Optimal Tax 05
Optimal Tax 05
Vincent A. Hildebrand
Glendon College, York University and SEDAP, McMaster University
Lecture 7
Frank Plumpton Ramsey (February 22, 1903 – January 19, 1930) was a British
mathematician, philosopher and economist.
To answer this, consider the following, somewhat arti cial problem : given the
total excess burden, let's try and raise as much revenue as possible.
That is, suppose that we raise the tax on food by one percent, and then lower
the tax on clothing by whatever amount makes the total excess burden
constant.
It's worth doing if the new policy — higher taxes on food, lower taxes on
clothing — leads to more tax revenue (T R). Then we've got more revenue for
the same amount of waste. By construction,
EB F = EB C
That is, dividing both sides by the equivalent changes in excess burden, if at the
margin
T RF T RC
F
<
EB EB C
This inequality suggests that it's a good idea to raise the tax on food and lower
the tax on clothing, since, at the margin, the excess burden per dollar of tax
raised is higher in the clothing market.
If the inequality goes the other way, then food taxation brings greater waste per
dollar raised, so it's better, at the margin, to lower the tax on food and raise the
tax on clothing.
The Ramsey Rule (4)
We should adjust the tax mix until the excess burden per dollar of revenue
raised, at the margin, is equal across commodities.
So that says that an optimal commodity tax system should be one for which
T R= EB is the same for every commodity.
Otherwise we could raise the revenue yield for a given excess burden ( or lower
the excess burden for a given revenue requirement ) by adjusting taxes, namely
by raising them on goods for which T R= EB is low.
Deriving the Ramsey Rule ( EB)
One of my approximate formulae for the area of the excess burden triangle is
@Q
EB (0:5)t( Q) (0:5)t2 (1)
@t
If I took the slope of the compensated demand curve as constant, then the
derivative of this with respect to t is
@Q
EB t Q (2)
@t
Notice I can actually get the rst part of this without approximation by using the
“area equals integral” formula
Z p+t
EB = Q(P )dP tQ(p + t)
p
TR Q EB Q
= = 1 (4)
EB EB EB
So I get T RF = EB F = T RC = EB C if
QF QC
F
=
Q QC
This is equation 17.9 of the text, the rst version of the so–called Ramsey rule
for taxation ( after the English economist who rst derived it ).This rule is true
EVEN IF goods are related (complements or substitutes.)
Interpreting Equation 17.9
If the commodity tax system minimizes the excess burden of raising a given
amount of revenue, then it should reduce demand for all goods by the same
proportion.
Notice that this rule does not say that the proportional tax rate on each good
should be the same. It says that the proportional reduction in demand for each
good, caused by the whole tax system, should be the same for each good.
Quite generally, a commodity tax system will minimize the total excess burden
only if the quantity purchased of each good is reduced by the same proportion
— approximately, and if people were compensated for the damage done by the
tax.
Alternative way of looking at the Ramsey Rule (1).
1 2 1
min L = [ X PX XtX ] +[ Y PY Xt2Y ] + [M PX XtX PY Y tY ]
tX ;tY ; 2 2
c
tX Y
= c
tY X
Summing Up our Results (1)
The “inverse elasticity” formula gives a very precise and speci c prescription for
commodity taxation. It says that the commodity tax system should not be
uniform — unless all goods had the same compensated own–price elasticity of
demand.
It does not tell me immediately how high all these rates should be, just the
relationship among them. But the revenue needs of the government will dictate
how high the absolute level of these taxes should be.
But there is one, potentially very big, catch here. This formula only works if
demand for each of the taxed goods is independent of the prices of the other
taxed goods.
The Corlett-Hague rule
Corlett and Hague looked at a more restricted problem : suppose that there
already is a tax in place, and that we cannot alter that tax.
What does the presence of that rst tax imply for the tax rates which we can
control. Chapter 17 tells me the answer: other things equal, we should raise the
taxes on substitutes to the taxed good, and lower the taxes on complements.
As an important example, consider the fact that we levy a lot of taxes on labour
income in Canada. What does the presence of those taxes imply for what we
should do with commodity taxes?
The Corlett-Hague rule
A tax on labour income is really a subsidy on leisure which also yield an excess
burden. So to reduce that excess burden, complements to leisure should be
taxed, and substitutes taxed less.
Suppose that we need to raise a given amount of money from the income tax.
Suppose further that I propose to make the burden of the income tax on the
poor less, by raising the basic exemption. Suppose that I raise the level of
income which is not subject to tax.
This change will make the income tax burden much less at low income. If I raise
the basic exemption from $10,000 to $12,000, ( and then tax all income above
the exempt amount at 17 percent ), then I'd reduce the tax paid by a family
earning $15,000 from $850 to $510, a pretty large percentage reduction.
Trade-off between Equity and Ef ciency (2)
Raising the exemption will lower the tax yield. To make up the revenue shortfall,
what can I do?
Raise the tax collected from higher income people by raising the marginal rate
they pay. However, my formula for the excess burden said that it increases with
the square of the tax rate.
Raising the marginal rate will de nitely increase the excess burden of the
income tax, and the numbers from the end of chapter 16 suggested that excess
burden was already pretty high.
So there is a con ict here, and a tough choice : higher marginal rates mean a
more progressive system, but also mean a greater excess burden.
The question in optimal income tax theory is how to weigh these costs and
bene ts, and what factors would lead to the costs of more progressivity being
higher or lower.
Optimal Income Taxation & Equity
To account for distributional (equity) issues, the Ramsey rule has been
modi ed.
Degree of departure from original rule depends on:
The British economist Edgeworth recognized this result about a century ago.
he asked : what would be the best income tax system, if income were truly
exogenous?
In particular, he wanted to know how the progressivity of the income tax would
depend on people's ( or the government's ) attitude toward inequality.
In general, we might think that a social welfare function had some aversion to
inequality : transferring a dollar from a rich person to a poor person would raise
social welfare, other things equal.
But there can be a wide range of views on how much inequality matters. Is a
dollar more to a poor person the same as $1.20 to a rich person? $2? $5?
Edgeworth's Contribution (2)
We might expect, the higher is our aversion to inequality, that is the more we
think transfers to the poor are more valuable than transfers to the rich, the more
progressive we would want the income tax to be.
Edgeworth showed that conclusion is wrong. As long as you think income is
worth more, socially, at the margin to the poor than to the rich, then the optimal
income tax should involve a marginal rate of — wait for it — 100 percent.
It doesn't matter whether you weight transfers to the poor as a little more
valuable than transfers to the rich, or a lot. Raising the marginal rate ( and
raising the exemption so as to keep the revenue yield constant ) will transfer
income from the rich to the poor.
Edgeworth's Contribution (3)
So if you do not think the ideal income tax system is one which con scates
everything and redistributes it equally, then probably one of two things are true :
2. Or you think that income taxes affect economic incentives, and that these
incentives are important.
The analysis becomes more dif cult if we drop the ( not very realistic )
assumption that income is exogenous.
Modern Studies: Endogenizing Labour Income
Now the average income has to be treated as a function of the tax system.
revenue = +t Income
If the government needs to raise $5000, then it must pick t and so that
t[a(t; ) ] = 5000
where a(t; ) is the average income in the country, which depends on the tax
rate and on the level of the exemption ( ).
Higher values of t mean more progressive tax but larger excess burdens.
Optimal income taxation comes down to nd the right combination of and t.
Stern Contribution (1974)
Solving this optimal at tax problem, Stern showed that how progressive the
income tax should be does depend on 3 factors:
Stern's nds that a value of t=20% would maximize social welfare....far from
Edgeworth 100%.
General Conclusion from Stern's model
Earlier, we mentioned that the excess burden of the income tax was
proportional to the compensated elasticity of labour supply.
The bigger is that elasticity, the higher the excess burden of the income tax, and
the less progressive is the optimal income tax.
More substitutability between leisure and goods raises the marginal cost ( in
terms of higher excess burden ) of a more progressive income tax.
The bigger is our aversion to inequality, the higher is the marginal bene t of
more progressive income tax, and the higher is the optimal income tax rate.
Very Recent Contributions
There is no reason to restrict attention to income tax schedules with constant
marginal tax rates. Economists have looked at the optimal overall shape of the
income tax schedule, without restricting that shape to be linear.
That is, if we can vary the marginal rate by income class, what should be the
optimal schedule?
It turns out that often it appears that it is not optimal to deviate much from a at
tax — although that conclusion is rather sensitive to the assumptions made.
Recent Contribution: Gruber and Saez (2000).
In a recent paper, Gruber and Saez (2000) showed that a tax system with
different marginal rate by income class is optimal (and yet progressive), if
people in higher-income tax brackets face lower marginal tax rates than people
in lower brackets.
(Intuitive) Explanation of this Paradox
Lowering the marginal tax rate on high income people induce them to supply
more Labour (L).
Increasing L increase tax revenue which can be used to lower the tax burden of
low-income earners.
Although marginal tax rates fall with income, the average tax rates rise with
income, so this optimal tax system is still progressive.
It is important to remember, that a lot of assumptions and value judgements lie
behind all these results. These considerations are really at the core of the
debate on optimal tax systems.