Job preferences as revealed by employee-initiated job changes

Abstract

Many previous studies try to discover job preferences by directly asking individuals. Since it is not sure whether answers to these surveys are relevant for actual behaviour, this empirical examination offers a new approach based on representative German data. Employees who quit their job and find a new one compare the two jobs with respect to eight job characteristics: type of work, pay, chances of promotion, work load, commuting time, work hour regulations, fringe benefits and security against loss of job. It is argued that the observation of many improvements (and few declines) for a certain attribute indicates a particular relevance and high preference for this attribute. It turns out that pay and type of work are most important for employees in this sense. Differences across subgroups of employees with respect to individual characteristics, such as sex and age, are explored. Those between East and West Germany diminish over time.

Keywords::

• job changes
• job characteristics
• job preferences
• quits

Acknowledgements

The author thanks Anja Göritz, Johannes Martin, Werner Nienhüser, Andreas Schmitt, Stefan Zimmermann and, in particular, Ingo Weller for helpful comments.

Notes

^1. Explicit tests of decision theories are only possible with some kind of experimental approaches (e.g. Huber, Daneshgar and Ford 1971) and not with such broad empirical studies using field data.

^2. Detailed information on the data and the questionnaire can be found at http://www.diw.de/english/soep/29012.html.

^3. We have, therefore, three values of each variable and cannot analyse the magnitude of changes in attributes. In principle, it would be possible to survey this for wages and commuting time, but less easy for other characteristics.

^4. There is information only for the reason of employees' job quit for single years of the 1980s unfortunately. Not surprisingly, there is some evidence that the fraction of improvements is somewhat higher for the subgroup of employees who quit their job and explicitly state that the reason was that they already have found a better job (Grund 2000).

^5. Commuting time may not be directly associated with work. Matiaske and Mellewigt (2001) provide more detailed evidence on the determinants of job satisfaction in Germany.

^6. Due to space limitations, the results for the job attribute dummies are not presented. In general, the results confirm the bivariate correlations of Table 1. Pseudo-R²s considerably increase. The results are robust with respect to another specification integrating the sum of changes of the seven other attributes instead of the set of dummy variables in the regression.

^7. This is in line with other direct estimations of job preferences such as the international study by Corrigall and Konrad (2006).