Health Insurance for the Poor: Myths and Realities
Based on a survey in seven locations, this article finds that most Indians are willing to pay 1.35 per cent of income or more for health insurance and most people prefer a holistic benefit package at basic coverage over high coverage of only rare events. The needs of the poor, and their demand for health insurance,
depend on local conditions.
DAVID M DROR
I
The analysis is based on data obtained through the largest comparative household survey conducted in 2005 in seven locations where micro health insurance units are in operation; the survey included both insured and uninsured persons. The seven locations are: Tamil Nadu (one urban and one rural location); Karnataka (one rural and one tribal rural location); Maharashtra (one rural and one urban location); and Bihar (mostly rural location). We also conducted focus group discussions (FGD), key-informant-interviews, and special sessions in which persons applied a decision-making simulation. The total size of for health insurance, and the lower their income, the less they are willing to pay for health. The reality: We collected field evidence in seven locations where micro health insurance units operate, using a bidding game to assess willingness to pay (WTP). The evidence shows that most people are willing to pay more than 1per cent of their income for health insurance. As can be seen in Figure 1, 50 per cent of the sampled population (15,668 persons in 3,204 households) stated a willingness-to-pay level of
1.35 per cent of annual household income for a health insurance package; and 75 per cent of the sampled population agreed to pay about 1 per cent of annual household income. Just as a reminder, median household income in this sampled population was Rs 41,400 per year (median income per person Rs 9,000). Consequently, this study shows that the majority of the sampled population were willing to pay about Rs 559 per household per year (value date mid-2005).
Furthermore, the poorer households agree to pay more as household income increases (Figure 2), but the poorest are willing to pay a higher percentage of
100
household income than less-poor households. This confirms that the poor prioritise access to some healthcare, and that this basic level is quite stable. The policy insight: The declared WTP levels are much higher than what has been assumed as feasible hitherto. Consequently, the demand for pro-poor and pro-rural health insurance at realistic premiums exceeds the supply available at present. Myth No 2: High costs of hospitalisation and surgery pose the greatest financial risk for poor households. The reality: The household survey yielded information about the breakdown of the cost of illness episodes by benefit type (Figure 3). Hospitalisations are rare and very expensive. Drug consumption occurs much more frequently, and sometimes can cost as much as hospitalisation, while in other cases would be much cheaper. Indeed, there is no significant difference between the aggregate costs of hospitalisations and drugs.
As for consultations, each consultation is usually not very costly, but the aggregate expense of this benefit type amounts to more than 50 per cent of the cost of hospitalisation. Policy insight: Health insurers and policymakers that aim to grant to poor households effective financial protection against the cost of illness would wish to ensure that the benefit packages should include drugs, tests and consultations, in addition to hospitalisations. Myth No 3: The larger the poor household, the less attention to health and therefore the more sickness among its members. Therefore large households pose a higher risk to the insurer. The reality: Larger households reported fewer illness episodes. As shown in Figure 4, there is a steep drop in illness
Figure 1: Willingness to Pay (WTP) as Per Cent of Annual Household(HH) Income
the sample has been 4,931 households, but
the effective sample for certain issues dis
cussed in this article differs according to
the number of valid replies.
The household survey, as well as the
FGD and the analysis have been con
ducted under the EU/ECCP project
‘Strengthening micro health insurance units
for the poor in India’ (www.microheal-
Per cent willing to pay more than...
90
1.35 per cent
80
1.3 5%
70
60
50
40
30
20
thinsurance-india.org). This article offers
10
evidence to show why the commonly held 0opinions are in fact myths.
0123456 7Myth No 1: The poor are unwilling to pay Premium (Per cent of HH income)
Economic and Political Weekly November 4, 2006 4541
Figure 2: WTP by Income Bands with more reported illness episodes. The
2.0 data reveals that among the poor, higher
income is associated with higher reported
1.8
prevalence of illness. Is it possible that the
same set of symptoms will be reported as
1.6
an illness in the more affluent families
1.4
40 1. 1. 1. % 900 800 700 600 500 400 300 Median WTP in Rs 1.79 per cent 1.56 per cent 1.50 per cent 0.84 per cent 1.04 per cent 800 600600 500 400 Median WTP as per centof annual income
while ignored by the poorest-of-the-poor?
We can say with certainty that within the
poor population (both urban and rural), the
poorest-of-the-poor subgroup does not
represent a higher risk for health insurers
than the more affluent subgroups.
1.2
1.0
0.8
However, when the socio-economic < 24k 24 k-36 k 36 k-48 k 48 k-72 k 72 k+ 0.6 status is measured by the quality of house Annual HH income (in ’000s) type rather than by income, higher-quality
—•— Median in INR (Y1) —+— Per cent of HH income (Y2) housing is associated with lower morbidity. Prevalence of illness in households
Figure 3: Cost Distribution of Illness Episodes
decreases as education increases. This finding is valid when the best educated person in the household is a man or a woman, be it the household-head or someone else.
400 350 300 250 200 150 100 50 0Mean cost per illness episode(in Rs)Females are more likely to be ill than males, and the under five age-group as well as +55 years’ age-group are very vulnerable. Yet, entire households are not very much influenced by their age and gender
Tests and Imaging Consultation Hospitalisation Drugs
composition, probably due to intra-
Figure 4: Lower Prevalence in Large Households
1.6
1.4
1.2 1
0.8
0.6
0.4
0.2 0
0246 810 HH size
household demographic smoothing. Policy insight: Intra-household information sharing, resource-and-asset sharing and demographic balancing within can lower prevalence of illness in households. Additionally, en bloc affiliation of households can lower the risk of adverse selection. Ignoring household features when calculating the premiums could result in premiums that are unjustified by the insured risk. Myth No 5: Poor people, who are often
12 illiterate and innumerate, are unable to make judicious rationing decisions regard-
Prevalence of illness perPrevalence of illness among
household member household members
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05 0
Figure 5: Prevalence of Illness and Income
ing the composition of a health insurance
Highest
medium or high coverage level. Participants chose first the benefits that cost most;
Lowest 2nd quintile 3rd quintile 4th quintile Income per HH member (by income bands)
episodes when household size increases from one to four persons, and is stable thereafter. Therefore, larger households represent a lower risk to insurers. (Sample size was 3,531 households, representing 17,323 persons; conducted in five
locations in Tamil Nadu, Maharashtra and Bihar). The average prevalence of illness in households for three months is 0.292. Myth No 4: Low income and low assets are indicators of higher risk of illness. The reality: Higher income is associated
these included: outpatient (OP); inpatient (IP); tests and imaging (T); and drugs (D). Table 1 lists the frequency of choices made:
The frequency of the choice stated by the participants reflects a clear preference for a broad benefit package, even at basic
Economic and Political Weekly November 4, 2006
Figure 6: Frequency of Choices of Minor Benefits Made in CHAT India
(Per cent) 100
45.7%42.1 45.7 77.8 90.4 56.0
Dental Medical Preventive Maternity Indirect Mental care equipment care costs
Figure 7: Prevalence of Illness in Households
0.663
Rs per household per year (in Rs) Illness episode per householdmember per quarter
0.128 0.138 0.291 0.23 level of coverage. The benefit type that was selected most frequently was drugs, followed by tests and hospitalisations, and consultations came next. The choices made by participants match the information on cost of healthcare obtained through a different survey, and reported above.
Additionally, participants selected benefits that cost less, and interestingly these choices, shown in Figure 6, provide protection to the weaker segments of the group (such as pregnant women or persons with disabilities). Policy implications: The results of this analysis demonstrate that the poor can participate actively in the design of the health insurance packages, and that they make judicious choices. The CHAT tool enables us to identify clients’ perceived priorities. Myth No 6: The poor are essentially quite similar to each other: with similar needs, a low ability to pay, low levels of education and a low demand for insurance. Therefore, uniform (“one size fits all”) insurance products are suitable for the poor. The reality: The healthcare needs of the poor are strongly context-dependant. This is evidenced by the difference in incidence of illness episodes in different locations and by the different cost associated with an illness episode in different locations. The demand for health insurance, evidenced by willingness to pay for it, is also strongly location-dependant. The evidence in Figures 7, 8 and 9 show the difference in prevalence of illness in households; the different levels of insurable cost of illness episodes (insurable costs include hospitalisation, drugs, tests and consultations); and the levels of willingness to pay for health insurance, respectively. Policy insight: Communities differ from each other significantly in their needs and priorities. For an insurance product to be
Table: Frequency of Choices Made
(Per cent)
| Choice | No of | Per Cent of |
|---|---|---|
| Groups | Individuals | |
| 1 OP(b)+IP(b)+T(b)+D(b) | 6 | 26.80 |
| 2 IP(b)+T(b)+D(b) | 8 | 31.80 |
| 3 OP(b)+T(b)+D(b) | 3 | 13.90 |
| 4 OP(b)+IP(b)+D(b) | 3 | 11.90 |
| 5 OP(b)+IP(b)+T(b) | 1 | 4.00 |
| 6 IP(m)+D(b) | 1 | 4.30 |
| 7 T(m)+D(b) | 1 | 4.00 |
| 8 IP(h)+T(h) | 1 | 3.30 |
Notes: (b) = basic coverage level; (m) = medium coverage level; (h) high coverage level OP = Outpatient, IP = Inpatient, T = tests and imaging, D = drugs.
M aharas htra
rural
2500 2000 1500 1000 500 0
1112
Maharashtra rural
Maharashtra Bihar Tam il Nadu urbanrural rural
Figure 8: Insurable Costs per Illness Episode
Maharashtra urban
Bihar Tamil Nadu rural rural
Mean
Me
Figure 9: Average Willingness to Pay for Health Insurance
1150
Tam il Nadu urban
Tamil Nadu urban
915 498 598 460 653
Maharashtra Maharashtra Karnataka Karnataka Bihar Tamil Nadu Tamil Nadu rural urban tribal rural rural rural rural urban
Economic and Political Weekly November 4, 2006 attractive to such diverse market, it must respond to context-specific needs, costs, and willingness to pay levels. The optimal adjustment between medical needs, their costs and willingness to pay must also take into account the perceived priorities of the prospective clients, and such perceptions may also be location-specific. Therefore, a “one-size-fits-all” insurance product is unsuited to the poor clientele and to the reality of India. Myth No 7: The premium that the poor are willing and able to pay cannot cover their essential needs. This puts them in a vicious cycle of inability to pay → no insurance → no access to care → lower health → lower capacity to earn → inability to pay… The reality: The household survey we conducted yielded data on the cost of illness in households during the three months preceding the survey (including consultations, tests and imaging, prescribed drugs and hospitalisations). It turns out that the contribution that households are wiling to pay toward health insurance, which is about Rs 30 per person per quarter, would generate enough funds to cover the medical needs of more than 50 per cent of the household, provided that the overall cost is spread on the entire sampled population. The principal mechanism that makes it possible to deliver low-cost health insurance is the affiliation of entire communities. The distribution of the costs reveals that in the absence of insurance, 5 per cent would face medical costs of Rs 1,000 or more per household member per quarter, 10 per cent of households would presumably pay more than Rs 500 per household member per quarter, and 20 per cent of the households could expect to pay more than Rs. 200 per member per quarter. With such high “outlier” costs, the poor cannot afford to remain uninsured! And the only question then is that the insurance package for the poor must cover the “outlier” expensive cases, which are the events that spell disaster on entire households, and for which no poor household can save enough to pay on its own. Insurance can reduce its own exposure to the outlier risk is by transferring it to reinsurance (which single households cannot do). Policy insight: The contribution that the poor are willing and able to pay toward their health insurance represents a significant part of the cost of insurance; but they cannot also pay the reinsurance premium, which others must cover (e g, government subsidy). This solution provides help to those who are willing to help themselves, and is cheaper, simpler and more equitable than alternatives.
Conclusion
Seven myths have been laid to rest in this article. We can state without doubt that there is a solvent market for health insurance among India’s poor. However, tapping this huge market is contingent on product development that starts from a deep understanding of the clients’ needs and wants. The insurance products must be adapted to the heterogeneity of the consumer-base. Community-based endeavours can be a powerful resource for process innovation and for gaining acceptance by the target population, because nobody is closer-to-client, and no other body is as effective as communities in implementing the local ethos that makes the local economy run. The communities are also best placed to mediate an optimal balance between needs, costs, resources and supply, all of which are context-specific. Minor adaptations of products developed for richer clients (and often in Europe or the US) are unlikely to find many willing takers in the slums and villages, where reality is completely different. Becoming familiar with the needs and priorities of the poor requires considerable innovation in processes; the logistics for data mining, access to clients, selling and servicing of the health insurance must be adapted to the context-specific social dynamics and local infrastructure. However, the long tail of the cost distribution, implying that outlier costs can be devastatingly high, makes it necessary to link local communities with a financial mechanism of reinsurance and risk equalisation, thereby enabling micro heath insurance schemes to benefit from economies of scale.
[liEmail: daviddror@socialre.org
Economic and Political Weekly November 4, 2006
