Why demotion of older workers is a no-go area for managers

Abstract

Demotion – the reduction of an employee’s rank and salary – is often mentioned by managers and policy-makers as a means of increasing the employability of older workers in an ageing labour force. However, so far in practice demotion is rarely applied. This paper is the first empirical investigation of how managers perceive demotion as an instrument of human resource management. By means of a survey and a vignette study among managers in the Netherlands (N = 355), we examine whether managers consider demotion of poorly performing older workers a fair solution. Three contributions stand out. First, based on attribution theory we find support for the hypothesis that managers judge demotion to be fair in those cases where deterioration in task performance is caused by controllable factors (such as work motivation) and unfair when the causes are uncontrollable (such as age). Second, the expectations of managers about the organization-wide consequences of introducing demotion as a human resource policy play a significant role in considering demotion. Most managers perceive negative organizational externalities (e.g. decrease in loyalty and motivation of staff) to arise when introducing demotion and are reluctant to apply demotion in practice. And a third contribution: positive (negative) beliefs of managers about the hard skills – e.g. creativity, willingness to learn, flexibility – of older workers make demotion less (respectively more) likely.

Keywords:

• Older workers
• demotion
• productivity
• managers
• stereotypes
• vignette study

JEL codes:

• M51
• M54
• J62

Acknowledgement

Comments by Douglas Hershey, Tinka van Vuuren, Mo Wang, Arjen van Witteloostuijn, Hannes Zacher and participants of seminars of the Catholic University of Leuven, at the University of Copenhagen and at the European Population Conference in Budapest and three anonymous referees of this journal are gratefully acknowledged.

Notes

1. Vignette studies are also known as conjoint analysis or factorial studies and are closely related to so-called ‘policy capturing studies’ (see e.g. Rotundo & Sackett, 2002).

2. We carried out a confirmatory factor analysis in order to examine the construct validity of the two stereotype dimensions. The results show that items were all significantly loaded on their respective latent factors. The information criteria for the two factor model were also obtained (AIC = 6067.8; BIC = 6187.9; RMSEA = .087). An alternative one-factor model was also specified by loading all items on the same latent factor (AIC = 6140.7; BIC = 62.56.9; RMSEA = .116). All three information criteria indicated a better fit for the two factor model.