Adaptation to Externally Driven Change: The Impact of Political Change on Job Satisfaction in the Public Sector.

This article uses a quasi-natural experiment to investigate the adaptation of job satisfaction to externally driven political change in the public sector. This is important because democratic government bureaucracies often experience changes in leadership after elections. The analyses are based on data drawn from a large longitudinal data set, the British Household Panel Survey. Findings indicate that the impact of political elections is largely weak and temporary and is only present for men. For women, the internal processes of the organization tend to be more important. These findings suggest that changes in political leadership may not be associated with fundamental changes in policy.

A growing literature has begun to investigate how employee job satisfaction adapts to events. These studies have looked at how both work-related and non-work-related events impact job satisfaction. For example, with regard to non-work-related events, job satisfaction has been shown to respond differently over time to life events such as marriage and having a first child (Georgellis, Lange, and Tabvuma 2012). Personality, disposition, and work–life conflict are non-work-related factors shown to affect job satisfaction (Dormann and Zapf 2001; Eby et al. 2005; Eby, Maher, and Butts 2010).

In contrast, studies exploring how job satisfaction responds to work-related events over time have focused mainly on job changes and workplace reorganization (Boswell et al. 2009; Chi, Freeman, and Kleiner 2006; Georgellis and Tabvuma 2010; Nelson, Cooper, and Jackson 1995; Pollard 2001). Other studies focusing on work-related factors have analyzed earnings, working hours, working environment, workplace socialization, autonomy, organizational control, and participation in training schemes with respect to job satisfaction (Agho, Mueller, and Price 1993; Arthur et al. 2003; Georgellis and Lange 2007).

These studies have carried out direct, causal examinations, but they do not explicitly investigate how job satisfaction adapts to external shocks. While some research has investigated the impact of external shocks, such as disability on well-being, these studies have exclusively focused on life domain and life satisfaction (Oswald and Powdthavee 2008). Although the impact of individual factors on job satisfaction is well known (Davis 2013; Fernandez and Moldogaziev 2013; Liu and Tang 2011; Pitts 2009; Pitts, Marvel, and Fernandez 2011), evidence on the direct, dynamic impact of organizational external events on workers’ job satisfaction is virtually nonexistent, especially for the public sector.

A fundamental determinant of employee job satisfaction is the fit between the employee and his or her organization. Compatibility of employees’ values and beliefs with those of the organization can result in increased job satisfaction (Kim 2012; Moynihan and Pandey 2008). Theories related to person–organization (PO) fit cover a wide spectrum of approaches, including selection and socialization with PO fit (Chatman 1991), mediation of the relationship between works attitudes and motivation (Kim 2012), work values and job choice decisions (Judge and Bretz 1992), conceptualization and operationalization of PO fit (Kristof-Brown 1996), and the fit of employee identity and organizational values (Johnson and Jackson 2009). However, these studies largely rely on data that are contemporaneous in nature instead of direct and causal examinations over time (Kristof-Brown 1996).

A fundamental determinant of employee job satisfaction is the fit between the employee and his or her organization.

We advance theory on PO fit and approaches to job satisfaction in the public sector by answering two questions: (1) How does job satisfaction in the public sector adapt to external random shocks? And (2) how does political preference, a form of PO fit, affect job satisfaction in the public sector? Whereas the second question can be considered reasonably standard but still worthwhile of being investigated, the first one, focusing on adaptation, brings a considerably new and interesting element to the study of PO fit. Studying adaptation patterns generates insights on how individuals cope with (un)expected events and how this translates into their professional lives.

According to Colquitt and Zapata-Phelan (2007), empirical research can make theoretical contributions through testing and building theory. This article emphasizes theory testing by exploiting a longitudinal data set for the analysis of PO fit in the public sector, with predictions grounded in PO fit theory. Additionally, we contribute to the adaptation literature and extend it to the public sector. The study of unexplored processes ranks high on the theoretical contribution taxonomy (Colquitt and Zapata-Phelan 2007). Therefore, in this article, we aim to cover the following points

First, we consider the results of political elections to be an external shock to the public sector. We use the results of political elections in a quasi-natural experiment to examine the adaptation of job satisfaction in the public sector. This leads to better understanding of how the job satisfaction and well-being of public sector workers responds to external events. This is especially important because job satisfaction is related to observable workplace behaviors, including absenteeism, organizational commitment, productivity, and quits (Clark, Georgellis, and Sanfey 1998; Judge et al. 2001; Schleicher, Watt, and Greguras 2004; Scott and Taylor 1985).

Second, we focus on the matching of employee political preference to the political type of the governing senior elected officials and its impact on job satisfaction of public sector employees. Ultimately the public sector is led by elected politicians whose decisions and actions are influenced by their political affiliation. We argue that such politically driven behavior simultaneously determines whether the culture, climate, and values of public sector organizations are congruent with employee values. (Dutton, Dukerich, and Harquail 1994). “Preference matching” occurs when an individual's preferred political party governs at the national level. Such value congruence may ultimately lead to increased job satisfaction (Kim 2012; Moynihan and Pandey 2008).

The results further have the potential to challenge standard PO fit theory, which relies on the assumption of value congruence. Values are considered to be relatively stable over time (Chatman 1989, 1991), and their fit with the values represented by the organization determines job satisfaction levels. However, if we find evidence for adaptation in the case of matched political preferences—that is, job satisfaction falls to baseline levels although values are aligned—PO fit theory can be criticized.

Our analyses are based on data drawn from the British Household Panel Survey, a large longitudinal data set. The present study extends previous PO fit research by making three major contributions: First, we contribute to the nascent body of research examining the adaptation of job satisfaction to work-related and non-work-related events by investigating adaptation in the public sector. Second, we focus our analysis on adaptation to an external shock. The literature on adaptation largely considers events that are entirely or partly endogenous (e.g., marriage, voluntary job change). It does not investigate adaptation to external events. Thus, this article opens new avenues for research on job satisfaction. Third, the effect of political preferences on job satisfaction is seldom investigated, although it might have a critical impact on the performance of public sector organizations. In addition, we highlight how the effects of political preference matching differ across gender.

Theoretical Development

The following sections outline approaches to PO fit and its relation to job satisfaction and political preferences as external shocks. We also discuss adaptation issues and gender differences in job satisfaction. Ultimately, each section derives hypotheses.

Person–Organization Fit and Political Preferences

PO fit considers the compatibility between individuals and organizations. It occurs when “(a) at least one entity provides what the other needs, (b) they share similar fundamental characteristics, or (c) both” (Kristof-Brown 1996, 4). PO fit is often understood as value congruence (Chatman 1989; Liedtka 1989). According to the theory of work adjustment, job satisfaction represents employees’ subjective evaluation of the degree to which their requirements are met by the work environment (Bretz and Judge 1994). The theory states that the fit between employees and the organization determines the level of job satisfaction experienced by employees.

Increasingly, the PO fit literature has focused on employee–organization value congruence. Value congruence between employees and organizations refers to the compatibility of the employee's values and beliefs with those represented by the organization (Ren 2010). Value congruence is particularly important because values are relatively stable over time and guide people's attitudes, judgments, and behaviors (Chatman 1989, 1991). Individuals’ political preferences are deemed to be a good proxy for deeply rooted individual beliefs and values. This is because they capture deeply embedded value systems in any given personality. Many voters think of themselves as long-term party supporters and remain loyal to the same party across different elections (Inglehart and Klingemann 1979).

Organizational values convey information about expectations and formal and informal rules, which connect the individual employee to the broader organizational context (Johnson and Jackson 2009). Studies have shown that high-ranking members of organizations, such as top management teams, strongly influence the values and culture of organizations (Hoffman et al. 2011). We argue that governing senior elected officials influence the culture, climate, and values of public sector organizations through their policy and management decisions. Because those decisions and actions are strongly influenced by the elected official's political affiliation, the political affiliation of senior elected officials will determine whether the culture, climate, and values of public sector organizations are congruent with employee values.

Having a political party preference match between public sector employees and senior governing elected officials is likely to result in a high level of value congruence between such public sector employees and the organization. This is because senior governing officials will influence the values and culture of the organization to reflect their political values and beliefs. This high degree of value congruence between employees and the organization will imply that matching employees will derive greater job satisfaction from working in the public sector. Consequently, we derive the following hypothesis:

A political party preference match between public sector employees and senior governing elected officials is likely to result in a high level of value congruence between such public sector employees and the organization.

Hypothesis 1: Political party preference matching between employees and senior governing elected officials is associated with higher job satisfaction in the public sector.

External Shocks

The outcomes of political elections are, to a large extent, external to the public sector. The change in senior elected governing officials can have both positive and negative effects on the job satisfaction of public sector workers with matching political preferences. First, as discussed earlier, job satisfaction will likely increase if the political preference of public sector workers matches the political affiliation of the new senior elected governing officials because of value congruence.

Second, internal processes of the organization are likely to change following an external change in public sector political leadership (Van Wart 2003). A change in leadership will result in some reorganization of the public sector. Research shows that job satisfaction decreases immediately before and during reorganization in the workplace. This decrease in job satisfaction is attributed to uncertainty during reorganization and is found in various forms of workplace restructuring, including restructuring of local government, privatization, and intracompany job transfers (Nelson, Cooper, and Jackson 1995; Pollard 2001). Because the public sector goes through significant reorganization when new senior officials govern following an election, we can expect employee job satisfaction to be adversely affected during that reorganization.

The change in senior governing officials will have a positive effect on the job satisfaction of preference-matching public sector employees if the positive effect of political party preference matching is greater than the negative effect of workplace reorganization on job satisfaction. The euphoria and positive emotion associated with a preferred political party winning an election is experienced immediately after the election. Typically, changes in the working of the organization only begin to be implemented several months after the election and continue for several years thereafter. Thus, the negative impact of workplace reorganization on job satisfaction will take place sometime after the election. Therefore, we expect the job satisfaction of preference-matching public employees to increase immediately after an election that elects their preferred political party.

Hypothesis 2: A change in public sector political leadership will positively impact the job satisfaction of preference-matching public sector employees.

Adaptation and Internal Processes of the Organization

Recently, researchers have become interested in analyzing patterns of adaptation to life events (Clark et al. 2008; Georgellis, Lange, and Tabvuma 2012; Georgellis and Tabvuma 2010). Adaptation is founded on the psychological literature on set point theory, which argues that individuals have happiness set points to which they inevitably return following disruptive life events (Headey and Wearing 1989). A number of studies have used large-scale longitudinal data to look for evidence of adaptation to economic and life events (Catalano and Dooley 1983; Clark et al. 2008; Frijters, Johnston, and Shields 2011). For example, Clark et al. (2008) found evidence of complete adaptation of life satisfaction to marriage, childbirth, widowhood, and divorce. However, they found no evidence of adaptation to unemployment. With respect to job satisfaction, Georgellis, Lange, and Tabvuma (2012) found evidence that job satisfaction does not adapt to the birth of a first child for women. Georgellis and Tabvuma (2010) found evidence that the increase in job satisfaction after a job change from the private sector to the public sector does not dissipate. The events that this literature has focused on are largely endogenous or partially endogenous events. Very few studies have looked at adaptation to external events. One exception is the study by Oswald and Powdthavee (2008), which found evidence of incomplete adaptation to disability, a largely external event. In this article, we contribute to the study of adaptation to external events by investigating the impact of election results, which are largely external, on the job satisfaction of public sector workers.

Adaptation theory implies that preference matching will only have a positive impact on job satisfaction for a limited period of time, as job satisfaction will eventually adapt back to a baseline level. The emotional response to an event is a main feature of adaptation processes. Events whose outcomes are below expectations or whose impact takes place over a short time period will result in emotional responses of shorter duration. In contrast, events whose impact is continuous and sustained will result in emotional responses over a longer duration. Therefore, the outcome of events experienced by the organization will significantly determine the length of time it takes for well-being in the workplace to return to pre-event levels. This is especially relevant when we consider the context of U.K. politics. The Labor and Conservative parties in the United Kingdom have converged dramatically on economic and social policy over the past two decades in an attempt to win votes (Adams, Green, and Milazzo 2012). Indeed, the Tony Blair government was accused of being a right-wing government in disguise (Adams, Green, and Milazzo 2012). As a result, its supporters were likely disappointed and frustrated by the economic and social policies pursued by the new governing party. This would be expected to lead to lower job satisfaction among public sector workers supporting this party. Also, adaptation to baseline occurs relatively quickly after the election.

Hypothesis 3: Job satisfaction quickly reverts to a baseline level after political party preference matching.

Gender Differences in Job Satisfaction

Gender differences in labor market attachment, job satisfaction, work–life conflict, and workplace values are well documented in the literature (Clark 1997; Hodson 1989; Sousa-Poza and Sousa-Poza 2003). A large body of empirical evidence has shown that women hold greater responsibility in household work and raising children than their male counterparts (Wiersma 1990). However, Clark (1997) argues that gender differences in job satisfaction should be transitory.

In addition, the analysis of gender and political preferences has provided interesting insights on the differences between men and women in political issue identification and prioritization. For example, women are more likely to prioritize education and health care issues, and men are more likely to select the economy as their most important election issue (Campbell 2004). Furthermore, men are found to be more committed and attached to political parties (Pratto, Stallworth, and Sidanius 1997). Such gender differences are likely to be an important moderating factor in the influence of external change on job satisfaction, as women may find it more difficult to adjust to changes brought about by new political leadership. Because there are important gender differences in political issue identification and commitment, gender may moderate the impact of external change on job satisfaction in the public sector. Therefore, we hypothesize,

The analysis of gender and political preferences has provided interesting insights on the differences between men and women in political issue identification and prioritization.

Hypothesis 4: Women's job satisfaction is more likely to be adversely affected by external change in the public sector than men's job satisfaction.

Data and Methodology

Sample

The data set comprises the first 18 waves of the British Household Panel Survey (BHPS). The BHPS is an annual survey consisting of a nationally representative sample of about 5,500 households recruited in 1991, containing approximately 10,000 individuals. The sample is a stratified clustered design drawn from the Postcode Address File. All residents present at those addresses at the first wave of the survey were reinterviewed each successive year, and they were followed thereafter, even if they moved to a new household. The BHPS is an unbalanced panel in that although approximately 10,000 individuals are interviewed year to year, some individuals drop out from the sample. Because of attrition, men remain in the survey for an average of 4.9 years, while women remain in the sample for 4.7 years from 1991 to 2008 in our final sample.

The longitudinal nature of the BHPS allows us to identify public sector workers before and after political elections, thus providing a quasi-natural experiment, as the results of political elections are largely external to the public sector. In addition, the BHPS provides a rich source of information on demographic and labor market characteristics, as well as information on individuals’ subjective evaluation of their jobs, their economic situation, and their political preferences.[1]

In our sample, we have 1,667 men whom we observe 8,159 times from 1991 to 2008 and 2,519 women whom we observe 11,893 times from 1991 to 2008. Thus, our final sample consists of 8,159 and 11,893 person-year observations for men and women, respectively. The sample consists of full-time workers in the public sector ages 16 to 65. Our sample does not include any part-time or unemployed individuals.

Identification of Political Preferences

Using exclusive categories, respondents are required to identify which political party they are closest to: Conservative, Labour, Liberal Democrats (Liberals/Social Democratic Party [SDP]), Scottish Nationalists, Plaid Cymru, Green Party, or other parties. Political party preference is identified after the respondent answer yes to either of the following questions: “Generally speaking do you think of yourself as a supporter of any one political party?” or “Do you think of yourself as a little closer to one political party than to the others?” Because of the small number of observations, the Liberal Democrats (Liberals/SDP), Scottish Nationalists, Plaid Cymru, and Green Party were reclassified as other parties. This question was asked in every year in which the panel data were collected. Table 1 shows the distribution of political party preference among public sector workers.

Table 1

Political Party Preferences of Public Sector Workers

Year	Men			Women
	Year Total	Labour	Conservative	Year Total	Labour	Conservative
1991	484	180	125	512	159	130
1992	163	47	33	276	54	51
1993	422	176	84	490	180	95
1994	408	184	73	491	186	67
1995	386	187	68	489	195	71
1996	387	180	81	509	187	77
1997	358	176	53	512	242	64
1998	384	173	52	553	216	80
1999	574	213	67	835	302	88
2000	560	204	77	834	294	92
2001	579	214	69	875	326	91
2002	515	170	63	773	272	66
2003	495	155	63	782	225	69
2004	494	154	69	793	228	77
2005	505	164	59	827	254	80
2006	500	152	73	811	227	79
2007	486	142	68	788	230	75
2008	459	138	69	743	176	81
Total	8,159	3,009	1,246	11,893	3,953	1,433

Matched Political Preferences

Matched political preferences occur when, at a given point in time, a public sector employee's preferred political party is governing at the national level. In our sample, the Conservative Party was in power from 1991 to 1996, and the Labour Party was in power from 1997 to 2008.[2] Four interaction terms are created to identify when political preferences matched and mismatched the governing political party.

Dependent Variable

The dependent variable is overall job satisfaction. Overall job satisfaction is measured on an ordinal Likert scale ranging from 1 to 7, where a value of 1 corresponds to “not satisfied at all” and a value of 7 corresponds to “completely satisfied.” The measure is constructed from individuals’ responses to the question, “I am going to read out a list of various aspects of jobs, and after each one I'd like you to tell me from this card (19) which number best describes how satisfied or dissatisfied you are with that particular aspect of your present job.”[3] Various studies show that a single-item overall job satisfaction measure performs well as opposed to multiple item and domain measures of job satisfaction (Georgellis and Lange 2012; Wanous, Reichers, and Hudy 1997).

Control Variables

We use a common set of control variables to minimize the impact of biasing effects. We control for demographic and job characteristics. The demographic characteristics that we control for are age, marital status, education, region, and health. In order to identify a nonlinear relationship between age and job satisfaction, we also include the square of individuals’ age as a control variable. The job characteristics that we control for are the log of individuals’ real income (gross monthly income), log of weekly working hours, union membership, promotion opportunities, and occupational classification at the one-digit level. Firm-level controls include job tenure (in years), firm size, and industry classification at the one-digit level.[4]

Model and Estimation

First, we compare the differences in means of job satisfaction in cases of matching and nonmatching political preferences. Second, we estimate the impact of these two cases on job satisfaction using fixed-effect panel regression. Controlling for a fixed effect ensures that the preference matching/mismatch dummy variables pick up the effect of political party preference matching/mismatching instead of selection effects, whereby fixed unobserved individual characteristics (e.g., personality or ability) are associated with higher job satisfaction and a higher probability of having a given political party preference. We present fixed-effects estimations that are based on simple ordinary least squares (OLS) regressions of job satisfaction. Ferrer-i-Carbonell and Frijters (2004) show that controlling for fixed effects is more important than respecting the ordinality of the variables. They also show that OLS produces coefficients that are easier to interpret in terms of orders of magnitude. Furthermore, the dynamics of adaptation to political preference matching are likely to be moderated by personality traits and individuals’ previous experiences. This highlights the importance of accounting for heterogeneity when investigating patterns of adaptation to events (Lucas 2007).

We control for various demographic and job characteristics, including marital status, education, age, log of real income (gross monthly income), log of weekly working hours, union membership, promotion prospects, job tenure, firm size, health, occupational dummies, industry dummies, and regional dummies.[5] Table 2 presents the correlations for the control variables. We estimate separate regressions for men and women.

Table 2

Correlation Coefficients for Control Variables

	Mean	SD	Age	Age Squared	Health	Education: High	Education: Mid	Married	Widowed	Divorced	Separated	Renter	Real Wage
Age	40.117	10.759	1.000
Age Squared	1.725	0.874	0.988**	1.000
Health	0.513	0.500	0.160**	0.162**	1.000
Education: High	0.644	0.479	0.033**	0.010**	0.005	1.000
Education: Mid	0.268	0.443	−0.169**	−0.155**	−0.031**	−0.700**	1.000
Married	0.575	0.494	0.443**	0.407**	0.039**	0.039**	−0.084**	1.000
Widowed	0.014	0.118	0.106**	0.113**	0.026**	−0.021**	−0.012**	−0.100**	1.000
Divorced	0.069	0.253	0.151**	0.143**	0.035**	0.003	−0.019**	−0.247**	−0.022**	1.000
Separated	0.017	0.128	0.029**	0.021**	0.010**	−0.009**	0.004	−0.137**	−0.012**	−0.030**	1.000
Renter	0.147	0.354	−0.162**	−0.147**	0.010**	−0.084**	0.007[*]	−0.222**	0.002	0.044**	0.039**	1.000
Real Wage	2.778	0.480	0.250**	0.200**	−0.024**	0.373**	−0.204**	0.212**	−0.025**	−0.003	0.002	−0.208**	1.000
Log Working Hours	3.595	0.147	−0.015**	−0.015**	−0.040**	−0.006	−0.019**	0.007	−0.029**	−0.022**	−0.003	0.036**	0.220**
Union Member	0.645	0.479	0.160**	0.144**	0.038**	0.090**	−0.060**	0.098**	0.025**	0.032**	−0.003	−0.096**	0.152**
Promotion Opportunities	0.582	0.493	−0.165**	−0.169**	−0.038**	0.123**	−0.030**	−0.071**	−0.018**	−0.024**	−0.005	−0.002	0.130**
Job Tenure	5.409	6.44	0.027**	0.027**	0.003	−0.002	−0.002	0.012**	0.002	0.005	−0.001	−0.007	0.008[*]
Firm Size: Small	0.195	0.397	−0.046**	−0.035**	−0.001	−0.066**	0.048**	−0.045**	0.002	−0.006	−0.003	0.077**	−0.182**
Firm Size: Medium	0.386	0.487	0.025**	0.021**	−0.000	0.011**	−0.014**	0.025**	−0.003	−0.004	0.003	−0.027**	0.029**
		Log Working Hours		Union Member		Promotion Opportunities		Job Tenure		Firm Size: Small		Firm Size: Medium
Log Working Hours		1.000
Union Member		−0.098**		1.000
Promotion Opportunities		−0.023**		0.178**		1.000
Job Tenure		−0.001		0.001		−0.015**		1.000
Firm Size: Small		0.050**		−0.173**		−0.143**		0.003		1.000
Firm Size: Medium		−0.002		0.025**		0.001		−0.003		−0.498**		1.000

Note: Correlation coefficients for the regional, industry, and occupational dummy variables are not included.

Indicates significance at a 1% confidence level

indicates significance at a 5% confidence level.

Third, the longitudinal nature of the BHPS allows us to identify public sector employees that experience a change in the political leadership and to follow them two years prior to and more than five years after the event. As shown in table A3 in the appendix, we identify 136 men and 150 women who prefer the Labour Party in the year the Labour Party took power from the Conservative Party. These employees experienced a change in public sector political leadership at time t, where t is the year 1997. To identify adaptation effects to a change in political leadership, we use lag dummies where we identify Labour-preferring employees from the time of the election in each year following the election. The last category is a catchall category for Labour-preferring employees. It captures all employees with a Labour preference starting five years after the election. If all the estimated lag coefficients are approximately the same, then we have no evidence of adaptation. However, if the coefficients of lags distant from the event are smaller than the coefficients of lags close to the time of the event, then this is evidence of adaptation, or at least partial adaptation. The adaptation coefficients should be interpreted according to whether they are significantly different from the baseline (other public sector workers) and not whether the coefficients are significantly different from each other. Table A3 summarizes the number of lag observations in our sample. As table A3 indicates, the number of observations decreases as we move further away from the event. For example, out of the 136 male public sector employees who prefer Labour at the time of the election, we can only follow 130 of them one to two years after the election, 117 of them two to three years after the election, and so on. This is attributable to: (1) sample attrition in the BHPS survey; (2) respondents having stopped indicating their Labour Party preference in the survey; or (3) respondents changing support to another political party.

We estimate fixed-effects regressions to test for adaptation effects from a change of political leadership in the public sector. We use the same methodology used in Georgellis, Lange, and Tabvuma (2012) by estimating the lag coefficients in the same regression to ensure that the coefficients are directly comparable. To measure well-being before the election, we look at two lead coefficients, Labour-preferring employees from the time of the election (time t) in the two years before the election. We expect the lead coefficients to be insignificant, as public sector workers are unlikely to predict the outcome of an election in the two years before the event. We estimate the following fixed-effects regression equation:

where S represents job satisfaction and X is a vector of demographic and job characteristics, including marital status, education, number of children, age, income (gross monthly income), firm size, health, regional, and industrial and occupational dummies. LAB represents the lead and lag dummy variables, θ is the estimated coefficient for the lead and lag dummy variables, and ε is an error term.

Results

Mean Satisfaction Comparisons

Table3 summarizes differences in mean job satisfaction between matching and nonmatching of political preference to the governing political party. Unsurprisingly, simple t-tests for differences in means reveal that overall job satisfaction is higher in cases in which political preferences match the party in power. Only women preferring the Labour Party deviate from this pattern. These results indicate that political preferences play a role in determining overall job satisfaction.

Table 3

Mean Job Satisfaction

	Men			Women
	LB × LBGVT	LB × CNGVT	T-Stat on Difference	LB × LBGVT	LB × CNGVT	T-Stat on Difference
Job satisfaction	5.259	5.130	2.52**	5.373	5.383	−0.22
Observations	2,061	948		3,009	944
	CN × CNGVT	CN × LBGVT	T-Stat on Difference	CN × CNGVT	CN × LBGVT	T-Stat on Difference
Job satisfaction	5.495	5.283	2.89**	5.724	5.561	2.58**
Observations	465	781		492	941

Indicates significance at a 1% confidence level

indicates significance at a 5% confidence level.

Job Satisfaction Regression Results

Table4 presents the estimated coefficients of the overall job satisfaction fixed-effects model. The coefficients for men are shown in column 1, while the coefficients for women are shown in column 2. We find that for men, overall job satisfaction is significantly higher when political preferences match the governing party for both Labour-preferring (coeff. q1= .144, p < .01) and Conservative-preferring (coeff. = .217, p < .01) public sector workers. For women, job satisfaction is higher only when there is a match of Conservative Party preference and Conservative Party government (coeff. =. 143, p < .05). We find no impact on job satisfaction when there is a preference mismatch. These results hold after controlling for many demographic and job characteristics.

Table 4

Fixed-Effects Results for Impact of Preference Matching on Job Satisfaction

	Men	Women
LB × LBGVT	0.144**	−0.010
	(0.048)	(0.0389)
CN × CNGVT	0.217**	0.143*
	(0.076)	(0.072)
LB × CNGVT	0.121	0.070
	(0.063)	(0.056)
CN × LBGVT	0.032	−0.031
	(0.067)	(0.060)
Observations	8,159	11,893
Individuals	1,667	2,519

Note: Standard errors are given in parentheses.

Indicates significance at a 1% confidence level

indicates significance at a 5% confidence level.

Other controls include age, age squared, number of children, health, education, marital status, renter, log of real wage, log of working hours, union membership, promotion opportunities, job tenure, firm size, and dummy variables for region and occupation.

These results provide evidence for hypothesis 1—that higher levels of value congruence, as proxied by matching political preferences with the current governing senior elected officials, are associated with higher job satisfaction in the public sector. The results also show that there are significant gender differences in the impact of matching political preferences with the current governing senior elected officials.

Higher levels of value congruence, as proxied by matching political preferences with the current governing senior elected officials, are associated with higher job satisfaction in the public sector.

Adaptation Results

The results of our trend analysis are given in table5. We find that for men, job satisfaction increases at the time of a change in political leadership and decreases immediately one year after the event. Job satisfaction and promotion prospects for women decrease in the first two years following a change in political leadership. Wages and working hours are unaffected by the change in leadership.

Table 5

Transition to Preference Matching: Trend Analysis

	Mean Job Satisfaction		Mean Real Wages (£/month)		Working Hours (Hrs/Week)		Promotion Prospects
	Men	Women	Men	Women	Men	Women	Men	Women
Public Sector Mean	5.25	5.44	2,026	1,654	38.40	35.70	0.62	0.56
Leads
1–2 years hence	5.08	5.36	1,727**	1,483*	37.66	36.25	0.65	0.58
Within the next year	5.07	5.34	1,726**	1,479*	38.26	36.12	0.58	0.53
Lags
0–1 years	5.26	5.21**	1,774**	1,541	38.48	36.13	0.60	0.47*
1–2 years	4.97*	5.25*	1,838*	1,535	38.59	36.35	0.59	0.47*
2–3 years	5.04	5.28	1,888	1,539	38.16	36.16	0.59	0.50
3–4 years	5.35	5.23	1,894	1,671	38.03	35.86	0.58	0.49
4–5 years	5.27	5.29	1,985	1,782	38.00	35.78	0.65	0.56
5 or more years	5.21	5.43	2,198**	2,341**	37.31**	35.63	0.57*	0.45**

Note: Mean satisfaction scores less than the corresponding mean satisfaction for stayers are in bold.

Indicates significance at a 1% confidence level

indicates significance at a 5% confidence level from public sector mean.

The fixed-effects results of the adaptation to a change in political leadership in the public sector are given in table6. The first column summarizes the results for men, and the second column summarizes the results for women. We perform separate analyses for men and women because of widely researched differences in terms of workplace values and beliefs, motivation, and labor market attachment. Particularly relevant in the present context is the fact that women tend to hold greater responsibility in household work and child rearing than their male counterparts, and therefore they are more likely to experience the negative effect of changes in the working of the organization.

Table 6

Fixed-Effects Job Satisfaction Regressions for Dynamic Impact of Change in Political Leadership

	Transition to Preference Matching
	Men	Women
Leads
1–2 years hence	0.006(0.100)	−0.165(0.105)
Within the next year	0.043(0.100)	−0.136(0.098)
Lags
0–1 years	0.213*(0.103)	−0.214*(0.098)
1–2 years	−0.084(0.106)	−0.239*(0.105)
2–3 years	−0.078(0.112)	−0.275*(0.114)
3–4 years	0.164(0.119)	−0.212(0.120)
4–5 years	0.115(0.121)	−0.129(0.126)
5 or more years	0.052(0.088)	0.063(0.082)
Age	−0.061**(0.019)	−0.051**(0.016)
Age Squared	0.653**(0.216)	0.171(0.182)
Health	−0.040(0.035)	−0.053(0.029)
Education: High	−0.018(0.149)	−0.348**(0.125)
Education: Mid	−0.161(0.157)	−0.310*(0.125)
No. of Children	0.048(0.025)	0.016(0.024)
Married	−0.143*(0.065)	0.033(0.050)
Widowed	0.172(0.221)	0.133(0.144)
Divorced	−0.035(0.110)	−0.038(0.083)
Separated	0.239(0.139)	0.178(0.102)
Renter	−0.012(0.064)	−0.008(0.052)
Real wage	0.284**(0.070)	−0.013(0.056)
Log Working Hours	−0.309*(0.156)	−0.129(0.104)
Union Member	−0.137**(0.040)	−0.100**(0.034)
Promotion Opportunities	0.380**(0.033)	0.267**(0.026)
Job Tenure	0.000(0.000)	0.000(0.000)
Firm Size: Small	−0.069(0.054)	−0.025(0.045)
Firm Size: Medium	0.100*(0.042)	0.004(0.038)
Constant	6.783**(0.717)	7.623**(0.521)
Regional Dummies	Yes	Yes
Industry Dummies	Yes	Yes
Occupational Dummies	Yes	Yes
Observations	8,159	11,893
Individuals	1,667	2,519

Before interpreting the lags and lead coefficients, it is important to consider the estimated coefficients of the control variables. Overall, the results of the control variables, in particular regarding the relation between job satisfaction, age, and education, support the findings of previous studies (Clark and Oswald 1996).

Income is positively associated with job satisfaction for men (β = .284, p < .01). We find no evidence of a positive relationship between income and job satisfaction for women. Working hours are negatively related to job satisfaction for men (β = –.309, p < .05). Unsurprisingly, we also find that the availability of promotion opportunities is positively associated with job satisfaction for both men (β = .380, p < .01) and women (β = .267, p < .01). Union membership is negatively associated with job satisfaction for both men (β = –.137, p < .01) and women (β = –.100, p < .01).

Table6 summarizes the estimated lag and lead coefficients for the effect of a change in political leadership in the public sector. We summarize the estimated lag and lead coefficients from the fixed-effects regressions in figures 1A and 1B. Figure 1A summarizes the results for men, while figure 1B summarizes the results for women. As it emerges from these results, there are no anticipation effects for both men and women to a change in political leadership. This is expected, as it is unlikely public sector employees can predict the outcome of political elections.

Figure 1

The Dynamic Effect of Preference Matching on Job Satisfaction

Discussion

In this study, we use longitudinal data to carry out a quasi-natural experiment to investigate how the job satisfaction of public sector employees adapts to external organizational change. Using recently developed methods underpinned by adaptation theory, we find a statistically significant increase in job satisfaction when political leadership changes to match the preference of male public sector employees. However, job satisfaction for men fully adapts to baseline levels after one year of preference matching. We also find that the job satisfaction of women is less prone to political influence than men. For women, the negative effect on job satisfaction of preference matching lasts for up to three years after a change in political leadership. We attribute this to the adverse effects of organizational change on job satisfaction, as women are less influenced by political preferences.

First, this suggests that, for men, the positive impact of preference matching is greater than the impact of workplace reorganization on job satisfaction soon after the election. Second, this increase is not sustained, and job satisfaction permanently decreases, becoming statistically insignificant one to two years after the event onward. This is strong evidence of adaptation and shows that the positive effect of an external political shock on job satisfaction is temporary.

Instead, for women, there is a statistically significant decrease in job satisfaction when the political leadership changes to match the preference of female public sector employees. Earlier results show that the job satisfaction of women that prefer the Labour Party is unaffected by political party preference matching. This suggests that the decrease in job satisfaction for women is caused by the negative effect of workplace reorganization on job satisfaction. Furthermore, job satisfaction is significantly lower for at least two to three years after the event. This is evidence that there is complete adaptation to a change in political leadership in the public sector only after two to three years. This longer adjustment period reflects the uncertainty and changes from new policies that are likely to affect work–life balance issues that are important to women and have been shown to adversely affect women's job satisfaction (Georgellis et al. 2012).

There is complete adaptation to a change in political leadership in the public sector only after two to three years.

The present study makes the following contributions to adaptation and PO fit research. First, we contribute to the nascent body of research examining the adaptation of job satisfaction to work-related and non-work-related events. We study adaptation to organizational change that is the result of an event external to the organization, that is, the outcome of political elections. This contributes to a better understanding of the temporal duration of the effects of organizational change on job satisfaction in the workplace. Our results show that the effect of political change on the job satisfaction of public sector workers is relatively weak and short term. This implies that the outcome of political elections is not crucial for the operation of public administration. This is especially true when we consider the political context in the United Kingdom. The two major political parties, the Labor and Conservative parties, have evolved toward very similar economic and social policies. Thus, a change in political leadership from Conservative to Labour (as we observe in our quasi-natural experiment) is unlikely to result in the types of changes in policy desired by party supporters. This result indicates that the initial euphoria of having one's preferred political party assume leadership of the public sector is not sustained. More fundamental reforms may carry with them a longer lasting impact on satisfaction.

Second, our study further extends PO fit research by investigating the effect on job satisfaction of employee political preferences matching the political affiliation of the senior governing officials. Prior to this study, very few studies attempted to examine political preferences in relation to the work context (Boddewyn and Brewer 1994). This is surprising given the variety of topics studied in the organization political behavior literature (Buchanan 2008). Therefore, we extend the analysis of PO fit and job satisfaction to include political preferences. This is important because this literature has hypothesized that a better match between organizations and individuals can result in more efficient and effective organizational performance (Besley and Ghatak 2005). However, while we find that political preferences can have a significant impact on job satisfaction in the public sector, this impact is weak and transitory. This implies that some forms of PO fit, such as the matching of political preferences, will have a smaller impact on organizational performance than expected.

Nonetheless, we find complete adaptation for men immediately after the event and adaptation for women after three years. This challenges PO fit theory because, according to PO fit theory, if values are aligned (i.e., matched preferences), job satisfaction should not decline. Thus, researchers may want to investigate the temporal aspects of PO fit theory to a stronger extent.

Third, we highlight how the effects of political preference matching differ across gender. Women seem to be less prone to political influence, even though they might still have strong political preferences. Their job satisfaction is less likely to be significantly influenced by political preferences. Men tend to display the opposite behavior. Their job satisfaction depends to a larger extent on their political preferences, in the short term at least. This finding confirms the previous argument that men and women organize their political views in different ways (Campbell 2004; Finseraas, Jakobsson, and Kotsadam 2012).

Managerial Implications

Our findings have several implications for public sector managers. First, our results show that a change in political leadership will only have a short-term impact on job satisfaction in the public sector. This suggests that public sector managers do not have to spend scare resources helping public sector employees become used to the new political direction in which the public sector will be led. Second, we also find that the change in public sector political leadership can adversely affect the job satisfaction of women. Public sector organizations should ensure they have policies and take action to mitigate the adverse effects of workplace reorganization on job satisfaction that follow a change in political leadership of the public sector. Such policies should be focused on reducing uncertainty, as previous research has shown that this is a major cause of the decline in job satisfaction before and during workplace reorganization.

Limitations and Future Research

Finally, we have to address a number of limitations that affect the interpretation of our findings. At the same time, these limitations are a good trigger for future research. First, our analyses use U.K. data only. Thus, the findings mainly relate to a U.K. working environment, and we cannot claim that the results are universally applicable. However, previous research has shown that the factors influencing the relation between job satisfaction and context factors are very similar across countries (Irvine and Evans 1995). Thus, the results presented here are of relevance for an international audience as well. Consequently, an option for future research arises with respect to cross-country comparisons of matched or mismatched political preferences and their effects on job satisfaction. Such an endeavor would help validate the findings presented here at an international level.

Second, we acknowledge the possibility of selection effects for which we were not able to control. However, our main approach was to investigate the patterns of adaptation, which are unlikely to be affected by selection effects.

Third, we rely on self-reported measures of satisfaction and do not apply objective measures of performance, as in Hekman et al. (2009). However, this limitation is compensated by the large sample that we use, in the sense that distortion of self-reported measures is unlikely to have a major influence. Nonetheless, it would be a worthwhile effort to extend the analyses presented here to more objective performance data.

Fourth, we only estimate how political party preferences affect the job satisfaction of supporters of the two major political parties in the United Kingdom. A sizable proportion of public sector employees in the United Kingdom support smaller parties, such as the Liberal Democrats (Liberals/SDP), Scottish Nationalists, Plaid Cymru, and Green Party. Our methodology does not allow us to investigate how political party preference matching affects these employees because none of these smaller political parties governed during our sample period.

Fifth, we consider the impact of a change in central government political leadership. While the senior governing officials of the central government are likely to affect the values, culture, and climate of the entire public sector through their policies and decisions, their impact is likely to vary depending on whether employees work for the central government, local government, or public sector organizations, such as the National Health Service. Our data and methodology do not allow us to investigate how political preference matching varies across these different levels of government because the sample sizes for preference matching become too small when divided into the different levels of government. It will be worthwhile for future research to extend our analysis to the different levels of government to investigate how the impact of political preferences on job satisfaction differs across the different levels of government.

Finally, BHPS data are collected annually. This limits our analysis to the year in which an event takes place instead of the actual date of the event. A recent study by Frijters, Johnston, and Shields (2011) improves reliability of the findings by using quarterly rather than yearly data to test for anticipation, adaptation, and selection effects in the Australian labor market. Further research needs to be done on the causal connection between preference matching and job satisfaction and future research can improve reliability and accuracy by using higher frequency data to more accurately investigate the dynamic impact of externally driven organizational change on job satisfaction.

A1

The Distribution of Job Satisfaction Measures

Rank	Men		Women
	Overall Job Satisfaction		Overall Job Satisfaction
	Count	%	Count	%
1	140	1.72	128	1.08
2	250	3.06	274	2.30
3	639	7.83	805	6.77
4	675	8.27	678	5.70
5	1,897	23.25	2,600	21.86
6	3,836	47.01	5,915	49.74
7	723	8.86	1,493	12.55
Total	8,159	100.00	11,893	100.00

A2

Variable Definitions

Variables	Definition	Mean
Age	Age of respondent	40.12
Age2	Square of respondent's age divided by 1,000	1.73
Health	Equals 1 if respondent reports no health problems	0.51
Higher Education	Equals 1 if respondent's education includes a higher degree, a first degree, a teaching qualification, or some other higher qualification	0.64
Medium Education	Equals 1 if respondent's education includes a nursing qualification, GCE A levels, or GCE O levels	0.27
No. of Children	The number of children each respondent has	0.56
Lower Education	Equals 1 if respondent's education includes a commercial qualification (with no GCE O level), CSE Grade 2–5 or Scot G, apprenticeship, other qualifications, or no qualifications	0.09
Married	Equals one 1 if respondent is married	0.57
Couple	Equals 1 if respondent is living with his or her partner	0.15
Widowed	Equals 1 if respondent is widowed	0.01
Divorced	Equals 1 if respondent is divorced	0.07
Separated	Equals 1 if respondent is separated from spouse	0.02
Never married	Equals 1 if respondent has never been married	0.18
Renter	Equals 1 if respondent rents the property he/she resides in	0.15
Log real wage	Log of respondent's usual inflation adjusted monthly pay	2.78
Log Work Hours	Log of the respondent's weekly working hours	3.59
Union Member	Equals 1 if respondent is a member of a labor union	0.64
Promotion Opportunities	Equals 1 if respondent believes his or her current employment offers promotion opportunities, 0 if he or she does not	0.58
Job Tenure	Length of time served in current job (in years)	5.41
Small Firm Size	Equals 1 if respondent works for a firm with 1–24 employees	0.20
Medium Firm Size	Equals 1 if respondent works for a firm with 25–299 employees	0.39
Large Firm Size	Equals 1 if respondent works for firm with 200 or more employees	0.41
Occupational Dummies	Occupation dummies at the one-digit level
Industry Dummies	Industry dummies at the one-digit level
Regional dummies	Equals 1 if respondent lives in Inner London, Outer London, Rest of South East, South West, East Anglia, East Midlands, West Midlands Conurb, Rest of West Manchester, Greater Manchester, Merseyside, Rest of North West, South Yorkshire, West Yorkshire, Rest of Yorkshire and Humber, Tyne and Wear, Rest of North, Wales, Scotland, or Northern Ireland

A3

Number of Observations of Lags and Leads

	Transition to Preference Matching
Men	Women
Leads
1–2 years hence	139	118
Within the next year	149	148
Lags
0–1 years	136	150
1–2 years	130	125
2–3 years	117	105
3–4 years	101	93
4–5 years	99	86
5 or more years	478	442