Estimating the Human
Opportunity Index
HOI in STATA
• Stata is a general-purpose statistical software
package.
http://www.stata.com/
• To install the HOI (Human Opportunity Index)
command type enter in the command window
ssc install hoi
and
ssc install indeplist
Poverty Reduction and Economic
Management Network
http://worldbank.org /visualize inequality
HOI in STATA: Syntax
hoi depvar [indepvars] [weight] [if] [in]
To limit the
Dependent List of If the individuals estimation, if
variable: the independent have an assigned necessary, to
opportunity. variables: weight (usually in certain sub-
circumstances. samples), this sample in the
variable identifies dataset.
such weights.
Poverty Reduction and Economic
Management Network
http://worldbank.org /visualize inequality
HOI in STATA: an example
Poverty Reduction and Economic
Management Network
http://worldbank.org /visualize inequality
Sub-national analysis
• If there are two regions in a nation and the HOI
increases in one region and remains the same in the
other region, then the overall HOI of the nation
should increase.
• However, even if HOI increases in one region, and
remains the same in the other, then it is still possible
that the overall HOI falls.
• This implies that we can not use the D-index based
HOI for sub-regional analysis. We have to use the
Geometric version of HOI.
Poverty Reduction and Economic
Management Network
http://worldbank.org /visualize inequality
Sub-national analysis
Poverty Reduction and Economic
Management Network
http://worldbank.org /visualize inequality
Sub-national analysis
The geometric HOI uses the weighted
geometric mean rather than the
arithmetic mean of the circumstance-
group specific coverage rates (or average
probability of access) for each of the
regions.
Poverty Reduction and Economic
Management Network
http://worldbank.org /visualize inequality
HOI in STATA: Shapley decomposition
• If we want to estimate the marginal
contribution of each circumstance to
inequality we use the Shapley decomposition
(see Shorrocks, 2002).
• To install the Shapley decomposition of the
HOI enter in the command window
ssc install hoishapley
Poverty Reduction and Economic
Management Network
http://worldbank.org /visualize inequality
HOI in STATA: Shapley decomposition
hoishapley depvar [weight] [if] [in],
shapley(indepvars) shapleyc(categorical indepvars)
List of independent variables List of categorical independent
that enter the decomposition variables that enter the
(not including categorical decomposition.
variables).
Poverty Reduction and Economic
Management Network
http://worldbank.org /visualize inequality
HOI in STATA: Shapley decomposition
• It seems that education of the household head is the
most important source of inequality, explaining 41.9%.
Poverty Reduction and Economic
Management Network
http://worldbank.org /visualize inequality
Changes in HOI
• Broadly the HOI can change for two reasons:
– Composition effect : A redistribution of the
population itself, where the size of the
circumstance groups changes, or
– Coverage effect : An improvement in the coverage
due to
• Scale effect : A change in the overall coverage for the
entire population without any changes in inequality.
• Equalization effect: A change in the equality of access
to the opportunity between the circumstance groups.
Poverty Reduction and Economic
Management Network
http://worldbank.org /visualize inequality
HOI in STATA: Time decomposition
hoi attendance male age urban edu_head male_head
[w=weight], by(year) decomp2
Additional syntaxis for
decomposition. Year can
be any time variable with
two periods, make sure it
is an integer.
Poverty Reduction and Economic
Management Network
http://worldbank.org /visualize inequality
HOI in STATA: Time decomposition
Estimation
for 1995
Estimation
for 2005
Poverty Reduction and Economic
Management Network
http://worldbank.org /visualize inequality
HOI in STATA: Time decomposition
Poverty Reduction and Economic
Management Network
http://worldbank.org /visualize inequality
Changes in HOI
• There was a reduction of 1.77 percentage points in the
HOI.
• 0.36 percentage points of the reduction were due to
change in the distribution of circumstances.
• 0.96 percentage points of the reduction were due to
reduction in the overall coverage.
• 0.45 percentage points were due to the equalization effect.
Poverty Reduction and Economic
Management Network
http://worldbank.org /visualize inequality
