed optimal, or whether boards simply pay lip service to sustainability.

The remainder of the paper is organized as follows. The next section reviews the literature and develops the hypotheses for testing. Then we describe the sample, data sources, and variables for testing. A discussion of the results and robustness checks follows. The final section considers the findings and provides some concluding comments.

2 Theory and hypotheses

2.1 Optimal-contracting theory

In a typical principal-agent model, CEO action (or level of effort) is not directly observable by shareholders. However, the outcome of a performance measure can be used as a signal about otherwise unobservable CEO action. Hölmstrom (1979) argues that optimal contracts are those that incentivize desirable CEO actions by including performance measures that are informative of such actions. Bushman et al. (1996) apply optimal-contracting theory to the inclusion of individual performance measures in CEO annual incentive plans. They conjecture that CEO annual bonus will be based in part on individual performance evaluation (IPE) when corporate financial measures do not fully capture value-enhancing CEO actions. Bushman et al. (1996) find a positive relation between the use of IPE in CEO bonus plans and both firm growth opportunities and firm product time horizon. Ittner et al. (1997) extend Bushman et al. (1996) by investigating the use of non-financial performance measures in CEO bonus plans. They distinguish between prospectors who follow a differentiation strategy, aggressively seeking growth opportunities; and defenders who follow a cost-leader strategy, focusing on efficiencies. They find that non-financial metrics are more popular with firms following a differentiation strategy, suggesting that financial measures alone do not incentivize managerial effort in strategic areas such as growth.

Similarly, Murphy & Oyer (2001) examine the role of discretion in executive incentive contracts. They again invoke optimal contracting to identify situations where CEO contribution to firm value cannot be fully evaluated through objective measures, instead requiring a subjective assessment of individual performance. Murphy & Oyer (2001) report a higher use of discretion in CEO bonus plans by larger firms and privately held firms. Larger firms, with more employees, suffer from more internal noise factors such as a temptation to free ride. The performances of other executives and employees will directly impact firm-wide objective measures, hence the need for individual performance assessments. Privately held firms do not have market-based measures of shareholder value, and therefore must rely more heavily on individual performance measures to provide incentives.

De Angelis & Grinstein (2015) examine performance criteria in CEO compensation contracts across both equity and non-equity awards. They discover that young and growing firms rely more heavily on stock price and sales-growth performance measures. In contrast, more mature firms with fewer growth opportunities rely more heavily on accounting returns. Although De Angelis & Grinstein (2015) obtain performance criteria data from CD&A disclosures, our paper differs from their study. We focus on the association between quantitative performance measures in CEO bonus plans and short-term operational objectives. De Angelis & Grinstein (2015) explore the influence that a long-term growth strategy has on the choice of performance measures, with results reported at an aggregate level across all performance-based awards (short- and long-term).

Inversely related to the informational value of performance measures is the noise inherent in these measures. Noise refers to the level of precision with which performance measures capture true managerial performance. Optimal-contracting theory suggests that the relative weight placed on a performance measure is a decreasing function of the exogenous noise in the measure. Lambert & Larcker’s (1987) findings support this hypothesis by reporting a negative relation between the weight placed on accounting performance in compensation plans and the variance of the accounting performance measure. Core et al. (2003) document similar results, albeit only in relation to CEO cash pay as opposed to CEO total compensation.

In summary, this strand of literature suggests that the choice of performance measures included in CEO incentive plans is important. Performance measures provide information about CEO actions, and actions that are value-enhancing should be incentivized through the use of informative measures.

2.2 Hypotheses development

The literature differentiates between long-term strategies and short-term objectives (Kaplan, 1994; Kaplan & Norton, 2008; Heesen, 2016). To execute strategy effectively, management must operationalize long-term strategies into short-term objectives. Long-term strategic plans are typically three-year or five-year plans (Steiner, 2010). Short-term objectives are those “whose outcomes are achievable in the next month, quarter, or year” (Heesen, 2016: 70). The pursuit of short-term objectives should be incentivized through annual bonus plans (Murphy, 1999; Murphy & Jensen, 2011; Larcker & Tayan, 2015). Therefore, if boards function effectively, we expect to see an alignment between specific bonus plan performance measures and short-term firm objectives. We formulate three hypotheses to examine this assumption of effective governance. Short-term objectives vary greatly across firms and it is difficult to identify the most pertinent objective of every firm. However, an analysis of firms’ financial ratios provides guidance on the relevance of particular operational objectives, depending on firm-specific circumstances.

For instance, firms with higher Market to Book (MTB) ratios are viewed as possessing higher growth opportunities (Ittner et al., 1997; Johnson, 2003; Angelis & Grinstein, 2015). Boards of such firms should incentivize their CEOs to exploit growth opportunities in the near term by increasing sales volumes, introducing new products, and growing market share. These actions are best captured through sales and growth performance measures in CEO bonus plans. In contrast, firms that report relatively higher amounts of operating expenses likely need to seek variable cost savings from improvements in operational efficiencies (Kaplan & Norton, 2001). Cost-reduction efforts should be incentivized through income measures, accounting-returns measures, and cost-control measures. Lastly, the cash-flow adequacy ratio reflects the capacity of firms to cover their immediate outgoings through operating cash flows (O’Regan, 2015). Firms with lower cash-flow adequacy ratios face more liquidity pressures to maintain their capital expenditures, dividend obligations, and debt repayments. These cash-strapped firms should concentrate efforts on improving cash flow and aggressively managing working capital (Boisjoly et al., 2020). A greater reliance on cash-based measures in CEO bonus plans should direct CEO effort towards cash-management objectives. This discussion leads to the following three hypotheses:

H1

Firm MTB ratio is positively associated with the use of sales and growth measures in CEO bonus plans.

H2

Firm operating expense ratio is positively associated with the use of income measures, accounting-returns measures, and cost-control measures in CEO bonus plans.

H3

Firm cash-flow adequacy ratio is negatively associated with the use of cash-based measures in CEO bonus plans.

3 Research design

3.1 Sample and data

The initial sample includes the S&P 500 set of firms.Footnote 3 We hand-collect data on performance measures in fiscal 2014 CEO annual bonus plans from firm CD&As, as published in 2014 and 2015 proxy statements.Footnote 4 Of the S&P 500, 445 firms award CEO bonus payments based on pre-established performance measures.Footnote 5 We extract accounting and board data from Compustat and Refinitiv Datastream.

3.2 Performance-measure variables

CD&A disclosures reveal boards’ choice of both more objective quantitative measures and more subjective qualitative measures of performance. Where boards rely on more than one quantitative measure, they tend to assign weights to the individual measures ex ante and assess their outcomes in a formulaic manner ex post. These weights allow us to clearly determine the measures on which boards place more reliance in particular situations. Conversely, qualitative measures are typically evaluated on an overall basis ex post without the use of individual weights. Based on these subjective evaluations, boards award discretionary adjustments to formulaic bonus payouts. As such, it is often unclear which individual qualitative measures better incentivize CEO actions. For this reason, we concentrate our analysis on quantitative performance measures. We capture boards’ reliance on each quantitative performance measure through the weight assigned to the measure, as disclosed in CD&As.Footnote 6

We observe six types of quantitative performance measures in CEO bonus plans: (1) sales and growth (SALES_GROW); (2) income-based (INCOME); (3) accounting returns and cost control (ACCRET_COST); (4) cash-based (CASH); (5) social and environmental (SOC_ENV); (6) other metric (OTHER). The appendix presents samples of the coding for each performance-measure type.

3.3 Explanatory variables

Following Ittner et al. (1997), Johnson (2003), and De Angelis & Grinstein (2015), we calculate the ratio of market value of assets to book value of assets to measure firm MTB, where the market value of assets equals the book value of assets minus the book value of equity plus the market value of equity.Footnote 7 We employ the industry-adjusted variant of this ratio (AMTB) by measuring the difference between the MTB of the firm and the respective average MTB ratio of all firms in the industry group. The firm operating expense ratio (OPEX) is computed by dividing total operating expenses by net sales. We calculate firm cash-flow adequacy (ADEQUACY) by dividing operating cash flow by the sum of long-term debt repayments, capital expenditures, and dividends paid.

Prior research has captured the long-term strategic orientation of firms by computing a rolling average of selected financial ratios over the prior five years (Higgins et al., 2015; Abernethy et al., 2019). However, our focus is on short-term objectives that are captured through short-term performance criteria. Similar to Delmas et al. (2015), we measure short-term performance through single-year observations of our selected financial ratios. Boards design CEO bonus plans at the start of the fiscal year for which CEO performance is to be reviewed. Therefore, in choosing performance measures for inclusion in bonus plans, short-term objectives at the end of the prior fiscal year should influence this choice. Thus, all explanatory variables are for the fiscal year preceding (i.e., 2013) the year for which performance measures are selected (i.e., 2014). The use of lagged explanatory variables also partially addresses endogeneity by alleviating the reverse causality problem.

3.4 Control variables

Previous research has found a relation between the noise in performance measures and the choice of measure to be included in CEO plans (Lambert & Larcker, 1987; Ittner et al., 1997). Noise is typically measured through the time-series variability in the relevant performance measure. To control for the noise in each performance-measure type, we identify the most common measure in each category (Table 1). The sales and growth category is represented through revenue (SALES). Earnings per share (EPS) denotes the income-based measures. We capture accounting-returns measures through return on capital (ROC). Free cash flow (FCF) proxies for the cash-based metrics.Footnote 8 The variability (natural logarithm of the standard deviations) in each of these measures over the previous five years is used as a proxy for the relevant measure type’s noise (NOISE).1

CEO power, or influence, over boards may also affect the design of CEO compensation (Bebchuk & Fried, 2004). For example, Morse et al.’s (2011) evidence suggests that powerful CEOs can persuade boards to direct their performance assessments away from unfavorable measures and emphasize the positive areas of performance. Cho et al. (2019) find that CEO power is significantly negatively related to the propensity of using non-financial performance measures in CEO bonus plans. We control for CEO-power effects in our model through three indicators that likely capture different aspects of CEO power (Morse et al., 2011): (1) board size (BSIZE); (2) proportion of independent directors (INDIR); (3) CEO duality (DUAL). Larger boards suffer from poor communication, lack of coordination, and less cohesion, suggesting that these boards are not as efficient in constraining managerial power (Bebchuk & Fried, 2004; van Essen et al., 2015). A higher proportion of independent directors should reduce CEO ability to influence decisions, as these outside board members are less exposed to conflicts of interest (Abernethy et al., 2015). Dual CEOs, who also hold the additional position of chair of the board, are considered to have more influence due to the concentration of decision-making power in one individual (van Essen et al., 2015). BSIZE is measured by the total number of directors who serve on the board. The proportion of independent directors on the board (INDIR) is the number of independent directors divided by the total number of directors. DUAL is coded one if the CEO is also the chair, and zero otherwise.

Firm life-cycle has also been found to influence performance-measure choice (De Angelis & Grinstein, 2015). We capture the life-cycle stage of the firm by measuring firm age (AGE), defined by the year the firm was founded. We control for firm size through the natural logarithm of lagged sales revenue (LNSALES). Finally, we include a set of dummy variables in all model specifications to represent firm industry.Footnote 9

4 Results

4.1 Descriptive findings

CEOs earned an average performance-based bonus of $2.38 million in fiscal 2014 (Table 2, panel A). Bonus payments vary considerably across firms, ranging from zero to $22.8 million. Panel A also reports the number of performance measures used in CEO bonus plans, showing that, on average, firms rely on three performance measures to evaluate CEO annual performance. The frequency distribution of the number of performance measures (Table 2, panel B) indicates that only 21% of firms rely solely on one measure, with the majority of firms (62%) including between two and four performance measures in CEO bonus plans.2

Table 1 Frequency distribution of performance measures in CEO bonus plans
Full size table

Table 3 reports descriptive statistics for the variables used in the main analysis. Income-based measures play a substantial role in annual bonus plans, comprising 54% of CEO performance on average. Sales and growth measures are also popular, with firms assigning an average weight of almost 19% to them. This is the first study of which we are aware to report firms’ usage of social-and-environmental measures in CEO incentive plans. The prior literature (Ittner et al., 1997; Angelis & Grinstein, 2015) has simply coded such measures as ‘non-financial’, suggesting an element of ambiguity in their evaluation. All the social-and-environmental measures included in our sample are quantified by firms through objective and verifiable scoring mechanisms such as survey results, safety incident rates, diversity percentages, etc.

Table 2 Sample characteristics
Full size table

4.2 Main results: hypotheses testing

As the dependent variables are proportions, their values are limited between zero and one and continuous within this range. The Ordinary Least Squares (OLS) model can produce inconsistent estimates for such data, with the Tobit regression model being proposed as a more appropriate solution (Long, 1997). We run two-limit Tobit estimations (left censored at zero and right censored at one) with the dependent variables being the proportions of CEO bonus assigned to sales and growth, income-based, accounting-returns and cost-control, and cash-based measures.Footnote 10 To address the possibility that outliers have arisen through measurement error, we winsorize all continuous variables at the top and bottom 1% and rerun the main analysis. The results are consistent and hence we retain the unwinsorized models.

H1 predicts a positive association between firm MTB ratio and the use of sales and growth measures in CEO bonus plans. We estimate the following model to examine H1:

$$SALES\_GROW_{t} {\text{ }} = a + \beta _{1} AMTB_{{t - 1}} + \beta _{{2{\text{ }}}} OPEX_{{t - 1}} + \beta _{{3{\text{ }}}} ADEQUACY_{{t - 1}} + \beta _{{4{\text{ }}}} NOISE\_SALES_{{t - 1\,to\,t - 5}} + \beta _{{5~}} BSIZE_{{t - 1}} + \beta _{{6~}} INDIR_{{t - 1}} + \beta _{{7~}} DUAL_{{t - 1}} + \beta _{{8~}} AGE_{t} + \beta _{9} LNSALES_{{t - 1}} + \gamma _{1} ~INDUSTRY~CONTROLS + \varepsilon $$
(1)

The results are reported in Table 4. Column 1 shows a significantly positive coefficient on AMTB (p < 0.01), suggesting that firms with higher growth opportunities tie a larger portion of CEO bonus to sales and growth performance measures. For example, the coefficient of 0.061 in column 1 indicates that, all else being equal, an increase of one in industry-adjusted firm MTB is associated with a 6% increase in the proportion of CEO bonus assigned to sales and growth measures. Such measures likely better incentivize CEO attempts at exploiting firm growth opportunities. Where firms face higher growth potential, boards should align short-term CEO incentives with specific objectives and goals such as increasing revenues and growing customer numbers in the near term. Our findings suggest that boards perform effectively in this regard. We also note that firms with relatively higher MTB ratios place less reliance on both income-based measures (p < 0.05) and accounting-returns and cost-control measures (p < 0.05), likely because such measures are less informative of growth objectives.

Table 3 Descriptive statistics of variables
Full size table
Table 4 Weight of quantitative performance measures in CEO bonus plans (H1–H3)
Full size table

Models 2 and 3 examine the short-term objective influencing the use of income-based measures and accounting-returns and cost-control measures respectively:

$$INCOMEt~ = \alpha + \beta _{1} AMTB_{{t - 1}} + \beta _{{2~}} OPEX_{{t - 1}} + \beta _{{3~}} ADEQUACY_{{t - 1}} + \beta _{{4~}} NOISE\_EPS_{{t - 1\,to\,t - 5}} + {\text{ }}\beta _{{5~}} BSIZE_{{t - 1}} + \beta _{{6~}} INDIR_{{t - 1}} + \beta _{{7~}} DUAL_{{t - 1}} + \beta _{8} ~AGE_{t} + \beta _{{9~}} LNSALES_{{t - 1}} + {\text{ }}\gamma _{1} ~INDUSTRY~CONTROLS + \varepsilon$$
(2)
$$ACCRET\_COST_{t} ~ = \alpha + \beta _{1} AMTB_{{t - 1}} + \beta _{{2~}} OPEX_{{t - 1}} + \beta _{{3~}} ADEQUACY_{{t - 1}} + \beta _{{4~}} NOISE\_ROC_{{t - 1\,to\,t - 5}} + \beta _{{5~}} BSIZE_{{t - 1}} + \beta _{{6~}} INDIR_{{t - 1}} + \beta _{{7~}} DUAL_{{t - 1}} + \beta _{8} ~AGE_{t} + \beta _{{9~}} LNSALES_{{t - 1}} + \gamma _{1} \,INDUSTRY~CONTROLS + \varepsilon$$
(3)

Columns 2 and 3 in Table 4 present the estimation results. Contrary to our expectations under H2, firm operating expense ratio has no relationship with either income-based measures or accounting-returns and cost-control measures in CEO bonus plans. The cost management literature accentuates the necessity for efficient cost structures if businesses are to deliver long-term sustainable success (Stenzel & Stenzel, 2003; Anderson, 2006). Firms with relatively higher operating expenses likely face considerable competitive pressure to rapidly reduce their variable costs. Our evidence suggests that boards of high-cost firms do not align CEO incentives with short-term cost-cutting objectives, despite such actions appearing value-enhancing in the circumstances.

Finally, we test the effects of firm cash-flow adequacy ratio on the use of cash-based measures through the following regression:

$$CASH_{t} ~ = \alpha + \beta _{1} AMTB_{{t - 1}} + \beta _{{2~}} OPEX_{{t - 1}} + \beta _{{3~}} ADEQUACY_{{t - 1}} + \beta _{{4~}} NOISE\_FCF_{{t - 1\,to\,t - 5}} + \beta _{{5~}} BSIZE_{{t - 1}} + \beta _{{6{\text{ }}}} INDIR_{{t - 1}} + \beta _{{7{\text{ }}}} DUAL_{{t - 1}} + \beta _{8} {\text{ }}AGE_{t} + \beta _{{9{\text{ }}}} LNSALES_{{t - 1}} + \gamma _{1} {\text{ }}INDUSTRY{\text{ }}CONTROLS + \varepsilon {\text{ }} $$
(4)

The results in column 4, Table 4 are consistent with H3: ADEQUACY is negatively related (p < 0.05) to the proportion of CEO bonus assigned to cash-based measures. Cash flow adequacy assesses the availability of firm cash flow to meet debt repayments, fixed asset purchases, and dividend payments. Firms performing poorly in this area should concentrate CEO efforts on cash generation to ensure costly external finance is not required to fund ongoing commitments. By linking CEO bonus payments to cash-based performance measures, boards govern strongly on this aspect.

Our findings partially support the prior literature on the inverse relation between performance-measure choice and performance-measure noise. Sales and growth measures are negatively associated with their noise proxy. The remaining noise coefficients, however, show no significance in our models. The informativeness of performance measures for short-term firm objectives seems to outweigh any board concerns about the variability of such measures. Lastly, we do not find strong evidence of a bias toward any particular performance measure in firms with more powerful CEOs.

5 Robustness checks

The key concern with our identification strategy is the potential endogeneity problem that arises due to our reliance on observational data without random assignment. While this is a customary limitation of all archival studies that draw causal inferences from observational data (Angrist & Pischke, 2008; Gow et al., 2016), we concede that one year of cross-sectional data further impedes our ability to identify causal relationships. In this section, we endeavor to alleviate each of the three sources of endogeneity, namely omitted variablesFootnote 11, simultaneity, and measurement error (Roberts & Whited, 2013).

5.1 Instrumental variables

First, similar to Boubaker et al. (2018), we employ an instrumental variable approach to resolve both omitted variables bias and simultaneity bias (Larcker & Rusticus, 2010). In the event that our hypothesized explanatory variables are endogenous, the biased coefficient estimates will lead to inaccurate inferences. Therefore, we attempt to identify suitable instruments for AMTB, OPEX, and ADEQUACY. Given that CEO compensation has been found to persist over one year (Conyon, 1997; Conyon & He, 2012), the use of one-year lagged predictors as instruments will not solve the endogeneity problem. However, we observe that 214 firms in our sample alter the choice of performance metrics in CEO bonus plans over the four-year period 2010 to 2014. Therefore, performance-measure weights are not stable over a four-year period and, as such, four-year lagged regressors should be sufficiently exogenous from the model. We use MTB2010, OPEX2010, and ADEQUACY2010 as instrumental variables for each of our three potentially endogenous explanatory variables respectively.

Table 5 reports the second-stage regression results of two-stage instrumental variable modelsFootnote 12 for sales and growth, income-based, accounting-returns and cost control, and cash-based performance measures. Before drawing conclusions from these results, we run several diagnostic checks on the instrumental-variable models. First, we test the validity of the instrumental variables. Valid instruments are those that are (1) correlated with the endogenous regressor and (2) uncorrelated with the second-stage error term (Larcker & Rusticus, 2010). Table 6 reports reasonably high partial R2 values and significant F statistics (p < 0.01) from the first-stage regressions, indicating that the instruments have significant explanatory power. Thus, the first condition of instrument validity is satisfied. It is not possible to test the correlation between instruments and the second-stage error due to the unobservability of the structural equation error term (Larcker & Rusticus, 2010). However, it is possible to correlate the instruments with the estimated error term in the second-stage equation. We regress the second-stage residuals on all exogenous variables. The coefficients on the instruments (Table 6) are all close to zero, indicating that the instruments also meet the second criterion for validity. We provide further additional evidence to support this conclusion in the form of the R2 values from the models, which are all close to zero.

Table 5 Two-stage instrumental variable models: second-stage regression results
Full size table
Table 6 Instrumental variable diagnostic tests
Full size table

Next, we run two post-estimation commands on Stata. For the tests of overidentifying restrictions, a statistically significant test statistic would indicate that the instruments may not be valid. We find insignificant p-values (p > 0.05) on both the Sargan (1958) and Basmann (1960) test statistics (Table 6), which again supports the appropriateness of the instruments. Finally, we run a test of endogeneity to determine whether the potentially endogenous regressors in the models are in fact exogenous. If the Durbin (1954) and Wu–Hausman (Wu, 1974; Hausman, 1978) test statistics are significant, the variables must be treated as endogenous. Table 6 shows that only AMTB should be treated as endogenous and, thus, the instrumental variable estimates for SALES_GROW are preferable to the standard Tobit estimates. As can be seen in Table 5, AMTB remains significant (p < 0.01) in the second-stage of the two-stage instrumental variable regression. This lends support to our original conclusion. As the standard Tobit estimates are preferable for INCOME, ACCRET_COST, and CASH, our original interpretations under these models are unaffected.

5.2 Panel data regression

We acknowledge that our single-year cross-sectional dataset fails to control for time-fixed effects. To eliminate the omitted variable bias caused by excluding these unobserved variables, we hand-collect additional data on performance measures in CEO bonus plans for fiscal year 2021 from SEC-filed firm proxy statements. Due to the laborious and lengthy nature of this task, we limit our updated fiscal 2021 CEO performance-measure dataset to a subsample of 50 firms from the original fiscal 2014 performance-measure dataset. Table 7 compares the mean performance-measure weights in both fiscal 2021 and fiscal 2014 CEO bonus plans for the same 50 firms. To test the differences between the paired weights, we run paired samples t-tests (similar to Martinov-Bennie & Mladenovic, 2015 and Lone et al., 2016). The paired samples t-tests reveal that boards place significantly less reliance on income-based measures (p < 0.05) in fiscal 2021 CEO bonus plans, with significantly higher weights now attached to both cash-based (p < 0.05) and social-and-environmental (p < 0.01) measures.

Table 7 Mean weights of performance-measure variables (subsample of 50 firms)
Full size table

Our additional data from fiscal year 2021 allow us to run a panel data regression with 100 firm-year observations. Similar to Van Vu et al. (2018), we include a year dummy variable (FISCAL2014, coded one if the fiscal year is 2014 and zero otherwise) to control for time-specific effects. Table 8 reveals that all our original interpretations remain the same, after holding time constant. AMTB remains significantly associated (p < 0.05) with the use of sales and growth measures in CEO bonus plans. OPEX has no significant effect on the use of either income-based measures or accounting-returns and cost-control measures. The coefficient on ADEQUACY remains negative and significant (p < 0.05) in the cash model. This again confirms that, while boards govern CEOs effectively in the pursuit of short-term growth and cash-generation objectives, they do not successfully align CEO incentives with cost-efficiency objectives. We also test for possible interaction effects between FISCAL2014 and our predictors of interest to establish if the relationships have strengthened or weakened over time. We find a significant interaction effect in our sales and growth model, where the effect of AMTB on SALES_GROW is stronger in fiscal 2021 (p < 0.01) in comparison to fiscal 2014.

Table 8 Panel data regression: weight of performance measures in CEO bonus plans
Full size table

5.3 Alternative proxy measures

Finally, our study employs observed financial ratios to proxy for unobservable short-term firm objectives. Any conceptual discrepancy between the true construct of interest and the archival proxy will lead to measurement error (Roberts & Whited, 2013). Therefore, similar to Abernethy et al. (2019), we check the robustness of the main results to alternative proxy measures for the explanatory variables. We use the ratio of capital expenditures to total assets as an alternative measure of firm growth opportunities. We employ the ratio of selling, general, and administrative expenses to sales as an alternative proxy for firm cost-control objectives. As an alternative to ADEQUACY, the dividend yield ratio proxies for firms’ cash-payment commitments, with higher yielding stocks requiring greater cash management. The alternative estimations are qualitatively similar in that the predicted associations remain statistically significant for H1 and H3.Footnote 13

6 Discussion and conclusions

This paper examines the linkages between short-term firm objectives and CEO bonus plan performance measures. CEO bonus plans are the primary compensation vehicle through which boards align CEO incentives with firm-specific short-term operational goals. We hypothesize which short-term CEO actions may be more desirable depending on firm-specific financial circumstances. We investigate boards’ ability to incentivize these desirable actions through relevant and informative bonus measures. Our results provide mixed evidence. While boards effectively incentivize executives’ pursuit of growth and cash goals, we find a lack of alignment between CEO bonus payments and short-term cost management objectives. Our chosen interpretation of this absence of optimal contracting is one of ineffective board oversight. Boards should ensure that CEO bonus metrics strongly align with short-term objectives, and shareholders should challenge boards on how executive pay is designed to incentivize the pursuit of cost efficiencies where these are necessary.

In addition to our principal study of financial-performance CEO measures, we also document boards’ weighting of quantitative social-and-environmental measures in CEO bonus plans. To our knowledge, we are the first to report such rich data. While some authors have reported the use of non-financial or subjective performance indicators in incentive plans (Ittner et al., 1997; Murphy & Oyer, 2001; Angelis & Grinstein, 2015), our manual examination of firm proxy statements searches for objective and measurable sustainability oriented performance criteria. By recording the quantitative weight placed on social-and-environmental measures, our analysis is finer than studies that simply code the use of such measures in incentives through a binary variable (Flammer et al., 2019; Haque & Ntim, 2020; Cohen et al., 2023).

We capture social-and-environmental performance through measures such as safety incident rates, number of environmental spills, customer-call-response time, and number of employee-training programs. Boards evaluate these short-term performance measures through quantifiable 1-year targets. We observe that boards place minimal weighting (mean 3.6%, Table 3) on social-and-environmental measures in fiscal 2014 CEO bonus plans. As further analysis, we regress the weighting of social-and-environmental measures on our main explanatory and control variables. Table 9 reports that firms in both the energy (p < 0.05) and utilities (p < 0.01) industries place significantly more weight on such measures in CEO bonus plans. These findings demonstrate that boards of heavily regulated firms ensure compliance requirements are met by setting appropriate executive incentives. Larger firms (LNSALES, p < 0.05) also display a significantly higher reliance on social-and-environmental measures. This is consistent with the theory that larger firms attract more attention from external stakeholders and thus likely face greater reputational risks from poor performance in these areas (Waddock & Graves, 1997; Minor & Morgan, 2011). We also note a significantly positive coefficient on OPEX (p < 0.05), indicating a correlation between operating expenditures and ESG incentives. Boards of high-cost firms may incentivize certain social-and-environmental actions to drive costs down (for example through energy efficiency, waste and water reduction, etc.). In such scenarios, CEO compensation contracts are efficient. However, an alternative explanation is that firms that take more social-and-environmental actions incur more operating expenses (for example through employee training, health and safety measures, etc.). This indicates a trade-off between short-term financial and ESG objectives. Boards should clearly disclose the rationale for their performance-measure choices in such situations.

Table 9 Weight of social-and-environmental performance measures in CEO bonus plans
Full size table

For our fiscal 2021 subsample of firms, we report a significant (p < 0.01) increase of 4.6% (Table 7) in the mean weighting of quantitative social-and-environmental performance measures in CEO bonus plans. While the literature suggests that sustainability issues have become more pertinent for board oversight (Flammer et al., 2019; Haque & Ntim, 2020), we previously did not have empirical evidence on the degree of importance attached to ESG measures in CEO bonus plans. Our performance-measure-weight data allows us to clearly establish this importance and, thus, responds to Cohen et al.’s (2023) call for future research on the relative weights attached to different performance metrics.

Investors demand that boards align executive pay with corporate social-and-environmental metrics (Kay, 2022; Ritz, 2022). In addition, proposed European Union regulation requires firms to take emission reduction plans into account when setting directors’ variable remuneration (European Commission, 2022). Yet, the practice of tying executive incentives to social-and-environmental targets remains largely underexplored. For instance, we have little knowledge of the extent of challenge within sustainability oriented executive performance targets. Undemanding sustainability targets in executive compensation plans will simply lead to more inflated executive pay and minimal improvements in sustainability performance. We also need more insight on the suitability of selected social-and-environmental measures in executive incentives. These measures should be properly aligned with firms’ corporate purpose, business strategy, organizational culture, and risks and opportunities. Including arbitrary ESG metrics in executive incentives to simply appear green will harm long-term sustainable business success. It would also be useful to determine the incentive effects of sustainability measures in CEO pay plans. For instance, does the inclusion of such metrics actually motivate better efforts towards sustainable business practices? All these questions are promising avenues for further academic inquiry.

Our study is subject to three limitations. First, due to the time-consuming nature of hand-collecting the data from firm proxy statements, the main analysis is restricted to one fiscal year (fiscal year 2014). To enhance the validity of our conclusions, we expand our dataset with a subsample of fiscal year 2021 CEO bonus plans (n = 50). The additional tests support our main causal inferences, thus strengthening our identification strategy. Second, we use board-level indicators to measure CEO power (similar to Davila & Penalva, 2006; Morse et al., 2011; and Abernethy et al., 2015). However, CEO power over compensation committees may also influence the design of CEO compensation (Conyon & Peck, 1998; Conyon et al., 2011; Cho et al., 2019; Curtis et al., 2021). Our models do not control for this possibility due to the unavailability of this data through an accessible commercial database. Third, we concentrate on the S&P 500 firms because of their size and importance. However, the design of CEO bonus plans might be different for smaller firms and non-US firms. Much of the literature on CEO bonus incentives is based on US data. We urge future researchers to explore the attributes of non-US executive pay plans. Notwithstanding these limitations, we maintain that this study provides important new evidence on boards’ aptitude to link short-term firm objectives to CEO bonus performance measures.

7 Appendix

Appendix: Sample coding by performance-measure type

Sales and growth

Accounting returns and cost control

Cash-based

Revenue/sales

Return on invested capital

Free cash flow

Revenue/sales growth

Return on equity

Operating cash flow

Sales volume

Return on capital employed

Cash-flow indicator

Pipeline score new products

Return on assets

Cash-flow multiple

Market share

Return on average net assets

Manufacturing cash flow

Operating revenue

Return on capital

Adjusted cash flow

Organic growth

Delivered cost-productivity savings

Cash-flow return

Orders growth

Operating expense ratio

Cash-flow performance

New sales growth

Cost per available seat mile

Funds from operations

Billed business growth

Cost management

 

Production growth

Expenses % of revenue

 

Average daily sales

Unit costs

 

New patents submissions

Selling, general, and administrative costs

 

New bookings/new contracts awarded

Reduction in operating expenses

 
 

Expenses’ efficiency ratio

 

Income-based

Social and environmental

Other metric

Operating profit

Lost workday case rate

Corporate performance rating

Earnings before interest and taxes

Call response time/calls answered

Working capital to net sales ratio

Earnings before interest, taxes, depreciation, and amortization

Incident rate/serious event rate

Book value per share

Operating income

Customer satisfaction score

Fundraising levels 

Earnings per share

Duration/number of interruptions in the power supply

Net debt 

Diluted earnings per share

Safety work index

Credit rating

Net income

Surveyed employees satisfied with job

Integrated risk index

Net operating income

Net promoter score

Loan to value

Operating earnings

On-time performance

Operations scorecard/project scorecard

Economic profit

Environmental, health and safety scorecard

Mechanical availability

Pre-tax income/earnings

Supplier and workforce diversity %

Write-off rate

Operating margin

Customer service levels

Manufacturing readiness

Gross profit

Food safety and quality rate

Quality rate

 

Spills to the environment

Risk-based capital ratio