When the skills are perfectly substitutable the optimal steady state policy subsidizes education to the tune of 170% and increases the tax rate to 37% from a baseline of 27.5%. The deductible increases just a little bit. The resulting welfare, measured in consumption, increases by 2.6%. In other words optimality implies a huge average tax rate – much larger than what is in place now, but preserves the incentives to obtain education by paying people to attend college. However, along the transition path towards the steady state large welfare losses are incurred. This is because it takes decades for the stock of college graduates to reach the level implied by the optimal tax parameters: while attendance reacts immediately, it takes a number of generations to achieve the new steady state. Effectively the large increase in taxes induces a recession early on, which imposes important welfare costs. When the authors factor in these welfare costs in the optimization problem, the resulting optimal tax rate declines to 23%, which is lower than the one in the baseline and the deductible declines a lot: we basically move almost to a flat tax system with essentially no progressivity. At the same time the subsidy rate for education is still high but declines to 120%. Interestingly, the authors use debt to push the costs of subsidizing education to later generations. Thus insurance is achieved by pushing more people into college. However, they avoid increasing taxes, which would provide more insurance because the upfront costs on the early generations of doing this are far too high. The welfare gains are, however, still substantial (1.66%).