Gaukhar Uvakbayeva

Kazakh National Al-Farabi University, Almaty, Kazakhstan

E-mail: Gauhar.Uvakbaeva@kaznu.edu.kz

ORCID 0009-0008-6287-4121

Do health-financing regime shifts matter for vulnerable employment? Evidence from a country - year state clustering approach

Nataliia Sheliemina

Sumy State University,

Sumy, Ukraine

E-mail: biletskayanat@gmail.com

ORCID 0009-0003-2120-0159

Lidia Mielczarek

National Medicines Institute, Warsaw, Poland

E-mail: l.mielczarek@nil.gov.pl

ORCID 0000-0003-2820-9773

Zsuzsanna Herman

University of Debrecen,

Debrecen, Hungary

E-mail: herman.zsuzsanna@unideb.hu

ORCID 0009-0008-1474-5755

Abstract. Rising out-of-pocket pressures and renewed attention to financial protection in health systems have intensified interest in how health-financing architecture may shape labour-market vulnerability. This article aims to investigate whether shifts in health-financing regimes, captured as data-driven country-year “states” of the financing mix, are associated with changes in vulnerable employment. An unbalanced annual panel merges OECD health-financing components (government, social insurance, private insurance, voluntary insurance, NPISH, enterprise financing, out-of-pocket and cost-sharing; % of GDP) with World Bank controls (GDP per capita; employment-to-population ratio) and the modelled ILO estimate of vulnerable employment; country–year observations are clustered on financing indicators (k-means) and linked to vulnerable employment using two-way fixed-effects models with country-clustered standard errors, complemented by switchers-only and entry/exit specifications. Relative to the modal social-insurance-dominant state (Cluster 5), the compulsory-private-insurance/high household-burden state (Cluster 6) is associated with higher vulnerable employment (β = 5.802, p < 0.001), a relationship that remains significant among switchers (β = 4.244, p < 0.01). Predicted vulnerable employment equals 21.77 in Cluster 5 versus 27.57 in Cluster 6 (baseline), rising to 34.77 for Cluster 6 when key channels (x3, x7, x8) are added. Entry into Cluster 6 is associated with a short-run decline (β = −1.024, p < 0.01), suggesting that transition dynamics may differ from long-run regime-level differences. 

Received: May, 2025

1st Revision: October, 2025

Accepted: December, 2025

DOI: 10.14254/2071-789X.2025/18-4/9

JEL Classification: I13, I18, J21, C38, C23

Keywords: health-financing regimes, vulnerable employment, out-of-pocket payments, cost-sharing, compulsory private insurance, k-means clustering, two-way fixed effects