A microeconometric model for analysing
efficiency and distributional effects of tax
reforms
Evidence from Italy and Norway
Rolf Aaberge (Research Department, Statistics Norway, Oslo)
and
Ugo Colombino (Department of Economics, University of Turin)
”La microsimulación como instrumento de evaluación de las
políticas: métodos y applicaciones”
Fundación BBVA, Madrid, 15-16 Nov. 2004
Various Modelling Approaches for
Analysing Tax Reforms
• Static microsimulation models
• Behavioural microsimulation models (Partial
equilibrium)
• General equilibrium models (CGE), where
labour supply is represented by a
representative agent
• Combining a behavioural microsimulation
model and a CGE
Outline of what follows
• The microeconometric model
• Labour supply elasticities (Italy and Norway)
• A simulation of some tax reform proposals (Italy)
• Looking for the optimal tax system (Norway)
• Testing the model: comparing model predictions to
the observed effects of a reform (Norway)
• Integrating the microeconometric model and a
General Equilibrium model (Norway)
We develop a model of labour supply
which features:
• simultaneous treatment of spouses’ decisions
• exact representation of complex tax rules
• quantity constraints on the choice of hours of
work
• choice among jobs that differ with respect to
hours, wage rate and other characteristics
Reference material
•
Aaberge, Colombino and Strøm, J. of Applied Econometrics, 1999
•
Aaberge, Colombino and Strøm, J. of Population Economics, 2000
•
Aaberge, Colombino and Roemer, Statistics Norway, Discussion paper
307, 2001
•
Aaberge, Colombino, Holmøy et al., Statistics Norway, Discussion paper
367, 2004
•
Aaberge, Colombino and Strøm, J. of Population Economics, 2004
•
+ some recent unpublished results
Labour supply elasticity
• The main purpose of behavioural
modeling is to account for labour supply
responses to policies
• Is labour supply really responsive, i.e.
elastic w.r.t. economic incentives?
Labour supply elasticity
• If, for example, we look at the overall labour
supply elasticity in Norway 1994, we read a
modest 0.12 ...
• …and then we would answer: NO, this is not
relevant, forget about behavioural modelling!
• But if we look BEHIND the aggregate figure
the picture changes quite a lot…
Labour supply elasticities w.r.t. wage
Married couples, Norway 1994
Household
income
decile
Female
Own
Male
Cross
Own
Cross
I
2.54
-0.29
1.77
-0.12
II
0.97
-0.67
1.17
-0.08
III-VIII
0.41
-0.47
0.31
-0.24
IX
0.20
-0.34
0.08
-0.14
X
0.26
-0.10
0.05
-0.42
All
0.52
-0.42
0.39
-0.23
Labour supply elasticities w.r.t. wage
Married couples, Italy 1993
Household
income
decile
Female
Own
Male
Cross
Own
Cross
I
4.44
0.82
0.32
0.06
II
2.31
-0.15
0.17
0.00
III-VIII
0.73
-0.24
0.10
-0.04
IX
0.20
-0.20
0.08
-0.03
X
0.13
-0.17
0.06
-0.02
All
0.66
-0.20
0.12
-0.02
A simulation of some reform proposals in
Italy
Two (old) ideas for reforming the taxtransfer system:
• Improving EFFICIENCY by flattening the
marginal tax rates
• Improving EQUALITY by introducing a
universal transfer or a minimum guaranteed
income
Gross income and net income under alternative tax
systems
Net income
Actual
Gross income
Gross income and net income under alternative tax
systems
Net income
FT
Actual
Gross income
Gross income and net income under alternative tax
systems
Net income
NIT
Actual
Gross income
Gross income and net income under alternative tax
systems
Net income
WF
Actual
Gross income
Percentage of “welfare-winners” under alternative tax
reforms
56
55
54
53
52
51
50
49
%
FT
NIT
WF
51,8
55
55,6
Percentage variations of Social Welfare and its
components
(Gini Welfare Function)
2.5
2
1.5
1
0.5
0
-0.5
-1
-1.5
Efficiency
Equality
Soc. Wel.
FT
NIT
WF
2.1
-1.2
0.9
0.8
0.7
1.5
1.1
0.5
1.6
Conclusions
• All the reforms are efficient
• FT is disequalising, but NIT and WF are equalising
• There is scope for designing tax systems that
produce bigger “cakes” and more equal “slices” too.
• Of course there might be even better reforms. In
what follows we look for optimal reforms …
Optimal taxation
An exercise for Norway
• We use the model to identify optimal
tax- transfer rules
• “Optimal” means maximizing a Social
Welfare Function
The model
max U(C, h, )
s.t.
C=f(wh, I)
(h, w, )  B
where:
U( ) = utility function
f( ) = tax-transfer rule
C = net income
h = labour supply
w = wage rate
I = exogenous income
 = other job characteristics
B = opportunity set
Simulating tax reforms
Given a new tax function t( ) and using the estimated
U( ) and B the simulation consists of solving for each
household
max U(C, h, )
s.t.
C=t(wh, I)
(h, w, )  B
to get new values of h and C
Optimal taxation
Class of tax-transfer rule
We consider 4-parameter tax-transfer rules:
Net = T - 1min(Gross,A) – 2max(0, Gross – A)
T = lump-sum transfer
1 and 2 = marginal tax rates for the two
brackets
A = cut-off value between the two brakets
Family of 4-parameter tax systems
Net income
2
1
T
A
Gross income
Optimal taxation:
Social Welfare Function
The Social welfare function is defined as the average
individual welfare () times (1 – Inequality Index)
There are many type according to how we define the
Inequality Index:
Bonferroni: W1    F1 (t) log tdt  (1  C1 )
Gini: W2  2 F1 (t)(1  t)dt  (1  G)
W3 
3
2
F
1
(t)(1  t 2 )dt  (1  C3 )
Utilitarian: W

 F1 (t)  
Optimal taxation: results
Social welfare in terms of individual welfare
W1
W2
(Bonferroni) (Gini)
Transfer b
7230
3650
(NOK)
Tax rate,
lowest
segment τ1
Tax rate,
upper
segment τ2
The lower
limit of the
upper
segment
A
Changes in
labour
supply
Income
inequality
(Gini coef.)
W3
W
(Utilitarian)
10510
930
Social welfare in terms of income
W1
W2
(Bonferroni) (Gini)
0
0
W3
W
880
(Utilitarian)
370
26
24
36
36
22
22
32
42
60
60
16
2
60
60
14
0
175000
475000
475000
475000 150000
475000 125000
150000
7.9
9.6
13.5
21.0
10.8
10.8
17.2
23.5
.222
.222
.255
.266
.219
.219
.252
.263
Out-of-sample prediction (Norway)
• In 2001 we are able to observe the effects of
a reform of the tax rule actually implemented
• We use the model estimated on 1994 data to
simulate the effects of the reform
• We then compare the model predictions to
the observed effects…
Out-of-sample prediction
Observed and predicted mean disposable income for couples,
single females and males in 1994 and 2001. 1000 NOK
Couples
Single males
Single females
Obs.
Pred.
Obs.
Pred.
Obs.
Pred.
1994
320
318
155
152
145
145
2001
456
452
207
218
184
192
Observed and predicted relative distributions of
disposable income in 2001
Couples
Single males
Single females
Deciles
Observed
Simulated
Observed
Simulated
Simulated
Simulated
1
50
49
41
42
45
47
2
68
64
54
55
56
61
3
77
74
65
67
68
71
4
83
83
76
76
79
79
5
89
90
87
86
90
88
6
95
98
97
97
101
98
7
102
107
107
108
111
108
8
111
117
119
121
123
121
9
125
131
137
141
139
138
10
199
187
218
207
189
188
Integrating the Micro- and the CGEmodel
Wage
Cash transfers
Capital income
CGE
Microeconometric
model
model
Labour supply
Integrating the Micro- and the CGEmodel
Simulation at 2050 of a flat tax that supports fiscal equilibrium
Flat Tax
1995
2050
24.0
32.0
18.3
22.9
Exogenous
(i.e. without the
Micro model)
Labour
Supply
Endogenous
(i.e. with the
Micro model)