valid asymptotically. If the instantaneous utility function of consumers is
strongly separable in consumption and leisure, and in addition is isoelastic
in consumption, welfare maximization implies that capital should be taxed
at 100 percent at the initial period and zero afterwards; this is because the
capital stock is inelastic in the short-run and therefore it must be efficiently
taxed at a confiscatory rate.
Lucas (1990), Jones, Manuelli and Rossi (1993 and 1997), Correia (1996b),
Atkeson, Chari and Kehoe (1999), and Judd (1999), among others, find that
the optimality of the zero capital income tax carries over a wide variety of se-
tups that incorporate human capital accumulation, perpetual growth, perfect
capital mobility and overlapping-generations.2
The second-best principle of capital taxation established by Judd (1985)
and Chamley (1986), however, is not an ineluctable law of dynamic pub-
lic finance as shown, for example, by Correia (1996a), Jones, Manuelli and
Rossi (1997), Chamley (2001), Erosa and Gervais (2002), and Abel (2006).3
Correia (1996a) discovers that the introduction of an additional factor of
production, which cannot be optimally taxed or subsidized, in a Ramsey-
Ricardo exogenous growth model leads to a violation of the zero capital tax
prescription. From a methodological perspective, Jones, Manuelli and Rossi
(1997) identify two types of changes in the neoclassical intertemporal frame-
work that lead to a capital income tax different from zero. The first change
is obtained if the capital stock appears in the pseudo-welfare function of the
social planner; this case is satisfied, for example, when the labor supply is
inelastic (i.e. pure rents enter the consumer’s budget constraint) and there
be untaxed, while labor (a final good as it appears in the utility function of consumers
through leisure) should be taxed.
2In particular, Atkeson, Chari and Kehoe (1999) provide a generality test of the
Chamley-Judd tax result by systematically relaxing one by one the hypotheses that sup-
port it and discovering that, in so doing, its validity remains unaffected.
3 Other contributions in which optimal capital income taxation may differ from zero are
by Pestieau (1974), Atkinson and Sandmo (1980), Lansing (1999), Boadway, Marchand
and Pestieau (2000), and Cremer, Pestieau, and Rochet (2003).