Climbing the Career Ladder Does Not Make You Happy: Well-being Changes in the Years Before and After Becoming a Leader

Subjective well-being tends to be higher in leaders vs. non-leaders. However, do these differences come from selection effects (e.g., because higher subjective well-being predisposes for occupational success) or from within-person well-being changes before and after becoming a leader? This question remains largely unresolved. Previous research suggests that becoming a leader might be a double-edged sword and affect subjective well-being positively but also negatively (e.g., due to more power but also more stress). Using data from the German Socio-Economic Panel Study (N = 25,674), we examined (a) well-being differences between employees who did vs. did not start a leadership position and (b) well-being changes before and after becoming a leader. Compared to non-leaders, leaders were more satisfied with their lives, happier, and less sad in the years before and after starting a leadership position. Leaders became slightly more satisfied with their lives in the five years before and five years after becoming a leader. Happiness, sadness, and anxiety did not change, but anger increased after starting a leadership position. These findings support the idea that differences in subjective well-being between leaders and non-leaders largely stem from selection effects, while starting a leadership position might even lower specific well-being facets.


Introduction
Previous research stresses the importance of leaders' subjective well-being not only for their own health but also for their employees' well-being and performance (Arnold, 2017;Byrne et al., 2014;Courtright et al., 2014;Harms et al., 2017;Inceoglu et al., 2018;Kaluza et al., 2020;Montano et al., 2017). For example, higher subjective well-being in leaders has been linked to more effective leadership (Byrne et al., 2014;Courtright et al., 2014;Kaluza et al., 2020), which in turn has been associated with more favorable health-and work-related outcomes among employees (Arnold, 2017;Harms et al., 2017;Inceoglu et al., 2018;Montano et al., 2017).
However, surprisingly little is known about changes in subjective well-being in the years before and after becoming a leader. Starting a leadership position often relates to benefits, including a higher status, income, power, and control. At the same time, being a leader typically relates to higher responsibilities, more work, higher time pressure, supervisory tasks, and representative functions, which can be stressful. Therefore, starting a leadership position might have favorable but -at the same time -also unfavorable effects on subjective well-being (Barling & Cloutier, 2017;Debus et al., 2019;Li et al., 2018).
Consistent with this idea, becoming a leader has been linked to higher stress (e.g., higher exhaustion) but also to higher job satisfaction (Debus et al., 2019). However, little is known about changes in general life satisfaction and different facets of affect (e.g., happiness, sadness, anxiety, and anger) before and after transitioning into a leadership role. Improved knowledge on this topic is crucial from a basic and applied perspective: From a basic perspective, it helps to resolve whether Set-Point Theory (Lucas, 2007) applies not only to major life events but also to major occupational transitions: Are well-being changes in emergent leaders transient or do they last for a long time? From an applied perspective, it helps to understand the mental health challenges that come along with being a leader, providing target points for interventional research and personnel development (Kaluza et al., 2020;Roche et al., 2014). For example, promoting specific well-being facets in (emergent) leaders might enhance leadership success, occupational health, and organizational growth. This study focuses on (a) well-being differences between leaders and non-leaders as well as (b) well-being changes before and after starting a leadership position.

Subjective Well-being
Subjective well-being is defined as subjective evaluation of one's life and comprises a cognitive and an affective component (Diener et al., 2009). Cognitive well-being refers to life satisfaction and affective well-being refers to positive and negative affect. Life satisfaction, positive affect, and negative affect have been shown to be moderately correlated but clearly separable dimensions (Diener et al., 2009). Moreover, affective well-being can be divided into more nuanced sub-facets such as happiness, sadness, anxiety, and anger (Möwisch et al., 2019).

Leadership Differences Between Leaders and Non-leaders
Previous research indicates that subjective well-being tends to be higher in leaders vs. employees in non-leadership positions (Jurkiewicz & Massey, 1997;Li et al., 2018;Sherman et al., 2012;Skakon et al., 2011). For instance, Li and colleagues (2018) compared psychological and physiological well-being indices between leaders and non-leaders from different samples and found that leaders tended to be more satisfied with their lives compared to non-leaders. (Skakon et al., 2011) found that leaders experienced lower emotional stress than non-leaders, but cognitive, behavioral, and physiological stress indices did not differ significantly between both groups. Sherman and colleagues (2012) showed that levels of anxiety and cortisol were lower in leaders vs. non-leaders and in leaders with more vs. less powerful positions. These effects were partially due to differences in perceived control. More broadly, a higher status and more power in social hierarchies in and outside occupational contexts have been linked to more positive and less negative affect (van Kleef & Lange, 2020;Witkower et al., 2020), lower stress, and lower cortisol secretion (Sherman & Mehta, 2020).

Theoretical Assumptions
Research on person-situation transactions (Buss, 1987;Caspi & Roberts, 2001) and personenvironment fit (Caplan, 1987;Rauthmann & Sherman, 2016) indicates that psychological characteristics and work-related experiences influence each other over time (Denissen et al., 2014;Nye & Roberts, 2019). It is thus plausible to assume that leadership emergence relates to (changes in) subjective well-being: Subjective well-being might affect whether people do or do not become leaders (selection effects). At the same time, subjective well-being might change in preparation for and in reaction to a leadership role.

Well-being Changes Before and After Becoming a Leader
According to the Job Demand-Control Model (Karasek, 1979), two job characteristics are essential for subjective well-being among employees: Job demands and job control. Job demands include, for example, high responsibilities, a heavy workload, time pressure, role conflicts, as well as physical and emotional demands. Job control refers to the ability to influence tasks and activities at work, as indicated, for example, by high autonomy and decision-making authority. The model assumes that higher job demands relate to lower well-being, while higher job control relates to higher well-being, which has been supported by previous research (Crawford et al., 2010;Gonzalez-Mulé et al., 2020;Häusser et al., 2010;Nixon et al., 2011;Bosma et al., 1997).
Leadership positions are typically characterized by high job demands (e.g., high responsibilities) and at the same time high job control (e.g., high decision-making authority; Barling & Cloutier, 2017;Li et al., 2018). Therefore, becoming a leader might promote but also hamper subjective well-being.
In their Dual-Pathway Model, Li and colleagues (2018) assume that being a leader relates to higher job demands, which in turn relates to lower subjective well-being. At the same time, leadership relates to higher job control, which in turn relates to higher subjective well-being. In two cross-sectional studies, the authors compared job demands, job control, and different well-being indices between employees in leadership and non-leadership positions. Consistent with their ideas, leadership had an indirect negative effect on subjective well-being (e.g., life satisfaction) through higher job demands. At the same time, leadership had an indirect positive effect on subjective well-being through higher job control. Moreover, longitudinal analyses revealed that job demands and job control were initially higher and increased more strongly in employees who did vs. did not start a leadership position over time (Li et al., 2018). However, longitudinal changes in subjective well-being in the years before and after becoming a leader were not examined.
Similarly, another longitudinal study found that starting a leadership position was indirectly related to higher exhaustion and work-to-family conflict through higher time pressure, an indicator of higher job demands (Debus et al., 2019). At the same time, becoming a leader was indirectly related to higher job satisfaction through higher participation in decision-making, an indicator of higher job control. However, whether the transition related to opposite changes in general life satisfaction and/or different emotions was not assessed.
Taken together, none of these previous studies focused on nuanced well-being changes in the years before becoming a leader (anticipation effects) and in the years after this transition (socialization effects). Thus, additional research is needed to resolve how life satisfaction and different facets of affect change at different junctions before, during, and after transitioning into a leadership role. Set-Point Theory (Lucas, 2007) assumes that subjective wellbeing fluctuates around a person-specific set-point over time. That is, subjective well-being might temporarily change due to positive or negative life experiences, but it bounces back to its set-point in the long term. In line with these ideas, leaders might experience particularly pronounced well-being changes in the first year of being a leader, which attenuate in the following years.

Gender Differences
To date, it has not been resolved whether well-being changes in the years before and after becoming a leader vary by gender. Role Congruity Theory posits that female (vs. male) gender stereotypes are less congruent with leadership roles, leading to less positive perceptions and evaluations of female leaders (Eagly & Karau, 2002). Consistently, previous research found that female and male leaders not only differed in their leadership behavior (Eagly & Johnson, 1990), but were also perceived and evaluated differently (Bass & Bass, 2009;Eagly et al., 1992). For example, women tended to lead in a more participative but less directive way (Eagly & Johnson, 1990) and were devalued more often when adopting a stereotypically masculine (e.g., directive) leadership style (Eagly et al., 1992). Thus, becoming a leader might be more difficult and stressful (Eagly & Karau, 2002) and thus relate to less favorable well-being changes in women vs. men.

Age Differences
Moreover, little is known about age differences: On the one hand, younger (vs. older) individuals might be more flexible and energetic, more easily adjust to work-related changes, and thus experience more positive well-being changes when starting a leadership position (Walter & Scheibe, 2013).
On the other hand, older people tend to have more knowledge, skills, experience, and occupational network contacts, which facilitates transitioning into a leadership role (Seibert et al., 2017;Walter & Scheibe, 2013). Thus, it is also plausible to argue that older individuals experience more positive well-being changes before and after becoming a leader. Aims.
Using data from the German Socio-Economic Panel (SOEP), this study focused on (a) well-being differences between leaders and employees in non-leadership positions as well as (b) well-being changes in the years before and after becoming a leader, including the role of gender and age. In the SOEP, not only cognitive well-being (i.e., life satisfaction) but also four facets of positive (i.e., happiness) and negative (i.e., sadness, anxiety, and anger) affect were assessed and considered in the analyses.
In the total sample (including leaders and non-leaders), we analyzed selection effects to investigate well-being differences between non-leaders and leaders-to-be (in the years before becoming a leader). We modeled post-transition differences to examine well-being differences between non-leaders and leaders (in the years after becoming a leader).
In leaders, we analyzed anticipation effects to study gradual well-being changes in the five years before starting a leadership position and socialization effects to study gradual well-being changes in the five years after starting a leadership position. We analyzed shortterm effects to examine well-being differences in the first year of being a leader vs. all other years. Finally, we modeled long-term effects to investigate well-being differences more than one year after starting a leadership position vs. the years before.
Furthermore, we explored changes in job characteristics in the years before and after becoming a leader: Weekly working hours were considered as an indicator of job demands. Leaders' monthly gross labor income was considered as a potential benefit because it typically increases through promotions and has been partially linked to subjective well-being, especially life satisfaction (although this link tends to decrease beyond a certain income level; Kahneman & Deaton 2010). (Job control was not assessed in the SOEP and thus could not be considered.) Hypotheses.
Based on previous findings on well-being differences between leaders and non-leaders (Jurkiewicz & Massey, 1997;Li et al., 2018;Sherman et al., 2012;Skakon et al., 2011), we hypothesize that leaders are more satisfied with their lives, happier, less sad, less anxious, and less angry than non-leaders, both in the years before becoming a leader (selection hypothesis) and in the years after this transition (post-transition hypothesis). Similarly, we assume that both female and male leaders are more satisfied, happier, less sad, less anxious, and less angry than same-sex non-leaders, respectively (gender hypothesis 1).
In line with the Job Demand-Control Model (Karasek, 1979) and Dual-Pathway Model (Li et al., 2018), we assume that becoming a leader relates to increased life satisfaction and positive affect (due to the benefits of a leadership position) but also to increased negative affect (due to the costs of being a leader). In line with Set-Point Theory (Lucas, 2007), we expect that these changes are most pronounced shortly after the transition and attenuate in the long run, leading to the following hypotheses: All well-being facets (i.e., life satisfaction, happiness, sadness, anxiety, and anger) increase in the years before becoming a leader (anticipation hypothesis). Life satisfaction, happiness, sadness, anxiety, and anger are higher in the first year (short-term hypothesis) and after the first year (long-term hypothesis) of being a leader (vs. all other years), but these effects attenuate over time (socialization hypothesis). In women, becoming a leader relates to less favorable well-being changes compared to men (e.g., a smaller increase in life satisfaction and happiness but a larger increase in sadness, anxiety, and anger; gender hypothesis 2). Interactions with age are tested exploratorily.

Study Sample
We used data from the German Socio-Economic Panel Study (SOEP), a nationally representative household panel study from Germany with multistage probability sampling. The SOEP started in 1984 and is still ongoing. Data are collected yearly and mostly stem from face-to-face interviews with all adult members of the respective target household. The initial sample from 1984 was regularly replenished with new participants to counteract attrition, to increase the sample size, and to allow for specific sub-group analyses. Therefore, panel members entered the study in different years. For the analyses, this means that not all individuals provided information on their employment status and subjective well-being over the entire course of the study. Our multilevel approach (see below) enables to deal with missing data of individual participants at individual waves.
More detailed information on the SOEP (including the sample structure, individual subsamples, and attrition analyses) has been previously presented (Goebel et al., 2019;Kroh et al., 2018) and is provided at https://www.diw.de/en/soep. A detailed description of all procedures and measures collected in the SOEP can be found at https://paneldata.org/soep-core. The SOEP data are available from the DIW Berlin after signing a data distribution contract (https://www.diw.de/en/diw_02.c.222829.en/access.html). Previous SOEP publications can be found at https://www.diw.de/sixcms/detail.php?id=diw_02.c.298578.en.

Assessment of Leadership
In the SOEP, participants were yearly asked about their employment status. Participants' current occupations were assessed from 1984 to 2017 with the International Standard Classification of Occupations from 1988 (ISCO-88) (Office, 1990). 1 The ISCO-88 system allows organizing jobs in clearly defined groups according to their tasks and duties. We distinguished between leaders (with occupations of ISCO-88 group 1) and non-leaders (with occupations that fall in other ISCO-88 groups).

Assessment of Subjective Well-being
General life satisfaction was assessed yearly since 1984 with a single-item measure ("How satisfied are you currently with your life as a whole?"), labeled on a 11-point scale, ranging from 0 ("completely dissatisfied") to 10 ("completely satisfied"). This measure of general life satisfaction is widely used in socio-economic and psychological research and its reli-ability and validity have been shown to be satisfactory (Schilling, 2006;Schimmack & Lucas, 2006;Schimmack & Oishi, 2005;Schimmack et al., 2008).
Since 2007, happiness was assessed as indicator of positive affect and sadness, anxiety, and anger were assessed as indicators of negative affect in the SOEP. Therefore, these four facets were considered. Specifically, participants were asked to indicate how often they had felt happy, sad, anxious, and angry during the past four weeks on a 5-point scale, ranging from 1 ("very rarely") to 5 ("very often"). The reliability and validity of this measure of affect have been previously supported (Anusic & Schimmack, 2016;Hudson et al., 2017Hudson et al., , 2020Schimmack et al., 2008).

Assessment of Working Hours
In each year, participants who were employed at the time of the survey were asked about their actual weekly working hours, including overtime work. Specifically, respondents were asked how many hours they work on average per week. Implausible answers (e.g., more than 80 h per week) were coded as missing values.

Assessment of Income
In each year, a variable was generated to indicate people's gross labor income in the previous month in participants who were employed at the time of the survey (in euros, including overtime payments but no irregular one-time payments such as bonuses). Item non-responses were imputed following a two-stage procedure. More detailed information on the imputation process can be found at https://www.diw.de/documents/publikationen/73/ diw_01.c.745953.de/diw_ssp0834.pdf.

Statistical Analysis
Stata 15 (StataCorp, 2017) was used for the analysis. Our study period reached from 2007 to 2018 because affective well-being was assessed for the first time in 2007 and both cognitive and affective well-being were assessed for the last time in 2018 so far. Individuals were considered who provided any well-being information between 2007 and 2018 and reported their occupations (according to ISCO-88) (a) at least twice during the entire study (i.e., from 1984 onwards) and (b) at least once during the study period (i.e., from 2007 onwards). We distinguished between employees who became leaders during the study period (i.e., in or after 2007, N = 1,426), and employees who never were in a leadership position throughout the entire study (i.e., from 1984 onwards, N = 24,248). Individuals who already were leaders before 2007 or when entering the panel were excluded from the analyses.
In leaders, we coded the year in which they started a leadership position (i.e., between 2007 and 2017) relative to the year in which they indicated their subjective well-being (i.e., between 2007 and 2018). Afterwards, we transformed the data from wide to long format to obtain fine-grained information on subjective well-being at different junctions before and after becoming a leader (because different people became leaders in different years). Based on these data, we modeled well-being changes from five years before until five years after becoming a leader. Similar to previous publications (Asselmann et al., 2022;Denissen et al., 2019), we used multilevel analyses with measurement occasions (Level 1) nested within persons (Level 2) and built separate models for life satisfaction, happiness, sadness, anxiety, and anger.

Well-being Differences Between Leaders and Non-leaders
First, we considered leaders and non-leaders to examine well-being differences between individuals who did and did not become leaders between 2007 and 2017 (selection and post-transition difference effects). Specifically, we simultaneously regressed the standardized score of the respective outcome (i.e., life satisfaction, happiness, sadness, anxiety, or anger) on gender (to account for gender differences), linear, quadratic, and cubic age (to account for linear and non-linear age effects), a testing variable (to account for effects due to repeated assessments of the respective indicator of subjective well-being), and a selection/post-transition difference variable. This categorical selection/post-transition difference variable was coded with 0 in non-leaders. In leaders, it was coded with 1 for well-being assessments before becoming a leader and coded with 2 for well-being assessments after becoming a leader. We compared category 1 vs. 0 to examine selection effects and category 2 vs. 0 to examine post-transition differences. Table S1 (upper and middle part) provides more detailed information on the definition and coding of each variable.

Well-being Changes Before and After Becoming a Leader
Second, we only considered leaders to examine well-being changes from five years before until five years after becoming a leader. Specifically, we simultaneously regressed the standardized score of the respective outcome (i.e., life satisfaction, happiness, sadness, anxiety, or anger) on gender, linear, quadratic, and cubic age, the testing variable, and four transitionrelated variables (anticipation, socialization, short-term, and long-term). These transitionrelated variables coded how the year of becoming a leader was temporarily related to the respective year in which the outcome was assessed. Based on these transition-related variables, we modeled nuanced well-being changes from five years before until five years after becoming a leader. Table S1 (lower part) specifies how the anticipation, socialization, as well as short-and long-term variables were defined and coded.

Random Effects
To account for individual differences with respect to mean-level well-being changes before and after becoming a leader, we repeated the analyses and included random effects for the anticipation, socialization, short-term, and long-term variable in the respective model.

Interactions with Gender and Age
Moreover, we tested whether the selection, post-transition difference, anticipation, socialization, short-term, and long-term effects varied by gender and age as moderators. Specifically, we repeated our main analyses and added an interaction term of the respective predictor with gender or age, respectively. Each interaction term was tested separately to avoid multicollinearity. Statistically significant interactions were decomposed to assess their direction of interaction. That is, the respective model was built separately in women and men or in younger and older individuals (grand-mean split).

Changes in Working Hours and Income Before and After Becoming a Leader
Furthermore, we analyzed changes in working hours and income from five years before until five years after becoming a leader. That is, we repeated our analyses in leaders with (a) their actual weekly working hours and (b) their monthly gross labor income (instead of subjective well-being) as outcome. Specifically, we log-transformed the income variables and standardized both variables in the total sample (i.e., in leaders and non-leaders). Afterwards, we simultaneously regressed the standardized score of working hours and income, respectively, on gender, linear, quadratic, and cubic age, and the four transition-related variables (anticipation, socialization, short-term, and long-term). Due to the more objective nature of working hours and income, no testing variable was used in these analyses. In the analyses on income, an imputation dummy (coded with 0 for non-imputed and 1 for imputed values, centered) was included to account for potential differences between imputed and non-imputed income values.

Alpha Level and Number of Tests
The alpha level was set at 0.05. Our main analyses on well-being changes before and after becoming a leader refer to five indicators of subjective well-being (life satisfaction, happiness, sadness, anxiety, and anger) and four transition-related effects (anticipation, socialization, short-term, and long-term). We did not adjust for multiple testing because each effect relates to another research question. However, researchers who believe that adjustment for multiple testing is necessary may refer to this number of estimated main effects.
Sample characteristics. Table S2 shows the number of leaders (N = 1,426) and non-leaders (N = 24,248), who indicated their subjective well-being between 2007 and 2018 (N = 25,674), including the mean number of well-being assessments per group.
There were 572 (40.11%) female and 854 (59.89%) male leaders as well as 12,777 (52.69%) female and 11,471 (47.31%) male non-leaders. The grand-mean age was M = 43.53 (SD = 11.06) years in leaders and M = 43.45 (SD = 12.89) years in non-leaders. On average, leaders were M = 42.44 (SD = 10.19) years old when starting a leadership position. Table S3 displays the number of observations per cell of the transition-related predictors in the total sample (including leaders and non-leaders, upper part) and leaders only (lower part). Correlations between different indicators of subjective well-being (i.e., life satisfaction, happiness, sadness, anxiety, and anger, across all waves) in the total sample and leaders only are shown in Table S4.

Well-being differences between leaders and non-leaders.
Well-being differences between leaders and non-leaders are shown in Table 1 2 . There were significant selection effects on life satisfaction (b = 0.09), happiness (b = 0.07), and sadness (b=-0.05). That is, leaders were slightly more satisfied with their lives, happier, and less sad than non-leaders in the years before becoming a leader. No selection effects were found for anxiety and anger.
Similarly, there were significant post-transition differences on life satisfaction (b = 0.12), happiness (b = 0.08), sadness (b=-0.06), and anger (b = 0.07). That is, leaders were slightly more satisfied with their lives, happier, less sad, and angrier than non-leaders in the years after becoming a leader. No post-transition difference was found for anxiety.
Interactions with gender. None of the selection effects differed between women and men. That is, the well-being differences between non-leaders and leaders before becoming a leader did not vary by gender.
However, the post-transition difference effect on anxiety differed between women and men (post-transition difference × gender: b = 0.10; 95% CI: 0.02, 0.18; p = .014). Decomposing this interaction revealed that the post-transition difference effect on anxiety was significant in women (b=-0.09; 95% CI: -0.15, 0.02; p = .011) but not men. That is, female leaders experienced anxiety less often than female non-leaders after becoming a leader. In contrast, male leaders were not less anxious than male non-leaders after becoming a leader.
In addition, the post-transition difference effect on anger differed between women and men (post-transition difference × gender: b = 0.11; 95% CI: 0.03, 0.19; p = .006). Decomposing this interaction revealed that the post-transition difference effect on anger was significant in men (b = 0.10; 95% CI: 0.05, 0.15; p < .001) but not women. That is, male leaders experienced anger more often than male non-leaders after becoming a leader. In contrast, female leaders were not angrier than female non-leaders after becoming a leader.
No gender differences were found for the post-transition difference effects on life satisfaction, happiness, and sadness. Taken together, female but not male leaders were less anxious, whereas male but not female leaders were angrier than same-sex non-leaders after they had started a leadership position.
No age differences were found for the selection effects on happiness, sadness, anxiety, and anger. Moreover, none of the post-transition differences varied by age.
2 To avoid multicollinearity, we estimated variance inflation factors (VIFs) for each model. The mean VIFs ranged (a) from 1.80 for life satisfaction to 1.81 for affect in the analyses based on the total sample (including leaders and non-leaders) and (b) from 2.62 for life satisfaction to 2.71 for affect in the analyses based on leaders only. All VIFs were smaller than 6 and thus clearly below the cut-off score of 10 that is frequently used Denissen, J. J., Luhmann, M., Chung, J. M., & Bleidorn, W. (2019). Transactions between life events and personality traits across the adult lifespan. Journal of Personality and Social Psychology,116(4), 612-633., indicating that multicollinearity was low.

Table 1
Well-being differences between leaders (N = 1,426) and non-leaders (N = 24,248)  Well-being changes before and after becoming a leader. As shown in Table 2; Fig. 1a, positive anticipation (b = 0.02 per year) and socialization (b = 0.03 per year) effect on life satisfaction indicated that life satisfaction increased gradually in the five years before and five years after becoming a leader. Moreover, a positive long-term effect on anger (b = 0.15) indicated that anger was higher after the first year of being a leader vs. before. At the same time, a negative socialization effect on anger (b=-0.03 per year) indicated that this post-transition difference in anger attenuated over time (Fig. 1b). There was no evidence that happiness, sadness, and anxiety changed before and after becoming a leader.

Random Effects
To account for individual differences in mean-level effects on subjective well-being in leaders, the analyses were repeated with random effects for the anticipation, socialization, shortterm, and long-term variable, which revealed highly similar results. The only exception was that the anticipation effect on life satisfaction was no longer significant.
Interactions with gender. The short-term effect on life satisfaction (b = 0.08; 95% CI: 0.00, 0.16; p = .045) and the socialization effect on anxiety (b = 0.03; 95% CI: 0.00, 0.05; p = .022) differed between women and men. However, the short-term effect on life satisfaction and the socialization effect on anxiety were neither significant in women nor in men. Therefore, we do not discuss these interactions further. Differences in happiness, sadness, and anger before and after becoming a leader did not differ between women and men.
Interactions with age. Testing interactions with age revealed no significant results. That is, well-being changes before and after becoming a leader did not vary by age.

Changes in Working Hours and Income Before and After Becoming a Leader
Furthermore, we aimed to unravel job changes that might explain the well-being changes in emergent leaders. Thus, we analyzed job characteristics (i.e., working hours and income) in the years before and after becoming a leader. Figure 2 depicts the (a) average actual weekly working hours and (b) average monthly gross labor income per year from five years before until five years after becoming a leader (in leaders only). Leaders worked considerably longer hours in and after the first year of being a leader than before. However, their workload diminished slightly in the years after starting a leadership position. Moreover, their income grew continuously from five years before until five years after becoming a leader.

Discussion
This study implies that levels of subjective well-being are higher in leaders vs. non-leaders and that these differences largely result from selection effects: Not only after but already before starting a leadership position, leaders were more satisfied with their lives, happier, and less sad than non-leaders. Subjective well-being marginally improved in the years before and after becoming a leader. Life satisfaction increased slightly, happiness, sadness, Table 2 Well-being changes before and after becoming a leader in leaders only (N = Note. b = coefficient from multilevel mixed-effect models. CI = confidence interval. p = p-value. Var.=variance and anger remained unchanged, and anger was even higher after becoming a leader than before. Well-being differences between leaders and non-leaders. We found that leaders were more satisfied with their lives, happier, and less sad than nonleaders before and after starting a leadership position. These findings are consistent with our selection and post-transition hypotheses and cross-sectional evidence that being a leader relates to higher subjective well-being and better health (Jurkiewicz & Massey, 1997;Li et al., 2018;Sherman et al., 2012;Skakon et al., 2011). The current study suggests that these well-being differences already exist before leaders start a leadership position, which stresses the importance of selection effects. In line with previous research (Anglim et al., 2020;Bono & Judge, 2004;DeNeve & Cooper, 1998;Furnham, 2018;Furnham & Crump, 2015;Judge et al., 2002Judge et al., , 2009Wells et al., 2016), individuals who feel better might, on average, tend to have higher leadership skills. Therefore, they might be more likely to self-select or be promoted into leadership positions, which could explain our results.
However, inconsistent with our selection hypotheses, there were no selection effects on anger. Only in the years after (but not before) starting a leadership position, leaders were more often angry than non-leaders (consistent with our post-transition hypotheses), which points toward transition-related changes (i.e., an increase in anger after transitioning into a leadership role).
Well-being changes before and after becoming a leader. There was little evidence for well-being changes in (emergent) leaders: Leaders became slightly more satisfied with their lives in the five years before and five years after becoming a leader. These results are consistent with our anticipation and socialization hypotheses that life satisfaction increases in leaders(to-be). However, inconsistent with Set-Point Theory, this effect did not diminish after the first year of being a leader but increased further over time. In terms of happiness, sadness, and anxiety, no effects were found. Finally, and consistent with our findings on group differences between leaders and non-leaders as well as our long-term and socialization hypotheses, leaders were more often angry after becoming a leader than before, and these changes attenuated over time.
In line with the Job Demand-Control Model (Karasek, 1979), the Dual-Pathway Model (Li et al., 2018), and previous research (Debus et al., 2019;Li et al., 2018), becoming a leader might decrease subjective well-being through higher job demands but also increase subjective well-being through higher job control. Such counteracting effects could explain why emergent leaders experienced an increase of life satisfaction but also more anger. The fact that no changes in other facets of affect (beyond anger) were found supports the idea that leadership positions tend to be stressful, leading to higher job demands and thus detrimental effects on mental health that -on average -cannot be balanced out by the benefits of being a leader.

Changes in Working Hours and Income Before and After Becoming a Leader
Moreover, our supplemental analyses on working hours and income provide some cues to explain why life satisfaction and anger may have changed in diverging directions: Emergent leaders tended to earn more and become more satisfied with their lives over time. Some previous research suggests that life satisfaction, the cognitive component of subjective wellbeing, might be particularly susceptible to "objective" changes in life circumstances (e.g., It is based on the anticipation effect multiplied by the time (in years) until becoming a leader. The second line indicates changes in life satisfaction/anger in the first year of being a leader. It is based on the socialization effect multiplied by the time after becoming a leader and the short-term effect. The third line indicates changes in life satisfaction/anger after the first year of being a leader. It is based on the socialization effect multiplied by the time after becoming a leader and the long-term effect. A black line indicates that any of the effects during the respective time frame (first line: anticipation effect; second line: socialization effect and/or short-term effect; third line: socialization effect and/or long-term effect) reached statistical significance (p < .05). In (a), all lines are marked in black because the anticipation and socialization effect were significant. In (b), the second and third line are marked in black because the socialization and longterm effect were significant income) (Kahneman & Deaton, 2010). Thus, one could speculate whether improvements in life satisfaction in our study were driven by improvements in income, status, and prestige.
Moreover, leaders worked longer hours and were angrier after becoming a leader than before. However, their working hours and levels of anger attenuated in the following years. In this regard, one could speculate whether higher levels of anger were driven by higher job demands, including a heavier workload and longer working hours (Debus et al., 2019). In the years after becoming a leader, working hours and anger might have decreased slightly because leaders might have adjusted at least partly to their new role over time. However, additional research is necessary to substantiate these ideas.

The Role of Gender
Partially consistent with our gender hypotheses, we found that women but not men were less anxious than same-sex non-leaders after starting a leadership position. Compared to men, women tend to be more anxious, which, however, is incongruent with leadership role stereotypes (Eagly & Johnson, 1990). Thus, especially female leaders might often need to adjust to their new role and be less anxious, which could explain these gender-specific effects.
Moreover, and partially consistent with our gender hypotheses, male but not female leaders were angrier than same-sex non-leaders after starting a leadership position. Being angry is consistent with male (but not female) gender stereotypes and fits also to leadership role stereotypes (e.g., being dominant and assertive) (Eagly & Johnson, 1990). Consistently, previous research has shown that anger expression at the workplace tends to be negatively evaluated in women but not men (Brescoll & Uhlmann, 2008;Livingston et al., 2012;Marshburn et al., 2020), Thus, men might be more likely to react with anger to elevated job demands after becoming a leader. At the same time, they might be penalized less when being angry, which could explain these results.

The Role of Age
There was little evidence for age-specific effects, except that especially older leaders were more satisfied with their lives than same-aged non-leaders before starting a leadership position. In other words, the discrepancy in life satisfaction between leaders-to-be and non-leaders was smaller in younger vs. older individuals. One could speculate whether especially younger leaders frequently tend to be highly ambitious and prone to put themselves under pressure (Seibert et al., 2017;Walter & Scheibe, 2013). Compared to older leaders, younger leaders might thus less often be more satisfied with their lives than same-sex non-leaders.
Strengths and limitations. We used data from a large and nationally representative household panel study from Germany (N = 25,674). Information on leadership and subjective well-being was yearly assessed. The comprehensive data allowed us modeling (a) well-being differences between leaders and non-leaders as well as (b) nuanced well-being changes in the years before and after becoming a leader (in leaders only). We considered multiple facets of subjective wellbeing and analyzed interactions with gender and age.
Nonetheless, our study is not without limitations: First, general life satisfaction was assessed with a single item, which might be less reliable than other, more comprehensive measures. However, the single item used in the SOEP is well established in socio-economic and psychological research and its psychometric properties have been shown to be satisfactory (Schilling, 2006;Schimmack & Lucas, 2006;Schimmack & Oishi, 2005;Schimmack et al., 2008).
Second, only a limited number of psychological items can be incorporated in the SOEP, so that only one facet of positive (happiness) and three facets of negative (sadness, anxiety, and anger) affect were assessed. Thus, future studies may not only focus on happiness but also on other facets of positive affect (e.g., alertness or pride) as, for example, assessed with the well-established Positive and Negative Affect Schedule (Watson & Clark, 1994).
Second, the frequency of happiness, sadness, anxiety, and anger was assessed retrospectively for the past four weeks at the respective yearly wave. One could speculate whether experiential measures of affect might be less susceptible to retrospective recall and memory biases and thus capture more reliable data. However, previous research has shown that global and aggregated experiential measures have similar psychometric properties when assessing overall subjective well-being (Hudson et al., 2020).
Fourth, comprehensive information on job demands and job control was not assessed in the SOEP. Thus, additional research is needed to examine whether changes in these job characteristics can explain well-being changes in emergent leaders.
Fifth, our findings come from a German sample and might not be generalizable to other groups outside Germany. For example, in countries with more traditional gender roles, wellbeing changes in emerging leaders might vary more strongly by gender.

Conclusion
Our study suggests that leaders feel better than non-leaders and that these differences already exist before they start a leadership position. However, their subjective well-being seems to improve little due to this transition. Instead, leaders even seem to be angrier after becoming a leader than before. These findings highlight the relevance of occupational health interventions to support emergent leaders and especially prevent an escalation of anger. Such interventions may help leaders to prepare for and adjust to their new role, cope with elevated job demands that may cause anger, and lead more effectively (Barling & Cloutier, 2017;Kaluza et al., 2020;Roche et al., 2014). Such programs could have beneficial effects on new leaders as well as their teams and organizations more broadly.
Future longitudinal studies with embedded ambulatory assessments and real-life observations may shed more light on the daily lives of emergent leaders. and disentangle the interplay between changes in job demands, job control, behavior, and subjective well-being before and after becoming a leader. Such research may capture even more fine-grained data on subjective well-being (e.g., in monthly increments) and additionally focus on promotions from junior to senior leadership positions and other job transitions (e.g., from leadership positions to unemployment or retirement). A more nuanced distinction between leaders in the public vs. private sector would also be useful.