ABSTRACT In this paper, we aim to provide robust estimates of wage returns to education in Ethiopia. Previous estimates vary widely based on the scope and size of the study sample, variables employed, and estimation method. We use two recent rounds of national labour force survey data and a pseudo- (or synthetic) panel framework to address the well-known potential endogeneity of schooling in wage regressions. Birth-year cohort based pseudo-panel estimate of the return to an additional year of education is between 14 percent and 16 percent, 4 to 5 percentage points higher than that from ordinary least squares (OLS). The significant downward bias in OLS is further confirmed by two-year pseudo-panel estimates and remains robust, to a large extent, as we use an alternate model, change the estimation sample, and address selection bias concurrently with endogenous schooling using a novel approach. We discuss some possible explanations in the context of Ethiopia.

中文翻译：

---

工资在埃塞俄比亚重返教育

摘要本文旨在为埃塞俄比亚的教育提供可靠的工资回报估算。根据研究样本的范围和大小，采用的变量和估计方法，先前的估计差异很大。我们使用最近两轮的全国劳动力调查数据和一个伪（或综合）面板框架来解决工资回归中众所周知的潜在的内生性内生性。基于出生年队列的伪面板估计，接受额外一年教育的回报率在14％至16％之间，比普通最小二乘法（OLS）高出4至5个百分点。两年的伪面板估计进一步确认了OLS的显着下降偏差，并且在很大程度上，当我们使用替代模型，更改估计样本，并在使用新方法进行内生教育的同时解决地址选择偏见。我们在埃塞俄比亚的背景下讨论了一些可能的解释。