Relative Performance Goals and Management Earnings Guidance

RP managers may strategically influence the selection of peers for performance benchmarks and compensation benchmarks (see, e.g., Bizjak et al., 2008, 2011; Carter et al., 2009; Faulkender & Yang, 2010, 2013; Gong et al., 2011). In this study, we take peer selection as a given and focus instead on the potential for managers to influence the performance of peers.

The subversion hypothesis is also called the sabotage hypothesis in the literature. Gibbons and Murphy (1990) note that paying workers based on RP goals can distort the worker’s incentives whenever the worker can influence the output of the reference group by taking the following actions: sabotage, collusion, choice of reference group, or production externalities.

Our sample is limited to the years for which we have available data for management earnings guidance from IBES and performance goals data from Incentive Lab.

As described later in the additional-test section, for a small sub-sample of firms for which appropriate data are available, we also perform a follow-up test and find that RP managers, on average, do meet their performance targets.

Tournament theory suggests that managers are rewarded based on performance rankings within a reference group.

As noted earlier, Gibbons and Murphy (1990) note that paying workers based on RP goals can distort the worker’s incentives whenever the worker can influence the output of the reference group by taking the following actions: sabotage, collusion, choice of reference group, and production externalities. The mechanism by which distortion occurs in our setting is sabotage. We do not expect CEOs with RP goals to collude with other CEOs because collusion among CEOs would constitute an antitrust violation and because collusion would require CEOs to communicate with each other, but as Gibbons and Murphy (1990) point out, “CEOs tend to have limited interaction with CEOs in the rival firms.” Sabotage is plausible in our setting because with public disclosures and the management earnings forecasts that we examine, there is no need for CEOs to interact with each other. Once a forecast is made, it becomes part of public information that everyone, including the competitors, can access and use.

Consistent with the idea that firms’ decisions can be influenced by disclosures by their peers, Gong et al. (2019) show that RP firms intentionally delay earnings releases so they can observe peers’ earnings and conduct last minute earnings management.

We use the OLS model because non-linear models such as Logit or Probit combined with fixed effects can yield inconsistent point estimates (Angrist and Pischke, 2009; Wooldridge, 2002).

We use the average forecast pessimism rather than the pessimism for the most recent forecast for two related reasons. First, we focus on managers’ incentives to guide peer firms’ expectations and outputs (which are harder to be adjusted in a short time) rather than incentives to guide analyst forecasts (which are easier to be adjusted). Secondly, the sabotage hypothesis implies that managers may want to issue downward guidance early in the year for competitors to lower their expectations and efforts. Untabulated results show that using the first management earnings forecast issued after earnings announcement for the previous year does not change our inferences for earnings forecast properties.

The inference remains the same using a logistic regression, where the coefficient on RP is positive and significant (p value < 0.05).

To ensure the robustness of our results, we control for seven of the eight matching variables in the forecast properties regressions shown in Panel B of Table 4; we do not control for common risk because the literature does not consider it to be a variable related to forecast properties. However, untabulated results show that our regression results are qualitatively the same even if we control for common risk.

We thank the referee for suggesting this analysis.

Demerjian et al. (2012) use the managers’ efficiency in generating revenues as a measure of managerial talent. Conceptually, managers who can transform corporate resources to a higher level of revenue than their industry peers are considered to have higher talent.

The financially negative words are developed by Loughran and McDonald (2011) based on actual usage frequency. These words are most likely associated with negative financial implications. They include some words from the Harvard-IV-4 TagNeg (H4N) list but also some other words not on the H4N list such as “litigation,” “restated,” “misstatement” that have clear negative implications in a financial context.

Our result for Year T + 1 is qualitatively similar if we do not require the availability of stock returns for Year T + 2.

Our results do not contradict the main findings in Tice (2020). Even though Tice (2020) finds that RP firms with high common risk exposure perform better than non-RP firms, she also finds that other groups of RP firms (e.g., firms with low common risk exposure) do not perform significantly better than non-RP firms.