Performance evaluations and employee turnover intentions: Empirical evidence from linked employer–employee data
Abstract
EconStor is a publication server for scholarly economic literature, provided as a non-commercial public service by the ZBW.
Full text
Pohlan, Laura; Steffes, Susanne Article — Published Version Performance evaluations and employee turnover intentions: Empirical evidence from linked employer– employee data Industrial Relations: A Journal of Economy and Society Provided in Cooperation with: John Wiley & Sons Suggested Citation: Pohlan, Laura; Steffes, Susanne (2024) : Performance evaluations and employee turnover intentions: Empirical evidence from linked employer–employee data, Industrial Relations: A Journal of Economy and Society, ISSN 1468-232X, Wiley, Hoboken, NJ, Vol. 64, Iss. 3, pp. 395-433, https://doi.org/10.1111/irel.12379 This Version is available at: https://hdl.handle.net/10419/323807 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/4.0/
Industrial Relations. 2025;64:395–433. | 395wileyonlinelibrary.com/journal/irel INTRODUCTION Firms are typically interested in preventing their employees from quitting, as the resources firms spend on recruiting and developing employees constitute a significant portion of total labor costs (see, e.g., Dube etal.,2010; Muehlemann & Leiser,2018). In this article, we argue that firms can use performance evaluations to reduce information asymmetries and provide pathways for future prospects within the firm, thereby increasing employee retention. The latter is crucial for the firm to prevent employee quits and clearly depends on good working conditions, decent salaries and promising career options (see, e.g., Clark, 2001; Cottini etal.,2011; Frederiksen,2017; Paul & Scott,2011). But existing career options need to be recognized by employees, and there is a growing perception that people make systematic mistakes when assessing their future prospects within or outside their current employer.1 Therefore, we 1An abundance of evidence from the lab has identified overand underconfidence of individuals regarding their ability. In the field, Hoffman and Burks(2020) or Huffman etal.(2022) detect significant and persistent overconfidence. Jäger etal.(2021) find that employees systematically have wrong beliefs about their outside options, i.e., the wages they could earn elsewhere. Received: 19 November 2023 | Accepted: 2 September 2024 DOI: 10.1111/irel.12379 ORIGINAL ARTICLE Performance evaluations and employee turnover intentions: Empirical evidence from linked employer– employee data LauraPohlan1,2,3,4 | SusanneSteffes4,5 This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2024 The Author(s). Industrial Relations published by Wiley Periodicals LLC on behalf of Regents of the University of California (RUC). 1Institute for Employment Research (IAB), Nuremberg, Germany 2IZA–Institute of Labor Economics, Bonn, Germany 3Labor and Socio-Economic Research Center (LASER), Nuremberg, Germany 4ZEW–Leibniz Centre for European Economic Research, Mannheim, Germany 5German Federal Statistical Office, Wiesbaden, Germany Correspondence Laura Pohlan, Institute for Employment Research (IAB) of the German Federal Employment Agency (BA), RegensburgerStrasse 104, D90478 Nuremberg, Germany. Email: [email protected] Abstract In this article, we study whether performance evaluations can serve as an instrument for firms to increase employee retention. Feedback on one's own performance may affect individual turnover intentions differently depending on the relative wage rank of workers among their peers. In line with these considerations, empirical evidence based on panel employer–employee data shows that relatively lowpaid employees decrease their turnover intentions after the implementation of a performance evaluation system at the establishment level. We find no effect for relatively highpaid employees. JEL CLASSIFICATION M51, M54, J63, C23
396 | POHLAN and STEFFES suggest that it is very likely that employees also have incomplete perceptions about their career options. In this study, we analyze the relationship between the introduction of firmlevel performance evaluations and employee turnover intentions empirically based on four waves of a linked employer–employee panel data set of German firms, the Linked Personnel Panel (LPP), which can be combined with social security records from the Federal Employment Agency for all workers in surveyed establishments. Turnover intentions reflect the employee's own impetus to leave the organization and can therefore be interpreted as a direct behavioral response to workrelated changes. The intention to leave might translate into actual quits, but dissatisfied employees who consider changing jobs but do not do so can also contribute to productivity losses and thus cause economic costs. Using the administrative data to calculate relative residual wage positions within firms, we analyze how the introduction of a performance evaluation system at the establishment level influences the employees' turnover intentions depending on their relative performance. To assess this relationship, we exploit withinindividual variation and estimate fixed effects and first difference specifications. In addition, the rich data allows us to shed light on mechanisms to some degree. In particular, we examine how job satisfaction and individual perceptions of fairness and promotion criteria change when performance evaluations are implemented. This article focuses on the overall management strategy to implement performance evaluations and does not consider the specific design of a performance evaluation system.2 Descriptive analyses indicate that establishments that introduce performance evaluation systems are comparable to other firms with respect to most dimensions, such as firm size, sector composition, wage premia, performance pay systems, support in decision making or promotion criteria. We further show that the introduction of performance evaluations often comes along with a bundle of HR practices that likely influence career prospects such as target agreements or personnel development plans. In the next step, we examine the impact of the introduction of a performance evaluation system on withinindividual changes in turnover intentions and document an overall negative impact on the intention to leave of professional employees.3 To separate the change in the provision of performance evaluations from further adjustments, we control for changes in other HR instruments and employment conditions at the individual or establishment level. Moreover, we estimate different specifications to study directly the impact of the introduction of clusters of HR practices. First, we make use of a survey question that asks establishments whether the introduction of the performance evaluation system only affected the area of employee assessment or whether changes also took place in other areas. In further specifications, we consider only the introduction of a performance evaluation system that is not accompanied by the introduction of target agreements or performance pay schemes or personnel development plans. In sum, these analyses show that the introduction of performance evaluations has a direct impact on the turnover intentions of employees irrespective of whether evaluations are implemented within a broader set of HR practices or not. Finally, we exploit information on the triggers for introducing performance evaluations 2The design of the performance evaluation system is matter of interest in many studies. In lab and field experiments, scholars establish individual versus relative performance evaluations, vary the information which is revealed to the employee, or vary the timing and the frequency of the feedback talks. As in Azmat and Iriberri(2016), performance feedback is often combined with other measures like the compensation system. 3Thereby our measured treatment does not capture whether an employee is actually treated by the implementation of the evaluation system. Although information on the actual treatment of individuals is available in our data, employees who are subject to performance evaluations might be a highly selected group, while the introduction of a performance evaluation system for all employees at the establishment level is less likely to be driven by the productivity of individual employees.
| 397 PERFORMANCE EVALUATIONS AND TURNOVER INTENTIONS at the establishment level. The most commonly indicated triggers are new ideas of the management, positive internal experience, and internal problems. In a sensitivity analysis, we distinguish two types of triggers, those that relate to innovations or external requirements and those that relate to internal experiences and problems. Our results are very similar for both types of triggers. Individual intentions to leave the organization are influenced by compensation, career prospects, and perceptions of organizational support, among other factors. Therefore, the impact of performance evaluations on employees' turnover intentions is likely to depend on the relative performance of the worker. Since we do not have a direct measure of relative performance, we use a proxy by assigning employees according to their observed rank position in the residual wage distribution of their peer group within the same establishment. The idea behind this approach is that firms often incentivize their employees by linking their pay to performance, which leads to wage dispersion between employees with similar characteristics performing similar tasks (see, e.g., Lazear, 2018). Specifically, we predict the wage of each individual conditional on observable characteristics such as gender, age, professional degree and the employment history. Then we calculate the residual wage within an establishmentoccupationjoblevel cell. In this way, we are able to determine the relative rank (below or above the median of the residual wage distribution) of an individual employee among a very similar group of coworkers. However, our relative rank measure only partially reflects relative performance in the economic sense, but rather the decision of employers to pay some workers within a comparable group better than others.4 Therefore, we use the term relative pay measure in the following. Our results suggest that belowmedian paid employees decrease their turnover intentions after the implementation of a performance evaluation system. For those paid above the median, we do not find a change in turnover intentions. These results are confirmed by looking at actual job search or employer switches. Moreover, we find that relatively low paid are more satisfied with their job and perceive promotion criteria as more objective after the introduction of performance evaluations. This article contributes to two different strands of literature. First, it contributes to the literature on relative wage positions and quit behavior which documents mixed results. Using linked employer–employee data from Germany, Pfeifer and Schneck(2012) find that workers with higher relative wage positions within their firms are, on average, more likely to quit their jobs than those with lower relative wage positions. The authors argue that this can be explained by the signal of few internal career advancement opportunities.5 In contrast, Mohrenweiser and Pfeifer(2023) document a positive correlation between quit intentions and wages of coworkers in the same firm using the first two waves of the LPP and Godechot and Senik(2015) find no significant correlations based on a French employer– employee database. Our study helps to shed light on the mixed findings on the relationship between relative wage positions and turnover by exploring the introduction of a performance evaluation system as an important moderator. Thereby we aim to extend the current literature in this direction to show whether the potential or intended revelation of internal career opportunities through a general performance evaluation system influences the intention to leave differently depending on the relative performance of workers among their peers. The revelation of relative performance could motivate the highperforming workers to search externally for jobs, as found by Pfeifer and Schneck(2012), who suggest that signaling few internal career prospects drives the quit behavior 4This decision may also be influenced by other aspects of wage determination, such as unobserved job characteristics, bargaining power or discrimination. 5See also, e.g., Clark etal.(2009) who find the same pattern for the relationship between relative wages and individual job satisfaction.
398 | POHLAN and STEFFES of employees with higher relative wage positions. But our results indicate that the introduction of a performance evaluation system in the firm does not significantly affect the turnover intentions of this group. This can presumably be explained by the fact that relatively highpaid employees are already sufficiently informed about their current wage rank and their future career opportunities within the organization. Instead, we observe a negative impact on the turnover intentions of lower paid employees that might partly be explained by the more precise signal of career and development prospects. Although we cannot empirically identify the potential drivers of this effect, our further analyses of other measures of affective behavior suggest that improvements in perceptions of fairness and organizational support related to career opportunities play some role. In addition, our study adds to the recent literature on performance evaluations and the impact on job transitions, which is only a small strand of the literature among the many studies on performance evaluations and other outcomes. Existing studies rely on personnel data of single firms and show a negative relationship between performance evaluations and job transitions. Based on the data of six large companies, Frederiksen etal.(2017) document a negative correlation between performance ratings and separation rates. In two firms, where it is possible to disentangle quits from dismissals, both types of exits are negatively correlated with performance, and the correlation between performance and dismissals is stronger. These correlations are also confirmed by Cappelli and Conyon(2018) who analyze data from one large U.S. Corporation. Using personnel records from a car sales company, Takahashi etal.(2021) find that lowerthanexpected evaluations lead to employee quits only when the supervisor is inexperienced, but there is no association with quits when the unexpectedly lower evaluation is given by an experienced supervisor.6 We contribute to this literature by showing that the revelation and discussion of performance through a general performance evaluation system in the firm could have an impact on the correlation between performance (measured in relative wage rank positions) and turnover intentions. Hence, a firm that is under pressure to retain its employees, even if they are not top performers, might therefore consider using such an evaluation system to motivate its employees to stay with the organization. The recent economic literature mainly focuses on the impact of performance evaluations on performance and motivational outcomes in lab experiments (see, e.g., Azmat & Iriberri,2016; Gill etal.,2019; Kuhnen & Tymula,2012). Azmat and Iriberri(2016) for instance, find in a lab experiment that relative performance feedback increases the subsequent effort of employees who perform under a piecerate incentive scheme. They do not find significant treatment effects under a flatrate system. Interestingly, this result holds independently of the content of the feedback; hence whether the feedbackreceiving person was performing above or below average. However, Azmat and Iriberri(2016) also find that the content of the feedback matters. Contrary to the findings on performance, the authors find very different effects on satisfaction for belowand aboveaverage performers: receiving positive (negative) feedback increases (decreases) individuals' subjective wellbeing and feeling of dominance but again only under a piecerate incentive scheme. To sum it up, our study is the first comprehensive analysis of the relationship between the provision of a performance evaluation system, relative ranks and individual turnover intentions. For this purpose, we use a unique source of field data, a longitudinal linked employer–employee survey combined with administrative records, which allows us to overcome the drawback that personnel records from single firms or data gained through lab or field experiments might be dominated by the specific setting in the firm or in the experiment. We first provide descriptive evidence on the type of firms that introduce a performance evaluation system, the triggers for the implementation and whether other HR practices are introduced simultaneously. Second, we analyze which group of workers, 6In a similar vein, Giuliano etal.(2011) find that workers are less likely to quit or be dismissed if they are the same race as their manager. Engellandt and Riphahn(2011) demonstrate the importance of flexible individual performance evaluations and the implicit cost of favoritism.
| 399 PERFORMANCE EVALUATIONS AND TURNOVER INTENTIONS measured by their relative wage rank, changes their turnover intentions after the implementation of a performance evaluation system and document that relatively lowpaid employees reduce their turnover intentions while relatively highpaid employees do not. Our results are also of direct value for practice: apart from monetary incentives, performance evaluations might represent an HR strategy to reach all employees and contribute to a work environment where personnel decisions such as promotions are more likely to be based on objective criteria. This can reduce individual turnover intentions and hence possibly also serve as an instrument for firms to increase employee retention. The rest of the article is structured as follows. Section2 provides theoretical considerations. Section3 describes the data source, the measurement of variables that are relevant for our analysis and the estimation sample. Section4 presents the empirical approach and the results. Section5 concludes. THEORETICAL CONSIDERATIONS From a theoretical perspective, the relationship between performance evaluations and employee turnover intentions is not obvious. In the following, we discuss three main drivers that could differentially affect turnover intentions, especially when heterogeneity in performance within narrow groups of workers is taken into account. The first driver is the firm's motive to align its employees' incentives with its own. The second is the provision of information on employee performance, which reduces asymmetric information about career prospects. The third are employees' behavioral responses to perceived organizational support. Primarily, firms might implement a structured performance evaluation system to better align incentives of their employees with their own incentives and increase in this way firm productivity.7 When setting incentives, individual bonuses are commonly based on the performance indicators that are determined through assessment processes (see, e.g., empirical evidence by Cappelli & Conyon, 2018; Frederiksen et al., 2017; Kampkötter & Sliwka, 2016; Kuhnen & Tymula,2012). In a work environment where workers perform many different tasks and work in teams, performance will be inadequately assessed by objective measures such as sales or financial figures. As a result, most performance evaluations in practice are based on subjective assessments by supervisors (Kampkötter & Sliwka, 2016). Theoretical literature relying on principalagent models has analyzed the use of subjective evaluation to incorporate hardtomeasure tasks in an incentive contract that ties compensation to performance (see, e.g., Baker etal.,1994; Chan & Zheng,2011; MacLeod,2003). Takahashi etal.(2021) provide empirical evidence based on personnel records from a car sales company and confirm that supervisors use the evaluations to incentivize employees to pursue nonsales tasks, such as mentoring junior representatives and building longterm customer relationships. Kampkötter and Sliwka(2016) and Frederiksen etal.(2017) discuss typical aspects in the design of a performance evaluation system that are often applied in companies. A typical evaluation process takes place on a regular and structured basis. Firms commonly use annual cycles: at the beginning of the year, employees and supervisors meet to set goals, which are reviewed at the end of the year using a performance scale that ranges, for instance, from 1 “low rating” to 5 “high rating”. In most incentive schemes, the performance evaluation results then serve as the basis for performance pay. In addition, evaluation outcomes are used in the personnel planning process 7Lazear(2000) documents that productivity increases significantly in an U.S. company that introduces a piecerate compensation scheme. The gain is due to both an incentive effect meaning workers becoming more productive under performance pay and changes in employee turnover patterns. The author argues and shows that performancebased compensation, which requires a measure of performance, is more attractive to high performers.
400 | POHLAN and STEFFES for decisions on promotions, reallocations or dismissals. For instance, Frederiksen etal.(2017) show based on personnel data of six large companies and Cappelli and Conyon(2018) based on data from one large U.S. corporation that subjective performance ratings are positively related to several career outcomes such as pay, promotions, demotions, dismissals and quits. The second driver focuses on reducing asymmetric information from the employee's perspective. Performance evaluations can involve the exchange of various types of information: First, performance evaluations provide direct “feedback” on where performance was good or not so good and what should be done differently in the future (Cappelli & Conyon,2018). This can help employees identify their potential strengths and weaknesses and improve their prediction of internal and external career prospects. In addition to assessing current performance, performance evaluations can also be the starting point to derive a perspective for future development within the firm, for example by discussing and planning of goalsetting procedures, promotions or training measures (Bakker & Demerouti,2007; Beenen etal.,2017). Thus, performance evaluations can serve as a tool to inform and guide employees by identifying internal pathways for advancement, such as reassigning tasks or investing in new skills. Finally, performance evaluations could include a discussion of why workers receive certain ratings and how they compare to their peers' ratings. The considerations so far suggest that the impact of performance evaluations on employees' turnover intentions is likely to depend on the relative rank of the worker in the performance distribution. Assuming that ratings are not biased and reflect actual differences in performance,8 highperforming employees receive a better rating than lowperforming employees in their comparison group. As performance reviews allow to reward good performance and to sanction poor performance more adequately, this might, in turn, decrease the turnover intentions of high performers and increase the turnover intentions of low performers. The same predictions are obtained if the performance assessment by the supervisor is more likely to deviate negatively from the employee's own assessment for low performers and to deviate positively for high performers. However, there might be opposing mechanisms that influence the impact of performance evaluations on the intention to leave the organization. First, better knowledge of internal career options could also influence the view of external career options. In line with this, Pfeifer and Schneck(2012) argue that higher wages of coworkers might signal better career prospects within the firm. Hence, relatively lowperforming individuals might reduce their turnover intentions as they have underestimated their future career prospects or overestimated their outside options thus far. In contrast, if a worker is already situated at the top of the pay distribution, further career advancements in the current firm are unlikely and the worker may consider quitting the job and joining another firm. Moreover, generous ratings and little differentiation between workers might lead to comparatively better ratings of low performers and worse ratings of high performers and in this way counteract the intended impact of performance evaluations. Finally, the third driver suggests that providing employees with information and guidance could have an overall positive impact on employee satisfaction and the decision to stay with the employer (Kampkötter,2017). The original idea relates to the psychological literature on perceived organizational support (POS), which states that employees' commitment to their organization might be influenced by the organization's perception of commitment to them (Eisenberger etal.,1986). Based on the organizational support theory, empirical studies from that field show a strong relationship between POS and behavioral outcomes of employees such as affective commitment or job satisfaction (Kurtessis et al., 2017). Moreover, the studies show that POS is affected by three basic antecedents, which can be summarized as “fairness”, “HR practices and working conditions”, and “leadership” 8Several studies point to the possibility that subjective evaluation by supervisors is systematically biased (see, e.g., Frederiksen etal.,2017; Kampkötter & Sliwka,2016, and the citations therein). For instance, subjective evaluations are known to be prone to “centrality bias”, meaning that supervisors not sufficiently distinguish between workers and to “leniency bias”, meaning that supervisors overstate their subordinates' performance.
| 401 PERFORMANCE EVALUATIONS AND TURNOVER INTENTIONS (Eisenberger etal.,2020). Kurtessis etal.(2017), for example, find that perceived supervisor support and the perception of organizational politics are strongly correlated with POS. We expect that the introduction of a performance evaluation system can influence perceptions of supervisor support because the supervisor must view the employee individually, regardless of common work structures and behaviors. The organizationwide implementation of performance evaluations could also be seen as a tool to combat structures of favoritism, lack of meritocracy and lack of frank discussion. The strongest relationship between the antecedents of POS and performance evaluations is through the family of HR practices and working conditions. The introduction of a performance evaluation system could also be accompanied by improvements in development opportunities and enrichment of job characteristics. Overall, these considerations suggest that there are opposing mechanisms and that we cannot make clear predictions about the effects of performance evaluations on the turnover intentions of lowperforming versus highperforming employees. Therefore, we analyze this relationship empirically based on linked employer–employee data. DATA Data source and definition of variables Our empirical analysis is based on the four waves (2012/13, 2014/15, 2016/17 and 2018/19) of the Linked Personnel Panel (LPP) which combines survey data of employers and employees. The data covers private sector establishments in Germany which employ at least 50 employees at the date of panel entry. The LPP can be linked with administrative data of the Federal Employment Agency which are based on employers' notifications to the social security authorities (LPPADIAB). While the employee survey covers a random sample of the establishment's workforce, the administrative records are available for ~85% of the employees. This fact makes the data exceptional and very well suited to our research question.9 Overall, the LPP is designed to conduct personnel economic research in Germany. The employer survey contains information on several HR instruments—prevalence, distribution and changes—and further information on establishment characteristics and policies. The employee survey collects information on working conditions, individual perceptions of personnel development and treatment by the employer, the supervisor and colleagues, as well as sociodemographics, personality traits and preferences. The administrative data include, alongside information on different labor market states, such as the start and end date of dependent employment, registered unemployment, registered job search or unemployment benefit receipt, individual information on (daily) wage records, job characteristics, as well as some sociodemographic information. The simultaneous observation of both employers and employees over time allows analyses of the relationship between (changes in) HR practices in the establishment and the employees' behavior in response to them. Moreover, it solves the problem of common method bias which is often discussed in survey studies. In particular, the high reliability of administrative wage information of almost all employees for each establishment is an important asset of the data. To measure the turnover intentions of employees, we make use of the question “How often have you thought of changing your current employer in the last 12 months?”. The answers can be: 1 “never”, 2 “several times a year”, 3 “several times a month”, 4 “several times a day” and 5 9See Kampkötter etal.(2016) and Mackeben etal.(2021) for a more detailed description of the data set.
402 | POHLAN and STEFFES “every day”. The data further include reasons why a job change was considered and we disregard cases of turnover intentions exclusively due to personal circumstances unrelated to the workplace.10 Our treatment variable relies on two employer survey questions. Employers are explicitly asked whether they use certain HR instruments and whether there have been any changes— such as an introduction—with respect to the three instruments performance evaluations, target agreements and performance pay in the last 2 years.11 In addition, employers are asked about the triggers for this change. In a sensitivity analysis, we look separately at those cases in which performance evaluations were introduced due to innovations or external requirements and due to internal experiences or problems, respectively. Obviously, the treatment variable does not contain any information on the content or design of the performance evaluations. In a largescale survey with several hundred establishments, the level of aggregation of such information must remain high. Therefore, the investigation of the impact of a performance evaluation design on turnover intentions is beyond the scope of this article. In our empirical analysis, we compare the results of two different definitions of our treatment variable. First, we define a variable used in fixed effects specifications, indicating whether the establishment has a performance evaluation system or not. For establishments that state they introduced performance evaluations between two consecutive waves, this variable is set to one in the current and subsequent waves and zero in all previous waves. The comparison group consists of both types of establishments: those that never had nor implemented performance evaluations and those that always had performance evaluations. Employers who state that they have abolished performance evaluations are excluded from the analysis. The second variable relies on a before–after comparison, used in first difference specifications, and is defined as an indicator equal to zero if the establishment does not use performance evaluations and equal to one if performance evaluations are implemented. In this case, the comparison group consists of establishments that never had a performance evaluation system in place during the observation period. FigureA1 in the Appendix shows by means of an example the definition of the two treatment variables. In the empirical specification, we also control for changes in other employee and employer characteristics, such as sociodemographic information (age and household composition), job characteristics (function, tasks, compensation, working hours), establishment size, and other HR practices (e.g., target agreements, personnel development plans, support measures for specific worker groups, and performance pay). To study the potential drivers of the effects, we consider job satisfaction and individual perceptions of fairness and promotion criteria as alternative outcome variables. Descriptive statistics of establishments Our empirical analysis relies on a comparison of individuals who are employed in establishments which introduce performance evaluations with individuals who are employed in establishments which do not experience a change in the performance evaluation system. However, establishments might differ in observable characteristics even before the introduction takes place. TableA1 in the Appendix presents descriptive statistics of establishments 10This concerns only a few cases and including them does not change our results. The estimates are available on request. 11The questions are: “Is the performance of employees in your establishment evaluated by supervisors at least once a year?”, “Does your establishment have written target agreements?” and “Does your establishment have a salary system with variable components?” as well as “We are now talking about changes to the performance evaluation system, the written target agreements and the salary system with variable components. Have there been any changes in these areas in the last 2 years? These include the introduction, abolition, expansion and reduction of the respective HR instrument.”
| 409 PERFORMANCE EVALUATIONS AND TURNOVER INTENTIONS and personnel development plans. The results in the Section3.2 indicated that these practices are often simultaneously implemented with a performance evaluation system and may act in a similar direction; particularly, in the context of career opportunities. We therefore expect an upward bias if we do not control for these instruments. The fourth specification includes—in addition to the former variables—information on special feedback for low performers, special training provided to low performers, and the promotion of women in management positions. It also includes information on the provision of training aimed at higher qualifications and whether employee surveys are conducted. Finally, the information about a personnel change in the executive board of the establishment is included in this specification. Finally, we investigate more deeply whether shortterm monetary incentives unrelated to future career prospects (at least not directly) could explain our findings. In particular, we examine whether a simultaneously introduced performance pay scheme at the establishment level has a joint impact on the turnover intentions. This could be the case as performance evaluations are often the floor where the magnitude of performance pay is discussed and is, thus, in some cases jointly implemented. The respective coefficients of performance evaluations of the fixed effects and first difference specification are summarized in Figure2. Although we already interacted the variable of interest with the group dummies, we compare the coefficients of the different specifications for every respective group first, before discussing the results regarding the two groups. The first dot from the left represents the respective coefficient of the baseline specification, whereas the other symbols represent the extended specifications in the above described order. Overall, the results point to the conclusion that our findings are robust to including further HR controls. The magnitude of the changes in coefficients is small and, hence, they are economically and statistically not significant. We also estimate specifications in which we control for the introduction of target agreements, a HR practice that is often jointly implemented with performance evaluations. Although the number of observations is substantially lower because of FIGURE 2 Effect of performance evaluations on turnover intentions by median of relative pay. Panel (a): Fixed effects specification; panel (b): First difference specification; outcome: Turnover intentions (1–5 scale); plotted coefficients: Interactions between conditional relative rank position and performance evaluations on the establishment level; baseline covariates: Partner, number of children, age dummies, leadership position, job strain, working hours (squared), job autonomy, multitasking, establishment size (squared); individual perfpay: Variable pay; HRhperf controls: Employee interviews, personnel development plans; HRlperf controls: Open discussion with low performers, personnel development of low performers, change position of low performers, dismissal of low performers, investment in higher qualifications, employee survey, promoting women in management positions, change in executive board, personnel management in the board of directors; intro perfpay: Performance pay; yearFE. N = 5403–5500 (panel a) and 1809–1857 (panel b). Source: LPPADIAB 7519, own computations. [Colour figure can be viewed at wileyonlinelibrary.com] (a) (b)
410 | POHLAN and STEFFES missing information on the use of target agreements, the results are in line with the findings above.18 However, in line with the literature on POS (Eisenberger etal.,1986), one may argue that complementary HR policies will have the maximum impact when they are introduced jointly and we would like to know the impact of a full cluster of complementary HR practices (see, e.g., Aral etal.,2012; Ichniowski etal.,1997, for different methods on how to bundle specific practices together). The logit regressions presented in TableA3 in the Appendix suggest that the implementation of a performance evaluation system at the establishment level comes along with other HR practices that likely influence career prospects, such as target agreements, employee interviews and personnel development plans. To study the impact of clusters of HR practices, we estimate different specifications which are implementable with our sample size and exploit information from questions that explicitly ask whether there have been changes—including an introduction—with respect to the HR instruments performance evaluations, target agreements and performance pay in the last 2 years. First, we make use of a survey question which asks establishments whether the indicated changes in the performance evaluation system were exclusive to the area of employee assessment or whether these changes occurred in combination with changes in other areas. The area of employee assessment may also contain changes in the usage of target agreements. Nearly 58% of the establishments that introduce a performance evaluation system indicate that this change only affects the field of employee assessment. In the first specification, we only concentrate on treated employees of these establishments. In the next specifications, we consider only the introduction of a performance evaluation system that is not accompanied by the introduction of target agreements (45% of the treated establishments) or by the introduction of a performance pay scheme (88% of the treated establishments). Finally, we look at the introduction of performance evaluations that does not come along with an introduction of personnel development plans (74% of the treated establishments).19 Figure3 shows that the estimated effects in these four specifications are comparable to the baseline specification. In sum, the analyses in this subsection show that the introduction of performance evaluations has a direct impact on the turnover intentions of employees irrespective of whether evaluations are implemented within a broader set of HR practices or not. Groupspecific results In this subsection, we turn to the quantitative and qualitative discussion of the differences in the estimated coefficients between belowand abovemedian paid employees. Our main focus is on the specification with all control variables included (last specification from the left of Figure2). The full set of coefficients is shown in TableA6 in the Appendix. The results suggest that employees with a conditional relative rank position below the median reduce their turnover intentions significantly by around 0.2 points on a 1 to 5 scale when performance evaluations are implemented in the establishment. This effect corresponds to a quarter of the overall standard deviation or 13% of the mean value of turnover intentions (see Table1). In contrast, the effect on the turnover intentions of employees with a conditional relative rank position above the median is zero, suggesting that the implementation of performance evaluations has no impact for this group of employees. 18The results are not shown in the paper but are available on request. 19Please note that the introduction of personnel development plans is not directly asked in the survey and we only rely on information of establishments that indicate not making use of personnel development plans in one period and making use of it in the next period.
| 411 PERFORMANCE EVALUATIONS AND TURNOVER INTENTIONS The impact of performance evaluations is expected to be more pronounced if the potential gains from an internal labor market are higher. Therefore, we estimate the treatment effects for different establishment sizes as large establishments might offer more scope for career development. The results presented in FigureA5 in the Appendix suggest that the negative effect on turnover intentions for relatively lowpaid employees is indeed driven by large establishments: Individuals with a conditional relative rank position below the median reduce their turnover intentions by 0.4 points if performance evaluations are implemented. For abovemedian paid employees, we document zero effects independent of establishment size. In another specification, we additionally control for percentiles of the actual wage level of the employee and the dispersion of wages within establishments as a larger variance in wages might offer more scope for potential wage increases. However, the results in Figure A6 in the Appendix show that the estimated coefficients do not change compared to the baseline specification. Finally, FigureA7 in the Appendix shows the results by job level. The second level comprises jobs which require a secondary level vocational education (e.g., apprenticeship training), and the third and fourth levels are usually staffed with persons who hold a tertiary degree and are qualified to take management responsibilities. Our prior would be that employees working in jobs at the second level face more information asymmetries in the sense that they are less well informed about their relative rank position than employees of higher job levels and the effect of performance evaluations might hence be more pronounced for the former group. The results presented in FigureA7 confirm that the negative effects for relatively low paid are driven by employees of job level 2. FIGURE 3 Effect of performance evaluations on turnover intentions by median of relative pay and the introduction of complementary HR practices. Panel (a): Fixed effects specification; panel (b): First difference specification; outcome: Turnover intentions (1–5 scale); plotted coefficients: Interactions between conditional relative rank position and performance evaluations on the establishment level; covariates of baseline model: Baseline covariates: Partner, number of children, age dummies, leadership position, job strain, working hours (squared), job autonomy, multitasking, establishment size (squared); individual perfpay: Variable pay; HRhperf controls: Employee interviews, personnel development plans; HRlperf controls: Open discussion with low performers, personnel development of low performers, change position of low performers, dismissal of low performers, investment in higher qualifications, employee survey, promoting women in management positions, change in executive board, personnel management in the board of directors; intro perfpay: Performance pay; yearFE; covariates of model “only introduction employee assessment”: Baseline covariates, individual perfpay and yearFE; covariates of model “no introduction target agreements”: Same as baseline model; covariates of model “no introduction performance pay”: Same as baseline model excluding intro perfpay; covariates of model “no introduction personnel development plans”: Same as baseline model excluding personnel development plans. N = 5102–5403 (panel a) and 1579–1809 (panel b). Source: LPPADIAB 7519, own computations. [Colour figure can be viewed at wileyonlinelibrary.com] (a) (b)
412 | POHLAN and STEFFES In sum, we see that the introduction of performance evaluations has a negative impact on the turnover intentions of relatively lowpaid and no impact on relatively highpaid employees. Two considerations might explain the finding that employees who are paid below the median reduce their turnover intentions: First, they may reduce their overall career aspirations because they see lower chances to benefit from internal or external movements. This reasoning is supported by recent findings of Jäger etal.(2021). The authors compare workers' subjective outside options against objective measures of pay premia from matched employer–employee data and document that many workers mistakenly believe their current wage is representative of the external labor market. Hence, their results suggest that objectively lowpaid (highpaid) workers are overpessimistic (overoptimistic) about their outside options. Second, comparatively lowpaid employees could have an incentive to stay and to invest in their human capital to increase future career options. As performance evaluations are often accompanied by a discussion on prospects of career development, it is not unlikely that a supervisor offers such options. TableA9 in the Appendix further suggests that turnover intentions in the pretreatment period are exceptionally high among belowmedian paid employees in establishments where performance evaluations are introduced in the future compared to individuals in establishments where performance evaluations have already been introduced or will never be introduced. The differences are less pronounced for abovemedian paid employees. After the implementation of performance evaluations, belowmedian paid employees reach a comparable level of turnover intentions as individuals in establishments that have already or do not introduce such a system. Different triggers for implementing performance evaluations Firms implement individual performance evaluation systems for different reasons, which may be more or less related to actual firm performance and employee turnover intentions. For example, establishments facing internal problems might implement a performance evaluation system with the explicit intention to reduce employee turnover, while other firms do not. The employer survey also contains information on the triggers for introducing a performance evaluation system. TableA7 in the Appendix shows the responses, classified as external and internal triggers. We classify triggers as external in case they relate to innovations or external requirements and as internal in case they relate to internal experiences or problems. However, we do not make the claim here that concentrating on external triggers eliminates necessarily endogeneity bias, as innovations might to some degree be related to firm productivity and the establishments may endogenously decide not to introduce performance evaluations. The figures show that the most often mentioned trigger is new ideas of the management team, which was true in 54% of cases. Almost 14% mention external experience and less than 10% mention other external triggers such as external requirements (e.g., set by collective agreements or the parent company) or new technologies (e.g., new IT solutions). Turning to internal triggers for the introduction of performance evaluations, 28% of the establishments indicate that they already gained positive internal experience with performance evaluations, also 28% state internal problems (e.g., due to the economic situation or employee dissatisfaction), 14% that expectations have not been met and 10% state it was at the advice of a consultant. To analyze whether triggers for the introduction might be a potential driver of the results, we study the effects separately for those individuals who work in establishments where performance evaluations are introduced due to external and internal triggers, respectively. Figure4 shows that the effects are the same for both types of triggers: relatively lowpaid employees
| 413 PERFORMANCE EVALUATIONS AND TURNOVER INTENTIONS reduce their turnover intentions significantly and for relatively high paid we do not find a significant effect. The role of affective behavior So far, we have seen that performance evaluations have a negative impact on the turnover intentions of belowmedian paid employees and that the simultaneous implementation of measures to support employees is likely. Next, we investigate potential channels for this relationship. In the following, we analyze the impact of implementing performance evaluations on employee perceptions of supervisors, organizational support, and fairness. Beside the content of the evaluation and the resulting emotions, the 1to1 meeting and the individualized attention of the supervisor might improve the relationship and perceptions of the manager and encourage individuals to stay in the establishment.20 Therefore, we consider in the following analyses overall job satisfaction, perceptions of fairness and management quality as well as the evaluation of the promotion criteria in the establishment. The results in Table2 (panel A) suggest that individuals do not change their view of being disadvantaged in a personnel decision when performance evaluations are implemented. We find weak evidence that abovemedian paid employees (weakly significant at the 10% level) are more likely to feel supported by the management. However, we document a substantial increase by 0.27 points of a standard deviation in the job satisfaction of belowmedian paid employees. 20See the organizational psychology literature on the importance of 1to1 meetings (e.g., Castro etal.,2022; Kim,2020). FIGURE 4 Effect of performance evaluations on turnover intentions by median of relative pay and triggers for implementing performance evaluations. First difference specification; outcome: Turnover intentions (1–5 scale); plotted coefficients: Interactions between conditional relative rank position and performance evaluations on the establishment level for external triggers for implementing performance evaluations (panel a) and for internal triggers (panel b); external triggers comprise “new ideas of the management”, “external experience”, “external requirement”, “new technology” and internal triggers “positive internal experience”, “expectations have not been met”, “internal problems” and “advice of consultant”; baseline covariates: Partner, number of children, age dummies, leadership position, job strain, working hours (squared), job autonomy, multitasking, establishment size (squared); individual perfpay: Variable pay; HRhperf controls: Employee interviews, personnel development plans; HRlperf controls: Open discussion with low performers, personnel development of low performers, change position of low performers, dismissal of low performers, investment in higher qualifications, employee survey, promoting women in management positions, change in executive board, personnel management in the board of directors; intro perfpay: Performance pay; yearFE. N = 1609–1656 (panel a) and 1619–1666 (panel b). Source: LPPADIAB 7519, own computations. [Colour figure can be viewed at wileyonlinelibrary.com] (a) (b)
414 | POHLAN and STEFFES In a next step, we assess the impact of performance evaluations on the employee's evaluation of promotion criteria. Interestingly, employees paid above the median do not change their view. Instead, we observe a significant decrease in the assessment that promotion decisions are based on nonobjective criteria among the belowmedian paid employees. This finding is in line with the evidence that for establishments that introduce performance evaluations target agreements and personnel development plans play a significantly larger role. In particular, the introduction of personnel development plans could be more relevant for comparatively lowpaid employees and coincide with a structured and objective way of defining promotion criteria. TableA8 in the Appendix shows further analyses of affective behavior. Additional outcome variables include perceptions of fairness and treatments shown by the supervisors of the establishment but also on clarity and goal communication by the management. Although some coefficients are large and positive for both belowand abovemedian paid employees (e.g., on the perception of supervisors showing trust towards employees, good guidance of supervisors), we do not find any significant relationship. This result leads us to the conclusion that the introduction of performance evaluations matters for the turnover intentions of employees and part of the impact is explained by fairness perceptions regarding career opportunities, which is in line with the POS literature. However, we do not find strong evidence on a positive impact on the perception of TABLE 2 Channels. Below median Above median Panel A: Effect of performance evaluations (FE specification) on Job satisfaction 0.270** 0.098 Perception of Being disadvantaged in personnel decision −0.081 −0.016 Being supported by management 0.022 0.176* Promotion criteria being based on Expected skillfit for new position −0.163 0.110 Expected skillfit of old position −0.053 0.133 Nonobjective criteria −0.328** 0.012 Panel B: Effect of performance evaluations (FD specification) on Job satisfaction 0.303** 0.060 Perception of Being disadvantaged in personnel decision −0.095 −0.019 Being supported by management −0.017 0.148 Promotion criteria being based on Expected skillfit for new position −0.271* 0.090 Expected skillfit of old position 0.022 0.137 Nonobjective criteria −0.427*** −0.01 Note: Panel (A): Fixed effects (FE) specification; panel (B): First difference (FD) specification; each row represents the regression output of the respective outcome variable; outcome variables are standardized; covariates: Partner, number of children, age dummies, leadership position, job strain, working hours (squared), job autonomy, multitasking, establishment size (squared), variable pay, employee interviews, personnel development plans, open discussion with low performers, personnel development of low performers, change position of low performers, dismissal of low performers, investment in higher qualifications, employee survey, promoting women in management positions, change in executive board, personnel management in the board of directors, performance pay, yearFE. N = 5243–5483 (panel A) and N = 1747–1837 (panel B). ***/**/*Significant at the 1/5/10% level. Source: LPPADIAB 7519, own computations.
| 415 PERFORMANCE EVALUATIONS AND TURNOVER INTENTIONS supervisor and management quality which goes beyond the career planning of the individual employee. Actual job search and employer switches Besides information on turnover intentions, the LPP also includes information on actual job search.21 In our sample, 10% are actively looking for a new job. FigureA8 in the Appendix shows the effects of performance evaluations on the incidence of searching for a new job in dependence of the relative performance of the employees. In line with the results for turnover intentions, employees paid below the median decrease their job search activities significantly by around 5% points when performance evaluations are introduced, whereas higher paid employees do not change their job search behavior. In a last step, we look at the effects of the introduction of performance evaluations on actually observed employer switches in the social security data. We are able to follow individuals up to the end of the year 2019 and examine whether or not individuals change their employer in the last observed year and the following year of the survey wave. These are very rare events: we only observe a change for 2% of the sample. FigureA9 in the Appendix shows the results for belowand abovemedian paid employees. The fixed effects specification in panel (a) suggests that for relatively low paid the probability to switch the employer is slightly reduced but the effects are not significant and for relatively high paid the effect is close to zero. The effects are more positive but still insignificant when looking at the first difference specification (panel (b)).22 CONCLUSION In this article, we study the effects of the implementation of performance evaluations at the establishment level on employees' turnover intentions. We document that establishments that introduce performance evaluation systems are comparable to other firms with respect to most dimensions, such as firm size, sector composition, wage premia, performance pay systems, support in decision making or promotion criteria. However, we also find that establishments that have a problem with relatively high turnover among those paid below the median are more likely to introduce performance evaluations and other forms of employee involvement and development. In our empirical analysis, we study the effects of the overall management strategy to implement a performance evaluation system while controlling for many other changes in employment conditions at the establishment and employee level. Our findings suggest that the implementation of performance evaluations has heterogeneous effects on workers' intentions to leave depending on their relative rank. Employees paid below the median significantly decrease their turnover intentions and are more satisfied with their job, whereas employees paid above the median do not change their turnover intentions. The same pattern in heterogeneities is confirmed when looking at other outcomes of job search behavior such as actual job search and employer switches. Moreover, we find that relatively low paid are more satisfied with their 21We use a measure of actual job search based on the question “Have you been actively looking for another job during the last 12 months or have you been approached by another employer?” and the answers: 1 “yes, I have been actively searching”, 2 “yes, I have been approached by another employer”, 3 “both is true” and 4 “no, neither”. We generate a dummy variable which equals one for categories 1 to 3 and zero for category 4. 22 We observe a similar pattern when we look at employer switches conditional on surviving, i.e. we only study the effects for individuals in waves where no employer switching has occurred in previous waves. The results are shown in FigureA10 in the Appendix.
416 | POHLAN and STEFFES job and perceive promotion criteria as more objective after the introduction of performance evaluations. Our study delivers useful insights about the impact of performance evaluations which might be helpful for the design of management strategies to retain workers. Our findings suggest that firms that introduce a performance evaluation system are successful in retaining a specific group of individuals who do not belong to the topperformers and might be unsure about their future development options within the firm. Performance evaluations might therefore be a good tool that not only provides performance incentives to employees but also provides credible information about individually tailored career paths; consequently keeping employees from pursuing their chances elsewhere. This could be an efficient strategy against the background of skilledworker shortages and large hiring costs. Nevertheless, retaining relatively low performers could also change the composition of the work force and have a negative impact on the productivity of the firm in the longrun. The consequences for firmlevel outcomes would be an interesting direction for future research. ACKNOWLEDGMENTS We thank the Research Data Center of the IAB, especially Stefanie Wolter, for data support and the ZEW Mannheim for financial support. We further thank two anonymous reviewers, Mario Bossler, Florian Englmaier, Matthias Fahn, Sarah McNamara, Jens Mohrenweiser as well as participants at the VfS Annual Congress 2021, EALE Annual Congress 2021, BGPE MiniWorkshop, LPPWorkshop and seminar participants at the IAB and the University of Leeds for helpful comments and suggestions. Susanne Steffes worked on this article while she was employed at ZEW Mannheim. The opinions expressed in this article are the author's own and do not reflect the view of the German Federal Statistical Office. Open Access funding enabled and organized by Projekt DEAL. CONFLICT OF INTEREST STATEMENT The authors declare that they have no conflict of interest. DATA AVAILABILITY STATEMENT This paper uses survey and administrative data from the IAB Research Data Center (FDZ), see https:// fdz. iab. de/ en/ ourdataprodu cts/ integ rated - estab lishm entandindiv idual - data/ lpp/ . Since data access is restricted by the laws of the German Social Code, the data is not freely available to the research community. Members of the research community can apply for and access data products at the FDZ. Access can take place, e.g., via onsite use or via remote data access. More detailed information can be found here: https:// fdz. iab. de/ en/ dataaccess/ . ORCID Laura Pohlan https://orcid.org/0000-0001-8143-764X REFERENCES Abowd, J. M., F. Kramarz, and D. N. Margolis. 1999. “High Wage Workers and High Wage Firms.” Econometrica 67(2): 251–333. Aral, S., E. Brynjolfsson, and L. Wu. 2012. “ThreeWay Complementarities: Performance Pay, Human Resource Analytics, and Information Technology.” Management Science 58(5): 913–31. Azmat, G., and N. Iriberri. 2016. “The Provision of Relative Performance Feedback: An Analysis of Performance and Satisfaction.” Journal of Economics and Management Strategy 25(1): 77–110. Baker, G., R. Gibbons, and K. J. Murphy. 1994. “Subjective Performance Measures in Optimal Incentive Contracts.” The Quarterly Journal of Economics 109(4): 1125–56. Bakker, A. B., and E. Demerouti. 2007. “The Job DemandsResources Model: State of the Art.” Journal of Managerial Psychology 22(3): 309–28.
| 417 PERFORMANCE EVALUATIONS AND TURNOVER INTENTIONS Beenen, G., S. Pichler, and P. E. Levy. 2017. “SelfDetermined Feedback Seeking: The Role of Perceived Supervisory Autonomy Support.” Human Resource Management 56(4): 555–69. Bellmann, L., B. Lochner, S. Seth, and S. Wolter. 2020. “AKM Effects for German Labour Market Data.” FDZMethodenreport 01/2020. Cappelli, P., and M. J. Conyon. 2018. “What do Performance Appraisals do?” ILR Review 71(1): 88–116. Castro, S., F. Englmaier, and M. Guadalupe. 2022. “Fostering Psychological Safety in Teams: Evidence From an Rct.” Working Paper Chan, J., and B. Zheng. 2011. “Rewarding Improvements: Optimal Dynamic Contracts with Subjective Evaluation.” The Rand Journal of Economics 42(4): 758–75. Clark, A. E. 2001. “What Really Matters in a Job? Hedonic Measurement Using Quit Data.” Labour Economics 8(2): 223–42. Clark, A. E., N. Kristensen, and N. WestergårdNielsen. 2009. “Job Satisfaction and CoWorker Wages: Status or Signal?” The Economic Journal 119(536): 430–47. Cottini, E., T. Kato, and N. WestergaardNielsen. 2011. “Adverse Workplace Conditions, HighInvolvement Work Practices and Labor Turnover: Evidence from Danish Linked Employer–Employee Data.” Labour Economics 18(6): 872–80. Dube, A., E. Freeman, and M. Reich. 2010. “Employee Replacement Costs.” Working Paper Eisenberger, R., R. Huntington, S. Hutchison, and D. Sowa. 1986. “Perceived Organizational Support.” Journal of Applied Psychology 71(3): 500. Eisenberger, R., L. Rhoades Shanock, and X. Wen. 2020. “Perceived Organizational Support: Why Caring about Employees Counts.” Annual Review of Organizational Psychology and Organizational Behavior 7: 101–24. Engellandt, A., and R. T. Riphahn. 2011. “Evidence on Incentive Effects of Subjective Performance Evaluations.” ILR Review 64(2): 241–57. Frederiksen, A. 2017. “Job Satisfaction and Employee Turnover: A FirmLevel Perspective.” German Journal of Human Resource Management 31(2): 132–61. Frederiksen, A., F. Lange, and B. Kriechel. 2017. “Subjective Performance Evaluations and Employee Careers.” Journal of Economic Behavior & Organization 134: 408–29. Gill, D., Z. Kissová, J. Lee, and V. Prowse. 2019. “FirstPlace Loving and LastPlace Loathing: How Rank in the Distribution of Performance Affects Effort Provision.” Management Science 65(2): 494–507. Giuliano, L., D. I. Levine, and J. Leonard. 2011. “Racial Bias in the ManagerEmployee Relationship an Analysis of Quits, Dismissals, and Promotions at a Large Retail Firm.” Journal of Human Resources 46(1): 26–52. Godechot, O., and C. Senik. 2015. “Wage Comparisons In and Out of the Firm. Evidence from a Matched Employer– Employee French Database.” Journal of Economic Behavior & Organization 117: 395– 410. Hoffman, M., and S. V. Burks. 2020. “Worker Overconfidence: Field Evidence and Implications for Employee Turnover and Firm Profits.” Quantitative Economics 11(1): 315–48. Huffman, D., C. Raymond, and J. Shvets. 2022. “Persistent Overconfidence and Biased Memory: Evidence from Managers.” American Economic Review 112(10): 3141–75. Ichniowski, C., K. Shaw, and G. Prennushi. 1997. “The Effects of Human Resource Practices on Manufacturing Performance: A Study of Steel Finishing Lines.” American Economic Review 87(3): 291–313. Jäger, S., C. Roth, N. Roussille, and B. Schoefer. 2021. “Worker Beliefs About Rents and Outside Options.” Working Paper Kampkötter, P. 2017. “Performance Appraisals and Job Satisfaction.” The International Journal of Human Resource Management 28(5): 750–74. Kampkötter, P., J. Mohrenweiser, D. Sliwka, S. Steffes, and S. Wolter. 2016. “Measuring the Use of Human Resources Practices and Employee Attitudes: The Linked Personnel Panel.” EvidenceBased HRM 4(2): 94–115. Kampkötter, P., and D. Sliwka. 2016. “The Complementary Use of Experiments and Field Data to Evaluate Management Practices: The Case of Subjective Performance Evaluations.” Journal of Institutional and Theoretical Economics 172(2): 364–89. Kim, N. Y. 2020. “Linking Individuation and Organizational Identification: Mediation through Psychological Safety.” The Journal of Social Psychology 160(2): 216–35. Kuhnen, C. M., and A. Tymula. 2012. “Feedback, SelfEsteem, and Performance in Organizations.” Management Science 58(1): 94–113. Kurtessis, J. N., R. Eisenberger, M. T. Ford, L. C. Buffardi, K. A. Stewart, and C. S. Adis. 2017. “Perceived Organizational Support: A MetaAnalytic Evaluation of Organizational Support Theory.” Journal of Management 43(6): 1854–84. Lazear, E. P. 2000. “Performance Pay and Productivity.” American Economic Review 90(5): 1346–61. Lazear, E. P. 2018. “Compensation and Incentives in the Workplace.” Journal of Economic Perspectives 32(3): 195–214. Mackeben, J., K. Ruf, S. Wolter, and P. Grunau. 2021. “LPP Survey Data Linked with Administrative Data of the IAB (LPPAdiab) 1975–2019.” FDZData Report (10/2021 (en))
418 | POHLAN and STEFFES MacLeod, W. B. 2003. “Optimal Contracting with Subjective Evaluation.” American Economic Review 93(1): 216–40. Matthes, B., H. Meinken, and P. Neuhauser. 2015. “Berufssektoren und Berufssegmente auf Grundlage der KldB2010.” Methodenbericht der Statistik der BA Mohrenweiser, J., and C. Pfeifer. 2023. “Reference Wages and Turnover Intentions: Evidence from Linked EmployerEmployee Data.” Applied Economics Letters 30(14): 1955–9. Muehlemann, S., and M. S. Leiser. 2018. “Hiring Costs and Labor Market Tightness.” Labour Economics 52: 122–31. Paul, O., and S. Scott. 2011. “Personnel Economics: Hiring and Incentives.” In Handbook of Labor Economics, Vol 4 1769–1823. Amsterdam: Elsevier. Pfeifer, C., and S. Schneck. 2012. “Relative Wage Positions and Quit Behavior: Evidence from Linked EmployerEmployee Data.” ILR Review 65(1): 126–47. Takahashi, S., H. Owan, T. Tsuru, and K. Uehara. 2021. “Multitasking Incentives and the Informative Value of Subjective Performance Evaluations.” ILR Review 74(2): 511–43. Van Praag, B., and A. Ferreri Carbonell. 2007. Happiness Quantified: A Satisfaction Calculus Approach. Oxford: OUP Oxford. How to cite this article: Pohlan, Laura and Susanne Steffes. 2025. “Performance Evaluations and Employee Turnover Intentions: Empirical Evidence From Linked Employer–Employee Data.” Industrial Relations: A Journal of Economy and Society 64 (3): 395–433. https://doi.org/10.1111/irel.12379. APPENDIX A.1 TREATMENT DEFINITION FIGURE A1 Definition of performance evaluations. FD, First difference; FE, Fixed effects. Source: Own presentation. [Colour figure can be viewed at wileyonlinelibrary.com]
| 425 PERFORMANCE EVALUATIONS AND TURNOVER INTENTIONS Implementation of performance evaluations in future No Yes pvalue (1) (2) (1) – (2) Promoting women in management positions 0.406 0.491 0.035 Change in executive board 0.270 0.311 0.269 Personnel management in the board of directors 0.437 0.447 0.810 Performance pay 0.698 0.671 0.467 Wage measures 0.066 0.043 0.264 Wage pct 0–5 Wage pct 5–25 0.221 0.292 0.038 Wage pct 25–50 0.317 0.217 0.009 Wage pct 50–75 0.252 0.342 0.012 Wage pct 75–95 0.129 0.093 0.183 Wage pct >95 0.015 0.012 0.808 Difference establishment wage max – pct 75 45.527 38.574 0.005 Difference establishment wage pct 75 – pct 50 21.945 20.454 0.129 Difference establishment wage pct 50 – pct 25 16.009 14.600 0.088 Difference establishment wage pct 25 – min 61.529 67.923 0.012 Number of individuals 2094 161 Note: Descriptive statistics are based on the first observation of individuals in the sample. Abbreviations: pct, percentile; UAS, university of applied science. Source: LPPADIAB 7519, own computations. TABLE A5 (Continued) FIGURE A2 Overall effect of performance evaluations on turnover intentions, different model specifications. Estimations from linear regression and probitadapted ordinary least squares (POLS); panel (a): Fixed effects specification; panel (b): First difference specification; outcomes: Turnover intentions (1–5 scale); plotted coefficients: Effect of performance evaluations on the establishment level; covariates: Partner, number of children, age dummies, leadership position, job strain, working hours (squared), job autonomy, multitasking, establishment size (squared), variable pay, yearFE. N = 5491 (panel a) and 1854 (panel b). Source: LPPADIAB 7519, own computations. [Colour figure can be viewed at wileyonlinelibrary.com] (a) (b)
426 | POHLAN and STEFFES FIGURE A3 Effect of performance evaluations on turnover intentions by quartiles of relative pay. Panel (a): Fixed effects specification; panel (b): First difference specification; outcomes: Turnover intentions (1–5 scale) and indicator for turnover intentions; plotted coefficients: Interactions between quartile of conditional relative rank position and performance evaluations on the establishment level; covariates: Partner, number of children, age dummies, leadership position, job strain, working hours (squared), job autonomy, multitasking, establishment size (squared), variable pay, yearFE. N = 5436 (panel a) and 1841 (panel b). Source: LPPADIAB 7519, own computations. [Colour figure can be viewed at wileyonlinelibrary.com] (a) (b)
| 427 PERFORMANCE EVALUATIONS AND TURNOVER INTENTIONS TABLE A6 Baseline specification: Effect of performance evaluations on turnover intentions divided by median of relative pay. (1) (2) FE specification FD specification PE interacted paid below median −0.194** −0.218** (0.088) (0.094) PE interacted paid above median −0.023 −0.030 (0.079) (0.080) Baseline covariates Partner −0.051 −0.146 (0.077) (0.136) Number of children 0.012 0.011 (0.028) (0.047) Age 30–39 0.069 0.081 (0.134) (0.201) Age 40–49 0.194 0.237 (0.159) (0.245) Age ≥50 0.234 0.301 (0.167) (0.262) Leadership position −0.003 −0.007 (0.049) (0.101) Job strain 0.021 0.069*** (0.013) (0.023) Working hours −0.006 −0.021 (0.008) (0.015) Working hours squared 0.000 0.000 (0.000) (0.000) Job autonomy −0.071*** −0.052** (0.016) (0.026) Multitasking 0.003 0.012 (0.017) (0.033) Establishment size −0.000 −0.001* (0.000) (0.001) Establishment size squared 0.000 0.000* (0.000) (0.000) Additional covariates Variable pay −0.084*** −0.150*** (0.029) (0.056) Employee interviews −0.011 −0.100** (0.039) (0.049) Personnel development plans 0.009 0.054 (0.034) (0.055) (Continues)
428 | POHLAN and STEFFES (1) (2) FE specification FD specification Open discussion with low performers 0.007 0.002 (0.013) (0.022) Personnel development of low performers 0.005 −0.005 (0.013) (0.020) Change position of low performers −0.006 −0.001 (0.011) (0.019) Dismissal of low performers −0.006 −0.019 (0.010) (0.017) Investment in higher qualifications 0.004 0.105 (0.035) (0.068) Employee survey −0.027 0.108 (0.035) (0.068) Promoting women in management positions 0.034 0.107* (0.028) (0.057) Change in executive board 0.013 0.019 (0.023) (0.040) Personnel management in the board of directors 0.025 0.001 (0.026) (0.047) Performance pay 0.113 0.149 (0.078) (0.125) Year 2014 −0.033 −0.051 (0.022) (0.039) Year 2016 −0.047 −0.085 (0.031) (0.059) Year 2018 −0.088** −0.199*** (0.042) (0.068) Constant 1.963*** 2.284*** (0.316) (0.552) Number of observations 5403 1809 Rsquared 0.019 0.059 Note: Outcome: turnover intentions (1–5 scale). Abbreviations: FD, first difference; FE, fixed effects; PE, performance evaluations. ***/**/* significant at the 1/5/10% level. Source: LPPADIAB 7519, own computations. TABLE A6 (Continued)
| 429 PERFORMANCE EVALUATIONS AND TURNOVER INTENTIONS FIGURE A4 Effect of performance evaluations on turnover intentions by median of relative pay, different handling of individuals earning wages above the ceiling of the social security contributions. Exclusion of establishmentoccupationjob level cells if more than 50% of the individuals in the cell have predicted wages above the ceiling of the social security contributions. Panel (a): Fixed effects specification; panel (b): First difference specification; outcome: Turnover intentions (1–5 scale); plotted coefficients: Interactions between conditional relative rank position and performance evaluations on the establishment level; baseline covariates: Partner, number of children, age dummies, leadership position, job strain, working hours (squared), job autonomy, multitasking, establishment size (squared); individual perfpay: Variable pay; HRhperf controls: Employee interviews, personnel development plans; HRlperf controls: Open discussion with low performers, personnel development of low performers, change position of low performers, dismissal of low performers, investment in higher qualifications, employee survey, promoting women in management positions, change in executive board, personnel management in the board of directors; intro perfpay: Performance pay; yearFE; wage ranks: Percentiles of the actual wage level; wage variance: Difference of quartiles of the wage distribution within establishments. N = 5332–5428 (panel a) and 1760–1808 (panel b). Source: LPPADIAB 7519, own computations. [Colour figure can be viewed at wileyonlinelibrary.com] (a) (b) FIGURE A5 Effect of performance evaluations on turnover intentions by median of relative pay and establishment size. Fixed effect specification; outcome: Turnover intentions (1–5 scale); plotted coefficients: Interactions between conditional relative rank position and performance evaluations on the establishment level for establishment size <250 employees (panel a) and establishment sizes ≥250 employees (panel b); baseline covariates: Partner, number of children, age dummies, leadership position, job strain, working hours (squared), job autonomy, multitasking, establishment size (squared); individual perfpay: Variable pay; HRhperf controls: Employee interviews, personnel development plans; HRlperf controls: Open discussion with low performers, personnel development of low performers, change position of low performers, dismissal of low performers, investment in higher qualifications, employee survey, promoting women in management positions, change in executive board, personnel management in the board of directors; intro perfpay: Performance pay; yearFE. N = 2880–2927 (panel a) and 2523–2573 (panel b). Source: LPPADIAB 7519, own computations. [Colour figure can be viewed at wileyonlinelibrary.com] (a) (b)
430 | POHLAN and STEFFES FIGURE A6 Effect of performance evaluations on turnover intentions by median of relative pay, controlling for administrative wage information. Panel (a): Fixed effects specification; panel (b): First difference specification; outcome: Turnover intentions (1–5 scale); plotted coefficients: Interactions between conditional relative rank position and performance evaluations on the establishment level; baseline covariates: Partner, number of children, age dummies, leadership position, job strain, working hours (squared), job autonomy, multitasking, establishment size (squared); individual perfpay: Variable pay; HRhperf controls: Employee interviews, personnel development plans; HRlperf controls: Open discussion with low performers, personnel development of low performers, change position of low performers, dismissal of low performers, investment in higher qualifications, employee survey, promoting women in management positions, change in executive board, personnel management in the board of directors; intro perfpay: Performance pay; yearFE; wage ranks: Percentiles of the actual wage level; wage variance: Difference of quartiles of the wage distribution within establishments. N = 5403–5500 (panel a) and 1809–1857 (panel b). Source: LPPADIAB 7519, own computations. [Colour figure can be viewed at wileyonlinelibrary.com] (a) (b) FIGURE A7 Effect of performance evaluations on turnover intentions by median of relative pay and job levels. Fixed effect specification; outcome: Turnover intentions (1–5 scale); plotted coefficients: Interactions between conditional relative rank position and performance evaluations on the establishment level for job level 2 (panel a) and job level 3 or 4 (panel b); baseline covariates: Partner, number of children, age dummies, leadership position, job strain, working hours (squared), job autonomy, multitasking, establishment size (squared); individual perfpay: Variable pay; HRhperf controls: Employee interviews, personnel development plans; HRlperf controls: Open discussion with low performers, personnel development of low performers, change position of low performers, dismissal of low performers, investment in higher qualifications, employee survey, promoting women in management positions, change in executive board, personnel management in the board of directors; intro perfpay: Performance pay; yearFE. N = 3732–3800 (panel a) and 1671–1700 (panel b). Source: LPPADIAB 7519, own computations. [Colour figure can be viewed at wileyonlinelibrary.com] (a) (b)
| 431 PERFORMANCE EVALUATIONS AND TURNOVER INTENTIONS TABLE A7 Triggers for implementing performance evaluations. External triggers New ideas of the management 0.542 External experience 0.139 External requirement 0.097 New technology 0.083 Internal triggers Positive internal experience 0.278 Expectations have not been met 0.139 Internal problems 0.278 Advice of consultant 0.097 Note: Multiple answers possible. N = 72. Source: LPPADIAB 7519, own computations. TABLE A8 Channels II. Below median Above median Panel A: Effect of performance evaluations (FE specification) on Perception of Being fairly treated by supervisor 0.102 0.206* Supervisors showing trust towards employees 0.134 0.179 Good guidance of supervisors 0.077 0.101 Supervisors openly explaining dissatisfaction −0.159 −0.015 With employees’ performance Clear communication of requirements and goals 0.043 0.038 Panel B: Effect of performance evaluations (FD specification) on Perception of Being fairly treated by supervisor 0.062 0.172 Supervisors showing trust towards employees 0.007 0.222* Good guidance of supervisors 0.073 0.098 Supervisors openly explaining dissatisfaction −0.153 0.052 With employees’ performance Clear communication of requirements and goals 0.02 0.03 Note: Panel (A): Fixed effects (FE) specification; panel (B): First difference (FD) specification; each row represents the regression output of the respective outcome variable; outcome variables are standardized; covariates: Partner, number of children, age dummies, leadership position, job strain, working hours (squared), job autonomy, multitasking, establishment size (squared), variable pay, employee interviews, personnel development plans, open discussion with low performers, personnel development of low performers, change position of low performers, dismissal of low performers, investment in higher qualifications, employee survey, promoting women in management positions, change in executive board, personnel management in the board of directors, performance pay, yearFE. N = 4461–5480 (panel A) and N = 1543–1837 (panel B). ***/**/* significant at the 1/5/10% level. Source: LPPADIAB 7519, own computations.
432 | POHLAN and STEFFES TABLE A9 Turnover intentions by treatment. Below median Above median Treatment group before treatment 1.701 1.405 Treatment group after treatment 1.481 1.452 Comparison group without performance evaluations in t01.511 1.573 Comparison group without performance evaluations in t11.508 1.519 Comparison group with performance evaluations in t01.486 1.445 Comparison group with performance evaluations in t11.537 1.455 Note: The table shows mean values of turnover intentions for individuals in establishments where performance evaluations are implemented (treatment group) and individuals where no introduction takes place (comparison group). Individuals from the comparison group are further distinguished by whether the establishment has already a performance evaluation system or not. The level of turnover intentions is shown in the wave before and after treatment for individuals of the treatment group and in the first (t0) and second (t1) observed wave for individuals of the comparison group. N = 77–84 (treated), 266–337 (controls without performance evaluations) and 689–802 (controls with performance evaluations). Source: LPPADIAB 7519, own computations. FIGURE A8 Effect of performance evaluations on actual job search by median of relative pay. Panel (a): Fixed effects specification; panel (b): First difference specification; outcome: Indicator for actual job search; plotted coefficients: Interactions between conditional relative rank position and performance evaluations on the establishment level; baseline covariates: Partner, number of children, age dummies, leadership position, job strain, working hours (squared), job autonomy, multitasking, establishment size (squared); individual perfpay: Variable pay; HRhperf controls: Employee interviews, personnel development plans; HRlperf controls: Open discussion with low performers, personnel development of low performers, change position of low performers, dismissal of low performers, investment in higher qualifications, employee survey, promoting women in management positions, change in executive board, personnel management in the board of directors; intro perfpay: Performance pay; yearFE. N = 4537–4613 (panel a) and 1469–1509 (panel b). Source: LPPADIAB 7519, own computations. [Colour figure can be viewed at wileyonlinelibrary.com] (a) (b)
| 433 PERFORMANCE EVALUATIONS AND TURNOVER INTENTIONS FIGURE A10 Effect of performance evaluations on employer switches by median of relative pay, conditional on surviving. Panel (a): Fixed effects specification; panel (b): First difference specification; outcome: Jobtojob transitions conditional on no switching in prior years; plotted coefficients: Interactions between conditional relative rank position and performance evaluations on the establishment level; baseline covariates: Partner, number of children, age dummies, leadership position, job strain, working hours (squared), job autonomy, multitasking, establishment size (squared); individual perfpay: Variable pay; HRhperf controls: Employee interviews, personnel development plans; HRlperf controls: Open discussion with low performers, personnel development of low performers, change position of low performers, dismissal of low performers, investment in higher qualifications, employee survey, promoting women in management positions, change in executive board, personnel management in the board of directors; intro perfpay: Performance pay; yearFE. N = 3891–3966 (panel a) and 1335–1375 (panel b). Source: LPPADIAB 7519, own computations. [Colour figure can be viewed at wileyonlinelibrary.com] (a) (b) FIGURE A9 Effect of performance evaluations on employer switches by median of relative pay. Panel (a): Fixed effects specification; panel (b): First difference specification; outcome: Jobtojob transitions; plotted coefficients: Interactions between conditional relative rank position and performance evaluations on the establishment level; baseline covariates: Partner, number of children, age dummies, leadership position, job strain, working hours (squared), job autonomy, multitasking, establishment size (squared); individual perfpay: Variable pay; HRhperf controls: Employee interviews, personnel development plans; HRlperf controls: Open discussion with low performers, personnel development of low performers, change position of low performers, dismissal of low performers, investment in higher qualifications, employee survey, promoting women in management positions, change in executive board, personnel management in the board of directors; intro perfpay: Performance pay; yearFE. N = 5486–5585 (panel a) and 1837–1887 (panel b). Source: LPPADIAB 7519, own computations. [Colour figure can be viewed at wileyonlinelibrary.com] (a) (b)