4. Furthermore, one organization is excluded because it is the supervisory board of another
organization in the list (Centrale Commissie voor de Statistiek), another one because it is an advisory board
without proper public authority (Commissie van advies inzake opleidingen en examens betreffende vervoer van
gevaarlijke stoffen door de lucht), and again another because it is not yet an independent agency, but still a
unit within the ministry (Nederlandse Emissieautoriteit).
5. As quite some organizations had a score of zero on the measures, I have actually calculated the
logarithms of (11the scores).
6. In the category of regulatory agencies, I have included organizations which regulate the activities of
organizations in the private sector, organizations which decide upon the market access of products
and services, and quasi-judicial bodies. The remaining organizations are categorized as executive or
service-providing agencies, and they discharge such tasks as distributing benefits and subsidies,
providing education and exams, administering public facilities, doing research on trends and
incidents, and registering products. One organization (College Sanering Zorginstellingen) clearly fell under
both categories as its two main tasks were to exercise oversight over reorganizations in the healthcare
sector (regulation) and to give subsidies to healthcare organizations which are closed down (serviceprovision).
As the exercise of oversight is the main task of the organization, it has been categorized as
a regulatory agency. However, I have also run the analysis without this organization, and this did not
change the results.
7. As the number of independent agencies per ministry vary considerably – ranging from two
organizations for the Ministry Social Affairs and Employment to 22 for the Ministry of Education,
Culture and Science – I have preferred the use of dummy variables for the ministries over the
estimation of a hierarchical model.
8. I have also run the analyses with the logged total amount of public funds which the agencies spend.
This does not change the results. However, as the data on the total amount of funds are missing for
many of the organizations, I only present the analyses with the dummy variable.
9. As a robustness check, the three models have also been estimated with a reduced dataset which
excludes the organizations which have one of the extreme scores on the formal political
accountability scale – that is, a score of zero (no provisions for accountability) or one (all provisions
for accountability). The results of these analyses are not substantively different, and the coefficients
only change slightly, but the p-values are overall a bit higher due to the lower number of
observations. As a consequence, the effect of government strength in Model 1 is no longer significant
at the 5 percent level.
10. The effect of the interaction terms was also estimated in three separate models, but no significant
impact of the interactions was found there either.
11. The mean scores on the formal political accountability scale per ministry are 0.48 for the Ministry of
Agriculture, Nature and Food Quality, 0.58 for the Ministry of Education, Culture and Science, 0.54
for the Ministry of Health, Welfare and Sport, 0.43 for the Ministry of Justice, and 0.20 for the
Ministry of Transport, Public Works and Water Management.