January 10, 2013
There is a growing ideological divide among Indian health researchers and policymakers with respect to the future direction of the country’s health policy. Those in favor of supply-side policies (i.e. the government should provide free or subsidized healthcare) contend that India’s health insurance scheme for the poor (called Rashtriya Swasthya Bima Yojana or RSBY) among the world’s largest programs of its kind is akin to an overly expensive white elephant, which brings more pride for the policymakers than actual results. Believers in RSBY consider this criticism to be mere white noise, and argue that demand-side policies such as RSBY have a complementary role to play. In this blog, I examine arguments from both sides of the debate.
The two sides of health policy
Targeted supply-side policymaking has historically been the cornerstone of India’s public healthcare delivery mechanism. As I discussed in my previous blog post, until the National Health Policy (2002), India generally took a piecemeal approach to health policy, often focusing on individual diseases or conditions. The 2002 overhaul promised to streamline public healthcare delivery through the integration of such standalone policies, improvement in quality of services, and renewed focus on lagging regions. This was followed by the National Rural Health Mission (NRHM 2005), which introduced elements of decentralized service delivery, with additional resources for the poorer states.
Once every few years, there is a renewed push for strengthening the supply side of public healthcare in India. The most recent one came in 2011, in the form of the High-level expert Group (HLEG) report on universal health coverage, instituted by the Planning Commission of India. Authored by leading public health researchers of India, this report once again advocates the integration of individual health policies into a National Health Package (NHP), emphasizes the need for more supply-side resources, and mentions innovations such as private care under contract from the government (where public healthcare delivery is inadequate). The authors envision a nation where each citizen will have access to free healthcare, through a mix of public and private providers.
However, this time around circumstances are different. Demand-side incentives have entered the domain of Indian health policy, starting with the successful safe motherhood scheme (Janani Suraksha Yojana) of 2005. In 2008, India launched RSBY, a health insurance scheme for poor families, under a public-private partnership model. At a nominal out-of-pocket enrollment fee of Rs. 30 (US $1 = Rupees 50, approximately), officially poor (below the national poverty line, or BPL) families of up to 5 members are covered for more than 700 medical treatments and procedures (at government-set prices) by the RSBY. The maximum annual coverage for a family is Rs. 30,000. Healthcare services are provided by government contracted hospitals, both public and private, and beneficiaries use a RSBY smart card with pre-loaded benefits, without the need for cash transactions.
With a current enrollment of over 33 million poor families, more than 12,500 empanelled hospitals, and a final target enrollment of 300 million people, RSBY is among the world s largest health insurance schemes. The coverage is provided by private health insurance companies, while the government pays the insurance premium (varies by district, negotiated between state government and insurance provider). The total cost of the program is divided between the central and state governments, with the center bearing a 75% share of the cost.
So, what s stopping RSBY from achieving its full potential, and why is there skepticism regarding its future? To understand this, we need a quick overview of the benefits of health insurance in developing countries.
Health insurance schemes: An international review
During recent decades, various low- and middle-income countries have implemented large social health insurance programs (not to mention thousands of other smaller community level schemes). Among the noteworthy programs are 3 Chinese health insurance schemes (1998, 2003, and 2007) that have already covered more than 1.1 billion people, and have a target of universal coverage. Other examples include Vietnam (1992), Philippines (1995), Nigeria (1997), Ghana (2003), Indonesia (2005), and Tanzania (2009).Acharya et al. (2012), a large systematic review, examines many social health insurance schemes, and their impact, across the developing world.
What are some potential benefits of such programs? Insurance could improve healthcare utilization (and possibly health, although that is difficult to measure), reduce out-of-pocket medical expenditure, provide additional financial protection so that households experiencing a health shock will not need to reduce consumption or finance healthcare through borrowing and asset sales.
However, as Acharya et al. (2012) find, the results are mixed. The social health insurance schemes across the developing world vary in terms of their target population, insurance co-payment mechanism, coverage level, and administrative characteristics. Therefore, it s no surprise that while some studies find a positive effect on one or more outcomes mentioned above, others find no effect, or even negative effect (e.g. in China and Zambia).
It is also important to note that an alarming number of studies analyzing the impact of health insurance suffer from methodological inconsistencies. Enrollment in health insurance, especially if voluntary, and if the scheme targets a particular population subgroup, brings a lot of methodological biases that a researcher must mitigate. First, a voluntary enrollment implies that certain individuals may self-select themselves into the program; and they may be systematically different from those not covered by the scheme. For example, people who often get sick may choose to participate more (adverse selection), creating an unfavorable risk pool for the insurance provider. While it is desirable from a societal point of view that the needy be covered, it may underestimate the overall effect of the insurance (it will look as if the insured people are experiencing more health shocks).
This bias is also present if there is selective program placement (i.e. the insurance scheme is rolled out only in needy areas). On the other hand, in developing countries/markets with incomplete information about insurance programs, there may be a so-called elite capture. People with higher standards of living, and hence better information and social connections, may be able to obtain benefits from the subsidized insurance schemes. In these cases, the perceived effect of the program will be overestimated.
Difference-in-difference methods (sometimes used in conjunction with matching methods), where data from pre- and post-implementation of an insurance program are used, may eliminate the initial (time-invariant) differences in unobservable characteristics between the insured and non-insured groups, and thereby reduce the self-selection or selective placement bias. However, biases originating from time-varying unobservable factors will still be unaccounted for.
These are some dangers in using observational data to draw conclusions about the effectiveness of social health insurance. The true impact may only be revealed by studies which are randomized in nature, such as the RAND experiment, Seguro Popular experiment, and SKY Cambodia experiment .
Evaluating RSBY
RSBY rollout has never been randomized. In fact, it has all the hallmarks of what I call a causal inference troublemaker the scheme is means-tested (only BPL people are eligible, recently expanded to include some other socioeconomically backward groups), enrollment is voluntary, and the scheme has unusually high enrollment rates with large regional variations that may be associated with socioeconomic and institutional factors or regional bottlenecks. Therefore, all RSBY evaluation studies are likely to suffer from some level of estimation biases, and claiming the success or failure of the program on the basis of observational estimates is extremely risky.
Methodological issues are the crux of an ongoing debate among RSBY researchers, as outlined by Vellakkal and Ebrahim (2013). A study by Selvaraj and Karan (2012) argued that RSBY and other health insurance schemes may have increased out-of-pocket expenditure for the poor in the state of Tamil Nadu. The authors were quick to denounce RSBY as a failure, and called for a replacement health financing mechanism. Dilip (2012) raised some methodological concerns about this study, which were rebuffed, in rather harsh language, by the authors of the original study. Vellakkal and Ebrahim (2013) supported some of the concerns of Dilip (2012), and proposed some analytical refinements for the original study.
Another brand new cross-sectional study of RSBY conducted in Gujarat (news article here) has come out with preliminary findings which indicate no significant difference between the out-of-pocket expenditures of insured and non-insured groups. An interesting fact noted in the study is that most of the out-of-pocket medical expenditure by RSBY beneficiaries is related to drug purchases. This only bolsters the argument in favor of both demand and supply side programs, whereby RSBY will encourage a poor individual to demand a treatment, which then should be provided in full (procedure and drugs) by the healthcare provider.
The rise of out-of-pocket expenditure post-RSBY may not necessarily be bad. There may have been a lot of pent-up demand for treatment, which never materialized due to high costs. The fact that most out-of-pocket expenditure is associated with drug purchase (also seen in Ghana), or not the actual medical/surgical procedure, signifies the presence of unmet needs. As long as there are no moral hazards (i.e. people engaging in risky health behaviors, now that they are covered), the rise in utilization and out-of-pocket expenditure may well be a short-term phenomenon, which will stabilize in the medium to long term. There may also be a seasonality of claims (people using coverage more towards the end of the year, right before the expiration), which would be interesting to analyze, if RSBY ever releases utilization data.
On the upside, Fan et al. (2012) points to some encouraging outcomes in Andhra Pradesh. Using fairly robust methods involving difference-in-difference and matching, the authors find that the state health insurance scheme significantly reduced out-of-pocket medical expenditure, more so for inpatient cases.
As this overview has made clear, the evidence on the intended effects of RSBY remains inconclusive. Further research involving better data and identification strategies are needed before the program s merit can be judged.
RSBY: Still a ways to go
Of course, there are administrative concerns about RSBY. The financial viability of the scheme in the medium and long term is uncertain. A study by Dror and Vellakkal (2012) presents a bleak picture, where the total premium cost of complete RSBY coverage (all BPL families) for the central government is estimated to be as high as Rs. 33.5 billion, or 0.3% of India s union budget in 2010-11. Instead, the actual union budget allocation for RSBY was only Rs. 3.15 billion in 2010-11, and Rs. 4.6 billion in 2011-12. In 2010-11, the budgeted resources could cover only about a third of the actual enrollment (not complete enrollment) cost of Rs. 9.29 billion.
Finally, as with many government schemes, there are problems with system leakage, insurance frauds, and other inconsistencies. A host of RSBY working papers examine the structural shortcomings that prevent RSBY from reaching the neediest, and materializing its full potential in protecting the poor from health and financial shocks. There is need for oversight to avoid unfortunate insurance frauds that may have a devastating effect on the reliability of the program, but then again RSBY is still developing (even the safe motherhood scheme, India s other flagship demand-side incentive program, had some rather hilarious hiccups).
