The cost of knowledge in a fragile state: Analyzing educational access and expenditure inequality in Somalia
Description
This dataset originates from the 2022 Somalia Integrated Household Budget Survey (SIHBS), a nationally representative, multi-topic household survey conducted by the Somalia National Bureau of Statistics (SNBS) with support from the World Bank. The survey covers urban, rural, and nomadic populations across 17 pre-war administrative regions in Somalia, achieving a 96% response rate. This specific analytical subset focuses on 25,186 individuals aged 5–24 years to examine the microeconomic determinants of educational expenditure. The data includes variables related to household demographics, remittance receipt, non-labor income, displacement (IDP) status, and educational spending (tuition, textbooks, uniforms, and transportation). Given that approximately 60% of households report zero educational spending, this dataset is structured to support censored regression analysis (Tobit models) to account for structural market exclusion and the "survival-first" consumption patterns prevalent in fragile and conflict-affected situations. It serves as a critical resource for understanding human capital investment, the "private school premium," and the impact of external financial flows in the context of the Somali education sector.
Files
Steps to reproduce
Data Acquisition: Obtain the 2022 Somalia Integrated Household Budget Survey (SIHBS) dataset from the Somalia National Bureau of Statistics (SNBS) via the Somalia National Data Archive (SoNADA). Sample Selection: Restrict the analytical sample to individuals aged 5–24 years. Data Cleaning: Perform listwise deletion to remove cases with missing data across socio-demographic and financial indicators (e.g., as done with the 209 dropped observations in the study). Variable Construction: Define educational expenditure (continuous, USD), remittance receipt (binary), non-labor income (aggregated), and include controls for household head characteristics, IDP status, school type, and place of residence. Statistical Modeling: Conduct a survey-weighted Tobit regression to account for the "zero-spending" left-censoring. Diagnostics: Perform a Variance Inflation Factor (VIF) test to check for multicollinearity and use a Wald Chi-Square test for overall model fit. Validation: Run OLS and Probit models as robustness checks to confirm the stability of the Tobit estimates.