· Description of the intervention
India’s health system was planned largely as a three tiered referral system of primary health care centers, district hospitals and specialized hospitals where care would be sought largely through the public sector and financing would be based entirely on a supply-side model. By the 1990s, it was quite clear that the system was broken. Studies showed that (a) the bulk of visits and out-of-pocket expenditures were incurred on a parallel private system of care (80 percent of all primary visits are to the private sector); (b) that the quality of care in the government sector was low due to the lack of incentives and; (c) that the government sector was barely progressive while investments in hospital care were regressive. In the 1990s, along with a package of reforms, the government slowly opened up insurance markets to private providers. Yet, by 2007, less than 15 percent of the population carried any kind of insurance and the vast majority of the insured were found in the formal sector and higher income workers. Health insurance among the rural poor was limited to a handful of microinsurance programs covering less than one percent of the population. Simultaneously, analysis of household survey data showed that illness-related shocks were one of the largest reasons for households falling into poverty.
To redress these problems, the government launched a flagship health insurance product for the poor in 2008. India’s Rashtriya Swasthya Bima Yojana (RSBY) offers a micro-insurance product for households designated as “below the poverty line” (BPL) and aims to cover up to 60 million households throughout the country over the next five years. The objectives of the RSBY are to provide financial protection for households affected by major health shocks and improve health outcomes for poor households. The RSBY tests an innovative approach to introducing pay for performance into a large-scale government-funded catastrophic insurance program by combining capitation payments with choice of care providers at the beneficiary level.
The RSBY is a demand-side, voucher-like intervention with several innovative features. First, insurance companies are selected by competitive bidding in each district and receive a premium for every household enrolled by them in the scheme (typically paid 75% by the national government, 25% by the state government). Second, insurance companies empanel in-patient care facilities (ICFs); they then reimburse ICFs for in-patient care provided to enrolled households. In sharp contrast to usual practice, these ICFs may be either public or private and for public facilities, the reimbursement received from the insurance company for an enrolled household seeking care at the facility can be retained by the facility in a trust managed by the facility, called a Rogi Kalyan Samiti (or RKS). Thus, private and public facilities face the same financial incentives, though intra-facility bargaining may be different. Third, enrolled households can seek care from any ICF and face the same set of prices. By empowering consumers to choose among competing facilities at the same time that they are given a demand-side subsidy, the hope of the scheme is that the increased competition will lead to better quality care.
Under the RSBY, BPL households (as identified by the concerned states’ BPL identification survey) must enroll and receive a smart card at the cost of Rs.30 per household per year. A maximum of 5 members may enroll from any family. This includes the head, spouse and up to 3 children/parents of the head of household. The smart card permits any registered member of the household to visit any in-patient care facility (ICF) for in-patient care for a pre-specified list of more than 700 surgical or medical procedures. The prices for each procedure are pre-specified by the state, and an enrolled household can use a maximum of Rs.30,000 of in-patient care each year. Transactions are cashless for households and require only the fingerprint-verification of the user to deduct the cost of the procedure from the remaining smart card balance.
Currently, the RSBY has been rolled out in over 100 districts in 17 states and in the first 16 months has covered more than 7 million households with nearly 25 million individuals. The smartcard data system that automates and centralizes data collection is working, and administrative data on enrolled households and procedures that have been performed at ICFs is available on a centrally maintained server. In addition to the rich administrative data available through the program, household surveys of over 5,000 households eligible for the RSBY were conducted in 2008 in several states.
· Primary research questions and outcome indicators
As the program is rolled out in the pilot districts, the World Bank has been approached by three states (Gujarat, Kerala and Uttar Pradesh) to provide assistance with Monitoring and Evaluation, supported by a loan to the Government of India. Part of the M&E effort requires identifying important new policy initiatives for the scheme, and piloting these initiatives in one or more of the three states to provide feedback to the state and to the GoI on the efficacy of these new policies. At this time, two important questions have been identified on which information can be provided through impact evaluation and the proposal is therefore focused on these.
The first set of questions is on the utilization of the program and uptake rates. Which BPL households choose to enroll in the RSBY? Given that enrollment rates varied highly across the pilot districts, what can be done to increase enrollment? Do sicker households enroll in the program? Can information campaigns increase take-up?
The second set of questions examine, for households that do enroll in the RSBY, the determinants of and barriers to utilization of the insurance benefits. Early pilot data shows that visits by enrolled households to ICFs were about the same as in the latest round of the National Sample Survey (NSS) of Indian households, which might be surprising given that pre-existing conditions are covered by the RSBY and there may be latent demand for treatment. Understanding factors associated with utilization of the insurance coverage is necessary for further improvements to the program.
Both questions can be examined using administrative transaction data (use of insurance) and data on household characteristics from the household survey subsamples.
· Evaluation design/ identification strategy. Include information on the unit of analysis and sample size
The issues of program uptake and benefit utilization will be studied through rigorous randomized evaluations. One such experiment will examine an information campaign designed to increase enrollment and another will examine health camps designed to improve utilization of insurance benefits by reducing diagnosis costs for households.
In Delhi, where initial enrollment rates were extremely low, a pilot information campaign was conducted by a local NGO to inform a sample of 7,500 BPL households randomly selected from a list frame of BPL households about the RSBY scheme using household visits and other strategies. A follow-up survey of a random sample of about 2,000 BPL households that had and had not received the intervention was then conducted after the enrollment period for RSBY had ended. This data, which has already been collected, can be used to examine the characteristics of households that chose to enroll and households that were induced to enroll by the information campaign. Because the targets of the information campaign were selected randomly, the impacts of the intervention on uptake of the insurance can be identified.
Two features of this evaluation bear elaboration. First, the current policy guidelines of the RSBY provide full incentives for insurance companies to enroll heads of households and then to minimize the number of members enrolled in the household. Since the enrollment of the head implies the issuance of an insurance card, the premium payment remains the same for any marginal member enrolled within the household. Data from 24 completed districts indeed show that the average number of household members enrolled (2.6) is far below the average number of eligible members in the household according to the BPL data (4.8). Therefore, the evaluation will focus not only on household enrollment but also the number of members enrolled with and without information. Second, the information campaign had two components. One component was the prominent display of posters and general information about the scheme; the second was visits by the team to individual households to explain the RSBY scheme and field queries about the scheme. Given that the first component was at the community level, the evaluation will identify the additional impact of household visits over and beyond the effect of community level information.
A second intervention, which has not yet been conducted, will examine the determinants of benefit utilization in the program through a village health camp intervention. The intervention will be modeled on the health insurance scheme in Andhra Pradesh, which predates the RSBY and mandates that hospitals as well as health insurance staff conduct health camps in villages. At these health camps, individuals who arrive are given health check-ups and, where required, are triaged to hospitals for further care. The health camp component of the Andhra Pradesh program is widely viewed as central to its (reputed) success in providing health insurance to a poor and not very well educated population. In the absence of such health camps, individuals seek hospital care on their own without a formal system of referrals.
Our model of the intervention is as follows. In a sample of villages chosen with equal probability from the list frame of all villages covered in the RSBY pilot, the government or NGOs will be contracted to run village health camps in which RSBY enrollees are screened locally for covered health conditions and then triaged to ICFs where services covered under the RSBY can be performed. The effect of the health camps on benefit utilization (and on future enrollment in the RSBY) can be examined using the administrative data on benefit utilization collected through the smartcard, as well as through household surveys in areas that did and did not have health camps. Sample size is still under determination, but will be sufficient to measure differences in utilization for subpopulations of enrolled households. Of particular interest is the possibility of using both NGOs and the government, so that we can not only compare the effects of the health camps, but also the relative efficacy of either implementation model.
Recent conversations with the government have provided greater clarity about the proposed form of this intervention. A model that is being discussed (but has not been finalized) is as follows:
a. Health camps will be held in 250 locations at headquarters of “Gram Panchayats”—typically a group of contiguous villages with a combined population of 20,000 individuals.
b. A camp will be held in each Gram Panchayat once in the year. On average, 25 camps will be held every month and the 250 camps will be held throughout the year, where the allocation to Gram Panchyats is conducted in an experimental fashion.
c. There is a possibility of home-visits to households who were recommended hospital care. If this is done in an experimental manner as well, further variation can be introduced at the household-level to examine finer hypothesis of the kinds of information that matter for hospital take-up.
Two sources of heterogeneity are particularly important in interpreting the results from such an exercise: the availability of hospitals and the quality of care in these hospitals. Although it will be difficult to conduct block-randomizations given the overall sample size, we will look at the effect of the health camps across villages that are far/close to hospitals. However, we do not have sufficient funds to measure the quality of care in hospitals and this important source will thus not be studied under the proposed evaluation. Neither do we plan to look at important, but more resource intensive issues, such as the crowding out of care in hospitals.
· Evaluation team: identify principal investigator(s) and describe roles/responsibilities for each team member. Distinguish between investigator(s) and evaluation advisors. Attach CVs for all proposed team members
On the World Bank side, the evaluations will be coordinated by the project TTL Robert Palacios (South Asia, Social Protection Unit) in combination with a team consisting of (in alphabetical order) Jishnu Das (DECRG), Xiaohui Hou, (SASHD), Arianna Legovini (DECVP), Jessica Leino (SASHD) and Changqing Sun (DEC). The evaluations will be undertaken in close collaboration with the relevant state government nodal agencies for the RSBY program (including one or more of the three states that have requested M&E support), as well as with the Ministry of Labour and Employment, Government of India..
· Describe the primary risks in completing the evaluation as designed. Include information on the client buy-in, potential obstacles, and contingency plans
The Government of India has complete buy-in to the project and the several State governments have requested assistance with M&E work to improve and expand the RSBY. Based on preliminary results from the household information campaign in Delhi, the Government has decided to expand the intervention to all eligible households in Delhi for the next enrollment period. The risk of the Government canceling or substantially altering the project is limited.