A multilevel analysis of job satisfaction in Britain - WERS 2004

1. IntroductionEconomists ventured into studying issues of subjective well-being and/or jobsatisfaction somewhat later than psychologists and sociologists. None the less,they have been fascinated by issues surrounding job satisfaction for it hasimportant economic consequences. In the economics circle job satisfaction hastraditionally been regarded as an important predictor of such labour marketbehaviours of workers as quit and absenteeism (e.g. Hammermesh 1997,Freeman 1978, Akerlof et al. 1988, Clark et al. 1998, Shields & Price 2002,Kristensen & Westergard-Nielsen 2004, Levy-Garbous et al. 2005). Both ineconomics and other areas of social research job satisfaction has also been linkedto individual well-being (Argyle 1989, Clark 1997, Sousa-Poza and Sousa-Poza2001, Kahneman & Krueger 2006), job performance and productivity (e.g.Iaffaldano & Muchinsky 1985), organisational performance (Ostroff 1992) andinnovation (Shipton, et al. 2006).There has been a growing interest in establishing the determinants of jobsatisfaction and many have argued as such (Hammermesh 2001). While mostprevious studies sought to relate job satisfaction to a host of worker and jobattributes (e.g. Clark 1996, Gazioglu and Tansel 2006), others have paid muchattention to some specific correlates of job satisfaction. These include, amongothers, job satisfaction and gender (Clark 1997, Bender et al. 2005), jobsatisfaction and age (Clark et al. 1996), job satisfaction and wage (Borjas 1979,Clark et al. 1998, Clark 1999, Lydon & Chevalier 2002), job satisfaction andrelative income (Clark and Oswald 1994, 1996), job satisfaction and unions(Borjas 1979, Berger et al. 1983, Bryson et al. 2004; 2006), job satisfaction andwork environment (Idson 1990), job satisfaction and work relations andmanagement (Gazioglu & Tansel 2004), job satisfaction and racial composition(Maume & Sebastian 2007).Though there exist such a long list of research into different aspects of jobsatisfaction and/or its consequences, there still exists a room for furthering ourunderstanding of the determinants of job satisfaction in particular. At least twoimportant reasons can be cited in this respect. First, almost all of the researchinto the determinants of job satisfaction fails to take into account unobservedindividual-level and/or workplace-level heterogeneity. 1 One important lesson thatresearch in labour economics, particularly where there is a linked/matchedemployer-employee data, underscores is the importance of such unobservedfactors in determining labour market outcomes (see, for example, Abowed et al.1999). 2 Job satisfaction as an outcome may not be too different in this regard,and one needs to account for unmeasured heterogeneity to address such issues as1 The few exceptions include Winkelmann & Winkelmann (1998), Clark et al. (2006), whoaccount for unmeasured effects using longitudinal data, and Bryson et al. (2004, 2006) who usethe 1998 sweep of WERS.2 Abowed et al. (1999) have used a matched longitudinal data. In WERS it is workplaces, and notemployees, that are followed in each of the surveys. Given this, the multilevel model offers thebest opportunity to account for employee-level and workplace-level heterogeneity in the contextof the cross-section (WERS 2004) data this study uses.3