The purpose of this paper is to present and discuss empirical findings concerning flexible work-time patterns, overtime and outcome among two groups of so-called knowledge workers: IT-consultants and university staff. Among the questions asked are to what extent contemporary so-called knowledge workers really are able to choose their working hours, if they make use of such choices, and if that really is a good thing. In a broader sense, we question whether it actually is reasonable to say that they have flexible jobs and that this is due to firm-level responses to increased uncertainty.
Using survey data of individual workers in two work organisations, we are able to link the ability to choose when to work (‘flexibility as autonomy/discretion’) with actual work-time patterns (‘flexibility as (non-)standard work arrangements’) at the individual level. This makes it possible to distinguish four groups of workers having either forced stability, forced flexibility, chosen flexibility or chosen stability. We further relate this information to particularly data on overtime, but also aspects of the organization and labour process, as well as e.g. health and sickness absenteeism/presenteeism.
Our overall empirical findings are hardly revolutionary in themselves and in line with much previous research (e.g. Sandberg & Movitz et al 2005; Allvin et al 2006; Movitz & Allvin 2012) – far from all employees resemble the popular notion of flexible workers but those that do have flexible work arrangements tend to work a lot of unpaid overtime even though they from e.g. a health perspective would benefit from having more normal and especially stable work hours.
Based on the findings and related to labour process theory, we however develop a theoretical discussion where we, first, contribute to the critique of the common environmental deterministic/contingency theoretical claim (which unfortunately at times is also found in labour process studies) that flexible working conditions results from straightforward strategic management decisions to increase firm-level flexibility in response to increased environmental uncertainty. Our argument is that this is a downward conflation that takes contemporary discourses – as well as workers phenomenological understandings (Sayer 1992) – at face value, overestimates the coherence and strategic capabilities of management (Hyman 1987) and leads to teleological and function explanations overshadowing the inherent exploitation of workers (Armstrong 1987).
Second, we criticize the simplistic assumptions that employees will actually use formal opportunities to choose working time and divert from standard working hours in similar ways (or at all), or in practice have the ability to do so (compare Kalleberg 2000). Workers naturally cannot be treated as uniform and merely automatically reacting to managerial strategies and rules in a similar fashion.
Third, we question the approach common in quantitative studies found in e.g. organizational psychology of simply viewing excessive overtime (as well as increased health risks) as a common and dependent outcome of flexible work arrangements since such co-variations at the empirical level decrease possibilities of detecting the real causal mechanisms (Sayer 2000; Fleetwood & Ackroyd 2004) and an understanding of the jobs and the labour process. We argue that it makes more sense to revise the image of certain jobs as flexible and instead understand them for what they are: essentially stable jobs with excessive workloads. When all slack is taken out of the system and workloads increase, the actual choice of when to work becomes rather artificial and those consistently working daytime and on top of that often work nights, weekends and vacations are hardly flexible – they simply work a lot. Perceived this way, flexible workers only make up a small minority of those we have studied, even though it is a group selected because it is has the most flexible conditions on the Swedish labour market (Allvin 2012).
Our main conclusion is not simply that future research can ‘correct’ this by altering the ‘operationalisation’ of flexibility in surveys so that overtime is taken into account, or include overtime as a mediating variable in statistical models in order to increase predictive powers. Instead, we argue that using the theoretical approach we sketch out above will contribute to labour process theory by increasing our understanding of what such jobs are like and the mechanisms causing them to look the way they do.