Fairness uncertainty and pay information exchange: Why and when employees disclose bonus pay to pay information websites
Summary
Having limited information regarding how pay is distributed in their organization, employees often find it difficult to assess the fairness of their pay. Uncertainty management theory (UMT) posits that fairness uncertainty is aversive and that individuals experiencing it search for information to reduce this uncertainty. Pay information exchange – the communication of one's pay-related information to others in return for information from that other – provides a mechanism to reduce pay information uncertainty. We focus on third-party mediated pay information exchange (such as via Glassdoor and PayScale), an increasingly prevalent form of exchange. Drawing on UMT, we investigate why and when individuals exchange their pay information with such agents. Using data from a field experiment we find that (a) the willingness of employees to disclose their pay to a pay information exchange platform is influenced by perceived utility of a-priori information offered by the exchange partner, but that this relationship depends on the salience of fairness uncertainty to the employee, and (b) employer pay communication restrictiveness only attenuates the impact of disclosure willingness on actual disclosure when individuals engage in deliberative thinking about such restrictiveness and its possible consequences. We discuss the implications for theory and practice.
1 INTRODUCTION
Pay fairness matters to employees (Bamberger, 2023; Colquitt, 2001), but accessing information to assess the fairness of pay can be difficult due to the widespread employer restrictions on pay communication (Hayes, 2017). Under a policy of restrictive pay communication, organizations provide only limited pay information and discourage employees from discussing their pay (Colella et al., 2007). Employers have argued that pay communication restrictions reflect employee preferences for privacy (Bierman & Gely, 2004). However, others note that such restrictions create a situation of pay information asymmetry (Bergh et al., 2019), with employers having full access to the pay information of all of their employees while employees are limited to knowledge of their own pay, even though many would like to know more than they currently do about their pay relative to that of others both within and across firms (Bamberger, 2023). Indeed, a multi-country survey of employees found that 69% wanted more information about others' pay and pay in their local labor market (Glassdoor, 2016). Interest in understanding the consequences of restrictions on pay communication is reflected in a surge of research interest (Brown et al., 2022), new regulatory initiatives such as the 2023 European Commission Pay Transparency Directive (Alterman et al., 2021; Aon, 2023), and the unilateral efforts of many enterprises, small (e.g., Buffer) and large (e.g., Whole Foods), to make pay more transparent (Bamberger, 2023).
Research suggests that in the absence of employer-provided pay information, individuals infer pay knowledge from whatever signals are available (Belogolovsky & Bamberger, 2014), though they tend to overestimate the pay of peers and underestimate both the pay of supervisors (Cullen & Perez-Truglia, 2018) and their own pay relative to the market (Smith, 2015). Overestimation of peer pay is associated with a decrease in employee hours worked (Cullen & Perez-Truglia, 2018) while restrictions on pay communication have a negative impact on individual job performance (Bamberger & Belogolovsky, 2010; Belogolovsky & Bamberger, 2014) and the retention of superior performers (Alterman et al., 2021; Belogolovsky & Bamberger, 2014).
Pay information asymmetry can have an adverse impact on the competitiveness of labor markets (Akerlof, 1970). Employers' relative ease of access to pay information boosts bargaining power, enhancing their ability to set wages below the economic value of the work performed (Rosenfeld & Denice, 2015). When employees lack information about others' pay, it is difficult to monitor and challenge pay discrepancies (Burn & Kettler, 2019). Employees have little choice but to accept employer claims about the distribution of pay (Roth, 2006). This can put workers at a significant disadvantage in negotiations, especially those with less bargaining power, including women and minority group members (Kim, 2015). Legislation in many countries (e.g., UK, US, Canada, Europe) has sought to facilitate greater employee access to others' pay information to reduce gender and racial pay gaps (Aon, 2023; Bennedsen et al., 2022; Kim, 2015).
Pay information exchange – the communication of one's pay-related information to others in return for information from that other – provides a mechanism to reduce pay information asymmetry. Direct pay information exchange is where one employee exchanges their pay information with a co-worker. It requires each employee to identify an appropriate co-worker with which to exchange pay information, convince them to engage in the exchange, and then assess the value and accuracy of the co-worker-supplied pay information, while potentially risking sanction from their employer for breaching organizational pay communication policies or norms. Although such exchange has been facilitated by laws enacted in some countries limiting the ability of employers to restrict such exchange (Aon, 2023; Bamberger, 2023), employees still report being uncomfortable asking a co-worker about their pay (Cullen & Perez-Truglia, 2018), often due to concerns about negative impacts on their relationships with colleagues and their self-esteem or to a perception that directness about pay is ‘taboo’ (Schnaufer et al., 2022). Further, employees who do engage in information exchange might disclose low-quality pay information (e.g., provide an underestimate or vague statement of their pay) (Smit & Montag-Smit, 2019). The direct approach requires employees to manage the issue of pay information asymmetry on their own and often generates only anecdotal evidence lacking the quantity and quality of data needed to accurately assess the fairness of their pay.
An increasingly popular alternative to direct pay information exchange is third-party-mediated pay information exchange whereby individuals exchange their pay information with others via a mediated third party, typically a website such as PayScale or Glassdoor. 1 By crowdsourcing pay information, website users have access to a greater quantity of others' pay information – more than would be possible under direct pay information exchange. The number of user entries on Glassdoor increased from around 290,000 in 2010 to approximately 1,100,000 in 2017 (Karabarbounis & Pinto, 2018) and Glassdoor reports over 60 million unique visits to their website each month. Third-party-mediated pay information exchange is popular with employees in organizations with restrictions on pay communications (Chamberlain, 2015), with users able to apply a variety of filters to identify relevant pay information, providing an almost immediate benefit in the form of others' pay information (Hann et al., 2007).
While third-party-mediated pay information exchange offers advantages over direct pay information exchange, it is not certain that all employees will participate. Some may be reluctant to engage with a pay information platform when the exchange involves disclosure of information that may be seen as personal and private (Pavlou et al., 2007). Further, as the number of users of third-party pay information exchange platforms increases, so have expectations about the quality of data available for exchange. Website users are asked to provide more fine-grained information about their current employment and pay circumstances, which carries the risk of identification by their employer.
Concerns have been expressed about the quality of data on pay websites (Bryant et al., 2020). However, the spatial distance from other website users can foster a sense of anonymity, facilitating the provision of accurate own-pay information. Individuals are less likely to have concerns about potential negative impacts of disclosure on workplace relationships (Smit & Montag-Smit, 2019) and expectations of reciprocity, as can happen in direct pay information exchange (Cullen & Perez-Truglia, 2018). Further, an evaluation by Karabarbounis and Pinto (2018) found that average salaries reported by Glassdoor are equivalent to those reported by the United States Bureau of Labor Statistics. Our paper investigates the individual-level dispositional and situational drivers of third-party pay information exchange.
Based on the assumption that pay information exchange via third-party agents is typically driven by an underlying motivation to manage uncertainties regarding pay fairness, we ground our theorizing on Uncertainty Management Theory (UMT; Lind & van den Bos, 2002). We propose a model of third-party-mediated employee pay disclosure governed by three principles. First, the perceived utility of information shared a-priori by the exchange partner (e.g., Glassdoor) influences individuals' willingness to disclose their own pay information, with the perceived utility of this shared information itself influenced by the magnitude and relevance of the information shared. Second, willingness to disclose – a situation-specific employee pay disclosure attitude – serves as a precursor to actual behavioral disclosure. Finally, these effects are moderated by individual differences and situational factors. In the case of disclosure willingness, we focus on the saliency of fairness uncertainty and, in the case of disclosure behavior, on the restrictiveness of employer pay communication policies and the degree to which employees engage in deliberate thought about these policies. We test these principles and the model they suggest (see Figure 1) in a field experiment using a representative sample of 642 Australian workers recruited to participate in a study involving engagement with a realistic third-party website designed expressly for this study.
Our findings offer several important and novel theoretical contributions. First, while research suggests that information exchange can be a means for reducing fairness uncertainty, little is known about the mechanisms and contingencies underlying and governing such exchange. Accordingly, our model explains why and when employees engage in third-party pay information exchange. The model contributes to the pay literature by identifying some of the more critical parameters impacting an increasingly important employee reward-related behavior and explains how these factors may interact in shaping employee agency in pay communication and knowledge. Second, our findings highlight the role of heuristic processes in explaining why, despite privacy concerns (Trotter et al., 2017), employees exchange pay information, even when the value of information that they may get in return is limited and may involve non-compliance with employer pay communication policy. Finally, we provide new insights into the impact of employer restrictions on pay communication. The intention of employer pay communication restrictions is to deter workers from disclosing their pay information to others (Colella et al., 2007). However, contrary to employer expectations, our findings demonstrate that employees may often be more likely to disclose their pay when such restrictions are in place.
2 THEORY DEVELOPMENT
Uncertainty Management Theory (Lind & van den Bos, 2002) offers a comprehensive, social-cognitive framework for understanding when and how individuals respond to fairness uncertainty (Alterman et al., 2021; SimanTov-Nachlieli & Bamberger, 2021). UMT posits that pay represents a central domain of fairness uncertainty for many employees and that pay-related fairness uncertainty is both aversive (Fiske & Taylor, 1991) and salient (van den Bos et al., 1998). Employees are thus driven to search for information to reduce this uncertainty from within (Lind, 2001) and/or outside of their organization (e.g., Bamberger, 2023; Lind et al., 1993). As suggested above, employees are increasingly engaging with third-party pay websites, exchanging their own pay information for the pay data these platforms make accessible. UMT offers an overarching framework for understanding the factors influencing such pay information exchange, highlighting both the calculative factors employees are likely to consider when invited to disclose their own pay information, as well as the heuristics likely to moderate such effects.
2.1 Linking the actions of third-party websites to Individuals' disclosure attitudes
Consistent with the literature on information exchange (Poortvliet et al., 2007) and resource seeking (Lim et al., 2020) UMT suggests that individuals are likely to be cautious when engaging in pay-related information exchange. UMT suggests that a “fundamental social dilemma” (Lind, 1995) faced by individuals confronting fairness uncertainty is how to balance the instrumental benefits of information exchange as a means of uncertainty reduction with the risks associated with it. Accordingly, UMT suggests that before engaging in information exchange as a means by which to address pay fairness uncertainty, individuals are likely to take two main considerations into account.
The first consideration refers to the potential risks associated with disclosing information to the exchange partner. Privacy concerns are important in website disclosure decisions (Son & Sung, 2008): employees concerned about website privacy are less likely to engage in exchange (Cichy et al., 2021). The risks to the employee include commercial misuse of their data, such as selling it to another party without their consent, as well as receiving unsolicited advertising and emails (Zhang et al., 2021). There is also a risk of being identified as having provided their pay information to others and being sanctioned by the employer for breaking implicit norms or explicit rules of pay secrecy (Colella et al., 2007).
The second consideration refers to the likelihood that the exchange will elicit the intended benefit, namely a reduction in pay fairness uncertainty. Building on fairness heuristic theory (Lind, 2001), UMT suggests that given the risks involved in such exchange, individuals search for information to the degree to which they can be confident that the exchange will yield the intended benefit. In situations in which “definitive trust” is lacking, individuals infer the credibility, relevance, and hence overall utility of the resources offered to them as part of the exchange based on implicit signals, ascribing a greater potential benefit to exchanges in which their partner signals goodwill and trustworthiness (van den Bos & Lind, 2002, p.10).
Typically, pay platforms address the first consideration by warning users about the risks of violating organizational policies while stressing their service's data security and promising confidentiality (Cichy et al., 2021; Hann et al., 2007). However, there is considerable variance among platforms with respect to the second consideration. Some websites take a minimalist approach to signaling the potential benefit of exchange by emphasizing the social value of such behavior (e.g., Glassdoor: “Everything you add helps others find a job and company they'll love”, Sundberg, n.d.), or offering a token of the kind of data that could be available to them by engaging in exchange (e.g., limited, employment-related information such as mean rates of job satisfaction or voluntary turnover in a given occupation and/or organization). In contrast, other platforms signal the potential benefit of exchange by sharing – a-priori – detailed and personalized pay information (e.g., minimum, maximum, and/or average pay for employees in their occupation with their level of work experience). By providing personalized pay data to potential exchange partners on an a-priori basis, websites likely aim to demonstrate their ability to deliver informative and highly tailored pay information (Pavlou et al., 2007), and hence implicitly reduce any apprehension individuals may have in disclosing their own pay information to the website (Karwatzki et al., 2017). Per the fair process effect of UMT (Folger et al., 1979) the a-priori provision of detailed personalized pay information even before the employee discloses their own pay information, may serve as a key fairness heuristic, with individuals potentially inferring greater trustworthiness of the pay information website and greater credibility and usefulness with respect to the information the website is able to share with them. Accordingly, we posit that:
Hypothesis 1a.The perceived utility of shared information will be higher for those assigned to the pay information provision condition relative to those assigned to a token employment information condition.
UMT assumes that individuals seek to maximize their perceived utility when selecting between alternative courses of action (Lind & van den Bos, 2002). The utility of the information shared a-priori by the website as perceived by a potential exchange partner is, therefore, a critical mechanism driving the latter's willingness to exchange their pay information with that website as a means by which to mitigate their pay fairness uncertainty. Willingness to disclose is an attitude that reflects how an individual thinks and feels about pay disclosure. Consistent with UMT, and as has been demonstrated across different types of websites (e.g., e-commerce, crowdsourcing), when the potential website user perceives a potential benefit associated with disclosure, they are more likely to express a willingness to disclose personal information to that site (Cheng et al., 2021; Cichy et al., 2021; Karwatzki et al., 2017). This suggests that:
Hypothesis 1b.The perceived utility of shared information is positively associated with individuals' willingness to disclose their own pay information to the soliciting, third-party website.
Further, since we posited that the a-priori provision of pay information by the soliciting agent (i.e., third-party website) is positively associated with individuals' utility perceptions (Hypotheses 1a), we also posit:
Hypothesis 1c.The perceived utility of shared information mediates the impact of a-priori provision of pay information (relative to token information provision) by the soliciting agent (i.e., third-party website) on individuals' willingness to disclose their own pay information.
2.2 The moderating effect of fairness uncertainty salience
While UMT posits that fairness uncertainty is an aversive state, it also acknowledges that the importance of fairness to individuals – i.e., fairness uncertainty salience (van den Bos, 2001, p. 932) – varies. Variance in fairness uncertainty salience can stem from trait-based differences (van den Bos, 2001). For instance, people vary in their interest in comparative information and the value they attach to it (Gibbons & Buunk, 1999). It can also stem from situational factors. For instance, reward fairness may be something employees are more cognizant of during a performance review/bonus allocation period.
Variance in fairness uncertainty salience can influence how individuals interpret and respond to unfairness. UMT research indicates that employees with high pay fairness uncertainty salience are more likely to experience a variety of negative emotional states, such as concern and anxiety about being under-rewarded or exploited, as well as strong negative beliefs about the consequences of uncertainty. As these are psychologically aversive conditions, those with high pay fairness uncertainty salience are likely to be more receptive to actions aimed at mitigating the situation generating the aversive emotional states (van den Bos, 2001). Indeed, among those for whom fairness uncertainty is more salient, the response to such situations is likely to be more automatic, leading them to take steps to mitigate this aversive situation with little regard for the utility of one form of action over another. A more automatic response among those for whom fairness uncertainty is more salient makes sense because, for such individuals, exposure to fairness uncertainty is likely to elicit stronger emotions (van den Bos, 2001), thus limiting deliberative thought about the potential costs and benefits of adopting one mode of response over another (Kahneman, 2011). Accordingly, when considering pay information exchange, those for whom pay unfairness uncertainty salience is high may rely on more automatic (i.e., less reflective) assessments of the utility associated with the shared pay information such that their willingness to disclose their own pay as a means by which to mitigate their own pay-related unfairness uncertainties is likely to be less sensitive to attributions of utility (Thau et al., 2009). We thus propose the following:
Hypothesis 2a.The association between the perceived utility of shared information and one's willingness to disclose is moderated by fairness uncertainty salience: the relationship is amplified under conditions of low pay fairness uncertainty salience and attenuated under conditions of high pay fairness uncertainty salience.
Furthermore, to the extent that we posited that the perceived utility of shared information mediates the effect of a-priori information provision by a third-party website on individuals' willingness to disclose their own pay information (Hypothesis 1c), Hypothesis 2b suggests that fairness uncertainty salience also moderates this mediated relationship. Accordingly, we posit:
Hypothesis 2b.The indirect effect of a-priori provision of pay information (relative to token information provision) by a soliciting agent (i.e., third-party website) on individuals' willingness to disclose their own pay information via the perceived utility of shared information is moderated by pay fairness uncertainty salience, such that this indirect effect is amplified under conditions of low pay fairness uncertainty salience and attenuated under conditions of high pay fairness uncertainty salience.
2.3 Contextual moderators of the willingness to disclose – pay disclosure relationship
Attitudes can be predictive of behaviors, though research on the relationship between willingness to disclose personal information (an attitude) and actual disclosure (a behavior) to a website are mixed (e.g., Cheng et al., 2021) suggesting that the relationship between willingness and actual disclosure to a website may be contingent on other relevant factors (Ajzen & Fishbein, 1977). Indeed, to the extent that variance in certain contextual conditions may weaken attitude-behavior correspondence, it is not surprising that Smit and Montag-Smit (2019) found the correspondence between trait pay information search preferences and past pay information-seeking behavior to be contingent on employer pay communication policy.
If, as suggested by Smit and Montag-Smit (2019), employer pay communication policy serves as a contextual factor moderating the link between employee pay information search preferences and actual seeking behavior, it is conceivable that employer pay communication policy also moderates the link between an individual's willingness to exchange their own pay information and the likelihood that they will disclose their pay. What is uncertain, however, is the form that this conditioning effect will take. We posit that unless employees engage in deliberative thinking about the potential implications of pay information sharing in more (versus less) restrictive pay communication contexts, the effect of pay communication context is likely to be automatic, driven by heuristic processes amplifying the association between the individual's willingness to disclose their pay and the probability of their actual disclosure.
2.3.1 Amplification via substitution and the fair process effect
UMT proposes that two heuristics, namely substitution (Kahneman, 2011) and the fair process effect (Folger et al., 1979), combine to explain individuals' responses to fairness uncertainty. The heuristic of substitution suggests that individuals apply “shortcut reasoning”, generalizing from one judgment to another (Lind & van den Bos, 2002, p. 199). Applying this principle, UMT proposes that lacking information about the fairness of effort-reward outcomes in their current employment situation, individuals may substitute information regarding the fairness of their employer's reward processes and practices (van den Bos & Lind, 2002). In a situation of uncertainty regarding the fairness of effort-reward outcomes (e.g., the degree to which their contributions were compensated relative to the compensation that others received for their contributions), individuals may draw inferences from pay communication policies as they are proximate to the fairness judgment in question (Belogolovsky & Bamberger, 2014). Smit and Montag-Smit (2019) and Montag-Smit and Smit (2021) note that the restrictiveness of the employer's pay communication policy is likely to serve as a highly salient source from which to draw inferences regarding outcome unfairness, with more restrictive employer policies likely associated with more negative inferences about the fairness of reward outcomes. When making these inferences, individuals may pay more attention to, and overweight, negative information (Baumeister et al., 2001). Individuals who pay more attention to negative information are likely to be suspicious of others' motives and intentions (Belogolovsky & Bamberger, 2014), potentially resulting in attributions of unfairness.
Building on the fair process effect, UMT posits that individuals who make such negative fairness judgments are likely to engage in resistance behavior (van den Bos, 2005), manifesting as higher levels of overt and covert disobedience (Huo et al., 1996). That is, the suspicion regarding the fairness of pay generated by more restrictive pay communication practices is likely to prompt implicit protest and resistance (Vermunt et al., 1996) as well as a sense of psychological license (Miller & Effron, 2010) for breaking restrictive pay communication rules. With such a response weakening employees' compliance with such restrictions (Lind et al., 1993), heightened pay communication restrictions are likely to have an effect precisely opposite to the one intended, namely amplifying the correspondence between a willingness to disclose pay information and actual pay disclosure. Accordingly, we propose that:
Hypothesis 3a.The restrictiveness of an employer's pay communication policy moderates the effect of willingness to disclose pay information on the probability of actual disclosure such that this relationship is amplified (attenuated) when pay communication is more (less) restricted.
2.3.2 Attenuation through deliberative thought
The theorizing above assumes that employees devote little conscious thought to employer pay communication policies. However, research on deliberative thought in organizations suggests that these same policies may have different implications on attitude-behavior correspondence regarding pay information exchange when employees reflect on the potential implications – positive and negative – of actual pay disclosure. Deliberative thinking requires employees to challenge the natural tendency to adopt viscerally attractive (hedonic) options (Derfler-Rozin et al., 2016) such as reducing fairness uncertainty and/or reasserting control by exchanging pay information. It does so by having them exert mental effort to contrast viscerally attractive considerations in light of potentially overlooked negative ones. Studies have found that reducing time pressure (and thus facilitating deliberative thought) is associated with diminished interpersonal deception (Shalvi et al., 2012), and that when individuals are asked to consider the potential consequences of alternative courses of action in a deception game, they engage in more ethical interpersonal behavior (Gunia et al., 2012). Likewise, Derfler-Rozin et al. (2016) found that task-demanded deliberative thought was associated with reduced organizational rule-breaking. This suggests that despite the tendency of more restrictive pay communication contexts to implicitly amplify the positive effect of willingness to disclose pay information on actual pay disclosure, deliberative thought about the potential consequences of such action is likely to weaken such an amplification effect. Accordingly, we propose the following three-way interaction:
Hypothesis 3b.Conscious deliberation on the likely costs and benefits of disclosure moderates the amplification effect of restrictive pay communication policy on the relationship between willingness to disclose and actual pay disclosure. This amplification effect is strongest when individuals do not engage in conscious deliberation of the potential benefits and costs of pay disclosure, and weakest when individuals do engage in conscious deliberation of the potential benefits and costs of pay disclosure.
3 METHODS
3.1 Design and procedure
We tested our hypotheses in an experimental field study design with two waves of data collected approximately four weeks apart. The study design was approved by the ethics panel of the fourth author's university (approval number 2018/248). Participants were randomly assigned to either a pay information condition or one of two token information conditions. Participants also received a subsequent random assignment to either a neutral or deliberative thought condition. For the first manipulation, we collected information at Time 1 about the employment circumstances of all participants, including experience, job title, employer location, and organization name. At Time 2, participants assigned to the pay information condition were given occupation-specific pay information, including the minimum, maximum, and average pay for employees in their occupation and region, and with their level of experience. 2 Participants in the token information condition were given either no detailed job information (directed to the first questions on the survey) or information about average job satisfaction for employees in their occupations. In neither of the token information conditions were participants provided with pay information. These two variants of token information, however, did not differ from each other with respect to the key outcomes of willingness to disclose (t-value = −1.00, p = .31) nor perceived utility (t-value = −1.73, p = .08) and were subsequently combined into a single token information condition. Furthermore, a manipulation check confirmed that there are statistically significant differences between the token and pay information conditions. 3 Participants in the token information condition (N = 421) reported a significantly lower understanding of their relative pay (M = 3.86; SD = 1.51) than those assigned to the pay information condition (N = 221; M = 4.25; SD = 1.28; t-value = −3.32, p < .01). Figure 2 provides an overview of the study design.
For the second manipulation, we randomly assigned participants to either a deliberative thinking or neutral condition. Following Derfler-Rozin et al. (2016), we prompted deliberative thinking by getting participants to reflect on an equal number of positive and negative consequences of pay disclosure: after providing information about their willingness to disclose, those in the intervention condition were prompted to consider the possible costs and benefits of pay disclosure by completing a 10-item (5 cost and 5 benefit) instrument that asked about agreement with statements about the possible consequences of pay disclosure such as, “disclosure could make me feel embarrassed,” “disclosure could help me improve my own pay level.” In contrast, those assigned to the neutral condition were prompted to go to the next section of the survey and were asked to disclose their pay (Pay Disclosure).
3.2 Sample
We engaged the Australian subsidiary of an international online panel data collection business (Pure Profile) to collect our data. To ensure we included participants with a variety of pay communication experiences, individuals had to be employed full-time, earn at least $AU45,000 per annum, work in the private or not-for-profit sectors, and be able to provide the name of their current employer. Employees earning less than $AU45,000 are likely to be paid based on a publicly known casual hourly rate (Goonrey & Banchetti, 2021). At the time of data collection, the average full-time equivalent wage and salary earnings in Australia was $AU82,188 (OECD, 2018).
Participants were required to confirm that their organization's base pay details were not publicly available (e.g., on the internet). The Time 1 (T1) sample (N = 1,715) closely resembled the gender and age composition of the Australian labor force (Australian Bureau of Statistics, 2018): 47% of our sample was female and 53% was male, while in the Australian workforce, the equivalent figures are 48% and 52%, respectively. Our sample was also broadly consistent with the age profile of the Australian labor force (Australian Bureau of Statistics, 2018). In the total Australian population, 7.6% of employees are 20 years and under (our sample 1.7% lower, most likely due to the minimum income and full-time working survey filter); 32.6% are 21–34 years (our sample 38.3%); 22.4% are 35–44 years (our sample 24.5%); 21% are 45–54 years (our sample 20.1%) and 16.3% are 55 and older (our sample15.4%).
At Time 2 (T2), 986 of the T1 participants completed a survey, though 46 participants failed the attention test and were dropped from the study. Following Goodman and Blum (1996), we tested for response bias by comparing the characteristics of people who responded to both T1 and T2 (coded 1) with those who only completed T1, using three characteristics from the T1 survey (frequency of pay-seeking, organizational tenure, and restrictive employer pay communication). Two of the independent variables (frequency of pay seeking and organizational tenure) did not differ between the two groups; however, employer pay communication restrictiveness did. Those dropping out of the study at T2 (when we requested pay disclosure) reported a significantly higher level of employer pay communication restrictiveness (M = 3.44, SD = .04) than those remaining (M = 3.33, SD = .04; t-value = 2.29, p < .05), suggesting that participants believed that the data requested at T2 might be publicly disclosed (a greater concern for those subject to more restrictive pay communication policies).
Because some base pay information, even if not publicly accessible, is often made available by employers to their employees in Australia, we focused on the disclosure of bonus pay. Individual and team bonuses comprise a significant part of total earnings, typically 6%–10% of the total earnings of non-executive employees (Gocardless, 2021) with bonus payments linked to individual performance being the most prevalent form (Shields, 2020). These bonus payments are intended to recognize the differential contributions of employees (Gerhart et al., 2009). Consequently, as in other countries (Worldatwork-Mercer, 2020), this form of pay is deemed highly confidential in most Australian organizations (Goonrey & Banchetti, 2021).
Of the 940 valid completions, 642 participants reported being eligible for bonus pay. A second response bias analysis comparing these 642 participants with the 298 dropped due to bonus pay ineligibility found no evidence that organizational tenure differed between the two groups. However, again dropouts differed from those remaining with respect to reported employer pay communication restrictiveness with those bonus-eligible (i.e., not dropped from the study) reporting a significantly higher mean level of employer restrictiveness (M = 3.40, SD = 1.05) than those ineligible (M = 3.19, SD = 1.15; t-value = 2.42, p < .05). Frequency of pay seeking also differed, with those eligible for a bonus reporting a higher mean frequency of pay seeking (M = .86, SD = 0.93) than those ineligible (M = 1.53, SD = 0.80; t-value = 4.94, p < .01).
3.3 Measures
3.3.1 Utility of shared information
At T2, using four items adapted from Lee et al. (2002), participants were asked about the utility of the information shared by the pay information website. Respondents were asked to, “please give us your assessment of the quality of this information by indicating your level of agreement with the following questions (7 = strongly agree; 1 = strongly disagree).” A representative item is “The information was useful to me” (alpha = .71).
3.3.2 Willingness to disclose
At T2 we assessed each participants' willingness to disclose their pay bonus using an instrument adopted from other measures of individuals' willingness to disclosure information on-line (Sun et al., 2015): “On a scale of one to five, how willing are you to disclose the following aspects of your pay in your current job?” Participants were then asked to indicate their willingness to disclose various aspects of their pay including “the exact amount of any annual bonus (e.g., total of individual and team bonuses)”. The response options were 1 = not at all willing to 5 = very willing. The sixth option “not relevant as my employer does not offer this form of pay” was used to exclude respondents from the analyses.
3.3.3 Moderators
Pay fairness uncertainty was operationalized by the frequency of pay seeking. Uncertainty salience has largely been examined based on priming (e.g., asking participants to think about things that make them feel uncertain) with manipulation checks relying on observer coding of participants' behavioral responses to the manipulation (van den Bos, 2005). Such behavioral assessment makes sense since self-reports, even as a manipulation check, could potentially have demand effects and even operate as a confounding prime. Accordingly, we operationalize fairness uncertainty salience based on a behavioral proxy, focusing on the intensity of recent pay-seeking behavior. Empirical evidence supports such a behavioral operationalization, with Smit and Montag-Smit (2019) finding pay information-seeking behavior to be significantly correlated with two salience-related constructs, namely general pay-seeking preference (r = .47) and intolerance to uncertainty (r = .19). The data was collected in T1 using four items developed for this study. The question stem read “How frequently in the last month did you engage in the following behaviors?” with a representative item being “Spoke with a recruiter about current pay rates for my occupation”. Participants could respond 1 = never; 2 = once; 3 = about every two weeks; 4 = weekly; 5 = daily (alpha = .89).
Employer pay communication restrictiveness was based on two items (collected at T1) from Tremblay and Chȇnevert (2008), one of which states: “My organization tries to discourage non-management employees from disclosing their pay to co-workers” (alpha = .81). Deliberative thinking was manipulated with participants randomly assigned to the neutral condition receiving a 0, and those assigned to the deliberative thinking condition coded 1.
3.3.4 Pay disclosure
At T2 we asked participants to disclose their actual pay bonus in the previous tax year (pay disclosure). Participants were asked to “select from the drop-down menu to specify the exact amount of any ‘annual bonus’ (i.e., total of individual and team bonuses) received in the most recent tax year in thousands/hundreds of dollars”. Those who provided their exact bonus pay information were coded 1 and those who selected “prefer not to disclose” were coded 0.
3.3.5 Control variables
We controlled for organizational tenure to account for participants' varying levels of familiarity with organizational pay policies using a single-item measure “How many years in total have you worked with your current employer?” Female (=1) was included as there is evidence that women are socialized to different role expectations around pay (Babcock et al., 2003). We controlled for dispositional differences in employee pay-sharing preferences using a six-item measure developed by Smit and Montag-Smit (2019) (alpha = .90) because the desire for pay privacy may override any desire to know others' pay information (Colella et al., 2007). Additionally, we controlled for employee age (measured in years) as attitudes regarding pay communication vary across cohorts (Trotter et al., 2017), as well as union membership (member = 1) as members are more likely to be concerned with issues of fairness at work (Freeman et al., 1984). Control variables were assessed at T1.
All statistical models were run with and without controls. With one exception detailed below, the significance level of the findings was robust to the exclusion of controls. Our tables and primary results include controls even though tests involving randomization to condition do not require controls (Bodner & Bliese, 2018). We opted to use controls because other components of the conceptual model (e.g., the link between the utility of shared information and reported willingness to disclose) are not randomized and likely to provide more accurate estimates when the variance from known confounds is removed.
3.4 Analytic approach
The conceptual model presented in Figure 1 presents an implied multi-faceted path model with direct effects, indirect effects, moderators, and moderated mediation. The hypotheses, however, center on the significance of the specific components of the model. These specific components of a path model can be estimated from the underlying regression equations (Cohen & Cohen, 1983). In presenting the results, we chose to test each component separately using the appropriate underlying model (OLS regression, logistic regression, tests for mediation, and moderated mediation). By testing each component separately, we can clearly articulate the underlying statistical model used to test each hypothesis and provide specific information about robustness to controls. 4
4 RESULTS
Means, standard deviations, correlations, and reliabilities for study variables are provided in Table 1. A confirmatory factor analysis was conducted to demonstrate discriminant validity among the study's four latent variables (frequency of pay seeking, employer pay communication restrictiveness, pay sharing preferences, and utility of shared information). The four-factor model fitted the data well ((χ2[96, N = 940] 410.89, p < .05, CFI = .96, TLI = .95, RMSEA = .06, SRMR = .04). The study then compared the 4-factor model to a one factor model ((χ2[102, N = 940] 4559.34, p < .05; CFI = .47, TLI = .38, RMSEA = .22, SRMR = .17), a two-factor model (combining frequency of pay seeking +utility of shared information, and pay sharing preference + employer pay communication restrictiveness: (χ2[101, N = 940] 2698.18, p < .05; CFI = .69, TLI = .63, RMSEA = .17, SRMR = .13) and a three factor model (pay sharing preferences+ frequency of pay seeking, with utility of shared information, employer pay communication restrictiveness: (χ2[99, N = 940] 2254.32, p < .05; CFI = .74, TLI = .69, RMSEA = .15, SRMR = .14). The four-factor model fit significantly better than the one factor model (Δχ2 = 4148.45 Δdf = 6, p < .01), the two-factor model (Δχ2 = 2287.29 Δdf = 5, p < .01) and the three-factor model (Δχ2 = 1843.43, Δdf = 3, p < .01). Further, all the factor loadings in the four- factor model were greater than .60 and all the cross loadings were less than .32 (Tabachnick & Fidell, 2001) except for the utility of shared information. The factor loadings for the reverse coded variable (“The information did not tell me anything I did not already know”) was 0.28. As the pattern of hypothesized results is unaffected by whether we used the three or four-item measure of utility of shared information, we decided to retain the four items from Lee et al. (2002).
| Mean | s.d. | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. Organizational tenure | 6.40 | 5.97 | - | |||||||||||
| 2. Female | 0.47 | 0.50 | −0.02 | - | ||||||||||
| 3. Sharing preference | 3.74 | 1.38 | −0.09 | −0.08 | .90 | |||||||||
| 4. Age | 38.27 | 12.26 | 0.37 | 0.07 | −0.25 | - | ||||||||
| 5. Union member | 0.08 | 0.28 | 0.12 | −0.04 | 0.04 | 0.04 | - | |||||||
| 6. Token v pay information | 0.34 | 0.48 | 0.01 | −0.01 | 0.08 | 0.06 | −0.03 | - | ||||||
| 7. Utility of shared information | 4.43 | 1.14 | −0.06 | 0.00 | 0.19 | −0.12 | 0.06 | 0.09 | .71 | |||||
| 8. Frequency of pay seeking | 1.86 | 0.92 | −0.11 | −0.27 | 0.27 | −0.45 | 0.07 | 0.05 | 0.12 | .89 | ||||
| 9. Willingness to disclose | 3.76 | 1.38 | −0.10 | −0.01 | 0.31 | −0.22 | 0.00 | 0.00 | 0.18 | 0.26 | - | |||
| 10. Deliberative thinking | 0.48 | 0.50 | 0.00 | −0.03 | −0.01 | 0.05 | −0.01 | 0.02 | −0.05 | −0.04 | −0.05 | - | ||
| 11. Employer pay comm restrictiveness | 3.40 | 1.05 | 0.04 | 0.08 | −0.14 | 0.05 | 0.02 | −0.03 | −0.03 | 0.03 | 0.07 | 0.00 | .81 | |
| 12. Pay disclosure | 0.78 | 0.42 | −0.03 | −0.10 | 0.16 | −0.18 | 0.05 | 0.00 | 0.14 | 0.23 | 0.47 | −0.02 | 0.13 | - |
- Note: n = 642. Reliabilities (in diagonals) based on individual respondents; correlations larger than |.08| significant at .05, two-tailed.
4.1 Utility of shared information and willingness to disclose
Hypothesis 1a proposed that participants assigned to the pay information condition would perceive the reported information to have higher utility than those in the token information condition. Model 1 in Table 2 provides results supporting Hypothesis 1a. As expected, relative to the token information condition, the utility of the shared information was significantly higher in the pay information condition (estimate = 0.21, p = 0.02).
| Model 1: Utility of shared information | Model 2: Willingness to disclose | Model 3: Willingness to disclose | Model 4: Willingness to disclose | |||||
|---|---|---|---|---|---|---|---|---|
| Coef. | Std. err. | Coef. | Std. err. | Coef. | Std. err. | Coef. | Std. err. | |
| (intercept) | 4.14*** | 0.24 | 2.78*** | 0.33 | 2.12*** | 0.37 | 1.05* | 0.59 |
| Organizational tenure | −0.00 | 0.01 | −0.01 | 0.01 | −0.01 | 0.01 | −0.01 | 0.01 |
| Female | 0.05 | 0.09 | 0.04 | 0.10 | 0.15 | 0.11 | 0.15 | 0.10 |
| Age | −0.01* | 0.00 | −0.02** | 0.00 | −0.01 | 0.01 | −0.01 | 0.01 |
| Union member | 0.28 | 0.16 | −0.03 | 0.19 | −0.08 | 0.19 | −0.04 | 0.19 |
| Pay sharing preferences | 0.13*** | 0.03 | 0.26*** | 0.04 | 0.23*** | 0.04 | 0.24*** | 0.04 |
| Utility of shared information | 0.14** | 0.05 | 0.14** | 0.05 | 0.35 | 0.10 | ||
| Frequency of pay seeking | 0.25*** | 0.06 | 0.87** | 0.28 | ||||
| Token v pay information conditions | 0.21* | 0.09 | −0.07 | 0.11 | −0.10 | 0.11 | −0.12 | 0.11 |
| Utility of sharing information x frequency of pay seeking | −0.13** | 0.06 | ||||||
| Model R square | 0.05 | 0.12 | 0.14 | 0.15 | ||||
- *** p < .001, two-tailed,
- ** p < .01, two-tailed,
- * p < .05, two-tailed, and
- † p < .10, two-tailed.
Hypothesis 1b proposed that the utility of the shared information would be positively related to willingness to disclose. Model 2 in Table 2 provides results in support of hypothesis 1b, showing that the utility of shared information was a strong predictor of willingness to disclose (estimate = .14, p = .00). Hypothesis 1c proposed that the relationship between information and willingness to disclose would be mediated by the utility of shared information. Mediation was tested using the mediation library in R (Imai et al., 2010; Tingley et al., 2014) and the non-parametric bootstrap with 5,000 draws to establish confidence intervals. 5 The indirect effect of .03 was significant with 95% CIs of [0.003, .07] providing support for Hypothesis 1c.
Hypothesis 2a proposed that the relationship between the utility of shared information and willingness to disclose would be moderated by the frequency of pay-seeking behavior. In support of the hypothesis, the interaction between the utility of shared information and the frequency of pay-seeking in Model 4 of Table 2 is significant (estimate = −0.13, p = .023). Figure 3 illustrates the form of the interaction. Individuals who reported a low frequency of pay-seeking behavior were more willing to disclose pay information as the utility of the shared information increased. In contrast, individuals who reported a high frequency of pay-seeking behavior were significantly less sensitive to utility manipulation, reporting a strong willingness to disclose their own pay information regardless of the utility of the information shared by the website.
The cumulative hypotheses to this point can be integrated into a formal test of a second-stage moderated mediation model (Edwards & Lambert, 2007) articulated as Hypothesis 2b. Results revealed that the previously identified mediation effect in Hypothesis 1c was significant for respondents with low pay-seeking behavior (indirect effect of .04, 95% CI [.002, .10]). In contrast, respondents with high pay-seeking behavior showed no evidence of mediation (indirect effect of −0.005, 95% CI [−0.04, 0.02]). A formal contrast of these two mediation effects indicated that they were significantly different from each other (mediation difference of 0.05, p = .045, 95% CI [0.0008, 0.12]). A robustness check testing the second stage moderated mediation model without controls returned the same difference in the indirect effect of 0.05, but a p-value of .08 suggesting that the inclusion of the controls helped reduce error variance. Overall, the results indicate that the utility of shared information was an important mediator between the information provided (pay versus token information) and willingness to disclose primarily for those who reported low pay-seeking behavior. In contrast, the utility of shared information was not an important factor for those who reported high pay-seeking behavior because they tended to show high willingness to share regardless of whether they believed the shared information would be useful.
4.2 Actual pay disclosure
A series of logistic regression models were estimated to test predictors of actual pay disclosure. The main effects reported in Model 1 of Table 3 reveal that females were significantly less likely to report actual pay (estimate = −0.59, p < .05), that willingness to disclose was a strong positive predictor of actual disclosure (estimate = 0.77, p < .001), and that employer pay communication restrictiveness was positively related to actual disclosure (estimate = 0.33, p < .01).
| Model 1: Main effects | Model 2: Two-way interactions | Model 3: Three-way interaction | ||||
|---|---|---|---|---|---|---|
| Coef. | Std. err. | Coef. | Std. err. | Coef. | Std. err. | |
| Intercept | −1.51* | 0.72 | −2.70* | 1.17 | −1.46 | 1.32 |
| Organization tenure | 0.02 | 0.02 | 0.02 | 0.02 | 0.02 | 0.02 |
| Female | −0.59* | 0.22 | −0.57* | 0.23 | −0.57* | 0.23 |
| Pay sharing preference | 0.04 | 0.09 | 0.04 | 0.09 | 0.05 | 0.09 |
| Age | −0.03* | 0.01 | −0.02* | 0.01 | −0.02* | 0.01 |
| Union member | 0.46 | 0.45 | 0.50 | 0.45 | 0.49 | 0.45 |
| Willingness to disclose | 0.77*** | 0.08 | 0.93*** | 0.26 | 0.54 | 0.33 |
| Deliberative thinking | 0.04 | 0.22 | 1.41 | 0.91 | −1.17 | 1.75 |
| Employer pay comm restrictiveness | 0.33*** | 0.11 | 0.68* | 0.28 | 0.28 | 0.36 |
| Willingness to disclose * deliberative thinking | 0.03 | 0.16 | 0.85† | 0.50 | ||
| Willingness to disclose * employer pay comm restrictiveness | −0.05 | 0.07 | 0.07 | 0.10 | ||
| Deliberative thinking * employer pay comm restrictiveness | −0.45* | 0.21 | 0.32 | 0.49 | ||
| Willingness to disclose * deliberative thinking * employer pay comm restrictiveness | −0.24† | 0.14 | ||||
| Pseudo R Square | 0.23 | 0.24 | 0.25 | |||
- *** p < .001, two-tailed,
- ** p < .01, two-tailed,
- * p < .05, two-tailed, and
- † p < .10, two-tailed.
Hypothesis 3a proposed a two-way interaction between willingness to disclose and the restrictiveness of employer pay communication policy. The results in Model 2 of Table 3 provide no support for this hypothesis (estimate = −.05, ns). It is likely, however, that this non-significant finding reflects the differential impact of the random assignment to deliberate thinking proposed as the higher-order three-way interaction in Hypothesis 3b. Model 3 of Table 3 provides the results of the three-way interaction (estimate = −0.24, p < .10). Hypothesis 3b was supported with the proposed 3-way interaction significant at a 90% confidence level. The form of the interaction (plus or minus one standard deviation for employer pay restrictiveness) is presented in Figure 4 and reflects the inherent non-linear nature of logistic regression (Long & Freese, 2014). Results are consistent with the hypothesized form, suggesting a less than a 5% chance of obtaining the observed form of the relationship if the null hypothesis was true (i.e., a p-value of less than .05 using a one-tailed test). Specifically, Panel A illustrates the neutral (no deliberative thinking) condition and reveals that the relationship between willingness to disclose and actual pay disclosure is unaffected by the level of employer restrictions on pay communications. That is, among participants not primed to engage in deliberative thinking about the possible consequences of pay disclosure, there was no interaction between willingness to disclose and participant perceptions of the restrictiveness of employer pay communication policy. In contrast, Panel B reveals that under the deliberative thinking condition (and where, due to a restrictive pay communication policy, the potential costs of disclosure are likely to outweigh the benefits), the relationship between willingness to disclose and actual pay disclosure is, as hypothesized, weaker. However, as shown in Figure 4b, this effect of deliberative reflection is reversed (amplification of the effects of willingness on actual disclosure), when – in the absence of such a policy – the potential costs may no longer outweigh disclosure benefits. The strength of the three-way interaction was invariant to the omission of controls.
4.3 Post hoc tests
Finding (H3b) that under deliberative thinking the relationship between willingness and actual pay disclosure is weaker when a situational factor (employer policy) is perceived as more restrictive, we explored the sensitivity of the willingness–disclosure relationship to employee turnover intentions. Measuring turnover intentions (at T1 using three items from Cammann et al., 1983), we assumed that for those contemplating leaving their current organization, deliberative thinking on the relative costs and benefits of pay disclosure would, by making the benefits of exchange more salient, amplify the impact of willingness to disclose on actual disclosure. Indeed, the findings indicate a significant three-way interaction of turnover intentions x willingness to disclose x reflection (estimate = 0.20, p = 0.07), even after controlling for the restrictiveness of the employers' pay communications. That is, our exploratory findings suggest that when individuals, such as those intending to leave their current employment, have little to lose and more to gain by engaging in information exchange, deliberate reflection on the utility of such behavior intensifies the association between attitudinal willingness to disclose pay information and actual pay disclosure itself.
5 DISCUSSION
Drawing from and extending UMT we identify factors that help explain the variance in third-party mediated employee pay disclosure behavior. We proposed and found that over and above the effects of dispositional differences in pay-sharing preferences, the a-priori provision of detailed and personalized pay information by a pay information website sequentially impacts individuals' perceptions of the utility of shared pay information, and through it, their willingness to disclose pay information, and as a result, actual pay disclosure. Further, the indirect effect of the perceived utility of shared pay information on employee pay disclosure is moderated by several heuristic processes. First, the salience of pay-related fairness uncertainty has a robust conditioning effect on the association between the perceived utility of shared information and one's willingness to disclose. Under low salience conditions, employee willingness to disclose positively varies as a function of the perceived utility of shared information. Under conditions of high fairness uncertainty salience, such predilections, while high, are largely invariant to the level of perceived utility of the shared information.
Second, the impact of individuals' willingness to disclose on actual pay disclosure behavior is moderated by (a) perceptions of the restrictiveness of employers' pay communication policy and (b) the degree to which employees engage in conscious deliberation of the benefits and costs of pay disclosure. We found that the moderating effect of employer pay communication restrictiveness on the willingness to disclose–pay disclosure relationship is dependent upon whether individuals give deliberative thought to the possible consequences of disclosure. Deliberative thinking weakened the association between willingness to disclose and actual disclosure when participants perceived their employer's pay communication policy as more restrictive, whereas among participants perceiving pay communication policies as less restrictive, deliberative thinking had the opposite effect, amplifying this association.
5.1 Theoretical implications
Our findings offer three contributions to the compensation management literature. First, extending UMT to the question of employee pay disclosure, we propose and demonstrate empirical support for a model explaining why and when employees may disclose their pay information to third parties such as pay information websites. Our findings offer important insights into why and when employees may violate the “social taboo” of disclosing their pay to others (Smit & Montag-Smit, 2019), and offer some of the first empirical evidence in support of Colella et al.’s (2007) untested observation that employees might be willing to trade off some privacy about their own pay to learn about others' pay. Second, we identify employee perceptions of the utility of the information shared a-priori by a potential exchange partner to be an important driver of individual willingness to disclose. While this effect is conditioned by fairness uncertainty salience, our findings support the notion that employees frame pay disclosure as an exchange, with employees willing to disclose their own pay-related information to reduce their own uncertainties regarding pay fairness but doing so only when they deem the implicit terms of such an exchange to be worthwhile. Our results indicate that pay information exchange partners such as pay information websites can influence utility perceptions (and subsequently, disclosure attitudes and behavior) by sharing detailed and personalized pay information prior to requesting the employee to disclose their own pay information. It follows that employee engagement with online information exchange services is contingent upon the degree to which these services can demonstrate a-priori that disclosure offers the employee the ability to access relevant, usable, and potentially valuable employment-related information.
Third, we provide novel insights into how and when a willingness to disclose leads to actual disclosure to a website, currently a poorly understood process (Cheng et al., 2021; Cichy et al., 2021). Our theoretical model and findings suggest that employee pay disclosure is nuanced by heuristic processing, and that models that assume employees engage in reflective utility maximization are likely to offer limited explanatory potential. Situational factors such as employer policies (i.e., restrictive pay communication policies) may be taken by employees as implicit signals of unfairness, unconsciously driving reactance (reduced rule compliance; Brehm, 1966), and thus increasing the likelihood of actual disclosure to third-party websites. Indeed, our results indicate that unless individuals are led to actively engage in reflecting on the potential benefits and costs of disclosure, such restrictive pay communication policies and practices may, consistent with reactance principles, in fact, bolster the precise behaviors they were intended to deter. Indeed, in the absence of deliberate reflection, there was no significant moderating effect of restrictive pay communication policies on the relationship between willingness and actual pay disclosure. This finding suggests that the confidentiality assured by pay information platforms, perhaps along with the relative anonymity associated with uploading information onto the internet and the expectation of receiving useful pay information, generated a rather automatic tendency among employees to disclose their pay information. If this is the case, it places employers in something of a predicament since the only way for them to deter such employee behavior would be to take steps to bolster employee awareness of the sanctions associated with such behavior; steps that may be illegal in an increasing number of jurisdictions.
5.2 Implications for policy and practice
The response by public policy makers to the widespread use of restrictive employer pay communication has been legislation prohibiting employer retaliation against employees who violate restrictive policies (Burn & Kettler, 2019). Our study shows that when employees are primed to think about the consequences, a restrictive employer pay communication policy can limit the extent to which an individual's willingness to disclose their pay translates into actual pay disclosure. Hence, legislation may have a role to play in facilitating employee pay information exchange. Facilitating greater individual access to others' pay information can generate broader labor market benefits, including diminished pay inequities (Bennedsen et al., 2022) and enhanced labor market and mobility efficiencies (Bamberger, 2023; Harris, 2018).
For managers, the rise of pay information platforms poses a significant challenge to the asymmetry in pay knowledge that employers have long enjoyed (Bergh et al., 2019). The quality of information obtained via pay information websites is enhanced by the volume of information received on any position and by the efforts of third-party websites to ensure their reputation by vetting the submitted information (Rosenfeld, 2017). The fact that over two-thirds of our study participants acceded to the request to disclose their bonus pay information suggests that employers may need to reconsider their restrictions on pay communication. What remains to be seen is how much pay information employers will make available and the level of specificity of that information (e.g., specific pay, average pay, pay range information) (Brown et al., 2022).
5.3 Limitations and directions for future research
Despite the significance of our findings, several aspects of our study design may limit the external and internal validity of our findings. The generalizability of our findings to employees' disclosure of their pay information to pay information platforms may be limited by our reliance on a pooled sample, the members of which received modest compensation for their participation in the study and who, in some cases, had a standing relationship with the survey company we contracted. Participants' preexisting relations with the survey company may have lowered the perceived risk of disclosure. While such design considerations may have inflated the proportion of participants disclosing pay information, other design considerations, particularly the advice in the initial phase of the experiment that the proposed pay information data base had yet to be fully developed, may have served as a disincentive, counter-balancing any heightened tendency to disclose. Future researchers might therefore sample those engaging with actual pay information exchange websites (whether ultimately disclosing their own pay information or not) to assess the degree to which our findings are generalizable, and perhaps further explore what types of signaling by those soliciting pay information may be most effective in driving employees to disclose their pay.
Additionally, our sample was limited to participants eligible for a performance-based bonus and our response bias test indicated an overrepresentation of participants reporting greater employer pay communication restrictiveness and a higher frequency of pay seeking. This does not however limit the generalizability of our findings as employer pay communication restrictions and pay information search are likely to be less relevant for and prevalent among those for whom pay information is publicly available (e.g., employees whose total pay is largely influenced by sector-wide agreements; Chamberlain, 2015).
Two internal validity issues serve as additional limitations. First, since our study examined the disclosure of performance-based bonuses, a more comprehensive model of employee pay disclosure should examine whether the same mechanisms that underlie bonus disclosure also underlie the disclosure of other pay forms. Second, to assess fairness salience, we used a self-report measure, namely the frequency of pay information seeking. Although others have used similar approaches to assess individual fairness salience (Smit & Montag-Smit, 2019), as factors other than fairness salience may also drive pay information seeking, in future studies there may be reason to apply more direct measures.
Our findings suggest several avenues for future research. First, researchers have typically focused on direct pay information exchange between co-workers in an organization (Smit & Montag-Smit, 2018, 2019), but new technologies have opened a variety of additional and alternative exchange mechanisms. In addition to pay platforms (such as Glassdoor), scholars should investigate the exchange of pay information that occurs privately between multiple co-workers in the same organization (e.g., ‘Young Microsoft FTE’; see Gershgorn, 2019) and software applications (e.g., ‘Blind Professional’) that provide a mechanism for an employee in one organization to anonymously engage in pay information exchange with an employee of another organization. While each of these exchange mechanisms has the potential to reduce an individual's pay fairness uncertainty, scholars might examine how employees select one form of exchange over another.
Second, scholars may wish to examine factors explaining the frequency with which employees search for pay information, as well as the extent to which they rely on more than just one pay information exchange mechanism. Participation in multiple pay information exchange mechanisms might reflect the intended use of the information, such as remaining with the organization or seeking employment elsewhere. Accordingly, understanding how and how often pay information is accessed is important as it can shed light on the choices that employees make, as well as impact pay equity and the competitiveness of labor markets (Harris, 2018).
Third, we identified some of the consequences of employer restrictions on pay communications, but the higher survey dropout rate between T1 and T2 of participants with employer pay communication restrictions suggests there is more to learn. Our measure of employer pay communication restrictions asked about the presence of a policy, but we currently know very little about when and how employers discipline an employee for a breach of the policy (Brown et al., 2022). Understanding employee perceptions of when (e.g., accidental, or intentional breach) and how (e.g., disciplinary action, dismissal) an employer enforces a policy will further our understanding of when employees engage in pay information exchange.
6 CONCLUSION
Theory and research on employee pay communication are largely grounded on the assumption that when it comes to pay communication, agency lies largely with the employer, and that employee privacy preferences underlie an overwhelming tendency towards employee non-disclosure of pay. However, as we have demonstrated, employees may play a critical role in pay communication, particularly given the increasing prevalence of, and employee engagement with, pay information websites. Moreover, our findings suggest that regardless of privacy preferences, employees may in fact be highly open to disclosing pay information, at least when they can do so via mediated online exchange platforms. Our findings lend important insights into the individual difference and situational factors underlying employee pay information exchange with such platforms and accordingly provide a model for understanding how and when contemporary employees disclose their pay to others.
ACKNOWLEDGEMENTS
This research was supported by a grant from the Australian Research Council (Discovery Project 1096775). This study was developed and executed while Professor Bamberger was serving as Professorial Fellow, Department of Management and Marketing, The University of Melbourne. Open access publishing facilitated by The University of Melbourne, as part of the Wiley - The University of Melbourne agreement via the Council of Australian University Librarians.
CONFLICT OF INTEREST STATEMENT
We have no conflicts of interest to disclose.
Biographies
Michelle Brown is a professor of human resource management at the University of Melbourne. Her research focuses on performance feedback systems, reward management, and employee share ownership.
Peter Bamberger is the Domberger Chair in Organization and Management at Tel Aviv University's Coller School of Management. His research examines rewards management, employee behavioral health, team effectiveness, and prosocial behavior. He is a Fellow of the Society for Industrial/Organizational Psychology and the Academy of Management.
Paul D. Bliese is the Jeff B. Bates Chair in the Department of Management at the Darla Moore School of Business, University of South Carolina. His research focuses on organizational stress and efforts to enhance the use of statistical methods to advance theory and practice.
John Shields is professor of human resource management and organizational studies at The University of Sydney Business School. His research interests include performance management, reward management, leadership, and corporate governance.
Open Research
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.
