Human Development

Network membership and demand for health insurance

  • Blog Post Date 17 January, 2025
  • Articles
  • Print Page
Author Image

Tanika Chakraborty

Indian Institute of Management, Calcutta

tanika@iimcal.ac.in

Despite being free and having liberal eligibility criteria, the adoption of public health insurance in India remains low. This article examines how informal networks influence adoption behaviour, in the context of Andhra Pradesh’s Aarogyasri programme. It shows that network type matters: information networks do not significantly impact the uptake of public insurance, whereas financial networks actually facilitate, rather than hinder, public insurance adoption.

Delivery of affordable healthcare remains a critical challenge for governments worldwide. Even among advanced countries, institutional mechanisms for healthcare provision vary widely. Some countries, such as Canada and the United Kingdom, have opted for State-provided healthcare, while others, like the United States, rely primarily on market-based systems.

In post-independence India, various paradigms of health policy have shaped healthcare delivery, utilising different institutional arrangements. During the pre-liberalisation period, the focus was on free public hospitals and health centres, an approach that had limited success. After 2005, recognising that public provision alone did not ensure access to quality healthcare, a new model was developed that involved free insurance to cover tertiary healthcare for the poor at both public and private facilities. The central government (in the form of Ayushman Bharat Pradhan Mantri Jan Arogya Yojana) and numerous state governments (for example, Rajiv Arogyasri in Andhra Pradesh, Swasthyashree in West Bengal, Bhamashah Swasthya Bima Yojana in Rajasthan) introduced such insurance programmes.

Despite being free and having liberal eligibility criteria, the adoption of these programmes remains low. According to the National Family Health Survey (NFHS-5), only 41% of the households in India had some health insurance (International Institute for Population Sciences and ICF, 2021). Understanding why India’s poor may not readily use free insurance is key to addressing high out-of-pocket expenditures on tertiary healthcare. In new research (Bhattacharya et al. 2024), we investigate the issue from an institutional standpoint.

In any resource allocation problem, three institutions play pivotal roles: State, market, and community. While State and market mechanisms are typically central in developed countries, the community also holds critical importance in developing contexts, in terms of healthcare decisions. We examine the role of community networks in influencing the adoption and utilisation of public health insurance.

Community network and formal insurance

While public insurance is free, adoption and utilisation involve different transaction costs that policyholders must bear (Naidu 2021, Rohit 2018). Community membership can influence the uptake and utilisation of public insurance by either increasing or decreasing these costs. A community network provides various services to its members, including informal insurance in the form of risk-sharing and financial support. Hence, network membership can crowd out the adoption of costly insurance, whether public or private, by acting as a substitute for formal insurance (Jowett 2003). On the other hand, networks can play a complementary role of information dissemination and reduce the transaction cost of information associated with formal insurance (Debnath and Jain 2020).

Taken together, these opposite effects suggest that the overall impact of network membership on adoption of formal insurance depends largely on the services that the network provides. We examine the distinct roles of financial and information networks in influencing the uptake and utilisation of public health insurance.

Empirical approach

We use the Young Lives Survey (YLS) panel data from the (present-day) Indian states of Andhra Pradesh and Telangana. YLS follows households of 3,000 children across six districts – Srikakulam, West Godavari, Anantapur, Kadapa, Karimnagar, Mahabubnagar, along with Hyderabad city – over five rounds between 2002 and 2016 (Young Lives, 2017). We are interested in the Rajiv Aarogyasri Scheme (RAS), a social health insurance programme introduced in undivided Andhra Pradesh in 2007. The programme continued to operate in the new states of Andhra Pradesh and Telangana, formed in 2013, by dividing erstwhile Andhra Pradesh.

A timeline of the survey and introduction of RAS is presented in Figure 1. Rounds 1 and 2 cover a period prior to the introduction of RAS and rounds 3 to 5 cover a period post the introduction of RAS. For most of the indicators we use, round 1 is not comparable to later waves. Hence, we consider rounds 2 to 5. While information on insurance adoption is available from round 3, we use round 2 for lagged information wherever relevant.

Figure 1. Timeline of Young Lives Survey and Rajiv Aarogyasri Scheme


We distinguish between the effects of financial and information networks on insurance adoption behaviour of households. Since utilisation is integrally linked to health shocks, we examine the relationship in the context of health shocks experienced by the households. The issue of measurement of the two networks is of critical importance here.

Financial network: We measure a household’s access to informal financial networks based on two questions: how the household would raise money in one week, and how the household responded to a shock. Households that responded that they would be relying on relatives and friends in their own or another community, or on informal loan are deemed to have an informal financial network.


Information network: We define informal information networks based on two questions: whether any household member is part of any social group, and whether any household member engages with the community. We consider a household to have access to an informal information network if any member responded in the affirmative.

Empirical findings

Overall, informal network membership increases adoption of formal health insurance. These findings are more in line with Debnath and Jain (2020), and stand in contrast to Jowett (2003). However, we find that only financial networks positively affect adoption. Membership in information networks does not matter. This result might appear counterintuitive as informal financial network is usually seen as a substitute for formal insurance. We explain our findings using a theoretical framework, and provide an essence of the argument below. However, before we do that let us look at the remaining findings.

Figure 2 depicts the coefficient plot of community membership on the probability of take-up and utilisation of public insurance. These coefficients show the main effect of community membership on take-up and utilisation. While membership in financial networks has positive significant impact on both take-up and utilisation, the effect is close to zero for information networks.

Figure 2. Effect of network membership on take-up and utilisation of public insurance


Figure 3 represents the result that emerges when we consider the behaviour of households in the context of health, by looking at the effects of interaction between health shock and community membership on take-up and utilisation of insurance. While network members are more likely to adopt health insurance on average, when faced with a health shock, non-members are more likely to take up free health insurance than members of financial networks. We explain the mechanisms in the conceptual framework given below. Once again, information networks do not matter.

Figure 3. Predicted probability of insurance take-up for members and non-members of financial networks, before and after health shocks

Notes: (i) Black bars (FN=0) represent predicted probability of insurance take up for people who are not members of financial networks. Grey bars (FN=1) represent the same for those that are members of financial networks. (ii) The left-hand side of the graph represents predicted probability for these groups before a health shock while the bars on the right-hand side represent post-shock.

Conceptual framework

To understand what might be driving these findings, we draw upon a very simple expected-utility framework to model people’s response to a health shock. Our results are based on the premise that free public insurance is not free in the real sense; both enrolment and claim settlement involve some transaction costs. People only enrol for the insurance if their expected benefit exceeds cost. The same is true for utilisation of insurance in the event of shock; people file claims if the expected receipt is higher than the transaction costs.

Enrolment decision before a health shock is realised: We assume that people differ in terms of their perceived probability of experiencing a health shock while the actual cost of health shock is same for everyone. But these set of assumptions mean that the expected cost of health shock – the product of perceived probability and the actual cost of health shock – is different for everyone.

Decision by non-members: People who are not part of any network (non-members), simply compare their expected health cost with the transaction cost of enrolling in the insurance. If the expected health cost is higher than the cost of enrolment, they enrol for the insurance, otherwise they do not. This formulation gives us a threshold value for the perceived probability of health shock – if one’s perceived probability of a shock is higher than that threshold (we call them the pessimists), they enrol for the insurance. On the other hand, non-members with perceived probability below the threshold (we call them the optimists) do not enrol for the insurance. So essentially, only a fraction of non-members (pessimists) opts for insurance.

Decision by members: For the members of financial networks, enrolment decisions are taken by network representatives for all members, rather than by individuals. Such network leaders may be respected village elders (Ambrus et al. 2014), as is common in the developing world. We argue that financial networks are a substitute for formal insurance; if a member faces a shock and does not have insurance, the network must bail them out. Therefore, financial networks have control over someone’s decision to adopt formal insurance. This is not the case for information networks and therefore, information networks cannot influence members’ decision to enrol, at least through this channel. In our formulation, the members of a financial network collectively evaluate the expected cost of a health shock using the average perceived probability of health shock. This can be either done by the leader of the network or using some other information aggregation mechanism. An example of the latter can be some democratic mechanism that uses the median voter’s perceived probability for calculating the expected cost of a health shock.

Comparison of members’ and non-members’ decision to enrol: The implication for such a formulation is that the enrolment decision becomes the same for all members – if the average perception of risk at the community level is greater than the threshold, all members adopt insurance and if not, no one does. But among the non-members, only a fraction (pessimists) enrols for the insurance. Together, this means that the members of financial networks are more likely to adopt insurance than the non-members.

Enrolment decision after a health shock is realised: Our theoretical expectations change for individuals who have experienced a shock. When an individual faces a health shock, they essentially upwardly revise their own perception of risk. This would mean that some of the optimists now become pessimists. For people outside financial network, this means that more non-members enrol for formal health insurance. Enrolment in formal health insurance can take place at primary health centres (PHC), empanelled network hospitals (NWH), or at health camps organised by PHC or NWH (Aarogyasri Health Care Trust, 2013). But how does it affect members? As a result of a realised shock, the community as a whole becomes more pessimistic as the average perceived probability goes up. But that cannot increase the enrolment rate among network members as in our theoretical model, all the members are already enrolled in formal health insurance.

Utilisation of health insurance: The basic intuition remains the same for utilisation as well. Only in this case, individuals differ in terms of their beliefs about the extent their insurance claims will be reimbursed. Optimists attach higher probability to the event of getting reimbursed for their claims, while pessimists attach a lower probability to the same. The rest of the model in this part is like the earlier section. 

Empirical validation of our theory

Our baseline theory supports our empirical finding only for a range of parameter values where average perceived probability of a health shock is higher than a threshold level. How do we know that this condition works in the real world? In our model, that threshold value is equal to the ratio of transaction cost of enrolment and health cost. Note that we cannot possibly have an accurate estimate of these variables. But our theory is likely to hold when the enrolment cost to health shock ratio is low. In that case, expected health cost for the community is higher than the transaction cost, and therefore the community leaders would encourage all members to adopt insurance. This is going to be the case when the real cost of health shock is high, such as in rural areas where preventive medical check-ups are difficult to obtain, and hospitals are so far away that people decide to hospitalise only when the degree of illness is high enough. So, if our theory is correct, our empirical results should be driven by observations from the rural areas and from areas where hospitals are very far. We then run heterogeneity analyses and find that the results are indeed driven by such observations.

Policy implications

The key takeaway from our research is the importance of working closely with informal community networks to encourage the uptake of public insurance. These networks can influence individuals' insurance decisions because, as alternative sources of support, they hold a degree of sanctioning power. Community leaders are also likely to welcome public insurance, as it can free up community resources traditionally reserved for assisting members in financial distress.

Further Reading 

  • Aarogyasri Health Care Trust (2013), ‘Guidelines for Rajiv Aarogyasri Scheme’. Retrieved 2022, available here.
  • Ambrus, Attila, Markus Mobius and Adam Szeidl (2014), “Consumption Risk-Sharing in Social Networks”, American Economic Review, 104(1): 149-182.
  • Bhattacharya, T, T Chakraborty and A Mukherjee (2024), ‘Demand for Health Insurance: Financial and Informational role of Informal Networks’, SocArXiv Papers.
  • Debnath, Sisir and Tarun Jain (2020), “Social connections and tertiary health-care utilization”, Health Economics, 29: 464-474.
  • Jowett, Matthew (2003), “Do informal risk sharing networks crowd out public voluntary health insurance? Evidence from Vietnam”, Applied Economics, 35: 1153-1161.
  • Young Lives (2017), ‘Young Lives Survey Design and Sampling (Round 5): United Andhra Pradesh’.
  • Naidu, TA (2021), ‘Kakinada Apollo, GSL Hospital fined for collecting fees from Aarogyasri patients’, The Hindu, 27 February.
  • Rohit PS (2018), ‘No cashless EHS support, hospital tells patients’, The Hindu, 3 March.
No comments yet
Join the conversation
Captcha Captcha Reload

Comments will be held for moderation. Your contact information will not be made public.

Related content

Sign up to our newsletter