Determinants of household healthcare expenditure: an analysis in Vietnam by using of VHLSS 2006

TABLE OF CONTENTS TABLE OF CONTENTS ACKNOWLEGEMENTS 2.1 Definitions 2.1.1 Healthcare 2.1.2 Household Healthcare expenditure Theoretical framework for Household Healthcare 2.2 Expenditure

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

DETERMINANTS OF HOUSEHOLD HEALTHCARE EXPENDITURE: AN ANALYSIS IN VIETNAM

BY USING OF VHLSS 2006

BY

LE PHUONG THAO

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, NOVEMBER 2011

- -··· · - - ~- - -~ -

-~ -1/79

VIETNAM- NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

DETERMINANTS OF HOUSEHOLD

HEALTHCARE EXPENDITURE: AN ANALYSIS

IN VIETNAM BY USING OF VHLSS 2006

A thesis submitted in partial fulfilment of the requirements for the degree of

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

By

LE PHUONG THAO

Academic Supervisor:

DR LE THI THANH LOAN

HO CHI MINH CITY, NOVEMBER 2011

in all the time of research and writing of this thesis And my sincere thanks also go to Associate Professor Dr Nguyen Trong Hoai, Co- Director of Vietnam - The Netherlands Program for M.A in Development Economics, who has always given me his encouragements and kindly during the course of my study and thesis research

I wish to thank my close friend, Pham Tien Thang, who supported me in finding working papers for references

Lastly, I owe my loving thanks to my parents and my husband Without their encouragement and understanding, it would have been impossible for me to finish this work

ABSTRACT

The main purpose of this study is to identify the determinants of household healthcare expenditure in Vietnam The main source data for the analysis are from Vietnam Household Living Standard Survey 2006 (VHLSS 2006) The analysis uses statistic analysis and Ordinary Least Squares (OLS) estimates to find out the determinants of healthcare expenditure First, statistic analysis gives us an overview

of household healthcare expenditure situation in Vietnam Second, we estimate the parameters of household healthcare expenditure model by using the Ordinary Least Squares (OLS) estimates

The statistic results indicate that in the total of household expenditure, the household healthcare expenditure made up only 6.37% and in total of household healthcare expenditure, 72.53% is used in paying user fees at health facilities (health expenditure for having treatment) The results also present that household healthcare expenditures differ by expenditure quintiles, health status, health insurance status, education of household head, gender of house head and ect,

The regression results bring out some findings First, household healthcare expenditures and household income (that household expenditure is a proxy) have significant relationship Second, important determinants of household healthcare expenditure were household expenditure, household size, health status and health insurance The age and education of household head are also important, but their effects on household healthcare are small Moreover, there were statistically significant differences in household healthcare expenditure across regions

Key words: households; healthcare expenditure, household expenditures, Vietnam

TABLE OF CONTENTS

TABLE OF CONTENTS ACKNOWLEGEMENTS

2.1 Definitions 2.1.1 Healthcare

2.1.2 Household Healthcare expenditure

Theoretical framework for Household Healthcare 2.2

Expenditure Function 2.2.1 Households and utilization of health care

Household characteristics and household healthcare 2.2.2

expenditure Community characteristics and household healthcare 2.2.3

CHAPTER3 METHODOLOGY AND DATA SET 35

HEALTHCARE EXPENDITURES IN VIETNAM

4.1 Overview household healthcare expenditure in Vietnam 43

4.1.1 Healthcare expenditure and household expenditure structure 43

in Vietnam 4.1.2 Household characteristics and healthcare expenditure 46 4.1.3 Community characteristics and healthcare expenditure 54 4.2 Estimated results and Explanation 56

LIST OF TABLES

Table 4.1 Household characteristics 46 Table 4.2 Income and household healthcare expenditure 48 Table 4.3 Health care expenditure by Education of head of household 49 Table 4.4 Healthcare expenditure by gender ofhead of household 50

i

Table 4.5 Health care expenditure by age of head of household 51 Table 4.6 Healthcare expenditure by health status 52 Table 4.7 Healthcare expenditure by Insurance 52 Table 4.8 Healthcare expenditure by household size 53 Table 4.9 Community characteristics 54 Table 4.10 Healthcare expenditure by urban/rural 55 Table 4.11 Healthcare expenditure by regions 55 Table 4.12 Variables - their definitions and expected signs 56

Table 4.13 Regression results of the determinants of household 59

healthcare expenditures Table 4.14 Regression results of the determinants of household 64

healthcare expenditures with significant variables

Structure of sources of health expenditures in VietNam,

2006 Household expenditure structure in year 2006 Household Healthcare expenditure structure in year 2006

healthcare and independent variables Checking for Multicollinearity Regression of the model with dependent variable 1s Household per capita healthcare expenditure

The best model: Regression of the model with dependent variable is Household per capita healthcare expenditure

Chapter 1: Introduction

1.1 Problem statement

Health is a fundamental dimension of well-being, a key component part of human capital and World Bank reports (1993) also indicated that economic growth (particularly poverty-reducing growth) and education are central to good health Therefore, the main challenge of the health care system is to protect households from the risks of the consequences of impoverishment from health spending and to ensure that all households receive health services when they need The financial burden of health expenditure may cause households to spend more than their available incomes and can lead a household into debt The "Vietnam Joint Annual Health Review 2007" show that around 34.5% of medium-income inpatients had to borrow to pay for health services and the burden of health expenditure has resulted in borrowings by many households The financial burden of healthcare also causes an endless cycle of poverty and ill-health - the burden of out-of-pocket health care payments on households

The "doi moi" (renovation) process of Vietnam started in 1986 and after over

20 years of "Doi moi'', Vietnam has gained significant achievements in both the economy and society, including important achievements in Health sector The health reforms have reached profound changes in healthcare utilization including the change

in healthcare financing, healthcare access, healthcare delivery More attention to promote the development of the private health sector and liberalization of the pharmaceutical industry are two of the most important reforms in health sectors Beside, the user fees for health services at higher level public health facilities and health insurance program also have introduced All of these reforms have had

extensive influence on the health sector, household healthcare expenditure and household health care-seeking behavior

Health insurance program started in 1992 by issuing Decree No 299/HDBT that the Regulation on Health Insurance was published Since then, health insurance has officially become a health financing source that support for healthcare expenditures The health insurance fund covers curative care expenditures for the people that enrolled in social health insurance schemes (compulsory and voluntary) Besides, the government also issues some health insurance policies that expanded subsidized health insurance to cover the poor, the near poor and children under 6 years old

And the VHLSS 2006's results show that more than 50% people recetvmg medical examination and treatment had health insurance, significant increase than the rate in 2004 even in rural areas However, coverage of health insurance remains limited; the financial sustainability of health insurance funding is still low Rural people had less opportunity to receive medical examination and treatment in state hospitals than urban people; they often had to go to commune health centers The rate

in richest quintile was higher than in the poorest quintile and the difference was bigger in the rate for out-patients Differences in utilization of health services between various populations groups have grown, it coupled with gaps in living standard Also according to the VHLSS 2006's results, expenditure for health care of households in

2006 was all increased than in 2004 The average expenditure per person of urban households and rich households was higher than of rural households and poor households respectively In Vietnam, health care expenditures of the poor make up a higher proportion of their income than the non-poor even though they often try to restrict their seeking-behavior

Moreover, the report by the World Bank (2001) indicated that there is a very large in relation to disposable income for the poor This is a big problem of the health sector especially in developing countries The expenditure for health care services is too high for the poor and many people does not confidence in the quality of local medial care services may cause this problem Therefore, understanding which determinants effect on household healthcare expenditure is very important, the findings will help policy makers solve the problems of health care sectors more effectively

In the world, many empirical studies research determinants of demand for health care or health care expenditure (Hjortsberg 1999, Mocan 2000, Rous and Hotchkiss 1998) These studies have investigated some factors that impact on health care expenditure such as household economic situations (income, wealth ), household compositions (age, gender, the number of males/females, the number of children, household size ), community characteristics (region, rural/urban ) and type of diseases In Vietnam, there are few studies that research on health care expenditure Trivedi (2002) has studied the major features of health care utilization patterns in Vietnam The study focused on "the determinants of largely self-prescribed, the use of pharmaceutical drugs, government hospitals, commune health centers, and private health facilities" Health insurance and household income are considered as the important factors that effect on health care expenditure Beside, seeking behavior of households to choice health care provider types is also analyzed

in the study By using regression method, the econometric models analyze health care expenditure in both individual and household level In another study on health care expenditure in Vietnam, CCSE - WHO group and Ministry of Health group (2006) pointed out many factors that had impact on catastrophic healthcare expenditure in Vietnam These factors consist: "household living standard status, household income;

education level of household head, ethnic status, number of inpatient visits, number

of outpatient visits, number of over-counter visits for self-treatment, number of children, number of fertile female and number of elderly persons in household, etc" And the roles social insurance and target subsidies to the poor in reducing the rate of household catastrophic health expenditure are also proved in the research

However, while many decisions are household decisions, the multivariate household-level studies on this issue has been very few There are some research papers that mentioned the producer of health is the family rather than the individual (Jacobson, 1999; Bolin et al, 1999), and utilization of health care is influenced by household income through the allocation of household budget Therefore, this paper will try to identify the determinants of household health care expenditure in Vietnam which are useful for planning an effective health care policy In addition, to improve equity in health expenditures, we examine vertical equity in health payments by examining burden of health care expenditures across five income quintiles This paper will use quantitative analysis and linear regression framework (the ordinary least squares (OLS) method) to estimate all key factors that may have impact on household health care expenditure The Vietnam Household Living Standard Survey (VHLSS)

2006 data with more detail questionnaires of health section is the main data source used for regression the model Besides, the data from the Ministry of Health reports is also used for descriptive analysis in this paper

1.2 Objectives of the study

The aim of this paper is to investigate the factors that may have influence on household health care expenditure in Vietnam More specifically, this study attempt to explore the follow questions:

-Which factors determine households' health care expenditure in Vietnam?

- What extent significant determinants impact on health care expenditure?

- Is there a difference between the health care expenditure patterns of poor households and those ofbetter-offhouseholds?

Knowing the answers to these questions is very important for policy makers, these can help them in making informed decisions regarding policies intended to improve social welfare For example, for the households that they are lack of the ability to spend more for healthcare, the government should provide only very basic healthcare at low price However, for the households that they have the ability and the willingness to spend more on healthcare and they ready pay for good quality health care, then the government can expand more options By offering a wider variety of health services, the government still recover a considerable fraction of the costs

- What the Government should do to reduce financial burden of healthcare expenditure for the poor?

The results of this research may help the planners give an effective health care policy that reduce the burden of health care expenditure for households, especially for the poor

This paper is organized in the following way It includes five chapters: Chapter

I is Introductory section; Chapter II - Literature review, this chapter introduces a theoretical framework that reviews theories and empirical studies related to the topic; Chapter III - Methodology and data set, it describes the data and the methodology that is used in the analysis; Chapter IV - Determinants of household healthcare in Vietnam, by using the descriptive method and regressing the econometric model, it analyses overview households' health care expenditure and examines determinants of households' health care expenditure in Vietnam; And final, Chapter V- Conclusion and Recommendations, this chapter summarizes all analysis and findings in previous chapters and gives some policy recommendations

- That affects the structure or any function of the human body

- Health care includes participation in research that, considering the risks and benefits

of participation, presents a reasonable prospect of direct medical benefit to an individual"

According to the Dependent Adults Act in the United State: "Health Care includes:

- Any examination, diagnosis, procedure or treatment undertaken to prevent any disease or ailment,

- Any procedure undertaken for the purpose of preventing pregnancy

- Any procedure undertaken for the purpose of an examination or a diagnosis,

I •

- Any medical, surgical, obstetrical or dental treatment, and

- Anything done that is ancillary to any procedure, treatment, examination or diagnosis"

2.1.2 Household Healthcare expenditure:

The MOH report (2008) definite that:

Household health expenditure is the total spending of a household on all of its health related needs, including preventive, promotive and curative care Household health expenditures can include pre-payment before an illness (e.g to purchase health insurance) or direct out-of-pocket health expenditures when using health services (e.g paying hospital user fees)

Direct out-of-pocket payment for health care refers to the expenditures households make directly when they use services, primarily purchase of drugs, payment of hospital user fees, diagnostic service fees and other indirect expenses related to seeking medical care at state or private facilities (including self-medication)

2.2 Theoretical framework for Household Health Expenditure Function

2.2.1 Households and utilization of health care:

First, we start with studies of utilization of health care because some studies used household health care expenditure is a proxy of health care utilization

' Base on the human capital theory, Grossman's model of health capital (Grossman, 1972,2000) is a formal model to analyze the relationship of health capital and income and education Income and education are the most influential factors for

health care utilization Bolin et al, 1999 also show the relationship between household income and health care utilization Some studies also concluded that income and education of household head have important impact on household health care expenditure (Himanshu, 2007; Parker, Wong, 1997) Beside, other factors such as age also impact on utilization because age reflects on perceived benefit and income And poverty impacts on limiting a patient's ability to pay for medical care so that it effect

on health care utilization largely (Blanchard, 2005)

In some studies, the authors mentioned about access factors that may influence

to utilization of health care In reality, the largest determinant of seeking care may be the expected access cost and the individual often take the first contact with the system healthcare Le5 Grand (1982) argues that access cost includes both monetary costs and time costs, it consists some factors such as out of pocket payments, distance to health facilities, waiting time at the facility etc, i.e More specifically, monetary costs embrace health services fees and costs for traveling to the health facilities, while time costs include time to reach the facility, waiting time at the facility and time to get advice from the health consultants Access costs are usually an important determinant

of health care utilization, especially it is more meaning in analyzing the differences in health care utilization across different social groups in developing countries (Gertler and van der Gaag, 1990)

Jacobson (2000) argues that the individual does not produces "good health",

"good health" is produced by the family Therefore, the Grossman's model is extended into a new model with the producer of health is the family With the new model, Jacobson (2000) concluded that the production of health not only use the individuals' own income but also the family's combined resources The family allocate the investments in health capital and it will not try to distribute the equal health capital to each member of the family Therefore, it leads to the marginal

benefits equal marginal net cost of health capital Moreover, the resource allocations within a household are influenced by the individual members of a household The most obvious examples indicate that there is a difference in allocating household budget between the households without children and the household with children The household with children will allocate a larger share of the budget to food than households without children A natural parallel to household health expenditure, it would be that households with children allocate a larger share of the budget to health expenditure compared to households without children

2.2.2 Household characteristics and household healthcare expenditure:

The next, we review some empirical studies that have considered how households allocate resources

The theoretical literature on household economics that Becker had given in

1964 and 1965 extended the neoclassical model of consumer demand to household In his model, the assumption is "all household members are assumed to maximize a household level welfare function" and therefore, the utility function is a joint utility function All available resources of the household are pooled and then reallocated base on a common rule and income is allocated in such a way that the marginal rate

of substitution between any two consumption goods is the same as for any other pair However, this model is not suitable in case intra household allocations It has been suggested that instead of considering the bargaining and negotiations that actually occur within a household, intra household allocations should be modeled with bargaining models (see e.g Manser and Brown, 1980; McElroy, 1990, Bolin et al, 1999) Moreover, there are some other models have been suggested, Behrman et al (1982; 1986) suggest that modeling intra household allocations should assume a specific structure for parental preferences, while others propose that a Pareto efficient

outcome should be used (Chiappori, 1988; Kooreman, 1990) While the Beckerian model pool all resources, these above model allow the differences between household members in preferences, resources are allocated towards goods that different household members desire

The next model that we review here is one of the simplest models of household consumption of Samuelson (1956) It assumed that "the household income always is divided in pre-specified proportions between household members" Each household member maximizes utility subject to the given budget constraint by choosing her or his own consumption bundle Applying this for heath care expenditure, we see that each household member would try to get her or his own utility of health care consumption and not the benefits for the household as a unit Therefore, for the household that do not have common preferences, Bargaining models from cooperative game theory may be the best choice in this situation Lundberg and Pollak (1996) had use Nash bargaining models for their research However, the experiences indicate that these models are only suitable in a two-person household

More specifically, almost the models above mention some main factors that impact on household healthcare expenditure as: household economics (income, wealth, poverty, employment ), demographic characteristics of household (household size, number of children, number of women in fertile age-group and the characteristics of household's head )

In most developing countries, the role of the head of household is relatively important In general, according to the hierarchical decision-making process in the households, the household head may decide on several issues, including the level of household healthcare expenditure The household heads have education is included to reflect knowledge of health and medicine in the household, as education and

healthcare utilization have positive correlation (e.g., Ichoku and Leibbrandt 2003; Lindelow 2004) It is commonly said that the decision of allocating the household expenditures on food is included to reflect the poverty or budget constraints and Makinen et al (2000) also show that "relative income influences both the decision to seek care and the type of care sought" Finally, resource-constrained households with multiple sick members face deciding who to treat through inpatient care and who to treat through other methods The three alternative-specific variables each reflect the costs and benefits associated with treatment at each type of facility; presumably, lower minimum spending thresholds, higher maximum benefit levels, and lower average costs of treatment increase the attractiveness of each hospital type Moreover, some other factor that likely influence both the decision to seek care and hospital choice such as age, sex, disabilities, and emigration status For example, when get sick, women are more likely to seek care for sickness than men in the U.S and China respectively (Gao and Yao 2006) Similarly, Reinhardt (2000) reveals that age have positive impact on both the quantity of health care expenditure and total spending The disabled are more likely to seek healthcare than people without physical limitations (Sommers 2006-2007) By contrast, people who emigrate have better self-reported health status and lower incidence of illness (Hesketh et al 2008), suggesting that they may have different preferences for healthcare than non-migrants

2.2.3 Community characteristics and household healthcare expenditure:

Final, we mention about some literatures of community characteristics and

household healthcare expenditure Obviously, each region has distinct features of geography, demography, and custom so that the household healthcare expenditure living in different regions also are different Place of residence, for example, whether one lives in a rural or an urban area, may indicate geographic proximity to a source of

care as well as local attitudes about health care (Woottipong, 2001) Household health care expenditure is higher for urban than rural household

In short, consumption behavior depends on demographic and socio-economic statuses From the above discussion of theories, we see that health care utilization may

be affected not only by the household composition (such as household size, the number of adult females, males, the number of fertile women and children within the household) but also individual characteristics of household members, household head's characteristic Some important determinants of healthcare utilization relating

to the household head's characteristics are Education level of household head, sex of household head, age of household (Himanshu, 2006) Besides, regions with different socio-economic conditions also have impact on household health care expenditure (Margherita and Theodore, 2002; Ha nguyen, Peter and Ulla, 2002) Moreover, the resident place of household (rural/urban) is an important factor that may have impact

on household healthcare expenditure (Woottipong, 2001 )

2.3 An Overview of the Empirical studies relates to household healthcare expenditure:

There are not many empirical studies mentions about determinants of household health care expenditure in developing countries as well as in Vietnam We can list here some main empirical studies:

Firstly, Himanshu (2006, 2007) studied the determinants of household

healthcare expenditure in Tribal and Urban Orissa (India) with three working papers

Two of these working papers explored the influence of household income and household head's education on household healthcare expenditure in Tribal and Urban Orissa The regression analysis and descriptive statistics is used to substantiate the

objective Three variables are used in the model, including: household health expenditure, household income and education of the household head; and the model that the author used in both studies is a linear regression model: PHE = ~ 1 + ~ 2 PHI +

~3EDN

While:

The model use per head health expenditure (PHE) to represent the household health expenditure, it is calculated by dividing total annual health expenditure of the household by the household size

Similarly, per head income of the household (PHI) is used for household income variable the regression analysis, it is calculated by dividing total annual household income by size of the household

And, education is a dummy variable in the regression analysis, education equal 1 if those head of the households is educated and equal 0 if those head of the households is uneducated

The results of these regression analysis showed that: the linear regression

models are fitted as: PHE = 31.37 + 0.43PHI + 0.06EDN for the case in Tribal area

and PHE = -696.046 + 0.82PHI + 0.03EDN for the case in urban area The results

indicates that in Tribal, rural and urban areas, both income and education have the positive influence of on health expenditure

The influence of income on healthcare expenditure is different between households living in Tribal, rural and urban areas In urban area, income has the most influence on healthcare expenditure and it has the least influence in Tribal area The reason is because of the lower per head income of the household in Tribal and rural area than in urban area

However, the influence of education on healthcare expenditure gtves an interesting finding: "healthcare expenditure in Tribal area is double that of rural and urban areas It means that an educated person on an average spends six paise more in

a rupee than the uneducated person on health expenditure in tribal area where as an educated person in both rural and urban areas, on an average spends only three paise more in a rupee than the uneducated person" The relative values of education for Trial people are more than the people of rural and urban areas cause this results

In the remaining working paper, Himanshu (2006) mentions about the impact

of gender on household healthcare expenditures in Urban Orissa By using the same methodology in the two studies above, the linear regression model was suggested as: PHE = Bl + B2 PMHE + B3 PFHE

PFHE is per female health expenditure, it is calculated by "dividing total annual female health care expenditure of the household by number of female members

of the household"

After running regress10n, the author has concluded that "biologically determined sex and socially constructed gender have strong bearing on the household out-ofpocket health expenditure The study shows that there is a significant difference between male and female out-ofpocket health expenditure in urban area"

However, out-of-pocket health expenditure of females living in urban is higher than

that living in rural and tribal areas The findings of the regression model also depicts that in urban area, the extent impacts of male on out-of-pocket health expenditure is higher than the female

The next is the study of Pravin K Trivedi (2002) that mentioned about the

household healthcare expenditures in Vietnam Health care expenditure is only a part

of this study The author used VHLSS 1997-1998 to analyze health care expenditure

in both individual and household level

The sample size for studying in individual level is 8081 A regression analysis

of medical expenditure is used in the study with the following variables:

The dependent variable is log of health care expenditure for each member of the household with the condition the health care expenditure for that individual is positive The health care expenditure here included all types of health care expenditure in the 4 week period preceding the survey

The independent variables are used in the study includes: household income and health insurance are the main independent variables because this analysis focus

on the impact of household income and health insurance on health care expenditure The author also controlled some other variables in the model such as the age, the gender (male/female), the marital status, education, and health status variables The health status variables included illness/not illness in 4 weeks before the survey, injury/not injury in 4 weeks before the survey, days of illness/injury in 4 weeks before the survey, and days of limited activity in 4 weeks before the survey

The results of individual healthcare expenditure indicate that whereas household income has strong impact on individual health care expenditure, the insurance variable is much less significant

To analyze the health care expenditure in the household level, the sample size used for analysis is 5006 The study analyze the aggregate health care costs of all households member is a useful check on the results of the individual data analysis By this way, it also help to estimate the Engle curve for health care expenditures This approach is limited because the health status of the household members are unable controlled The model regression also controlled some of other relevant variables such

as household size, gender, age, household heads' education levels and location (urban

or rural) The linear regression model was also used for the household level analysis and the result shows that:

Age and sex of the household head have significant impact on household healthcare expenditure On average, the households with a female household head paid more for health care than ones with a male household head and households with older heads also spend more for health care However, household size and educational level of the household head do not have significant impact on household healthcare expenditures

Location variable is a significant factor that determine the household healthcare expenditure The analysis indicates that the urban household spend more for health care than rural household

Household income is a significant determinants of household healthcare expenditure It has positive impact on household healthcare expenditure, the household with higher income send more on healthcare Besides, the point estimate of income elasticity for household healthcare expenditure is larger than the corresponding estimate for individual health care expenditure

We continue with the study of Catharina Hjortsberg ( 2000), this paper analyzes the determinants of total health care expenditure of a household and explains health care expenditure among households from different socio-economic groups in Zambia The household's economic situation is also analyzed in this paper and a particular interest is focus on "the impact of household economic circumstances on household healthcare expenditures"

Based on utilization healthcare theory ("households are constrained by monetary and time resources"), household economic theory and the assumption that

"households obtain utility from their household members' health and other consumption", the study has suggested a linear regression model with three groups of independent variables: Economic circumstance, Household characteristics and Access variables Details of the model as below:

Dependent variable: Total health care expenditure for the household

Explanatory variables:

Economic circumstances: Monthly total expenditure of household; Monthly total

expenditure on other than food of household; Monthly income for household; Self assessed poverty level; Rented or owned

Household characteristics: Education of head of household; Age of head of

household; Sex of head of household; Total household size; Number of boys; Number

of girls; Number of male adults; Number of female adults; Number of women in fertile age-groups ( 15-49); Number of children in schooling age attending school

Access variables: Distance (Distance to the nearest health care facility in km);

Vehicle (If the household own its own vehicle or not); Location (Indicates if the household is located in a rural area)

In this paper, the researcher used regression methods to estimate health care expenditure as: first, regressing the model to estimate the equation of the household's health care expenditure and then to estimate the equation, the author used limited dependent variable techniques (censored regression using Tobit technique) Data source for analysis is the data from the 1998 Living Conditions Monitoring Survey (LCMS)

The paper has empirical analyzed determinants of household healthcare utilization using household health care expenditure as a proxy for utilization and the main findings: almost the independent variables are suggested in the model have significant impact on household health care expenditure

The estimate results indicates that the households' economic circumstances and access to health care facilities have directly impact on health expenditures by Zambian households Zambian households' healthcare expenditure are influenced by total monthly expenditure and monthly expenditures on other than food Household size and the wealth of the household can directly relate to both of these variable However, when considering the difference among three poverty groups, it becomes more clear that poor households or moderately poor households are more sensitive to the level of expenditure on other than food than non-poor households Ownership of house is a dummy variable and is also a proxy of economic circumstances And the result indicates that households who actually own their own house spend more on healthcare than households not owning their own house

VNP

Class 13

The results of estimations also show that access to facilities variables are important factors that effect on health care expenditure All access variables including the distance to the nearest health care facility, vehicle and location have significant effect on the household healthcare expenditure The households that own a vehicle spend more on health care than those without a vehicle Moreover, the household healthcare expenditure level is also affected by the distance to the nearest health care facility However, the result show that distance does not affect on health care expenditure of non-poor households With the respect to the household location, the regression result present that the level of rural household health care expenditures is lower than urban households It takes longer to reach a health facility in rural areas than it does in urban areas with given the same distance The less developed infrastructure in rural areas may be the reason of this

Lastly, household health care expenditure level is also influenced by the demographic characteristics of the household Household size is a significant variable that impact on total health care expenditures, household have larger number of members tend to spend more on healthcare Otherwise, household head's age is also

an important determinant of household health care expenditure

Final, Maathai K.Mathiyazhagan, (2003) also analyzed the relationship between Rural Household Characteristics and Health expenditure in India

In this paper, the author also used literature of health care utilization for studying household healthcare expenditure Base on the household economics theory,

it is assumed that "households get utility or satisfaction from consuming goods and services", and to desire for consumption "household members must produce many of the commodities" And the paper also assumes that "the utilization of health services

VNP

Class 13

is a derived demand from the production of health" Therefore, the regression model

is formulated as follows:

Where:

(h-exp); Are the expenditures on health care by household i

Y; is the household income

n; is household size

lt;k is the number of household members in age-group k

z 1 represents a vector of other household socio-economic and demographic characteristics and e; is a random error term

The dependent variable is used in the regression equations is the total household health care expenditure or total expenditure on drugs and medicines And the method uses to estimate here is two-stage least-squares (2SLS) regression In the first stage, to compute estimated values of the problematic predictor(s), the author use instrumental variables that are uncorrelated with the error terms After that, in the second stage, the computed values that computed in the first stage are used to estimate

a linear regression model of the dependent variable

The study uses four groups of explanatory variables to analyze, including

"household income variables, household composition variables, risk variables, and socio-economic variables"

i'

Group 1 consists household income variables Household income variables

classify to low income and high-income percentiles

Group 2 comprises household composition variables such as the household

s1ze (total number of household members), percentage of women of fertile ages

(15-49 years old) and percentage of children with respect to the household size

Group 3 includes risk variables such as health status variables and the number

of times for doctor's advice The health status variables consists some variables as general health conditions of the household members, number of visit to the hospital, the number of working days lost due to ill health, health seeking behaviors (using public or private health care providers) of the households and source of health care utilized

Group 4 embraces socio-economic variables such as health insurance,

education level of household head Health insurance variable is considered whether the household head or dependents have health insurance or other benefits

Moreover, to assess the differences in cost of living with severe morbidity across regions, the paper used regional dummies in all regressions

The next, we consider the estimated results of the paper The following parts is some main findings of the paper:

Firstly, the regression result show that the household income variables have strong impact on household healthcare expenditure Changes in household income levels have sensitive effects on the household healthcare expenditure in rural India With high income groups, the elasticity of health expenditure with respect to income

is largest This finding suggests that at times of economic crisis and recession, the

households of the high income group tend to reduce proportionately healthcare expenditure more than the low-income group This also shows that health care is a luxury good and it is especially for the upper income group When the real per capita income increased, household health care expenditure will be made a greater proportion of income

Secondly, the regression results of the composition variables indicate that the proportion of fertile age group women in the households has a significant negative effect on household healthcare expenditure The results also indicate that for both low and high income groups in short term function, the high percentage of children has positive and significant effect on household health expenditure However, for long term function, the percentage of children in the household do not have any significant influence on the health care expenditure

Thirdly, the findings suggest that almost risk variables relate significantly to the household health care expenditure The number of working days lost due to illness has positive and significant impact on both total health expenditure and drugs and medicines expenditure The results is also similar for the frequency of doctor's consultation variable Moreover, the findings ·reveal that the choice of health care provider of the rural households have negative impact on total health care expenditure

Fourth, other sources of health benefits of the household is illustrated that it has a significant negative relationship with total health expenditure Beside, occupational status, literacy of household head are also significant determinants of household health expenditure

Finally, another significant factor that impact on household health expenditure

is the household members' residing location The estimates results presented that the household members living in the different regions have difference in spending for health care

Chapter 3: Methodology and data set

This chapter will specify a model from the theoretical framework m the chapter 2 and give the methodology to estimate the key determinants of household health care expenditure

3.1 Econometric Model of the study:

The model will be built based on both theoretical and specific situation in Vietnam, the choice of the variables is also influenced by the availability of data As discussion in Chapter 2, we can expect that household healthcare expenditure is impacted by social, economic, and demographic characteristic features of the households Therefore, the model is suggested for this research is:

Xh is the vector of Household characteristics factors that may have a direct impact

on household health care expenditure

Xc is the vector of Community factors that may have a direct impact on household health care expenditure

Jli is summarizing all unobservable characteristics of economic, household and

community that affect on total household health care expenditure

Because of some limitations, the regression equations only includes only the following variables The variables of the model are detail as below:

Dependent variable:

HHEXP: Household healthcare expenditures, we use logarithm of expenditure on health care According to the VHLSS 2006, through personal interviews, individuals are asked to recall their total household health care expenditures in last 12 months (expenditures on drugs, diagnosis and/or treatment, transportation, lodging, buying health care insurance, care-taking and other cost for health care)

Explanatory variables:

In our model, there are two sets of explanatory variables:

Household characteristics variables

Household characteristics variables are used in our model:

• EXPEND: Household Expenditure per capita It is calculated by dividing the total household expenditure by the number of household members

In developing countries, expenditures are commonly used to proxy for effective income because expenditures are less likely to vary than income and are less prone to reporting biases Therefore, instead of household mcome variable, in the model, we use household expenditure variable

• Household head's characteristics variable:

The head of a household is a key person who makes decision bearing on health service utilization of all household members Three selected characteristics of the household head were used as the variables in this study:

* EDUChh: Education of head of household It is the highest diploma of the household head and it is measured by years of schooling The years of schooling and the corresponding educational levels are defined as follows:

No diploma

Primary school

Secondary school

Higher Secondary

Short - term vocational training

Long- term vocational training/ Professional high school

Junior College Diploma

: 12 :13 : 14 : 15 : 17

: 19 :23

* SEX: Sex of head of household, this is a dummy variable that is set equal 1

if Sex of head of household is a male and 0 otherwise

• HHSIZE: Household size is the number of people in a household (members of the household)

• HESTA: Health status, we use a dummy variable of illness/injury which members of household get in 04 weeks recall ( 1 if any member of household got illness/injury in 04 weeks recall and o otherwise)

• INS: Health care insurance status (health care insurance or non health care insurance) Health insurance is an important factor that have influence on health care expenditure Health insurance help the household reduces

household out-of-pocket expenditure for health care services In the econometric model, insurance variable is a dummy variable: equals 1 if at least one member of the household is insured and 0 otherwise

Communities' characteristics:

• AREA: urban/rural, this is a dummy variable that is set equal 1 if household lives in urban and 0 otherwise

• REGION: Because of distinct features of geography, demography, and custom

of each region, household health care expenditure also varies across different regions Vietnam is comprised of eight regions that are Red River Delta, North East, North West, North Central Coast, Central Coast, Central Highlands, South East and Mekong River Delta North West is used as a benchmark and six dummy variables are used for remaining regions

3.2 Data set

3.2.1 Data source

The data of this analysis ts derived from the Vietnam Household Living Standards Surveys (VHLSS) household survey 2006 collected by the General Statistical Office of Vietnam with the assistance of the UNDP and World Bank The survey was conducted nation-wide, involving a sample scale of 45,945 households (36,756 households for income survey, 9,189 households for income and expenditure survey) in 3,063 communes/wards, representative for whole country, 8 regions, urban/ rural area and provinces Organizationally, the survey was conducted to collect information in 2 rounds, 2006 and by direct interviews with headed households and key commune officials The survey include 13 main sections: Demography; Education; Labor - Employment; Health and health care; Percentage of people with

disability; Childbirth; Smoking; Income; Expenditure; Housing, electricity, access to safe drinking water, sanitary and durable goods; Income gap and inequality; Involving

in the Hunger alleviation and poverty reduction Programs and Credit; Characteristics

of commune

The health- healthcare, expenditure, demography, education components of the

2006 VHLSS is the main source of the data used here The responses to questions about expenditure on health care refer to a period of 12 months

Limitation of the data: People hide income, long interview, no reliable estimates for district or commune levels, limited budget

3.2.2 Some definitions of VHLSS 2006

• RuraVurban areas:

According to VHLSS data, the rural area including households live in communes and the urban area including household live in wards

(b.ttp:/ /www.gso.gov vn/nada/ddibrowser/? id=4)

• Income quintiles (expenditure quintiles):

Households are divided to 5 income quintiles base on each household's income Each quintile makes up 20% of the total number of households from the poorest quintile (Quintile 1) to the richest quintile (Quintile 5)

• Household:

The term "household" adopted by the VHLSS was "a number of people who share lodging, income, expenditure for at least 6 months out of the last 12 months"