Professional Documents
Culture Documents
Dr Snje
j Reiche
i h
Part 4
Motivation
Casual use of Social Indifference curves
Where do these come from?
How are the interests of different individuals weighed up in
creating such social indifference curves?
N
Normative
ti issues
i
(rather than positive analysis of GE)
2
Books
Varian Ch. 33
Cowell Ch. 9.1, 9.4, 9.5
xf y
yfz
xfz
Distinguish:
g
Bergson-Samuelson Social
Welfare Function
S
Social
i l welfare
lf function
f ti (which
( hi h corresponds
d to
t
individual utility function)
Mechanism for achieving this social welfare
function (through combining individual
preferences)
Arrow:
Social Choice Rule,
Social Welfare Functional
or
5
Constitution
Resource allocations
R
ll
i
should
h ld be
b judged
j d d only
l through
h
h their
h i
effects on individuals in the economy
JJohn
h L
Locke
k (1632-1704),
(1632 1704) Adam
Ad Smith
S i h (1723-1790),
(1723 1790) JohnJ h
Stuart Mill (1806-1873)
6
Non-Paternalism
2. Individuals are the best judges of their own
welfare
Alternative: Society
y as an Organism
g
Plato (424-348 BC), Karl Marx (1818-1883)
Thomas Hobbes (1588-1679)
Rulers determine goals for the organism (e.g.
q
, glory
g y of God,, triumph
p of ideology)
gy)
conquest,
Individuals matter only insofar as they can be used
as tools to achieve those goals
II.e.,
e A heart muscle cell has no independent
importance beyond helping pump blood
Examples: Divine right of kings: Britain (formerly!)
Communism
Fascism
8
I li it Assumptions
Implicit
A
ti
Individualism and Non-Paternalism
Process Independence
Anonymity
9
Implications
Ordinally measurable
2.
Key: an ordinal
K
di l utility
ili function
f
i is
i unique
i
up to a positive,
ii
monotonic transformation
u1a = xy
u1b = log x + log y
u1c = x 3 y 3
all represent the same preferences
The units have no meaning: convey no information about the
i t it off individual
intensity
i di id l 1s
1 preference
f
Hence, the units used for individual 1 convey no information that
Hence
can compare utility to individual 2s level of utility
11
u2
C
B
A
INCOMPLETE ORDERING
BUT: Pareto improvement possible
f
from
interior,
i
i so necessary condition
di i
u1
(although precise shape of UPS depend on the particular representation)
12
Compensation Tests
Cannot rank Pareto efficient allocations
defines set of all efficient allocations
Compensation Tests
Take x and y with u1 ( x) > u1 ( y ), u 2 ( x) < u 2 ( y )
Ask: How much would 1 (and 2) be willing to pay (accept) in
terms of monetaryy compensation
p
to gget x instead of yy?
If CV1 ( y, x) + CV2 ( y, x) > 0 individual 1 can compensate 2
for ggettingg x rather than yy. Can sayy that x is better than yy.
No need for interpersonal value judgement!
But:
Only hypothetical reallocation. Otherwise: real pareto improvement.
Implicit assumption that value of 1 is worth the same to all
individuals (so it does not matter who actually benefits).
Can lead to Paradox: Both x and y can p
pass the compensation
p
test.
New test: CV1 ( y, x) + CV2 ( y, x) > 0 and CV1 ( x, y ) + CV2 ( x, y ) < 0 ?
14
Criterion is not transitive!
Bergson-Samuelson
g
Social Welfare
Functions
Criteria to rank different allocations using explicit distributional
value judgements (i.e. degree of concern for equity)
Benevolent
B
l t social
i l planner,
l
defines
d fi
preference
f
ordering
d i using
i
particular utility representation of individual preferences
W (x ) = W (u1 (x ), u 2 (x ),..., u N (x ))
xf y
if andd only
l if
W (x ) > W ( y )
Cardinal
C
di l utility
tilit
function!
15
P
Properties
ti
W
0
uh
Paretian
u2
u1
16
Forms of B
B-S
S SWF: Utilitarianism
N
W = u1 + u2 + ...+ uN = uh
h=1
u2
u1
This does not imply that giving 1 to individual 1
has the same benefit as giving 1 to individual 2
17
u2
u1
Utilityy of individual 1 is a pperfect
f complement
p
for the utilityy of
individual 2
18
B-S SWF
Basis for showing how attitudes to inequality affect optimal allocations
u2
Utilitarian
R li
Rawlsian
45
Egalitarian
UPF
u1
19
Summary
Evaluation of value to society of different allocations
Individualistic
I di id li ti
Pareto
Outline: Arrow
Definition
Desirable Properties
Three possible resolutions of impossibility!
Focus on information requirements for individual
utility
tilit functions
f ti
Books:
Varian: Ch. 33
Cowell: Ch. 9.2
21
Example
p
3 Individuals in society, 3 alternatives A, B, C:
Preferences are given by
Af Bf C
1
B 2 C f 2 A
C f 3 A 3 B
Utility representation:
u1 (A) = 8
u2 ((A)) = 7
u1 (B) = 6,
u2 ((B)) = 280
u3 (A) = 5
u3 (B) = 5
u1 (C) = 1
u2 ((B)) = 280
u3 (C) = 4
23
Example:
Utilitarian B-S SWF
Arrows SCR,
Example
W ( A)) = W ((8,7,5)
, , ) = 4
W (B) = W (6,280,5) = 281
W (C) = W (1,280, 4) = 283
Bf AfC
24
2.
F must be
b defined
d fi d over all
ll possible
ibl sets off iindividual
di id l preference
f
orderings
Pareto principle
i i l [P]
3.
4.
Non-dictatorship [ND]
E ample continued
Example
contin ed
In the Example: There are 63=216
216 strict preference
ordering profiles
A f1 B f1 C
A f1 B f1 C
A f1 B f1 C
A f1 B f1 C
A f1 B f1 C
A f1 B f1 C
B f2 C f2 A
B f2 C f2 A
B f2 C f2 A
B f2 C f2 A
B f2 C f2 A
B f2 C f2 A
A f3 B f3 C
A f3 C f3 B
B f3 A f3 C
C f3 A f3 B
B f3 C f3 A
C f3 B f3 A
ordering
gI
CCX more!
26
Universality
The Social Choice Rule must map each set of preference
orderings
d i
i
into
a social
i l ordering
d i over the
h 3 alternatives,
l
i
for
f
example:
F ( I ) = F ( II ) = F ( III ) = F ( IV ) = { A f B f C }
F (V ) = F (VI ) = {C f A f B}
until
til F(CCXVI)
Pareto Principal
The SCR that sets F ( III ) = {C f A f B} violates [P].
Test: Does [P] say anything
hi about
b
F(IV)?
( )
27
Non-Dictatorship
Non
Dictatorship
The SCR that sets F f 2 violates [ND].
I:
A f1 B
II:
A f1 B
III:
B f1 A
IV:
Test:
Af B
Bf A
Af B
What possible SCRs satisfy all 4 requirements?
2
B f1 A
B f2 A
Impossibility Theorem
afb
bfc
cfa
not transitive
30
Solutions to Impossibility
Relax universality
Majority voting
works if single-peaked preferences
difficulties if multi-dimensional preferences
31
Single-peaked preference
For each individual
there is a uniquely
optimal policy. The
closer a policy is to
the optimal policy the
higher it is ranked.
utility
toptimal
tariff
32
Single-peaked
Single
peaked preferences
utility
capitalists
land owners
workers
t t
tariff
33
3: c f3 a f3 b
Individual 3
does not have
single peaked
preferences!
f
!
a
c
34
Relaxing IIA
First
First-past-the-post:
past the post: largest number of votes wins
Example :
A f1 B f1 C
A f1 B f1 C
A f2 B f2 C
C f2 A f2 B
B f3 A f3 C
B f3 A f3 C
C f4 B f4 A
B f4 A f4 C
A
B
1. Individuals rankings between A and B are the same in
the two sets but the social ordering changes
2 C is a competitor who is never elected but his presence
2.
35
affects the election between A and B.
Relaxing IIA
Borda voting: assign a score to each alternative, (1 to
worst, 2 to next etc.),
) add upp scores and rank alternatives
according to the total.
particular point system matters (eg.
(eg 1,
1 2,
2 4,
4 8 instead)
discarding the worst and recalculating may affect the
outcome
A f1 B f1 C
A f2 B f2 C
B f3 A f3 C
C f4 B f4 A
(1,2,3)
A=9
(1,2,4)
A = 11
B=9
B = 10
C=6
C =7
ABfC
Af BfC
discarding C :
A=6
B=6
AB
36
Imposing Cardinality
The only information for each individual allowed
b Arrow
by
A
iis an individuals
i di id l preference
f
ordering
d i
Corresponding utility functions are
ordinallyy measurable
non-comparable
[ ONC ]
37
C id the
Consider
th B-S
B S SWF with
ith ONC utility
tilit information
i f
ti
1 N
u = uh
N h=1
1
W = u + min
i (uh u ),
2
Allocation Y Allocation Z
u1
u2
4 1/2
5
38
)
u1 = 0.5u1
)
u2 = 4u2
u1
Allocation
Y
2
Allocation
Z
4
u2
24
16
7 1/2
39
Relaxing ONC
What extra measurability and comparability assumptions are
necessary to resolve Arrow?
Suppose individual utility:
cardinally measurable for each i
)
ui = i + i ui
i > 0
)
ui = + ui
> 0, i
Social choice rule exists and satisfies all 4 conditions: need extra
information on intensity and comparability
40
ui ( x ) ui ( x ' ) ui ( x ) + ui ( x ' )
ui ( x) ui ( x' )
=
=
u j ( x) u j ( x' ) u j ( x) + u j ( x' ) u j ( x) u j ( x' )
Summary
Arrow: how are individual preference orderings aggregated to
give society
societyss preference ordering?
Impossible under minimal assumptions (U,P,IIA,D)
(U P IIA D)
Limited information on individual preferences: only ordinal
information, so no intensity or comparability
Extra information enables aggregation to B-S SWF
42
Books:
Cowell Ch.
Ch 9.3-9.3.2,
9 3 9 3 2 13
13.7
7
Varian Ch
Ch. 31.9-31.13,
31 9 31 13 32.7-32.14
32 7 32 14
Second Best
an allocation
ll ti which
hi h is
i Pareto
P t efficient
ffi i t subject
bj t to
t resource
constraints and additional constraints (eg. imperfect
information)
45
46
max
x A , y A , x B , y B , K x , Lx , K y , L y
u A (x A , y B )
subject to
u B ( xB , y B ) = u B
f x ( K x , Lx ) = x A + x B
K = Ky + Kx
f y (K y , L y ) = y A + y B
L = L y + Lx
Production Efficiency
y
max
K x , Lx , K y , L y
f x (K x , Lx ) subject to:
f y (K y , L y ) = y
K = K y + K x ; L = L y + Lx
max f x (K x , Lx ) + y f y (K K x , L Lx )
K x , Lx
F.O.C.s
f y (K K x , L Lx )
f x (K x , Lx )
=
Ly
Lx
f y (K K x , L Lx )
f x (K x , Lx )
=
K y
K x
f K x , Lx
f K y , Ly
y
Lx
L
=
x
x
f K , L
f K y , Ly
K x
K y
48
Production Efficiency
Oy
x
y
MRTS KL
= MRTS KL
Potential allocation
of inputs
Area of Pareto
improvement
Ox
f K x , Lx
f K y , Ly
y
Lx
L
=
x
x
f K , L
f K y , Ly
K x
K y
49
Consumption Efficiency
OB
y
Area off P
A
Pareto
t
improvement
Potential allocation of
consumption goods
MRS xyA = MRS xyB
u
OA
u ( x , y ) u ( x , y )
B
x A
x
=
A
A
u ( x , y ) u ( x B , y B )
y A
y B
A
50
x A , y A , xB , y B
F.O.C.
u A
=0
x A
u A
+ F ' = 0
y A
subject to
u B ( xB , y B ) = u B
x A + xB = F ( y A + y B )
u B
=0
xB
u B
+ F ' = 0
y B
MRTxy
Overall Efficiency
y
MRTxy
MRTxy = MRS xy
MRSxy
PPF
x
MRTxy = MRS xy
52
No externalities
No
N public
bli goods
d
No asymmetries of information
54
(Aside on externalities)
Consumption externality
u A ( x A , y A , xB )
Assumption
i is
i NOT that
h this
hi sort off utility
ili function
f
i is
i ruled
l d
out
Rather
R th the
th assumption
ti is
i that
th t there
th is
i a price
i for
f eachh goodd
if there
th is
i a price
i (market)
(
k t) for
f eachh good,
d then
th the
th
externality becomes a pecuniary externality
55
px
MRS =
py
i
xy
So, if all
ll consumers face
f
the
h same prices,
i
then
h consumption
i efficiency
ffi i
Si il l with
Similarly
ith optimising
ti i i firms:
fi
eachh sets
t
Allocation of production between firms:
w
MRTS =
r
px
MRTxy =
py
j
KL
56
Caveats
Agents might not be price-takers if markets are
monopolized
li d or monopsonized
i d prices
i
do
d not
measure the marginal value of all activities.
Markets mayy not be complete:
p
how likelyy is it that
there is a market for clean air noise etc.? Future
markets,, markets for state contingent
g
commodities?
Equity vs Efficiency
FT1: market leads to the Pareto frontier
Pareto optimality: which point on the Pareto frontier
maximises
i i social
i l welfare?
lf ?
Is there a cost between moving to a point on the
frontier which improves social welfare? Is there a
trade off bet
trade-off
between
een efficiency
efficienc and equity?
eq it ?
58
Assumption 3: Convexity
Individual indifference curves are convex
rules out a preference for extremes (over mixtures)
59
60
61
Competitive equilibrium,
no intervention
OB
e yA + e yB
SWF*
Initial
endowment
e2
OA
exA + exB
e1
x
Nott supported
N
t d by
b
competitive equilibrium
prices
p
OA
P
Pareto
Effi
Efficient
i
S mmar
Summary
Necessary conditions for Pareto efficiency
First Welfare Theorem: competitive
p
market leads to Pareto
efficiency
S
Second
dW
Welfare
lf Theorem:
Th
any Pareto
P t efficient
ffi i t point
i t is
i achievable
hi bl
by LST and competitive markets
allows us to separate out efficiency from equity
65
Implications of FT2
Decentralisation result: redistribute with limited
government informational
i f
i l requirements
i
decentralised decisions will lead to a first-best Pareto
allocation
ll
i (satisfy
( i f the
h marginal
i l conditions)
di i )
Implications
Value of FT2
FT2 provides rationale for why government
i
intervention
i may be
b desirable
d i bl
Assumptions (particularly on information) are
veryy strong:
g
FT2 is not valuable to understand what form
interventions should take
second best analysis of policy intervention
68
MRS MRT
First best solution: is it possible to remove the distortion
directly? If so, then move back to the first best.
If not:
try to satisfy as many Pareto conditions as possible (i.e. ensure efficiency
in all other markets ignoring distortion )
or introduce offsetting distortion (second best policy)
69
Example
private
social
pcar = MCcar
< MCcar
2.
But cars and public transport are substitutes: price below marginal cost
induces switch away from car use
social
pcar MCcar
social
imposes
p
deadweight
g loss in public
p
transport,
p , but negligible
g g
since p public = MC public
CAR USE
p
PUBLIC TRANSPORT
Deadweight
g loss
SMC
PMC
q*
Reduction in
deadweight loss from
marginal reduction in
q: MB - SMC
MC
p*
q*
Introduced
deadweight loss from
marginal increase in
q: (P-MC)
(P MC)
71
u (c ) + v( z )
Assume v concave
73
c h , cl , z h , z l
s.t. ch + cl = ah (H z h ) + al (H zl )
First-order conditions for consumption:
u (ch )
ch :
=
ch
u (cl )
cl :
=
cl
74
v( z h )
zh :
= ah
z h
v( zl )
zl :
= al
zl
v( z h ) / z h ah
=
v( zl ) / zl
al
ah
>1
al
v( z h ) v( zl )
>
z h
zl
z h < zl
ah = wh and al = wl
(Th , Tl )
c i = a i (H z i ) T i
a h (H z h ) T h = a l (H z l ) T l
Since a h > a l and we want z h < z l we must have
Th > 0, Tl < 0
High ability workers should work more and be taxed, low
ability workers should work less and receive a benefit.
78
Implications
Impossible
p
to differentiate lump-sum
p
taxes by
y unobservable
characteristics!
Taxes must be based on observed behaviour (or characteristics)
When characteristics are private information, redistribution must
involve taxes that are distortionary.
1. Some conditions for first best Pareto Efficiency will not hold
2. Theory of the second best suggests Pareto Efficiency in
remaining markets may not be first best neither
Summary
Second Welfare Theorem: any Pareto efficient point is achievable
by LST and competitive markets
allows us to separate out efficiency from equity
80