How do income and education affect health inequality: Evidence from four developing countries

Satis Devkota, Mukti Upadhyay

Research output: Contribution to journalArticlepeer-review

3 Scopus citations


Using household survey data from four countries ‒ Albania, Nepal, Tajikistan and Tanzania ‒ this article calculates income-related inequality in health care utilization. We measure health disparity separately for generally and chronically ill individuals by constructing two models: one for the probability of a visit to a physician and another for the number of visits. Following model-based measurements, we decompose inequality into two major parts: one accounted for by identity-related factors and another by socioeconomic and other factors such as education, geography and distance to a clinic.We propose a new method to quantify the effect of changes in income and education on health disparity. One of our important findings suggests that health disparity is pro-rich in all our sample countries. The pro-rich disparity is prevalent among generally ill as well as chronically ill patients, in both visit probability and visit frequency models. Health inequality seems primarily driven by income differences followed by nonidentity factors. Further, the principle of equal treatment for equal need is not fulfilled in any of our countries. Among policy implications, increasing average income and education in a way that also reduces disparity in income and education, respectively, will substantially shrink inequality in health care utilization.

Original languageEnglish (US)
Pages (from-to)5583-5599
Number of pages17
JournalApplied Economics
Issue number52
StatePublished - 2015

Bibliographical note

Publisher Copyright:
© 2015 Taylor & Francis.


  • Developing countries
  • Health concentration index
  • Health inequality
  • Socioeconomic determinants of health care demand


Dive into the research topics of 'How do income and education affect health inequality: Evidence from four developing countries'. Together they form a unique fingerprint.

Cite this