Stata in the measurement and
analysis of poverty in Mexico
2009 Mexican Stata Users Group Meeting
April 2009, Mexico city
1
National Council of Evaluation of Social Development
Policy (CONEVAL)
Creation of CONEVAL
General Law of Social Development (January 2004)
Object of the Law:
“To guarantee the total exercise of the social rights
established in the Political Constitution of Mexico ”
Article 81: Establishes the creation of the Council
Income Poverty Measure in Mexico
(recent history)
In 2001 the Ministry of Social Development created
the National Committee for Poverty Measure (CTMP).


7 academics and
4 government members: CONAPO, INEGI, Ministry of
Social Development, and Presidencia)
In 2002 The Committee proposed a methodology:
http://www.sedesol.gob.mx/archivos/801588/file/Docu01.pdf
National Council of Evaluation of Social
Development Policy
The Council is a public decentralized organism of the
federal public administration with technical
autonomy
The direction of the Council is given by:


Six academic researchers and
Executive secretary
Responsibilities:
1) Establish the criteria to define, identify, and
measure poverty, and
2) Rule and coordinate the evaluation of the
national policy of social development
Right now, CONEVAL is working on a new
methodology for multidimensional poverty measure
2
Stata and CONEVAL
Stata and the measurement of poverty
•
Why do we use Stata?
To use survey and census data and generate inputs, indicators, and other relevant
information to measure, characterize, and analyze the phenomenon of poverty;
and help in the decision making process to alleviate it.
•
Content of presentation:
1) Inputs in poverty measurement
2) Construct poverty indicators
3) Poverty analysis
4) Poverty mapping
3
Income poverty, 1992 -2006
National, urban and rural
Income poverty incidence, 1992 - 2006
30.0
15.0
0.0
percentage
45.0
60.0
(food poverty line)
1992
1994
1996
1998
2000
2002
2004
2006
years
National
Rural
Urban
Source: estimates based on ENIGH 1992-2006
4
1) Inputs in poverty measurement
Construction of food poverty line (example)
Adjustment coefficient:
Adjustment coefficient by moving quintiles
National level
1.1
1.2
1.3
AC = consumed calories/required calories
per household
Reference households stratum:
.8
.9
1
Used to construct an observed food
basket and determine the (food) poverty line
1
11
21
31
41
51
Moving quintiles
Reference households
Adj. coeff. = 1
61
71
81
Adj. coeff.
2006 Official (food) poverty line:
Urban: $809.87 (mxn pesos)
Rural: $598.70 (mxn pesos)
Source: estimates based on ENIGH 2006
5
1) Inputs in poverty measurement
Non-food poverty lines: Inverse of Engel coefficient
Engel coefficient:
Ratio that measures the expenses on food in households
as a proportion of the expenses needed to cover:
- health and education: Capabilities line, and
- public transport, clothing, and housing: Assets line
The ratio is calculated for rural and urban areas in a
reference stratum
6
1) Inputs in poverty measurement
Standard errors and hypothesis testing
Standard errors:
Income poverty incidence, 1992 - 2006
# delimit ;
foreach x in 1992 1994 1996 1998 2000
2002 2004 2005 2006 { ;
use “$data\poverty `x’.dta”, clear ;
svyset upm [w=factorp], strata(est) vce(linearized) ;
svy linear, level(95): mean povlp1 ;
} ;
20.0
30.0
40.0
(food poverty line)
10.0
Hypothesis testing:
1992
1994
1996
1998
2000
2002
2004
2006
years
Poverty Incidence
95% Conf. Int.
95% Conf. Int.
Source: estimates based on ENIGH 1992-2006
7
2) Poverty indicators
Poverty gap and squared poverty gap
(food poverty line)
0.0
0.0
5.0
5.0
FGT(2) x 100
10.0
15.0
Squared poverty gap, 1992 - 2006
(food poverty line)
10.0 15.0 20.0 25.0
Poverty gap, 1992 - 2006
1992
1994
1996
1998
2000
2002
2004
2006
years
National
Rural
1992
1994
1996
1998
2000
2002
2004
2006
years
Urban
Source: estimates based on ENIGH 1992-2006
FGT(α) :
Foster, J., J. Greer, and E. Thorbecke (1984), “A Class of
Decomposable Poverty Measures”, Econometrica, vol. 52, pp.
761-765.
National
Rural
Urban
Source: estimates based on ENIGH 1992-2006
# delimit ;
gen fgt0 = cond(income<pov_line,1,0) ;
gen fgt1 = cond(fgt0==1,(pov_line - income)/pov_line,0) ;
gen fgt2 = cond(fgt0==1,((pov_line - income)/pov_line)^2,0) ;
tabstat fgt* [w=factorp], stats(mean) by(area) format(%6.4f) ;
8
2) Poverty indicators
Child poverty indicators
9
3) Poverty analysis
Poverty profile
10
3) Poverty analysis
Components of changes in poverty measures
11
3) Poverty analysis
Microsimulation of an intervention (example)
Income distribution, microsimulation
Rural areas, 2006
Urban areas, 2006
.0002
0
.0001
Density
.0005
.0003
Income distribution, microsimulation
0
200
598.7
Income
Observed
800
1200
Simulated
Source: estimates based on ENIGH 2006
0
200
600
Income
Observed
809.87
1200
Simulated
Source: estimates based on ENIGH 2006
Microsimulation :
Using the income and expenditure survey of 2006, the
microsimulation consists in increasing by $180 pesos
the households’ income of a public programme net
12
4) Poverty mapping
Stata and the income poverty maps
•
Poverty mapping
National level indicators often hide important differences between regions or
areas. The analysis of poverty interventions consequently requires a focus on
poverty information that is more geographically disaggregated.
•
Stata and poverty mapping
1) Social gap index
2) Estimate income poverty and a set of indicators from survey data
3) Generate the same set of indicators from census data (very hard work!)
4) Validate poverty measures with other indices
5) Compute changes in poverty
13
4) Poverty mapping
Social gap index 2005
Methodology
Principal component analysis (PCA) using
Census data 2005
Variables defined in the General Law of Social
Development
Index stratification:





Very low
Low
Medium
High
Very high
Disaggregation levels:
 Entities


Municipalities
Localities
Components
1. Population over 15 years illiterate
2. Population between 6 and 14 that doesn’t attend
to school.
3. Population over 15 years with incomplete basic
education
4. Households with people between 15 and 29 years
with at least one member with less than 9
years of education
5. Population without health security
6. Dwellings without washing machines
7. Dwellings without refrigerator
8. Dwellings with sand floor
9. Dwellings without toilets
10. Dwellings without tubed water of the public
network
11. Dwellings without sewage
12. Dwelling without electric energy
13. Overcrowding
14
Social gap index
Localities, 2005
Social Gap Degree
Very low
Low
Medium
High
Very high
15
Poverty mapping
Income poverty and other indicators
Income poverty and Social gap index (SGI)
Municipalities, 2005
Municipalities, 2005
.75
.5
.25
Y = 2.13 – 2.39 X
adj. R2 = .7177
.4
Y = 0.33 + 0.17 X
adj. R2 = .8032
0
0
.25
.5
.75
Income poverty FGT(0)
1
1
Income poverty and Human Development Index (HDI)
.6
.8
1
-2.5
0
HDI
Municipalities
2.5
5
SGI
Fitted values
Source: estimates based on CONEVAL and UNDP
Municipalities
Fitted values
Source: estimates based on CONEVAL
16
Food poverty map
Municipalities, 2000
Ranks
Total
municipalities
[1.6 - 20.6)
522
[20.6 - 39.7)
582
[39.7 - 58.7)
577
[58.7 - 77.8)
500
[77.8 - 96.8]
272
17
Food poverty map
Municipalities, 2005
Ranks
Total
municipalities
[0.11 - 16.9)
562
[16.9 - 33.7)
848
[33.7 - 50.5)
552
[50.5 - 67.2)
355
[67.2 - 84.0]
137
18
Changes in income poverty
Municipalities, 2000 - 2005
Changes in poverty
75
50
0
25
Income poverty 2005
100
Municipalities, 2000 - 2005
0
25
50
Income poverty 2000
Municipalities
75
100
45° line
Source: estimates based on CONEVAL
19
Changes in food poverty map
Municipalities, 2000 - 2005
Rank
Significant increase
Total
municipalities
46
Non significant increase
1474
Significant decrease
933
20
Income poverty and Social gap index
Five municipalities with highest poverty rates and very
high social gap level
San Pablo Cuatro Venados
Population: 1,267 Hab.
Food poverty: 81.1%
Social gap degree: Very high
Chalchihuitán
Population: 13,295 Hab.
Food poverty: 81.4%
Social gap degree: Very high
San Juan Cancuc
Population: 24,906 Hab.
Food poverty: 83.7%
Social gap degree: Very high
Santiago el Pinar
Population: 2,854 Hab.
Food poverty: 84.0%
Social gap degree: Very high
Chanal
Population: 9,050 Hab.
Food poverty: 83.1%
Social gap degree: Very high
21
Food poverty map (number of population in poverty)
Municipalities, 2005
Ranks
Total
municipalities
[0.00 - 2,500)
1037
[2,500 - 15,000)
1083
[15,000 - 100,000)
327
[100,000 - 172,271)
7
22
CONEVAL online
• Please visit us at:
www.coneval.gob.mx
• Do files available at:
http://www.coneval.gob.mx/coneval2/htmls/medicion_pobreza/HomeMedicionP
obreza.jsp?categorias=MED_POBREZA,MED_POBREZA-med_pob_ingre
• Surveys available at:
http://www.inegi.org.mx/est/contenidos/espanol/soc/sis/microdatos/enigh/defa
ult.aspx?s=est&c=14606
Authors:
Héctor H. Sandoval (hhsandoval@coneval.gob.mx)
Rodrigo Aranda Balcazar (ranohead@gmail.com)
Martín Lima (jlimav@gmail.com)
23