Does politics crowd out professional competence? The organisation of ministerial advice in Denmark and Sweden

Abstract

The use of politically appointed ministerial advisors has increased noticeably in many Western countries, but we know little about how this development has affected the civil servants recruited on merit. The article asks whether political appointees accentuate or blur the line between politics and administration. Do political appointees take over political-tactical advice and leave policy advice to the permanent civil service, or do they cause permanent civil servants to be even more influenced by political considerations? And do political appointees make it easier or more difficult for the permanent civil service to be politically responsive? A Most Similar Systems Design comparison of Denmark and Sweden allows an assessment of the effects of political appointees. It is found that a large number of political appointees decreases functional politicisation of the permanent civil service; that functional politicisation tends to crowd out tasks related to more classic policy advice; and that functional politicisation increases political responsiveness.

Keywords:

• Political advisers
• political appointees
• Denmark
• Sweden
• politicisation

Notes

1. One reason for the very high Danish response rate was that the survey was endorsed by the permanent state secretary of the Danish Prime Minister’s office. There was no such endorsement in Sweden.

2. Chief executives in Denmark are defined as permanent state secretaries and vice state secretaries (afdelingschefer). In Sweden they are defined as departmental executives (departementsråd).

3. The five largest departments in Denmark are the Ministries of Business and Growth, Finance, Foreign Affairs, Justice and Taxation. The five largest departments in Sweden are the Ministries of Finance, Foreign Affairs, Health and Social Affairs, Justice and Enterprise, Energy and Communications.

4. Two of our dependent variables are on an ordinal scale, but because it is easier to interpret the coefficients, we use XTMIXED in STATA 13.1. When we use the ordinal command, MEOLOGIT, the results are robust.