The institutional choice of refuse collection: Determining variables in the Netherlands
General empirical evidence suggests that contracting out refuse collection results in a cost decrease in the order of 20%. However, although the method of contracting out refuse collection has become more popular, it is still less common than in-house provision. This paper investigates the reasons behind this phenomenon. Recently, L?pez-de-Silanes et al. (1997) tried to explain the reservedness of local authorities towards contracting out with US-data and show that political patronage is an important explanation. In this article we give such an empirical assessment using Dutch data. We base our empirical research on the combination of the theories around efficiency, interest group influence and ideology. To test these theories we model the choice between private and public provision of refuse collection on the one hand side and the choice between in-house and out-house provision on the other side. Data are available for nearly all Dutch municipalities. A most striking conclusion is that nearly all political parties in the Netherlands do have a preference for public and in-house provision of refuse collection. We find only modest evidence for the hypothesis that a high level of real estate tax (proxy for the soundness of municipal financial affairs, the efficiency argument) or a low level of unemployment (the interest group argument) raises the probability of private and out-house provision. We find more evidence for the assumed relation between the size of municipalities and private collection. In all cases a smaller municipality has a higher chance of private collection. Therefore, scale effects are important for the choice between public and private provision. For the choice between out-house and in-house collection in relation to scale less evidence exists. Compared with other studies we estimated more general models, allowing for higher order effects and heteroskedasticity. We show that the standard Logit model is too restrictive, both for the choice between private and public provision and the choice between in-house and out-house provision. Signs and significance do differ between the different models. Especially, for the choice between in-house and out-house provision the applied model matters. In this case, only unemployment and municipality size has influence. Thus, more attention is needed for the implications of model choice for the explanation of the raised questions.