This paper asks whether tax cycles can represent the optimal policy in a modelnwithout any extrinsic uncertainty. I show, in an economy without capital and where labor is the only choicenvariable (a Lucas-Stokey economy), that a large class of preferences exists, where cycles are optimal, as well as a large class where they are not. The larger government expenditures are, the larger the class of preferences fornwhich cycles are optimal becomes. nTax cycles are also more likely to be optimal if frictions (deviations of the model from Walrasian markets) are added. While this cannot be shown inngeneral and will not be true for arbitrary frictions, I demonstrate this in two specific worlds. I consider an economy with search frictions in the labor market, and one with frictions in the goods and credit market. A reasonablenparametrization of both economies shows that results change considerably. Even with constant relative risk aversion, cycles can be optimal, whereas this class of preferences rules out cycles in the Lucas-Stokey economy.nFinally, I characterize the optimal policy. No more than two tax rates are needed to implement the Ramsey policy both in the Lucas-Stokey economynand in the model with frictions.