on participation. Alternatively, our analysis carries through with minor modifications
under the proviso that it pertains to those skill types whose participation is affected by
the policies modeled in this article.
Following Diamond (1980), the government maximizes a utilitarian social welfare
function.3 Given the participation decisions, social welfare can be written
W=
pn
J n
/»m* (c(n),b,r) mm(n)
u(c(n), m)f (m, n)dm +
⅜/ m(n) m m* (c{n),r,r)
u(b, r)f (m, n)dm dn.
(3)
To focus on the issue of optimal redistribution, we assume that the sole motives for
taxation are to finance the public welfare benefit and go cover an exogenous revenue
requirement R. Additional, so as to not conflate other effects of workfare with its role
in influencing labor market participation, we assume that workfare is completely unpro-
ductive. Under these conditions, the public constraint of the government can be written
nn p
nm
`n
■m* (c(n),b,r)
m(n)
rnmn))
[n - c(n)]f (m, n)dm -
m* (c(n),b,r)
bf (m, n)dm dn = R.
(4)
The government’s decision problem is to choose c(n), b and r to maximize the social
welfare criterion (3), subject to the resource constraint (4).
The goal of our analysis is to examine the conditions under which workfare is a
desirable addition to the policy mix. Thus, we proceed in two steps. First, we consider the
optimal tax-transfer mix (the choice of c(n) and b) for an arbitrary intensity of workfare.
Second, we ask if, starting from no workfare — that is, r = 0 — a small increase in the
intensity of workfare enhances social welfare. The first step is formally identical to the
existing literature on optimal taxation with work choice along the extensive margin.
The results from this literature that we need below are summarized in the Proposition
1 below. In order to simplify the statement of this result, we define the participation tax
for individuals of type n, denoted τ (n) to be sum of the tax paid when employed and the
public welfare benefit, which is withdrawn upon taking up employment. Formally,
τ(n) = t(n) + b = n - c(n) + b.
(5)
3Our results extend to more general Bergson-Samuelson social welfare functions.