The differential effects of politicized appointments on input and output legitimacy: a survey experiment with elite and citizen respondent

According to several observers, public bureaucracies are increasingly politicized as individuals with a party-political background are recruitment into leading positions in the public bureaucracy. Concerns regarding politicized recruitment include its potential for driving out competence, colouring the policy advice bureaucrats give to political executives, and negatively affecting the performance of public agencies. Despite a growing empirical literature and theoretical advances, we are only starting to understand the effects of politicized bureaucracies, especially in traditional meritocratic systems. In this paper we focus on the relationship between the appointment of top civil servants on the one hand and the input and output legitimacy of government on the other hand. How does the politicization of appointments affect input and output legitimacy, and are they affected in an equal manner? We hypothesize that input legitimacy (views on qualifications and the quality of policy advice) is more negatively affected by politicization than output legitimacy (views on responsive policy implementation). Next, is the effect of politicization of appointments conditioned by the appointee’s traditional merit qualifications? We hypothesize that negative effects on legitimacy are stronger if political appointees lack merit qualifications. Finally, do elite and citizen respondents perceive the dangers of politicization in the same way? We hypothesize that government elites (politicians and civil servants) have a more accommodating view of politicization than citizens and journalists. We test these hypotheses with unique data from a survey experiment with four groups of respondents in Norway: citizens, journalists, bureaucrats and politicians. The panel of citizens is based on a probability sample of the Norwegian population, the panel of politicians includes actors from national, regional and local level; the panel of bureaucrats includes ministry and agency officials; and the journalist panel includes respondents working for both national and local media outlets. The experimental data have been collected, the experiment was pre-registered, and preliminary analyses confirm most of our expectations.