The Political Power of Bureaucrats: Not only Agenda Shaping, but also Preference Shaping?

A strong civil service is vital in a democracy. Skilled bureaucrats with specialized knowledge are necessary for politicians. Effective political control builds upon a bureaucracy that can prepare and implement new policies. The more skilled and specialized they are, the more potentially valuable for politicians. However, skill and knowledge leave bureaucrats with political power. A good bureaucrat is a case of a loyal Dr Jekyll and a powerful Mr Hyde with his own agenda. This dilemma is a classic in public administration. Max Weber worried about the political dilettante facing the bureaucratic expert. The early public administration literature (Gulick; Simon) confirmed these worries, and gave rise to theories about bureaucrats’ ability to influence the political agenda (Niskanen, 1973). The worries were supported by studies showing that politicians and top bureaucrats are everywhere in close dialogue about policy formulation, both at central and local government levels (Aberbach, Putnam & Rockman). We argue that this classic literature, however pessimistic it is, underestimates the power potential of the bureaucracy. The bureaucrats’ exclusive access to politicians can be used for more than agenda influence. It can also be used to influence political preferences. We make this argument by applying insights from political psychology (framing theory, prospect theory and motivated reasoning) about the importance of information for attitude formation to the relationship between bureaucrats and politicians. To test the argument, we study politicians in four different political systems: USA, Italy, Belgium and Denmark. We use experiments embedded in surveys to (i) assess the importance of Aberbach, Putnam & Rockman’s “intense dialogue” between bureaucrats and politicians today; (ii) investigate the importance of the bureaucracy is a provider of information; (iii) test whether bureaucrats can influence their politicians’ preferences by strategically framing policy proposals; (iv) by exploiting that politicians are risk averse and place different value on potential losses and gains; and (v) by exploiting that politicians tend to prefer information sources which they ideologically agree with.