L06: Redistribution and Social Assistance

1. Overview

• The design of social assistance programs is problematic from various

perspectives.

2. The Normative Argument for Redistribution: Insurance against Uncertainty

• Idea: Individuals want to escape anarchy but there is uncertainty as to

whether one will be rich or poor. Redistribution then serves as a
insurance.
• Model:
• Post constitutional incomes are: Yi , with Y2 > Y1
• The number of post-constitutional poor and rich are p and r.
• Individuals maximize

O = π 2U 2 (Y2 − T ) + π 1U1 (Y1 + B )

with
p r
π1 = , π2 = and rT = pB
p+r p+r

Then, maximizing O with respect to B

r p
max O = U 2 (Y2 − T ) + U1 (Y1 + B )
B p+r p+r

56
 dO r dU 2 p p dU1
=− +
dB p + r dY r p + r dY
r dU 2 p p dU1
=
p + r dY r p + r dY

dU 2 dU1
=
dY dY

An individual who is uncertain about her post-constitutional wealth wants
to make sure that the marginal utilities of all incomes are the same.

• Why do insurance markets fail?

o Assume two periods. All individuals are healthy in period 1. There is

though a chance that an individual gets disabled in period 2, which
leads to a lower income, YD < YH . An individual has therefore an
incentive to buy insurance.
o The population likelihood of becoming disabled is π D .The
individual’s perceived likelihood of becoming disabled is π i . A fair
insurance charges the (tax) premium T = π D B in order to pay a benefit
T
of B = .
πD

o Let the individual now maximize the expected utility from buying a
disability insurance (ignoring any discount factor):

max EU = U (YH − T ) + π iU (

YL + B ) + (1 − π 1 )U (YH )
T
   
YToday YTomorrow when disabled

dU (YH − T ) π i dU (YL + B )
EU ' = − + =0
dY πD dY

dU (YH − T ) π i dU (YL + B )
=
dY πD dY

57
 o Thus, if the individual’s risk self-assessment is the same like the
population average, it will buy T in order to assure that tomorrow’s
income is the same as today.
o People who know or think that their disability risk is below the
average

dU (YH − T ) dU (YL + B )
<
dY dY

will buy less T.

o Similarly, people who know or think that their disability risk is

above the average

dU (YH − T ) dU (YL + B )
>
dY dY
will buy more than T.

Private insurances will then have too few low risk and too many high
risk customers and therefore go bankrupt.

A mandatory insurance can therefore be Pareto-efficient.

3. The Positive Argument for Redistribution: Outcome of Democratic Practices

• Consider an economy with N consumers whose incomes range between a

minimum of zero and a maximum of ŷ .
• The income distribution is skewed to the right

58
 Count

Income
Mode Median Mean

• The government provides a public good G with G’>0 and G’’<0 that is
financed by a proportional tax t.
• The utility function of consumer i with income yi is

ui ( t , G ) = (1 − t ) yi + b ( G ) = (1 − t ) yi + ln G

• Denoting µ the mean income, the government’s budget constraint is

G
G = tN µ so that t =
Nµ
• Therefore,

 G 
max ui ( t , G ) =  1 −  yi + ln G
G
 Nµ 
with the f.o.c.

∂ui y 1
=− i + =0
∂G Nµ G

59
 Nµ
G* =
yi

• The median voter will determine the optimum level of G. Everyone with
an income less than the median voter preferred even more G.
• The optimum G is determined by

Nµ
G* =
ym
where ym is the income of the median voter.
µ
• The term is some kind of an inequality indicator. The further away is
ym

the mean income from the median income, the greater will be the size of
redistribution.

4. How to transfer income and design social assistance? The Basic Micro
Framework

• A household maximizes the Cobb-Douglas utility function

U = X α Y 1−α

s.t the budget constraint

I = PX X + PY Y

• The Lagrange function is

60
 L = X α Y 1−α + λ [ I − PX X − PY Y ]

with
LX = α X α −1Y 1−α − λ PX = 0

LY = (1 − α ) Y −α X α − λ PY = 0

and

Lλ = I − PX X − PY Y = 0

• Rearrange to

α X α −1Y 1−α P
= X
(1 − α ) Y X PY
−α α

αY P
= X
(1 − α ) X PY

with

Y=
(1 − α ) PX X and X=
α PY Y
α PY (1 − α ) PX

• Solve for Marshallian Demand functions

α PY Y  α  PY Y
I − PX X − PY Y = I − − PY Y = I − PY Y  1 + =I− =0
(1 − α )  1−α  1−α

61
 I
Y * = (1 − α )
PY

Similarly,

I
X* =α
PX

• The optimum utility is then

α 1−α
 I   I  1−α I
U = α   (1 − α )  = α α (1 − α )
*

 PX   PY  PX PY1−α
α

Obviously, an increase in the price of x leads to two effects: A lower

quantity demanded and a lower utility.

• What would be the demand function if the household could keep the
utility constant? This is the idea of the compensated or Hicksian demand
function. Rearrange U* to get

U * PXα PY1−α
(
I U * , PX , PY =) α α (1 − α )
1−α

• Inserting I (U * , PX , PY ) into Marshallian demand functions

1−α
αU * PXα PY1−α α 1−αU * PY1−α  α PY 
X = α
*
= =U*  
α PX (1 − α )
1−α
PX (1 − α )
1−α 1−α  (1 − α ) P
 X 

Similarly,

62
 α
 (1 − α ) PX 
Y =U 
* *

 α PY 

• Example: α=0.5, PX=0.25, PY=1, I=2

I 2 I 2

Marshallian demand: X * = α = 0.5 = 4 ; Y * = (1 − α ) = 0.5 = 1
PX 0.25 PY 1

Optimum utility: U * = 4 0.510.5 = 2

Assume that Px changes from PX=0.25 to PX=1

I 2 I 2

Marshallian demand: X * = α = 0.5 = 1 ; Y * = (1 − α ) = 0.5 = 1
PX 1 PY 1

Optimum utility: U * = 1 0.510.5 = 1

 Both demand for X and utility decrease.

What if only price changed, while utility of 2 could be held constant?
This asks for the compensated demand.

1−α
 α PY  1
0.5

Compensated demand: X =U 
* *
 = 2  =2
 (1 − α ) P 1
 X 
α
 (1 − α ) PX  1
0.5

Y =U 
* *
 = 2   =2
 α PY  1

Therefore, the Substitution Effect is S.E.=-2, the income effect

I.E.=-1. The good is a normal good.

Graphically:

63
5. Lessons from the Basic Micro Framework

• A lump-sum tax is more efficient than taxing one good

In the above example, assume the increase in price for good X is due
to a tax. The state then collects tax revenues from the individual of
TR=0.75.

It is now easy to show that if the state imposed a lump-sum tax of
0.75, the individual would have a utility of

α 1−α
 I   I  1−α I 1 2 − 0.75
U = α   (1 − α )  = α α (1 − α ) = = 1.25
*
α 1−α
 PX   PY  PX PY 2 0.250.510.5

which is greater than the utility of U=1 associated with distorted

prices.
∂U * ∂U *

The proof is easy: Just compare − vs. and show that
∂I ∂PX

64
 −U I*
<1
U P*X

The intuition is easy too: Both lump sum and commodity tax reduce
real income by the same margin. In addition to the real income
reduction, the commodity tax additionally distorts relative prices.

• A cash transfer is better than a price subsidy

Same proof as above

• A cash transfer is always as good as or better than an in-kind transfer.

An in kind transfer of, for example, XT=3, creates the budget

constraint

I + Px X T = PX X + PY Y for X>XT

while a cash transfer CT=PxXT creates the budget constraint

I + Px X T = PX X + PY Y for all X

Under the cash transfer, the household has therefore at least as many
consumption opportunities as under an in-kind transfer or more.

65

Graphical example: I=4, PX=1, PY=1, XT=4, CT=(1)(4)=4

Y
B+PX X
8
PY
In-kind transfer
7

Cash transfer
6

B
4
PY In-kind transfer

X
1 2 3 4 5 6 7 8
B B+PX X

PX PX

1 3
• Exercise: Assume the utility function U ( X , Y ) = X 4Y 4 . Using the
numbers from above graphical illustration, determine the optimum
consumption plan (X*, Y*, and U*) for the situation prior to any social
assistance as well as for a cash transfer of CT=4 and the in-kind transfer
of XT=4.

Solution initial equilibrium: X * = 1 , Y * = 3 , U * = 2.280

Solution CT=4: X * = 2 , Y * = 6 , U * = 4.559

Solution XT=4: X * = 4 , Y * = 4 , U * = 4 (why?)

• An Exception to the Rule:

66

Sometimes the donor of an in-kind transfer has a different objective
function than the recipient of a cash-transfer.

For example, international organizations such as the World Food
Program often give refugees an in-kind rather than a cash-transfer for
the following reasons:
o First, where refugees live, there are often no markets. And if
there are markets, they are very thin and a cash-transfer will
drive up prices. Inflation then prevents that the shift of the
budget constraint reaches its desired location. In other words,
the assumption of perfectly competitive markets often does not
hold. This is especially true in conflict areas.
o Second, a cash transfer may lead to perverse consumption
effects. For example, the World Food Program would like to
prevent that a cash transfer in a crisis region is spent on
weapons, drugs, prostitution and other non-basic needs items.
Often, however, refugees understate the long-term disutility
from an undersupply with basic needs and overstate the long-
term benefit from the consumption of non-basic needs like food.
o Example: In the above example, assume that X is a basic need
like food and Y a non-basic needs items. The
recipient (R) of social assistance gives consumption
of Y a higher weight than the donor (D). Let the two
1 2 1 1
objective functions be U R = X 3Y 3 and U D = X 2Y 2 .
Prior to any social intervention, the recipient chooses
the optimum consumption plan X*=1 and Y*=3. The
recipient perceives a utility of U R* = 2.28. . For the
donor, however, the utility is only U D* = 1.73. Assume
that the donor considers a supply of X=4 as essential

67
 to secure long-term health. As seen above, an in kind
transfer of X=4 will assure that the social assistance
recipient indeed consumes X*=4. With a cash-
transfer, however, the aid recipient consumes only
X*=3. The utility of a cash-transfer to the donor is
only U D* = 3.464 , as opposed to a utility of
U D* = 4 associated with an in-kind transfer. The aid

recipient, of course, preferred the cash-transfer.

o Graphical Illustration: Cash vs. In-kind transfer when donors and

recipients have different objective functions.

11
UR | Cash

UD | Cash

10

UR | In-kind
3
UD | In-kind
2
UR | Initial equilibrium
1
UD | Initial equilibrium
0
0 1 2 3 4 5 6 7 8

68
6. Further reading

Nicholson, W. (2005), Microeconomic Theory, Basic Principles and

Extensions, Ninth Edition, Thompson – South Western, Mason, Ohio,
Chapters 4-6 (as relevant to lecture notes).

69