Achieving Equity in Health through Community-based Health

This article was downloaded by: [Aradhna Aggarwal] On: 16 November 2011, At: 20:42 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Development Studies Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/fjds20

Achieving Equity in Health through Community-based Health Insurance: India's Experience with a Large CBHI Programme Aradhna Aggarwal

a

a

Department of Business Economics, South Campus, University of Delhi, India Available online: 16 Nov 2011

To cite this article: Aradhna Aggarwal (2011): Achieving Equity in Health through Community-based Health Insurance: India's Experience with a Large CBHI Programme, Journal of Development Studies, 47:11, 1657-1676 To link to this article: http://dx.doi.org/10.1080/00220388.2011.609586

PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.tandfonline.com/page/termsand-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sublicensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages

Downloaded by [Aradhna Aggarwal] at 20:42 16 November 2011

whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

Journal of Development Studies, Vol. 47, No. 11, 1657–1676, November 2011


Achieving Equity in Health through Community-based Health Insurance: India’s Experience with a Large CBHI Programme ARADHNA AGGARWAL Department of Business Economics, South Campus, University of Delhi, India

Final version received 8 April 2011

ABSTRACT This article analyses equity in enrolment, renewal of enrolment, and utilisation of community-based health insurance with special reference to the Yeshasvini health care programme. The analysis employs a primary survey conducted in rural Karnataka using a random sample of 4109 households. The study identifies quantifiable variables covering various dimensions of vulnerability and assesses their relationship with enrolment, renewal of enrolment, and utilisation using logistic regression techniques. The results demonstrate that inequities do exist even though they are less pronounced in utilisation than in enrolments and renewals. While community-based health insurance (CBHI) may be used as a mechanism to reach the disadvantaged population, they can not be considered as substitute for government-created health infrastructure.

1. The Study The new international thinking on health that emerged during the late 1990s underlines the notion that the governments in developing countries should promote and strengthen community-based health insurance (CBHI) programmes1 as a viable option in providing financial protection to the poor (WHO, 2000, 2001).2 As a result, governments in many developing countries have taken initiatives to strengthen these programmes as part of their health policy. This has led to a proliferation of these programmes across developing countries (ILO, 2005; Tabor, 2005 for survey). There is evidence albeit weak that such programmes help to improve financial access, utilisation, and quality of health care services through cooperative and community efforts (Jakab and Krishnan, 2004; Ekman, 2004; Radermacher et al., 2009). However, there are gaps in knowledge in determining how equitable such programmes are. While examining the issue of equity, most studies focus on the Correspondence Address: Aradhna Aggarwal, Department of Business Economics, South Campus, University of Delhi, Benito Juarez Road, Delhi-110021, India. Email: [email protected] ISSN 0022-0388 Print/1743-9140 Online/11/111657-20 ª 2011 Taylor & Francis http://dx.doi.org/10.1080/00220388.2011.609586


1658 A. Aggarwal economic composition of current members of these programmes. Generally it is believed that for equity reasons, membership should not be biased towards the better off and be effectively open to economically vulnerable groups. However, equity aspects of a programme cannot be assessed in terms of its ability to address economic vulnerability alone. In the recent literature, multi-dimensional measures of vulnerability are emerging. In a broader sense, vulnerability is defined (SchmidtThome´ and Jarva, 2003) as a set of conditions and processes resulting from physical, social, economic, locational and environmental factors, which determine the susceptibility of an individual/community to the impact of catastrophic events (deterioration of health in the present context). Thus the narrow focus on vulnerability arising out of economic factors alone cannot fully capture the notion of vulnerability. Further, most studies have focused on the determinants of enrolment; few have examined the characteristics of those who have been renewing registration or have actually been the claimants of the benefits. The present study addresses these gaps in the literature. The main aim of the study is to examine, using a broad based concept of vulnerability, equity in enrolment, renewal of membership, and utilisation of services in the context of one of the largest CBHI programmes in India: Yeshasvini Health Care programme. The programme offers the rural poor an opportunity to take benefits of advanced and highly expensive surgical treatments which otherwise would be non accessible to them. Evaluations have demonstrated that the programme has contributed to improving access and quality of care (Aggarwal, 2010). This study examines whether the programme reaches out to the most vulnerable sections of the rural poor. The multidimensional concept of vulnerability used in the study covers: . . . . .

social vulnerability, gender-related vulnerability, health vulnerability, economic vulnerability, and location specific vulnerability

The rest of the article is planned as follows. Section 2 proposes a theoretical framework for analysing the impact of CBHI on health equity. Section 3 focuses on the Yeshasvini health insurance programme. It describes the core characteristics of the programme and discusses their implications for equity. Section 4 describes the research methodology and database while Section 5 discusses the empirical results. Finally, Section 6 concludes the analysis.

2. CBHI and Health Equity: A Theoretical Framework People purchase health insurance if the utility of the expected benefits of coverage in the form of expected covered expenditures plus the value of protection from financial risk, exceed the premium. Algebraically, let Un be the utility from not obtaining insurance, such that: Un ¼ Un ðYn ; M; Tw Þ þ en

India’s Experience with a Large CBHI Programme 1659 where, Yn is income in the uninsured state, M are out-of-pocket medical expenses when uninsured (assumed to be zero when insured), Tw is a shift variable capturing the addition to utility from non-participation for individuals with weak preferences for coverage, and en is a stochastic error term. Similarly, let Ui be the utility from insurance:


Ui ¼ Ui ðYi ; P; Ts Þ þ ei It depends upon income in the insured state, premium P, a shift factor, Ts, accounting for the gain in utility from enrolling in health insurance, and error term ei. Out-of-pocket medical expenses are assumed to be zero in the insured state. An individual will enroll if Ui(Yi,P; Ts) – Un (Yn, M; Tw) 4 en – ei. The difference is a linear function of Y, P, M and the difference in tastes for coverage, Ts, and Tw. Thus: Enrolment ¼ fðY; P; M; Pf Þ In the model, Y captures economic factors such as income, wealth, and income sources while M is a function of health related factors represented by age, family health background, sanitation conditions, and natural conditions in surrounding areas. Pf is the difference between Ts and Tw (Pf ¼ Ts 7 Tw) and denotes individual preference. Individual preference is affected by socio-cultural and location specific factors. Socio-cultural factors cover social status, education, access to information, gender, and family size. Location specific factors on the other hand capture health and transport infrastructure, presence of cooperative societies and the quality of governance. Thus, our model is: Enrolment ¼ fðEð:Þ; Hð:Þ; Sð:Þ; Lð:Þ; Zð:ÞÞ

ð1Þ

In the model E, H, S, and L represent economic, health, social and locational factors, respectively while Z represents control variables. The same set of factors is also likely to influence renewals and the utilisation of health care services. Besley (1989) argues that just as demand for health services is derived from demand for health, demand for health insurance is derived from demand for health services. Thus the same set of factors is expected to affect the utilisation of health care also. Theories of optimal risk taking behavior suggest that households which are more likely to face vulnerability exhibit a higher coefficient of risk aversion (Lin 2009) and hence greater preference for insurance. Thus, in principle insurance demand should be negatively related with economic well being, health and social status, and location-specific well being. This would mean that insurance would always be equity enhancing. Empirical evidence however suggests that commercial health insurance is positively related with these factors (see Bhat and Jain, 2006 for discussion). It excludes the vulnerable population from insurance coverage and hence exacerbates health inequities, further. In this context, it is usually argued that the community involvement in the financing of health care can extend social protection to the most vulnerable sections that would otherwise have no financial protection against the cost of illness in the developing countries (Tabor, 2005). CBHI in particular a large


1660 A. Aggarwal successful programme removes obstacles to obtaining health insurance coverage for the vulnerable population and hence is likely to promote equity in health. Theoretical explanations for the success of CBHI in reaching out to the poorest are set within the fold of the social capital framework (Atim, 1999; Jakab and Krishnan, 2004; Dror and Preker, 2002; Criel and Waelkens, 2003; Hsiao, 2001; Jowett, 2003, 2008; Kiwanuka-Mukiibi et al., 2005; Meessen et al, 2002; Ron, 1999; Schneider, 2004; Zhang et al., 2006).3 Low-income households are generally benefited by informal risk protection through family and relatives, community links, institutional links or societal links. These social and community links, if institutionalised through insurance programmes, may serve as social capital and can effectively play an important role in promoting social welfare (Dror and Preker, 2002: 48). Social connections as well as trust and local community control over these programmes overcome the informational disadvantages and high transaction costs involved in providing insurance to a low income population. This in turn improves willingness to pay even by the poorest of the poor. Further, communities are set within highly differentiated economic, cultural, demographic, political and epidemiological contexts. The technical designs of micro insurance programmes are influenced by local contexts and offer the parties considerable flexibility to negotiate a contract, reducing transaction costs and scaling up participation (Churchill, 2006). Finally, given their close knit character, CBHI programmes are likely to possess sufficient information to detect any deviation by its members from any contracted level of consumption. This can considerably reduce the problem of moral hazard which affects the willingness of the poorest to pay for insurance by reducing the attractiveness of the service package and inflating costs (Lahkar and Sundaram-Stukel, 2010). Community level monitoring of service providers can achieve success in preventing producers’ moral hazard also. Thus, social bonding, solidarity, trust, intra- and extra-community networks, vertical civil society links and state-society relations at the local level are important factors in explaining the success of CBHI in reaching out to the poorest. Theoretically, therefore, CBHI may be an effective mechanism for ensuring equitable distribution of health protection among the rural population in the resource-poor settings where governments have limited financial and institutional capacity and formal mechanisms of social protection for vulnerable populations are missing. Despite the strong theoretical arguments for a positive equity impact of CBHI programmes, existing empirical evidence is ambiguous. In their review of literature on equity, Preker et al. (2002), Ekman (2004), Jakab and Krishnan (2004) and Radermacher et al. (2009) find that while CBHI is effective in reaching the low income group people, the poorest and socially excluded groups are not automatically included. This exclusion effect has been observed among others by Jutting (2004) for Senegal, and Ranson et al. (2006, 2007) and Sinha et al. (2006) for India in the context of the selected CBHI programmes. In contrast, Onwujekwe et al. (2009) and Polonsky et al. (2009) find strong evidence of equity in a successful large CBHI programme in Nigeria and Oxfam’s CBHI programmes in Armenia, respectively. In a qualitative investigation of demand for health insurance in rural West Africa, De Allegri et al. (2006) suggest that the equity aspect of a CBHI programme largely depends on its technical design. The most important reasons for inequity (Jakab and Krishnan, 2004; Criel and Waelkens, 2003) are believed to be the inability to afford


India’s Experience with a Large CBHI Programme 1661 the premiums and poor accessibility to hospitals providing services under these programmes. Onwujekwe et al. (2009) attribute inequity to implementation strategies, insufficient community involvement in the programme planning, lack of trust in the programme or its managers or voluntary membership strategy while Lahkar and Sundaram-Stukel (2010) find moral hazards and high administrative cost as important factors which inflate the cost and discourage the poorest from enrolling. Preker et al. (2002) in their survey of literature conclude that trained and competent management with strong involvement and ownership of the community contribute to the objective of inclusion. Apparently, the equity effects of CBHI programme are expected to vary across countries, CBHI types and designs; and are subject to empirical analysis. The present study focuses on the ‘Yeshasvini health care programme’ in India and empirically analyses its equity dimensions using Model (1).

3. Core Characteristics of Yeshasvini and Their Implications for Equity The programme was introduced in June 2003 in an Indian state of Karnataka as a cooperative venture between the public, private and cooperative sectors to insure the rural poor against surgical procedures. It targets the rural population organised in cooperative societies. The programme is marked by both good and bad practices (Kuruvilla and Liu, 2007; Radermacher et al., 2005; ILO, 2006). While good practices are likely to promote equity in health bad practices can make it inequitable. In what follows we discuss both good and bad practices and propose hypotheses for empirical testing. Good Practices Broad based administrative set up. The administrative set up of Yeshasvini involves efficient partnership arrangements between the government, private and cooperative sectors to exploit their respective strengths to deliver health to the targeted disadvantaged group. While the programme is governed by an independent charitable trust: ‘Yeshasvini Cooperative Farmers’ Health Care Trust’,4 it is being run under the auspices of the Department of Cooperation (DOC). The vast administrative infrastructure of the DOC has facilitated the collection of revenue at no additional cost. Its involvement ensures government backing of the programme and proves to be an important trust building factor for the programme. Involvement of the DOC also provides the managers of the programme access to the cooperative network which assists in mobilising membership and implementing the programme at the grass-roots level. Further, the programme benefits from the societal capital generated by a vast network of Cooperative Societies5 in Karnataka which organises diverse rural farmers and other informal sector rural workers in a strong institutional framework, and acts as a communication channel between the government and the rural population. These societies not only mobilise membership but also help members in seeking treatment from the network hospitals. Finally, the designated health care providers are mainly private hospitals, although charitable, public sector and cooperative sector hospitals are also participating in the programme. Some are

1662 A. Aggarwal


super specialty hospitals. In 2008–2009, the number of network hospitals was 349. These were spread across all 27 districts of the state. Thus the administrative structure of Yeshasvini has been designed to tap the government for its vast administrative machinery to mobilise membership and monitor the programme without creating additional administrative infrastructure; the private sector for quality health care services; and the cooperative sector network for the societal capital generated over the years through business linkages between these societies and their members. Low contribution. Initially, the premium was fixed at Rs (rupees) 60 per person per year but was subsequently raised to Rs 120 per person per year. Recently, it has been further raised to Rs 130. A rebate of 15 per cent is offered on contribution for families of five or more members. At the current level of premium, financial sustainability is not achievable even with a vast membership base. This is because the programme covers high end medical treatment. In order to augment its resources, therefore the Trust accepts donations from private and government bodies, collects 3 per cent of the profits from the profitable societies as donations and most importantly, receives government subsidy on an annual basis which is almost 42 per cent of revenues.6 The management has also created a contingency fund which is invested in interest yielding assets to earn interest income. Long enrolment schedules. Enrolment schedules are spread over five months namely, January to May. These months are also the harvesting time of cash crops such as cotton and sugarcane which make it easier for farmers to make the payment. Flexible mode of payment. The mode of payment is also flexible. It is decided by the local cooperative societies depending on the local conditions. Some societies accept monthly payments during the enrolment period while others demand lump sum payment. Recognising that an annual subscription may present an obstacle to membership, some societies accept bimonthly payment also. Credit cooperative societies generally deduct the subscription amount with the consent of the member while lending money. Attractive service package. The benefit package is well defined. It focuses on surgical procedures the cost of which could be catastrophic for the poor households. This makes moral hazard on the clients’ side unlikely as one can assume that few seek surgery for the sake of surgery (Radermacher et al., 2005). Free out patient department (OPD) consultations and diagnostic laboratory tests at concessional rates are optional. The maximum coverage per person per year amounts to Rs. 200,000 with free OPD. The benefits are reviewed from time to time and appropriate changes are introduced in the package depending on the demand. For instance, recently, normal deliveries and medical emergencies such as snake bite, bull gore and dog bites are also included in the package, keeping in view the growing demand for such coverage. Cashless transactions. Even while the programme is implemented by a third party administrator and not by an insurance company, it offers the feature of cashless

India’s Experience with a Large CBHI Programme 1663 transactions. Patients are not involved in any administrative process. Since the literacy rate is rather low among the most vulnerable population in rural areas, any programme that involves paperwork is likely confined to the better off population. This is therefore an important equity-enhancing feature of the programme.


High quality of network hospitals. A hospital willing to join the Network can apply if it has 25 or more beds. It is given a self-assessment form to assess itself. Each medical facility provided by the hospital is assigned specific marks in this form. A minimum of 88 marks is to be scored to be a part of the network. Hospitals claims are verified through inspections by the Trust before making the final decision on its enrolment. Monitoring of network hospitals. The third party administrator undertakes regular inspections of network-hospitals to monitor the quality of their services, and to ensure that the latter remain true to their commitments made to the Trust. If they are found defaulting, strict action is taken against them including a permanent ban on their Yeshasvini membership. The above characteristics have translated into the success of the programme in terms of enrolment and benefits generated at the macro level. The programme has effectively created a large membership base. In the first year of the programme itself 1.6 million cooperative members enrolled with the programme. In 2008–2009, enrolment increased to 3.0 million which marked a 29.3 per cent increase over the increased base of the last year. In terms of benefits, there has been a rapid increase in the number of surgery cases. In absolute terms, in 2007–2008, a total of 60,668 surgeries took place. In 2009, during April–October alone 23,883 surgeries had been performed under the programme. A wide range of surgical processes are being offered by the programme. These include: heart, obstetrics and gynaecology (OBG), ophthalmology, orthopaedics, ENT (Ear, Nose and Threat), uro- and general surgeries. Heart surgeries accounted for almost 20 per cent of the total surgeries performed over the period of three years: 2003–2004 to 2006–2007, followed by general and OBG surgeries. In the initial years, the distribution of beneficiaries was highly concentrated in five of the 27 districts: Mandya, Kolar, Hassan, Bangalore and Davangere. For instance, in 2004–2005 these districts accounted for almost half the beneficiaries. Within a short span of time, however, there has been a considerable change in the distribution of beneficiaries by location. The Hirschman Herfindahl (HH) index of spatial concentration of claimants was 670 in 2004–2005; it fell to 323.3 in 2006–2007. Further, in 2004–2005, 148 hospitals registered 17,407 surgeries under Yeshasvini. Of this, 1700 (9.81%) were performed in Narayana Hrudayala hospital alone. In 2006– 2007, the number of hospitals registering Yeshasvini patients increased to 290 and they covered 79 cities across the state. The institutional, organisational and technical features of the programme, and its success in creating a wide membership and effective service provisions are expected to ensure that the programme is equitable. But there are inequitable practices as well which impinge on the growth of the programme. For instance, after seven years the scheme has only reached about half of its stated target of six million subscribers; enrolment has been flat since 2008.7

1664 A. Aggarwal Bad Practices


Open membership. While the programme is designed to help the rural poor access expensive surgeries that they otherwise could not afford, there is no mechanism to ensure that only poor segments of the population join as the programme is open for all cooperative members. As a result, rather affluent members are more likely to make use of the programme. Flat premium. The rich and the poor pay the same insurance premiums, with no regard for age, gender and social status during the process of enrolment. This practice directly contravenes the notion of vertical equity in health care financing and provision because the poor have greater health needs but less money to pay for them than the rich do (Radermacher et al., 2005). Bureaucratic set up. While government involvement has been a critical factor in the success of the programme, it has also introduced hierarchies in the administrative set up. As a result, the community control over the programme is weakened and direct communication between the managers and members is missing. Lack of information flows. The programme has not developed appropriate mechanisms to ensure information flows. Many clients are not aware of the details of the benefit package and how to obtain them. Further, frequent changes are introduced in the policy from time to time. Even the network hospitals’ status keeps changing. They may be dropped out for some time or on a permanent basis if they are found to have indulged in frauds. But there are no effective channels to disseminate information on such changes. In the absence of complete information, fears of losing money predominate among the poorest. Exclusions. The programme does not cover inpatient admission without surgery. Further, while surgery procedures are free of cost, implants such as stents, valves, pacemakers, coils or devices come at a cost. The list of exclusions indicates that a considerable risk of high health costs is still not covered and needs to be borne by the patient. Further, while a well specified list of exclusions is provided to all the hospitals, there is little awareness about them among the poorest. They demand services which are not included in the package and once they do not get them, they get disillusioned with the programme. Since the marginal utility of money is rather high for them, they feel that they do not get good value for their money. Long waits. Cashless health care services provided by the programme require preauthorisation for surgery which can take a long period to be issued (Radermacher et al., 2005). This constitutes a burden to poor clients as they might need to travel to the hospital several times or face the (opportunity) costs of waiting. Adverse selection. There is no screening mechanism at the time of subscription and no pre-existing illness is excluded. Therefore the possibility of adverse selection is rather high. This can escalate the cost of the programme and result in tight

India’s Experience with a Large CBHI Programme 1665 monitoring of financial reimbursements to the service providers. This in turn can affect the services offered to programme members, particularly the most vulnerable ones due to their ignorance and illiteracy.


Fixed surgery prices. An important feature of the programme is that the price for surgery paid to a network hospital is significantly below the normal market charges and it is not even adjusted to the current rate of inflation. This is a source of dissatisfaction among service providers and is manifested sometimes in an unfriendly attitude towards Yeshasvini patients discouraging in particular the least well off who are illiterate and do not have access to correct information. Location of network hospitals. As the network hospitals are usually big and with modern health facilities, they are mainly located in and around better off areas. Clients who stay in far flung areas have to travel long distances. Thus any dissatisfaction and word-of-mouth adverse publicity affects their enrolment, utilisation and renewal of membership. Poor monitoring. Although periodic ‘quality audits’ are conducted they tend to focus on structural conditions and not other areas of quality. According to a reliable source8 disempanelment ratio is a mere 0.8 per cent which is rather low and reflects on the quality of these audits. Further, even though empanelled hospitals are expected to provide free OPD, there are no mechanisms to monitor that they actually follow this practice. Finally, there is no system whereby the Trust can monitor the functioning of the third party administrator. In a nutshell, the programme seems to have several innovative features which are likely to have positive effects on equity but at the same time, there remain several challenges as well. It is at best an open empirical question whether the programme promotes equity in health.

4. Empirical Estimation and Database For the empirical testing of equity effects of the programme we estimate Model (1). For this, we express both dependent and independent variables in terms of quantifiable variables, as under: Dependent: Enrolment ¼ 1, if a at least one member of the household is enrolled with the programme ¼ 0, otherwise; Renewal ¼ 1, if at least one member of the household has been member of the programme for the past three years ¼ 0, otherwise; Claimant/Beneficiary ¼ 1, if at least one member of the household has claimed benefit in the past three years. ¼ 0, otherwise.

1666 A. Aggarwal Independent: Economic factors. variables:

Economic factors are captured by the following income related

Annual income (annual_inc.): Annual household income


Concen_inc: HH index of concentration of income sources using data on percent income generated from different occupations. Household wealth (hh_wealth): Wealth indices constructed from household possessions and amenities and dwelling characteristics, using principal component analysis (PCA) Health related factors. The likelihood of a health shock may vary systematically with observable characteristics, such as age, prevalence of chronic health problems in the family, and ecological and sanitation conditions in the surroundings in which households live. Age is captured by two variables: Headage: Age of the head of the family Age_dividend: Proportion of household members in the working age group. The health status of the family which is an important determinant of health vulnerability is represented by a dummy: chron_health. chron_health: ¼ 1, if at least one member is suffering from chronic health problem. ¼ 0, otherwise. Ecological and sanitation conditions are proxied by: V_wat_san: Index of living conditions constructed from water and sanitation conditions based on principle component analysis and, V_Naturalcdn: Number of beneficiaries of natural disaster as percentage of total population at village level. Social and cultural factors.

These are captured by the following variables:

Sh_female: Share of female members in the household SC_group: ¼ 1, if belongs to schedule caste/schedule tribe9 (SC/ST) ¼ 0, otherwise Aveduyears: Average education years of the household Access_TV: Regularity of watching TV on a likert scale of 1 to 4 Access_Paper: Regularity of reading newspaper on a likert scale of 1 to 4 Membershg: ¼ 1, if member of self-help groups (SHGs) ¼ 0, otherwise.

India’s Experience with a Large CBHI Programme 1667 Locational specific variables. Village and district specific health attributes include health infrastructure, distance from the nearest health facility, and distance from the nearest Yeshasvini facility, frequency of natural calamities, and water and sanitation conditions.


V-hlthinfra: Index of government health facilities constructed from the number and quality of health facilities using the principal component analysis (PCA) technique V-hlthdistance: Distance from the nearest health facility Y-dist: Distance of the nearest Yeshasvini facility D_healthinfra: Index of the quality of district-level health infrastructure D-tpt: PCA based index of district level transport facilities. Control variables: These variables are as under. Size: Household size D_R_f_rat: Rural female literacy rates D_f_m_gp: Female members in gram panchayats as ratio of total members V_copop: Cooperative societies per capita D_pcy: Per capita income at the district level. In general, the model is specified as: Z ¼ bX þ U where U is a stochastic error term. The model estimated is of discrete choice type, which explains the probability that a person chooses to enroll, renew membership and utilise the programme benefits in alternative specifications. Logit regression methods are employed for estimation of the specified functions. The binary logistic regression model is stated in terms of the probability that Y ¼ l given X: PðY ¼ 1jXÞ ¼

1 : 1 þ exp ðXbÞ

The probability that the household is member/claimant is given by: Probi ¼ Pr½P0i ¼ 1nXi ¼ Pr½Ui < b1Xi ¼ y½1 b1Xi Where X ¼ a vector containing all the independent variables. Data Secondary and primary data sources are used to create a database for the study. The database contains three levels of hierarchy: district level, village level and the household level. Household data. Primary data were collected through a fully structured household questionnaire. It was administered to a random sample of 4109 households across 82

1668 A. Aggarwal villages drawn from a sample of 16 districts in Karnataka. A multi-stage stratified design was adopted for deriving the coverage. The survey was conducted during the months of December 2007 to May 2008.


Village level data. The village level information was collected through primary and secondary sources. The ‘Department of Rural Development and Panchayat Raj’ maintains a vast information system on the basic amenities available in the villages of Karnataka. We compiled this information for 82 sample villages. This was supplemented with the primary data collected from these villages. District level information. The district level information was collected from various government departments. These included, the ‘Directorate of Economics and Statistics’, the ‘Planning and Statistics Department’ and the office of the ‘Registrar of Cooperative Societies’ of the Government of Karnataka.

5. Empirical Findings Our household sample comprised of two broad groups of households: cooperative households (CH); and non-cooperative households (NCH). A pertinent question was whether the representative sample should comprise of only the eligible (cooperative) households or the full sample. The argument in favour of the former is that the programme covers only cooperative households and therefore the impact of the factors affecting membership can be more meaningfully captured by the regression based on the censored sample. However, the participation in cooperative societies itself is voluntary, and therefore the possibility of unobservable self-selection bias could not be ruled out. Any regression based only on the sample of cooperative members will not capture the systematic difference between the two groups. If the regression is based only on the cooperative households then it is necessary to adjust it for the selection bias. Since we do not have any idea of the bias involved, we used both, the censored and the full sample in alternative specifications. The logit model estimations show a relatively good fit of the model, expressed by Chi-squared statistics in both the specifications (Table 1). The results are also found to be consistent and stable. This shows that the results are robust and predictive. In what follows, we discuss the results. Economic equity. Enrolment and renewal of enrolment, both are found to be inequitable in economic terms. The variable hholdasset is significant with a positive sign in all the specification of the logit model. There is thus clear evidence that enrolment remains disproportionately higher in favour of the wealthier classes even though income and concentration of income sources do not appear significantly different from zero.10 Interestingly, both income and wealth related factors turned insignificant in the utilisation equation. This means that there are no significant differences in the economic status of beneficiaries and non-beneficiaries. Apparently, while a well-off household is more likely to become a member, a beneficiary is not necessarily economically better off than his non-Yeshasvini member counterpart seeking

75.66E 7 06 (70.49) 0.041919 (0.97)

0.205351a (2.69) 0.002079 (71.08)

Enrolment

7 0.14818 (70.57) sc_grp 7 0.7004a (76.11) aveduyears 0.141982a (3.72) access_paper 7 0.08404b (72.26) access_tv 7 0.0767c (71.76) membershg 0.157691b (2.39) Economic vulnerability Annual_inc 0.092432 1.23 hh_wealth 0.23584a (3.14)

Social vulnerability sh_female

V_naturalcdn

V_wat_san

headage

Age-dividend

Health vulnerability Chron_health

Variable

0.018317 0.21 0.263417a (3.32)

0.203141 (0.69) 7 0.61464a (73.99) 0.037442b (2.28) 7 0.09165b (72.02) 7 0.06488 (71.18) 0.065381 (0.88)

2.45E-05c (1.78) 0.082117c (1.88)

0.28371a (3.5) 70.00029 (70.12)

Renewal of membership

Sample of cooperative households

7 0.01091 70.06

7 0.16761 (70.95)

1.398251c (1.74) 7 1.13899a (72.53) 0.035622 (0.88)

0.035462a (2.92) 7 0.26624 (70.99)

0.240782 (1.26)

Utilisation

0.1879 2.7 0.336971a (5.00)

7 0.28732 (71.27) 7 0.75175a (77.24) 0.0823a (6.11) 7 0.10225a (73.08) 7 0.11038a (72.81) 0.171098a 2.89

7 6.83E-06 (70.67) 0.06901c (1.77)

0.213358a (3.1) 0.001 (0.57)

Enrolment

0.152372c 1.82 0.32368a (4.29)

0.093576 (0.34) 7 0.69427a (74.66) 0.049656a (2.99) 7 0.10597b (72.41) 7 0.09833c (71.84) 0.087044 (1.22)

2.12E-05c (1.67) 0.103093c (2.5)

0.290039a (3.68) 0.000356 (0.16)


Full sample

Table 1. Logit estimations of the probability of enrolling, renewing and utilising the programme


(continued)

0.046913 0.24

7 0.15119 (70.84)

1.277904c (1.71) 7 1.18738a (72.68) 0.068074c (1.7)

0.036535a (3.27) 7 0.32946 (71.29)

0.286547 (1.55)

Utilisation

India’s Experience with a Large CBHI Programme 1669

Enrolment

7 0.25307 (70.37) 5.778508 (0.28) 7 0.69997 (70.92) 229.56 2756

0.024448 (71.35)

7 0.05231c (71.71) 0.152201a (3.11)

7 0.26049 (70.33) 102.317a (4.49) 7 2.28857a (72.6) 177.78 2756

0.037321c (1.88) 4.78E-05a (3.66)

7 0.06052c (71.67) 0.134603a (2.55)

116.0558c (1.69) 7 4.0531c (71.89) 36.14 395

4.65E-05 (0.96) 0.182434c (1.83)

0.091721 (0.6)

c

1.049763c (1.68) 30.74835c (1.67) 7 2.69936a (73.98) 444.16 3772

0.0201 1.15 7 5.84E-06 (70.67)

7 0.08136a (72.84) 0.020483 (0.49)

7 0.15594a (73.08)

7 0.25207 (71.39) 7 0.08953 (70.77) 7 0.00782c (71.87)

7 0.12809c (71.87)

Enrolment 7 3.50E-05 (71.71)

Utilisation

7 1.70E-05 (70.63)


Notes: aSignificant at 1 per cent; bsignificant at 5 per cent; csignificant at 10 per cent.

LR statistics NOB

_cons

V_copop

D_panchay_

D_f_mem_gp

D_pcy

Control variables hsize

D_tpt

D_health_infra

Yesh_dis

7 2.70E-05 (71.17) Location specific vulnerability V_hlthdistance 7 0.03196 (70.55) V_hlthinfra

Concen_income

Variable

Sample of cooperative households

Table 1. (Continued) Full sample

0.541192 (0.7) 114.3783a (5.21) 7 3.51732a (74.17) 296.01 3772

0.036624c 1.82 4.25E-05a (3.61)

7 0.08004b (72.23) 0.067954 (1.35)

7 0.2102a (73.26)

7 2.60E-05 (71.00)



167.355a (2.62) 7 5.28689b (72.47) 49.57 483

5.11E-05 (1.13) 0.203465b (2.14)

0.123493 (0.86)

7 0.3198c (71.82) 7 0.10007 (70.89) 7 0.00699c (71.74)

Utilisation

1670 A. Aggarwal



surgery. Thus benefits of the programme do not necessarily go to wealthier households within the programme. Health equity. The variable chron_health turns significant in all the equations of enrolment and renewal. This implies that the health status of the family is an important driver of enrolment and renewal. However, it is not a significant determinant of utilisation of health services. Health service utilisation appears to increase with age, a commonly reported finding. As age increases, there is increasing likelihood of experiencing an episode of ill-health, and this greater health need accounts for the increased levels of utilisation among individuals. Poor water and sanitation conditions impose a serious health risk. Our findings suggest that the Yeshasvini programme does not effectively address this health vulnerability. Renewals are negatively related with this variable indicating that the programme has not penetrated into high risk villages. After controlling for other effects however, households living in villages susceptible to natural disasters are more likely to enrol. Thus the programme appears to effectively address health vulnerability. Gender equity. Our logistic regressions reveal that the probability of enrolling into the CBHI programme is gender neutral. The variable Sh_Female turns insignificant in all the specifications of enrolment (including renewals). This indicates that the programme offers equal access to health care services to all irrespective of the gender. Interestingly however, the variable turns positive and significant in the estimation of beneficiaries’ function indicating that the women are more likely to be the beneficiaries of the programme. The programme has gender positive effects in terms of the distribution of benefits. Evidence suggests that in developing countries females are more susceptible to illness than men due to their social status. Nutritional status of women and girls is compromised by unequal access to food, by heavy work demands, and by special nutritional needs (such as for iron). Furthermore, they are often trapped in a cycle of ill health exacerbated by childbearing, exposure to heavy smoke from kitchen fires, and hard physical labour, especially in agricultural areas. Despite their serious health problems they do not get adequate levels of preventive care. The consequences of women’s unfavourable status in the inherently inequitable social system prevalent in most developing countries include discrimination in the allocation of household resources, such as food, health care and education. Women’s health and nutritional status is thus inextricably bound up with social, cultural, and economic factors that influence all aspects of their lives. Therefore, women comprise the key stakeholders in health policy debates and are the most vulnerable section of society. The focus of any social health insurance programme should therefore be to ensure that the programme provided is just as attractive to women as it is to the men and that it is effective in attracting their participation and providing health access. Our results show that the Yeshasvini health care programme offers an opportunity for the rural women to access quality health care services at low contribution. Other dimensions of social equity. While the programme is equitable in terms of female participation, it is inequitable among different social groups. After


1672 A. Aggarwal controlling for other household and location specific characteristics, H_SC turned significant with a negative sign in all the specifications of enrolment, renewal of enrolment and utilisation. This implies that there is a problem of social exclusion in enrolment and utilisation. The programme enrolment is biased in favour of the empowered classes of the society. Since independence, there has been political and economic mobilisation of SCs in India. However, this mobilisation seems to have had little impact on the social status of this group of population. Education, access to information (access_tv, access-paper) and membership of self-help groups, are found to be empowering factors that increase the likelihood of joining the programme and renewing membership. Thus the population groups that suffer from social vulnerabilities in terms of lack of education, access to information and community groups have a lower probability of joining the programme. The programme is socially inequitable in that sense. Utilisation of the programme however is influenced only by the levels of education; other variables are not significant. Thus, social inequities are less skewed in terms of the distribution of benefits. Locational equity. Both enrolment and renewal of membership are negatively related with D_hlthinfra. This implies that the programme is doing well in reaching out to the rural population in the areas where the quality of government health facilities is poor. However, the probability of enrolment and renewal is negatively related with distance from a health facility and positively related with transport facilities. By disproportionately favouring the people in well connected villages and districts it exacerbates inequities. Thus the net effects of the enrolment into the programme on locational vulnerabilities remain uncertain. In the utilisation equation however the only variable that is consistently significant with a negative sign is distance from the Yeshasvini facility. All other variables are insignificant. Locational vulnerabilities except the distance from the Yeshasvini facility do not seem to affect the utilisation of the programme. Apparently, the most important contribution of the programme is the empowerment of females. Further, the programme has also successfully extended the coverage to those who are susceptible to health shocks. However, the poorest people living in far flung areas with little connectivity and access to health infrastructure still remain excluded. Despite low premiums and other attractive features the most vulnerable did not participate in the programme. Inequities are less prounced in utilisation indicating that the benefits of the programme are more equitably distributed. Though this provides evidence of empowerment, equity can not develop until the most vulnerable group of population enrols. Among the control variables, the density of cooperative societies emerges significant in all the specifications of enrolment, renewal and utilisation. Another interesting variable that needs attention is the proportion of females in village panachayats. It is significant with a positive sign in the equation for utilisation. It shows that women representatives in local governance can play an effective role in promoting health care utilisation in rural areas. This finding needs further explorations.



6. Conclusion This study analyses whether micro health insurance programmes can serve as an effective instrument in providing health security to the most vulnerable sections of the society. More specifically it examines how equitable is enrolment, renewal, and utilisation of community-based health insurance? While doing so, it focuses on one of the largest and the most innovative CBHI programmes in India: the Yeshasvini health care programme. Data was collected using a questionnaire that was administered to 4109 respondents in rural Karnataka selected by stratified random sampling. Our results based on logit models reveal that inequities do exist. However, they are less pronounced in the distribution of benefits than in enrolments and renewals. Clearly, the programme has the potential to increase access of the most vulnerable groups of the population to health care and provide them financial protection (Aggarwal, 2010), but it remains unexploited if the poorest are not included. This calls for concerted efforts to enrol the most vulnerable sections of population with the programme. For this, managers of the programme may introduce a differential pricing strategy to subsidise the poor. Alternatively, differential pricing may be accompanied by differential service packages to charge premium prices from the better off sections. Further, all out efforts need to be made to increase the number of enrollees, so as to increase the pool of funds and risks. Well designed strategies should be adopted to increase levels of trust by improving community participation in the programme management and developing effective information channels. Besides, additional health checks may be organised for Yeshasvini members on an annual basis as an incentive to enrol. The findings also suggest that the network of service providers needs to be extended to locationally disadvantaged areas to ensure service delivery on time. Although the level of concentration of beneficiaries by location has decreased over time, 35 per cent of utilisation occurs in just 3.7 per cent of facilities, which are concentrated in larger cities.11 This may be an impediment to utilisation by lower income beneficiaries living in distant areas. Most importantly however our results show that CBHI programmes are useful in themselves, but no substitute for hospitals and hospital networks that can respond to the needs of the most vulnerable sections of the population. Governments must increase the national budget for health. This is the only proven method for achieving equity in health. The National Rural Health Mission (NRHM) was launched in 2005 by the Government in India to improve the availability of and access to quality health care, especially for the vulnerable population. An early impact evaluation of the programme (Gill, 2009) however indicates that there are many problems in its implementation, so that delivery is far from what it ought to be. These problems need to be overcome with the necessary political will and commitment at all levels.

Acknowledgements This study was funded by the Bill & Melinda Gates Foundation. I would like to thank the Global Development Network (GDN) for giving me an opportunity to


1674 A. Aggarwal carry out this study. I gratefully acknowledge the help of the Karnataka State Department of Cooperation and other State Government departments in providing relevant data and information. The paper was presented at the British Northern University, India Forum (BNUIF) seminar ‘India’s Service Sector: Who does it serve?’ Hyderabad, 24–26 March 2010. I would like to thank the participants and discussants for their useful comments and suggestions. A revised version of the paper was written while I was serving as Visiting Professor at the Institute of Economics and Business Administration, University of Kobe, Japan. I thank the Institute for providing me with the necessary research support. Particular thanks are due to an anonymous reviewer of this journal whose comments were most helpful in further revision of this paper. I would also thank the managing editor of the journal for his useful and constructive suggestions which helped me improve the paper. Notes 1. CBHI programmes are known by different names in different parts of the world, including: microinsurance, community health finance organisations, mutual health insurance programmes, prepayment insurance organisations, voluntary informal sector health insurance, mutual health organisations/associations, community health finance organisations, and community self-financing health organisations. 2. The Commission on Macroeconomics and Health (WHO, 2001: 60), for example, recommended that, ‘out-of-pocket expenditures by poor communities should increasingly be channeled into community financing programmes to help cover the costs of community-based health delivery’. 3. Mladovsky and Mossialos (2008) however use the social capital framework to argue how the same factors may affect the growth of CBHI adversely. 4. The Trust consists of 12 members and is chaired by the Principal Secretary of the Department for Cooperation. Other members include employees of the Karnataka Department of Cooperation; health professionals representing network hospitals; and the Director of the Health Department. 5. There were a total number of 32,900 cooperative societies in Karnataka in 2007. Of them, 28,099 co-operatives were actively working. 6. My thanks are due to the anonymous referee of this article for making this point. 7. I would like to thank the referee for this information. 8. I am grateful to the referee for providing me with this statistic. 9. Scheduled Castes (SC) and Scheduled Tribes (ST) are Indian population groups that are explicitly recognised by the Constitution of India; they were previously called ‘depressed classes’ by the British. 10. The significance of the income variable in the full sample model of enrolment could be capturing the effect of economic status of cooperative members. 11. I thank the anonymous referee for this information

References Aggarwal, A. (2010) Impact evaluation of India’s ‘Yeshasvini’ community-based health insurance programme. Health Economics, 19(S1), pp. 5–35. Atim, C. (1999) Social movements and health insurance: a critical evaluation of voluntary, non-profit insurance programmes with case studies from Ghana and Cameroon. Social Science and Medicine, 48(7), pp. 881–886. Besley, T. (1989) The demand for health care and health insurance. Oxford Review of Economic Policy, 5(1), pp. 21–33. Bhat, R. and Jain, N. (2006) Factoring affecting the demand for health insurance in a micro insurance programme. Working Paper No.2006-07-02. July 2006, Ahmedabad: IIM. Churchill, C. (ed.) (2006) Protecting the Poor: A Microinsurance Compendium (Geneva: ILO).


India’s Experience with a Large CBHI Programme 1675 Criel, B. and Waelkens, M. P. (2003) Declining subscriptions to the Maliando Mutual Health Organisation in Guinea-Conakry (West Africa): what is going wrong? Social Science and Medicine, 57(7), pp. 42–48. De Allegri, M., Sanon, M. and Sauerborn, R. (2006) To enrol or not to enrol? A qualitative investigation of demand for health insurance in rural West Africa. Social Science and Medicine, 62(6), pp. 1520– 1527. Dror, D. M. and Preker, A. S. (eds) (2002) Social Reinsurance: A New Approach to Sustainable Community Health Financing (Washington DC: World Bank and Geneva: ILO). Ekman, B. (2004) Community-based health insurance in low-income countries: a systematic review of the evidence. Health Policy and Planning, 19(5), 249–270. Gill, K. (2009) A primary evaluation of service delivery under the National Rural Health Mission (NRHM): findings from a study in Andhra Pradesh, Uttar Pradesh, Bihar and Rajasthan. Working Paper 1/2009 – PEO, Planning Commission of India, May 2009. Hsiao, W. (2001) Unmet Health Needs of Two Billion: Is Community Financing a Solution? Working Group 3, Background report to Commission on Macroeconomics and Health, 2001. ILO (2005) Community-based schemes: India: an inventory of micro insurance programmes. Working Paper no. 2, International Labour Organisation, Geneva. ILO (2006): India Yeshasvini Cooperative Farmers Health Programme Social Security Extension Initiatives in South Asia, International Labour Organisation, Geneva. Jakab, M. and Krishnan, C. (2004) Review of the strengths and weaknesses of community financing, in: A. S. Preker, and G. Carrin (eds) Health Financing for Poor People: Resource Mobilisation and Risk Sharing (Washington, DC: The World Bank), pp. 53–118. Jowett, M. (2008) Sustainable health financing in low income countries: does social capital offer any insights? Oxford Policy Institute, UK. Accessed at http://www.opi.org.uk/publications/documents/ Socialcapitalandhealthfinancing_JowettApril2008.pdf (downloaded in December 2009). Jowett, M., Contoyannis, P. and Vinh, N. D. (2003) The impact of public voluntary health insurance on private health expenditures in Vietnam. Social Science and Medicine, 56(2), pp. 333–342. Jutting, J. (2004) Do community-based health insurance schemes improve poor people’s access to health care? Evidence from rural Senegal. World Development, 32(2), pp. 273–288. Kiwanuka-Mukiibi, P., Derriennic, Y. and Karungi, G. (2005) The Good Practice Model: Community Participation in Luweero District, Uganda. The Partners for Health Reformplus Project, Bat Associates Inc., Bethesda, MD. Kuruvilla, S. and Liu, M. (2007) Health security for the rural poor? A case study of a health insurance programme for rural farmers and peasants in India. International Social Security Review, 60(4), pp. 3– 21. Lahkar, R. and Sundaram-Stukel, R. (2010) Protecting the poor through community based health insurance. Accessed at http://www.isid.ac.in/*pu/conference/dec_10_conf/Papers/RatulLahkar.pdf (downloaded on 28 March 2011). Lin, F. T. (2009) Does the risk aversion vary with different background risk of households? International Research Journal of Finance and Economics, 34, pp. 69–82. Meessen, B. Criel, B. and Kegels, G. (2002) Formal pooling of health risks in sub-Saharan Africa: reflections on the obstacles encountered. International Social Security Review, 55(2), pp. 71–93. Mladovsky, P. and Mossialos, E. (2008) A conceptual framework for community-based health insurance in low-income countries: social capital and economic development. World Development, 36, pp. 590–607. Onwujekwe, O. et al. (2009) Is community-based health insurance an equitable strategy for paying for healthcare? Experiences from southeast Nigeria. Health Policy, 92(1), pp. 96–102. Polonsky, J. et al. (2009) Equity in community health insurance programmes: evidence and lessons from Armenia. Bulletin of the World Health Organisation, Geneva. Preker, A. S. et al. (2002) Effectiveness of community health financing in meeting the cost of illness. Bulletin of World Health Organisation, 80(2), Geneva. Radermacher, R. et al. (2005) Yeshasvini Trust, Karnataka, India. Series: good and bad practices in microinsurance, CGAP Studies, no. 20, Geneva: ILO. Radermacher, R. et al. (2009) What do we know about the impact of micro insurance?, paper submitted to the Microinsurance Conference 2009 in Dakar. Accessed at http://www.munichre-foundation.org/ NR/rdonlyres/3577DCF9-4C72-4AF6-8526-F9AFA4C852F1/0/S3_MIC09_Paper_Radermacher.pdf (downloaded on 15 March 2011).


1676 A. Aggarwal Ranson, M. K. et al. (2006) Making health insurance work for the poor: learning from the Self-Employed Women’s Association’s community-based health insurance programme. Social Science Medicine, 62(3), pp. 707–720. Ranson, M. K. et al. (2007) Equitable utilisation of Indian community based health insurance programme among its rural membership: cluster randomised controlled trial. British Medical Journal, 334(7607), pp. 1309–1309. Ron, A. (1999) NGOs in community health insurance programmes: examples from Guatemala and the Philippines. Social Science and Medicine, 48(7), pp. 939–950. Schmidt-Thome´ P. and Jarva J. (eds) (2003) ESPON project 1.3.1 the spatial effect and management of natural and technological hazards in general and in relation to climate change, 2nd Interim Report, August 2003, GTK, Espon. Schneider, P. (2004) Why should the poor insure? Theories of decision making in the context of health insurance. Health Policy and Planning, 19(6), pp. 349–355. Sinha, T. et al. (2006) Barriers faced by the poor in benefiting from community-based insurance services: lessons learnt from SEWA Insurance, Gujarat. Health Policy Plan, 21(2), pp. 132–142. Tabor, S. R. (2005) Community based health insurance and social protection policy social protection, Discussion Paper series No. 0503, Washington, DC: World Bank. World Health Organisation (2000) Health systems: improving performance. The World Health Report 2000 (Geneva: World Health Organisation). World Health Organisation (2001) Macroeconomics and health: investing in health for economic development, Report of the Commission on Macroeconomics and Health, Chaired by Jeffrey D. Sachs (Geneva: World Health Organisation). Zhang, L., Wang, H., Wang, L. and Hsiao, W. (2006) Social capital and farmer’s willingness-to-join a newly established community-based health insurance in rural China. Health Policy, 76(2), pp. 233– 242.