Causal effects of HIV on employment status in low-income settings

Research output: Contribution to journalArticlepeer-review


This paper estimates the causal impact of being HIV positive on individual employment status using a recursive bivariate probit with male circumcision as the instrument to overcome the endogeneity arising from simultaneity bias. The results show that being HIV positive reduces the probability of being employed by 5 percentage points among males in Uganda. The effect is greater for individuals employed in manual labor than non-manual labor. When limiting the sample to mainly individuals employed in subsistence agriculture, we find a 4 percentage point reduction in the likelihood of employment, suggesting that the effect occurs primarily through reductions in labor supply as opposed to demand. This is supported by additional analysis using univariate probit regressions to assess the association between different levels of HIV illness (as measured by CD4 cell count) and the likelihood of employment. The magnitude of the association increases as CD4 cell count decreases. Having a CD4 cell count of 200permm(3) or below is associated with a 9 percentage point reduction in employment compared to individuals with CD4 cell counts above 200permm(3).

Original languageEnglish
Pages (from-to)248-260
Number of pages13
JournalEconomics and Human Biology
Issue numberPt A
Publication statusPublished - 8 Sep 2017

Bibliographical note

© 2017 Elsevier B.V. All rights reserved.This is an author-produced version of the published paper. Uploaded in accordance with the publisher’s self-archiving policy.


  • Journal Article
  • Instrumental variable
  • CD4
  • Employment
  • Recursive bivariate probit
  • Severity of Illness Index
  • Models, Econometric
  • Humans
  • Middle Aged
  • Uganda/epidemiology
  • Developing Countries/statistics & numerical data
  • Male
  • CD4 Lymphocyte Count
  • Socioeconomic Factors
  • Employment/statistics & numerical data
  • Young Adult
  • Adolescent
  • Occupations/statistics & numerical data
  • Adult
  • HIV Infections/epidemiology

Cite this