ABSTRACT
This article examines the relationship between citizen expectations, citizen satisfaction, agency reputation and agency behaviour in public management. We expand upon the assumptions of the expectancy disconfirmation model (EDM) and present a game-theoretic model that explores the interplay between agency reputation, citizen expectations, citizen satisfaction and agency behaviour by integrating perspectives from reputation and commitment problems literature. Our study highlights the uncertainty generated by bureaucracies with misaligned motives, which occasionally prioritize their own interests over the public’s. We delve into the dynamics of agency and citizen behaviour over time, considering potential shifts in agency priorities. We also explore how citizen satisfaction conditions future expectations and influences the behaviour of service providers in improving public service performance. We show that agencies with misaligned priorities with the citizenry provide poorer current performance when past citizen satisfaction is greater. Yet, from the citizens’ perspective, better past citizen satisfaction is associated with higher current expected service quality. Our findings complement existing literature on citizen satisfaction and expand our understanding of how satisfaction affects government service providers. They underscore the importance of addressing misaligned motives within bureaucracies and the role of citizen satisfaction in shaping agency behaviour. These results have important implications for bureaucratic behaviour, administrative burden and organizational reputation theories and practices.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1. See the model section for a more detailed discussion on this connection.
2. As mentioned above, the model is borrowed from Phelan (Citation2006) and applied to the current context. We include the relevant formal proofs required for the arguments below in the appendix, but refer the reader to that paper for an in-depth formal treatment, including the proof of uniqueness (Theorem 8).
3. We use a dichotomous variable for service quality assessment due to mathematical tractability and interpretability. A dichotomous variable clearly differentiates between high- and low-quality services and provides clarity in presentation. Using a continuous service assessment would induce greater realism by making changes in citizens’ beliefs smoother and not discontinuous, but would not qualitatively change our results.
4. More formally, assume a unit measure of non-atomic citizens indexed by . Let
be the probability of citizen
using the service. We can define the fraction of the population which uses the public service as
.
5. We are modelling commitment A as a nonstrategic action type but interpret this as a proxy for a strategic payoff type which strictly prefers to exert effort. e.g. suppose for a given effort level and
fraction of citizens utilizing the service, commitment A pays the cost
, where
. This means A faces higher costs for not exerting effort, hence strictly prefers exerting maximum effort. Further assume that, akin to the belief restriction in Kreps and Wilson (Citation1982, 265), citizens experiencing low-quality service would not think that A is more likely to be a commitment type. We would obtain the same equilibrium dynamics with this setup. Our choice of modelling A as an action type follows the standard approach to reputations in economics since Kreps and Wilson (Citation1982), serves to simplify the analysis, economizes on notation, and facilitates the focus on the behaviour of the non-commitment type. Relatedly, see Weinstein and Yildiz (Citation2013) for the proof that any infinitely repeated game with commitment types is strategically equivalent to a game with incomplete information about the stage-game payoffs.
6. Since we have a unit measure of identical citizens, a given citizen’s probability of using the public service equals the fraction of the population using the service.
7. See the Appendix for a formal description of Markovian equilibria, including the formal expressions for, and a discussion on, an individual citizen’s expected payoffs and .
8. This is a simplifying assumption to lend the model mathematical tractability. We could have allowed the commitment type to make occasional mistakes by assuming that it also provides low-quality service occasionally due to factors outside of its control. Then citizens would still lower the agency’s reputation at the end of the period after observing poor performance, but these revisions would not be as dramatic as in here. This would not, however, change the qualitative dynamics in equilibrium and would complicate the analysis and presentation.
9. Citizens do not directly observe the Agency’s effort on how the reputation affects its behaviour in improving or not the service quality, hence we use dotted lines. The change from t to t + 1 signifies the change in reputation in time.
10. See Phelan (Citation2006, 40) for the discussion on starting from any other prior would yield the same equilibrium dynamics.
11. Generically A’s reputation will not be exactly equal to at the Nth step. Therefore, we ignore that knife-edge scenario and maintain that A’s reputation strictly exceeds
at the Nth step.