Estimation of the Household Sharing Rule

Drug Consumption and
Intra-household Distribution of Resources:
The Case of Khat in Djibouti
Seminar
Federico Perali
Department of Economics
University of Verona, Italy
Assistance in Institutional Strengthening and Agricultural Policies
Damascus, September 15 2002
Motivation
Who is better off?
• A poor child living in a rich household or a rich child
living in a poor household?
Economists may have something to say in terms of
material needs ... less in terms of immaterial needs...
Objectives
• Identify the sharing rule between adults and children in Djiboutian
families
•
We use Khat consumption and expenditures on children’s education to
learn about the intrahousehold decision process
•
Definition. Sharing rule: is the rule governing the intrahousehold
allocation of resources (money, time, goods).
• It is a measure of household inequality and of the distribution of power
among household members.
• Being a function it is a policy interesting object.
Introduction to the analysis of the intrahousehold allocation of resources
• TRADITIONAL APPROACHES: FAMILY AS A
“BLACK BOX”
– UNITARY FRAMEWORK: THE FAMILY UTILITY
FUNCTION IS INCREASING IN THE UTILITY OF THE
MEMBERS, INDIVIDUAL AND HOUSEHOLD WELFARE
MOVE IN THE SAME DIRECTION
• ALTERNATIVES APPROACHES: FAMILY AS A
DYNAMIC ORGANIZATION OF INDIVIDUALS
Introduction to the analysis of the intra-household allocation of
resource
• THE INTRA-HOUSEHOLD ALLOCATION OF RESOURCES IS
EXPRESSION OF THE DISTRIBUTION OF THE BARGAINING
POWER WITHIN THE FAMILY AND OF THE UNDERLYNG
DECISIONAL PROCESSES
• TAKING INTO ACCOUNT THE INTRA-HOUSEHOLD
DISTRIBUTION OF RESOURCES IS IMPORTANT TO
MESASURE BOTH INDIVIDUAL AND HOUSEHOLD
WELFARE
• THE WELFARE OF THE FAMILY MEMBERS DEPENDS ON
HOW THE INCOME RECIPIENTS DECIDE TO SHARE THEIR
RESOURCES BETWEEN INDIVIDUAL CONSUMPTION AND
THE HOUSEHOLD BUDGET
Sociological models
• For a traditional family with the husband as the breadwinner
• ALTRUISTIC HUSBAND
• PATERNALISTIC AND BENEVOLENT HUSBAND
• EGOISTIC HUSBAND
Game theory as an instrument for interpreting
the allocation mechanism
Family decisions as a bargaining problem
• COOPERATIVE MODELS
max[ Um( xm, lm) | Uf ( xf , lf )]
max[ Um( xm, lm)  (1   )Uf ( x f , lf )]
m, f N
m, f
å å p x xij £ å wi hi  y ,
i j =1
i
li = T  hi
xij  0, li  0, hi  0,
Game theory as an instrument for interpreting
the allocation mechanism
• THE COOPERATIVE NASH SOLUTION AS THE MAX OF
THE HOUSEHOLD WELFARE FUNCTION
W = (Um  Vm)  (Uf  Vf ) 
• COOPERATIVE MODELS: PARETO EFFICIENCY IMPLIES
THE EXISTENCE OF THE SHARING RULE
The Edgeworth’s Box and the cooperative Nash
solution
Um
Um
W=(Uf-Vf(Um-Vm)
A
Core
Set of
negotiation
Contracts
curve
Uf
e Threat point
B
Uf
Collective models and Sharing Rule
• THE COLLECTIVE MODEL ENCOMPASSES THE COOPERATIVE
AND NON COOPERATIVE MODELS WHEN THE SOLUTION S
PARETO EFFICIENT
• GIVEN THE EFFICIENCY CONDITION, A TWO STAGE DECISION
PROCESS IS ASSUMED:
– FIRST, THE TOTAL HOUSEHOLD INCOME IS ALLOCATED AMONG
THE DIFFERENT MEMBERS ACCORDING TO THE SHARING RULE
ESTIMATED FROM THE OBSERVABLE BEHAVIOR;
–
SECONDLY, EACH PARTNER DECIDE WHAT CONSUME ON THE
BASE OF AN INDIVIDUAL BUDGET CONSTRAINT
The Collective Model and the Sharing Rule
The Pareto household program can be rewritten as an
individual problem (Chiappori 1992)
max U i ( xi , li ) i = m, f
N
åp
j =1
x
xij £ wi hi  i wm , w f , y  = Y
li = T  hi
xij  0, li  0, hi  0,
where phi(w,y) is the sharing rule describing the intra-household
allocation process.
Khat in Djibouti
•
An individual .... Good (bad ?) and a social plague
•
Women oppose khat consumption, an amphetamine, by the males because
khateur household-heads are likely to be more irritable and are less aware of
the family needs.
•
Djibouti is characterized by widespread child malnutrition. Especially among
poor households there are significant differences in the nutritional status of
children and women between households with and without khateur members.
•
While the shares of education and health in total expenditure are low across all
population quintiles, the share of khat is relatively high across all population
segments indicating substitution between khat and other goods, especially
among the poor.
•
Khat share of total budget: 8% poor, 18 % medium, 23 % rich segments of the
population
Definitions
•
Definition 1: an exclusive good or bad is a private good-bad used by only one
member of the household
– An exclusive good or bad is thus a good or bad for which it is possible to
unequivocally identify the user.
•
Definition 2: an assignable good or bad is a private good or bad whose
consumption by each member of the household can be observed.
– Assignability depends on the fact that it is possible to observe who consumes the
good and in what proportion.
•
In the case of Djibouti it is appropriate to consider khat and cigarettes an
exclusive good of the head of the household and of the adult males in general.
Assumptions
•
Assumption 1: Consumption Set up. Partner's labour supply fixed and absence of child
labour .
•
Assumption 2: Children Representative. Children elect the mother (not the father) as
the representative of their preferences and wills.
•
Assumption 3: Individual Preferences. Wife's preferences are egoistic with respect to
the husband and altruistic with respect to the children; husband's preferences are caring.
•
Assumption 4: Exclusivity. Mother and children do not consume khat. Khat is an
exclusive good consumed by the adult males and education an exclusive good of
children.
•
Assumption 5: No addiction. Khat consumption does not generate addiction in the strict
sense of physical dependence.
•
Assumption 6: Non-Consumption. The consumption of khat (or education) is absent in
some families. This implies the presence of corner solutions.
Khat Consumption and Labour Supply
Decisions
• NEGATIVE HEALTH
CONSEQUENCES OF
CHEWING KHAT
Quintiles
• DOES KHAT
CONSUMPTION AFFECT
THE LABOUR SUPPLY
DECSIONS AND THE
LABOUR
PRODUCTIVITY OF THE
KHATEURS MEMBERS?
Househol heads
KHATEURS
NON KHATEURS
first
67.37
40.41
second
68.15
61.32
third
80.4
64.51
fourth
78.01
79.41
fifth
78.42
84.13
percentage of employed household-heads in kahteur
and non khateur families
Khat consumption and Labour supply
decisions
• DOES KHAT
CONSUMPTION AFFECT
THE LABOUR SUPPLY
DECSIONS OF THE NON
KHATEURS MEMBERS?
Quintiles
Wifes
KHATEURS
NON KHATEURS
first
16.51
6.8
second
11.12
6.6
third
8.61
10.42
fourth
9.82
7.21
fifth
18.45
21.92
percentage of employed wifes in kahteur
and non khateur families
Household-head labour supply estimation: two stage Heckman estimation method
Annual labour income
Variables
education
age
informal sector
public administration
services
trade
region
nationality
expenditure share on khat
public transfers
privat transfers
quintile di appartenenza
children 0-5 years
constant
probit equation
number of students
number of workers males
number of workers females
if the wifes works
region
nationality
education
age
age square
children 0-5 years
expenditure share on khat
public transfers
privat transfers
if common budget administration
if own house
quintile
constant
rho
sigma
lambda
coefficients
0.0318
0.021
-0.094
0.016
0.083
0.146
-0.013
-0.001
-0.63
8.15E-07
1.27E-07
0.156
-0.039
12.074
0.161
-0.13
0.167
-0.577
-0.132
0.0325
0.023
0.078
-0.001
0.088
1.404
-2.66E-07
-1.71E-07
0.468
0.226
-1.865
-0.508
0.986
-0.502
standard
error
0.0074
0.003
0.08
0.076
0.11
0.093
0.089
0.149
0.228
2.98E-07
8.83E-08
0.027
0.025
0.235
0.031
0.084
0.082
0.111
0.102
0.167
0.01
0.029
0.034
0.313
3.13E-01
9.68E-08
2.63E-07
0.154
0.085
0.033
0.661
.
0.021
0.011
t
P>|t|
4.294
6.112
-1.168
0.212
0.762
1.568
-0.144
-0.01
-2.767
2.737
1.441
5.793
-1.539
51.289
0,000
0,000
0.243
0.832
0.446
0.117
0.885
0.992
0.006
0.006
0.15
0,000
0.124
0,000
5.219
-1.552
2.034
-5.211
-1.286
1.951
2.251
2.721
-3.542
2.595
4.482
-2.751
-6.49
3.03
-4.982
6.787
-2.822
0,000
0.121
0.042
0,000
0.198
0.051
0.024
0.007
0,000
0.009
0,000
0.006
0,000
0,002
0,000
0,000
0,005
Wife labour supply estimation: two stage Heckman estimation method
Annual labour income
Variables
education
age
informal sector
public administration
services
trade
region
nationality
expenditure share on khat
public transfers
privat transfers
quintile di appartenenza
children 0-5 years
constant
probit equation
number of students
number of workers males
number of workers females
if the husband works
region
nationality
education
age
age square
children 0-5 years
expenditure share on khat
public transfers
privat transfers
if common budget administration
if own house
quintile
constant
rho
sigma
lambda
0.055
0.039
0.025
-0.09
-0.178
-0.349
0.021
0.243
0.246
-1.38E-07
1.00E-06
0.221
0.208
10.349
standard
error
0.034
0.011
0.195
0.28
0.297
0.258
0.318
0.368
0.597
4.26E-07
1.99E-07
0.067
0.075
1.116
0.043
-0.137
0.226
-0.561
0.416
0.254
0.114
0.105
-0.001
-0.059
1.005
-2.28E-07
-6.09E-07
0.041
-0.133
0.007
-4.025
-0.081
0.938
-0.076
0.036
0.108
0.081
0.109
0.16
0.191
0.015
0.037
0.001
0.042
3.55E-01
1.35E-07
2.89E-07
0.205
0.101
0.039
0.756
0.322
0.053
0.304
coefficients
t
P>|t|
1.612
3.845
0.131
-0.341
-0.599
-1.352
0.076
0.662
0.413
-0.324
0.511
3.276
2.748
9.724
0.107
0,000
0.896
0.734
0.549
0.177
0.939
0.508
0.681
0.746
0.61
0.001
0.006
0,000
1.167
-1.265
2.823
-5.104
2.596
1.333
7.547
2.844
2.341
-1.378
2.827
-1.691
-2.111
0.201
-1.317
0.187
-5.321
0.234
0.206
0.005
0,000
0.009
0.182
0,000
0.004
0.019
0.168
0.005
0.091
0.035
0.841
0.188
0.851
0,000
Khat consumption and labour supply decision
• KHAT CONSUMTION SIGNIFICANTLY AFFECTS THE
LABOUR SUPPLY DECISIONS OF BOTH KHATEURS AND
NON KHATEURS MEMBERS
• INCREASING LABOUR FORCE PARTICIPATION OF THE
WIFE IS NECESSARY IN ORDER TO PROVIDE THE
FINANCIAL MEANS FOR COVERING THE ESSENTIAL
NEEDS OF THE FAMILY
Khat Consumption and Intra-household
Resource Allocation
• DOES KHAT CONSUMPTION AFFECT THE FAMILY WELFARE
AND THE BUDGET SHARE DEVOTED TO OTHER GOODS?
• TWO EXCLUSIVE GOODS:
KHAT AS AN ADULT PRIVATE GOOD AND PULIC BAD
EDUCATION AS A CHILDREN GOOD
• INTRA-HOUSEHOLD SHARING RULE BETWEEN ADULT MALES
AND CHILDREN
Khat Consumption and Intra-household
Resource Allocation
• A PARAMETRIC MODEL FOR ESTIMATING THE SHARING RULE
• Estimation: generalized Heckman procedure to account for zero
consumption
• TWO DEMAND EQUATIONS: EXPENDITURE ON KHAT AND
CIGARETTES AND EXPENDITURE ON EDUCATION
ln( ek ) =  0   k ' yk   x ln( ek ) =  0   k ' yk   x ln(  ( z ) x)
ln( ec ) =  0   ' yc   x ln( ec ) =  0   c ' yc   x ln( x(1   ( z ))
• THE SHARING RULE EQUATION IS A FUNCTION OF EXOGENUS
VARIABLES AFFECTING THE DECSIONAL PROCESS BUT NOT
INDIVIDUAL PREFERNCES
 ( z) =  o   ' z
Estimation
with two exclusive goods the sharing rule is identified up to a
constant
• METHOD: JOINT ESTIMATION
wc =  o  å  ci y ci   x ln( x  (1   ( z )) =  o  å  ci y ci   x ln( x  (0.5  å  cj z j ))
i
i
j
wk =  o  å ki y ki   x ln( x   ( z )) =  o  å ki y ki   x ln( x  (0.5  å  kj z j ))
i
given
wk  wc = 1
i
0  0 = 1
 i   i = 0 "i
 x   x = 0 "x
z j ( x kj   x cj ) = z j ( x kj   x cj ) = z j x ( kj   cj ) = 0
®  kj =  cj ,
j
RESULTS: SHARING RULE ESTIMATES
Average estimated sharing rule: 0.64
Parameter
G21
G22
G23
G24
G25
G26
D10
D11
D12
D13
D14
D15
D16
D17
Definition
children financing
income pooling
share of m wage
age difference
wage
educ difference
constant
age of head
house ownership
region
children education
income
pdf
n. of children<18
D100
D111
D112
D113
D114
D115
D116
D117
constant
avg age of children
house ownership
region
wife education
income
pdf
age of wife
Estimate
-0.0210
0.1969
-0.0294
0.0219
-0.0213
0.0035
-7.3270
-0.0035
0.0038
0.1268
-0.0186
0.5506
0.8953
-0.0117
t-statistic
-1.1551
5.1528
-1.5610
0.6001
-2.4898
3.2194
-28.2241
-5.1550
0.2278
9.0592
-2.1747
30.3863
18.8847
-5.7934
-1.5388
0.0016
0.0357
0.0112
-0.0120
0.1256
-0.0080
0.0004
-13.8623
1.0853
4.1874
1.7952
-2.1829
15.9965
-0.3496
1.4839
A Graph of the Sharing Rule
Estimates (Y1*m/Y)
.736298
.335932
1800
18800
Household full income
Conclusions I
• The estimates of the sharing rule and labour supply
decisions show the negative impact of khat consumption
on the allocation of resources within the family and on the
welfare of non khateurs family members
• khat is at the same time a private and a public bad
• The Sharing Rule average value, 64%, indicates that khat
is an adult good the consumption of which detracts
resources that could be allocated amongst other members
of the household and reinvested in education.
Conclusions II
• Aspects related to income formation and estimates of the
sharing rule are inextricably linked: in Djibouti
unemployment is a long-term structural problem. In
general, males have greater access to the labour market:
they are more educated and find better qualified
occupations.
• Because the head of the household is the main wage
earner makes households dependent upon how the head
chooses to divide his earnings between private
consumption and the family budget.
• Income redistribution policies would need to take into
account the equality of household resource allocation to
eradicate poverty: children are recipients of residual
resources and women are negatively affected by an
unbalanced resource distribution.