INSURING THE POOR

(1)

INSURING THE POOR

– A combined micro health insurance feasibility and demand study for rural Nepal –

Anne de Jong

Anne de Jong (s1622137) Supervisor: Robert Lensink Co-supervisor: Karlijn Morsink

Research Master thesis IE&B (EBM897A30) and Master thesis IB&M (EBM719A25) University of Groningen

17 July 2012

Abstract: This research set out to combine a feasibility of and demand study for a micro health

(2)

ACKNOWLEDGMENTS

(3)

EXECUTIVE SUMMARY

For several decades, microfinance has been a key term in development economics, gaining a substantial body of scientific research, and at the same time appealing to the masses. Currently, many microfinance institutions are expanding their financial aid with additional services, such as savings possibilities, education and training, or microinsurance products, the latter one being the focus point of this thesis.

Microinsurance aims to reduce poor people’s vulnerability to various shocks that have a depreciating effect on their income. Especially poor health is often seen as a vicious circle with regard to poverty: Poverty is a major underlying cause of disease in the developing world, while consequently, falling ill can have disastrous outcomes for one’s income. Next to the evident payment of medical bills, the household has to deal with lost income and lowered productivity as well. Micro health insurance (MHI) attempts, with the payment of small regular premiums, to prevent households from becoming financially distressed once a family member falls ill.

The setting of this thesis is Nepal, and especially Dhulikhel Hospital, which, next to a central hospital with over 300 beds, operates 15 small rurally located ‘health centres’ in order to provide access to healthcare for some of the most remote areas of the country. Dhulikhel Hospital has had a microfinance program in place for 5 years, and is currently examining the possibilities of expanding the program with a MHI scheme. The two specific research questions that will be answered in this thesis are:

 Research question 1: What are the needs and demands for micro health insurance of the rural

Nepalese women participating in Dhulikhel Hospital’s microfinance program, and how can such a health insurance scheme be designed in a sustainable manner? (labeled the ‘feasibility study’)

 Research question 2: What are the factors that influence demand for enrolling in a micro health

insurance scheme? (labeled the ‘demand study’)

There are several delivery models in which MHI schemes can take place. Dhulikhel Hospital pursues to operate a ‘provider model’, in which the hospital will simultaneously be the insurer and the healthcare provider. As with the other delivery models, the provider model has both advantages and flaws, however, in the thesis it is argued that for Dhulikhel Hospital the disadvantages might be less worrisome than for other healthcare providers trying to implement a MHI scheme.

(7)

The demand study, assessing variables which will influence an individual’s demand for MHI, finds 10 out of 13 demand factors stemming from literature to be insignificant predictors of demand for insurance. Explanations for this can be first the small sample with which the coefficients are estimated, and second the weaker measurement of demand for health insurance (‘willingness’ to enroll rather than ‘actual’ enrolment).

The three variables that were able to significantly explain willingness to enroll are household income, household size, and household riskiness. First, higher level of household income is associated with a higher likelihood of becoming a member in the insurance scheme, since these households might be less burdened with paying the insurance premium. Second, larger households are also found to display a higher demand for insurance. This might be due to larger households having higher income in this sample. Lastly, household riskiness proves to be a significant positive indicator for enrolling in the insurance scheme, in essence signaling adverse selection for the scheme by showing that more risky households in terms of illness are more likely to demand health insurance. A model which includes only these three significant contributors is able to correctly classify individuals becoming members or nonmembers of the scheme four out of five times.

With regard to the feasibility study, respondents stated a vast level of willingness to enroll in the scheme, with 77.8% of the respondents indicating that for 50 Nrs per person per month, they would become members. Willingness to pay averaged 72.06 Nrs, which then would include all five healthcare services provided by the health centre: 1) consultations, 2) one day hospitalization, 3) pharmacy services, 4) laboratory facilities, and 5) the safe motherhood program.

A sustainable premium was calculated on the basis of historical utilization rates, and was evaluated in five different scenarios (based on different units of enrollment, co-payment mechanisms, and exclusion of certain services from the coverage package). Evidence from this model suggests that the scenario in which the pharmacy services will be excluded is most suitable, since in this setting the insurance scheme does not suffer from moral hazard behavior of the insurance clients. In this scenario, the individual monthly premium equals 15.54 Nrs.

Other results from the feasibility study imply a monthly premium paid per individual is preferred among its prospective clients. Moreover, mandatory enrollment for all microfinance participants largely is agreed upon.

(8)

1 INTRODUCTION

For several decades, microfinance has been a key term in development economics, gaining a substantial body of scientific research, and at the same time appealing to the masses. Microfinance, with its grassroot philosophy and community-based practices, has even been labeled the solution towards ending world poverty (United Nations, 2003). Ever since in 1976, when Mohammed Yunus granted the first microloans to rural women in Bangladesh (Yunus, 2007), microfinance institutions in a multitude of developing countries have emerged. The Microcredit Summit campaign reported that by the end of 2009 almost 3,600 microcredit institutions were active on a global scale, reaching more than 190 million borrowers (Reed, 2011). Moreover, this vast number of institutions includes only those able to produce complete financial statements. Thus, the total number of institutions, including those which do not report such official statements, potentially is considerably larger.

In essence, microcredit is the provision of collateral free small loans, often focused on women, in order to enable them to develop household based micro enterprises (Hamid et al., 2011a). These loans can be used for numerous purposes, ranging from the purchase of supplies for small businesses to purchasing livestock or crops. Microfinance is the overarching term for all programs in place aimed at breaking the circle of poverty (Armendáriz and Morduch, 2010). Even though for many people the terms are interchangeable, some argue that this shift in terms also gave rise to a focus on commercially-oriented microfinance institutions, rather than solely providing loans to the very poor. Regardless, this thesis will apply the microfinance term, since by default this entails the broader set of activities a (either not-for-profit or commercially-oriented) development organization performs.

Microfinance thus has a much broader focus than merely offering credit to the poor. Many microfinance institutions are expanding their financial aid with additional services, such as savings possibilities, education or training, or microinsurance products. In the literature, these complementary services are coined ‘non-credit tie-ins’ (Smith, 2002), ‘non-financial services’ (Godquin, 2004), or ‘credit-plus activities’ (Biosca et al., 2011). As Smith (2002) notes, such additional services are established due to the conviction that even though providing credit is necessary, it is insufficient.

One of these credit-plus activities is the focus point of this thesis. Microinsurance aims to reduce poor people’s vulnerability to various shocks that have a depreciating effect on their income. There are various types of microinsurance offered to the poor, of which life-, and accidental death and disability insurance are the top two microinsurance products, with 56% and 20% respectively of the total insurance products sold. Health-related products account for a mere 11% of the total market, while contrastingly, a health insurance is demanded most by the poor (Roth et al., 2007).

(9)

two channels. Firstly, improvements in an individual’s health status results in higher productivity and lower workdays lost, both associated with a higher income level. Secondly, due to reduced uncertainty of health expenses an individual might decide to increase his investments in other forms of capital, including high return assets. This in turn might also lead to a higher household income. Even though microfinance institutions offer MHI as a complementary (or sometimes mandatory) service to their clients for only approximately a decade, once one more closely examines the existing literature, it actually appears that from the mid-1980s onwards a stream of literature on this topic has been published (Carrin et al., 2005). Under the label ‘community-based health insurance’, or CBHI for short, numerous studies on offering microinsurance to the poor have been written. As Preker et al. (2001) and Dror and Preker (2002) note, the term ‘community financing’ increasingly has become a generic term used to label and summarize different financing initiatives, such as microinsurance, community health funds, and mutual health organizations. However, all of the mentioned initiatives have different objectives, policies, and most importantly, management. Micro health insurance thus can be covered under CBHI, however it is not a synonym, rather an example of. Micro health insurance in this thesis is defined as ‘the protection, offered on a community-based

level, of low income people against health-related shocks, in exchange for premium payments proportionate to the likelihood and cost of the risk involved’ (adapted from Cohen and Sebstad,

2005:1). The development of CBHI and MHI will be discussed in a later chapter of this thesis.

1.1 Research questions and innovative aspects

This thesis integrates two objectives, and hence also answers two research questions. The first objective is related to a feasibility study performed for Dhulikhel Hospital, which has had a microfinance program in place for five years, and now is examining the possibilities of adding a micro health insurance scheme to it. Please refer to the next section for an elaboration on the context of this study, Dhulikhel Hospital. From here onwards, all writings related to the introduction of this MHI scheme will be referred to as the ‘feasibility study’.

The second objective is trying to shed more light on the factors that influence an individual’s demand for MHI. Existing demand studies report very low enrolment ratios between 10 and 20% of the target population (Kamuzora and Gilson, 2007; Thornton et al., 2010; Jehu-Appiah et al., 2011). By increasing our understanding of variables that influence demand for microinsurance, potentially these take up rates can be increased. The factors examined in the literature to date are individual characteristics, such as age and education, household characteristics, such as family income and household size, and behavioral and financial features, such as trust, credit constraints, and risk aversion. These demand factors will be assessed simultaneously with the feasibility study, and hence will be conducted with regard to Dhulikhel Hospital’s microfinance program as well. The variable of interest that is assessed here is ‘willingness to enroll’ in the MHI scheme. From here onwards, all writings related to these demand factors will be referred to as the ‘demand study’.

The two specific research questions that will be answered in this thesis are:

 Research question 1: What are the needs and demands for micro health insurance of the rural

(10)

 Research question 2: What are the factors that influence demand for enrolling in a micro health

insurance scheme?

Even though the two research questions might appear rather distinctive at first sight, in reality they are very intertwined. Evidently, when one wishes to assess demand for a certain insurance scheme, it is very helpful to also have knowledge on what constitutes demand in the first place. By taking into account both sides of the coin, ultimately a high-quality insurance scheme can be designed for which enrolment rates will be better than the 10-20% currently measured for schemes. Moreover, as will also become evident from the conclusion of this thesis, in fact some results from the demand study have been incorporated in the recommendations for Dhulikhel Hospital’s insurance scheme settings. In other words, not only the theory and context, but also the results of the feasibility study and the demand study will be entangled.

Research question 1, on behalf of Dhulikhel Hospital, incorporates several innovative perspectives by its approach and setting. First, the feasibility study is largely based on the preferences of the insurance scheme’s prospective clients: The women currently enrolled in the microfinance program. As Dror et al. (2007) argued, there is only little literature on the preferences of income and low-education target groups with little or no experience with insurance. However, when designing a MHI scheme, having knowledge on the needs and demands of the target population is key. Actively exercising client participation is therefore considered to be a necessity, and by means of this feasibility study the literature’s knowledge gap on preferences of rural, poor, illiterate insurance clients is partly filled. A second important feature stems from the study’s context. Various feasibility studies in different regions have been conducted, yet, to the author’s knowledge, this will be the first study assessing demand and preferences for a micro health insurance scheme in Nepal.

The second research question as well is scientifically innovative for a number of reasons. Most importantly, the demand study will not be conducted on an isolated MHI scheme, rather, the scheme will be an integrated part of an already operational microfinance program. Scholars advocate that microfinance institutions are promising and appropriate delivery agents for providing MHI schemes (Matin et al., 2002; Churchill, 2003; Thornton et al., 2010). Several theoretical rationales lie behind this statement. First, they are seen as appropriate providers for microinsurance schemes due to their penetration in the ‘low-income’ market already. Due to their existing client base, it is believed that schemes of existing institutions will quickly reach a sustainable client population (Thornton et al., 2010). A second reason is the fact that, as Giesbert et al. (2011) find, there might exist a reinforcing pattern between insurance and other services provided by a microfinance institution, such as credit or savings possibilities. The authors argue that, among other reasons, existing clients are more financially literate, enabling them to perceive the benefits of purchasing an insurance more easily. Matin et al. (2002), moreover, note that by increasing the range of financial services offered, the poor’s access to different sources of funding is improved. Furthermore, as Churchill et al. (2003) argue, due to the provision of credit for the poor, the microfinance institutions are vulnerable to the same risks as their clients. By for instance offering a life- or health insurance the institution itself is less prone to defaulted clients due to dead or sickness. Last, having a prior relationship with insurance clients potentially increases the degree of trust in the insurer and the scheme, whereas a lack of trust is an often-cited reason for non-enrolment in an insurance scheme (Basaza et al., 2008; Cai et al., 2010; Morsink and Geurts, 2011).

(11)

and Hamid et al. (2011b) note, despite these theoretical motivations, to date there are few studies known to research the impact of adding MHI to an already existing microfinance package. This study, even though it researches ‘willingness to enroll’ rather than actual enrollment, might be one of the first to assess demand for a health insurance scheme in an existing microfinance context.

A second novelty of research question 2 is that it takes into account not only the already heavily researched demand factors in the existing literature, such as age and household income (Schneider and Diop, 2001; Jütting, 2002; Chankova et al., 2008; Jehu-Appiah et al., 2011), but it combines these factors with relatively novel concepts, such as riskiness, hyperbolicity, risk aversion, and being credit constrained. To the author’s knowledge, only few studies take this holistic approach in conducting a demand study (Giné et al., 2008; Bauer et al., 2010; Ito and Kono, 2010; Cole et al., 2012). A last new feature of this thesis is the context of Nepal. Even though various demand studies so far have been conducted, these studies mainly reside in either India or sub-Saharan Africa. To date, no demand study in a Nepalese context has been executed.

1.2 Context of this thesis: Dhulikhel Hospital

Data for this thesis has been gathered using primary data collection. The microfinance institution chosen for the analysis is Dhulikhel Hospital in Nepal, which, being a hospital, has a natural motivation to pursue a micro health insurance scheme. A microfinance program has been in place for five years to date at this hospital, and the program combines microloans with a mandatory health education training for all of its participants.

Due to the mountainous geography of Nepal, Dhulikhel Hospital operates 15 ‘health centres’, in order to provide access to healthcare for even the most rural areas of the country. It is in five of these health centres where the microfinance program is implemented. Hence, the ‘central’ hospital does not participate in the microfinance program. In this thesis, ‘Dhulikhel Hospital’ will be used to indicate the central institution, while ‘health centre’ will be used to refer to the small-scale, community-based health posts part of the larger Dhulikhel Hospital.

The current microfinance scheme of Dhulikhel Hospital is aimed entirely at female participants, and indicates that each woman (managed in groups of 10-13) receives a loan between 5,000 NRs (Nepalese Rupee) and 7,000 NRs (approximately between €50 and €70), which is to be repaid in semi-annual installments for four years. The interest charged is only 4% annually, while in contrast most formal loan institution charge interest rates up to 20% per annum. The loan is required to be used for income-generating purposes, and the majority of women purchases livestock with her loan.

In addition to this loan, the participating women are required to attend monthly meetings at the health centre, where they receive health education from the hospital staff. These sessions cover topics related to women’s and children’s health, featuring special attention on the most prevalent diseases in rural Nepal. After the women have repaid the loan, they are not required to attend these sessions anymore. Hence, attending the health education sessions is mandatory, and free of charge. The total runtime of the program (microcredit and health education) equals four years.

(12)

microinsurance scheme would become mandatory for all participating women. This mandatory health insurance has the benefit of eliminating adverse selection, which will further be elaborated on in chapter 2.

The feasibility study component of this thesis draws on a predesigned health insurance scheme which is currently piloted among 10 women and one of their children, in total a group of 20 people. For a monthly premium of 50 NRs, the woman and her child receive the following healthcare benefits:

 The out-patient service area will be covered

 One day hospitalization services will be covered

 The micro-insurance program will provide basic medicines that are required

 Every lab facility available in the health centre will be provided

 Each and every area of the safe motherhood program will be covered (including antenatal care, delivery, postnatal care, and child health)

 Chronic illnesses are not covered

However, currently, the scheme is not financially viable. This study will examine whether there is demand for a MHI scheme among the target population, which services should be included, and what premium would make for a long-term financially sustainable health insurance scheme.

As mentioned previously, next to the feasibility study, this thesis also will provide a ‘demand study’, assessing which factors influence an individual’s demand for micro health insurance, or ‘willingness to enroll’. This demand study will be executed simultaneously with the feasibility study, using the same respondents of the microfinance program of Dhulikhel Hospital. The demand study broadens our knowledge of what constitutes the low demand for MHI measured in existing microinsurance schemes, as well as gives an opportunity to design a high-quality MHI scheme for Dhulikhel Hospital which takes into account these demand factors, ensuring higher enrolment rates and a better fit with the target population.

1.3 Outline of this thesis

(13)

(14)

CHAPTER 2 INSURANCE FROM A THEORETICAL PERSPECTIVE

Everybody faces certain risks in daily situations. This holds both whether you are a car owner in the Netherlands facing the risk of car theft or damage, or whether you are a rice farmer in India facing the risk of a flood damaging your harvest. Uncertain shocks that stem from risk can have a detrimental impact on one’s income and wealth, and in general, when the size of the loss is significantly high, people take steps to smoothen their income.

According to Ray (1998), people can smooth their income in several ways. The first method is to engage in self-insurance. This refers to the act of smoothing one’s income from uncertain shocks by means of one’s own wealth. Practical applications of self-insurance are for instance using savings from regular income, or selling assets when an uncertain shock occurs. In essence, individuals who self-insure are ‘protecting themselves against uncertainties that they cannot control’ (Ray, 1998: 591).

A second approach to income-smoothing is taking on credit. Whether this credit stems from formal financial institutions, informal moneylenders, or (especially relevant in developing countries) microcredit institutions, it provides a fast and readily available source of money that may alleviate the burden of income constraints due to unexpected shocks. Both self-insurance and taking on credit are forms of coping mechanisms for unexpected health shocks. Coping mechanisms will be discussed in depth in chapter 4, in which the current situation in developing countries is assessed.

A third method, and the focus point of this thesis, is the use of insurance. Other mechanisms, such as adjusting labor supply or making conservative production choices (Morduch, 1995) are outside the scope of this thesis.

This chapter proceeds as follows. First, an example is used to illustrate a perfect insurance model related to health issues. However, such a perfect model only exists theoretically, in practice this model is not feasible due to several limitations, which are discussed in the second section. The last section deals with differences in insurance characteristics of traditional insurance and microinsurance.

2.1 The perfect insurance model

Insurance in its simplest structure will be illustrated by a health-related insurance product, in line with the focus of this thesis. Two people, A and B, have the same income of 1,000. In addition, they both have the same probability of falling ill and being hospitalized, with a probability of 50% (for illustration purposes, even though this figure is not very realistic), and costs of 500. In other words, they have a 50% probability of consumption of 1,000, and 50% probability of consumption of 500. In the following example, only the payment of hospital bills is considered. However, in practice, one not only faces reduced consumption due to unexpected payments, but also faces indirect shocks from falling ill, such as lower income due to the inability to work, or lowered productivity.

(15)

the shock on income they experience due to hospital bills. However, in scenario 3 and 4, when either one of them falls ill, they can smoothen their income by means of this mutual insurance scheme. Consider the idea that in scenario 3 or 4, the ‘healthy’ individual gives 250 to the ‘sick’ individual, such that both their individual consumption in these scenarios is 750. Their individual consumption without the mutual insurance scheme is either 1,000 (when staying healthy) or 500 (when falling ill), and hence their expected consumption is 750. Under the mutual insurance scheme, their expected consumption in both scenarios equals 750 as well. Theory reasons that risk-averse people prefer to engage in the insurance scheme, since even though this renders them a constant consumption of 750 in each possible contingency, utility is higher under the insurance scheme. As Eeckhoudt et al. (2005) argue, the diversification of risk results in a ‘mean-preserving’ transfer of wealth from the extreme events to the mean. A risk-averse individual always prefers receiving the expected outcome of a ‘lottery’ with certainty, in this example a certain income of 750, rather than the ‘lottery’ itself, which refers to an income of either 500 or 1,000. This is based on Jensen’s inequality theorem for a concave utility function, and is visualized in figure 2.1. Expected utility of wealth ω(750) (green dot in the figure, point C) exceeds the expected utility of the average of the utility function with 0.5*ω(500) + 0.5*ω(1000) = 750 (red dot in the figure, point B).

The rationale behind the preference of a certain payoff over an uncertain one lies in the decreasing marginal utility function. The potential loss of 250 when one falls ill indicates a larger reduction in utility compared to a potential gain of 250 when one remains healthy (Eeckhoudt et al., 2005). Figure 2.1 translates this as a larger difference between points A-B compared to the difference between points C-D on the y-axis.

(16)

Under mutual insurance, person A and B share the risk of an unexpected shock to their consumption due to costs associated with hospitalization, and they hence dampen the effect this has on their consumption. Such a scheme only works when the outcomes are relatively independent from one another, otherwise the probability that scenario 3 and 4 occur (recall, these are the only scenarios under which this insurance structure is viable) is extremely low. Since the event of individual A falling ill is unrelated to the event of person B falling ill (at least in the majority of circumstances), this independence is guaranteed.

Of course, one can extend this mutual insurance scheme in such a way that there are many people contributing to a common fund, from which sick people can withdraw in case of hospitalization. Referring back to the four scenarios outlined previously, this insurance scheme now implies feasibility not only in scenario 3 and 4, but also in scenario 1 and 2, simply since there are more people engaging in risk-sharing. All people are able to smooth their consumption completely, and do not suffer from any unexpected health shocks that will decrease their consumption levels. Ray (1998) labels such a scheme ‘perfect insurance’.

Positioning this illustration in a mathematical formula, it shows that an individual without participating in an insurance scheme has a consumption of

𝐶 = 𝑌 − 𝜖 − 𝛳

Where Y represents the individual’s income in a given period, ϵ represents idiosyncratic shocks the individual faces in terms of value, and ϴ represents systematic shocks of the person in terms of value. Idiosyncratic shocks stem from the risks that affect only a single person. In contrast, systematic, or aggregate shocks, are faced by an entire village or community. With regard to health insurance, most risks faced are idiosyncratic: the risk of an individual falling ill and the necessity to be hospitalized. With systematic risks, one can think of the risk of an epidemic reaching a community. When this aforementioned person engages in health insurance, the costs of the idiosyncratic (‘individual’) risks he faces can in essence be transferred to a ‘common pool’ to which multiple individuals have contributed. His consumption formula becomes

𝐶 = 𝑌 − 𝑃 − 𝛳

Where P represents the premium one needs to pay in order to opt in the health insurance scheme. Since the size of the premium P is only a small portion of the total amount of the idiosyncratic shock ϵ, the person’s individual consumption is higher than compared to the situation without insurance. In practice, however, the perfect insurance model does not exist. There are limits to the ability of individuals to insure one another, which will be discussed next.

2.2 Limitations to the perfect insurance model

(17)

defaulting insurance company not paying valid claims, resulting in an ‘unfair’ price for the insurance. With regard to behaviors of the insured individuals, Radermacher et al. (2006) argue that moral hazard, adverse selection, and fraud are considered to be the main issues. These three matters are certainly not exhaustive, but are believed to be most pressing with regard to issuing a health insurance scheme, and hence for this study. The three aforementioned limitations will be discussed next with regard to their special relationship with health insurance.

Moral hazard, fraud and adverse selection are believed to especially present in health insurance, due to the ‘subjective nature of the insured events’ (Radermacher et al., 2006:69). Consider for example a life insurance, which pays out in case of the much more ‘objective’ event of death. Another example is the index-based crop insurance, which pays out when the rainfall in a certain area has been less than a preset minimum. Sickness, on the other hand, is much more subjective. Related to subjectivity, as a result falling ill suffers from high costs of verification. Whereas the events under life insurance and crop insurance are relatively easily verified, to establish whether one has fallen ill, (often high-cost and multiple) consultations and examinations are needed. Due to both the subjective nature of falling ill and the difficulties arising with verifiability, it is believed that health insurance is more affected by the aforementioned problems of moral hazard, adverse selection, and fraud than other types of insurance. A last limitation is enforceability, which is equally likely in all types of insurance. As is in line with the focus of this thesis, the aforementioned limitations will be discussed in regard to health insurance specifically.

2.2.1 Moral hazard and fraud

Moral hazard stems from an information asymmetry, and can take several forms in the case of health insurance (Armendáriz and Morduch, 2010). First, insured individuals may be less likely to take necessary precautions, since, once they are insured, they believe they are more protected from unexpected health expenditures (called ex ante moral hazard). A second form of moral hazard stems from overusing facilities, that is, seeking treatment for only minor illnesses, for which no treatment would have been sought without insurance (called ex post moral hazard).

Related to moral hazard is fraud, which can be committed by several actors in the insured-insurer relationship. The insured individual, for example, might be obtaining treatment for persons that are not covered by the insurance scheme, by means of impersonation. The healthcare provider also has several options to commit fraud, such as aggravating claims by prescribing the most expensive, but not necessary, drugs, or submitting false claims to the insurer (Radermacher et al., 2006).

2.2.2 Adverse selection

A third, and very important limitation in health insurance is the issue of adverse selection. Adverse selection arises when households exposed to more risk (such as less healthy households) are more likely to opt in the health insurance scheme. The insurer hence faces a relatively less healthy client base when compared to the overall population’s health, implying more risk for the insurer (Armendáriz and Morduch, 2010). As Radermacher et al. (2006) note, adverse selection causes the actual costs of insuring individuals to be higher than expected, especially if the group size is small, and hence could threaten the financial sustainability of an insurance scheme.

(18)

voluntary health insurance. In a study considering demand for an Indian micro health insurance scheme, Ito and Kono (2010) find also signs of adverse selection being present.

With adverse selection, the individuals who do purchase health insurance are more inclined to make claims, reducing the feasibility of risk sharing. Due to information constraints, the insurer fails to gain knowledge about the relative health of the household, and hence is unable to set different prices for different households. Such an information asymmetry can be overcome by the insurance company requesting additional information, however, with costs associated. Mechanisms to counteract adverse selection are considered in the next section.

2.2.3 Enforceability

Ray (1998) lists another limitation apparent when issuing insurance: Limited enforceability. Consider a situation in which an individual in a certain period decides not to contribute to the common insurance fund, and hence is deviating from the insurance arrangements. Obviously, the gain from deviating is a higher current consumption, since the individual does not need to pay the health insurance premium. On the other hand, the costs of deviation are twofold: First, there is the possibility of exclusion from future access to insurance. Second, an individual might experience social sanctions from the community.

Due to all these aforementioned limitations, perfect insurance as described in section 2.1 cannot be accomplished. As an alternative for a perfect insurance scheme, a so-called ‘second-best’ insurance scheme now becomes an option. The second-best scheme does not provide full, but rather incomplete insurance. This incomplete insurance scheme is needed to account for the various limitations that are associated with engaging in insurance.

2.3 Counter mechanisms for moral hazard and adverse selection

Several ‘cost-sharing’ mechanisms, such as deductibles or co-payments, can alleviate moral hazard (Armendáriz and Morduch, 2010). The Economist (1995) argues that deductibles are appropriate for ex ante moral hazard behavior, whereas copayments are more suitable for ex post moral hazard behavior. A deductible acknowledges claims only over a certain minimum amount spent; the insured individual has to pay the medical expenditures up to a threshold, despite being insured. This encourages more precautions counteracting ex ante moral hazard. Co-payments require the insured individual to account for a specific percentage of the total medical expenditures, and hence limits overusing facilities (ex post moral hazard) since insured individuals have an incentive to keep the size of the claim low. If both types of moral hazard behavior exist in a certain context (or when the insurer is unaware which type is causing problems), a combination of both a deductible with co-payments is not advisable. Whereas ultimately timely healthcare sought can reduce the overall claimed amount, the existence of a deductible might actually discourage early action. Therefore, it is recommended that the insurer designs an insurance scheme with ‘larger-than-usual’ co-payments.

(19)

capital as the ratio of a district’s population living in small and isolated communities, they found that the existence of social capital reduces moral hazard. It should be noted, though, that both duos of authors specifically model informal agreements, whereas Dhulikhel Hospital’s MHI scheme will display more formal characteristics.

Specifically for counteracting adverse selection, the scheme’s design can, among others, feature household enrolment versus individual enrolment, the establishment of a waiting period between enrolment and being allowed to submit claims, and the exclusion of coverage for chronic or pre-existing illnesses (Weber, 2002; Ranson, 2003). All three mentioned approaches ensure that not only high-risk individuals enroll in the scheme, rather that the total risk of the insured population reflects the overall population. For instance, with individual enrolment, only the family member who is ill most often will enroll, while with family enrolment, also the healthier individuals need to enroll. By means of a waiting period, the scheme is prevented from enrolment of currently ill people who receive free healthcare, and once their illness is resolved, they drop out of the scheme. Lastly, the exclusion of chronic illnesses avoids that only these patients will enroll, requiring healthcare to a great extend, and consequently making the scheme most likely unsustainable in the future.

Next to the aforementioned methods, the most intuitive counter mechanism for adverse selection is designing a mandatory, rather than voluntary, health insurance scheme. In this way, all households engage in risk sharing, regardless of their health status. Moreover, next to improved claim ratios by reduced risk and reduced adverse selection, compulsory coverage also leads to cost reductions due to higher volumes (Wipf et al., 2006; PlaNet Finance, 2011).

Research on mandatory insurance coverage, however, shows also negative aspects of such a compulsory additional product in order to receive a microfinance loan. Manje (2005), researching a Zambian microfinance institution with compulsory life– and funeral insurance, found that clients see the insurance as a condition for borrowing, rather than a beneficial product. Other problems associated with mandatory insurance are a lack of understanding of the insurance benefits, the estimation of the height of the premium paid that highly exceeds the actual amount, and, in some cases, clients that are unaware they have insurance in the first place (Manje, 2005; McCord et al., 2005; Churchill and Cohen, 2006).

To counteract the difficulties related to compulsory coverage, Churchill and Cohen (2006) suggest to take a demand-driven perspective, and to treat the insurance as a complementary service that will actually benefit the clients. Key here is the client’s understanding of the benefits of the insurance scheme, as well as an active participating role of the prospective clients with regard to designing the specific features of the scheme.

(20)

2.4 Differences between traditional insurance and microinsurance

The example provided in section 2.1 describes, in essence, the working of a (perfect or second-best) insurance, however, in the real world, insurance schemes often do have different characteristics from such a theoretical model. Most importantly, instead of the people contributing and withdrawing from a common pool, the majority of insurance schemes involves a third party.

When purchasing insurance, one (called the insured) ‘transfers’ the risk of a financial loss associated with a certain object or activity to a third party (called the insurer). In exchange for a premium (dependent on both the probability and size of the loss) paid to the insurer by the insured, the insurer guarantees to compensate the insured for the financial loss associated with that specific risk.

Regardless of insurance in developed or developing countries, the concept of insurance as the ‘transfer of risk’ is identical. However, distinctive differences can be perceived in the characteristics both types of insurance possess. The main differences between ‘traditional’ insurance in developed countries and microinsurance in developing countries can be found in table 2.1.

Table 2.1: Differences between traditional insurance and microinsurance1

Traditional insurance Microinsurance Clients • Low risk environment • High risk environment, high

• Established insurance culture vulnerability

• Little understanding and knowledge about insurance

Distribution models • Sold by licensed intermediaries or insurance companies with a broad knowledge about insurance

• Sold by non-traditional

intermediaries with little experience of insurance

Policies • Complex policy documents with many exclusions

• Policies characterized by simple langue and few, if any, exclusions

• Group policies

Premium calculation • Good statistical data • Little historical data • Pricing based on individual risk • Group pricing

• A very price sensitive market

Premium collection • Monthly to yearly payments • Frequent and irregular payments

Control of insurance risk • Limited eligibility • Broad eligibility

(adverse selection, moral hazard)

• Significant documentation required • Limited controls

• Screenings, such as medical tests, may be required

Claims handling • Complicated processes • Extensive verification

documentation

• Simple and fast procedures for small sums

1

Adapted from Lloyd’s 360° Risk Insight, 2010.

(21)

many exclusions and limited eligibility. Microinsurance is also characterized by limited control of insurance risk, in order to minimize transaction costs. Instead of requiring complete documentation, or a health screening in order to be accepted by a health insurance, microinsurance provides broad eligibility.

(22)

CHAPTER 3 HEALTH INSURANCE IN DEVELOPED COUNTRIES

Among the developed countries in the world, the majority of nations employs some type of universal

healthcare system. Universal healthcare can be defined as ‘physical and financial access to necessary healthcare of good quality for all persons in a society’ (Kutzin, 2000: 2), and hence is available to all,

irrespective of income or status.

Naturally, such nationwide health insurance systems did not develop overnight. Abel-Smith (1999) provides a brief historical description of how Western European countries initiated and extended the coverage of their health insurance schemes, up until the point all residents of a country are similarly covered under one health insurance system.

As Abel-Smith (1999) notes, many of the first health insurance schemes were initiated by workers themselves, either linked by a similar profession, or bound by geographical borders. Initially, only few insurance funds were started by employers. With the expansion of the system, healthcare providers themselves also started operating health insurance schemes, in order to ensure that their profits were not eroded by independent insurance collectives who had a strong bargaining position.

In 1883, Germany was the first country to engage in a mandatory health insurance scheme for all workers (Saltman and Dubois, 2004). Such a system had the advantage that all employers were required to contribute as well, facilitating also the lower income groups to participate in the scheme. Moreover, the premiums did not depend on health status or family size, resulting in an equal, nationwide coverage for all inhabitants.

From there on, the system was extended to the self-employed sector (particularly farmers, fisherman and the like, who did not have an employer to share the premium), by a variety of different methods employed by different countries. Examples were highly subsidized state hospitals, such that the premium was low for all insured individuals, or subsidizing only the self-employed individuals. Moreover, premium collection for the self-employed could be based on taxes paid on the amount of land, on the harvest produced, or on the farmer’s stated income (Abel-Smith, 1999). The initial universal healthcare coverage was a fact.

(23)

3.1 Universal healthcare coverage through three approaches

The financing of a country’s healthcare expenditures typically stems from four sources: 1) general taxation, 2) social insurance contributions, 3) private insurance, or 4) out-of-pocket (OOP) payments (Wagstaff and Van Doorslaer, 1992). Hence, how such a universal health care system is financed can differ greatly between countries. A general distinction can be made between tax-financed systems, social-insurance systems, and a private system. These three systems will be discussed in turn by taking three developed countries, each with its own health insurance system, as an example. These are Denmark with a tax-based system, France with a social-insurance system, and lastly the U.S. with a private health insurance system. Please consult figure 3.1 for a graphical display of the distinctive ways in which a national health insurance system can be arranged.

Figure 3.1: Three approaches to funding a national health insurance system1

Health care in Denmark is mainly financed through a centrally-collected tax set at 8% of taxable income, and hence can be viewed as a tax-financed system. Moreover, around 36% of the Danish population purchases complementary private health insurance (Vrangbaek, 2011). An example of a social-insurance system can be found in France, where the health insurance scheme is financed by employer and employee payroll taxes, a national income tax, and state subsidies (Durand-Zaleski and Chevreul, 2011). Lastly, the health care system of the United States can be earmarked as a private system. In the U.S. there are two social health insurance programs (Medicare for the elderly and for the disabled under age 65, and Medicaid for certain groups of the poor), and over 1,200 not-for-profit and for-not-for-profit health insurance companies offering private insurance (The Commonwealth Fund, 2011). Technically, a private health insurance system cannot be labeled as universal coverage, since by definition people with a higher income are then able to purchase a better, or more covered, health insurance.

(24)

3.2 The Dutch health insurance system

Up until 2006, the Netherlands had a similar system as the French system described previously. The Dutch healthcare system was known as a social-insurance system, in which the majority of the public health expenditures is financed with social insurance contributions. All Dutch residents with earnings up to approximately €30,000 (US$42,116) per year were statutory covered by public sickness funds. People exceeding this threshold level, and their dependents (roughly 35% of the Dutch population) were automatically excluded from statutory coverage. This part of the population could purchase healthcare coverage from private health insurers (Westert and Klazinga, 2011). Due to increased dissatisfaction with the dual system of both public and private coverage, the Dutch healthcare system has been reformed as from January 1, 2006.

Since that date, all residents of the Netherlands are lawfully obligated to purchase health insurance coverage under the Health Insurance Act (Zorgverzekeringswet, or Zvw2). The state determines the

rather broad basic health insurance package, which is mandatory ‘statutory coverage’ for all residents. Next to this basic coverage, individuals can purchase complementary or supplementary ‘voluntary health insurance’ (VHI). Such VHI provides for instance coverage for extra visits to a physiotherapist, or additional coverage for visiting the dentist (Westert and Klazinga, 2011).

The health insurers are legally required to accept anyone applying for the statutory coverage, charging people the same premium regardless of health status or age. With regard to the VHI, on the other hand, the insurers are not obliged to accept all applications. Moreover, different VHI premiums are charged by different insurers.

The statutory health insurance is financed by means of both an income-related contribution and individually-paid premiums. The income-related contribution is set at 6.9% of the first €32,369 (US$45,442) of annual taxable income. The contribution is officially paid by employers, however the employee is required to pay taxes over this amount (Westert and Klazinga, 2011). With regard to the individually-paid premiums, these are directly paid by the individual clients. Clients with a certain minimum earnings threshold are entitled to a ‘state health care allowance’, which aids in paying the individual monthly premium (Westert et al., 2010). Lastly, all clients have an annual ‘own-risk-payment’, or deductible, which needs to be paid by the individual before the insurer pays out any restitution.

The advantage of the Dutch healthcare system lies in her risk equalization system. By means of an equalization pool, people who have a low-risk profile and as a result are relatively cheap to insure, such as young people, will even out the risks associated with the high-risk clients, such as elderly or people with poor health. This risk equalization funds is regulated by the Dutch government, and insurers receive compensation from this regulator on the basis of relative claims. Hence, insurance companies with high payouts receive a larger amount than insurance companies with lower payouts. In other words, this minimizes the insurers’ motivation to select clients on the basis of risk (Westert et al., 2010). As a result, all individuals in the Netherlands can be statutory covered at relatively low costs.

(25)

The problems associated with health insurance, as mentioned in section 2.2, do not apply to the Dutch health insurance system. First, adverse selection does not occur since all Dutch residents are by law required to acquire at least statutory coverage. The Dutch hence are all insured at basic terms, meaning the insured population does not carry significantly more health risks than the entire population. In addition, people may opt to purchase additional voluntary health insurance, for which medical screening might be a prerequisite. Due to such a screening, health insurers are able to make a difference between healthy and less healthy individuals, and can charge different premiums according to health status.

(26)

CHAPTER 4 COPING WITH HEALTH-RELATED RISKS IN

DEVELOPING COUNTRIES

This chapter starts with providing argumentation for why universal coverage might not be feasible in developing countries. Next, an alternative solution is provided, seen by many as a transitional phase towards a nationwide insurance system. Community-based health insurance, or CBHI for short, is introduced, its impact on various outcomes is discussed, and different delivery-options are described. Moreover, next to seeking coverage from CBHI or MHI schemes, people in low-income countries often have multiple coping mechanisms to deal with risks, which are discussed next. Lastly, this chapter gives not only insights into coping mechanisms on a general level, but also delves deeper into the risks and associated coping mechanisms of rural women in Nepal.

4.1 Impediments for implementing universal coverage in developing

countries

As became evident from the previous chapter, various best-practices in achieving universal healthcare through means of insurance are available. Yet, most developing nations fail to institutionalize a national, sustainable health insurance. Literature suggests a plethora of reasons for this lack of universal coverage in low-income countries (Carrin, 2002; Carrin, 2005; Carrin et al., 2005).

Recall that the three main insurance systems are either tax-financed, socially funded, or privately financed. As Carrin (2002) notes, the most effective methods of attaining universal coverage are either the system based on taxation, or the social health insurance scheme. However, a tax-based health insurance system may be problematic to implement due to a limited tax base in developing countries. Increasing this tax base might be difficult due to the often only modest economic growth in the country, as well as the combination with a low degree of ‘formal’ market transactions. Moreover, often there is a low institutional capacity to effectively collect taxes in the first place, by enforcing tax obedience and avoiding tax evasion by the population. Lastly, even with a sufficient tax base in place, this does not ensure that the informal sector participates in the insurance system by paying taxes.

With regard to social-insurance systems, these are usually initiated by firstly insuring workers with a formal payroll. Subsequently, the insurance can be extended to also include the remainder of the inhabitants of a country, such as self-employed people and non-workers. However, in developing countries it is common that the informal sector, and especially self-employed inhabitants, represent a large majority of the population. Hence, a social-insurance system can be tremendously demanding to initiate in a developing country, since there is only a small ‘starting base’ of formal workers. Moreover, large income inequality in a country does not provide a proper foundation either. Middle- and high income earners are likely averse to contributing a higher premium than the poor.

Lastly, private systems are known to be relatively costly, and hence by definition, are not suitable to attain universal coverage for low-income people.

(27)

infrastructure and a lack of ‘basic healthcare’ all add up. Moreover, extensive knowledge on how to collect premiums, and how to organize reimbursements is required for a financially sustainable and trustworthy insurance scheme, for which again developing countries might be lacking expertise.

4.2 An alternative for nationwide coverage – Community-based health

insurance

Even though the knowledge and resources are often lacking, developing countries do recognize the need for a sound and sustainable healthcare system. Therefore, next to a universal, nationwide health insurance, other options are also initiated. One of these options is community-based health insurance (CBHI), which provides people with health coverage on a community-, rather than national, level. Providing insurance on a community level is often associated with easier communication with the target population, and, also important, easier collection of the premiums (Carrin, 2003). A second advantage is the fact that communities might have more confidence and trust in a more local management (Carrin, 2002).

As Fonteneau and Galland (2006:379) rightfully state: ‘It is difficult to give a standard definition of

community-based micro [health] insurance. Literature on the subject has almost as many definitions of the model as there are community-based organizations or specialists’. Indeed, when examining 10

studies on the topic of CBHI, one easily finds 10 different definitions of the concept. Even though most definitions share more or less the same core idea, there are also vastly different characteristics that are highlighted. For instance, Roth et al. (2007) especially note that community-based organizations are ‘member-owned and member-managed’, indicating that the community itself is actively involved in the day-to-day management of the insurance scheme. On the other hand, Bennett (2004:147) defines a CBHI scheme as ‘any scheme managed and operated by an

organization, other than a government or private for-profit company, that provides risk pooling to cover all or part of the costs of healthcare services’.

This latter definition is also the one used in this thesis. CBHI is the overarching label of, among others, micro health insurance (MHI), however, it is not a synonym. In this thesis, the literature on CBHI is also taken into account, since for my purpose, the most important features of a scheme are a design for low-income families, with a strong community-based approach. Both characteristics are evident in MHI and CBHI. Moreover, in this study also MHI schemes are considered that are offered by different providers, such as NGOs and healthcare organizations, as long as the insurance is offered on a community-based level, rather than at a national level. The keyword here indeed is ‘community’, indicating that all programs taken into account in this thesis are small-scale, grass root initiatives.

(28)

argues, CBHI schemes are especially beneficial for self-employed individuals, since this group often does not receive financial protection from other sources. Individuals with formal employment potentially can be covered under some form of organization-based insurance scheme. The final phase ultimately transforms these smaller-scale initiatives into a nationwide, tax-financed or socially-financed (or some hybrid form) health insurance scheme.

Figure 4.1: The process towards universal coverage

4.3 The impact of CBHI schemes

As has been described in the introduction of this thesis already, the idea of community financing is not novel. Even though microfinance practitioners only recently discovered the concept of complementing microcredit with micro health insurance, CBHI has been an innovative model since the mid-1980s (Carrin et al 2005).

To date, several review studies have been conducted on the impact of CBHI schemes on various outcome variables. For instance, Ekman (2004) in a systematic review of the studies written on the effects of CBHIs, notes there is strong evidence suggesting that CBHI schemes do reduce the out-of-pocket (OOP) spending of its members, thereby giving financial protection against unexpected health-related shocks. Moreover, there is some evidence that CBHI increases access to healthcare in the areas the scheme is operating, however, there is no support for increased quality or efficiency of healthcare services. Lastly, he discovers an ‘exclusion effect’ in most CBHI schemes, referring to the fact that most programs are unable, or perhaps even unwilling, to include the ‘least well-off’ groups in the scheme’s operating areas. Compared to the CBHIs specific definition and goals, this notion is contrasting. A second review is conducted by Radermacher et al. (2012), who state that even though numerous variables have been researched in the existing literature so far, only OOP spending and utilization rates are considered to be positively affected by MHI across the board.

Next to review studies, individual articles report the effects CBHI or MHI schemes have on health awareness, utilization rates, health outcomes, and income protection (Schneider and Diop, 2001; Jütting, 2004; Schneider and Hanson, 2007; Chankova et al., 2008; Thornton et al., 2010; Hamid et al., 2011a; Magnoni and Zimmerman, 2011), however, evidence is highly mixed. For all mentioned outcome variables one can find studies that either show either a positive or a negative correlation with enrollment in a health insurance scheme, while another category of studies finds no statistically significant relationships.

Palmer et al. (2004) provide a literature review on CBHI schemes, and find especially the small size of most schemes to be noteworthy: 50% of all reviewed schemes had less than 500 members. Roth et al. (2007) argue that scale in providing health insurance is a significant problem. A scheme needs to create a ‘critical mass’ in order to be financially viable in the long run.

No financial

protection

•Reliance on OOP

Intermediate

coverage

•Different small-scale initiatives

Universal

coverage

(29)

4.4 Different delivery options of CBHI schemes

Ranson (2003) and Devadasan et al. (2006) both identify three distinct forms of how a community-based health insurance can model its ownership and management. Figure 4.2 provides a graphical representation of the working of these three models.

In the ‘provider model’ the insurance scheme operator has its own facilities for (primary and/or secondary) healthcare. Moreover, the scheme operator collects the insurance premium itself, as well as manages its own insurance fund. In other words, the scheme’s management bears the financial risks of the insurance scheme. In the ‘insurer model’, the scheme operator acts as the insurer, but not the healthcare provider. Again, the organization collects the premiums and bears the associated risks, however the institution ‘purchases’ healthcare from independent healthcare providers, whether these are for-profit or not-for-profit. The insured individuals either pay their medical bills upfront, after which they receive reimbursement from the insurance fund, or they can make use of some form of third-party payment mechanism. In the last type, the ‘partner-agent model’, sometimes also labeled the ‘linked model’, the insurance scheme operator has an intermediary role, linking the members of the community, a third-party insurance company, and (one or more) healthcare provider(s) to form an insurance scheme. The organization collects the individual premiums from the community, and forwards the total group premium to an insurance company. The insured community members then can either receive reimbursement of their medical bills, or enjoy a third-party payment mechanism.

Figure 4.2: Three delivery models of community-based health insurance (CBHI) schemes

(30)

services, while the provider increases its potential customer base, receives revenue from people that otherwise would have foregone treatment, and restricts patients to choose alternative healthcare facilities.

One the other hand, the combining role of both provider and purchaser of the service, as is the case in this provider model, is a potential source of conflicting interests (Ahsan and Mahmud, 2012). The healthcare provider has an incentive to either under-provide services, when the premium is used directly for operating the healthcare facility, or an incentive to over-provide, in the case of fee-for-service payments. Moreover, actuarial expertise to calculate sustainable premiums often is lacking, even though the provider has the best possible data regarding healthcare expenses. A last disadvantage of the provider-based model is its absent access to reinsurance. Reinsurance can be defined as the transfer of liability from the primary insurer, which is the healthcare provider in this case, to another insurer, called the reinsurance company (Outreville, 2002). Reinsurance significantly increases the long-term sustainability of the insurance scheme, since the primary insurer can safely accept more and higher risks to insure.

Even though these disadvantages of the provider model should not be overlooked, it has to be noted that no delivery type of CBHI schemes is free of flaws. In the partner-agent model for instance, the scheme operator acts as an agent of the insurance company towards the community, while in reality it might affiliate more with the community. In the case of claim settlements, this might result in a conflict of interest, decreasing the community’s trust in the NGO (Radermacher and Dror, 2006). Another example is the insurer model, which requires a considerable human and financial investment, as well as actuarial knowledge, before being able to operate the scheme. Moreover, the scheme operator has to negotiate with multiple healthcare providers about prices, copayments, and premiums. For these reasons, many practitioners actually advocate to incorporate the beneficial features of all models into one comprehensive delivery model for community-based health insurance (Ahsan and Mahmud, 2012).

Ekman (2004) reviewed the evidence on 3 different types of delivery models: provider-based schemes, community-based schemes (in the study defined as ‘operating outside of service providers’), and national health insurance schemes. Evidence suggests that the provider model has a positive effect on resource mobilization, and shows no signs of any exclusion effects (in other words, even the least well-off groups are represented). However, there is no evidence on financial protection by increased access the healthcare.

Contrastingly, community-based schemes do have strong exclusion effects, as well as limited impact on resource mobilization. However, such schemes seem to do offer increased access to healthcare. Lastly, the nationally-based schemes are found to have limited resource mobilization and limited financial protection.

INSURING THE POOR