Download StudyGeneral

Study Overview

Housing and Human Capital: Condominiums in Ethiopia
Study is 3ie funded:
Study ID:
Initial Registration Date:
Last Update Date:
Study Status:
In Development

Rapidly growing urban centers have led housing policy to become increasingly important across the developing world. In this project, we leverage random lotteries for government-subsidized condominiums in Ethiopia to understand the medium-run effects of winning a condominium on the educational and labor market outcomes of children. This policy affects children directly by influencing their neighborhood of residence, exposing them to different peers and educational and labor market opportunities. The direction of the policy’s direct impacts are theoretically ambiguous given potentially offsetting features of the condominium neighborhoods. The condominium policy may also indirectly affect outcomes by increasing parental wealth. Through detailed household surveys, supplementary administrative data, and a structural selection model, we separately identify the relative importance of the direct and indirect consequences of this policy.

Registration Citation:
Economic Policy
Urban Development
Additional Keywords:
housing, human capital
Secondary ID Number(s):

Principal Investigator(s)

Name of First PI:
Daniel Agness
UC Berkeley
Name of Second PI:
Tigabu Getahun
EconInsight - Ethiopia

Study Sponsor

Weiss Fund for Development
Study Sponsor Location:
United States

Research Partner

Name of Partner Institution:
EconInsight Center for Development Research
Type of Organization:
Private firm

Intervention Overview


In 2005, the Ethiopian government launched apublic housing policy to build hundreds of thousands of residential units for urban dwellers in Addis Ababa. The stated goals of the program were to provide housing for low- and middle-income urban dwellers and support the domestic construction industry. Since its inception, the policy has been massively oversubscribed. Thus far, there have been two rounds of registrations taking place in 2005 and 2013. An estimated 50% of all households in Addis Ababa have registered for the program, with over 900,000 applications to date. Through 2019, nearly 200,000 units had been successfully completed and transferred to residents across 13 lottery rounds. This policy continues to this day -- tens of thousands of new units are expected to be occupied in 2022. 

The criteria for eligibility are: (1) only one application per household; (2) the heads of household cannot own property in Addis Ababa; (3) the household must have resided in Addis Ababa for at least six months. Households are free to choose the size of the desired unit but not the location.

Condominiums are allocated via random lottery. In order to be eligible for the lotteries, after submitting an application, the household must open a bank account at the Central Bank of Ethiopia (CBE) and make deposits towards a down-payment. After winning, there is no requirement that the household move into the unit that they win; they are free to rent it out or leave it unoccupied. Winning households may only sell their condominium unit after an embargo period of five years.

Theory of Change:

We leverage random lotteries for government-subsidized condominiums to identify how neighborhoods of residence and household wealth, each of which change dramatically for lottery winning households, impact medium-run economic, educational, and health outcomes for the children of winning households. This policy functionally reallocates families from low-quality, dense housing in the city center to higher-quality housing on the outskirts of the city. By virtue of their location on the city's outskirts, the areas in which the government-constructed condominiums are located have worse labor market access, less developed social networks, and may have lower quality education and health infrastructure. Relative to programs in the United States and North America that focus on moving families from ``bad" neighborhoods to ``good" ones, the heterogeneity of neighborhood quality along multiple dimensions in the Ethiopian context makes the effects on children's outcomes unclear ex-ante.

While nearly all winners purchase their unit after winning a lottery, a substantial share (~40%) do not move into the unit but choose to rent it out as a supplementary source of income. Only children in households that move into the condominiums will experience the change in housing quality, educational and health infrastructure access, and peer characteristics. Regardless of whether their family moves into the unit, children may benefit from the lottery due to an increase in parental wealth via a government-subsidized asset. Identifying the relative importance of these two channels, neighborhood of residence and parental wealth, is a primary aim of this study.

Multiple Treatment Arms Evaluated?

Implementing Agency

Name of Organization:
Addis Ababa Housing Development and Administration Bureau
Type of Organization:
Public Sector, e.g. Government Agency or Ministry

Program Funder

Name of Organization:
Household downpayments and public subsidies
Type of Organization:
Public Sector, e.g. Government Agency or Ministry

Intervention Timing

Intervention or Program Started at time of Registration?
Start Date:
End Date:
Evaluation Method

Evaluation Method Overview

Primary (or First) Evaluation Method:
Natural experiment
Other (not Listed) Method:
Additional Evaluation Method (If Any):
Instrumental variables
Other (not Listed) Method:

Method Details

Details of Evaluation Approach:

Using the list of all condominium registrants, we will draw a stratified random sample of lottery winners and losers. We will stratify by the location at the time of registration, number of bedrooms requested, and applicant gender. For winning households, we will select a subset of condominium sites across rounds in a first step randomization. We will oversample from early lottery rounds to increase the share of children (1) with long exposures to treatment and (2) who have entered the labor market or completed post-secondary education. 

Random allocation of lotteries allows for identification of ToT and ITT effects of a child’s family winning a lottery on later life outcomes. This is a rich empirical setting and has served as the basis for most analysis of similar policies to date. We will instrument the lottery treatment with a binary indicator for winning a lottery to manage the small share of non-compliers who do not successfully purchase a condominium after winning. As the literature shows that neighborhood effects vary with exposure, we will further estimate a linear exposure model where treatment effects are a function of the child’s age at the time their parents win a lottery. We will estimate a set of models that incorporate family fixed effects, such that the models are identified using within-family variation in treatment timing and exposure across siblings and use variance decomposition methods to identify the relative importance of family and neighborhood characteristics.

Finally, we will extend the structural selection model developed in Kline & Walters (2016) to account for endogenous selection into moving into, renting, or selling the condominium that a household wins. This model interacts the lottery instrument with plausibly exogenous baseline household characteristics in a 2-stage least squares framework to identify models with multiple treatment and fall-back states.

Outcomes (Endpoints):

We have three sets of primary outcomes for children: (1) educational attainment and test scores; (2) employment, income, labor market participation, and formal sector employment; (3) aspirations and well-being.

Educational attainment outcomes will be measured as whether a child is enrolled in school, has completed primary school, has completed secondary school, has enrolled in post-secondary education, or has completed post-secondary education. Each of these will only be considered on grade-relevant sub-samples of children. Test scores are measured in grades 8, 10, and 12.

Well-being will be measured based on strength and difficulties questionnaires administered to parents (SDQ). These will be scored following standard SDQ scoring using both the integer and categorical scores. Aspirations are measured by asking children about their desired occupations and educational attainment as well as the likelihood of these desires occuring. We map these responses into cardinal occupation and education aspiration measures. We consider weighted versions of these measures where the weights are determined by the likelihood of occurance. Finally, we combine the two aspirational measures into a normalized aspiration index following Anderson (2008).

We have 15 secondary outcomes: 1. Numeracy; 2. Literacy; 3. Fluid intelligence; 4. School quality; 5. Dwelling quality; 6. Neighborhood quality; 7. Social cohesion; 8. Parental investment in their children; 9. Child labor rates; 10. Occupation; 11. Family structure/marriage age; 12. Parental aspirations; 13. Household economic well-being index; 14. Household wealth; 15. Parental income.

Outcomes 1-8, 13, and 14 will be built as indexes following Anderson (2008).

Unit of Analysis:
Individual child

Treatment is defined as an individual's parent winning a lottery for a condominium unit when they were a child. We have three primary hypotheses:

1. Treatment will increase educational attainment and test scores for children in families that win a lottery.

2. Treatment will increase employment, income, labor market participation, and formal sector employment for children in families that win a lottery. 

3. Treatment will increase the aspirations and improve the well-being of children in families that win a lottery. 

We have 14 secondary hypotheses releated to the secondary outcomes mentioned above:

1. Treatment will increase measures of numeracy, literacy, and fluid intellligence.

2. Treatment will not change school quality.

3. Treatment will increase housing quality. 

4. Treatment will increase neighborhood quality.

5. Treatment will decrease social cohesion.

6. Treatment will increase parental investment. 

7. Treatment will decrease child labor rates. 

8. Treatment will lead to occupational upgrading. 

9. Treatment will delay marriage age and age of first child.

10. Treatment will increase parental aspirations for their children.

11. Treatment will increase an index of household economic well-being, household wealth, and parental income. 

12. Winning households will be more likely to move to condominiums in higher quality neighborhoods. 

13. Winning households will be more likely to move to condominiums closer to where they were previously living.

14. Winning households will be more likely to move to condominiums if they were previously living in a low quality neighborhood.

Unit of Intervention or Assignment:
Number of Clusters in Sample:
The intervention is assigned at the household level but we select a subset of condominium sites.
Number of Individuals in Sample:
3,100 households, 5,000-6,000 children
Size of Treatment, Control, or Comparison Subsamples:
Treatment: 1,600 households, 2,600 children; Control: 1,500 households, 2,400 children

Supplementary Files

Analysis Plan:
Other Documents:

Outcomes Data

We will combine household surveys with administrative data on test scores, school quality, neighborhood characteristics, and wages.
Data Already Collected?
Data Previously Used?
Data Access:
Data Obtained by the Study Researchers?
Data Approval Process:
Approval Status:

Treatment Assignment Data

Participation or Assignment Information:
Data Obtained by the Study Researchers?
Data Previously Used?
Data Access:
Data Obtained by the Study Researchers?
Data Approval Process:
Approval Status:

Data Analysis

Data Analysis Status:

Study Materials

Upload Study Materials:

Registration Category

Registration Category:
Prospective, Category 1: Data for measuring impacts have not been collected

Completion Overview

Intervention Completion Date:
Data Collection Completion Date:
Unit of Analysis:
Clusters in Final Sample:
Total Observations in Final Sample:
Size of Treatment, Control, or Comparison Subsamples:


Preliminary Report:
Preliminary Report URL:
Summary of Findings:
Paper Summary:
Paper Citation:

Data Availability

Data Availability (Primary Data):
Date of Data Availability:
Data URL or Contact:
Access procedure:

Other Materials

Survey Instrument Links or Contact:
Program Files:
Program Files Links or Contact:
External Link:
External Link Description:
Description of Changes:

Study Stopped