https://orcid.org/0000-0001-8468-3141
Figures
Abstract
The presence of ex-ante moral hazard could undermine the potential gain from expanding health insurance coverage in developing nations. To test the proposition, this study utilizes a nationally representative longitudinal survey with Indonesia’s health insurance for poor policy in 2014 as the quasi-experimental case study. The country represents developing nations that undergo a massive and rapid expansion of health insurance coverage. The empirical approach combines a matching and difference-in-differences method to obviate potential bias of the selectivity nature of health insurance provision and time-invariant unobserved factors. The findings suggest the presence of ex-ante moral hazard in the form of the less people using trash cans associated with the introduction of the subsidized health insurance premium. The results add empirical findings of a negative side effect of expanding health insurance coverage in developing nations.
Citation: Gitaharie BY, Nasrudin R, Bonita APA, Putri LAM, Rohman MA, Handayani D (2022) Is there an ex-ante moral hazard on Indonesia’s health insurance? An impact analysis on household waste management behavior. PLoS ONE 17(12): e0276521. https://doi.org/10.1371/journal.pone.0276521
Editor: Alison Parker, Cranfield University, UNITED KINGDOM
Received: January 6, 2022; Accepted: October 10, 2022; Published: December 15, 2022
Copyright: © 2022 Gitaharie et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Data is available from https://www.rand.org/well-being/social-and-behavioral-policy/data/FLS/IFLS/ifls4.html.
Funding: The authors are grateful for the support of the PUTI Grant by the Directorate of Research and Development, Universitas Indonesia (NKB-2043/UN2.RST/HKP.05.00/2020).
Competing interests: This research and publication has no competing interests related to any patents, patent application, or products in development or for market.
1 Introduction
The governments of developing nations use various schemes to reach universal health coverage [1]. Among countries’ schemes, Indonesia implements subsidized health insurance premium to poor households known as Premiums Assistance Recipient (PBI), launched in 2014 (2) and as an earlier version as Askeskin [2]. One of the key motivations was to protect these vulnerable groups from catastrophic health spending upon illness [3–5]. Despite this merit, the scheme is believed to potentially create an ex-ante moral hazard [6–8]. In this study, we revisit the issue by investigating the preventive behavior of lower-income households, in the sense of living hygienic, particularly domestic waste management, after they obtain subsidized health insurance from the GoI (Government of Indonesia). We expand the hypothesis of the effects considering the conventional adverse effect in the form of the increased careless behavior and the improved health awareness because of the increase in healthcare visits.
The presence of moral hazard is an important consideration in developing an optimal health insurance design [9]. Its existence may reduce the effectiveness of health insurance provision. Discussion on ex-ante moral hazard has received little subsequent attention in empirical work of health insurance [10]. The existing empirical research in several countries that confirm the presence of ex-ante moral hazard in health insurance are more related to unhealthy lifestyle, such as consuming high-calorie foods [11], smoking and drinking alcohol [12–14], becoming obese [15], doing less physical activities/exercises [14], and spending a large amount of time in sedentary activities ([11] and [16]). Insured households in Ghana are also less likely to use bed nets to prevent malaria [17]. To the best of our knowledge, literature on ex-ante moral hazards that relate health insurance to a clean environment or hygienic living conditions is lacking. The relation is still not unveiled. For that reason, we intend to fill the empirical gap by investigating the impact of subsidized health insurance provision for lower-income households on their waste management behavior in Indonesia.
The empirical approach of this study utilizes a nationally representative longitudinal survey in Indonesia, namely IFLS (Indonesia Family Life Survey) of the Rand Corporation. The data represents the nation’s population, and the samples mainly come from the country’s major islands, including Sumatra, Java, and Kalimantan islands. We exploit the variation of the household’s health insurance between the baseline period of 2007 and the end line period of 2014 as the most suitable recent data that is available for analysis. To the best of our knowledge, there is no other potential data for such purpose. The socioeconomic national survey (SUSENAS) has some thematic survey in particular year containing health aspect, such as in 2017. In addition, the Demographic Health Survey/DHS (SDKI) and the Basic Health Research Survey (RISKESDAS) are potential but do not contain our variables of interest. Nevertheless, these datasets are cross-sectional so they are less ideal for our purpose.
The focus of our study is to investigate whether there is a different behavior of waste management between the insured and uninsured lower-income households. We exercise differences in pre-treatment level between insured and uninsured to directly compare both groups’ characteristics and perform propensity score matching combined with a difference-in-difference procedure. These approaches are used to ensure that our impact estimate is a result of the subsidized insurance and not due to the common behavior of a specific group. We also run a heterogeneous analysis and coefficient stability for robustness.
This study aims to test the causal inference between the provision of subsidized health insurance for lower-income households and how these households manage their waste. This study contributes to empirical evidence by examining whether ex-ante moral hazard behavior exists among Indonesia’s poor and vulnerable groups. There are two plausible competing effects in which subsidized health insurance could affect individuals—adverse and positive effects. The adverse effect arises when the behavior of the health insurance holder does not take care of his health because he knows the insurance will reimburse his medical expenses if he becomes ill. The positive effect works through the availability of health promotion information and greater contact with medical professionals. Health insurance can be perceived as an important factor in increasing preventive care and possibly improving health behavior [18].
Based on our estimates, we find evidence of ex-ante moral hazard in the subsidized health insurance recipients on how they manage their household waste. The insured households are less likely to use trash can to dispose their domestic waste than the uninsured. The effects vary by demography and geography, suggesting some important implications for focusing the mitigation efforts. The finding of this study is expected to help the Indonesian government to design a better intervention concerning adverse side effect of health insurance subsidy.
The following section describes the plausible competing effects on recipients of subsidized health insurance. Section 3 presents the existing condition of household waste management and health insurance in Indonesia. Section 4 depicts the empirical strategy and data employed. Section 5 displays the empirical result and discussion. Lastly, Section 6 concludes the paper.
2 The competing effects of subsidized health insurance: Ex-ante moral hazard vs. the improved awareness of hygiene
Theoretically, there are two plausible effects in the discussion of health insurance: one positive and the other one is adverse. One positive effect of owning health insurance is that the owner would have more opportunities to communicate and interact with medical professionals. If healthcare professionals advise their patients about health risks, the increased interaction and the patients’ improved knowledge could lead to better health behavior [19, 20]. A study in Saudi Arabia finds that health insurance holders are more likely to seek medical attention—the effects are higher amongst non-Saudi nationals compared to Saudi Arabian citizens—and are encouraged to get specialized medical exams for high cholesterol, diabetes, and hypertension [21]. Meanwhile, a study in China explains that public health insurance (PHI) has heterogenous effects depending on age and income [22]. The findings show that the effect of PHI is stronger for both middle-aged and elderly. When it comes to the lower income group, PHI significantly increases the probability of having no chronic diseases, self-reporting good health, and good life satisfaction (mental health). The low-income group also tends to have a more significant increase in health care utilization and is more likely to feel relieved of a medical financial burden. In the rural area of Uganda, the presence of voluntary community-based health insurance increases the likelihood of utilizing a mosquito net and getting dewormed by 26% and 18%, respectively [23]. They hypothesize that health insurance has an impact through three different channels: the financial protection channel, the use of healthcare services, and the dissemination of information and social learning.
Even though health insurance is meant to protect the holders from health and financial risks, its ownership is not always followed by healthy living behavior. There are still insurance participants who behave recklessly, leading to an ex-ante moral hazard and resulting in an adverse outcome. To mention some of the ex-ante moral hazards in health insurance, as described earlier in Section 1 of this paper, are the findings of [11–17] which is in contrast to [23].
Mixed results on health behavior are also found in the case of the Affordable Care Act (ACA) study conducted by [18, 19, 24]. A study by [18] finds that the after-one-year ACA expansion increases insurance coverage, access, and the use of certain forms of preventive care; but they do not find evidence of ex-ante moral hazard. Meanwhile, the three-year ACA expansion also increases preventive care utilization but increases risky behavior, hence, ex-ante moral hazard occurs [24]. For the five-year ACA expansion, as in [18] finds the expansion increases utilization of certain forms of preventive care, reduces heavy drinking and smoking, and increases the probability of exercise [19].
Many studies, some as mentioned above, focus more on the effect of health insurance on lifestyle, for example, consuming high-calorie foods, smoking and drinking alcohol, doing less physical activities/exercises, spending a large amount of time in sedentary activities, and becoming obese. As far as we can tell, there is no literature yet on ex-ante moral hazards that link health insurance to hygienic living conditions. Poor living conditions which are commonly found in developing countries, including improper domestic waste management issues, are associated with health problems. This paper questions how would the causal effect of subsidized health insurance on domestic waste management in Indonesian households, specifically in the lower-income group.
3 National health insurance and household waste management: The Indonesia context
3.1 Indonesia national health insurance
Since its establishment during the colonial period, Indonesia’s health insurance program has evolved. One of the milestones is initiating universal health coverage in 2004 through Askeskin, which provides health insurance for the poor [25]. A study conducted by [2] shows that the Askeskin program has targeted the poor and the vulnerable, despite the non-trivial leakages to the non-poor. Askeskin has led to an increase in outpatient healthcare utilization among the poor. [26] extends the study of [2] by breaking down the samples into subgroups and investigates the impact of Askeskin on both outpatient and inpatient healthcare. The author finds that the program had a larger effect on the use of outpatient healthcare by females than males, but inversely on inpatient healthcare.
Askeskin then transformed to Jamkesda (Jaminan Kesehatan Daerah) in 2005 and Jamkesmas (Jaminan Kesehatan Masyarakat) in 2008 [27]. Both were dedicated for the poor and the vulnerable. In term of their sources of financing, Jamkesmas is financed by the State Budget (APBN-Anggaran Pendapatan Belanja Negara), whilst Jamkesda by the Regional/Local Government Budget (APBD-Anggaran Pendapatan Belanja Daerah) to cover the shortfall in funding from Jamkesmas in the region.
Intending to provide comprehensive, fair, and equal health coverage to the entire population, including the poor and vulnerable [28], the GoI launched JKN (Jaminan Kesehatan Nasional or the National Health Insurance) in 2014. Not only for free treatment, but JKN also provides the opportunity for people to be healthy by reducing risk and screening those at risk [29]. To that end, the BPJS (Badan Penyelenggara Jaminan Sosial or Social Health Insurance Administration Body) has curative and preventive action programs as the managing entity for JKN. JKN has covered 82.3% of Indonesia’s population as of March 2021, but it still has a long way to meet the 98% target by 2024.
To protect the poor and vulnerable, the GoI offers PBI (Penerima Bantuan Iuran-health insurance premium assistance beneficiary). The targeted group and the amount of contribution are both determined by the government—through the Ministry of Social Affairs Regulation No. 21 of 2019 and the Indonesian Presidential Regulation No. 64 of 2020, respectively. The criteria of the targeted include: (i) not having a source of livelihood and or having but do not meet basic needs, (ii) having expenses only to meet basic needs, (iii) being unable to afford for seeking medical attention, (iv) unable to afford for buying clothes once a year for household members, (v) able to send their children to junior high school, (vi) having the walls of the house made of bamboo/wood/wall in poor condition/low quality, (vii) having the floor of the house made of soil or wood/cement/ceramic with poor condition, (viii) having the roof of the house made of palm fiber/rumbia or tile/tin roof/asbestos with poor condition, (ix) having house lighting not from electricity or electricity without a meter, (x) having small house floor area, less than eight meters-squares /person; and (xi) having drinking water sources come from wells or unprotected springs/river water/rainwater/other. To these targeted groups, the health insurance contributions for PBI participants are paid in full by the government with 42,000 IDR per month per eligible recipient.
With the implementation of JKN, Jamkesda and Jamkesmas participants automatically become PBI participants. The PBI participation has increased by 40% since the scheme was first launched in 2014 to 2019. Fig 1 indicates that, as of July 2019, 83% of the total population is insured, and 60% of which is that of PBI.
Source: Social Security Agency for Health.
The other beneficiary group is the non-PBI which covers (i) wage workers and family members, (ii) non-wage workers and family members, (iii) non-workers and family members. The non-PBI participation has also grown by 132% for 2014–2019. Non-PBI participants, on the other hand, must pay contributions as depicted in Table 1.
Despite its growing participants, BPJS has faced main challenges-the number of participants remains below the target, there is still inequality in access to health services, and there are issues related to health service financing [30]. Since its first establishment, BPJS has still run deficits in its financial statement. It is likely because participant contribution is lower than the actuarial calculation value (see Fig 2). In addition, it is recorded that 25,326 companies manipulate employee data potentially cause revenue losses of IDR 6.19 billion [31]. Fig 3 shows that the total expenditure for health insurance has exceeded its contribution revenue since the JKN program first launched in 2014, except in 2016 and 2019.
Source: The Audit Board of The Republic of Indonesia.
Source: Social Security Agency for Health.
Table 2 shows that JKN utilization among PBI participants remains far below that of non-PBI participants. It indicates that the GoI needs to focus on increasing health literacy among the poor. People with poor health literacy do not look for medical services when needed. Despite health coverage [32], they habitually visit health care facilities only when their diseases have reached an advanced stage. Being well informed is essential to promote positive results in healthier lives.
3.2 Household waste management in Indonesia and related diseases
As in other developing countries, household waste management remains a major concern in Indonesia. The country is ranked fourth in the world in generating municipal solid waste (3.55%) after the US (4.40%), India (18.05%), and China (18.75%) [33]. Indeed, households in Indonesia generate waste the most (45.4%) [34]. The four largest types of waste are food waste (40.1%), plastic (17.2%), wood/leaves/twigs (14.2%), and paper/cardboard (11.9%) [34].
Most households do not manage their waste properly. Regardless where they reside, households burn (rural: 68.2%; urban: 34.1%; Indonesia: 49.5%), dispose their garbage in the river/sewer/sea (rural: 11.3%; urban: 4.9%; Indonesia: 7.8%), and anywhere improperly (rural: 10.1%; urban: 2.5%; Indonesia: 5.9%) [35]. Particularly, among the lower income group (quintiles 1 and 2) 55.95% of them do not manage waste based on our calculation with IFLS 2014.
Households need to be aware that waste evolves in time. Waste material composition becomes more complex and frequently contains toxins and harmful elements [36]. Waste sorting is, therefore, urgent to do. A survey by KataData Insight Center [37] indicates that 50.8% of households interviewed do not separate nor sort their domestic waste. They argue that—they do not want to bother with waste sorting (79%), waste will eventually be mixed at the temporary/final disposal sites (17%), it is useless to sort waste (3%), and other (1%).
The home environment is one crucial determinant of resident health [6]. People with lower incomes are more likely to have poor health [4, 38]. It is globally indicated that more than 8,000 people die every day from poor sanitation and hygiene conditions [39]. Based on BPS data in 2019, the Indonesia lower-income group or people living in poverty covers 9.41% of the population [40]. As for rural-urban division, 43.3% of population are living in rural and 56.7% in urban areas; and 13.86% living in an unclean environment [41]. The untreated waste—44% of 252 square meters of the estimated household waste production per day [42]–may become vectors of diseases, including malaria and dengue [36]. The number of dengue cases in Indonesia has increased from 68,410 in 2017 to 137,760 cases in 2020 [43]. In terms of home environment-related diseases [6–8], specifically, the subsidized health insurance recipient households suffer from chronic diseases, primarily hypertension, digestive disease, and arthritis/rheumatism (see S7 Table in S1 Appendix), and acute diseases—headache, runny nose, cough, stomach ache, skin infection, and diarrhoea (see S8 Table in S1 Appendix). As a preventive action, improving the home environment, sanitation, and hygiene could reduce the incidence of sanitation-related diseases.
Waste management in Indonesia is a reasonably complex issue. Indonesia faces both demand and supply constraints. There is an increasing volume and production of waste from the demand side as the population grows and urbanization rapidly expands. Slums in urban areas have also doubled—rising from 4.09 percent in 2018 to 9.04 percent in 2019 [42]. The public lacks knowledge in proper waste management. On the supply side, there is inadequate provision of waste facilities, equipment, and technology. Only 27.15% of household waste in Indonesia was collected by the garbage collector in 2017 [42] and served by 3,873 garbage trucks across 34 big cities in the county. The availability of garbage handcarts that usually serve door-to-door and temporary waste storage (TPS-Tempat Pembuangan Sementara) has declined by 20% and 4.84%, respectively, in 2019 [42].
4 Data and empirical strategy
4.1 Data
The data for this study comes from the Indonesia Family Life Survey (IFLS). IFLS is a comprehensive and nationally representative longitudinal survey conducted by Rand Corporation in Indonesia. The surveys consist of five waves in which we utilize the last two waves of IFLS 4 (the year 2007) and IFLS 5 (the year 2014) for our purpose of the study. The ethical review and clearance follow the IFLS ethical clearance. The survey covers socio-economic and health variables that allow us to identify household ownership changes toward health insurance and associate them with the reported actions of responsible health and environmental behavior representing the ex-ante moral hazard measures. The behavioral measures include toilet ownership, where the household drains its sewage, how the household disposes of its garbage, and whether the household participates in health funds by the community.
4.2 Empirical strategy
The main objective of our study is to examine the impact of lower income group household health insurance (SIht) ownerships status on ex-ante moral hazard behavior, measured by household’s waste management behavior measures (Yht). We use Propensity Score Matching (PSM) (34) concerning the subsidized insurance ownership selection issue in the pre-estimation step. Subsidized insurance in Indonesia is a household targeted program with the eligibility criteria of living under poverty. The matching strategy aims to create a comparable control group for those who own subsidized insurance. Therefore, our impact estimate will better reflect the implication of the subsidized insurance on waste management behavior and is not because of the different characteristics between the two groups. A similar approach of matching for analyzing subsidized insurance is a norm in the literature. See for example [17] or [2].
Some anti-poverty programs in Indonesia have imperfect targeting performance [44, 45], with PBI or Askeskin as no exception. There are cases of inclusion (non-eligible households receiving program) and exclusion error (eligible households but not receiving the program). Fig 4 indicates these targeting errors. However, the imperfect targeting allows us to construct a counterfactual group using similar households who did not receive the program at the baseline period (SIi,2007) with the following matching strategy:
(1)
in which Zh,2007 represent a set of household characteristics that the Indonesian government uses as the eligibility criteria for subsidized insurance targeting. It includes the age of the household head, per capita expenditure, household size, gender of household head, marital status of household head, house conditions (has a bamboo wall, soil floor, has the vehicle, has an electronic asset, has electricity access, and ownership of toilet). The estimation for
utilizes limited dependent variable estimator of a Probit model.
Source: Author’s calculation with IFLS 2007.
Subsequently, if both treated and comparison group resulting from matching strategy has similar (parallel) growth pattern of their potential ex-ante moral hazard behavior in the absence of subsidized insurance, we estimate Eq (2) to the observations that are in the common support of the propensity score matching obtained from Eq (1) with a double difference framework. For household h and time t, the main specification of the canonical difference-in-differences estimator to identify the impact of SIht ownership on waste management outcome (Yht) between pre and post period indicated by POSTt with the parameter of interest β3 is the following:
(2)
The point of estimate of β3 captures the difference-in-differences in the outcome variables concerning waste management between households that have SI and similar ones that do not have SI across times before and after SI implementation. The nature of endogeneity of measuring the impact of SI ownership on ex-ante moral hazard could originate from unobserved time variant (such as degree of risk aversion) and time invariant factors (such as expansion of supply-side health-related public infrastructure), apart from the selectivity problem. Therefore, to control for time-varying observable confounding factors, we include a set of household-level covariates of Xht that covers family years of education, degree of risk aversion, the availability of waste management supply at the district level, population density, and rural-urban dummy in addition to unbalanced covariates explained in Section 5.2. Last, we also include island-year dummies of ηrt to limit heterogeneity bias originated from macro aggregate shocks and the varying development level in the Indonesian archipelago. We also perform the coefficient stability test using the framework developed by [46]. The test uses assumptions of the maximum R-squared equals 1.3 times the existing model with complete control variables and “the proportionality assumption” of 0.545.
We also perform heterogeneous analysis using subgroups which include gender, education, income quintiles, expenditure per capita, location—both rural-urban and region, and waste service to observe the group that drives our main result.
5 Result and discussion
5.1 Defining treatment and control group
To identify the treatment and control groups, in IFLS, there are four categories of individuals based on their status of subsidized insurance ownership: always treated, never treated, newly treated, and untreated. Between 2007 and 2014, always treated individuals were those who identified held subsidized insurance in both periods. In 2007 subsidized insurance is elicited by Askeskin as the only available survey question item, whereas in 2014 it expanded into a wider type of subsidized insurance including Jamkesda, Jamkesmas, Jamkessos, Jampersal and JKN as the new terminologies and policies. Those who never had subsidized insurance in baseline or end line period are never treated. Newly treated did not receive Askeskin in 2007 but then held any subsidized insurance in 2014. Last, the untreated had Askeskin in 2007 but no longer had any subsidized insurance in 2014. We only use always and never treated as treatment and control group to avoid bias from subjects with switching status as they created negative weighting issues in the estimation [47]. Effectively, we use only 1,642 households as the treatment group and 5,677 households as the control group among a total of 10,676 balanced panel households. The flow of subsidized insurance status changes between 2007 and 2014 in IFLS is provided in Fig 5 and Table 2 Panel B. The descriptive statistics is presented in S9 Table in S1 Appendix. They also show that the expansion of subsidized insurance has been massive from 2007 to 2014.
Source: Author’s calculation using IFLS Wave 4 & 5.
5.2 Pre estimation
Table 3–Panel A presents the pre-treatment balance test results for the covariates to determine the probability of getting subsidized insurance. The result shows that most of the variables have a statistically significant difference between the mean for the control and treatment groups. These characteristics gap between the two groups indicates that subsidized insurance ownership is a selective event.
Furthermore, when we estimate Eq (1) the statistical significances are also picked up by the same variables in the t-test mentioned above. As presented in Table 3 Panel C, the result indicates that households with good economic conditions are less likely to participate in subsidized insurance. A household with higher per capita expenditure is less likely to be insured. Moreover, households who own electronic assets and vehicles are 1.8% and 4.5%, respectively, less likely to be insured. On the other side, households who live in a house with a wall made of bamboo and have no toilet at home are 6,1% and 5,2%, respectively, more likely to have participated in subsidized insurance.
To limit the potential bias from the selection mechanism, we perform PSM and calculate the propensity score of each individual. We obtained 35 off-support individuals using the matching method of an Epanechnikov kernel with a 0.06 bandwidth. The estimate suggests using 1,642 treatment individuals and 5,642 control individuals after excluding 35 samples from the untreated group. However, the balancing test in Panel B of Table 3 suggest that not all of the selection variables are balanced. Only household head gender is balanced between the treatment and control group but not for the rest. To anticipate bias associated with the imperfection in the matching step, we add all of the unbalanced covariates in the main estimates. The distribution of propensity score by treatment status and on-support is provided in Fig 6.
Source: Author’s calculation using IFLS 2007.
5.3 Main estimate
This study hypothesizes that participating in subsidized health insurance could lead to two possible effects in the form of a hygienic home environment. One is a more responsible behavior toward better hygiene and environmentally responsible behavior resulting from the indirect health educational-process obtained from health services. Second, the opposite direction originated from incentive effect in which insurance holders tend to be careless when the adverse health risk is covered.
The main results of estimating Eq (2) only for on-support samples are presented in Table 4 with full covariates. The stepwise estimations by a different specification of covariates are shown in S10 Table in the S1 Appendix section. We employ difference-in-differences regression to estimate an ex-ante moral hazard in subsidized health insurance provision on how households manage their waste. The coefficient of interest is the interaction between subsidized insurance and post-period. We found evidence of ex-ante moral hazard concerning the behavior of disposed waste in the trash can, but we see no impact for the other two types of outcomes. As shown in column (1) of Table 4, it is implying that 4 out of 100 insured households are less likely to dispose of waste in the trash can after receiving the treatment compared to their counterfactual.
The ex-ante moral hazard is typical behavior of an insured who does not mind about his/her health and tend to avoid preventive care due to his/her confidence that the medical expenses will be covered by the insurance company. Having insured does not necessarily improve subsidized insurance recipients’ knowledge of living a hygienic lifestyle. It contrasts to what is found in previous studies, i.e. [18] and [48].
To this adverse effect of subsidized health insurance provision, we suppose there is a low awareness of proper domestic waste disposal from the demand side, which may be due to a lack of knowledge and habit of doing so [49]. The subsidized insurance recipients may not be aware of the danger of unproper waste disposal since most subsidized insurance recipients have a low educational background. Solid waste harms human health. Household waste, in general, has contaminated water and soil in 21% and 2.7% of Indonesian villages, respectively. It has been reported that 28% of households do not have improved drinking water. In 2016, the contaminated river, which is still used for daily needs or as infiltrated groundwater, caused 17 million cases of diarrhea in Indonesia, 39.6% of which were untreated [42].
Our empirical result in Table 4 confirms [35] that individuals who were living in rural areas, have low educational attainment, and come from a disadvantaged socioeconomic background, tend to not properly dispose of their solid waste in the trash can. From the supply side, the lack of funding causes the limited availability of waste facilities and services in Indonesia [50]; and the distance to final disposal sites from home is not a walking distance. Households do not want to be too bothered about taking the trip. In addition to that, the GoI has issued laws and regulations on waste management, such as Law No. 8 of 2008 on the Management of Municipal Solid Waste and Presidential Regulation No. 97 of 2017 on the Indonesian National Strategy Policy on Managing Domestic Waste and Domestic Waste Equivalents. However, the enforcement is still ineffective, easily caught and seen by everyday improper practices and violation of the rules in the neighborhood without penalties. The GoI, then, must provide the needed facilities and infrastructures to overcome these habits. The GoI must also enforce the regulation effectively.
5.4 Heterogeneous effect estimates
We subsample observations by demography and geography to observe whether our estimates’ magnitudes are consistent or not across different sub-group of interest. Furthermore, the analysis identifies the priority target population for policymaking in reducing the ex-ante moral hazard by, for example, improving the waste management on the supply side. Overall, the government might target low-income groups, low education levels, and the population living in urban areas. The heterogeneous impacts are summarized in Table 5. The detailed results are as follows.
As for the outcome variable of disposing of waste to the trash can, lower-income households’ sub-population consistently to have ex-ante moral hazards, even with higher magnitudes of effects. In addition, the negative effects are also found for female-headed households, low-educated, households living in Non-Jawa Island, and households living in urban areas and population living with the presence of waste collection system. The treatment group of these sub-populations has a smaller number of households that manage waste in the range of 5.6 to 7.1 percentage points. We found the consistent null effect for outcome variables of burning trash and throwing the trash to land or open dumping.
5.5 Robustness check
Our estimates in Table 4 might not be completely free from biases. Therefore, we perform a coefficient stability test based on [46]. The results are summarized in Table 6. Overall, the adjusted coefficients are lower in absolute terms for disposing the waste to the trash can as the outcome variable (first-column) but higher for burning trash (second-column) and throwing the trash to land or open dumping (third-column). The adjusted coefficient in the first column makes the conclusion unchanged, it is still negative and statistically significant at ten percent level. There is an effect of subsidized insurance on the probability of households disposing of the waste to the trash can. It could be the case that the reduced number of households throwing the waste to trash can is shifted to burning trash. The inspection to the data reveals that there have been numbers of household shifting from throwing trash to land to become burning trash among the subsidized ones. It is about 176 households or about 39.64% of the total households practicing burning trash in 2014.
Moreover, the adjusted coefficient in the second column indicates the tentative finding that subsidized insurance increases the probability of households burning trash as the magnitude becomes larger. Last, the adjusted coefficient in the third column holds the case that subsidized insurance has no effect on the probability of households throwing the trash to land or open dumping. Based on the robustness test, our overall conclusion is unchanged.
6. Conclusion
This paper investigates the impact of a subsidized health insurance program on how lower-income households manage their domestic waste. We statistically find evidence of a projection of ex-ante moral hazard of fewer people to properly dispose of their waste into the trash can after the selected households obtained the program. On the external validity, we acknowledge the potential low base effect in which the point of estimates might be overstated compared to the recent situation as the share of the subsidized group becomes larger. Accordingly, we infer that our estimates serve as the upper bound of the effects. It indicates irresponsible behavior of the insured households toward their home environment, which would then imply their health condition.
The finding of this study will help the government not only focus on the curative action but more importantly on preventive ones by imposing stricter prohibitions for subsidized insurance recipients to keep their good behavior of disposing waste to trash can. However, the government needs to find more effective approaches for behavior change initiatives. Incentives are one example of the initiatives and are considered economically attractive for lower-income households. Another example is to improve household awareness of the benefits of domestic waste management. Social campaigns and marketing through various media, such as electronic, mass, and social media, are possible ways to raise awareness. Through this social dissemination, social learning is expected to occur. Households are then expected to engage more actively in waste management. The government can also increase waste sorting and recycling training for low-income households, as this will provide them with knowledge on how to make objects from recycled materials. This creates new opportunities for income-generating activities. Furthermore, social empowerment, particularly for women, can hasten the implementation of domestic waste management, including the 3R program (reuse, reduce, and recycle), and, ultimately, realizing the circular economy (zero waste).
Due to limited lower-income household knowledge, limited waste disposal facilities, and not all households being capable of paying for private garbage collector services, households tend to dispose of their waste improperly. Households then must constantly be informed of the danger of improper waste disposal and littering as it will bring long-term impacts to their physical and environmental health. The government should also prioritize the provision of adequate waste disposal facilities across the country. The imminent reduced improper waste disposal will lead to a healthier household living environment which, in turn, will reduce the government’s potential extra budget to mitigate health-related problems due to improper waste disposal and the pollution it may cause.
Acknowledgments
We thank Vid Adrison, Ph.D. and M.H. Yudhistira, Ph. D. for their constructive comments in improving the manuscript during the preliminary dissemination seminar at the Department of Economics, FEB UI.
We also thank the reviewers for their valuable input and suggestion in improving the manuscript during the review process for the publication.
References
- 1. Lagomarsino G, Garabrant A, Adyas A, Muga R, Otoo N. Moving towards universal health coverage: health insurance reforms in nine developing countries in Africa and Asia. The Lancet. 2012;380: 933–943. pmid:22959390
- 2. Sparrow R, Suryahadi A, Widyanti W. Social health insurance for the poor: Targeting and impact of Indonesia’s Askeskin programme. Soc Sci Med. 2013;96: 264–271. pmid:23121857
- 3. Geldsetzer P, Manne-Goehler J, Marcus ME, Ebert C, Zhumadilov Z, Wesseh CS, et al. The state of hypertension care in 44 low-income and middle-income countries: a cross-sectional study of nationally representative individual-level data from 1·1 million adults. The Lancet. 2019;394: 652–662.
- 4. Brook RD, Weder AB, Rajagopalan S. “Environmental Hypertensionology” the effects of environmental factors on blood pressure in clinical practice and research. J Clin Hypertens. 2011;13: 836–842. pmid:22051429
- 5. McIntyre D, Thiede M, Dahlgren G, Whitehead M. What are the economic consequences for households of illness and of paying for health care in low- and middle-income country contexts? Soc Sci Med. 2006;62: 858–865. pmid:16099574
- 6. Adamkiewicz G, Spengler JD, Harley AE, Stoddard A, Yang M, Alvarez-Reeves M, et al. Environmental conditions in low-income urban housing: Clustering and associations with self-reported health. Am J Public Health. 2014;104: 1650–1656. pmid:24028244
- 7. Rubin E, Farber JL. Environmental diseases of the digestive system. Medical Clinics of North America. 1990;74: 413–424. pmid:2181212
- 8. Deane KD, Demoruelle MK, Kelmenson LB, Kuhn KA, Norris JM, Holers VM. Genetic and environmental risk factors for rheumatoid arthritis. Best Pract Res Clin Rheumatol. 2017;31: 3–18. pmid:29221595
- 9. Einav L, Finkelstein A. Moral hazard in health insurance: What we know and how we knowit. J Eur Econ Assoc. 2018;16: 957–982. pmid:30220888
- 10. Dave D, Kaestner R. Health insurance and ex ante moral hazard: Evidence from Medicare. Int J Health Care Finance Econ. 2009;9: 367–390. pmid:19277859
- 11. Qin X, Lu T. Does health insurance lead to Ex ante Moral Hazard? Evidence from china’s new rural cooperative medical scheme. Geneva Papers on Risk and Insurance: Issues and Practice. 2014;39: 625–650.
- 12. Courbage C, De Coulon A. Prevention and private health insurance in the U.K. Geneva Papers on Risk and Insurance: Issues and Practice. 2004;29: 719–727.
- 13. Tavares AI. Health insurance and lifestyles. Appl Econ. 2014;46: 1910–1923.
- 14. Putra FN. Ex Ante Moral Hazard Pada Sistem Jaminan Kesehatan Nasional (JKN) di Indonesia. Jurnal Ekonomi Kesehatan Indonesia. 2020;5.
- 15. Klick J, Stratmann T. Diabetes treatments and moral hazard. Journal of Law and Economics. 2007;50: 519–538.
- 16. Stanciole AE. Health insurance and lifestyle choices: Identifying ex ante moral hazard in the US market. Geneva Papers on Risk and Insurance: Issues and Practice. 2008;33: 627–644.
- 17. Yilma Z, van Kempen L, de Hoop T. A perverse ‘net’ effect? Health insurance and ex-ante moral hazard in Ghana. Soc Sci Med. 2012;75: 138–147. pmid:22507951
- 18. Simon K, Soni A, Cawley J. The Impact of Health Insurance on Preventive Care and Health Behaviors: Evidence from the First Two Years of the ACA Medicaid Expansions. Journal of Policy Analysis and Management. 2017;36: 390–417. pmid:28378959
- 19. Soni A. The effects of public health insurance on health behaviors: Evidence from the fifth year of Medicaid expansion. Health Economics (United Kingdom). 2020;29: 1586–1605. pmid:32822116
- 20. Erlangga D, Suhrcke M, Ali S, Bloor K. The impact of public health insurance on health care utilisation, financial protection and health status in low- And middle-income countries: A systematic review. PLoS ONE. Public Library of Science; 2019. pmid:31461458
- 21. Al-Hanawi MK, Mwale ML, Kamninga TM. The effects of health insurance on health-seeking behaviour: Evidence from the Kingdom of Saudi Arabia. Risk Manag Healthc Policy. 2020;13: 595–607. pmid:32607027
- 22. Fan H, Yan Q, Coyte PC, Yu W. Does Public Health Insurance Coverage Lead to Better Health Outcomes? Evidence From Chinese Adults. Inquiry (United States). 2019;56. pmid:30975010
- 23. Nshakira-Rukundo E, Mussa EC, Nshakira N, Gerber N, von Braun J. Impact of community-based health insurance on utilisation of preventive health services in rural Uganda: a propensity score matching approach. Int J Health Econ Manag. 2021;21: 203–227. pmid:33566252
- 24. Courtemanche C, Marton J, Ukert B, Yelowitz A, Zapata D. Effects of the Affordable Care Act on Health Behaviors after Three Years. 2018. Available: www.iza.org
- 25. Agustina R, Dartanto T, Sitompul R, Susiloretni KA, Suparmi Achadi EL, et al. Universal health coverage in Indonesia: concept, progress, and challenges. The Lancet. 2019;393: 75–102.
- 26. Shrestha R. Health Insurance for the Poor, Healthcare Use and Health Outcomes in Indonesia. Bull Indones Econ Stud. 2021;57: 85–110.
- 27.
World Health Organization. State of Health Inequality Indonesia. Geneva: World Health Organization; 2017. Available: http://apps.who.int/iris/handle/10665/259685
- 28.
TNP2K. The Road to National Health Insurance (JKN). Jakarta: National Team for the Acceleration of Poverty Reduction; 2012. Available: http://www.tnp2k.go.id/images/uploads/downloads/FINAL_JKN_road to national health insurance.pdf
- 29. Widjaja FF. Universal health coverage in Indonesia–The forgotten prevention. Medical Journal of Indonesia. 2014;23: 63–64.
- 30. Aspek T, Gadjah U, Pkmk M. Laporan Reviu Kebijakan Program Jaminan Kesehatan Nasional (JKN).
- 31. Indonesia Media. Perlu Sistem yang Efektif Atasi Fraud [Need an Effective System to Overcome Fraud]. In: Humaniora [Internet]. 2020 [cited 1 Dec 2021]. Available: https://mediaindonesia.com/humaniora/339143/perlu-sistem-yang-efektif-atasi-fraud
- 32. El Omari S, Karasneh M. Social health insurance in the Philippines: do the poor really benefit? Journal of Economics and Finance. 2021;45: 171–187.
- 33. Statista. Global population and municipal solid waste generation shares. In: Statista [Internet]. 2018 [cited 21 Nov 2021]. Available: https://www.statista.com/statistics/1026652/population-share-msw-generation-by-select-country/
- 34. Ministry of Environment and Forestry Republic of Indonesia. No Title. In: Grafik komposisi sampah berdasarkan jenis dan sumbernya (Graph of waste composition by type and source) [Internet]. 2021 [cited 29 Dec 2021]. Available: https://sipsn.menlhk.go.id/sipsn/
- 35. Badan Penelitian dan Pengembangan Kesehatan. Laporan_Nasional_RKD2018_FINAL.pdf. Badan Penelitian dan Pengembangan Kesehatan. 2018. p. 198.
- 36. Ferronato N, Torretta V. Waste Mismanagement in Developing Countries: A Review of Global Issues. Int J Environ Res Public Health. 2019;16: 1060. pmid:30909625
- 37.
Kata Data. No Title. In: Piilah Sampah Jadi Berkah [Internet]. 2019 [cited 23 Aug 2021]. Available: https://katadata.co.id/timpublikasikatadata/infografik/5e9a4c4914ffa/pilah-sampah-jadi-berkah
- 38. Geldsetzer P, Manne-Goehler J, Marcus M-E, Ebert C, Zhumadilov Z, Wesseh CS, et al. The state of hypertension care in 44 low-income and middle-income countries: a cross-sectional study of nationally representative individual-level data from 1·1 million adults. The Lancet. 2019;394: 652–662.
- 39. Addo HO, Dun-Dery EJ, Afoakwa E, Elizabeth A, Ellen A, Rebecca M. Correlates of domestic waste management and related health outcomes in Sunyani, Ghana: A protocol towards enhancing policy. BMC Public Health. 2017;17: 1–10. pmid:28673275
- 40. Badan Pusat Statistik/BPS–Statistics Indonesia. No Title. In: Persentase Penduduk Miskin Maret 2019 Sebesar 9,41 Persen [Internet]. 2021 [cited 1 Dec 2021]. Available: https://www.bps.go.id/pressrelease/2019/07/15/1629/persentase-penduduk-miskin-maret-2019-sebesar-9-41-persen.html
- 41. Badan Pusat Statistik/BPS–Statistics Indonesia. No Title. In: Persentase Rumah Tangga Kumuh Perkotaan (40% Ke Bawah), Menurut Provinsi (Persen), 2017–2019 [Internet]. 2019 [cited 28 Dec 2021]. Available: https://www.bps.go.id/indicator/23/1561/1/persentase-rumah-tangga-kumuh-perkotaan-40-ke-bawah-menurut-provinsi.html
- 42. BPS-Statistics Indonesia. Statistik Lingkungan Hidup Indonesia 2020. Statistik Lingkungan Hidup Indonesia 2020 [Indonesian Environmental Statistics]. 2020.
- 43.
Statista. No Title. In: Number of dengue cases in Indonesia from 2017 to 2019 [Internet]. 2021 [cited 23 Oct 2021]. Available: https://www.statista.com/statistics/705200/number-of-dengue-cases-in-indonesia/
- 44. Alatas V, Banerjee A, Hanna R, Olken BA, Tobias J. Targeting the Poor: Evidence from a Field Experiment in Indonesia. American Economic Review. 2012;102: 1206–1240. pmid:25197099
- 45. Alatas V, Banerjee A v., Hanna R, Olken BA, Purnamasari R, Wai-poi M. Self-Targeting: Evidence from a Field Experiment in Indonesia Abhijit Banerjee Rema Hanna Ririn Purnamasari Matthew Wai-Poi. Journal of Political Economy. 2016;124.
- 46. Oster E. Unobservable Selection and Coefficient Stability: Theory and Evidence. Journal of Business & Economic Statistics. 2019;37: 187–204.
- 47. Goodman-Bacon A. Difference-in-differences with variation in treatment timing. J Econom. 2021;225: 254–277.
- 48. Kenkel DS. Health Behavior, Health Knowledge, and Schooling. Journal of Political Economy. 1991;99: 287–305.
- 49. Brotosusilo A, Handayani D. Dataset on waste management behaviors of urban citizens in large cities of Indonesia. Data Brief. 2020;32. pmid:32775569
- 50. Meidiana C, Gamse T. Development of waste management practices in Indonesia. European Journal of Scientific Research. 2010;40: 199–210.