ECON 7035A: Public Economics and Fiscal Policy (July-October 2019)

TOPIC 7
TAX INCIDENCE
& OPTIMALITY

MICHAEL SACHS
Adjunct Professor | Southern Centre for Inequality Studies
University of the Witwatersrand | Johannesburg
Outline
2

▪ This week
– Partial equilibrium
– Incidence
– Efficiency and deadweight loss
– Commodity taxes
▪ Next week
– General equilibrium
– Income taxes (on labour and capital)
– Equity and progressivity
Readings
3

▪ Institute for Fiscal Studies (2011) Tax By Design: The Mirrlees Review
– Chapter 2 The Economic Approach to Tax Design
▪ Stiglitz, J. E. and Rosengard, J. K. (2015) Economics of the Public Sector
– Chapter 18 Tax incidence
– Chapter 19 Taxation and economic efficiency
– Chapter 20 Optimal taxation
But also …
– Chapter 21 Taxation of capital
– Chapter 22: The personal income tax
▪ Hindriks, J. and Myles, G. D. (2013) Intermediate Public Economics
– Chapter 15 Commodity taxation
– Chapter 16 Income Taxation

▪ Also: Black, P. A., Calitz, E., Steenekamp, T. J. and Black, P. A. (2015) Public Economics
– Chapters 11, 12, 13 and 14
– Good coverage of incidence and efficiency, less on optimal taxation
 4

Preliminary issues in
taxation
Tax systems Mirlees Review
5

▪ The tax system is “enormous and fearsomely complex”

▪ For a given distributional outcome the tax system should aim for:
– Minimize negative effects on welfare and economic efficiency
– Minimize compliance costs
– Be procedurally fair and non-discriminatory
– Transparency
▪ This is achieved by a simple, neutral and stable tax system
– Neutrality: The same activities attract the same rate of taxation
– Simplicity: complexity and avoidance dynamic
– Stability: compliance costs and long term planning
▪ The trade-off between redistribution and efficiency is at the centre of debates about the tax
system.
▪ The theory of optimal taxation tries to provide consistent and rigorous tool of evaluation
 6

Introduction to tax
incidence
Atkinson and Stiglitz

“In principle, the analysis of the incidence of a tax is a straightforward matter.

We calculate the general equilibrium of the economy before the change in
taxation or expenditure, and we recalculate the equilibrium afterwards; the
changes provide a description of the incidence of the tax.
We can say whose income has gone up or down and by how much, and what
prices have changed.
In almost all situations, the real incomes of individuals other than those upon
whom the tax is levied will change.
Quite often the change in real income of those upon whom the tax is levied is
smaller than the magnitude of the tax. We say then that the tax has been
shifted to others in the economy.”
Who bears the burden of a tax?
8
Economic impacts Concepts Analytical tools
▪ Consumption patterns ▪ Statutory and economic incidence ▪ Partial equilibrium
▪ Distribution ▪ Shifting forwards and backwards ▪ General equilibrium
▪ Supply of factors ▪ Equivalent taxes
> Growth, employment, industrial ▪ Equivalence of taxes and subsidies Economic factors that influence
structure incidence
Allocative efficiency ▪ Elasticities
Distributive equity ▪ Degree of competition

Owner of land Factors of production

or resources Shareholders (equity)
Workers

Producer Other stakeholders

Producer
Creditors (debt)
Suppliers
Producer Communities

Retailer

Consumer
Tax incidence
9
▪ Tax incidence is the study of the effects of tax policies on prices and the distribution of
utilities

▪ Ideally, we would characterize the effect of a tax change on utility levels of all agents in the
economy

▪ In practice: aggregate economic agents into a few groups

– Producer vs. consumer (tax on cigarettes)

– Source of income (labour vs. capital)

– Income level (rich vs. poor)

– Region or country (local property taxes)

– Across generations (social security reform)

▪ In general tax incidence analysis is informative about signs and comparative statics but is
inconclusive about magnitudes

Source: Chetty (2012)

Understanding tax incidence is very important
10

▪ Consider the following argument: Government should tax capital income because it is
concentrated at the high end of the income distribution

▪ Neglects general equilibrium price effects

– Tax might be shifted onto workers

– If capital taxes result in less savings and/or capital flighyt then over time the capital
stock may decline, driving return to capital up and wages down

– Some argue that capital taxes are paid by workers and therefore increase income
inequality
Tax incidence in partial equilibrium
11
▪ Assumes
– Two good economy
• Only one relative price so partial and general equilibrium are same
• Might be realistic if:
the market being taxed is small
there are no close substitutes/complements (demands is independent)
• We can think of good one being the taxed good, and good two being all other
commodities.
– Tax revenue is “thrown into the ocean”
– Perfect competition among producers
▪ Commodity taxation - Government levies an excise tax on good x
– Excise or specific tax
– Ad-valorem tax: fraction of prices (e.g. VAT)

Source: Chetty (2012)

 Specific excise duties in South Africa
12

Table 4.6 Changes in specific excise duties, 2019/20

Current excise Proposed excise Percentage change

Product duty rate duty rate Nominal Real

Malt beer R95.03 / litre of absolute alcohol R102.07/ litre of absolute alcohol 7.4 2.2
(161,56c / average 340ml can) (173,51c / average 340ml can)

Traditional African beer 7,82c / litre 7,82c / litre – -5.2

Traditional African beer 34,70c / kg 34,70c / kg – -5.2
powder
Unfortified wine R3.91 / litre R4.20 / litre 7.4 2.2
Fortified wine R6.54 / litre R7.03 / litre 7.4 2.2
Sparkling wine R12.43 / litre R13.55 / litre 9.0 3.8
Ciders and alcoholic fruit R95.03 / litre of absolute alcohol R102.07/ litre of absolute alcohol 7.4 2.2
beverages (161,56c / average 340ml can) (173,51c / average 340ml can)

Spirits R190.08 / litre of absolute R204.15 / litre of absolute 7.4 2.2

alcohol (R61.30 / 750ml bottle) alcohol (R65.84 / 750ml bottle)
Cigarettes R15.52 / 20 cigarettes R16.66 / 20 cigarettes 7.4 2.2
Cigarette tobacco R17.44 / 50g R18.73 / 50g 7.4 2.2
Pipe tobacco R4.94 / 25g R5.39 / 25g 9.0 3.8
Cigars R82.31 / 23g R89.72 / 23g 9.0 3.8
Source: National Treasury
Specific tax in a competitive market
13
▪ A specific tax on a firms output
▪ Effective cost of production is increased by the amount of the tax
▪ In aggregate, the supply curves shifts up by the amount of the tax

Price S1

S0
p0 + t

Tax

p0

Q0 Quantity
Specific tax in a competitive market
14
▪ The burden of the tax is shared between consumers and firms
▪ What happens if the tax was paid by the consumer (or the
retailer at the point of sale)?

Price S1

S0

E1
Price paid by consumer pc
E0
Pre tax price p0

Price received by firms pf

Q1 Q0 Quantity
Specific tax in a competitive market
15
▪ What happens if the tax was paid by the consumer (or the retailer at the point of sale)
▪ The statutory burden is different, but the economic burden is exactly the same.

Price
(received by producer)

S0

Price paid by consumer pc

E0
Pre tax price p0

Price received by firms pf

E1 Tax
p0 - t
D0
D1
Q1 Q0 Quantity
Elasticity: limiting cases
16
D Inelastic
Elastic demand
supply S1
p1
S1 p1
S0
p0 S0 p0

Q1 Q0 Q0=Q1

Elastic S1
Inelastic
demand
supply
S0 p0=p1
p0=p1 D
D0

Q1 Q0 Q0=Q1
Tax incidence summary
17
▪ Economic incidence describes who actually bears the burden – whose real income or welfare
changes as a result of the tax
▪ It does not depend on who pays over the money (statutory incidence)
▪ In competitive markets, incidence depends on the elasticity of demand and supply
– If demand is perfectly elastic, consumers bear no burden
– If supply is perfectly elastic, producers bear no burden.
– If demand is perfectly inelastic, consumers bear the full burden
– If supply is perfectly inelastic, producers bear the full burden
Factor markets
18

▪ The same principles apply to factor markets

– The incidence of a tax on a factor in a competitive market depends on the elasticity of
supply and demand for the factor.
– Factors with inelastic supply bear the burden of the tax
– Elastic factors can shift the burden.
– Mobility of capital and labour?
– More on this next week.
▪ Land tax
– The supply of (unimproved) land is fixed
– Therefore the total burden of the tax will fall on landowners
– Distinguishing land value from improvements
– Consider a tax on oil or platinum
 19

Introduction to
taxation and efficiency
Welfare analysis of distortionary taxation
20
▪ For the same amount of revenue what is the
difference in welfare (utility) between a
Good 2
distortionary tax and a lump sum tax?
▪ Equivalently: for the same effect on individual
welfare, how much extra revenue (money)
would a lump sum tax have raised?
▪ The difference is the deadweight loss.

E*
E

U’

Good 1
Effects of taxation on behaviour
21
▪ Income effect
– All taxes take income away from individuals, it makes them worse off.
– Individuals make different decisions when their income changes.
– Because they are poorer they postpone their retirement, they cannot enjoy much leisure,
spouses take on work instead of looking after children…

▪ Substitution effect Efforts to avoid taxes by substituting non-taxed for taxed activities
– Not all activities are taxed, or taxed at the same rate.
– Taxation changes relative prices from their economic equilibrium.
– This diverts economic activity from taxed to untaxed areas, or from areas of higher taxes
to areas of lower tax.
– Leisure, household production, subsistence agriculture do not attract taxation. People
seek out untaxed areas – for instance more leisure may mean putting in less effort on the
job, retiring earlier, working shorter hours, or beginning work later in life.
– (Note the contrary impacts on work effort and retirement choice)

▪ Financial effects
– Real activity can be mediated through different forms of payments
– People may prefer payment in cash or in kind to avoid taxation.
– Remuneration may shift from income to capital
– Finance may be provided in the form of debt rate then equity.
Atkinson and Stiglitz p27
Income and substitution effects
22

Good 2
▪ A lump sum tax reduce the consumer’s income
▪ Equivalent to a proportional tax on all consumption goods
▪ Utility falls, but with no substitution effects

Government’s
revenue

U0
b
U1

Good 1
Income and substitution effects
23
▪ A lump sum tax reduce the consumer’s income
Good 2
▪ Utility falls, but with no substitution effects

▪ A commodity tax on good 1 shifts relative prices

▪ Government revenue is the same at points b and c
rc

rb

c
a

U0
b
U1

Good 1
Income and substitution effects
24

Good 2 ▪ A commodity tax on good 1 shifts relative prices

▪ Government revenue is the same at points b and c

▪ But at point c, utility is lowered even further.

▪ U1 – U2 is DWL measured in utility terms: i.e. the
fall in welfare that exceeds the gain in revenue

c
a

U0
b
U1

U2

Good 1
Income and substitution effects
25
▪ But at point c, utility is lowered even further.
Good 2
▪ U1 – U2 is DWL measured in utility terms: i.e. the fall in welfare that
exceeds the gain in revenue

▪ A lump sum tax that generates U2 is at d

▪ The difference between the two budget lines (at
b and d) is a monetary measures of deadweight
loss
– The tax results in:
• Income effect (a-d)

c • Substitution effect (d-c)

a

U0
b
d
U1

U2

Good 1
Magnitudes of the income and substitution effect
26
▪ For the same amount of revenue what is the
Good 2 difference in welfare (utility) between a
distortionary tax and a lump sum tax?
U1 – U2
▪ Equivalently: for the same effect on individual
rc welfare, how much extra revenue (money)
would a lump sum tax have raised?
rd rd - rc

c
a

U0
b
d
U1

U2

Good 1
Magnitudes of the income and substitution effect
27

Good 2

The size of the substitution

effect depends on the
elasticity of substitution

c
a

d U0

U1
U2

Good 1
Magnitudes of the income and substitution effect
28

Good 2

c
a

U0
b
d
U1

U2

Good 1
Magnitudes of the income and substitution effect
29

Good 2 ▪ If the two commodities are

perfect complements, there is no
substitution effect in demand
▪ The non-neutral tax is more like
a lump sum tax
▪ There is no dead-weight loss: dwl
is cause by substitution (not
income effects)

a
U0

b
U1

Good 1
Deadweight loss (excess burden)
Price ▪ Constant cost supply
a – Infinitely elastic supply means that the
consumer bears the full burden of the tax
increase
– No producer surplus

Consumer
surplus Marginal
b social cost
p

Marginal social
benefit
(Compensated
demand curve)

X0 Quantity
Deadweight loss (excess burden)
Price
a The deadweight loss is the extent to which the
welfare loss exceeds the revenue raised

Consumer DWL = ½ t dX
surplus c
q

t = dP Tax revenue
(tX1) Marginal
DWL b social cost
p

Marginal
social benefit

dX

X1 X0 Quantity
Deadweight loss (excess burden)
Price 1 %∆𝑄 𝑑𝑄/𝑄
𝐷𝑊𝐿 = 𝑑𝑃 𝑑𝑋 𝜀 𝑑 = %∆𝑃 =
2 𝑑𝑃/𝑃
a
(NB: assume constant elasticity in log space)
𝑃 𝑑𝑋
𝜀𝑑 ≡
Consumer 𝑋 𝑑𝑃
surplus 𝑑𝑃 𝑑
q c => 𝑑𝑋 = 𝜀 𝑋
𝑃
𝑡 𝑑
t = dP Tax revenue or 𝑑𝑋 = 𝜀 𝑋
𝑃
(tX1)
DWL b
p

dX

1 𝑡 2 𝑑
=> 𝐷𝑊𝐿 = 𝜀 𝑋𝑃
2 𝑃

▪ DWL is proportional to:

– The square of the tax rate
– The elasticity of demand
Proportional to tax rate
Price
a

q c

t
b
p

X1 X0 Quantity
Proportional to tax rate
Price
d

t
q c

t
b
p

▪ DWL is proportional to the square of the tax rate

▪ As the tax rate increases, the excess burden increases by a multiple of the tax rate
▪ Indicates efficiency properties of low rate, broad base
▪ i.e. low rates on all commodities, rather than high rates on a few commodities is likely to be
more efficient

X1 X0 Quantity
Proportional to elasticity
Price
a

q c

t
b
p

X1 X0 Quantity
Proportional to elasticity
Price
a

c
q

t
b
p

▪ DWL is proportional to the price elasticity of demand

▪ The efficiency loss (or welfare loss) is less for inelastic commodities
▪ Concomitant to this: price inelastic commodities generate higher revenues per unit of
welfare loss.

X1 X0 Quantity
Question?
37

▪ Leaving aside distributional concerns (which can be

addressed on the expenditure side)
▪ … focussing on how to finance a given quantum of transfers
to the poor …
▪ Should we introduce two VAT rates, one for basic
commodities, one for luxury goods?
Principles of taxation again
38

▪ Taxes burdens can be shifted.

▪ The ability to shift a tax burden depends on the elasticities of supply and
demand.

▪ The size of the excess burden (or efficiency cost of taxation) is determined by
price elasticities, the size of the tax base and the tax rate.

▪ The more inelastic demand the greater the revenue collection and the smaller
the excess burden.

▪ As the tax rate increases, the excess burden (distortion and welfare impact)
increases by a multiple of the tax rate

▪ Low tax rates on a wide base will lead to lower excess burdens and more tax
revenue than high tax rates on a narrow base.

▪ Excess burden implies efficiency losses compared to the market benchmark. At

the same time there are distributional consequences for where the tax is shifted.
 39

Introduction to optimal
taxation
What is optimal taxation?

▪ Pareto efficient tax structures minimize distortions

▪ How do we compare or order two pareto-efficiency tax systems?

▪ Combine (positive) analysis about efficiency and incidence to reach (normative) conclusions
about the optimal design of the whole tax system

▪ The set of taxes that results in the highest level of welfare while raising the revenue
required by government

– an objective function that takes account of efficiency and equity concerns

– Consumers left free to choose their most preferred consumption plans at the resulting
prices

– Firms must continue to maximize profits

– Government sets taxes on uses of income in order to accomplish two objectives: Raise a
total amount of revenue (R), Minimize the utility loss for agents in the economy

– Market clearing prices

Ramsey approach to optimal taxation
41

▪ Mainly deals with commodity taxes (which are linear)

▪ Rule out the possibility of lump sum taxes

▪ At least one good is not taxed (e.g. leisure) (so relative prices must change)

▪ Standard assumption : Production prices are fixed (infinitely elastic supply curve)

▪ For now, we assume that income tax is not an option (which is the case in many
developing countries)

▪ The questions is: What is the least distortionary pattern of commodity

taxation?
 One consumer, two goods
42

▪ Two goods: Labour (input) and consumption (output)

▪ One consumer: supplying labour, consuming output (no equity x
issues!) consumption,
output
▪ Labour is the numeraire
and therefore wages are 1
▪ The price vector is U(x,ℓ)
orthogonal to the budget U’(x)>0

constraint
𝑝𝑥 = ℓ U’(ℓ)<0

l
1 labour use,
input
Production
43
▪ One firm (demanding
labour supplying output)
x
▪ The price of output (p) is consumption,
set in competitive output
equilibrium (!) at the profit
maximizing level – i.e
parametric
▪ Production function:
constant returns to scale
▪ So the firm is indifferent to
points on the production Production
frontier frontier
▪ Inside the frontier there is
production inefficiency Feasible
production p
set

l
1 labour use,
input
No taxes
44

▪ Utility is maximized subject to budget

constraint x
consumption,
▪ Production is at the limit of possibility output

▪ Everybody happy

ent

l
1 labour use,
input
A tax on output of the firm
45
▪ Government needs to
raise revenue R
x
▪ The first R units of labour consumption,
input are commandeered output

by government

▪ We don’t want to alter

the relative prices faced
by the producer (to
maintain production
efficiency)

l
1 labour use,
input
R
Lump sum tax
46
A lump sum tax
would change the
consumers income, x
without affecting the consumption,
output
relative price of
labour and 𝑝𝑥 = ℓ − 𝑅
consumption
But this is ruled out
by assumption

elst

l
1 labour use,
input
R
A tax on consumption
47
▪ This is achieved by driving a wedge between the cost of production
(p) and the purchase price of output (q)
▪ Labour is not taxed, but this is a normalisation not a real restriction x
▪ A second best world: different agents face different levels of consumption,
taxation output

▪ Now, what is the optimal rate of taxation on consumption?

(𝑝 + 𝑡)𝑥 = ℓ

q
t
p

l
1 labour use,
input
R
Two good economy - production
48

▪ What is the optimal rate of taxation on consumption? (q-p)

x
consumption,
output

l
1 labour use,
input
R
Two good economy - production
49

▪ What is the optimal rate of taxation on consumption? (q-p)

x
consumption,
output

q
p

l
1 labour use,
input
R
Two good economy - production
50

▪ What is the optimal rate of

taxation on consumption? x
(q-p) consumption,
output
▪ The offer curve shows points
consistent with utility
maximization as commodity
taxes vary q

▪ Each point on the offer

curve is associated with a
budget constraint that runs Offer curve
through the origin and an
indifference curve tangent
to that budget constraint.

▪ Utility increases along the q

offer curve
p

l
1 labour use,
input
R
Optimal commodity taxation
51
▪ What is the optimal rate of
taxation on consumption? (q-
p)
x
▪ The optimal commodity tax is consumption,
determined at the highest output
point on the offer curve in the
production set.
▪ The optimal tax rate is t*, which
ensures that:
– The consumer chooses
point e
U0
– By construction this point Offer curve
raises R = t* x*, where x* is
consumption at point e
elst
▪ The optimal rate e, is second best
because elst raises the same e
revenue with a higher level of
utility q
t*
p

l
1 labour use,
input
R
Diamond-Mirlees production efficiency lemma
52

▪ The producer is allowed to maximize leaving relative prices in tact.

▪ The economy must maximize output from the resource endowment

▪ Optimal commodity taxation should not disrupt production efficiency

(MRTSkl=r/w)

▪ The optimum with commodity taxation must be on the boundary of the

production set

▪ Intermediate goods should not be taxed

▪ All distortions are focused on consumer choice.

Take away so far
53

▪ Optimal taxation: The set of taxes that results in the highest level of welfare while raising
the revenue required by government
▪ Modelling approach
– Firms maximize output on the production efficiency frontier
– Constant returns to scale fixes producer prices, and ensures that any demand is met
with supply
– The problem is then divided into two problems
• The consumers problem: Consumer left free to choose most preferred consumption
plans, maximizes utility to derive the offer curve in the context of inevitable price
distortions
• The government's problem: achieve the revenue constraint while maximizing the
consumers welfare by choosing the maximum point on the offer curve.
– The solution is the determined at the highest point of the offer curve in the production
set
– An optimal allocation is chosen (e), and the tax rate is inferred from this point
▪ Commodity taxes are second best, relative to lump sum taxes (the level of welfare is always
lower
Questions unasked
54

▪ In this example we had only one consumer, so equity was not at issue

▪ We also had only one commodity: What is the solution if we have more than one
commodity?

▪ Should all goods have the same rate or should taxes be related to the
characteristics of the good
Inverse elasticity rule Model set up
55

▪ Using only commodity taxes, maximize social welfare subject to raising required
revenue
▪ Labour (input) produces commodities (output)
▪ Labour (x0) is the numeraire , it is the only good not taxed (t0 = 0)
▪ Two commodities (xi where i = 1, 2)
▪ Independent demands (i.e. no cross-price effects) – this is a very strong simplifying
assumption, which is relaxed in the more general model
▪ Commodities are produced competitively at constant returns to scale:
– pi = MC = ci where ci units of labour needed to produce commodity i
– Equilibrium prices are independent of the level of demand
– In equilibrium supply will meet any level of demand.

▪ Producer price: pi = ci
▪ Consumer price: qi = pi + ti
▪ Governments revenue requirement: R = t1x1 + t2x2
Inverse elasticity rule Consumers problem
56

▪ The consumers buys two taxed goods and sells labour

𝑚𝑎𝑥 𝑈 𝑥0 , 𝑥1 , 𝑥2 𝑠. 𝑡. 𝑞1 𝑥1 + 𝑞2 𝑥2 = 𝑥0

ℒ = 𝑈 𝑥0 , 𝑥1 , 𝑥2 + 𝛼 𝑥0 − 𝑞1 𝑥1 − 𝑞2 𝑥2

▪ First order conditions 𝜕ℒ

= 𝑈 ′ 𝑥0 + 𝛼 = 0 𝑈 ′ 𝑥0 = −𝛼
𝜕𝑥0

𝜕ℒ
= 𝑈 ′ 𝑥1 − 𝛼𝑞1 = 0 𝑈 ′ 𝑥1 = 𝛼𝑞1
𝜕𝑥1

𝜕ℒ
= 𝑈 ′ 𝑥2 − 𝛼𝑞2 = 0 𝑈 ′ 𝑥2 = 𝛼𝑞2
𝜕𝑥1

▪ Where α is the marginal utility of income

▪ NB the consumer does not internalize (is not affected by) government’s budget
Inverse elasticity rule Government’s problem
57
▪ Government’s budget constraint is R = t1x1 + t2x2
▪ Since ti = qi – pi we can rewrite this as q1x1 + q2x2 = R + p1x1 + p2x2
▪ The government must choose an optimal tax rate that maximizes the consumer’s utility
subject to the government’s revenue constraint.

max ℒ = 𝑈 𝑥0 , 𝑥1 , 𝑥2 + 𝜆 𝑞1 𝑥1 + 𝑞2 𝑥2 − 𝑅 − 𝑝1 𝑥1 − 𝑝2 𝑥2
𝑥1 ,𝑥2
▪ NB
– the labour supply is endogenously determined through the consumer’s maximization, so
we can substitute x0 for 𝑥0 = 𝑞1 𝑥1 + 𝑞2 𝑥2
– The consumer’s choices of xi feed directly to the governments revenue constraint (this is
the source of deadweight loss)
– Inverse demand function 𝑞𝑖 = 𝑞𝑖 𝑥𝑖
(since demand is independent; otherwise its too complicated)

– Which means that: 𝑥0 = 𝑞1 (𝑥1 ) ∙ 𝑥1 + 𝑞1 (𝑥2 ) ∙ 𝑥2

▪ So the first order condition (for commodity 1) becomes

𝜕ℒ 𝜕𝑞1 𝜕𝑞1
= 𝑈 ′ 𝑥1 + 𝑈 ′ 𝑥0 𝑞1 + 𝑥1 + 𝜆 𝑞1 + 𝑥1 − 𝑝1 = 0
𝜕𝑥1 𝜕𝑥1 𝜕𝑥1
Inverse elasticity rule Government’s problem 2
58
𝜕ℒ 𝜕𝑞1 𝜕𝑞1
= 𝑈 ′ 𝑥1 + 𝑈 ′ 𝑥0 𝑞1 + 𝑥1 + 𝜆 𝑞1 + 𝑥1 − 𝑝1 = 0
𝜕𝑥1 𝜕𝑥1 𝜕𝑥1

▪ But from the consumer’s optimisation we have: 𝑈 ′ 𝑥1 = 𝛼𝑞1 𝑈 ′ 𝑥0 = −𝛼

▪ And remembering that ti = qi – pi
▪ We can rearrange and simplify to get 𝛿𝑞 𝛿𝑞
−𝛼 𝑥1 𝛿𝑥1 + 𝜆𝑡1 + λ 𝑥1 𝛿𝑥1 = 0
1 1

▪ Now, with respect to the post-tax price, the definition of elasticity is

𝑞 𝜕𝑥 1 𝑥 𝜕𝑞
𝜀𝑑 ≡ =
𝑥 𝜕𝑞 𝜀𝑑 𝑞 𝜕𝑥
▪ So we can substitute to get
𝑞1 𝑞1
−𝛼 + 𝜆𝑡𝑖 + λ =0
𝜀1𝑑 𝜀1𝑑

▪ And further rearrange and simplify to conclude and state the inverse elasticity rule

𝑡1 𝜆−𝛼 1
=−
𝑝1 +𝑡1 𝜆 𝜀1𝑑
Inverse elasticity rule
59

𝑡𝑖 𝜆−𝛼 1 • α is the consumers marginal utility of income

=− •
𝑝𝑖 +𝑡𝑖 𝜆 𝜀𝑖𝑑 λ is the utility cost of government’s marginal revenue
• Since λ includes deadweight loss λ > α
• εd < 0 so the tax rate is positive.

▪ The proportional rate of tax on good i should be inversely related to its price
elasticity of demand
𝜆−𝛼
▪ The constant of proportionality is the same for all goods
𝜆

▪ The goods with inelastic demand have lower deadweight loss, and must
therefore attract a higher rate of taxation.
▪ The assumption of independent demand is relaxed to give a more general
version
▪ Very efficient, but is it equitable?
Ramsey tax rule Stiglitz and Rosengard version
60

𝑡𝑖 1 1
=𝑘 𝑑+ 𝑠
𝑞𝑖 𝜀𝑢 𝜀

Where: k depends on revenue requirement

t tax per unit
q after tax price
εd compensated elasticity of demand
εs elasticity of supply

The main conclusion as εs → ∞ t/q → k/ εd

 61

Optimal income taxation

Personal income tax brackets in South Africa
62
Table 4.4 Personal income tax rates and bracket adjustments
2018/19 2019/20
Taxable income (R) Rates of tax Taxable income (R) Rates of tax
R0 - R195 850 18% of each R1 R0 - R195 850 18% of each R1
R195 851 - R305 850 R35 253 + 26% of the amount R195 851 - R305 850 R35 253 + 26% of the amount
above R195 850 above R195 850
R305 851 - R423 300 R63 853 + 31% of the amount R305 851 - R423 300 R63 853 + 31% of the amount
above R305 850 above R305 850
R423 301 - R555 600 R100 263 + 36% of the amount R423 301 - R555 600 R100 263 + 36% of the amount
above R423 300 above R423 300
R555 601 - R708 310 R147 891 + 39% of the amount R555 601 - R708 310 R147 891 + 39% of the amount
above R555 600 above R555 600
R708 311 - R1 500 000 R207 448 + 41% of the amount R708 311 - R1 500 000 R207 448 + 41% of the amount
above R708 310 above R708 310
R1 500 001 and above R532 041 + 45% of the amount R1 500 001 and above R532 041 + 45% of the amount
above R1 500 000 above R1 500 000
Rebates Rebates
Primary R14 067 Primary R14 220
Secondary R7 713 Secondary R7 794
Tertiary R2 574 Tertiary R2 601
Tax threshold Tax threshold
Below age 65 R78 150 Below age 65 R79 000
Age 65 and over R121 000 Age 65 and over R122 300
Age 75 and over R135 300 Age 75 and over R136 750
Source: National Treasury
The EITC
63

Chetty (2012)
Lump sum taxes and distortions
64

▪ From an efficiency point of view we should fund expenditure with lump sum taxes
▪ Poll tax (efficient but highly regressive)
▪ If we are concerned with equity or redistribution we need individual specific lump sum tax
related to intrinsic ability
▪ E.g. A height tax:
– In a world where height was was a perfect predictor of income, we could impose a lump
sum tax on height
– Height is observable
– It is endowed and cannot be changed (leaving aside childhood nutrition etc)
▪ But in the real world
– We cannot directly observe individual’s ability
– Revealing ability is not incentive compatible
– Therefore we must tax economic outcomes such as income or consumption
– When we tax outcomes we change incentives to attain those outcomes (e.g. reducing the
income associated with labour may reduce the supply of labour)
▪ This leads to distortions
▪ And distortions lead to a trade off between equity and efficiency
Optimal taxation and Mirrleesian approach
65

▪ Consideration of the optimal system must be flexible enough to consider a variety of forms

– Whereas commodity taxation is typically restricted to τX or (p+t)X

– Income taxation schemes take the form of t(X)

▪ Take account of efficiency objectives (labour supply)

– Positive question: how does income taxation affect labour supply

▪ Take account of redistributive objectives (equity)

– Normative question: how should the income tax structure be determined

▪ Optimal tax system: maximizes social welfare (equity and efficiency) subject to raising
governments revenue requirement
Taxation and labour supply model
66

▪ Workers vary hours of work/leisure to optimize income/consumption subject to a budget

constraint
▪ Assuming there are no savings, so consumption is identical to income

𝑧
𝑧 = 𝑤ℓ ℓ=
𝑤
𝑧
𝑈 = 𝑈 𝑥, 𝐿 – ℓ = 𝑈(𝑥, ℓ) 𝑈 = 𝑈 𝑥,
𝑤
s.t. 𝑝𝑥 = 1 − 𝑡 𝑤ℓ s.t. 𝑝𝑥 = 1 − 𝑡 𝑧
𝑝𝑥 = 1 − 𝑡 𝑧
Consumption

Consumption
1 − 𝑡 𝑤𝐿
𝑝

𝒙∗ 𝒙∗

L - ℓ∗ Leisure 𝒛∗ Before tax income

 Tax increase and labour supply (intensive margin)
67
▪ What is the impact of a tax increase on hours worked?
– Substitution effect is always the right sign
– Income effect can move either way
▪ The outcome is therefore indeterminate (without specification)
▪ NB:
– Movement from a to b is the compensated effect
– Movement from a to c is the uncompensated effect
Consumption

a Consumption a
b

Leisure Before tax income

 Tax threshold
68
▪ Tax is now paid only on earnings above z*
▪ At point a no tax is paid, at point c tax is paid
▪ Many consumers will “bunch” at b
– At b an extra unit of labour receives net-of-tax return of [1-t]w, whereas the previous marginal
unit received w.
– If a consumer is optimizing at b, a change in the marginal tax rate might not change the
optimal point
– While there will be no substitution effects, income effects still apply
▪ Interior solutions (a, c) and corner solution (b)
▪ Most tax systems have several threshold points
Consumption

Consumption
c
c

b
b

a
a

L - z*/w Leisure z* Before tax income

 Labour force participation (extensive margin)
69

▪ What if the number of working hours cannot be varied continuously to arrive at preferred
income
▪ In practice, the hours of work are fixed: there is a minimum working day, regulated by firms
and firm-union agreements
▪ Workers must “choose” whether on not to work or not (useful to think of non-working
spouse)
▪ A worker who was indifferent between working and not working is now committed to not
working
Consumption

Consumption

L - ℓ𝑚𝑖𝑛 Leisure z* Before tax income

Evidence?
70
▪ Theory suggests three major issues in the relationship between labour supply and taxation
– Interior solution can go either way (income and substitution)
– Corner solutions could make labour supply insensitive to tax changes
– Participation decision can be sensitive to taxation
▪ Survey evidence
– “Surveys on labor supply have normally arrived at the conclusion that changes in the tax
rate have little effect on the labour supply decision”
– Lawyers and accountants at the intensive margin
– Weekly overtime
– Suggestive of a vertical labour supply function
– Income effect offsetting the substitution effect?
▪ Econometric estimation of the labour supply elasticities
– Income effect always negative, an usually outweighs substitution effect
– Men seem to have almost vertical/negative supply function
Labour-supply elasticities
– Women more elastic
Married women Married men Single Mothers
USA UK USA UK USA UK
Uncompensated wage 0.45 0.43 -0.03 -0.23 0.53 0.76
Compensated wage 0.90 0.65 0.95 0.13 0.65 1.28
Income -0.45 -0.22 -0.98 -0.36 -0.18 -0.52
 Optimal income taxation Model setup
71
▪ Two commodities: consumption (x) and labour (ℓ)
▪ 𝑧 𝑠 =𝑠∙ℓ 𝑠 s measures the hourly output of the consumer, equal to the wage rate
▪ What is the optimal income tax function: T(z)
▪ 𝑥 = 𝑐 𝑧 = 𝑧 − 𝑇(𝑧)
▪ 𝑐 ′ 𝑧 = 1 − 𝑇′(𝑧) 𝑥 = 𝑐(𝑧)

▪ Note that:
– On 45-degree line 𝑇(𝑧)Ƹ
𝑐 ′ 𝑧 = 1 𝑠𝑜 𝑇 ′ 𝑧 = 0
𝑥ො 𝑐 𝑧 = 𝑧 − 𝑇(𝑧)
– At point a c
𝑐 ′ 𝑧 > 1 𝑠𝑜 𝑇 ′ 𝑧 < 0 b
– At point b
𝑐 ′ 𝑧 < 1 𝑠𝑜 𝑇 ′ 𝑧 > 0
▪ A negative (marginal) tax rate implies a a
marginal subsidy
▪ The after tax wage for additional work will
be greater than the before tax wage (e.g.
EITC) 45○
z
𝑧Ƹ
Optimal income taxation High and low skill consumers
72
𝑧
▪ All consumers have the same utility function: 𝑈 = 𝑈 𝑥,
𝑠

▪ But different levels of skill: sL and sH

▪ A high-skill consumer takes less labour time to achieve a given level of income than a low skill
consumer
𝑥 = 𝑐(𝑧)
▪ So, at any income consumption
pair 𝑥,ෝ 𝑧Ƹ the indifference curve
of the high-skill consumer is UL
flatter (agent monotonicity)
UH
▪ Taxation is on income choice
[T(z)] not on skill (s) so both
𝑐 𝑧 = 𝑧 − 𝑇(𝑧)
face the same consumption
function
𝑥ො
▪ High-skill consumer will always
(choose to) earn strictly more
than low-skill consumer

z
𝑧Ƹ
Optimal income taxation Maximum marginal tax rate
73

▪ Between a and b, increased income (work effort) leads to lower consumption

▪ Given preferences as stated, no consumer would choose a point between a and b
▪ 𝑐 𝑧 = 𝑧 − 𝑇(𝑧) => 𝑐 ′ 𝑧 = 1 − 𝑇′(𝑧)
▪ Therefore: 𝑐 ′ 𝑧 ≥ 0
𝑥 = 𝑐(𝑧)
▪ Which implies: 𝑇′(𝑧) ≤ 1

▪ So the optimal marginal tax rate

must be strictly less than 100%
𝑐 𝑧 = 𝑧 − 𝑇(𝑧)

z
Optimal income taxation Lower limit?
74

▪ 𝑐 𝑧 = 𝑧 − 𝑇(𝑧) => 𝑐 ′ 𝑧 = 1 − 𝑇′(𝑧)

▪ Note: On 45-degree line: 𝑐 ′ 𝑧 = 1 𝑠𝑜 𝑇 ′ 𝑧 = 0 marginal tax rate is zero
At point a 𝑐 ′ 𝑧 > 1 𝑠𝑜 𝑇 ′ 𝑧 < 0 marginal tax rate is negative
At point b 𝑐 ′ 𝑧 < 1 𝑠𝑜 𝑇 ′ 𝑧 > 0 marginal tax rate is positive

𝑥 = 𝑐(𝑧)
▪ We showed that (optimally)
the marginal tax rate should be
less than 1
▪ But is there an optimal lower
𝑐 𝑧 = 𝑧 − 𝑇(𝑧)
limit for T’(z)?
▪ To answer we must say 𝑥ො
something about the b c
welfare function and equity

45○
z
𝑧Ƹ
Utilitarian egalitarianism
75

Δ𝑈 𝐻

Δ𝑈 𝐿

Y
R1000 R1000
Optimal income taxation Can marginal tax rate be negative?
76

▪ Utilitarian welfare function

▪ Two types of consumers –
low skill and high skill c1
▪ The c1(z) has a gradient
greater than 1
UH

h1
𝑥ℎ1

UL

𝑥𝑙1 l1
45○
z
𝑧𝑙1 𝑧ℎ1
Optimal income taxation Negative marginal tax rate?
77

▪ The c1(z) has a gradient

greater than 1
▪ The c2(z) gradient is 1 c1
▪ Keeping income fixed the
new consumption function c2
changes consumption by the
same amount – positive for
the low skill, negative for the
h1
high skill
𝑥ℎ1
▪ Govt revenue has not
changed (aggregate income 𝑥෤ℎ
and consumption are the ℎ෨ 1
same)
▪ But aggregate welfare has
improved (because dU/dc is
greater for the lower income 𝑙ሚ1
individual) 𝑥෤𝑙

𝑥𝑙1 l1
45○
z
𝑧𝑙1 𝑧ℎ1
Optimal income taxation Negative marginal tax rate?
78

▪ Now let consumers optimize

▪ Relative to 𝑙ሚ1 and ℎ෨ 1 utility is
improved for both c1
▪ Government has not lost
revenue as the movements are
c2
along a zero marginal tax rate
curve
▪ So, with the same government
revenue, c2 must have higher
social welfare than c1 𝑥ℎ2
𝑥෤ℎ h2
▪ So c1, with a negative marginal
tax rate cannot be optimal ℎ෨ 1
▪ An optimal tax rate must be non-
negative
▪ 𝑇′ 𝑧 ≥ 0 𝑙ሚ1
𝑥෤𝑙
▪ Note: we have not made any l2
conclusion on the average tax 𝑥𝑙2
rate (𝑇(𝑧)Τ𝑧) which could still be
negative 45○
z
𝑧𝑙2 𝑧𝑙1 𝑧ℎ1 𝑧ℎ2
Optimal income taxation Top end
79

▪ What marginal tax rate should we impose on the highest skill people?
▪ Lets post ABC as a candidate for optimality
▪ At point B the gradient of the function is less than 1, due to a positive marginal tax rate
▪ The high income consumer pays over tH to the government

C
tH

45○
z
Optimal income taxation Top end
80
▪ Along alternative tax schedule ABD, the line BD is parallel to the 45 degree line (i.e. with
zero marginal tax rate)
▪ The consumer moves to point b
▪ The government still gets tH
▪ There is a pareto improvement (nobody is worse off as a result of the welfare improvement)
▪ The optimal marginal tax rate for the highest skill person is zero
x
tH D

b C

45○
z
Optimal tax in theory and practice
81

▪ Optimal tax theory generates three key results:

– The highest marginal rate must be strictly less than 1
– Marginal tax rates should always be greater than zero
– The marginal rate on top income earners should be zero
▪ Only the first conclusion is observed in practice (and its not a very interesting result)
▪ The other two results are honoured in the breach:
– The EITC has a negative marginal rate
– Top marginal rates are the highest
▪ And the overall conclusion that optimal tax rates should be between zero and one does not
provide much practical guidance.
82