Accountability – are We Better off Without It?

Abstract

Public managers increasingly lament about negative effects of accountability. Despite these reports and the importance of accountability in public organizations, it is yet unknown when it has positive or negative effects on managers. Overcoming two major obstacles in accountability research, this study therefore investigates ‘how accountability affects public manager's work behaviour’. Firstly, this study applies a cognitive theory offering a promising way out of the current theoretical dead-end. Secondly, the quasi-experimental research design makes the complex concept of accountability more tangible and findings transferable to practice. Moreover, individual level characteristics such as motivation are used to investigate a potential interaction between contingency and individual level characteristics. Results indicate both positive and negative effects of accountability on work behaviour and hence strongly challenge the positive normative connotation of accountability.

Key words:

• Accountability
• public manager
• new public management
• quasi-experimental

ACKNOWLEDGEMENTS

For most helpful comments, their support and advice while writing this paper (which is based upon a thesis written during the author's enrolment in the Research Master at the Utrecht School of Governance), I would like to thank Wouter Vandenabeele, Thomas Schillemans, Mark Bovens, Mirko Noordegraaf, Martin Noack, Daniela Behr, Andreas Tiemann and all participants of the European Group for Public Administration (EGPA) permanent study group on public sector performance in Toulouse, September 2010, as well as the participants of the accountability workshop at Utrecht School of Governance in November 2010. Moreover, I highly appreciate the constructive comments by the two anonymous reviewers of Public Management Review.

Notes

For a more detailed overview on the concepts of results, output, outcomes and impact, please see Bouckaert and Halligan (2008).

The focus on these two accountability mechanisms allows to go beyond ‘quantity’. Scholars so far have focused on the effects of the amount of accountability, introducing terms such as ‘accountability overload’ and ‘accountability deficit’ (Bovens, 2007). This study, however, analyses the effects of qualitatively different characteristics of accountability on managers’ work behaviour.

From now on, for reasons of parsimony this study refers to the term work behaviour.

Please note that the term political as used in the concept PB as it has been utilized in social psychology is not to be confused with political as applied in the field of political science. In the specific context of this study it rather refers to ‘pleasing’ behaviour.

As outlined above, the normative terms desirable and undesirable behaviours rest on the assumption that PSB leads to better performance while PB exerts a negative effect on performance.

One participant only filled out two out of the three vignettes.

All survey items cited here are freely translated from Dutch.

Not that while this situation includes several stimuli (restructuring, no knowledge on supervisor preferences, etc.), these are held constant across the three vignettes.

Both CFA models have been estimated using maximum likelihood estimators. These have been shown to yield effective as well as efficient results with ordinally scaled variables. Furthermore, Maximum Likelihood (ML) estimation is robust against partial non-normal distribution of the variables (Kline, 2004). To confirm the robustness of ML estimation against non-normality, the unweighted least squared estimation method has been applied, as well. While the parameter estimates are naturally the same, the fit indices are also in line with ML-findings.

If we were interested in predicting the dependent construct as a whole, a high R ² would have been mandatory. However, this study aims to determine the unique effects of accountability and motivation on work effort independent from other potential predictors (e.g. salary). Therefore, the R ² is of secondary interest.

However, the multiplicative nature of the interaction terms generally leads to an overestimation of the standard errors, causing a too frequent rejection of the effect (Jaccard, 2003).

PSB II's predictive power even decreases by .5 per cent.

This reasoning would run against the line of argument put forward here.

A more detailed interpretation is refrained from given that it would go beyond the statistical results produced.

This finding indicates the potential to improve work behaviour by satisfying, for example, managers’ basic needs, therewith increasing their autonomous motivation.

Nevertheless, the importance of this comparison is of course acknowledged and recommended for future assessment.

As Achen already argued in 1977 (806), ‘replac[ing] the unmeasurable by the unmeaningful is not progress’.

In less knowledge-intensive environments, bureaucrats could, for example, be supposed to engage in less (reflective) PSB in order to increase work pace.