This paper presents a theory of optimal public finance in which a succ
ession of governments sets welfare-maximizing fiscal policy through ti
me. Such a setting allows, for the first time, estimation of stochasti
c Euler equations associated with the government's fiscal choice probl
em. While there has been theoretical work on the dynamic application o
f the theory of optimal taxation, no serious effort has been made to t
est this theory with data. This paper deals empirically with the dynam
ic application of the theory of optimal taxation in general, and the i
ssue of time-consistency in particular. We show that U.S. data do not
reject the implications of optimal public finance given by the Euler e
quations.