OPTIMAL LINEAR INCOME TAXATION WITH GENERAL EQUILIBRIUM

EFFECTS ON WAGES AUTHOR: FRANKLIN

ALLEN

A Review by KOLAWOLE BASHIR OLAYINKA at the AERC, CPP, JFE-

2010

Following the work of Mirlees (1971) where it formulated a model for the investigation of
the distortion on the choice between labour and leisure, and the subsequent contributions by
Sheshinski (1972), Atkinson (1973), Stern (1976), and Helpman and Sadka (1978), this study
also joined the large and growing discussions on income taxation.

As such, this paper was motivated by the need to investigate the distortion of income taxation
on the choice between labour and leisure by considering the effect of local changes in the
linear tax at the market solution; and examining the sign of the tax required to maximise
Rawlsian social welfare.

In its analysis, the paper assumed two types of labour: skilled and unskilled (entrepreneur and
raw labour), with ability groups, A1 and A2 whose relative size is one to n. Also, earnings of
the A1’s group are taken to be higher than the A2’s group’s. Such that X = w1L1/w2L2 > 1,
where wi and Li are the wage and labour supply of the members of group A i. The author
further assumed constant returns to scale, a linear tax function and a single homogeneous
output. Individuals choose their consumption and labour supply to maximise their utility
subject to their budget constraint. Production cost function is represented by g(w1, nw2) and
equilibrium in the output market holds only when g(w1, nw2) = 1.

Denoting the proportionate component of the linear tax by (1-β) and lump-sum grant by α,
the study analytically presented that in each labour market, equilibrium, with respect to the
theory of cost functions, requires that, Li(βwi, α) = Ygi(w1, nw2), i = 1,2, where Y is total
output and g1= δg/δw1, g2 = δg/δ(nw2). Finally, the author showed that the overall equilibrium
requires that the government’s budget constraint is satisfied when α = (1-β)k, where k is per
capita income and k = w1L1 + nw2L2/n+1.

As such, using the illustration of local changes in the tax at the market solution, the paper
stressed the fact that the three equilibrium conditions stated above could be used to
investigate the effect of a change in the proportionate tax rate (1-β) while altering the lump-

1
sum grant α to keep the government’s budget constraint satisfied. The consequential change
in utilities from a reduction in the marginal tax rate will result in wiLi + dα/dβ, which is the
direct distribution via the fiscal system; and change in net wages, βwiLi d log wi/dβ, which is
the general equilibrium effect.

If one considers possible cases where a large number of assumptions concerning the labour
supply of the two groups are conceived, then it follows that the enjoyment of the two types of
work may be different, and leisure opportunities may not be the same because of different
abilities, and so on. Given a conventional type of labour supply curve a higher wage
corresponds to a lower elasticity so that (ε2-ε1) will be positive. Where εi=βwiLiw/Li. Provided
leisure is a normal good the ratio Lit/Li will be negative. Since a uniform lump-sum grant is a
small component of income the higher is a person’s earnings, it seems likely that the ratio
will increase with income so that (L2t/L2 - L1t/L1) will be negative.

In the critic of the findings by Feldstein (1973), the study analysed that the results by
Feldstein were due to the use of a Cobb-Douglas production function since it ensures that the
contribution via the fiscal system and due to the change in wages were reinforcing. The
author, however, noted that this was not, generally, the case. Such that, in plausible situations
they can be offsetting and it is possible for the general equilibrium effect to dominate so that
the people that are worst-off are helped by a linear tax consisting of a wage subsidy and a
uniform lump-sum tax, which redistributes income from poor to rich.

The author demonstrated that not only that the general effect at the market solution may have
both signs, but that it could outweigh the direct effect. The study was able to show that it was
possible for the optimal tax to have either sign. In the analysis the general equilibrium effect
was predominant, as it was characterised by a low elasticity of substitution between the types
of labour with negative labour supply elasticity for the highly paid. This, however, seems to
be unreasonable. But what is required for the sign to change is for the high earners’ labour
supply elasticity to switch from a positive to a negative value.

This result was corroborated with a similar result by Stern (1982) for the non-linear income
tax. It was shown analytically that, with a production function in two types of labour, and a
positive marginal tax rate for the lowest paid workers if leisure is a normal good, and a
negative marginal rate for the highest paid if the marginal social utility of their consumption
is lower than that of the other group.

2
This result may, however, based on the Stern’s result, have difficulties in implementing a
wage subsidy if the optimal linear tax is negative. This is due to the fact that there’s problem
in enforcing a positive tax as people tend to conceal their true incomes and declare their
earnings to be lower than they actually are. The problems of doing this on a large scale,
however, are usually such that it is worthwhile only for those with high incomes to do it, and
then, there is an upper bound to the amount of tax that can be avoided. But with a negative
tax, where the proportionate component takes the form of a subsidy, everybody has an
unlimited incentive to conceal their true incomes and inflate their declared earnings. It might
be very difficult to enforce such a system since artificial jobs paying large amounts can
usually be created without problems and there is no upper bound to the amount that can be
claimed, because the purpose of the system is to subsidise higher incomes the most. A
negative linear tax is equivalent to a proportionate subsidy on consumption and it might be
more feasible to implement the tax in this way though it may still be fairly easy to create false
transactions and exploit the system.
In conclusion, the paper observed that because of changes in wages, attempts to alleviate
poverty by progressive income taxation may not be very successful and may even exacerbate
the problem, especially if the elasticities of labour supply are negative and different types of
labour are complements rather than substitutes. However, the study omitted the utilitarian
case, in its analysis, as a corollary to the findings on the sign of the Rawlsian optimal tax.