رضایت شغلی و درآمد نسبی در گذار اقتصادی : وضعیت یا سیگنال؟ در مورد مناطق شهری چین

We use two datasets for urban China to examine whether an increase in reference group income lowers or increases job satisfaction. The former is consistent with a status effect — an increase in the income of others lowers my satisfaction because I feel jealous. The latter is consistent with a signal effect — an increase in the income of others might make me jealous, but it also provides an information signal about my future prospects. When we use a single item indicator of job satisfaction we find no support for a status or signal effect; however, when we use a psychometrically valid instrument to measure job satisfaction, we find some support for the existence of a status effect. We consider the components of job satisfaction through which the status effect operates. We find that the status effect operates through satisfaction with co-workers, operating procedures, pay and supervision.

A large economics literature on the determinants of subjective well-being has focused on the role of relative income in explaining job or life satisfaction (see Clark, Frijters, & Shields, 2008 for a survey). This literature has generally concluded that relative income is important for determining job and life satisfaction. An important implication of this result is that the neoclassical utility function, in which one's utility is a function of the individual's own consumption or income, should be extended to incorporate relative consumption or income. The relationship between relative income and job satisfaction is important for several reasons. First, within the firm the link between relative income and job satisfaction is important for human resource managers. Job satisfaction has been shown to be associated with absenteeism (Leicht & Shapelak, 1994); labour turnover (Crampton & Wagner, 1994) and productivity (Harder, 1992). Second, income comparisons are important for the functioning of the labour market and public policy. Concern about relative status affects the supply of labour and wage profiles (Frank, 1984); savings (Kosicki, 1987) and optimal tax policy (Oswald, 1983). Income comparisons are relevant for whether policies should be put in place to smooth income distribution (Luttmer, 2005). If relative income matters, the case for economic growth becomes less clear and the case for progressive taxation is strengthened (Oswald, 1983). Akerlof and Yellen (1990) argue that involuntary unemployment and other macroeconomic phenomena can be explained in a model in which relative income matters. Third, job satisfaction is positively correlated with life satisfaction (see eg. Knight, Song, & Gunatilaka, 2009). Thus studying the relationship between relative income and job satisfaction is relevant to the broader economics literature examining the economics of happiness. Conceptually relative income could have a negative or positive effect on job satisfaction. The former refers to a status effect, while the latter refers to a signal effect. The status effect is where the higher earnings of my reference group make me jealous, lowering my sense of well-being. The signalling effect is where the higher earnings of my reference group increase my sense of well-being: the more others earn, the happier I am. The signalling effect is consistent with Hirschman (1973) tunnel effect — while others wage increases might make me jealous, it also provides information about my own future prospects. If others are doing well, this acts as a signal that I too have better prospects of doing well in the near future. To this point most studies for stable industrialized capitalist studies have found evidence consistent with a status effect. However, Clark et al., 2009 and Panos and Theodossiou, 2007 find support for the signal effect using Danish and British panel data respectively. It is likely that there will be more evidence of the signal effect in transitional economies, than stable western industrialized economies. Senik, 2004 and Senik, 2008 finds evidence of a signalling effect in transitional economies. In an unstable labour market, as is often the case in transitional economies, what happens to others who have similar characteristics to me today might be thought of as providing a signal about my own future labour outcomes. In this respect, Senik (2008) argues that the respective importance of status and signalling effects depend on the level of economic uncertainty and labour market mobility. China represents an interesting case study to examine the signalling and status effects of relative income on job satisfaction. Senik (2004, p. 2100) describes transition in Russia “as a natural experiment characterized by an unusually high variance in absolute and relative incomes”. This is also true for China. China's market reforms have created tremendous opportunities for people to climb the labour market ladder, not only through fast wealth creation in the non-state sector, but also through more flexible labour markets in the state sector. At the same time, China's market reforms have generated marked increases in income inequality. We examine the effect of relative income on job satisfaction in urban China using two datasets. Both have their advantages and disadvantages. The first is matched employer–employee data from Shanghai for a sample of 784 employees in 78 firms. The advantage of this dataset is that earnings of all workers in the same establishment can be used to examine the role of reference income. Others working in the same establishment is arguably a more appropriate reference group to examine the effect of relative income on job satisfaction than other potential peer groups, such as friends, neighbours or those living in the same city, given that what others in the same establishment earn will be a better signal of my future earning capacity than what my neighbours can earn. A disadvantage of the Shanghai sample, however, is that it is relatively small and to measure worker well-being it uses a single item indicator of job satisfaction in the form, ‘how satisfied are you with your job’? While economists have almost universally used single item indicators of job satisfaction of this form to measure worker well-being and this research has been published in the leading journals (see eg. Clark et al., 2009), such an approach has been criticized in the psychology literature on job satisfaction, in which multi-item indicators have long been widely used on two grounds. The first is that the researcher cannot estimate the internal consistency of a single item indicator, with the result being that such indicators are subject to low levels of internal reliability (see Oshagbemi, 1999, Pollard, 1996 and Wanous et al., 1997). Andrews and Whithey (1976) found single item indicators of personal well-being to have relatively low reliabilities (test–retest correlation 0.40–0.66), even when asked twice in the same session 1 h apart. Test–retest correlations for multi-item indicators tend to be much higher — in the range 0.82 to 0.84 for temporal intervals of up to 10 weeks (Krueger & Schkade, 2007). The second is that single item indicators are not able to capture the multidimensionality of psychological constructs and hence construct validity is compromised. The second sample that we employ is from a survey of job satisfaction among approximately 2800 employees across six Chinese cities; namely, Chengdu, Dalian, Fushun, Fuxin, Fuzhou and Wuhan. The advantage of this dataset is that the sample size is larger and, to measure job satisfaction, it contains data from the Job Satisfaction Survey (JSS) (Spector, 1997). The JSS, which is a 36 item multidimensional measure, subsumes nine components of job satisfaction and is one of the most widely used instruments to measure job satisfaction in the psychology literature. The disadvantage of this dataset is that it does not contain information on average establishment earnings in the firm in which the respondent works. Thus, to measure relative income, we adopt two approaches. One approach, following Clark and Oswald (1996), among others, is to estimate a conventional earnings equation on the whole cross-section of employees and then use this regression equation to predict an earnings level for each person. These predicted earnings levels correspond to the income of ‘typical’ employees with given characteristics. The other approach, following Blanchflower and Oswald, 2004 and Luttmer, 2005, among others, is use a geographical definition of reference group.

China's market reforms have resulted in a substantial improvement in the living standards of most of the urban population. However, at the same time they have also generated marked increases in income inequality. And while many have benefited from the marketization process there are also those who have been left behind, who have become increasingly vocal in criticizing rising income inequality and reminding the state of its socialist claim to legitimacy and of promises of egalitarianism made during the Maoist past. The findings reported here are evidence that China's urban population are concerned about rising income inequality and that the signalling effect of higher wages has not been sufficient to outweigh the status effect of relative deprivation with rising reference group income. One of the contributions of this study has been to highlight the need to examine the effects of changes in reference group income on different components of job satisfaction, using a psychometrically valid instrument. The results suggest that the status effect operates through some components of job satisfaction associated with a more competitive work environment, but not others. A potential limitation of this study is that we do not control for personality traits. One approach to controlling for personality traits has been to use panel data, which we do not have. The other approach has been to control for personality traits by including variables to measure attitudes on social issues (Smyth, Mishra, & Qian, 2008), mood (Knight et al., 2009) or mental health indicators (Ferrer-i-Carbonell & Gowdy, 2007). However, none of these variables accurately depict personality traits as they are conceptualized in the psychology literature. Future research should control for stable aspects of the personality, such as the locus of control and dispositional optimism. Locus of control, defined by Rotter (1966) as a person's perception of their control over event outcomes, has been found to impact on personal well-being in a range of cultural contexts (see eg., Garcia et al., 2002, Kulshrestha and Sen, 2006 and Spector et al., 2001). Dispositional optimism, defined by Scheier and Carver (1985) as the propensity to generally expect favourable outcomes over unfavourable ones, has also been demonstrated to co-vary with perceived well-being (see eg., Isaacowitz, 2004). We would encourage future studies to use psychometrically valid and reliable measures of these psychological constructs which capture the complex conceptualisation of these variables. Clark et al. (2009) emphasise the composition of the reference group and that finding a signal effect is more likely if the reference group is work colleagues than neighbours. A possible reason why we fail to find a signal effect is that our Shanghai sample with which we investigated the effect of establishment earnings on job satisfaction was too small. A useful avenue for future research would be to combine the use of the JSS (or some other psychometrically valid instrument to measure JSS) with matched employee–employer data for a large sample, in order to combine the advantages of both datasets employed here.