Effect of Access to Free Formal Rural Healthcare on Labour Market Outcomes

9152 words (37 pages) Full Dissertation in Full Dissertations

06/06/19 Full Dissertations Reference this

Disclaimer: This work has been submitted by a student. This is not an example of the work produced by our Dissertation Writing Service. You can view samples of our professional work here.

Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of UK Essays.

1. Introduction

While there is much known about the labour market impacts of access to better quality healthcare in developed countries, relatively little is known about these effects in developing countries. Given that illness impacts the poor disproportionately more, it is important to understand whether access to better healthcare is welfare improving overall. A large body of literature shows that better health increases labour supply, wages and productivity in developing countries (see e.g. Strauss and Thomas 2007). Better quality healthcare through insurance, by mitigating the adverse impacts of health shocks either through a reduction in the time spent looking after sick individuals or through directs effects on health can impact labour market outcomes.[1] It is also plausible that health insurance alters the utility associated with leisure. On the one hand, people may enjoy leisure more if they are healthier. However, risk averse consumers might enjoy leisure less if leisure results in more uncertainty about health care expenditures (Currie and Madrian, 1999). Understanding how the provision of health insurance impacts labour supply decisions of the household is important not only because labour market outcomes are indicators of overall welfare but also because provision of non-contributory insurance to the underprivileged may be an important redistributive policy tool.

This study examines whether access to free formal rural healthcare has a causal effect on labour market outcomes by studying a plausibly exogenous policy change. In 2008, India introduced the Rashtriya Swasthya Bima Yojana (RSBY) which provides each below poverty line (BPL) household access to health insurance at no cost at selected private and public hospitals in the country. Designed to increase access to healthcare and reduce the financial burden of health expenses, RSBY has enrolled about 36 million households as of 2014 (about 12% of the total number of households in the same year). While enrolling into the program is voluntary and thus households self-select into the program, I take advantage of the phased rollout of the program from 2008 through 2012 at the district level and employ a differences-in-differences approach to examine the impact of the program on labour market outcomes. In particular, I assess the value to a household of an increase in access to health insurance in terms of hours of labor supply gained and improved efficiency of labor allocation between home and market work and between child and adult labor.

Individual out-of-pocket payments (OOP) constitute the largest proportion of total health expenditures in India and are one of the highest in the world. According to the National Health Accounts Data (2004-05) individual OOP account for 78 percent of total expenditures while government expenditures account for only 20 percent (Azam, 2016).  Given the large share of OOP payments in health care in India, RSBY relies on providing cashless health services to the beneficiary households without any paperwork with the use of smart cards with only a marginal enrollment/renewal cost of 30 Indian Rupees (INR) (or about $0.5) per year.

While there is a growing literature on the impact of RSBY on enrollment into insurance and on out-of-pocket expenditures (Johnson and Krishnaswamy 2012; Nandi, Ashok, & Laxminarayan, 2013; Sun, 2010; Rajasekhar, Berg, Ghatak, Manjula, & Roy, 2011), there is little evidence on the labour market impacts of the program. Illness exposes households in developing countries to increased risk. Risk pooling strategies such as social safety nets or formal health insurance are often lacking and this leaves households to adopt costly risk coping strategies which include depleting savings, selling durable assets, and reallocating labour away from productive activities. Health insurance reduces the price of treatment faced by a household, and hence has the potential to both, lower the burden of out-of-pocket health expenditures on the household and the risk of impoverishment as well as improve healthcare utilization and health. Given the scope for government intervention, large scale healthcare reforms have been introduced by many countries in the last two decades. These usually entail the provision of free or heavily subsidized health insurance. For instance, in 2003, China adopted a new health insurance system, the New Cooperative Medical Scheme (NCMS), in rural areas where 80% of people did not have health insurance of any kind (Wagstaff et al., 2009).  In this context, using nationally representative data from the Indian National Sample Survey (NSS) this paper seeks to understand the extent to which a national rural health insurance program, the Rashtriya Swasthya Bima Yojana (RSBY), impacts labour market outcomes of households in India.

Existing evidence on the impact of the RSBY program on out-of-pocket expenditures is mixed. Karan, Yip and Mahal (2017) find that the reduction in out-of-pocket health expenditures and catastrophic inpatient spending was small and statistically insignificant. Using primary data from Amaravati district in the state of Maharashtra and Patan district in the state of Gujara, Rathi, Mukherji and Sen (2012) and Devadasan et al. (2013) respectively, find that a large proportion of families enrolled in RSBY continue to incur OOP spending despite the fact that RSBY is a cashless scheme with no co-payment or fees at point of service. However, Johnson and Krishnaswamy (2012) find that the program led to a decrease in out-of-pocket expenditures and thus, total medical expenditures. Similarly, Azam (2016) finds that while out-of-pocket health expenses were not impacted by the program, expenditures on medicines declined significantly in rural areas. One reason why Johnson and Krishnaswamy (2012) seem to find a significant reduction in outpatient expenditures is because they look at the impact of the program only up to 2009-2010 while Karan, Yip and Mahal (2017) include program effects up to 2011-2012.

In contrast, with respect to hospital utilization, the literature consistently finds that RSBY led to an increase in utilization rates. Azam (2016) finds a positive impact on hospital utilization by households in rural areas. Further, he finds that conditional on having received medical treatment for major morbidity, the program led to a rise in the probability of being hospitalized and being treated by a government doctor in rural areas. He also finds that RSBY led to lower expenditures on medicines in rural areas. Similarly, Johnson and Krishnaswamy (2012) find that the program led to an increase in the number of households that have had a hospitalization case.

There is limited research on the labour market impacts of non-contributory health insurance, especially in developing countries. Moreover, most of the research that exists focuses on the shift in employment from the formal to the informal sector. This is because in most countries (e.g. Mexico, Thailand, Colombia) non-contributory health insurance is only offered to those working in the informal sector. In a study from Colombia, Camacho, Conover and Hoyos (2013) find that public health insurance led to an increase in informal employment of approximately 4 percentage points.

The present study is different from previous work in at least four ways. First, this is the only study that examines a health insurance program, in a developing country setting, whose eligibility criteria was not linked to one’s labour market status but instead had a pure income criterion. Thus, this study indirectly examines the ability of health insurance to alleviate poverty because of health shocks and thereby impact labour market outcomes. Second, this is the first study that examines the impact of variations in household access to health insurance, an important determinant of relative productivity for individual members, on short term allocations across different types of work for children. Third, it contributes to the nascent literature on the economic consequences of provision of health insurance in developing countries. While much is known about the impact of health shocks, little is known about how households reallocate labour supply as a result of access to higher quality healthcare. The exception in this regard is Adhvaryu and Nyshadham (2012) who estimate the effects of higher quality healthcare through access to formal healthcare on labour supply of acutely sick individuals in Tanzania and find that the ability to choose formal care led to individuals spending more labour hours on the farm. Given that impoverishment due to health-related expenses is a leading cause of poverty in India, it is important to understand the extent to which social insurance programs trigger direct and indirect behavioural responses that can mitigate this impoverishment. Finally, my paper contributes to the literature on gender differentiated impacts of health insurance. It links the impact of health insurance to a wider literature on the role of economic development in helping either mitigate or reinforce gender inequalities.

Overall, I find significant positive effects of the RSBY program on the labour supply of women in the private casual labour market. On average, household level access to free health insurance increases the number of days spent in the past week by women on private casual labour by 0.286 days in early treatment districts (i.e. districts that received RSBY treatment on or before March 2010). Mean days spent in private casual work by women at baseline in 2004-05 is 0.45 days in a week. Thus, an impact of 0.286 days is large; time spent in private casual work approximately increases by 50 percent for women. This translates into approximately 15 more days of work annually for women in the private labour market. There was also a significant increase in time spent, for women, in the private agricultural wage market by 0.254 days in the past week. Importantly, this increase in labour supply in private sector work is accompanied by a significant decrease in the number of days spent in the past week on domestic work by 0.694 days.

I examine various mechanisms that may explain the relationship between the program and women’s labour supply. I consistently find that, for all labour supply measures, the effect of accessing health insurance is largest for women in households with fewer working age members and higher number of dependents. Further, I provide suggestive evidence that the impact of the program may be due to both the, impact on health and healthcare utilization, and the reduction in time spent at home by women in caregiving tasks. For men, there is no significant change in labour supply in private sector work. However, I find that time spent by men in self-employment activities is significantly lower in the post-RSBY period in early districts and time spent out of the labour force is consistently higher. This points to the fact that there is increased healthcare utilization by men which improves their health and enables them to reduce time in self-employment activities at home. However, men who are now healthy and attempt to seek work in the private labour market, find it harder to obtain employment which leads them to withdraw from the labour force. Indeed, I find that time spent out of the labour forcr for men increases significantly. Moreover, for households with children, I find that children substitute for adults in the two domains that men and women are moving away from: domestic work and self-employment activities. Consequently, I find that for children, time spent at educational institutions declines. Overall, given positive changes in female employment (and significant negative changes in male self-employment), I attribute the effect of the program to a reduction in caregiving tasks for women and increased healthcare utilization by men.

Vulnerability to health shocks is a serious concern in India. Nearly 10% of India’s population record out-of-pocket expenditures in excess of 25% of non-food consumption (World Bank, 2008). There is a significant literature that shows that households in India use productive assets to smooth consumption in the face of income shocks. Rosenzweig and Wolpin (2017) find that in rural India, bullocks, while also used as sources of mechanical power in agricultural production, are sold to smooth consumption in the face of income shocks. Consumption is therefore smoothed at the cost of crop production efficiency. Jodha (1978) again using data from India, argues that sales of productive assets when faced with shocks (a drought in this case) is very common. Even though the specific shock that I consider in this paper is different, the basic story remains. The poor engage in costly consumption smoothing techniques in the short run in response to catastrophic health shocks which can have severe long-term impacts, both in terms of productive activity and labour market outcomes.

In recent times, the World Bank has ratified the elimination of OOP expenditures for health care at the time of use through the prepayment of insurance as an important step toward reducing the financial hardship associated with paying for health care (Hsiao et al., 2007). In this context, accounting for close to USD 160 million in the union budget of 2012-13, RSBY is one of the largest programs of its kind in a developing country. In this setting, this study contributes to the ongoing policy debate regarding the benefits of increased access to quality healthcare. While existing literature has failed to find a large impact of RSBY on out-of-pocket health expenditures, it has consistently found an increase in health care utilization rates. This implies that the poor are availing of the RSBY program which impacts their labour supply; but the maximum coverage of Rs. 30,000 (about 440 USD in 2010-11) under the program is insufficient. Thus, impacts on labour supply due to RSBY are not the result of an income effect of the program, but instead are a consequence of the improved health of men and reduced time in caregiving tasks for women.

The paper is organized as follows. Section 2 provides a brief review of the previous developments in the literature on health insurance in India. It also provides a description of the RSBY. Section 3 provides a basic theoretical framework linking individual labour supply decisions to the availability of health insurance. Section 4 discusses the data. Section 5 describes the identification strategy and the econometric model. Section 6 presents the results. Section 7 examines the impact on health and healthcare utilization and Section 8 performs robustness checks. Section 9 concludes.

2. Context

2.1 Background of Health Insurance in India

Illness and poverty are closely linked to each other. The poor are often unable to smooth consumption during periods of ill health and catastrophic health expenses often push families into poverty (Xu et al., 2007; Rajasekhar et al., 2011). Vulnerability to shocks is an important cause for deprivation (Dercon, 2001). This is compounded by the presence of weak financial instruments and the adoption of sub optimal coping mechanisms such as asset sales, migration and child labour (Rosenzweig and Wolpin, 2017; Haughton and Khandker, 2009). According to Krishna (2004), the most common reasons given by poor people for their descent into poverty are ill health and health related expenses (Rajasekhar et al., 2011). According to Shahrawat and Rao (2012) the extent to which health related out-of-pocket (OOP) expenditures exacerbate poverty is at a maximum for households below the poverty line (BPL) in comparison to those above (i.e. the poverty gap increased by Rs. 10.45 due to OOP payments for health care for BPL households compared to Rs. 1.49 for above poverty line households).[2] The only government owned health insurance company that exists in India is the ‘New India Assurance’. New India Assurance provides health insurance at slightly lower rates than private providers. One has to purchase health insurance directly through this company as one would purchase from a private provider. Consequently, prior to the introduction of the RSBY, access to health insurance of any kind depended on one’s ability to pay for insurance through the marketplace.

In the absence of adequate social safety nets households can become impoverished not only because of out-of-pocket expenses and ill health, but also because of missed work, disability or premature death (Fan, Karan and Mahal, 2012). These asymmetries in health service access and quality have resulted in large health inequalities between the insured and the uninsured. To bridge this gap, the government launched the Rashtriya Swasthya Bima Yojana (RSBY) in 2008. Given that nearly 25 percent of Indians are below the poverty line and do not have access to formal safety nets, health insurance can potentially reduce the financial risk arising from the combination of out-of-pocket medical expenditures and income losses.

Available evidence on the impacts of Indian health insurance on household economic outcomes is nascent, in part because until recently there were no insurance schemes with large enough coverage to be of policy interest. Acharya et al. (2012) provide a comprehensive review of the literature on social health insurance schemes and the extent to which they enhance access to care and offer protection from financial risk to poor households. The few small-scale schemes assessed for their likely effects are ‘community-based health insurance’ (CBHI) and ‘micro-insurance’, and are predominantly operated by non-profit, non-government and civil-society organizations (see e.g. Aggarwal, 2010; Devadasan et al., 2004; David Mark, Radermacher and Koren, 2007; Ranson, 2002). Aggarwal (2010) assesses the impact of the Yeshasvini CBHI scheme, the largest of these, in the state of Karnataka by using propensity-score matching for 4,109 households. She estimates that people who enrolled in Yeshasvini insurance significantly reduced total payments from savings, income, and other sources by up to 74 percent and total borrowings by 30 – 36 percent.

While the impact of health insurance on health utilization and expenditures is relatively well studied, evidence on the impact of rural health insurance on labour market outcomes in developing countries is limited. Given the importance of such programs in mitigating the effects of a contraction in incomes due to health shocks and hence impacting labour supply, this lack of evidence seems stark. In one of the first papers in this domain, Wagstaff and Manachotphong (2012) examine the universal health coverage scheme in Thailand which was rolled out in 2001. Using staggered rollout to identify program effects, the study finds that universal coverage encouraged employment especially among married women, reduced formal-sector employment among married men but not among other groups, and increased informal-sector employment especially among married women.

Levy and Meltzer (2008) have argued that social insurance schemes, like rural health insurance, can have unintended consequences. Since such schemes provide people with an incentive to work in the informal sector rather than the formal sector, individuals can obtain similar if not identical health coverage without making any additional health specific contribution. However, given that employer provided health insurance in India is rare coupled with the fact that, unlike health insurance schemes in other countries, RSBY is not restricted to informal workers, a decrease in the share of formal sector workers post-RSBY is unlikely. The most critical part of RSBY is that it allows households to transfer risk to the government while simultaneously increasing access to health services. Impacts on labour supply are thus driven either through impacts on own health or through decreased time spent caring for sick dependents.

2.2 Overview of the Rashtriya Swasthya Bima Yojana

In 2008 the Indian government committed to expanding health insurance nationwide. Subsequently a National Health Insurance Scheme or RSBY was announced in early 2008.[3] The scheme aims at improving access of quality medical care to below poverty line (BPL) families for treatment of diseases involving hospitalization and surgery through an identified network of health care providers.[4],[5]  It also aims to provide financial protection against catastrophic OOP expenditures (RSBY, 2009). RSBY provides cashless coverage up to Rs. 30,000 (about 440 USD in 2010-11) each year to each enrolled household for hospitalization procedures in empaneled private or public hospitals. This coverage is large in purchasing power terms: the median level of income of the average person in my sample is about 616 USD and the average household spends about 7-10% of their income towards healthcare costs. The policy covers hospitalization, day care treatment and related tests, consultations, medicines and pre- and post-hospitalization expenses. While the scheme covers most surgical and non-surgical procedures, not all diagnostic tests are covered. Pre-existing conditions are included as is maternity care. Further there is a provision for transport allowance subject to a cap of Rs. 1000 per year. The RSBY does not cover any expenses related to outpatient treatment.[6]

Under the scheme each BPL household can register up to five members. Each household is issued a smart card in which the names, ages, photographs and thumb impressions of enrolled members are recorded. Beneficiaries can obtain cashless treatment by presenting the smart card at any participating (‘empaneled’) hospital. Hospitals are issued with the technology required to access the data stored in the cards. Treatment costs are reimbursed to the hospital by the insurance company according to fixed rates. The scheme aims to improve poor people’s choice of care provider by empaneling both private and public hospitals. There is also a provision for ‘splitting’ a card so that migrant workers can avail of RSBY benefits from any empaneled hospital in the country.

2.3 Enrolment and Utilization of RSBY

RSBY is implemented at the state level. State governments choose a public or private insurance company to implement the scheme in their state through a bidding process. The financial bid is essentially an annual premium per enrolled household. The premium cost for enrolled beneficiaries under the scheme is shared by Government of India and the state governments in a 75:25 ratio. Once a qualified insurer wins the bid in a particular district, the premium payments that it would receive depend on the number of BPL households that it manages to enroll during a four-month period.[7]

The state government is expected to provide the selected insurance company with the list of eligible (BPL) households in the state. In order to ensure widespread coverage states are required to prepare in advance a roadmap for the enrollment campaign in each village in a district or taluk, and give the village prior notice of the enrollment team’s visit.[8] A list of eligible households is required to be posted on the enrollment station in each village. This ensures that eligible households know in advance that they are eligible for the scheme and can plan to be present (or not) when the enrollment team visits the village. Smart cards are issued on the day of enrollment. Households have to pay Rs. 30 (45 cents) as annual registration fees. During the enrollment process households are handed pamphlets that contain a list of participating hospitals, a summary of the procedures and tests that are covered under the policy and a toll-free telephone number in each district through which one can obtain information about the scheme. Prior schemes of the government have been plagued by insufficient publicity and lack of prior notice. The enrollment process of the RSBY aims to correct both these flaws. Policies are issued for one year and are renewed on an annual basis. Beneficiaries can utilize the scheme from the start date of the policy which is usually three to four months after the enrollment process has been completed.

Existing literature points to low to moderate levels of awareness of the RSBY program among the target population. Rajasekhar et al. (2011) examine the implementation of RSBY in the state of Karnataka from the initial political and planning processes through the first six months of operation. They use a large survey of eligible households and interviews with empaneled hospitals in the state. The authors find that six months after the introduction of the program in early 2010, almost 85% of eligible households in the sample were aware of the program. Further, about 68% had been enrolled.  However, knowledge about how to obtain treatment under the scheme was still low. Moreover, a large proportion of beneficiaries were yet to receive their cards. Das and Leino (2011) find that an Information and Education Campaign in Delhi to popularize RSBY is not associated with higher enrollment.

The state-wise number of beneficiaries enrolled from 2012-2015 in major states is given in Appendix Table I. As of September 2016, more than 41 million health cards (signifying enrolment in RSBY) have been issued, covering almost 150 million poor people, with nearly 460 districts participating in the program. Although the share of eligible households enrolled in the program (enrolment ratio) was 57% nationally, there was considerable variation across districts. Enrolment ratios varied from a low of 3% in Kannauj and 6% in Kanpur Dehat districts in Uttar Pradesh, to nearly 90% in many districts of Chhattisgarh and Kerala (Karan, Yip and Mahal, 2017). As Karan et al. (2017) point out, overall enrolment rate of 57% suggests that a significant fraction of households were not enrolled despite being eligible for the program. There could be at least three reasons for this: one, it is possible that these households are living in districts that have not yet participated in RSBY (Palacios et al., 2011; Sun, 2011). Two, even in participating districts, enrolment agencies may not yet have reached all eligible households (Sun, 2010; Rathi et al. 2012). The third reason for the low enrolment rate is that, eligible households may have simply fallen through the cracks and ended up not getting enrolled (e.g. lack of adequate outreach by enrolment agencies, or absence at the time of enrolment) (Rajasekhar et al. 2011; Sun, 2010; Rathi et al. 2012; Devadasan et al. 2013). Finally, though large scale ‘adverse selection’ in RSBY is unlikely as the scheme is non-contributory, it is plausible that some households with healthier family members may have less incentive to enroll in RSBY (Sun, 2010).

2.4 Rollout of RSBY

RSBY was introduced at the national level in 2008 and each state government was expected to adopt the scheme in a phased manner over the next five years. States are responsible for selecting districts for inclusion into the scheme. In proposing districts for inclusion there were three criteria that were considered. Selected districts should have: (a) an adequate network of hospitals/health facilities which meet minimum standards for service delivery, (b) an adequate presence of potential intermediaries which can partner with health insurers to ensure effective outreach and, (c) basic infrastructure necessary for implementation such as roads and electricity. States began rolling out the program from 2008 in different districts and in different years.

3. Theoretical Framework

This section formalizes the intuition that in the presence of health shocks, the standard labour-leisure choice model of a poor household, without adequate coping mechanisms, will be affected by their demand for health insurance. In this model, it is assumed that health insurance can potentially impact the labour supply choice in two ways. First, it may impact the labour force participation margin of the working age population. This could also include the formal-informal work margin since it is largely people with informal jobs who stand to benefit most from the RSBY.[9] This could arise both from improved health as a result of the program or the ability to now enter the labour market since caregiving tasks are reduced.

Second, the program can change the number of hours spent in the labour market. This can happen in two ways. First, better health impacts productivity. If we assume that people don’t alter their consumption of leisure because of better health, then health insurance should increase labour supply. Further, RSBY covers hospitalizations only. Since major morbidities are more likely to result in hospitalizations than short term illnesses, the potential for RSBY to impact labour supply is greater.

Table 1 presents health related summary statistics in the last 4 weeks from the Indian Human Development Survey (2004-05). From the table, it is clear that a significant amount of time is lost because of major and minor illnesses especially for those in the age group 0-5 years or in the age group 55 and above. Access to free healthcare has the potential to treat these illnesses quickly, which can impact labour supply for both men and women. Secondly, in the absence of health insurance, individuals in households with a higher number of dependents would spend a large amount of time in caregiving activities for the sick. RSBY can potentially reduce the burden of caring for sick dependents and allow household members to reallocate time towards the labour market. Given that household members from uninsured households spend a significant of time in helping cope with illness, RSBY can potentially help in freeing up household resources towards more productive activities. For a typical household on average about 5 days in a month are spent with at least one household member being unable to work due to short-term illness and about 4 days because of some major illness (see Table 1).[10] It is also plausible that the provision of RSBY could lead to health gains; particularly among dependents.[11] Prompt medical attention coupled with the greater vulnerability of these sub-groups means that RSBY is well-equipped to generate health impacts. This could be happening through both, the availability of and accessibility of prompt medical attention that the RSBY generates. Johnson and Krishnaswamy  (2012) find that health care utilization has increased after the introduction of the RSBY. They find that RSBY has increased hospital utilization rates by as much as 20% of the pre-treatment baseline utilization rates.

3.1 Basic Model

The influence of these incentives is captured in a simple extension of the agricultural household model. I introduce a health production function, h(.). Time spent at home is modeled as a household public good, such that individual utility depends on leisure and the home production of all members via h(.), such that leisure and home production affect household utility both directly and indirectly through their impact on health. In this framework, utility, given a set of household characteristics, φ and resource endowments E, is an increasing function of per capita leisure, consumption, and home production. Home production, in turn is determined by the following three parameters: total hours of household time at home, an exogenous parameter, HI, which reflects the households access to health insurance, and τ, which includes all other inputs the household has access to such as paid private and public doctors or other social programs. In this model, time spent at home can have two components: one, time spent at home and unable to work because of a major morbidity and two, time spent at home because of caregiving tasks.

I make the following assumptions. First, I assume that households maximize per capita leisure and not the leisure of individual members. In the empirical section, I provide individual level estimates. Second, I assume that there is no outside market for home production. This missing labour market can be justified because it is uncommon to employ outside people to engage in caregiving tasks of family members.[12] Third, I assume that leisure and home production are perfect substitutes.[13] Moreover, I assume that home production includes, among other things, time spent at home and unable to supply labour because of a major morbidity and time spent on caregiving tasks. Finally, I assume that this is a unitary household and all household members face a wage, w.[14]

Let the labour hours of household members be divided between work at home, Hfand work in the outside market Ho. Time spent at home, Z is divided between work at home, Hfand leisure, L. The value of labor at home is given by the production function q (Hf), while the value of work outside the home is the market wage, w.[15] Households then maximize utility, U (

;

; h; E; φ), subject to per capita consumption

x̅, per capita leisure

l̅and health, h and h=h (Z, HI, τ). That is, the health production function depends on home production, access to health insurance, and other medical inputs. Here U(.) and h(.) are twice continuously differentiable and increasing in each argument. The choice variables for the household are: Hf, Ho, X, L and h where L =

∑i=1Nliand X =

∑i=1NXi(for i = 1, …, N household members). The constraints facing the household are:

h = h (Hf + L, HI, τ)

pX = wHo + q (Hf)

T = L + Ho + Hf

L; Ho; Hf; X ≥ 0

where q(.) satisfies decreasing marginal productivity (q’> 0, q’’<0). Normalizing prices to one, the household’s optimization problem can be written:[16]

maxHo,Hf      U   T-Ho-HfN   ;  wHo+qHfN  ;  h(T-Ho, HI, τ)

The first order conditions for an interior solution (Hf > 0; Ho > 0; Hf + Ho < T) are:

wUx̅N = Ul̅N +Uh̅ hH0

qHfUx̅

=

Ul̅

The first, first order condition establishes that at the optimum households equate the marginal value of an additional hour of outside labor with the marginal utility of leisure. The second first order condition establishes that households also equate the marginal utility of leisure with the marginal value of an additional hour of work at home. The solution to these set of equations implicitly define demand functions for labour hours in the outside market and home production which depend on HI, w and τ:

H*f = Hf (w; HI; τ)        

H*o = Ho (w; HI; τ)

Differentiating with respect to HI yields the following predictions:

∂Hf ∂HI<0; ∂Ho ∂HI>0

For households involved in both types of labor, an increase in access to health insurance will increase work hours in two scenarios. One, health insurance, by increasing healthcare utilization and thus health, will decrease time spent at home for those suffering from a major morbidity. Two, work hours at home will decrease for those involved in caregiving tasks. In both cases, work hours in the outside market will increase.[17] Intuitively, this reflects the fact that an exogenous increase in access to health insurance corresponds to a decrease in the household’s need to spend time on home production activities, thereby lowering the opportunity cost of outside labor force hours. At the corner solution, Hf = 0,

∂Hf ∂HI=0; ∂Ho ∂HI>0. That is, households with zero home production hours ex-ante (Hf = 0) will increase total hours in the labour market after the program. Thus, in aggregate, increasing access to health insurance decreases work hours at home and increases hours outside the home.

3.2 Extensions

Given average consumption level, x, the effects found above are decreasing in the number of working age members, N.

∂2Hf* ∂HI∂N>0; ∂2Ho* ∂HI∂N<0

In large households, it is likely that some members stay at home, independent of either whether they’re sick or because of caregiving concerns, thus large households should be less distorted by health shocks and the concomitant impacts on labour supply.[18] For households with fewer dependents, the labour market response is ambiguous. If the household has a high number of working age members then we should see a smaller labour market response through both mechanisms (improvements in healthcare utilization or decrease in caregiving tasks). However, if there are few working age members and lots of dependents then we might see a larger labour market response.

4. Data

4.1 Program Rollout Data

Information about the rollout of the RSBY comes from administrative records maintained by the Ministry of Health, Government of India. This is available online on the RSBY website from where this information was accessed. Figure 1 (in the supplementary appendix) maps the district wise rollout of the program as of September 2015. In the figure, ‘rounds’ refer to the number of years of coverage for a particular district. For instance, those districts in say Round 5, have been covered for 5 years under the program as of 2015.

4.2 Household Survey Data

To study the effect of RSBY on the labour market, I use repeated cross sections of a nationally representative district level Employment and Unemployment Survey (NSS Survey) carried out by the National Sample Survey Organization (NSSO). My analysis uses cross-sectional data for the years 2004-2005, 2007-2008 and 2011-2012 which are the 61st, 64th and 68th rounds respectively.

Beginning in the 1950s, the NSSO has conducted nationwide surveys with successive rounds. The surveys conducted in 2004-2005 (61st Round), 2007-2008 (64th Round) and 2011-2012 (68th Round) are large-sample surveys. All the rounds used in this paper are representative at the district level. The surveys are implemented by a stratified multi-stage sampling design where the first-stage units (FSUs) are 2001 census villages in rural areas and urban frame survey blocks in urban areas. Each district of a state is included as part of either the rural or urban stratum of that district. Beginning on 1 July and ending on 30 June (of the following year), households are sampled evenly in each of four quarterly sub-rounds, with equal numbers of sample villages and blocks (i.e. FSUs) allotted in each sub-round. The NSS oversamples some types of households and therefore provides sampling weights. All statistics and estimates computed using NSS data in this study are adjusted using these sampling weights.

The NSS includes information on various individual and household level characteristics for individuals of all ages including age, caste, religion, marital status, employment status and household consumption. My identification uses the phased rollout of the program at the district level. Districts are administrative units within states. While constructing the sample the following exclusion rules have been used:  one, the states of Tamilnadu, Karnataka and Andhra Pradesh have been excluded since these states had similar state run rural health insurance programs during the period under consideration in this study.[19] Two, I also exclude the 3 Union Territories: Andaman & Nicobar Islands, Dadra & Nagar Haveli and Daman & Diu. Three, individuals between 18 and 60 years have been included for the adult labour market outcomes, while those between 6 and 17 years have been included for the child outcomes. Four, I drop observations that have missing information for age and gender. Finally, I only include districts that have rural populations. The final sample includes households from 531 districts.

Implementation of the RSBY began in 2008-09. Thus, the 2011-12 wave (68th round of the NSS) represents the post-intervention period. The 2004-05 and 2007-08 waves (61st and 64th rounds of the NSS respectively) represent the pre-intervention period. Districts that began implementing the RSBY in 2008 did so in the latter half of the year from August to December, and thus the 2007-08 NSS survey (64th round) can be treated as pre-program data since it is canvassed from July 2007 – June 30, 2008, before any district began implementing RSBY.[20]

4.3 Outcomes of Interest

My main outcomes of interest are individual measures of employment. I construct the employment outcome as follows. The NSS includes detailed questions about the daily activities of all persons over the age of four in surveyed households for the last seven days. For each day and each activity, the NSS survey records whether the activity was performed at an intensity of 0, 0.5 or 1 day. I then compute for each person, the number of days in the past seven days spent into six activities: (a) private casual work (b) private salaried work (c) private agricultural wage work (d) domestic work (e) not in the labour force and (f) self-employment. Private casual work and private salaried work together constitute private wage work. I also examine private agricultural wage work separately which can come under either private casual work or private salaried work. Not in the labour force includes attendance of educational institutions, pensioners, disabled and beggars. Domestic work could arguably be categorized as not in the labor force. However, given that most households engage in small-scale agriculture, many activities could equally well be categorized as domestic work. Additionally, domestic work also includes time spent in caregiving tasks and thus I study this employment measure separately.

4.4 District Level Information

To construct district controls I use data from two sources: first, I use individual-level data from the 2002-2004 District Level Household Facility Survey, Wave 2 (DLHS-2) aggregated to the district level. The District Level Household Facility Survey is a nationwide repeated cross-section survey which is representative at the district level. The main district controls from the DLHS-2 are: the proportion of villages in a district connected by a road, proportion of villages in a district with a primary health center, proportion of villages in a district with a government hospital, proportion of villages in a district with a health sub-center and the average distance of a village in a district to the nearest town. I also use the 61st Round of the National Sample Survey (NSS 2004-05) to control for baseline district-level characteristics –  fraction of scheduled castes and scheduled tribes, illiteracy rate, male and female labor force participation and fraction living under the poverty line.[21] These controls are time invariant and I interact them with a dummy for post-program status to pick up trends correlated with the controls.

5. Empirical Framework

Previous studies have used several different approaches to identify treatment effects of the RSBY program. For example, both Karan et. al (2015) and Johnson and Krishnaswamy (2012) use NSS surveys in 1999-00, 2004-05 and 2009-10. In the former study, the authors treat all eligible households residing in RSBY districts as treated and all eligible households residing in non-RSBY districts as non-treated. They proxy eligibility of the households by restricting their sample to bottom two quintiles of consumption expenditure. In the latter study, the authors first match districts based on certain characteristics and then use a difference-in-differences strategy across RSBY districts vs non-RSBY districts.  Ravi and Begkvist (2012) also use a similar district level difference-in-difference strategy and identify intent-to-treat (ITT) effects. Azam (2016) identifies RSBY beneficiary households based on a household’s response to a direct question about the RSBY card and estimates the average treatment effect on the treated (ATT) using a matching with difference-in-difference strategy.

5.1 Defining Treatment and Control Groups

While the eligible population was initially only the Below Poverty Line (BPL) households, this was expanded in later years to cover other defined categories of unorganized workers.[22],[23] In this context, I employ two definitions of treatment and controls groups. In Sample I, all ‘poor’ households in RSBY implementing districts are taken as the treatment group with the poor in non-RSBY districts being the control group. This is similar to the strategy followed by Karan et al. (2015).  I use households in the bottom two quintiles of consumer expenditure as a proxy for ‘poor’ households.[24] However, given that by 2011-12 the eligible population also includes National Rural Employment Guarantee Scheme (NREGS) workers and other unorganized sector workers, in Sample II, I use an expanded sample and include all households in the bottom two quartiles of consumer expenditure as a proxy for ‘poor’ households.

Because RSBY effects might vary with years of exposure, I consider two discrete cut-off points, March 2010 and June 2012, to identify two treatment groups with differing lengths of program (RSBY) exposure. These are: (a) poor households living in districts which began participating in RSBY on or before March 2010 (‘early’ treatment districts) and, (b) those living in districts which began participating between April 2010 and June 2012 (‘late’ treatment districts). Out of a total of 531 districts there are 191 districts which introduced the policy on or before March 2010 (these will be called the ‘early’ districts), 200 districts that introduced the policy between April 2010 and June 2012 (these will be called the ‘late’ districts) and 140 districts that either introduced the program after June 2012 or have yet not introduced the program (these are the ‘control’ districts).[25]Appendix Table 2 details the state-wise number of districts in early and late treatment groups.

5.2 Intent-to-treat effect – Identification Strategy

State governments played a major role in deciding the order in which RSBY was rolled out at the district level. While labour supply considerations and in particular, female labour supply considerations were never factored into program placement decisions; even so, a major methodological concern remains that placement decisions may not be orthogonal to other factors that could affect labour market outcomes. Since districts that received the program early could have been purposively selected based on variables that are correlated with labour market outcomes, a simple comparison of early and late districts is not likely to be informative. Thus, I compare changes over time between districts that received the program early (‘early’ districts) and districts that received the program later (‘late’


[1] According to the efficiency wage theory the link between health and labour supply would be stronger in less healthy populations, and to the extent that health insurance impacts own health, this could lead to impacts on labour supply as well.

[2] 1 US Dollar = 63 Indian Rupees.

[3] The Labour and Employment Ministry of Government of India was as in charge of RSBY, however, effective from April 1, 2015, RSBY has been moved to Ministry of Health and Family Welfare.

[4] The Below Poverty Line (BPL) is a threshold to identify poor households that require government aid. The BPL list used for RSBY is based on a census undertaken in 2002. It uses 13 socioeconomic parameters such as food security, literacy and sanitation and uses different criteria for rural and urban geographies to identify BPL families (Azam, 2016). There are concerns about exclusion criteria in the 2002 BPL list. To overcome this, some states cover additional households that they consider poor (Krishnaswamy and Ruchismita, 2011).

[5] The decision to include NREGA workers in late 2009 effectively expanded the coverage to non-BPL families (source: http://archive.indianexpress.com/news/insurance-scheme-to-include-nrega-workers/511643/). There is little information on the overlap between the BPL list and NREGA rolls, but some estimates suggest that half of NREGA workers do not have BPL status (Azam, 2016 & Palacios, 2010).

[6] RSBY also includes many day care surgeries/procedures which do not require stay at hospital. A list of day care surgeries covered under RSBY are: hemodialysis, parenteral chemotherapy, radiotherapy, eye surgery, lithotripsy (kidney stone removal), tonsillectomy, D&C, dental surgery following an accident, surgery of hydrocele, prostrate, few gastrointestinal Surgery, genital surgery, surgery of nose, throat, ear, and urinary system, treatment of fractures/dislocation (excluding hair line fracture), contracture releases and minor reconstructive procedures of limbs which otherwise require hospitalization, laparoscopic therapeutic surgeries that can be done in day care, identified surgeries under general anesthesia, and any disease/procedure mutually agreed upon (Azam, 2016).

[7] While more than one insurer can operate in a particular state, only one insurer can operate in a single district at any given point in time.

[8] The insurer usually contracts out the enrollment work to a smart card service provider (SCSP) who organizes the enrollment process under the supervision of the insurer and the State Nodal Agency. Besides the enrollment team, a designated district-level government officer must be present to oversee the enrollment process. Specifically, this official, in his role as Field Key Officer (FKO), must verify the identity of the household head and must insert his own smart card into the enrollment application to issue the personalized smart card.

[9] Section 2 outlines the details of the scheme. Eligibility was based on one’s poverty status. A household was deemed eligible for RSBY if the per capita consumption expenditure of that household was below the government specified poverty line. While one’s employment status was not included in the definition of eligibility, most informal workers are part of BPL population.

[10] Here major illnesses refer to: cataract, tuberculosis, heart disease, leprosy, cancer, polio, paralysis, epilepsy, mental illness, asthma and diabetes.

[11] Levy and Meltzer (2008) review the literature on the impact of health insurance on health.

[12] I could however, extend this model to incorporate extended family members.

[13] While this may seem unreasonable because leisure that one spends inside or outside the home is different from

leisure constrained as home production, inside the house; making this assumption does not change the comparative

statics of the model.

[14] Assuming equal wages for all family members might be an unrealistic assumption. It might be more reasonable

to assume wmen ≥ wwomen. However, that should not change the main predictions of the model.

[15] Incorporating a market for hired labor in home production does not affect the model’s predictions. Inseparability

in this model comes from the lack of substitutability of household members between the labour market and the

production of health, not q(.).

[16] For the remainder of the analysis, household characteristics and resource endowment are assumed to be fixed and omitted from the arguments of the utility function.

[17] This is under the assumption that there is no time that is spent on home businesses. This is because there will be a negative program effect on those households who spend a larger fraction of their time in home businesses. This is not an unreasonable assumption since, the sample being examined is poor and a very small proportion of these households are engaged in home businesses.

[18] It is also possible that smaller households with only a single member working in the labour force and many dependents, such as a single mother with children, could also see a smaller labour supply response since that member needs to work in labour market since he/she would have no time to spare for caregiving even in the absence of the program.

[19]At the state level, by 2012, Andhra Pradesh had Rajiv Aarogyashri scheme, Karnataka had Yeshasvini and Vajpayee Arogyashri scheme, while Tamil Nadu had implemented Kalaignar scheme (Forgia and Nagpal, 2012). Azam (2016), Johnson and Krishnaswamy (2012) and Karan et al. (2015) also drop these states from their sample.

[20] Concerns of bias due to inter-district migration is likely to be low. Migrating from a rural district to another rural district for employment is uncommon. According to Imbert and Papp (2015), rural to rural migration for employment among adults aged 18-60 with secondary education or less, within the last one year is less than 0.1%. Similarly, the proportion of adults from 18-60 with secondary education or less who migrated for employment from rural to urban areas in the past year is 0.11 percent.

[21] The poverty rate is constructed using Round 61 (2004-05) of the NSS Consumer Expenditure Survey.

[22] The unorganized workers category covers building and other construction workers registered with the welfare boards, licensed railway porters, street vendors, National Rural Employment Guarantee Scheme (NREGS) workers who have worked for more than 15 days during the preceding financial year, beedi workers, domestic workers, sanitation workers; mine workers, rickshaw pullers, rag pickers, and auto/taxi drivers.

[23] The decision to include NREGA workers in late 2009 effectively expanded the coverage to non-BPL families. There is little information on the overlap between the BPL list and NREGA rolls, but some estimates suggest that half of NREGA workers do not have BPL status (Palacios, 2010).

[24] According to Karan et al. (2015) a comparison of households’ self-reported BPL status and consumption expenditure per capita in 2004-05 and 2011-12 suggests that the two lowest per capita expenditure quintiles account for about 65 per cent (more than 70 per cent in the RSBY intervention districts) of households with BPL status.

[25] Broadly speaking, my use of staggered implementation as an identification strategy follows Gerber et al. (2008), who assess the impact on religious participation of the repeal of ‘’Blue Laws’’ in U.S. states, and Field (2002) who studies a nation-wide titling program in Peru.

Cite This Work

To export a reference to this article please select a referencing stye below:

Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.

Related Services

View all

DMCA / Removal Request

If you are the original writer of this essay and no longer wish to have the essay published on the UK Essays website then please:

McAfee SECURE sites help keep you safe from identity theft, credit card fraud, spyware, spam, viruses and online scams Prices from
£124

Undergraduate 2:2 • 1000 words • 7 day delivery

Order now

Delivered on-time or your money back

Rated 4.1 out of 5 by
Reviews.co.uk Logo (30 Reviews)