ON THE EXAMINATION OF OUT-OF-POCKET HEALTH EXPENDITURES IN INDIA, PAKISTAN, SRI LANKA, MALDIVES, BHUTAN, BANGLADESH AND NEPAL

. The aim of this study is to analyze the healthcare expenditures in seven South Asian countries namely, India, Pakistan, Sri Lanka, Maldives, Bhutan, Bangladesh and Nepal. The longitudinal data has been taken for 19 years from 1995 to 2013. We specifically examine the out­of­pocket healthcare expenditure in these countries. The per­capita health expenditure differences have been compared. We also develop panel data pooled OLS model for out­of­pocket expenditure with the factors affecting it, i.e. per capita health expenditure, household final consumption expenditure and public health expenditure. The work is in line with the earlier studies of determinants of out­of­pocket health expenditures. The results suggest that Maldives has the highest per capita health expenditure while out­of­pocket health expenditure as a percentage of total expenditure on health is highest for the India. The fixed and random effect is evidenced on health expenses across the years and cross section based on various determi­ nants. The novel aspect of the work is that, this is an attempt to explain healthcare financing in the developing economies. The key determinant of out­of­pocket expenditure is the final household expenditures as the percentage of gross domestic product.


Introduction
The outofpocket health expenditure (OOPHE) is one of the main issues in the policy decisions in the national set tings. The rising health expenses in the emerging economies have drawn special attention toward building of suitable health shelter plans and basic healthcare facility that mi nimizes the OOPHEs. Health, as defined by World Health Organization, is "a state of complete physical, mental and social wellbeing and not merely an absence of disease or infirmity". Almost 37 million people die each year due to diseases (both communicable and noncommunicable). Hence, the health of population is a key issue in public policy discourse in any society of today's world. The main task of the work is to analyze healthcare expenditures in seven South Asian countries namely, India, Pakistan, Sri Lanka, Maldives, Bhutan, Bangladesh and Nepal. The rising To link to this article: https://doi.org/10.3846/btp.2017.003 economies and India is no exception to this trend. The tre mendous significance of health care imparts huge impor tance to this issue due to which healthcare reforms are being initiated in many countries all over the world. Healthcare reforms in India were introduced as early as in 1980s but the main thrust in these reforms came in 1990s. However, the government was slow in implementing these reforms but still they have had substantial impact on the structure, organization and delivery of health care services around the country.
The introduction of user fees as a part of the India Eighth FiveYear Plan was one of the most significant policy shift in public health care spending of that time. These user fees were waived for people who were below the poverty line. But, at that time the definition of poor or what we call the poverty line was arbitrary that led to limited benefit for most of the poor people. The private health care sector grew a lot when the public spending on health care was reduced in 1990s and private sector moved in to exploit this op portunity. One of the major reforms in the recent times is the launching of National Rural Health Mission (NRHM). Under this mission, many a states have reported consid erable improvements in the various health indicators like outpatient cases, institutional deliveries, disease control, total immunization and family welfare services.
Managing health care costs is a tedious task and one of extreme importance for the policy makers. It depends on the approach to health care. One of the approaches to health care is marketbased where the private organizations and individuals provide healthcare services. The main aspect of this market based system is the competition amongst the players which although detrimental to the welfare of poor people, as it leads to increase in costs, promotes and incentivizes these institutions to develop new medical ad vances and bring them into the market as soon as possible. The other approach is the governmentfinanced system in which the government tries to provide some minimum health coverage for all its citizens using the taxes and other charges. The reach of the governmentfinanced systems is more as compared to the marketbased systems.
One of the biggest challenges to public expenditure policies for all the governments around the world is the rapid growth in the direct medical costs and prescription medicines' cost as well. Most of the countries have adopted copayment systems (mix of market based and government financed) as a measure to contain the health care expendi ture by avoiding consumers' moral hazard and minimizing the unnecessary use of health care services. Previous studies on this issue have found that these copayment systems can reduce unnecessary medical utilization.
Even though all the developing countries have some form of the government financed healthcare system, still the health costs are so high that people have to pay out of their pockets. Catastrophic payments for health can be related with the household resources. The ratio of health expen diture to total consumption expenditure can easily show the household health expenditure in relation to aggregate consumption expenditure. A commonly used threshold of 10 percent implies that the households may be forced to give up other basic necessities or sell their belongings for healthcare expenditure (e.g. Pradhan and Prescott 2002, Ranson 2002, Russell 2004.
The rest of the work is organized as: Section 1 deals with the literature survey, Section 2 and 3 discusses the data sources and empirical model employed, Section 4 offers the empirical results and discussion and last section ends with the conclusion.

Literature review
A fair amount of research has been undertaken by many researchers on healthcare financing, analysis of outofpoc ket expenditure on healthcare and the welfare implications of healthcare financing reforms in many countries all over the world. The efficiency and equity implications of the different strategies have also been extensively studied. Many such researches have also been undertaken in the Indian context. Pannarunothai and Mills (1997) analyze data from household health interview survey for a large urban area in Thailand. An inequitable pattern of outofpocket health expenditure has been observed in context of income quin tile and per capita. The disadvantaged have been found to be least likely to be covered by the government health care schemes while the civil servants made minimal outof pocket payments and did not even contribute to the gov ernment medical benefit fund. This shows the inequality in access to not just good facilities but also to government ben efit schemes which are mainly directed towards deprived people but do not benefit them much. Kutzin (2001) defines a conceptual framework that has the objective of enhancing the insurance aspect of health care systems. This framework has been proposed to be used as a tool for descriptive analysis of the policies and functions of existing health care systems as well as for identifying new policies for the same. It highlights the need for coordinated reforms instead of focus on particular organizational forms of health insurance. Mugisha et al. (2002) in their work on examination of outofpocket health care expenditure in Burkina Faso found high levels of outofpocket health care expenditure (almost 80%) by the households. They used descriptive analysis and a multivariate analysis using the Tobit model on 800 urban and rural households in Nouna health district. Xu et al. (2003) investigate the catastrophic outof pocket health care expenditures for various countries as the initial step for policy reforms. Regression analysis has been used for a crosscountry household survey of 59 coun tries using variables associated with outofpocket health care expenditure. They define catastrophic expenditure as exceeding 40% of household income remaining after meet ing the basic needs. The payments were found to be highest for the countries in transition and in some Latin American countries. They argue that people can be protected from catastrophic healthcare expenditure by reducing reliance on outofpocket payments and providing financial risk protec tion to the poor households. They suggested that health care services should be improved and made more easily available and accessible to the poor people. Falkingham (2004) investigates the level and distribu tion of outofpocket healthcare expenditure in Tajikistan and examines the extent to which these payments hinder the access to health care for poor people. The period of study ranges from 1990-1999. The analysis shows that there are considerable differences in healthcare utilization rates across the various socioeconomic groups. The differences have been observed to be related to the ability to pay. High outofpocket health expenditure costs are taking a toll on the social welfare with the population having to sell their assets for better health facilities. Damme et al. (2004) examines the effect of outof pocket healthcare expenditure on income of households and how it can lead to debt in the rural areas. They survey 26 households for two years to conclude that in Cambodia even modest outofpocket healthcare expenditure can lead to indebtedness and subsequently poverty. It has been suggested to correct this by a accessible public healthcare system with safety nets for the poor. Glied (2008) has examined the effect of alternative health care system financing strategies using data from OECD countries and Canadian health care system for the study. He concludes that in Canada the effects of health insurance, which is publicly funded, are modestly redis tributive. He has also suggested that there is no systematic relationship between efficiency with which the health care system operates and the form of health care financing used (cost being used as a proxy for efficiency). Further, he men tions that health care financing has negligible impact on distribution of wellbeing in society. Raban et al. (2013) describes outofpocket payments and catastrophic health expenditures from household sur veys in India. They conclude that the catastrophic health expenditures have been rising over time. The other inpa tient and outpatient costs have also risen from 200405 to 200910.
There are quite good number of attempt (e.g. Acharya et al. 1993, Sauerborn et al. 1996, Fabricant et al. 1999, Hotchkiss and Gordillo 1999, Pannarunothai and Mills 1997) that deals with the macroeconomy and health, household curative expenditures and household strategies. Therefore, the outofpocket expenditures and health issues became one of the main concern for the policy decisions and economic development.
Some of the recent studies (e.g. Galárraga et al. 2010, Truffer et al. 2010, Bernard et al. 2011, You and Kobayashi 2011, Marshall et al. 2011, Martin et al. 2012, WHO 2015 that examines the Health insurance for poor, Health spend ing projections, OOP expenditures for Nonelderly Adults, Determinants of OOP health expenditures, Risk of OOP, US health spending and Global health issues. However, there are no studies that document the healthcare financing in de veloping countries like India, Pakistan, Sri Lanka, Maldives, Bhutan, Bangladesh and Nepal.

Data sources
The data for this study has been collected from the World Bank database for all the countries. This data is updated annually and adjusted according to the publicly available reports from Ministry of Finance, Central Bank, National Statistics Offices, WHO and other authentic sources for respective countries. Also, the estimates are sent out to Ministries of Health of the countries prior to their publis hing.
The data has been collected for seven South Asian coun tries for a period of 19 years, from 1995 to 2013. We have taken the data on various parameters to make countrywise analysis. We have formed the longitudinal dataset and panel data pooled OLS has been employed. The parameters are outofpocket health expenditure (% of total expenditure on health), outofpocket health expenditure (% of private expenditure on health) and health expenditure per capita (current US$). For running the regression on outofpocket health expenditure, the regressors used are: health expen diture per capita (current US$), total health expenditure (% of GDP) and household final consumption expenditure (% of GDP). Since the data on household final consump tion expenditure was not available, for the complete period under consideration for Maldives, we run panel regression taking into account those missing values.

Empirical model
The methodology for countrywise comparison involves the use of two statistical tools namely, descriptive statistics and regression analysis. For comparison of the countries on the parameters: outofpocket health expenditure (% of total expenditure on health), outofpocket health expendi ture (% of private expenditure on health) and health expen diture per capita (current US$), descriptive statistics have been calculated. Graphical representation has also been used for the analysis (see, Fig. 1). The summary statistics are reported in the next section.
Next, we develop a model for estimating outofpocket health expenditure by regressing it on the following inde pendent variables: health expenditure per capita (current US$), total health expenditure (% of GDP) and household final consumption expenditure (% of GDP). These indepen dent variables have been taken according to the literature as well as because it is suspected that these variables will have an impact on the outofpocket health expenditure. To explain the structure of OOPHE in the developing nations, (i) the health expenditure per capita, (ii) health expenditure as percentage of GDP and (iii) household final consumption expenditure, are the three main factors provide an insights on the health expenses delivery. The scholars like Damme et al. (2004), Falkingham (2004), Glied (2008), and Raban et al. (2013) mainly take into account these factors to elu cidate the outofpocket health expenses in the national settings. We have built the following panel regression model to account the determinants of health expenses in the Asian countries.
The logtransformed equation with logarithm is given as: LNOOPHE it = Ѳ 0 + Ѳ 1 · LNHEPC it + Ѳ 2 · LNTHE %GDP it + Ѳ 3 · LNHHFCE%GDP it + ε it . (2) These two equations are estimated for all the six coun tries using pooled OLS and fixed and random effects model. The main task of the work is to analyze healthcare expen ditures in seven South Asian countries. The pooled OLS adopted for the empirical design is quite capable to explain such relationship.

Empirical results and discussion
First, we summarize the value of descriptive statistics for all the countries for the parameter, outofpocket health expenditure (% of total expenditure on health). The results are tabulated in Table 1. A graphical representation has also been shown for better understanding (see Fig. 1). We obser ve that the mean of outofpocket health expenditure as a percentage of total expenditure on health is the highest for India and lowest for Bhutan. On the other hand, the devia tion is highest for Bhutan and lowest for Bangladesh while the range is highest for Maldives. This implies that India has the highest outofpocket expenditure as a percentage of total health expenditure which shows that in India most of the healthcare cost is borne by the individuals themsel ves. From the Figure we can observe that the except for Maldives, it has somewhat remained constant for all the other countries. Next we analyze the outofpocket health expenditure as a percentage of private expenditure on health. The re sults are shown in Table 2. We observe that the mean is highest for Bangladesh and lowest for Maldives while the deviation is highest for Maldives and lowest for Bhutan. India is almost inbetween the set. Range is also the high est for Maldives followed by Nepal. From the Figure 1 we observe that it is almost between 85-100% for all the countries except Maldives for which it is initially low and has risen over the years. This implies that the condition has worsened over time for Maldives and the outofpocket expenditures are rising.
Another parameter that is analyze is the per capita health expenditure (current US$). This is the highest for Maldives followed by Bhutan, Sri Lanka, India, Pakistan, Nepal and the least for Bangladesh. The Figure 1 shows that it has been rising over time for all the countries with the rise being steepest for Maldives. The deviation is also the highest for Maldives followed by Bhutan and the least for Bangladesh. This shows that the health condition of the population of Maldives is very poor as compared to its neighboring South Asian countries. India is at the fourth place well behind Bangladesh, Nepal and Pakistan. This implies that we still have a long way to go.
Next, we run the pooled OLS regression on outofpock et health expenditure (as shown in Tables 3 and 4) using the regressors: health expenditure per capita (current US$), total health expenditure (% of GDP) and household final consumption expenditure (% of GDP). Two different regres sions have been performed, Table 3 with raw data and Table  4 using the logarithmic values. Both the results are summa rized in the Table 3 and 4. Since the data for Maldives was not completely available, the pooled regression has taken care of missing values. When we investigate the impact of health expenditure per capita and health expenses as per centage of GDP and final consumption over the period, the above tables show significant impact on OOPHE. The cross section analysis explain the impact of internal growth of each countries, the random effect model shows significant shock only in terms of health expenses per capita and the disposable income available in each developing countries 1 .
The output reported in Panel A, Table 3, the last row shows the slope of household income as one of the main de terminant of health expenses in these south Asian countries. The slope appears to be positive and statistically significant. This implies that disposable household income positively Note: * significant @1%, ** significant @5%,***significant @10%. Note: * significant @1%, ** significant @5%,***significant @10%.
influence the rate of private health expenses. More specifi cally, if we look at the logtransformed results in Panel B, the slope of health expenses as percentage of GDP and house hold income remains the main determinants of out of pock et health expenses in these emerging Asian economies. The positive significant slope explains that out of pocket health expenses are positively associated with the health expenses incurred and availed disposable income. The Table 4 does not provide any important consequences based on private health expenses. The results are in line with the previous outcome but not significant. Another thing, which is evi dent, is that the per capita health expenditure is negatively correlated with outofpocket health expenditure while the household final consumption expenditure is also negatively correlated but not significant. This shows that a trend is not evident and we cannot generalize this observation.

Conclusions
Healthcare is of prime importance to everyone in the world, and rising healthcare expenses remain the main concern for the policy maker in emerging economies and well de veloped nations. The rising outofpocket health expen ditures all over the globe are a cause of worry for all the policy makers and economists. This is mainly worrisome for the developing countries as much of their population is below poverty line and so, providing affordable healthcare to them is very important and equally difficult. The study investigates the outofpocket healthcare expenditure for seven South Asian countries and all these seven countries are regarded as developing countries, by preparing the longitudinal panel dataset. Next, we performed the po oled OLS for outofpocket health expenditure using the selfgoverning variables as health expenditure per capita (current US$), total health expenditure (% of GDP) and household final consumption expenditure (% of GDP). The results suggest that Maldives has the highest per capita health expenditure while outofpocket health expendi ture as a percentage of total expenditure on health is the highest for India. The fixed and random effect is evidenced on health expenses across the years and cross section based on various determinants. The novel aspect of the work is that, this is an attempt to explain healthcare financing in the developing economies. The key determinant of outof pocket expenditures is the final household expenditures as the percentage of GDP.