From micro data on actual social assistance receipts, I estimate expected payments for a standardized user in each municipality
to represent the actual levels of municipal social assistance generosity. I find in Section 4 that the coefficient of variation for benefit
norms is significantly reduced after the introduction of the norm, while variation is unchanged for expected payments. Like
Fiva and Rattsø (2006), I find hardly any correlation between benefit norms and expected payments. In Section 5, I study to
what degree benefit norms and expected payments are decided by local determinants, combining the empirical strategies of
Fiva and Rattsø (2006) and Borge et al. (2014). I find that local demand determinants are more important for actual payments
than for benefit norms, both before and after the introduction of national guidelines in 2001. These results support the hypothesis
that caseworkers are governed, at least partly, by local political preferences.