Organized session
Examining the Effects of Social Health Insurance in low- and middle-income countries: Enrollment, Utilization, Financial Protection, and Labor Supply
Talks
Does Free Enrolment Set an Equal Opportunity? Evidence from Social Health Protection Program in Pakistan
Vendula Stepanik, University Erlangen-Nürnberg
Introduction
Background: In this study, we focus on enrolment in a social health protection program in Pakistan, known as the Sehat Sahulat program. The program enrolment was a gradual process. It began with the poorest 21% of the population in 4 pilot districts in early 2016. It quickly expanded to cover the entire province, extending eligibility to 51% of the poorest population in August 2016 and eventually encompassing the bottom 69% of the entire population in early 2017. Specifically, we examine enrolment differences based on household poverty status and identify household characteristics driving this difference. We aim to contribute to the literature on the shift towards universal health coverage through publicly funded health insurance schemes. Methods: We matched the extensive administrative data that consists of the universe of eligible and enrolled households in the Khyber Pakhtunkhwa province with the sub-sample from the population census through a unique identifier. Since the official eligibility dates do not correspond with the start of the enrolment in each union council, we identify the effective eligibility dates per each union council and enrolment period. Using Kaplan-Meier survival curves and stratified Cox proportional hazard regression, the relationship between deprivation quintiles and enrolment timing was evaluated, accounting for geographic location and eligibility period. Blinder-Oaxaca decomposition assessed the impact of household characteristics on enrolment timing and share. Results: By 2018, 62.26% of eligible households were enrolled, with the highest share (78.92%) among the poorest 21% (first period). Enrolment rates were 73.32% and 23.52% in the subsequent periods. Within the first period, the poorest households enrolled slower and less than the second poorest, with 1.5 and 1.1 percentage point differences in pilot and non-pilot districts. Days to enroll differed by 18 and 11 days in pilot and non-pilot districts, respectively. These differences were partly attributed to household education, yet were not economically significant. Conclusion: In conclusion, the poorest segment showed the highest enrolment share in the program. However, minor differences existed within this segment, with the poorest subgroup enrolling slightly slower. Policymakers should consider these findings when establishing such programs and provide targeted support to ensure equitable access for vulnerable populations.
Health Insurance Coverage and the Hypertension Care Cascade: Evidence from A District Fixed Effects Study in Ghana
Jakob Pfeifer, University Frankfurt
Introduction
Introduction: Hypertension is a pressing public health concern in Ghana, demanding a comprehensive exploration of the Hypertension Care Cascade to inform effective interventions. We investigate how health insurance coverage affects the various stages of the Hypertension Care Cascade (diagnosis, treatment, and control) for individuals aged 15-49, using data from the 2014 Ghana Demographic and Health Survey (GDHS). Methods: We utilized cross-sectional data from the 2014 GDHS with a two-step sampling process for national representation. Our study focused on individuals aged 15-49 and incorporated comprehensive face-to-face interviews and blood pressure measurements, using standardized health-related questionnaires. Fixed effects models were used to assess the influence of health insurance on each cascade stage, accounting for demographic and health factors, with modified Poisson regression for dichotomous variables. Results: Among 1,355 hypertensive individuals, 80.1% with health insurance demonstrated higher rates of diagnosis (48.1%), treatment (34.6%), and control (18.6%) compared to the uninsured (27.4%, 11.9%, 5.9%). Health insurance significantly increased the likelihood of diagnosis by 25% compared to the uninsured (IRR = 1.25, 95% CI [1.05, 1.50], p = 0.014). However, no significant impact was observed in the treatment (IRR = 0.99, 95% CI [0.79, 1.26], p = 0.964) or control stages (IRR = 0.89, 95% CI [0.54, 1.47], p = 0.650). Wealth positively influenced treatment initiation and control, while gender disparities persisted. Females exhibited a 38% lower likelihood of diagnosis compared to males (IRR = 0.62, 95% CI [0.50, 0.77], p = 0.000), a 23% lower likelihood of receiving treatment (IRR = 0.77, 95% CI [0.60, 0.99], p = 0.043), and a 36% lower likelihood of achieving hypertension control (IRR = 0.64, 95% CI [0.51, 0.80], p = 0.000). Discussion: Our study examines the connection between health insurance status and health outcomes for Ghanaians with hypertension amidst the growing threat of non-communicable diseases (NCDs). While health insurance significantly boosts hypertension diagnosis rates, its influence diminishes in subsequent stages. Expanding insurance coverage alone may not fully optimize health outcomes. To achieve comprehensive improvements, enhancing the benefits package and ensuring access to prescribed drugs for those enrolled in the local national health insurance scheme (NHIS) is crucial.
Effects of the Indian National Health Insurance Scheme (PM-JAY) on Hospitalisations and Out-of-pocket Expenditures
Christoph Strupat, German Institute of Development and Sustainability (IDOS)
Introduction
India launched one of the world’s largest health insurance programmes, PM-JAY, targeting more than 500 million Indians from economically and socially disadvantaged groups. PM-JAY is publicly funded and covers costs related to hospitalisations in public and private facilities. Between December 2019 and February 2020 we conducted a household survey covering six Indian states, and collected data on socioeconomic and demographic information, health status, hospitalisations, and PM-JAY for more than 50,000 PM-JAY eligible individuals. Using a series of multivariate regression models, we estimated whether PM-JAY was associated with any changes in hospitalisations, out-of-pocket expenditures (OOPE) and catastrophic health expenditures (CHE), and whether these differed across public and private facilities. We found that PM-JAY was not associated with an increase in hospitalizations, but it increased the probability of visiting a private facility by 4.6% points (p<0.05). PM-JAY was associated with a relative reduction of 13% in OOPE (p<0.1) and 21% in CHE (p<0.01). This was entirely driven by private facilities, where relative OOPE was reduced by 17% (p<0.01) and CHE by 19% (p<0.01). Given the complex Indian healthcare system with the presence of parallel public and private systems, our study concludes that for economically and socially disadvantaged groups, PM-JAY contributes to improved access to secondary and tertiary care services from private providers. As curative care is concentrated in the private sector and the size of OOPE is several times higher in private-sector hospitalisations, it appears that PM-JAY also reduces the immediate financial burden for beneficiaries. However, for universal health coverage, India needs also to invest in primary health care and the public health sector, given that private facilities are expensive and concentrated in urban areas.
Good Policy Gone Bad: Health Insurance Premium Changes and Labour Supply in a Low-Income Country
Emmanuel Nshakira-Rukundo, Leibniz-Institute of Economic Research (RWI)
Introduction
The health insurance – labour supply relationship has been a longstanding theme of health and labour economists, establishing that provision of health insurance was central to labour supply (Currie & Madrian, 1999; Gruber & Yelowitz, 1999). The advantages of health insurance go beyond the immediate financial protection to extensive income effects emanating from unspent health care costs (Nyman, 2001, 2008). Gaining health insurance (especially employer-provided) is therefore likely to increase labour supply (Garthwaite, Gross, & Notowidigdo, 2014; Yörük & Xu, 2019). However, one understudied question, especially in the context of low income countries is how labour market participation might change due to changes in health insurance costs (premiums). This paper studies the effects of a premium policy change in a large health insurance programme in Rwanda. Specifically, the policy change introduce on the one hand, premium increases to higher socioeconomic status individuals and on the other, premium waiver to lower socioeconomic status individuals. Targeting was through a community-based targeting system, which, due to it’s flaws and bottlenecks (Hasselskog, 2018; Sabates-Wheeler, Yates, Wylde, & Gatsinzi, 2015), presented a quasi-random set up in which policy dimension exposure was more likely exogenous than endogenously determined. We use three rounds of nationally representative repeated cross-sectional data, including a pre-policy change round that acted as baseline. We employed a differences-in-differences with matching framework (Uysal, 2015) to assess the effects on total labour supply, agricultural, non-agricultural and wage labour supply. Our results show two main findings. Firstly, both premium increases and premium waivers had a negative effect on labour supply. Secondly, significant labour supply reductions were experienced in the medium term than the short term. These results point to learning and adjustments driven by both the income effect – for individuals experiencing premium waivers and potential manipulation effects – for individuals exposed to the premium increase. This behaviour, especially potential welfare under-reporting among higher SES households is consistent with previous studies on health insurance (Shi, 2016) and in tax policies (Saez, 2010), where bunching around the targeting thresholds is observed. Men and individuals in rural areas were the ones most likely to reduce their labour supply. The results serve a cautionary note on social policy and social welfare subsidies in low income countries.