Previous literature has shown that public provision of private goods can be a welfare-enhancing device in second-best settings where governments pursue redistributive goals. However, three issues have so far been neglected. First, the case for supplementing an optimal nonlinear income tax with public provision of private goods has been made in models where agents differ only in terms of market ability. Second, the magnitude of the welfare gains achievable through public provision schemes has not been assessed. Third, the similarities/differences between public provision schemes and tagging schemes have not been thoroughly analyzed. Our purpose in this paper is therefore threefold: first, to extend previous contributions by incorporating in the theoretical analysis both heterogeneity in market ability and in the need for the publicly provided good; second, to perform numerical simulations to quantify the size of the potential welfare gains achievable by introducing a public provision scheme, and to characterize the conditions under which these welfare gains are sizeable; finally, to compare the welfare gains from public provision with the welfare gains from tagging.