Policy networks is a powerful analytical concept. However, it must be
combined with a model of the actor if the analyst is to move beyond de
scription and into the more interesting field of policy explanation. T
his article argues that this may be done by the use of 'new institutio
nalism'. It first demonstrates that policy networks may be understood
as institutions, i.e. as rules constraining the actions of the partici
pating actors. It then outlines the contribution which institutional t
heory may make to policy network analysis. This is done by giving an i
nstitutional answer to three questions which the policy network litera
ture has not successfully addressed. Why do networks come into existen
ce? Why do networks change? Why are networks so persistent? The instit
utional logic is illustrated in a case study of a Danish intergovernme
ntal policy network: the annual budget negotiations between the state
and local authorities. The main lesson to draw from this exercise is t
hat when the insights from policy network analysis are combined with t
he theoretical insights from 'new institutionalism', it is indeed poss
ible to start providing explanations of policy networks and their impa
ct on policy outcomes.