The demand of part-time in European companies: a multilevel modelling approach

Abstract

Part-time work is one of the most well-known ‘atypical’ working time arrangements. In contrast to previous studies focusing on the supply side, the originality of our research is to investigate the demand-side of part-time work and to examine how and why companies use part-time work. Based on a large and unique sample of European firms operating in 21 member states, we use a multilevel multinomial modelling in a Bayesian environment. Our results suggest that the variations in the extent of part-time workers at the establishment level is determined more by country-specific features than by industry-specific factors.

Notes

¹ Including flexible working hours, overtime, part-time work, work at unusual hours (shift work, night work and weekend work), parental leave and phased and early retirement. A common feature of all these time arrangements is that they deviate from the standard working time regime (i.e. full time hours, Monday to Friday) with respect to duration, distribution or timing of working time.

² The only sectors excluded are ‘Agriculture’, ‘Forestry’, ‘Private households’ and ‘Extraterritorial organizations’. In these sectors the number of establishments with 10 or more employees is negligible in the countries surveyed.

³ TNS infratest Sozialforshung, Munich coordinated the fieldwork.

⁴ In the ESWT part-time employment is defined as ‘less than the usual full-time hours’.

⁵ This descriptive part relies heavily on a report commissioned by the European Foundation for the Improvement of Working and Living Conditions, see Anxo et al. (2007).

⁶ As mentioned above the distribution of firms according to working time is as follows: 46% of the surveyed establishments do not make use of part-time workers, 40% of establishment displays a low incidence of part-time work (less than 20% of the work force) and 24% of establishments have a high incidence of workers working part-time (20% of the work force or more). Furthermore, the multilevel multinomial model using these cutting points has shown to fit the data better than the more aggregate binomial model (including only full time and part time) and a number of other multinomial models with different cutting points.

⁷ i.e. low incidence of part-time workers (s = 1) and high incidence of part-time workers (s = 2).

⁸ The ICC gives the proportion of variance attributable to between group differences (in our case countries or sectors within countries).