CGE Modeling and Microsimulations by Dr. Dario Debowicz

Macro-micro modeling:
linking CGE models with micro-
simulation models
Dario Debowicz
April 2013

Different approaches to link macro and micro model
in macro-micro modeling.
Microsimulations
models combined
with macro models
Integrated
Layered
Non-behavioral
Through
households
Through the factor
markets
Behavioral Top-down
Top-down/ Bottom-
up
Cockburn 2006
Pdv. analysis for PSSP
Vos and Sanchez 2010
Bourguignon et al 2004
Debowicz 2007
Savard 2010,
Debowicz & Golan 2013

Illustration:
Micro-simulating the effects of capital
outflows on employment, poverty
and inequality for Argentina
(Debowicz 2007)

The macro shock and macro model
Non-residents deposits at domestic banks fall 60.8% in Argentina, from 32.9 billion
dollars (December 2000) to 12.9 billion dollars (December 2002).
Real-financial CGE model where money is included in the production function following
the tradition of Levhari and Patinkin (1968): liquidity affects the efficiency with which
the economy uses its real resources by allowing for greater specialization and
exchange.

The behavioural approach
“Layered” behavioural microsimulations approach developed by Anne-Sophie
Robilliard, François Bourguignon and Sherman Robinson (2008), RBR, which
captures the way rationing occurs in an imperfect labour market.
The main use of the microsimulation model (MSM) is to select individuals who are
barred from (or let in) jobs, making the selection depend on individuals’
characteristics, i.e. who is fired when the employment level shrinks
Stata code used in RBR provided by Anne-Sophie Robilliard.
Application: capital outflow suffered by Argentina Dec 2000 – Dec 2001.
Behavioural MS vs. Non-behavioural ones.

Five steps in behavioural MS
1. Specify a household income model consistent with the CGE
model
2. Estimate the household income model
3. Simulate capital outflow in the macro CGE model, and let the
CGE communicate the effects to the household income
model
4. Attribute the communicated macro changes at micro level
5. Compute and evaluate distributional outcomes at micro level

CGE (macro) model Micro model
The labor market is segmented into formal
skilled, formal unskilled and informal
unskilled components
Individuals supplying labor are assigned
into one of these segments
The labor supplies are exogenous and fixed
in the short run
They remain in original segment
In the informal segment there is full
employment
All individuals informally employed remain
as such
In the formal segments there is some
unemployment
Individuals supplying labor in the formal
segments need to be assigned among
employed and unemployed alternatives in
each simulation
The unskilled unemployed are located into
the formal segment
1. Specification of household income model

1. Specification of household income model
𝑌𝐻𝑕 = (𝑊𝑖
𝑠
𝐼𝑊𝑖
𝑠
+ 𝑌0𝑖)
𝑖 𝑕
(1)
= 𝐼𝑛𝑑 𝛼 𝑠
+ 𝑍𝑖
𝑠
𝛽 𝑠
+ 𝑢𝑖
𝑠
> 𝐶𝑉
𝑈
(2)
𝑌0𝑖 = 𝐷𝐼𝑉𝐷𝑖 + 𝐹𝐼𝑁𝑇𝑖 + 𝑂𝑇𝐻𝑌𝑖 (4)
𝑙𝑜𝑔𝑊𝑖
𝑠
= 𝑎 𝑠
+ 𝑋𝑖
𝑠
𝑏 𝑠
+ 𝑣𝑖
𝑠
(3)
𝐼𝑊𝑖
𝑠
= 𝐼𝑛𝑑(𝐶𝑉𝑖
𝑠
> 𝐶𝑉 𝑈
)
: nominal income of household h
: dummy variable identifying labor status (1 for employed, 0 otherwise) in
labour segment s of individuals i in household h
: nominal wage of individual i in household h working in labour segment s
: non-labour income of individual i in household h
Employment equation:
Wage equation:
Non-labor income equation:
Household income equation:

*: significant at 5% level
D: for a discrete change
M: marginal and impact effects reported by segment for a married male heading a household in Great Buenos Aires who has not completed
education level corresponding to his skill category (primary for unskilled, university for skilled) and has mean experience (25.9 years for
unskilled, 17.7 years for skilled).
𝐼𝑊𝑖
𝑠
= 𝐼𝑛𝑑(𝛼 𝑠
+ 𝑍𝑖
𝑠
𝛽 𝑠
+ 𝑢𝑖
𝑠
> 𝐶𝑉
𝑈
) (2) P(𝐼𝑊𝑖
𝑠
= 1 𝑍𝑖
𝑠
=
𝑒 𝛼 𝑠+𝑍 𝑖
𝑠 𝛽 𝑠
1+𝑒 𝛼 𝑠+𝑍 𝑖
𝑠 𝛽 𝑠
Variable Formal skilled Formal unskilled
Coef dy/dxM
Coef dy/dxM
MaleD
0.0393
(0.0560)
0.0035
(0.0050)
0.2333*
(0.0651)
0.0581*
(0.0162)
MarriedD
0.4145*
(0.0643)
0.0431*
(0.0071)
0.6360*
(0.0586)
0.1573*
(0.0142)
Household HeadD
0.2747*
(0.0691)
0.0270*
(0.0071)
0.5901*
(0.0666)
0.1462*
(0.0161)
Completed Education LevelD
0.9702*
(0.0705)
0.0583*
(0.0054)
0.7799*
(0.0825)
0.1762*
(0.0204)
Experience 0.0900*
(0.0072)
0.0079*
(0.0008)
0.0997*
(0.0083)
0.0246*
(0.0020)
Experience squared -0.0013*
(0.0001)
-0.0001*
(0.00001)
-0.0014*
(0.0001)
-0.0003*
(0.00003)
Household Size -0.0613*
(0.0133)
-0.0054*
(0.0012)
-0.0483*
(0.0116)
-0.0119*
(0.0028)
Region NorthwestD
0.1752*
(0.0830)
0.0144*
(0.0069)
0.1277
(0.0884)
0.0313
(0.0216)
Region NortheastD
0.3896*
(0.1037)
0.0293*
(0.0077)
0.0793
(0.1052)
0.0195
(0.0258)
Region CuyoD
0.3618*
(0.1060)
0.0275*
(0.0079)
0.1742
(0.1057)
0.0425
(0.0257)
Region PampaD
0.0674
(0.0749)
0.0057
(0.0065)
-0.0770
(0.0800)
-0.0190
(0.0198)
Region PatagoniaD
0.6654*
(0.1056)
0.0449*
(0.0072)
0.9434*
(0.1000)
0.2071*
(0.0220)
Constant 0.5730*
(0.0996)
-2.5913*
(0.1637)
N 14,574 6,858
McFadden-R2
0.0952 0.1252
Prob > 𝜒2 0.0000 0.0000
2. Estimation of household income model

*: significant at 5% level
Variable Formal
skilled
Formal
unskilled
Informal
unskilled
Male 0.3538* 0.1800* 0.4347*
(0.0140) (0.0241) (0.0164)
Completed Education Level 0.3692* 0.1027* 0.2563*
(0.0229) (0.0365) (0.0223)
Experience 0.0209* 0.0033 0.0406*
(0.0031) (0.0045) (0.0021)
Experience squared -0.0003* -0.00001 -0.0005*
(0.00005) (0.00006) (0.00003)
Married 0.0594* -0.0386 0.1753*
(0.0166) (0.0251) (0.0175)
Region Northwest -0.5441* -0.2794* -0.3334*
(0.0226) (0.0273) (0.0273)
Region Northeast -0.6392* -0.3000* -0.4162*
(0.0273) (0.0324) (0.0308)
Region Cuyo -0.5720* -0.2731* -0.3440*
(0.0283) (0.0333) (0.0319)
Region Pampa -3.3764* -0.1500* -0.1115*
(0.0214) (0.0253) (0.0261)
Region Patagonia -0.0891* 0.0713 0.2595*
(0.0277) (0.0374) (0.0320)
Inverse Mills Ratio 2.3143* 0.8279*
(0.1990) (0.1296)
Constant 6.2963* 6.2981* 4.4198*
(0.0705) (0.1614) (0.0420)
N 10,627 3,386 8,636
R2
0.3182 0.2240 0.2109
Prob>F 0.0000 0.0000 0.0000
𝑠
= 𝑎 𝑠
+ 𝑋𝑖
𝑠
𝑏 𝑠
+ 𝜆(𝛼 𝑠
+ 𝑍𝑖
𝑠
𝛽 𝑠
)ρs
+ 𝑣𝑖
𝑠
(3)

Impute unobservables and criterion value for base option (unemployment)
𝑢𝑖
𝑠
values are randomly drawn from the inverse of the logistic 𝑝𝑑𝑓 assuring consistency
with the observed employment status.
𝐶𝑉 𝑈
= 𝐸 𝛼 𝑠
+ 𝑍𝑖
𝑠
𝛽 𝑠
The criterion value associated with unemployment is arbitrarily
set; for convenience, at the mean of the index function of the employed alternative
𝑣𝑖
𝑠
imputed from regression residual when existent; otherwise randomly from
N(0,G2
𝑣𝑖)
 Every element in the Household Income Model has been determined
𝐼𝑊𝑖
𝑠
= 𝐼𝑛𝑑(𝛼 𝑠
+ 𝑍𝑖
𝑠
𝛽 𝑠
+ 𝑢𝑖
𝑠
> 𝐶𝑉
𝑈
) (2)
𝑠
= 𝑎 𝑠
+ 𝑋𝑖
𝑠
𝑏 𝑠
+ 𝜆(𝛼 𝑠
+ 𝑍𝑖
𝑠
𝛽 𝑠
)ρs
+ 𝑣𝑖
𝑠
(3)

3. CGE results of a capital outflow
Non-residents deposits at domestic banks fall 60.8% in Argentina, from 32.9 billion
dollars (December 2000) to 12.9 billion dollars (December 2002).
Nummeraire: CPI.
This leads to a contraction of the economy, with the following changes
communicated to the microsimulation module:
%58.1EXR
PA: price of primary good; PI: price of industrial good; YS: income of skilled RHG; YU:
income of unskilled RHG; YC: income of capitalist RHG

4. Communications from the CGE to the
microsimulation model
13
𝑁𝐹𝑆, 𝑁𝐹𝑈
𝑊𝐹𝑆, 𝑊𝐹𝑈, 𝑊𝐼𝑈, 𝑃𝐴, 𝑃𝐼, 𝐸𝑋𝑅
𝐷𝐼𝑉𝐷, 𝐹𝐼𝑁𝑇
Sim.1
Sim.2
Sim.3
Sim.H: Non-behavioural microsimulations linked to CGE through the households
SYH
𝑌𝐻𝑆, 𝑌𝐻 𝑈, 𝑌𝐻𝐶
Cumulative effect for behavioural microsimulations
Sim.F: Non-behavioural microsimulations linked to CGE through the factor markets
In the line of Vos and Sanchez (2010).

4. Attributing results at micro level
Keeping unchanged the observed and unobserved characteristics of the individuals, the
parameters in the household income model need to change to allow wages and
employment status to adjust consistently with the CGE macro results.
Following the methodology designed by RBR, the changes in the coefficients are done
assuming “neutrality” with respect to individual characteristics, changing the intercepts:
=> fall in probability of being employed for everyone in the labor segment,
where the fall depends only on initial probability and not on individual characteristics
=> proportional fall of all wages in the labour segment
↓𝛼 𝐹𝑆
, ↓ 𝛼 𝐹𝑈
↓𝑎 𝐹𝑆, ↓ 𝑎 𝐹𝑈, ↓ 𝑎𝐼𝑈
𝑙𝑜𝑔𝑊𝑖 = 𝑎 + 𝑋𝑖 𝑏 + 𝑣𝑖 (3)
P(𝐼𝑊𝑖
𝑠
= 1 𝑍𝑖
𝑠
=
𝑒 𝛼 𝑠+𝑍 𝑖
𝑠 𝛽 𝑠
1+𝑒 𝛼 𝑠+𝑍 𝑖
𝑠 𝛽 𝑠 (2)

4. Newton’s technique to change
intercepts
𝑥 𝑛+1 = 𝑥 𝑛 −
𝑓(𝑥 𝑛 )
𝑓′(𝑥 𝑛 )
Distance to
macro target
Intercept

4. Implementing Newton’s technique
𝑁𝑓
∗
= 𝑁𝑓,0. (1 + 𝑁𝑓)
𝑊𝑓
∗
= 𝑊𝑓,0. (1 + 𝑊𝑓),
𝑥 = 𝛼 𝐹𝑆 𝑎 𝐹𝑆 𝛼 𝐹𝑈 𝑎 𝐹𝑈 𝑎𝐼𝑈 intercepts
𝑓 𝑥 = 𝑁 𝐹𝑆,0, 𝑁 𝐹𝑈,0 𝑊𝐹𝑆,0, 𝑊𝐹𝑈,0 𝑊𝐼𝑈,0
𝑓∗
𝑥 = 𝑁𝐹𝑆
∗
, 𝑁𝐹𝑈
∗
𝑊𝐹𝑆
∗
, 𝑊𝐹𝑈
∗
𝑊𝐼𝑈
∗
macro targets
5x5
.

4. Regression Intercept changes
Intercept Regression Simulation 1
(N falls)
Simulations 2 & 3
(N and W falls)
0.5730 0.5403 0.5403
6.2963 6.2944 6.2931
-2.5913 -2.6094 -2.6094
6.2981 6.3095 6.3052
4.4198 4.4198 4.4102
FS
FU
FUa
IUa
FSa

5. Percentage change in household per capita income by percentile
Simulations 1 and 2
shape dominated by people getting fired
labor income large share of income at the bottom

Simulations 2 and 3

Simulations 3 and RHG

5. Percentage changes in employment by percentile
Simulations Behavioural (NSIM3) and ‘Non-parametric approach’
(via factor market)

Indicator BASE SIM1 SIM2 SIM3 SIMRHG
Per capita income 328.7 326.0 325.4 327 325.5
Inequality
Entropy Index (α=2) 69.9 70.4 70.6 71.9 69.6
Gini Index 51.1 51.3 51.3 51.5 51.1
Poverty
Official Extreme Poverty Line
Head-Count Index (P0) 11.8 12.1 12.2 12.2 11.9
Poverty Gap Index (P1) 7.1 7.4 7.5 7.5 7.2
Poverty Severity Index (P2) 5.9 6.2 6.2 6.2 5.9
Official Moderated Poverty Line
US$ 1 a day Poverty Line
US$ 2 a day Poverty Line
5. Per capita income, inequality and poverty by simulation
Official poverty rates are in line with those reported by World Bank-UNLP SEDLAC (Socioeconomic
Data for Latin America and Caribbean): for 2001, 9.4% and 28.0%. CEDLAS estimation of 2.5 DLS a d
Line for 2001 is also in line: 18.7%(P0), 9.1(P1), 6.1(P2)

Conclusions
•In the behavioural microsimulations, as per capita income falls and inequality
increases, the poverty headcounts, the poverty gaps and the poverty severity indexes
go up for the different poverty lines. The increase is mainly due to the employment
fall, though there are slight increases due to the wage fall, and no change at all due
to the capital income changes.
•As in RBR, it is found that “the selectivity of labour market rationing is the channel
through which economy-wide policies have the most distributional impact”.
•Graph 3 (Behavioural vs. Arithmetic MS) gives a clear indication of the power of
behavioural microsimulations to capture the heterogeneity of income changes in
different parts of the income distribution due to a macro shock, as opposed to
arithmetic microsimulations.

Illustration:
The impact of Oportunidades on
human capital and income
distribution: a top-down/bottom-
up approach
(Dario Debowicz and Jennifer Golan, 2013)

Rationale, method and results
• Effects of Oportunidades conditional cash transfer program on
human capital and labour markets, accounting for its partial
and general equilibrium effects.
• Linking a microeconometric and a general equilibrium model
in an iterative bidirectional way.
• Our results suggest that partial equilibrium analysis alone may
underestimate the program effects.

Oportunidades
Around 14 billion of Mexican pesos (Mexican Ministry of Finance 2011), or 1.1 billion
US$ per year spent on Oportunidades, reaching 5.8 million households.
The program aims at developing the human capital of poor households. It provides
cash to poor households under the condition that they behave consistently with the
accumulation of human capital.
The program has three components: education, nutrition and health. The largest
transfer of the program is the educational one.
Beneficiaries are targeted at the micro level. However, given its national scale, it is
expected to provoke interesting macro level effects with meaningful interactions with
the direct micro-level effects of the programs that will jointly affect income
distribution and poverty.

Oportunidades educational transfers
Monthly Scholarship, July to December 2008
Primary education Boys and Girls
Third grade $130
Fourth grade $155
Fifth grade $195
Sixth grade $265
Secondary eduction Boys Girls
First grade $385 $405
Second grade $405 $450
Third grade $430 $495
Upper secondary/ High
school
Men Women
First grade $645 $740
Second grade $695 $790
Third grade $735 $840
Source:
http://www.normateca.gob.mx/Archivos/46_D_1786
_.pdf

Observed time allocation of children
Distribution of children by occupational
choice
Choice Frequency Percent
Not Studying 3,543 11.68
Work and School 1,702 5.61
School only 25,086 82.71
Total 30,331 100
Authors' calculation based on ENIGH 2008
0
10
20
30
40
50
60
70
80
90
100
6 7 8 9 10 11 12 13 14 15 16 17%ofchildrenbyalternative
Age
No School Work&School School only

Our macro-micro approach
Computable General Equilibrium Model
Factor and output markets with countrywide supplies and demands
Behavioural Microsimulation Model capturing
transfer conditional on school attendance
Oportunidades
CCTs
Labor supply
(% ch)
Occupational
choice
of children
Child labor
wage
(% ch) Other household
Income
(% ch)
Aggregation
Representative
household
groups
Other factors
income
(% ch)
Household Income
Poverty and inequality indicators at household level
IFPRI Standard
CGE model
Bi-directional
iterative
Link á la Luc
Savard
Bourguignon,
Ferreira and
Leite (2003)

Behavioral micro-simulation model
(0) Works
(1) Works and attends school
(2) Attends school
Mincer equation for wage
MNL model for time allocation of children based on ARUM (Amemiya and Shimono 1989)

Simulations
1) Absence of Oportunidades transfers (conditional and
unconditional)
2) Extension of the conditional transfers to all the moderately poor
children according to the existent program rules. Existing transfers
remain, but coverage extends with CCT=f(school grade child were
to assist, gender) for moderately poor who are not yet
beneficiaries.
3) Lagged human capital effect of the existing transfers

Simulated changes in time allocation
No Program: Partial Equilibrium General Equilibrium
Base Choice Base Choice
Simulated Choice
Not Studying
Work and
School
School onlyNot Studying
Work and
School
School only
Not Studying 100.0 0.3 0.1 100.00 0.0 0.0
Work and School 0.0 99.7 0.0 0.0 99.94 0.0
School only 0.0 0.0 99.9 0.0 0.06 100.00
Program
Expansion Partial Equilibrium General Equilibrium
Simulated Choice
Not Studying
Work and
School
Work and
School
School only
Not Studying 87.4 0.00 0.00 87.9 0.52 0.05
Work and School 0.8 100.00 0.00 0.5 96.38 0.0
School only 11.8 0.00 100.00 11.6 3.10 99.95
Program Skilling Partial Equilibrium General Equilibrium
Simulated Choice
Not Studying
Work and
School
Work and
School
School only
Not Studying -- -- -- 99.8 0.0 0.0
Work and School -- -- -- 0.0 99.9 100.0
School only -- -- -- 0.2 0.1 0.0

Simulated changes on income distribution:
No-program and program-expansion simulations,
partial and general equilibrium
%changeinhouseholdpercapitaincome
centiles
NOPROG_GE NOPROG_PE
PROGEXP_GE PROGEXP_PE
0 50 100
-20
0
20
40

SKILLING
%changeinhouseholdpercapitaincome
centiles
0 50 100
0
10
20
30
Simulated changes on income distribution:
Lagged human capital effect
general equilibrium

Simulated changes on income
distribution and poverty indicators
Simulated poverty and inequality indicators by simulation and type of equilibrium
Base No Program Program Expansion
Program
Skilling
FGT(0) PE GE PE GE GE
National extreme poverty line 6.7 8.3 8.9 5.7 5.3 5.8
National moderate poverty line 29.2 30.3 31.1 27.4 26.5 27.8
1.25$ a day line 1.7 2.8 3.1 1.2 1.1 1.2
2$ a day line 5.2 6.6 7.0 4.4 4.0 4.4
FGT(1)
1.25$ a day line 0.4 0.9 1.0 0.3 0.3 0.3
2$ a day line 1.5 2.3 2.5 1.2 1.1 1.2
FGT(2)
1.25$ a day line 0.2 0.4 0.5 0.1 0.1 0.1
2$ a day line 0.6 1.1 1.3 0.5 0.4 0.5
Gini Coefficient 0.511 0.517 0.521 0.506 0.501 0.504

Conclusions
 We combine a micro-simulation model with a general equilibrium model to search for
an equilibrium that satisfies both the utility-maximizing decisions regarding the time-
allocation of the children in school age and the equilibrium of the country-wide factor
and commodity markets.
 Applying our model to the Oportunidades conditional cash transfers in Mexico, we find
that partial equilibrium analysis alone may underestimate the distributional effects of
the program.
 By raising the opportunity cost of work through the provision of transfers conditional
on school attendance, and hence reducing child labor supply, Oportunidades increases
the wages earned by children at work in a double-digit percentage change magnitude.
 This indirectly benefits poor households who retain their children at work: in partial
equilibrium analysis the program causes a 1.8 percentage-point drop in
poverty, accounting also for the general equilibrium effects is estimated to lead to a
drop in poverty of up to 2.7 percentage points.

Conclusions (cont.)
 The skilling of the future workforce generated by the lagged human capital acquisition
allowed by Oportunidades further increases the incomes of the poor
households, decreasing the poverty rate roughly in the order of 1.4 percentage points.
 Careful re-calibration of the parameters of the model we presented would allow
researchers to apply it to other national-level conditional cash transfer program in a
developing country, and identify its partial and general equilibrium effects.
This combined model has two serious limitations:
 The model is static and, as such, does not allow us to see the time path of the
distributional changes provoked by the program.
 The micro-simulation model does not model existing intra-household decision-making
mechanisms that are important in the determination of the time-allocation choices of
children and indirectly in identifying the distributional effects of the program.

References
Cockburn, J. (2006). Trade liberalisation and poverty in Nepal: a computable general equilibrium micro
simulation analysis. In M. Bussolo & J. Round (Eds.), Globalization and poverty: channels and policies
(pp. 171-194). London: Routledge.
Vos, R., & Sanchez, M. (2010). A Non-Parametric Microsimulation Approach to Assess Changes in
Inequality and Poverty. International Journal of Microsimulations, 3(1), 8-23.
Bourguignon, F., Robilliard, A. S., & Robinson, S. (2004). Representative versus real households in the
macroeconomic modelling of inequality. In T.J.Kehoe, T.N.Srinivasan & J. Whalley (Eds.), Frontiers in
Applied General Equilibrium Modelling. Cambridge: Cambridge University Press.
Savard, L. (2010). Scaling Up Infrastructure Spending in the Philippines: A CGE Top-Down Bottom-Up
Microsimulation Approach. International Journal of Microsimulations, 3(1), 43-59.
Debowicz, D. (2007). The Effects of Globalisation on Poverty and Distribution in Argentina in the 1990s.
Thesis draft. University of Sussex (attached, see chapter 7, p.141/171).
Debowicz, D and J Golan (2013), The Impact of Oportunidades on Human Capital and Income
Distribution
A Top-Down/Bottom-Up Approach, IFPRI Discussion Paper 1257. (attached)

General equilibrium model
Selected equations

Macro-micro approaches
Microsimulations models
combined with macro
models
Integrated Layered
Non-behavioral Behavioral
Top-down
Top-down/ Bottom-up

Econometric explanation of child wage
Table 1 Child wage equation
(1) (2)
Age 6 to
17
Age 12 to
17
Work & School -0.644*** -0.593***
(0.060) (0.058)
Log average
federative wage 0.346*** 0.438***
(0.064) (0.072)
Male 0.214*** 0.213***
(0.048) (0.053)
Years of schooling 0.009 0.153***
(0.044) (0.047)
Years of
schooling^2 -0.001 -0.012***
(0.003) (0.003)
Age 0.389*** 0.366
(0.090) (0.334)
Age^2 0.001 -0.001
(0.003) (0.011)
Female head 0.135*** 0.127**
(0.052) (0.056)
Rural -0.243*** -0.326***
(0.051) (0.054)
Noreste 0.219 0.096
(0.152) (0.166)
Noroeste 0.306*** 0.337***
(0.100) (0.108)
Occidente 0.151 0.167
(0.103) (0.111)
Oriente -0.062 0.053
(0.112) (0.115)
Centronorte 0.122 0.110
(0.099) (0.104)
Centrosur 0.178* 0.206**
(0.101) (0.105)
Sureste -0.023 0.141
(0.100) (0.107)
Constant -5.882*** -5.086**
(0.528) (2.484)
Observations 3,021 2,285
R-squared 0.596 0.262

Econometric explanation of children’s
time allocation
Table 1 Multinomial logit estimates of child labor supply, all children
(1) (2) (3) (4) (5)
Multinomial logit estimates Marginal effects at the mean of data
Work & School School only Not studying Work & School School only
Y_i 0.019*** 0.022*** -0.001*** -0.000 0.001***
(0.004) (0.003) (0.000) (0.000) (0.000)
Log mean fed. wage 0.411*** -0.157*** 0.008** 0.025*** -0.033***
(0.089) (0.054) (0.003) (0.003) (0.005)
Male 0.493*** -0.206*** 0.011*** 0.031*** -0.042***
(0.063) (0.040) (0.003) (0.002) (0.003)
Years of schooling 0.000 -0.093*** 0.006*** 0.004*** -0.010***
(0.037) (0.025) (0.002) (0.001) (0.002)
Years of schooling 0.008** -0.002 0.000 0.000*** -0.000***
(0.003) (0.002) (0.000) (0.000) (0.000)
Rank of child 0.355*** 0.396*** -0.025*** -0.001 0.026***
(0.036) (0.022) (0.001) (0.001) (0.002)
No child 0<age<6 -0.058 -0.122*** 0.008*** 0.002 -0.010***
(0.042) (0.026) (0.002) (0.002) (0.002)
No child 5<age<13 0.297*** 0.289*** -0.018*** 0.001 0.017***
(0.034) (0.022) (0.001) (0.001) (0.002)
No child 12<age<18 -0.613*** -0.705*** 0.044*** 0.002 -0.046***
(0.043) (0.025) (0.002) (0.002) (0.002)
No people age>17 -0.278*** -0.240*** 0.015*** -0.002* -0.013***
(0.034) (0.019) (0.001) (0.001) (0.002)
Rural -0.412*** -0.211*** 0.015*** -0.009*** -0.005
(0.073) (0.045) (0.003) (0.003) (0.004)
Educ. Head 0.101*** 0.169*** -0.011*** -0.002*** 0.013***
(0.009) (0.006) (0.000) (0.000) (0.001)
Age Head 0.006* 0.019*** -0.001*** -0.001*** 0.002***
(0.003) (0.002) (0.000) (0.000) (0.000)
Noreste -0.160 -0.289** 0.020** 0.005 -0.025**
(0.205) (0.115) (0.009) (0.009) (0.012)
Noroeste 0.618*** 0.076 -0.007 0.029*** -0.023***
(0.135) (0.082) (0.005) (0.007) (0.009)
Occidente 0.371*** -0.153* 0.008 0.027*** -0.035***
(0.144) (0.087) (0.006) (0.008) (0.010)
Oriente 0.253 0.171* -0.010** 0.004 0.006
(0.154) (0.090) (0.005) (0.006) (0.008)
Centronorte 0.112 -0.068 0.004 0.008 -0.012
(0.138) (0.080) (0.005) (0.006) (0.008)
Centrosur 0.042 0.153* -0.009* -0.004 0.013*
(0.140) (0.080) (0.005) (0.005) (0.007)
Sureste 0.631*** 0.108 -0.008* 0.029*** -0.020**
(0.136) (0.081) (0.005) (0.008) (0.009)

Conceptual and Numerical SAM
capturing Oportunidades in Mexico
Activities Commodities Factors Households Government
Saving-
Investment
Changes in
stocks
Rest of world Oportunidades
Other public
transfers
Activities
Domestic
supply
Commodities
Private final
consumption
Public final
consumption
Fixed
investment
Change in
stocks
Exports
Factors
Value added at
factor cost
Households
Households factor
income
Foreign
remittances
Oportunidades
Other public
transfers
Government Activity taxes Tariffs Direct taxes
Saving-Investment
Private
saving
Public saving
Foreign
saving
Changes in stocks
Change in
stocks
Rest of world Imports
Net factor income
of non-residents
Public transfers to
non-residents
Oportunidades Oportunidades
Other public
transfers
Other public
transfers
Activities Commodities Factors Households Government Saving-Investment Changes in stocks Rest of world Oportunidades Other transfers Total
Activities 12,165 12,165
Commodities 7,856 1,307 2,696 612 3,417 15,889
Factors 10,964 10,964
Households 10,867 281 14 1,002 12,164
Government 1,201 35 813 2,049
Saving-Investment 3,494 -368 182 3,309
Changes in stocks 612 612
Rest of world 3,689 97 94 3,880
Oportunidades 14 14
Other transfers 1,002 1,002
Total 12,165 15,889 10,964 12,164 2,049 3,309 612 3,880 14 1,002 64,097

Actors in Mexican CGE model
Activity Sectors (14)
Agriculture, Livestock, Forestry, Fishing and Haunting; Mining; Electricity, water and
gas provision by tube to final consumer; Construction; Manufacturing; Trade; Transport,
mail and stocking; Information in massive media; Financial and insurance services;
Professional, scientific and technical services; Education; Health and social assistance;
Public services; Other services.
Production Factors (15)
Labor (13): Male skilled informal; Male unskilled informal; Male semi-skilled informal;
Male skilled formal; Male unskilled formal; Male semi-skilled formal; Female skilled
informal; Female unskilled informal; Female semi-skilled informal; Female skilled
formal; Female unskilled formal; Female semi-skilled formal; Child labour
Others (2): Capital, Land.
Representative Household groups (16)
Non-oport non-poor urban male; Non-oport non-poor urban female; Oport non-poor
urban male; Oport non-poor urban female; Oport poor urban male; Oport poor urban
female; Oport poor rural male; Oport poor rural female; Oport non-poor rural male;
Oport non-poor rural female; Non-oport poor urban male; Non-oport poor urban female;
Non-oport poor rural male; Non-oport poor rural female; Non-oport non-poor rural male;
Non-oport non-poor rural female.
Other Accounts (10)
Government; Income tax; Imports tax; Activity tax; Oportunidades; Rest of transfers
from government to households; Change in stock; Saving-Investment; Rest of the World.

Evaluation of CCTs following BFL
A Summary of the Microsimulation Literature on Conditional Cash Transfers following Bourguignon, Ferreira and Leite
Study Country Sample Year (data) Program Name Scenarios Impact
Bourguignon,
Ferreira and Leite
(2002)
Brazil Children aged 10-
15 and children
aged 10-15 living
in poor households
1999 Bolsa Escola 1. Bolsa Escola transfer, 2. doubling Bolsa
Escolar transfers, 3. age-contingent transfer,
4. means-test raised, 5. combination of 2 and
4, 6. combination of 3 and 4, 7. no
conditionality.
Simulation 1: 2.1 percentage point reduction in
children out of school, 1.1 percentage point
increase in fraction of children working and
going to school, 1.8 percentage point increase
in children attending school. Effect more
pronounced for the poor: initially 9.1 percent
working decreases in simulation to 4.7 percent,
23.7 percent working and studying increases to
24.7 percent and 67.3 percent increases to 70.6
percent attending school only. Doubling the
transfer reduces fraction of children out of
school by an additional percentage point, age-
contingency of transfer does not alter to a great
extent the results, amount seems more relevant
than means-test. In terms of poverty, program
in Simulation1 would reduce poverty by 1
percentage point and inequality by 0.5
percentage points, Simulation2 would reduce
headcount by 1.3 percentage points.
Azevedo and
Robles (2010)
Mexico Children aged 12-
18, 12-15, 16-18
1996 Oportunidades in 2005 1. Oportunidades design in 2005, 2. Increase
UCT by 26 percent, 3. suspension of transfer
to students in third to fifth grade of primary
school and proportional increase in transfer
to the rest maintaining UCT component, 4.
suspension of transfer to grade three and five
of primary school plus triplicating existing
transfers, maintaining UCT, 5. triplication of
the CCT plus UCT as in 2, 6. quadruples
transfer to secondary and high school
students, maintain UCT, 7. transfer design
based on opportunity costs as measured by
average reported wage in 2005, 8. Reduction
of 2005 transfer design by one half.
Simulation 1: 2005 CCT design increases
school attendance by 1.18 percentage points
for 12-15 year olds, by 1.17 for 16-18 year
olds to study only and 0.34 percent work and
study and 1.17 for 12-18 year olds to study
only and 0.14 percent of work and study. The
effect is more pronounced for the poor.
Modifying the 2005 design for poor children
according to the different scenarios yields in
Simulation 2. does not alter occupational
choices, Simulation 3 increases the fraction of
students attending school only from 44.04 to
44.66, the fraction of students working and
going to school from 10 to 10.8. Simulation 4
and 5 increase the fraction of children going to
school only to 48.2 and 50.4 and the fraction
of children going to school and working 10.91
and 11.25. Simulation 6 reduces the fraction of
children not attending school by 4.71
percentage points, Simulation 7 by 0.56
percentage points and Simulation 8 increases
the school dropouts by 2.58 percentage points.
The program reduces poverty by 2 percentage
points nationally and by 4 percentage points in
rural areas. Also, inequality is reduced with the
Gini dropping from 0.527 to 0.512.
Amarante, Arim, de
Melo and Vigorito
(2008)
Ururguay Boys and girls aged
14 -17
2006 Asignaciones familiars 1. New asignaciones familiares regime, 2.
Change in the transfer amount for children
aged 0-12 and 13-17, 3. Equal transfer
amount to all children aged below 18, 4.
Simulation of transfer design on adult labor
supply (assuming first children's labor supply
is decided upon and then adults)
Simulation 1: increase in school attendance by
about 2.5 percentage points. 14 % of children
that are out of school in baseline return to
educational system and 20% who initially
work and study, study only in simulation. This
effect does not vary much by the different
simulation scenarios. The authors find a
reduction in poverty by 1 percentage point and
2 percentage points for households with
children. Also, the authors find the program to
reduce inequality. The transfer is only reduces
spouses labor supply and is simulated to
reduce hours of work for eligible households.

Statistics by Occupational choice
Table 1 Summary statistics by Occupational Choice
Not Studying
Work and
School
School
only
Wage (yearly) 7275 6189 0
Total hh income (yearly) 109709 133661 140243
Rural 0.40 0.26 0.28
Age 15.11 12.49 10.98
Male 0.53 0.65 0.50
Years of schooling 6.74 6.01 4.68
No child 0<age<6 0.57 0.50 0.52
No child 5<age<13 1.06 1.33 1.48
No child 12<age<18 1.68 1.14 0.94
No people age>17 2.94 2.56 2.63
Education head 4.77 6.92 7.59
Education spouse of head 3.50 5.37 5.96
Age head 45.93 43.34 43.61
Age spouse of head 32.51 30.38 32.10
Female headed household 0.24 0.26 0.21
Rank 2.04 2.02 2.12
Hrs. worked 19.54 16.13 0.00
N 3,543 1,702 25,086
Authors' calculation based on ENIGH 2008

Elasticities in CGE model
Factor substitution Armington CET
Agriculture, livestock, forestry, fishing, hunting 0.6 4 4
Mining 0.6 3 3
Electricity, water and gas provision 0.6 3 3
Construction 0.6 3 3
Manufacturing 0.6 3 3
Trade 0.6 3 3
Transport, mail and storage 0.6 3 3
Media 0.6 2 2
Financial and insurance services 0.6 2 2
Profesional, scientific and technical services 0.6 2 2
Education 0.6 2 2
Health and social assistance 0.6 2 2
Other private services 0.6 2 2
Public services 0.6 2 2
Source: IFPRI model for Mexico – Rebecca Lee Harris and David Coady (2000)

Frisch and expenditures elasticities in
LES consumption demand
Non-oport
non-poor
urban
male hhd
Non-oport
non-poor
urban
female
hhd
Oport non-
poor
urban
male hhd
Oport non-
poor
urban
female
hhd
Oport poor
urban
male hhd
Oport
poor
urban
female
hhd
Oport
poor rural
male hhd
Oport
poor rural
female
hhd
Oport non-
poor rural
male hhd
Oport non-
poor rural
female
hhd
Non-oport
poor
urban
male hhd
Non-oport
poor
urban
female
hhd
Non-oport
poor rural
male hhd
Non-oport
poor rural
female
hhd
Non-oport
non-poor
rural male
hhd
Non-oport
non-poor
rural
female
hhd
Frisch parameter -1.0 -1.0 -1.0 -1.0 -1.0 -1.0 -1.0 -1.0 -1.0 -1.0 -1.0 -1.0 -1.0 -1.0 -1.0 -1.0
Agriculture, livestock, forestry, fishing, hunting 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8
Mining 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Electricity, water and gas provision 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Construction 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Manufacturing 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Trade 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Transport, mail and storage 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Media 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Financial and insurance services 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2
Profesional, scientific and technical services 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2
Education 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2
Health and social assistance 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2
Other private services 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2
Public services 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Source: IFPRI model for Mexico – Rebecca Lee Harris and David Coady (2000)

CGE Modeling and Microsimulations by Dr. Dario Debowicz

More Related Content

What's hot (20)

Similar to CGE Modeling and Microsimulations by Dr. Dario Debowicz (20)

More from International Food Policy Research Institute (20)

Recently uploaded (20)

CGE Modeling and Microsimulations by Dr. Dario Debowicz

Editor's Notes