A considerable amount of research on labor turnover and job search behavior exists in economics (for example, Belzil 1996; Farber 1994; Lillien and Hall 1986; Parsons 1991; Schettkat 1996; Veum 1997). This is not surprising since labor turnover and job search behavior are essential ingredients in an imperfect world where employers' and employees' characteristics are not known to all: only well-directed mobility can assure that anything resembling an optimal allocation of jobs to workers arises. A topic strongly related to job mobility is "turnover intentions." Turnover intention reflects the (subjective) probability that an individual will change his or her job within a certain time period. As opposed to labor turnover, turnover intentions are not definite. And, although turnover intentions are often associated with job search behavior, this need not always be the case. (1)
So why is the analysis of turnover intentions important? First, it should be noted that a vast literature on the analysis of turnover intentions exists in psychology (e.g., Cohn 2000; George and Jones 1996; Hom et al. 1992; Mobley 1977; Sager et al. 1998; Wright and Cropanzano 1998). Psychologists argue that voluntary labor turnover is an important topic since such movements represent potential costs to organizations in terms of loss of valuable human resources and the disruption of ongoing activities (Cascio 1991). Economists generally accept this point although they often tend to stress the benefits of labor turnover. For psychologists, turnover intentions are accordingly important, since such intentions are the immediate precursors to actual turnover. As is well and extensively documented in the psychology literature, a close relationship between the intention to quit and actual turnover exists (e.g., Mobley 1977; Mobley et al. 1979; Price and Mueller 1986; Rusbult and Farrell 1983; Steers and Mowday 1981). Thus, these intentions are good at forecasting actual quits (e.g., Mobley et al. 1979; Sager et al. 1998; Steel and Ovalle 1984). (2)
There is also a second important reason for analyzing turnover intentions: to out knowledge, very few international comparative studies on actual turnovers exist. This is not surprising since there are severe data limitations in such cross-national studies. The main reason is that information on job fluctuations, when analyzed with microdata, usually require a panel, and cross-national panels are very rare. Quoting Simon Burgess, "Ideally, we would want to examine individual level surveys for a wide variety of countries, conducted on exactly the same basis and including the same rich set of conditioning variables. Sadly but unsurprisingly, this is not possible. (3) (1999). These data limitations can be avoided by analyzing job turnover intentions) Since turnover intentions and actual (primarily voluntary) turnovers are strongly correlated, such an approach presents an interesting alternative for analyzing job mobility in an international, comparative way.
Four cross-national studies on actual turnovers which we are aware of are those of J. van Ours (1990), R. Schettkat (1997), and S. Burgess (1994, 1999). Van Ours used aggregate pooled cross-sectional time-series data from various sources in order to analyze job mobility in six OECD countries (France, Japan, the Netherlands, Sweden, United Kingdom, and United States). He showed that job mobility is positively correlated with the growth of employment and negatively correlated with the unemployment rate. Furthermore, the results appear to indicate that no structural differences in job mobility between the USA, Sweden, France, and the UK exist, whereas job mobility in the Netherlands and Japan is structural[y lower than in the other countries. Schettkat used aggregate data from the European Union's Labor Force Survey, with which he analyzed job mobility in the years 1982-83 and 1987-88 in twelve EU countries. The main aim of this study is to capture the effects of labor market regulations on labor mobility. He concluded that "there are clear indications of the impact of national regulations on employment stability, but at the same time variations in stability are observed which can hardly be explained by changes in labor market regulations" (117). He also showed that industrial structure and the macroeconomic situation influence job mobility. Burgess (1999) used aggregate data from different national surveys on elapsed job tenure in ten countries (France, Germany, Holland, Italy, Japan, Poland, Spain, Sweden, United Kingdom, and United States). These data are used to characterize the nature of labor reallocation and to try to isolate the effect of country-specific factors. Emphasis is placed on the role of employment protection legislation (EPL) and the trainability of the workforce. He found that both of these have a significant role to play in affecting the reallocation of labor. Burgess (1994) used aggregate data (disaggregated by industry) on employment from the OECD International Sectoral Database for ten countries (Belgium, Denmark, France, Germany, Great Britain, Holland, Italy, Japan, and Sweden). The main aim of this paper is to analyze the relationship between the tightness of employment protection legislation (EPL) and job mobility. The results provide some support for the idea of a link between EPL and the speed of labor reallocation.
All these studies have data limitations in the sense that only a limited number of countries are considered and/or different national data sets are used, tending to make comparisons difficult and only possible at an aggregate level. Samples sizes are also usually very small. Furthermore, and if at all, very few control variables are included.
The aim of this study is to analyze turnover intentions in a cross-national setting based on a single survey. With the aid of an interesting data set covering twenty-five countries, we believe we are able to make two contributions to the existing literature on job mobility: first, and as was mentioned above, although a few studies on cross-national research on actual turnovers exist, they all face data limitations. Out data set, besides covering numerous countries, includes a rich set of control variables, not only with regard to demographic and job-related characteristics but also with regard to job perceptions. Especially perceptions such as job security and job satisfaction are seldom considered in empirical mobility studies, despite their obvious and accepted importance. Second, we are unaware of such a detailed cross-national analysis of turnover intentions, neither in the economics nor in the psychology literature. (4) We believe that such a comprehensive analysis of turnover intentions is not only interesting in its own right but it is most probably the only feasible way for obtaining rich, comparable, and microdata-based information on job mobility from a single survey covering so many countries.
The Determinants of Turnover Intentions
Economists have little to say about the direct determinants of turnover intentions since this topic usually falls into the realm of psychology. Since, however, actual turnovers follow directly from turnover intentions, economic theory on labor flows should also tell us something on turnover intentions. Based on numerous theoretical perspectives (e.g., human-capital theory, search theory, matching theory, and labor market segmentation theory), economists have identified several determinants of job turnovers. It is beyond the scope of this paper to comprehensively discuss all possible determinants. Instead, we will focus on variables which will be incorporated in the subsequent empirical analysis. Most of these variables are usually encountered in labor-turnover studies or in psychological studies on turnover intentions.
Gender has been shown to influence actual turnovers (e.g., Blau and Kahn 1981; Royalty 1998). It is sometimes observed that, depending on the level of education, women are less likely to change jobs (see Royalty 1998). Furthermore, in some countries, women have higher levels of job satisfaction, which generally reduces job-change inclinations (see Sousa-Poza and Sousa-Poza 2000a). Booth and Francesconi (1999) round no large differences in job-to-job mobility between genders. Booth et al. (1999) showed, however, that, for older cohorts, job-quitting behavior is more pronounced for men, while leaving a job for other reasons is more common for women.
Age has been shown to be negatively correlated with the probability of changing a job (e.g., Campbell 1997; Kidd 1991, 1994). The main argument for this observation is that the available time to amortizise the costs associated with a job change diminishes with age, thus making a job change less attractive (see also Shapiro and Sandell 1985).
Marriage could have a negative effect on the probability of changing a job, since it is usually more costly if a family (as opposed to an individual) has to move (see Holmlund 1984). Furthermore, due to the traditional gender roles, married and employed women are more likely to be less mobile than corresponding men. The empirical evidence is, however, not very robust (see, for example, Sicherman 1990; Zimmermann 1998).
It is sometimes assumed that the level of education has a positive effect on the probability of changing jobs since a high education is often associated with better labor-market alternatives (e.g., Royalty 1998). Most studies, however, do not reveal a significant correlation (e.g., Booth and Francesconi 1999; Campbell 1997).
Working time may influence job-to-job mobility in a positive manner since lower working hours (and especially part-time employment) could imply that a worker is less integrated in a firm (e.g., Garcia-Serrano 1998). It is, however, also conceivable that long working hours (often in the form of involuntary overtime) may also increase the desire to change one's job. Thus, a U-shaped...