INTRODUCTION
Problem Statements
Foreign language education, particularly in English, has become a significant issue globally, especially with the rise of information and communication technologies that connect nations English serves as a common intermediary language, facilitating understanding and cooperation across cultures Most scholarly and technical knowledge, including books and academic papers, is predominantly published in English, reinforcing its global dominance According to Bolton (2006), English is the fastest spreading language worldwide, with over 1.2 billion speakers as of 2003, making it the most widely used language for communication across diverse linguistic backgrounds.
According to Crystal (2003), English is spoken in 75 countries worldwide, and Graddol (2008) predicts that the number of people learning English as a second language could reach nearly two billion in the next decade Modiano (2006) notes a shift in Eastern European countries, where English has replaced Russian as a primary school subject Recognizing the critical role of English education, Nergis (2011) emphasizes the need for government investment in this area In developing nations like Turkey, there is a consensus that enhancing English proficiency is essential for meeting modern communication demands, facilitated by the Turkish language reform.
In Vietnam, Thinh (2006) noted a remarkable surge in English language acquisition over the decade leading up to the country's accession to the WTO Statistics reveal that over 90% of foreign language learners in Vietnam are studying English, significantly outpacing other languages like Chinese, Japanese, and French.
The demand for English language skills has surged in Ho Chi Minh City, with over 900,000 learners flocking to language centers due to the necessity for job-seeking and overseas studies This trend has prompted the introduction of English programs in universities, high schools, and government agencies Following Vietnam's accession to the WTO and the subsequent influx of foreign investment, the need for a qualified labor force capable of communicating with international partners has become crucial As a result, learning English has emerged as an urgent priority in the context of today's global economic integration Many parents are now prioritizing English language education for their children from an early age to ensure a strong linguistic foundation.
This study aims to examine the significance that parents in Ho Chi Minh City place on English education by analyzing their spending on their children's English language education at the primary school level The findings will highlight the crucial role of English language education for individual development and the broader advancement of the country, emphasizing the need for investment from both private individuals and the government in Vietnam.
Research Objectives
General objective: To investigate household’s expenditure on English language education for children at starting school age (primary school)
(1) To identify the importance of English language education in a household’s decision
(2) To identify the determinants of household’s expenditure on English education for primary school children in Ho Chi Minh City h
Research Questions
This study investigates the influence of various demographic factors, including parents' income, work environment, English proficiency, age, gender, marital status, location, home ownership, and employment status, on their expenditures for children's English education We aim to determine whether these factors significantly impact financial investment in English learning and to quantify their contributions.
Research methodology
This study builds on the education expenditure theory proposed by Pritchett and Filmer (1999) and utilizes both descriptive statistics and regression analysis to assess the significance of English education and its expenditure determinants Initially, descriptive statistics provide an overview of English education spending among households in Ho Chi Minh City, considering factors such as total income, total expenditure, and household characteristics Subsequently, ordinary least squares (OLS) regression identifies influential factors affecting English language expenditure for primary school-aged children Additionally, logit regression examines whether these determinants influence households' decisions to invest in English education Finally, ordered logit regression addresses the issue of the dependent variable, English education expenditure, being ordinal rather than continuous, ranking expenditures from zero to six levels that are not equal.
The structure of the study
This study is structured into five chapters, beginning with an introduction in Chapter 1 Chapter 2 explores three theories related to education expenditure, including the household production function and the income elasticity of education spending, while also referencing recent empirical research on education expenditure in Vietnam and globally Chapter 3 details the research methodology, encompassing data collection methods, variable explanations and measurements, as well as model specifications.
Chapter 4 provides an overview of education in Ho Chi Minh City, with a specific focus on English education It examines the influencing factors that affect English education expenses for primary school-aged children, including parents' total income, expenditure, education level, English proficiency, and various demographic characteristics of the household.
LITERATURE REVIEW
Theoretical literature
In order to comprehend the purpose of this research, we first discuss education expenditure theories
Theory of Education expenditure: Education production function
Pritchett and Filmer (1999) explore the theory of education expenditure through the lens of the education production function, which illustrates how various inputs contribute to educational outcomes Their research focuses on the causal relationships between school inputs—such as teacher education levels, class size, teacher experience, and educational resources like textbooks—and student achievement Additionally, they highlight the impact of non-school inputs, including family factors, environmental influences, and children's innate abilities, on educational outcomes.
The specific function of education production is defined as below:
In this framework, Cit represents children's outputs, Sit signifies school inputs, and Fit indicates non-school inputs from the family, while Ii refers to children's congenital abilities The term used encompasses the fixed student contribution, acknowledging the absence of data sets to measure nonfigurative variables such as congenital ability.
Theory of household production function
Becker (1965) and Muth (1966) introduced a household production function model to illustrate how commodities purchased in the market serve as inputs for household production Specifically, they applied this model to the production function related to child health.
Whereas, Yj are other good affect child’s health; Ik present for health input and à is the family-specific health endowments such as genetic characteristics or environmental factors
Later on, from this household production function, they developed household’s reduced-form demand function:
Whereas, p is the price of goods or services; F is the exogenous income and à is the family specific endowment
To measure household expenditure on English language education, it is essential to view education as a consumable good This analysis examines various factors influencing the household production function, including household income, family-specific endowment factors, and other demographic variables Specifically, family-specific endowment factors considered in this study include the English proficiency of parents, reflecting genetic traits, and the parents' working environment, representing environmental influences.
Theory of income elasticity of educational expenditure:
Benson (1961) highlights that household income significantly influences educational expenditure, as indicated by the income elasticity of education Households are categorized into low, middle, and high-income groups Specifically, low and high-income households typically exhibit an income elasticity of education ranging from zero to one In contrast, middle-income households demonstrate a different value of income elasticity, reflecting their unique financial dynamics in relation to educational spending.
Middle-income families tend to prioritize their children's education, often investing more in educational expenses compared to other income groups In contrast, low-income families show less concern for the quality of schooling, leading to a projected increase in education spending that is lower than their overall income growth.
Use the Lorenz curve to measure the distribution of income to the education expenditure The form of income elasticity of educational expenditure as follows:
Whereas E i is the elasticity of income v i (x) is the Engel function of the i expenditure items.
Empirical literature
This research aims to analyze the factors influencing expenditures on English language education, viewing it as a human capital investment, as highlighted by Espenshade (1997) The study underscores that spending on English education is multifaceted and cannot be explained by a single independent variable While previous studies have primarily focused on general education expenditures, there is a lack of detailed analysis specifically addressing English language education Key determinants of educational spending identified in earlier research include household income, overall household expenditure, parental education levels, employment sector, employment status, age and gender of the household head, maternal employment, household size, and homeownership status (Kanellopoulos, 1997; Tansel, 2006; Donkoh, 2011) Building on these insights, we propose a new model to better understand the complexities of English language education expenditure.
8 determinants of spending on English language education We will talk about the specific model later in the methodology section
Household income significantly influences education expenditure, as wealthier families tend to prioritize investments in their children's education to improve their living standards This trend highlights the importance of education as a key factor in enhancing future opportunities for the next generation (Donkoh, 2011; Glewwe, 1999; Tansel and Bircan).
(2006) and Donkoh and Amikuzuno (2011) found that when the household’s income rise they are more willing to pay for the private tutoring
This article examines the relationship between income and education expenditure using income educational expenditure elasticity as a proxy Research indicates that the Tobit model is commonly used to assess this elasticity Notably, Hashimoto and Heath (1995) found that middle-income households exhibit an income elasticity of education expenditure greater than one, suggesting that their education spending increases significantly with income In contrast, lower and higher-income households show an elasticity ranging from zero to one, indicating a lesser impact of income changes on their education spending Additionally, studies by Kanellopoulos and Psacharopoulos (1997) in Greece and Hashimoto and Heath (1995) in Japan report income elasticity values of 3.18 and 2.35, respectively, both concluding that education expenditure functions as a luxury good However, Tansel and Bircan (2006) challenge this notion, finding unitary elasticity in Turkey, which suggests that education is not viewed as a luxury good in all contexts.
This research examines the role of informal English language education, particularly private tutoring According to Stevenson and Baker (1992), households are increasingly investing in informal educational activities to ensure their children achieve higher education and successful careers Consequently, the likelihood of spending on educational activities rises with household expenditure levels Notably, families allocate approximately 11.2% of their total expenses to tutoring, with over 63% of this amount dedicated to foreign language education, as highlighted by Kanellopoulos and Psacharopoulos (1997) Their findings indicate a significant correlation between parental spending on education and overall household expenses.
To address the issue of simultaneity in variable selection, some researchers argue that the relationship between total and individual expenses can introduce bias To mitigate this, they recommend using instrumental variables like household income (Liviatan, 1961) Tansel and Bircan (2006) tested the exogeneity of total household expenses using a Tobit model, as defined by Smith and Blundell (1986), which involves two steps: first, regressing total spending on household income, and second, incorporating the residuals from this regression into the Tobit model for additional education spending to assess significance Their findings, which reject the null hypothesis at the one percent level, support the use of total household expenses as an explanatory variable over household income.
Numerous studies have explored the relationship between education expenditure and parental education, including research by Knight and Shi (1996), Kanellopoulos and Psacharopoulos (1997), Tansel and Bircan (2006), Qian and Smyth (2011), and Donkoh and Amikuzuno (2011) Most of these studies indicate a positive correlation, demonstrating that higher levels of parental education lead to increased spending on children's education Specifically, Knight and Shi (1996) highlight two key factors influencing this relationship.
Research by Kanellopoulos and Psacharopoulos (1997) reveals that parental educational attainment is the most significant factor influencing children's education expenditure Notably, a father's education has a greater impact than a mother's education Households led by individuals with higher education levels are more inclined to allocate a larger portion of their income to educational expenses Specifically, for each additional year of education attained by the household head, there is an approximate 2.2% increase in the likelihood that the family will invest in their children's education.
Research highlights the significant influence of a mother's education on family decisions regarding educational spending According to Qian and Smyth (2011), households where mothers have completed senior secondary school or higher tend to invest more in their children's education Similarly, Donkoh and Amikuzuno (2011) suggest that well-educated individuals are more likely to provide their children with similar educational opportunities, fostering an expectation for them to achieve higher education levels.
Research by Kanellopoulos and Psacharopoulos (1997) and Andreou (2012) indicates that households in the private sector are more inclined to invest in their children's education Additionally, Qian and Smyth (2011) found that families with fathers in professional occupations tend to allocate more resources towards education Therefore, the private sector plays a crucial role in influencing educational spending within households.
According to Kanellopoulos and Psacharopoulos (1997), Donkoh and Amikuzuno
In 2011, research indicated that parents' occupations significantly influence their willingness to invest in their children's education There are two prevailing perspectives on this matter: one suggests that a parent's occupation positively impacts educational investment, while the other posits that lower-income households may be less inclined to prioritize education due to financial constraints.
Many upper-class parents tend to compensate for their lack of education by investing heavily in their children's education, often viewing it as a necessary expense However, some believe that education holds relatively low value compared to their overall household wealth, leading them to underestimate its importance Despite this perspective, Andreou (2012) suggests that parents' occupations do not significantly influence their private spending on education.
Numerous studies highlight the age of the household head as a crucial factor influencing educational investment decisions Research by Kanellopoulos and Psacharopoulos (1997), Donkoh and Amikuzuno (2011), and Andreou (2012) indicates that older household heads tend to exhibit a negative correlation with their educational spending choices.
Donkoh and Amikuzuno (2011) analyze the impact of household head age on education spending, finding that older heads are less willing to invest in education They suggest that younger household heads tend to be more progressive and recognize the long-term benefits of education, contrasting with the attitudes of their older counterparts.
Kanellopoulos and Psacharopoulos (1997) categorize age into six groups: under 25, 25-34, 35-44, 45-54, 55-64, and over 65 Their research reveals that the highest education spending is among individuals aged 35-44, with approximately 62.6% of heads in this group willing to invest in education Notably, the likelihood of education expenses decreases in the subsequent age groups.
Tansel and Bircan (2006) presented a different perspective on private tutoring expenditure in Turkey, revealing that as the age of the household head increases, there is a greater willingness to invest in their child's tutoring This study aims to examine the impact of the household head's age on educational spending for children.
Conceptual framework
Overall of the above empirical literature, the study is going to test the relationship between the private expenditure for English education and the explanation variables as the following: h
Figure 2.1 Analytical framework for the impact explanatory variable on English education expenditure
Age Gender (dummy V) Marital status Location (dummy V) Home ownership (dummy V) Employment status Number of children
PRIVATE EXPENDITURE for English language Education
RESEARCH METHODOLOGY
Sampling strategy and data collection
The study employed face-to-face surveys for primary data collection, utilizing a questionnaire that addressed key topics such as total household income, parents' education and English proficiency, parental careers, and various demographic factors including age, gender, and marital status The surveyed households consisted of families with children enrolled in pre-primary and primary education in Ho Chi Minh City Participants were randomly selected from different districts and sub-districts, ensuring a diverse representation by choosing households randomly from a designated ward within each district.
20 households were interviewed in each districts/sub-districts The internal districts in Ho Chi Minh City comprise district one to twelve, Tan Phu, Tan Binh, Binh Tan, Binh Thanh,
Go Vap, Phu Nhuan The sub-districts include Binh Chanh, Hoc Mon, Cu Chi, Nha Be whereas we excluded the Can Gio’s observation because of its low population density
In cases where a surveyed household does not have children of primary school age, the next neighboring house is approached for replacement If the subsequent three households also lack children in the targeted age group, the sample is left empty Additionally, some selected households may submit blank answer sheets due to personal reasons As a result, certain districts failed to provide the required 20 responses, leading to a total of 267 observations for this study A detailed table of the questionnaires can be found in the annex.
Variables’ measurement and explanation
3.2.1 Current expenditure on English education
Expenditure on English is a complex and varied factor, regarded as an investment in human capital (Espenshade and Fu, 1997) It encompasses the costs associated with this investment and the anticipated returns that individuals expect from their spending.
This study explores the factors influencing English education expenditure, addressing the scarcity of research in this area Building on findings from Kanellopoulos and Psacharopoulos (1997), Tansel and Bircan (2006), and Donkoh and Amikuzuno (2011), we analyze variables such as total household income, overall expenditure, parental education levels, employment sectors, age and gender of the household head, maternal employment status, household size, and homeownership The goal is to determine how these factors affect household spending on English education.
This study evaluates parental interest in children's English education through current expenditure levels, categorized into six distinct ranks A rank of zero indicates no expenditure, while rank one reflects spending between 0 and 300,000 VND Rank two encompasses expenditures from 300,001 to 500,000 VND, rank three ranges from 500,001 to 1,000,000 VND, rank four includes spending from 1,000,001 to 2,000,000 VND, and rank five signifies expenditures exceeding 2,000,000 VND These classifications are derived from average fees for extra English classes at schools and English centers, with data gathered from ten schools and eleven English centers catering to primary children at the beginning and pre-intermediate levels A comprehensive list of the schools and English centers is available in the appendix.
The previous chapter highlighted the significant influence of household income on education expenditure As households become wealthier, they prioritize improving their living standards, with education being a key area of investment (Donkoh and Amikuzuno, 2011) Research by Glewwe and Desai (1999), Tansel and Bircan (2006), and Donkoh and Amikuzuno (2011) indicates that rising household income correlates with an increased willingness to invest in private tutoring.
English education, a form of private tutoring, is significantly influenced by a household's total income To assess the impact of income on education expenditure, two evaluation methods are employed: direct and indirect evaluation The direct evaluation utilizes OLS and Probit/Logit models to analyze the effects of varying income levels on current payments for English education In contrast, the indirect evaluation categorizes income into low, middle, and high brackets to calculate the income elasticity of English education expenditure, using the Tobit model for analysis.
In this study, the household’s total income variable is also divided into five levels:
The classification of income groups can be assessed through two methods: direct and indirect evaluation In the direct evaluation method, income brackets are defined as follows: below 5,000,000 VND, from 5,000,000 VND to 10,000,000 VND, from 10,000,000 VND to 20,000,000 VND, and over 20,000,000 VND Conversely, the indirect evaluation method categorizes income levels into low, middle, and high income According to the World Bank's 2012 classification, low income is designated for groups with a Gross National Income (GNI) per capita of $1,035 or less.
Income classifications are categorized as follows: low income ranges from $1,036 to $4,085, equating to 1,811,250 VND or less per person per month; lower middle income spans from $4,086 to $12,615, corresponding to 1,813,000 to 7,148,750 VND per person per month; upper middle income is defined as $12,616 or more, translating to 7,148,750 to 22,076,250 VND per person per month; and high income is set at 22,078,000 VND or more per person per month In this context, the low income group is identified as level (1), below 5,000,000 VND, while the middle income group includes levels (2), (3), and (4), and the high income group is classified as level (5).
Empirical studies indicate that middle-income households exhibit the highest income elasticity of education expenditure, exceeding a value of one In contrast, lower and higher income classes show elasticity values ranging from zero to one This suggests that as income increases, middle-income households significantly increase their education spending compared to other income groups Furthermore, research by Hashimoto and Heath suggests that lower and higher income families appear less concerned about the quality of their children's education.
(1995) However, some studies found no difference between these income groups which affect on education demand For these reasons, the paper is going to investigate whether h
The study investigates whether an 18% increase in income would positively influence expenditure on English education and examines the differences in income elasticity of English expenditure across low, middle, and high-income groups Building on theoretical frameworks and previous empirical research, the study aims to test this hypothesis.
H1: The household’s total income has positive relationship with the English education expenditure children
This research examines the relationship between total household expenditure and spending on English education Stevenson and Baker (1992) found that households are willing to invest more in informal educational activities when they aspire for their children to attain higher education and successful careers As a result, there is a clear correlation between the level of household expenditure and the likelihood of investing in educational activities Kanellopoulos and Psacharopoulos (1997) highlight that spending on foreign language education constitutes over sixty-three percent of total educational expenditures Thus, it is evident that parental spending on their children's education significantly increases in line with overall household expenses.
In this study, we utilized multiple-choice responses to assess total expenditure, categorizing it into six levels based on the VHLSS 2010 dataset: (1) below 2,000,000 VND, (2) 2,000,000 to 3,500,000 VND, (3) 3,500,000 to 5,000,000 VND, (4) 5,000,000 to 10,000,000 VND, and (5) over 10,000,000 VND The research aims to examine two key aspects: first, the relationship between household total expenditure and English education expenditure; second, whether the simultaneous analysis of total and individual expenses introduces bias when using income as an instrumental variable, as discussed by Liviatan (1961) To achieve this, Tansel and Bircan (2006) employed a Tobit model to test the exogeneity of household total expenditure.
Smith and Blundell (1986) outline a two-step process for analyzing household spending The first step involves regressing total spending against household income In the second step, the residuals from this regression are incorporated into a Tobit model to evaluate the significance of additional education spending If the regression results indicate a significant finding that leads to the rejection of the null hypothesis, total household expenses can be utilized as an explanatory variable in place of household income.
In summary, the study is going to test the proposed hypothesis as following:
H2: The level of household’s total expenditure has positive relationship with the English education expenditure
H3: The current expenditure on extra education relates to the English education expenditure
Many studies found the positive impact of parental education to their spending for children’s education such as Knight and Shi (1996), Kanellopoulos and Psacharopoulos
(1997), Tansel and Bircan (2006), Qian and Smyth (2011) and Donkoh and Amikuzuno
A study by Knight and Shi (1996) emphasizes that a father's education is a more significant factor than a mother's education This research aims to explore the relationship between parents' education levels and their expenditure on English education Following the VHLSS 2010 framework, education levels are categorized from one to eight, corresponding to the completion of Primary school, Junior high school, Senior high school, Primary vocational school, Colleges, University, Master’s, and Doctorate The study seeks to test the stated hypothesis.
H4: Parents’ education level has a positive relationship with the expenditure on English education
H5: The level of English of Parents is positively related to their spending on English education h
Households employed in the private sector tend to invest more in their children's education, particularly in English, as highlighted by Kanellopoulos and Psacharopoulos (1997) and Andreou (2012) This study aims to explore how different employment sectors influence English education expenditure Specifically, it examines whether parents working in foreign companies or NGOs in Ho Chi Minh City are more likely to allocate funds for their children's English education compared to those in other sectors The employment sectors are categorized into four groups: public sector, domestic private sector, foreign private sector, and others, with the latter being treated as a control variable in the regression analysis Ultimately, the research tests the hypothesis regarding the impact of these employment sectors on education spending.
H6: Parents’ sector of employment could explain the current expenditure on English education
Research indicates that mothers and fathers exhibit differing spending patterns on their children To explore this further, we will examine how the gender of the household head influences decisions regarding English education expenditures Drawing from empirical studies, we will utilize a dummy variable where zero represents female heads and one represents male heads Subsequently, we will test the following hypothesis.
H7: There is a difference between a mother household’s head and father household’s head in their way of expense on their children’s English education h
Generally, the age of head appeared to be negative efficiency to their choice of spending according to Kanellopoulos and Psacharopoulos (1997), Donkoh and Amikuzuno
Model specification
To investigate the impacts of all suggested independent variables to the English education expenditure, this study use five models to test the twelve proposed hypotheses
This study employs Ordinary Least Squares (OLS) to identify the factors influencing English language expenditure among primary school-aged children, illustrating the relationship between explanatory variables and English spending To assess the impact of these determinants on household decisions regarding English education expenditures, logit regression is utilized, as the standard OLS model is insufficient for this analysis Additionally, ordered logit regression addresses the scenario where the dependent variable, English education expenditure, is not continuous but rather ranked from zero to six, with unequal intervals between levels The overall model will be structured as outlined in the study.
In this study, E signifies the current expenditure on English education, while X represents income levels, Y indicates the English proficiency of parents, and Z refers to the parents' work environment Additionally, T encompasses various demographic factors, including parents' age, the gender of the household head, whether the wife is employed, the number of children, household location, and ownership status.
Various regression methods are utilized to achieve the research objectives, beginning with the application of Ordinary Least Squares (OLS) regression to assess the relationship between dependent and independent variables.
In this case, there are two specific models employed as following:
Model 3 and model 4: Logit models with income (model 3) and expenditure (model 4)
The logit regression is utilized to examine the impact of various determinants on households' decisions to invest in English language education Given that the spending behavior is a binary variable, a logit model is employed for this analysis Unlike OLS regression, this model estimates the probability (P) of household expenditure on education based on different characteristics.
The probability (P) of a household's spending on education is influenced by various independent variables, including household characteristics such as income, the age and gender of the household head, and location, as identified by Kanellopoulos and Psacharopoulos (1997) This study utilizes two regression models: Model 3, which focuses on income and other explanatory variables, and Model 4, which examines total expenditure alongside other explanatory variables for comparison Additionally, the marginal effects of each explanatory variable are calculated after applying the logit model to assess their impact on English education expenditure.
The English education expenditure in this study is categorized into six ordinal levels, ranging from zero to five, making it an ordinal dependent variable Therefore, the ordered logit model is an appropriate method for examining the relationship between this ordered dependent variable and the independent variables in the analysis.
The ordered outcomes are estimated as a linear function of explanatory variables and cutpoints According to Fu (1998), the probability of observing a specific outcome corresponds to the estimated probabilities of independent variables, adjusted by an error term, within the defined cutpoint levels This relationship can be represented by a specific function.
The current expenditure for English education at level 0 is defined as no expenditure, while levels 1 through 5 correspond to increasing expense ranges: level 1 for expenses below 300,000 VND, level 2 for expenses from 300,000 VND to 500,000 VND, level 3 for expenses from 500,000 VND to 1,000,000 VND, level 4 for expenses from 1,000,000 VND to 2,000,000 VND, and level 5 for expenses exceeding 2,000,000 VND The independent variables, represented as x1, x2, …, xkj, include the household's total income, total expenditure, current payments for extra classes, parents' education levels, parents' English proficiency, and various demographic characteristics such as the parents' age, the gender of the household head, whether the wife is employed, the number of children, the household's location, and ownership status The coefficients β1, β2, …, βk represent the explanatory variables in this context.
K 1 , K 2 ,… K k-1 , are the cutpoints u j is the logistically distributed of error term h
Table 3.1: Variable description and expected sign for Model 1
Variable name Description Units Expected sign
CPE Current payment for English education VND dIncome2 Dummy household’s total income variable Equal to income rank from
5,000,000 to 10,000,000; otherwise equal to income below 5,000,000 VND Positive (+) dIncome3 Dummy household’s total income variable Equal to income rank from
10,000,000 to 20,000,000; otherwise equal to income below 5,000,000 VND Positive (+) dIncome4 Dummy household’s total income variable Equal to income rank over
20,000,000; otherwise equal to income below 5,000,000 VND Positive (+) dtotalexp2 Dummy total expenditure variable Equal to rank from 2,000,000 to
3,500,000; otherwise equal to income below 2,000,000 VND Positive (+) dtotalexp3 Dummy total expenditure variable Equal to rank from 3,500,000 to
5,000,000; otherwise equal to income below 2,000,000 VND Positive (+) dtotalexp4 Dummy total expenditure variable Equal to rank from 5,000,000 to
The study examines various financial variables affecting household expenditures and education levels A total expenditure dummy variable is established, indicating a positive correlation for households with incomes exceeding 10,000,000 VND, while those earning below 2,000,000 VND show a different trend Additionally, current payments for extra education are positively associated with household financial health The education level of the wife, ranked from 1 to 8, also demonstrates a positive impact on overall expenditures.
The education level of a husband is categorized on a scale from 1 to 8, indicating varying degrees of educational attainment Additionally, the husband's proficiency in English is assessed with a dummy variable, where a score of 1 signifies knowledge of English, while a score of 0 indicates a lack of proficiency Similarly, a corresponding dummy variable evaluates the wife's English proficiency, following the same scoring system.
The article analyzes various factors influencing household dynamics, highlighting the impact of employment and location on family status The dummy variable for the wife's employment status indicates a positive correlation when she stays at home, while the household's urban location also shows a positive effect Conversely, the ages of both the husband and wife present negative correlations, suggesting that as their ages increase, certain dynamics may shift unfavorably Additionally, homeownership is positively associated with the household's stability, reflecting the significance of owning a house in enhancing family well-being.
Nchildren Number of children Child (+)/(-)
EMPIRICAL ANALYSIS RESULTS
General information of the Household’s characteristic in HCM City
A recent survey conducted in Ho Chi Minh City reveals that the education levels of parents with children in primary school predominantly fall within the senior high school and university categories Notably, the educational distribution between husbands and wives is nearly equivalent; however, husbands generally possess higher education levels Specifically, 29.96% of husbands have completed university education compared to only 25.09% of wives Additionally, 7.87% of households have husbands who hold a Doctor of Philosophy degree, while no wives have attained this level of education.
A comparison of the Vietnam Household Living Standard Survey (VHLSS) 2010 reveals significant findings regarding family education levels Firstly, it highlights that husbands tend to have a higher education level than their wives Additionally, in Ho Chi Minh City, the educational attainment of parents is predominantly concentrated in no education, primary, junior high, and senior high school levels, with university-educated parents comprising a notable 11.69% of the total.
Figure 4.1 The education level of the parents in Ho Chi Minh City – Author’s survey 2013
Figure 4.2 The education level of the parents in Ho Chi Minh City – VHLSS 2010
General information of English education in Vietnam
In Vietnam, the English in general education is assigned as a compulsory subject from junior high school and senior high school From the year 1982 to 2002, it was set into h
Vietnam has implemented two English learning programs: a three-year curriculum for students in grades 10 to 12 and a seven-year program for those starting from grades 6 to 12 Since 2002, English has become a mandatory subject in junior and senior high schools and an elective in primary schools In primary education, English is taught from grades 3 to 5 with two periods per week, while junior and senior high schools have three periods per week This comprehensive English language program aims to equip students with essential skills in listening, speaking, reading, and writing, ensuring they possess a foundational proficiency in English by the time they graduate from high school, enabling them to comprehend textbooks at a similar level with the aid of a dictionary.
Between 2008 and 2013, Vietnam saw a significant increase in foreign language centers, rising from 843 to 1,935, as reported by the Ministry of Education and Training (MOET) This growth aligns with the National Foreign Languages 2020 Project, established by Decision No 1400/QD-TTg on September 30, 2008, aimed at enhancing foreign language education within the national system The project focuses on developing a comprehensive qualifications framework with six levels that align with international standards It also mandates new foreign language training programs in basic education, ensuring students achieve specific qualifications upon graduation: level 1 for primary, level 2 for junior high, and level 3 for senior high Additionally, the project includes training university students in new foreign language programs and integrating English instruction into core subjects for final-year students.
However, the actual situation of teaching and learning English in Vietnam is still insufficient Along with learning English at school, some parents also equip their children h
Vietnam's demand for English education has led to the rise of numerous domestic and international language centers, such as the British Council, VUS, VASS, Apollo, and ILA, particularly in Ho Chi Minh City and Hanoi These centers are not only operating efficiently but also generating significant profits, prompting plans for large-scale national expansion Additionally, many smaller centers and various educational institutions, both private and public, have emerged to cater to the growing need for English language proficiency across the country.
Table 4.1 The development of number of foreign language center through the years
Source: Ministry of Education and Training
Importance of English expenditure in household’s decision
Figure 4.3 illustrates that households in Ho Chi Minh City allocate a significant portion of their total education expenditure to English education.
Basic education accounts for an average of 61.4% of total school education fees In relation to household finances, it represents 4.7% of total income and 6.7% of total expenditure The study indicates that English education is viewed similarly to private tutoring, with 84.7% of households willing to invest in English classes and 84.3% for additional subjects Most households spend between 300,000 VND and 500,000 VND per month on these educational services For detailed statistics, refer to the appendix.
Figure 4.3 The distribution of current English education expenditure in basic education, total income and total expenditure h
Empirical analysis results
This section presents the empirical analysis results and discussion, beginning with descriptive statistics for the dependent and explanatory variables, alongside their correlation matrix It then summarizes the estimated results from three regression models, including the marginal effects obtained from these regressions and a detailed discussion of the findings.
This section presents the descriptive statistics derived from the survey data, serving as a foundation for subsequent model discussions The primary dependent variable is illustrated through histogram graphics, while the explanatory variables are statistically summarized in Table 4.1, which includes metrics such as mean, frequency, maximum, minimum, and standard deviation Notably, 84.7% of households are willing to invest in their children's English education, with the majority spending around 300,000 VND or less per month, accounting for approximately 41% of all households.
Figure 4.4 presents a histogram illustrating the current expenditure on English education in Ho Chi Minh City The expenditure categories are as follows: 0 indicates no spending on English education, while 1 represents expenditures ranging from 0 to 300,000 VND Categories 2 and 3 reflect spending levels between 300,000 to 500,000 VND and 500,000 to 1,000,000 VND, respectively Furthermore, category 4 encompasses expenditures from 1,000,000 to 2,000,000 VND, and category 5 includes any spending exceeding 2,000,000 VND.
Figure 4.5 Histogram graphic of the dummy variable of current payment for English education in Ho Chi Minh City (0 – no expenditure on English education, 1 – present expenditure on English education)
Dummy variable of current payment for English education
The survey shows that the groups of household which gain total income from 5,000,000 VND to 10,000,000 VND per month mostly predominate over the households in
In Ho Chi Minh City, the distribution of household income reveals that the middle-income group dominates, accounting for nearly 70% of total households The low-income group follows at 23.60%, while the high-income group comprises only 6.74% Among the five income levels, households earning between 5 to 10 million VND represent the largest segment at 40.07% Other notable groups include those earning below 5 million VND at 23.60% and those earning between 10 to 20 million VND at 20.60% Conversely, the highest income bracket, with earnings over 30 million VND, is the least populated, comprising only 6.75% of households This data underscores the significant presence of the middle-income demographic in Ho Chi Minh City.
In Ho Chi Minh City, household income distribution is categorized into two main formats: (a) five specific income brackets—below 5 million VND, 5 to 10 million VND, 10 to 20 million VND, 20 to 30 million VND, and over 30 million VND; and (b) three broader classifications—low income, middle income, and high income.
Most households surveyed spend between 3,500,000 VND and 10,000,000 VND per month, accounting for 74.15% of total expenditures, with the 3,500,000 VND to 5,000,000 VND bracket representing 40.82% This aligns with the predominant income group, as middle-income families primarily allocate their earnings to essential needs such as food, education, and healthcare A strong correlation exists between total income and total expenditure, indicating that as income increases, so does spending capacity Additionally, a significant majority of households invest in private tutoring for their children, with only 15.67% opting out, highlighting the growing emphasis parents place on education.
The study explores the demographic characteristics of households in Ho Chi Minh City, revealing that over 35% of parents have English proficiency, with husbands showing a higher percentage than wives In terms of education, 29.96% of husbands have completed university, while the percentage for wives is lower Employment data indicates that 25.09% of husbands and 28.46% of wives work in the public sector, and 34.08% of husbands and 25.47% of wives are employed by domestic companies, with fewer working for foreign companies or NGOs Notably, about 25% of individuals are self-employed, engaging in market sales or temporary labor Additionally, 18.66% of wives are homemakers, and 77.9% of household heads are male The age range for husbands is 26 to 58 years, while wives are between 25 and 55 years old The majority, 75%, reside in urban areas, and over 64% own their homes Lastly, the average number of children per household is 1.65, reflecting a trend towards smaller families with one to two children.
Table 4.2 Summary statistic of variables (a)
Cpe cpe0 cpe1 cpe2 cpe3 cpe4 cpe5
Income dincome1 dincome2 dincome3 dincome4 dincome5
Total expenditure dtotalexp1 dtotalexp2 dtotalexp3 dtotalexp4 dtotalexp5 dtotalexp6
1.50 8.99 40.82 33.33 11.99 3.37 cpec dcpec1 dcpec2 dcpec3 dcpec4 dcpec5
15.67 44.78 26.49 8.58 4.48 eduh Primary Senior HS Junior HS Primary VS College University Master PhD
6.74 15.36 28.84 8.99 0 29.96 2.25 7.87 eduw Primary Senior HS Junior HS Primary VS College University Master PhD
Englishh_0 38.20% husband know English 61.80% don’t know English
Englishw_0 35.96% wife know English 64.04% don’t know English
Private (Foreign.) NGOs Stay at home
Private (Foreign.) NGOs Stay at home
Wifeworking 81.34% wife are working 18.66% wife stays at home
Genderhh 77.9% HH are male 22.1% are female
Local 75% HH in urban 25% HH in Suburban
Ownh 64.18% own a house 35.82% not own a house
Table 4.2 Summary statistic of variables (b)
Variable Mean Std Dev Min Max
Table 13 in the appendix illustrates the correlation among explanatory variables, revealing that only the education levels of both spouses and their ages exhibit a correlation This finding aligns with the social characteristics of Vietnam Additionally, the data indicates that the highest income group (group five) tends to spend more, leading to a significant correlation between this income group and total expenditures in groups five and six Furthermore, there is a notable correlation between total expenditure and current payments for English education, supporting Benson's (1961) theory of income elasticity in education expenditure.
The analysis reveals a clear correlation between household income and expenditure on English education Households earning below 5 million VND per month predominantly allocate less than 300,000 VND (37%) for English education, with 15% spending between 300,000 and 500,000 VND, and 27% contributing nothing at all In contrast, households with incomes exceeding 30 million VND show a marked increase in spending, particularly in the 1,000,000 to 2,000,000 VND and above 2,000,000 VND categories Similarly, total household expenditure plays a significant role, with those spending more overall also investing more in English education, particularly in the higher expenditure brackets Conversely, households with lower total expenditures prioritize basic living costs and consequently invest less in English education.
Table 4.3 The relationship of total income and current payment for English education
Current payment for English education (VND)
20 - 30 millions VND 2% 8% 6% 23% 12% 0% over 30 millions VND 0% 3% 6% 3% 40% 25%
Table 4.4 The relationship of household’s total expenditure and current payment for
Current payment for English education (VND)
10 - 15 millions VND 2% 5% 13% 23% 36% 25% over 15 millions VND 0% 2% 4% 3% 12% 25%
This chapter addresses the research questions outlined in Chapter One and relies on the hypotheses presented in Chapter Three The study focuses on the significance of English education in Vietnam by examining how factors such as parental income, working environment, English proficiency, and other demographic variables influence expenditures on English education for children Table 4.5 presents the results of two regression methods, as defined in the methodology section, to analyze the relationship between these explanatory variables and the current payments for English education among children in Ho Chi Minh City.
Table 4.5 The results of the models
Current payment for English education
Coefficient Coefficient Coefficient Marginal effect (dy/dx)
Coefficient Marginal effect (dy/dx)
Current payment for extra class
0.327 *** 0.309 *** 0.368 0.020 0.188 0.011 (0.00) (0.00) (0.16) (0.18) (0.47) (0.48) Dummy English var of husband
0.009 0.049 -1.121 -0.069 -0.971 -0.063 (0.96) (0.80) (0.08) (0.14) (0.13) (0.18) Dummy English var of wife
-0.393 ** -0.356 * -0.489 -0.028 -0.267 -0.016 (0.04) (0.06) (0.42) (0.47) (0.68) (0.70) Dummy Public sector of husband
0.096 0.082 -0.365 -0.021 -0.300 -0.018 (0.60) (0.66) (0.59) (0.60) (0.64) (0.66) Dummy Public sector of wife
( * ): Statistic significance at 10% level, ( ** ): Statistic significance at 5% level, ( *** ): Statistic significance at 1 % level
Models 1 and 3 examine the relationship between total income and various explanatory variables, while models 2 and 4 focus on the correlation between total expenditure and current spending on English education The findings from model 1 indicate a positive relationship between total income, current payments for extra education, and the education level of the wife with expenditures on English education Notably, households in the highest income bracket, earning over 20 million VND, show a statistically significant influence on English education spending at a 1% significance level This supports the perspectives of Glewwe and Desai (1999), Tansel and Bircan (2006), and Donkoh and Amikuzuno (2011), which suggest that higher income correlates with a greater willingness to invest in additional education.
Model 1 reveals that spending on extra classes positively influences expenditures on English education at a 1% significance level Interestingly, a wife's English proficiency negatively impacts English education spending at a 5% significance level This phenomenon can be attributed to the cultural traits of Vietnamese women, who often dedicate time to teaching their children at home Consequently, wives with higher English skills can educate their children themselves, thereby reducing the need for additional English education expenses, aligning with findings from Huy's research.
In 2012, a significant correlation was found between a wife's education level and her willingness to invest in her children's education, particularly in English, with results indicating a 1% significance level.
The analysis indicates that a wife's age positively correlates with spending on English education, with a significant level of 5% Additionally, households that own their homes tend to allocate more resources towards English education This is attributed to the financial stability provided by homeownership, which alleviates the burden of rent and allows families to redirect funds towards their children's education.
The study found that variables such as total expenditure, parents' education, employment sector, wife's employment status, gender of the household head, husband's age, household location, and number of children do not significantly affect the researched object In model 2, which examines the relationship between total expenditure and English education expenses, households with expenditures over 10 million showed a positive impact on English education spending at a 10% significance level However, when total income and total expenditure were analyzed together, results indicated that both factors positively influence household spending on English education at a 5% significance level Other controlled variables remained consistent with model 1, except for home ownership, which showed no significant effect on English education spending in this model.
CONCLUSIONS, RECOMMENDATIONS AND LIMITATIONS
Conclusions and recommendations
This study explores the significance of English education in Ho Chi Minh City and identifies the factors influencing households' decisions regarding English education spending for primary school-aged children The findings reveal that 84.7% of households invest in English education from the start of their children's schooling Key factors affecting English education expenditure include household characteristics such as income, the wife's education level, and her English proficiency In contrast, other variables like the husband's education level, working status, age, location, and number of children do not significantly impact spending on English education Notably, the results highlight the critical role of the wife in making educational expenditure decisions, underscoring her influence within the family.
The study highlights the critical importance of English education in Ho Chi Minh City and across Vietnam, urging the government to take action It reveals that English education expenditure is influenced by household income, the educational background and English proficiency of the wife, and home ownership status Notably, there is a significant disparity in spending on English education among different income groups, with only children from households earning over 20 million VND being able to access high-quality English classes.
Low-income households often struggle to afford education, creating inequities in children's learning opportunities To address this issue, government intervention is necessary to promote educational equity across different income levels Raising awareness among mothers about the importance of English education can encourage them to prioritize their children's learning Research indicates that many families value English education and enroll their children in English classes from a young age Therefore, the government should consider implementing English as a foreign language in primary schools, similar to successful bilingual education programs in various European countries.
Limitations
The study has notable limitations in data collection and methodology Firstly, due to privacy concerns, households were unable to provide exact figures for their total income and expenditure, only offering approximate monthly amounts Consequently, information on income and expenditure was gathered using rank values and dummy variables, which may introduce bias Secondly, calculating expenditures on English education is challenging for households with children in international schools, as these costs cannot be distinctly separated from general education expenses As a result, the research primarily focuses on households whose children attend regular schools Lastly, due to time constraints, the study is limited to analyzing the English education status in Ho Chi Minh City, without providing a comprehensive overview of English education across Vietnam, particularly in rural areas Future research should aim to broaden the scope of investigation and align with the Government's VHLSS survey.
The study highlights the reverse causality effect of English education expenditure on future income Additionally, it notes the inability to evaluate the income elasticity of English expenditure due to the dependent variable not being a string variable Therefore, future research should focus on calculating the income elasticity of English expenditure to provide a comprehensive understanding of household spending on English education.
Andreou, S N (2012) Analysis of Household Expenditure on Education in Cyprus Cyprus
Becker, G S (1965) A Theory of the Allocation of Time The economic journal, 75(299), 493-
Benson, C S (1961) The economics of public education: Houghton Mifflin Boston
Bolton, K (2006) World Englishes today The handbook of world Englishes, 240-269 Crystal, D (2003) English as a global language: Ernst Klett Sprachen
Dang, H.-A (2007) The determinants and impact of private tutoring classes in Vietnam
Donkoh, S., & Amikuzuno, J (2011) The determinants of household education expenditure in
Ghana Educational Research and Reviews, 6(8), 570-579
Espenshade, T J., & Fu, H (1997) An Analysis of English-Language Proficiency among U.S
Immigrants American Sociological Review, 62(2), 288-305 doi: 10.2307/2657305
Fu, V K (1998) Estimating generalized ordered logit models Stata Technical Bulletin Reprints, 8, 160-164
Glewwe, P., & Desai, J (1999) The economics of school quality investments in developing countries: An empirical study of Ghana: Macmillan London
In the context of global English, Graddol (2008) suggests that its prevalence may signal a decline in the traditional concept of 'English as a Foreign Language.' Additionally, Hashimoto and Heath (1995) analyze the income elasticities of educational expenditure among different income classes in Japan, highlighting the relationship between household income and educational investment Furthermore, Huy (2012) examines the key factors influencing educational expenditure in Vietnam, providing insights into the economic determinants of education in a developing context.
Jelani, J., & Tan, A K (2012) Determinants of participation and expenditure patterns of private tuition received by primary school students in Penang, Malaysia: an exploratory study
Asia Pacific Journal of Education, 32(1), 35-51
Kanellopoulos, C., & Psacharopoulos, G (1997) Private education expenditure in a ‘free education’country: the case of Greece International Journal of Educational Development,
Kim, S., & Lee, J.-H (2002) Private tutoring and demand for education in South Korea Paper presented at the a World Bank seminar
Knight, J., & Shi, L (1996) Educational Attainment and the Rural‐Urban Divide in China Oxford
Bulletin of Economics and Statistics, 58(1), 83-117
Liviatan, N (1961) Errors in variables and Engel curve analysis Econometrica: Journal of the
Modiano, M (2006) The Handbook of World Englishes, eds Braj Kachru, Yamuna
Kachru and Cecil L Nelson, Oxford: Blackwell
Muth, R F (1966) Household production and consumer demand functions Econometrica:
Journal of the Econometric Society, 699-708
Nergis, A (2011) Foreign language teacher education in Turkey: A historical overview Procedia-
Pritchett, L., & Filmer, D (1999) What education production functions really show: a positive theory of education expenditures Economics of Education Review, 18(2), 223-239
Qian, J X., & Smyth, R (2011) Educational expenditure in urban China: income effects, family characteristics and the demand for domestic and overseas education Applied Economics, 43(24), 3379-3394
Smith, R J., & Blundell, R W (1986) An exogeneity test for a simultaneous equation Tobit model with an application to labor supply Econometrica: Journal of the Econometric Society, 679-685
Stevenson, D L., & Baker, D P (1992) Shadow education and allocation in formal schooling:
Transition to university in Japan American Journal of Sociology, 1639-1657
Tansel, A., & Bircan, F (2006) Demand for education in Turkey: A tobit analysis of private tutoring expenditures Economics of Education Review, 25(3), 303-313
Thinh, D.H (2006) The role of English in Vietnam's foreign language policy: A brief history Annual
EA Education Conference 2006, Perth, Australia 9, 1-12.
Stevenson, D L., & Baker, D P (1992) Shadow education and allocation in formal schooling:
Transition to university in Japan American Journal of Sociology, 1639-1657
Tansel, A., & Bircan, F (2006) Demand for education in Turkey: A tobit analysis of private tutoring expenditures Economics of Education Review, 25(3), 303-313 h
Table 01 – OLS regression result – Model 01
_cons -1.363547 5194084 -2.63 0.009 -2.386623 -.3404717 Nrofchildren -.0440684 0860152 -0.51 0.609 -.2134919 1253552 ownh 2855928 1319733 2.16 0.031 0256458 5455399 local 031588 1384509 0.23 0.820 -.2411179 304294 agew 0436632 0186145 2.35 0.020 0069983 0803282 ageh -.004596 0168757 -0.27 0.786 -.0378359 0286438 genderhh 0470405 1475008 0.32 0.750 -.2434909 3375719 wifeworking -.2590881 173704 -1.49 0.137 -.6012319 0830556 soew_0 -.2628781 1896199 -1.39 0.167 -.6363711 110615 soeh_0 0964413 1813795 0.53 0.595 -.2608209 4537035 englishw_0 -.3926812 1897022 -2.07 0.040 -.7663365 -.0190259 englishh_0 0091718 19381 0.05 0.962 -.3725746 3909181 eduw 3274595 0780133 4.20 0.000 1737971 4811218 eduh -.0630879 0575058 -1.10 0.274 -.1763568 0501809 cpec 5003437 0665682 7.52 0.000 3692248 6314627
2 -.0009339 1591595 -0.01 0.995 -.3144295 3125616 incomet cpe Coef Std Err t P>|t| [95% Conf Interval] Robust
Root MSE = 94909 R-squared = 0.4592 Prob > F = 0.0000 F( 17, 245) = 15.75 Linear regression Number of obs = 263 h
Table 02 – Multicolinearity test of OLS – Model
Nrofchildren 1.15 0.869824 ownh 1.24 0.807576 local 1.18 0.846103 agew 3.40 0.293784 ageh 3.49 0.286877 genderhh 1.06 0.947707 wifeworking 1.22 0.819173 soew_0 1.64 0.610243 soeh_0 1.59 0.627896 englishw_0 2.37 0.421661 englishh_0 2.26 0.441967 eduw 3.38 0.295742 eduh 2.96 0.337550 cpec 1.19 0.841647
Table 03 – Logit regression result 01 – Model 03
_cons -3.657033 1.82147 -2.01 0.045 -7.227048 -.0870184 Nrofchildren -.3496952 3001152 -1.17 0.244 -.9379102 2385198 ownh 1.009829 4438265 2.28 0.023 1399455 1.879713 local 3828772 4142548 0.92 0.355 -.4290471 1.194802 agew 1760758 0688444 2.56 0.011 0411433 3110083 ageh -.0614185 0621596 -0.99 0.323 -.1832491 0604121 genderhh -.2093613 5090875 -0.41 0.681 -1.207154 7884318 wifeworking -.2818758 5803707 -0.49 0.627 -1.419381 8556299 soew_0 -.6166213 6564968 -0.94 0.348 -1.903331 6700889 soeh_0 -.3652005 6732687 -0.54 0.588 -1.684783 954382 englishw_0 -.4890562 6118155 -0.80 0.424 -1.688192 7100802 englishh_0 -1.120923 6422311 -1.75 0.081 -2.379673 1378264 eduw 3677952 2614649 1.41 0.160 -.1446667 880257 eduh -.1257198 1681974 -0.75 0.455 -.4553806 2039411 cpec 1.744994 4311084 4.05 0.000 9000376 2.589951
2 -.351869 6083459 -0.58 0.563 -1.544205 840467 incomet dcpe Coef Std Err z P>|z| [95% Conf Interval] Robust
Log pseudolikelihood = -80.705588 Pseudo R2 = 0.2910 Prob > chi2 = 0.0598 Wald chi2(17) = 26.89 Logistic regression Number of obs = 263 h
Table 04 – Marginal effect after Logit regression – Model 03
(*) dy/dx is for discrete change of dummy variable from 0 to 1
Nrofch~n -.0186697 01686 -1.11 0.268 -.051718 014379 1.61977 ownh* 0628274 03593 1.75 0.080 -.00759 133245 642586 local* 0223223 02697 0.83 0.408 -.030542 075187 749049 agew 0094004 00422 2.23 0.026 001126 017675 35.7529 ageh -.003279 00332 -0.99 0.323 -.009788 00323 38.4981 genderhh* -.0106239 02469 -0.43 0.667 -.059014 037766 779468 wifewo~g* -.0139541 02664 -0.52 0.600 -.066159 03825 809886 soew_0* -.0381279 04914 -0.78 0.438 -.134437 058181 247148 soeh_0* -.0209971 04035 -0.52 0.603 -.100073 058079 281369 engl~w_0* -.0278595 03859 -0.72 0.470 -.103501 047782 365019 engl~h_0* -.0690716 04716 -1.46 0.143 -.161507 023364 387833 eduw 019636 0147 1.34 0.182 -.00917 048442 3.85551 eduh -.006712 00924 -0.73 0.468 -.024831 011407 4.20532 cpec 0931626 02272 4.10 0.000 048635 13769 1.41065 dinco~t4* 0625143 03029 2.06 0.039 003152 121876 159696 dinco~t3* -.0221728 05822 -0.38 0.703 -.136289 091944 209125 dinco~t2* -.0194514 03526 -0.55 0.581 -.088556 049653 39924 variable dy/dx Std Err z P>|z| [ 95% C.I ] X
Table 05 – OLS regression result – Model
_cons -1.209754 5903896 -2.05 0.042 -2.372664 -.0468432 Nrofchildren -.0535274 0896321 -0.60 0.551 -.2300789 123024 ownh 2427965 1341195 1.81 0.071 -.0213831 5069762 local 0542165 1396164 0.39 0.698 -.2207907 3292237 agew 0433232 0181667 2.38 0.018 0075395 0791068 ageh -.0051432 0165262 -0.31 0.756 -.0376953 027409 genderhh -.0242883 1474851 -0.16 0.869 -.3147946 2662181 wifeworking -.2760428 1772245 -1.56 0.121 -.625128 0730423 soew_0 -.2352022 1909597 -1.23 0.219 -.6113421 1409377 soeh_0 0817378 185508 0.44 0.660 -.2836637 4471393 englishw_0 -.3558126 1872655 -1.90 0.059 -.7246759 0130506 englishh_0 0488498 1904441 0.26 0.798 -.3262744 4239741 eduw 3086596 0809731 3.81 0.000 1491642 468155 eduh -.0539666 0577517 -0.93 0.351 -.167722 0597888 cpec 4552241 073478 6.20 0.000 3104921 5999561 dtotalexpt5 6874581 3585757 1.92 0.056 -.0188406 1.393757 dtotalexpt4 2198905 3040206 0.72 0.470 -.3789491 8187302 dtotalexpt3 0035058 2393392 0.01 0.988 -.4679287 4749402 dtotalexpt2 -.0409011 251212 -0.16 0.871 -.535722 4539198 cpe Coef Std Err t P>|t| [95% Conf Interval] Robust
Root MSE = 95694 R-squared = 0.4525 Prob > F = 0.0000 F( 18, 244) = 18.02 Linear regression Number of obs = 263 h
Table 06 – Multicolinearity test of OLS – Model
Nrofchildren 1.19 0.840553 local 1.21 0.826600 wifeworking 1.24 0.808125 ownh 1.25 0.800912 cpec 1.29 0.775292 soeh_0 1.59 0.630849 soew_0 1.67 0.597484 englishh_0 2.26 0.443037 englishw_0 2.34 0.426472 eduh 2.85 0.351058 agew 3.42 0.292610 ageh 3.49 0.286735 eduw 3.52 0.283793 dtotalexpt2 7.02 0.142376 dtotalexpt5 12.43 0.080422 dtotalexpt4 19.33 0.051727 dtotalexpt3 19.45 0.051414
Table 07 – Logit regression result – Model 04 (a)
_cons -6.003235 2.556035 -2.35 0.019 -11.01297 -.9934975 Nrofchildren -.1674733 3214882 -0.52 0.602 -.7975786 462632 ownh 736127 4496512 1.64 0.102 -.1451731 1.617427 local 39491 4160752 0.95 0.343 -.4205823 1.210402 agew 1924707 0741705 2.59 0.009 0470992 3378422 ageh -.0646416 065009 -0.99 0.320 -.1920569 0627737 genderhh -.4665157 5253627 -0.89 0.375 -1.496208 5631762 wifeworking -.3144039 6618085 -0.48 0.635 -1.611525 9827169 soew_0 -.5504442 6030762 -0.91 0.361 -1.732452 6315634 soeh_0 -.2999803 6442077 -0.47 0.641 -1.562604 9626436 englishw_0 -.2666764 6403972 -0.42 0.677 -1.521832 988479 englishh_0 -.9710477 6419314 -1.51 0.130 -2.22921 2871147 eduw 1883084 2600612 0.72 0.469 -.3214021 698019 eduh -.0994752 1741595 -0.57 0.568 -.4408215 2418711 cpec 1.64758 4723387 3.49 0.000 7218131 2.573347 dtotalexpt5 3.888525 1.759674 2.21 0.027 4396263 7.337423 dtotalexpt4 2.419262 1.435419 1.69 0.092 -.3941084 5.232632 dtotalexpt3 2.263512 1.261546 1.79 0.073 -.2090731 4.736098 dtotalexpt2 2.143534 1.294358 1.66 0.098 -.3933617 4.680429 dcpe Coef Std Err z P>|z| [95% Conf Interval] Robust
Log pseudolikelihood = -81.443602 Pseudo R2 = 0.2845 Prob > chi2 = 0.0023 Wald chi2(18) = 39.67 Logistic regression Number of obs = 263 h
Table 08 – Marginal effects after Logit regression – Model 04
(*) dy/dx is for discrete change of dummy variable from 0 to 1
Nrofch~n -.0096229 01836 -0.52 0.600 -.045598 026352 1.61977 ownh* 0469965 03286 1.43 0.153 -.017401 111394 642586 local* 0248229 02937 0.85 0.398 -.032734 08238 749049 agew 0110593 0044 2.51 0.012 002439 019679 35.7529 ageh -.0037143 00365 -1.02 0.309 -.010873 003444 38.4981 genderhh* -.0240351 02436 -0.99 0.324 -.071773 023703 779468 wifewo~g* -.0166229 03164 -0.53 0.599 -.078632 045386 809886 soew_0* -.0359703 04483 -0.80 0.422 -.123833 051893 247148 soeh_0* -.0182967 04127 -0.44 0.658 -.099185 062592 281369 engl~w_0* -.0158438 04037 -0.39 0.695 -.094976 063289 365019 engl~h_0* -.0628338 04692 -1.34 0.181 -.154794 029127 387833 eduw 0108201 01541 0.70 0.482 -.019376 041016 3.85551 eduh -.0057158 01031 -0.55 0.579 -.025929 014498 4.20532 cpec 094669 02155 4.39 0.000 052431 136907 1.41065 dtota~t5* 1043288 04656 2.24 0.025 013071 195586 155894 dtota~t4* 1148973 07852 1.46 0.143 -.039007 268802 334601 dtota~t3* 1239615 08326 1.49 0.137 -.039218 287141 406844 dtota~t2* 0637695 03063 2.08 0.037 003736 123804 087452 variable dy/dx Std Err z P>|z| [ 95% C.I ] X
Table 09 – Statistic summary of variable cpe/cpbe, cpe/income, cpe/totalexp cpetotalexp 267 06704 0515155 0 4705882 cpeincome 267 0470075 0435199 0 3 cpecpbe 267 6139513 7110703 0 4
Variable Obs Mean Std Dev Min Max
Table 10 – The description of cpe and cpec Total 268 100.00 over 1,000,000 12 4.48 100.00
0 42 15.67 15.67 extra class Freq Percent Cum.
Table 11 – The relationship between total income and total expenditure to cpe
_cons 25 5538639 0.45 0.652 -.8406895 1.340689 dtotalexp6 1.727824 7812856 2.21 0.028 1892874 3.266361 dtotalexp5 1.950373 6677354 2.92 0.004 6354436 3.265303 dtotalexp4 1.484905 6184192 2.40 0.017 267091 2.70272 dtotalexp3 1.022503 5846873 1.75 0.082 -.1288853 2.173891 dtotalexp2 6666667 5982413 1.11 0.266 -.5114126 1.844746 dincome5 1.104559 4581295 2.41 0.017 2023931 2.006725 dincome4 2338345 363168 0.64 0.520 -.4813296 9489985 dincome3 1816083 3129473 0.58 0.562 -.4346592 7978759 dincome2 -.0075167 2249593 -0.03 0.973 -.450515 4354815 cpe Coef Std Err t P>|t| [95% Conf Interval]
Total 411.198502 266 1.54585903 Root MSE = 1.1077 Adj R-squared = 0.2062 Residual 315.354599 257 1.2270607 R-squared = 0.2331 Model 95.8439027 9 10.6493225 Prob > F = 0.0000 F( 9, 257) = 8.68 Source SS df MS Number of obs = 267 h
The correlation matrix reveals significant relationships among various explanatory variables, with notable negative correlations observed between dincome2 and dincome3 (-0.42), as well as dincome4 and dincome5 (-0.22) Positive correlations were found between dtotalexp3 and dtotalexp4 (0.41), indicating a strong relationship Additionally, the variables eduh and eduw exhibited a high positive correlation (0.77), suggesting a strong link between educational attainment and income levels Other variables, such as wifeworking and genderhh, showed minimal correlations, highlighting the complexity of interactions within the dataset Overall, the matrix provides valuable insights into the interdependencies of income, education, and demographic factors.
The analysis reveals various correlations among demographic and economic factors, including the influence of gender, age, and local ownership on income and expenditures Notably, the presence of a working wife positively correlates with certain income levels while showing a negative relationship with specific expenditures Additionally, age appears to have a complex interplay with other variables, influencing both household dynamics and financial outcomes Understanding these relationships is crucial for assessing the socio-economic landscape and developing targeted interventions.
Table 14 - The results of the models
Model 1 Model 2 Model 3 Model 4 Model 5
OLS 1 OLS 2 Logit Ordered Logit Tobit dincome2 -0.1141 -0.0149 -0.8824 -0.3152 -0.1141 dincome3 -0.1302 0.1100 -1.1158 -0.2819 -0.1302 dincome4 0.0631 0.2830 0.7016 0.2856 0.0631 dincome5 1.1466 *** 1.2753 *** (omitted) 2.5108 *** 1.1466 *** dtotalexp2 -0.0173 1.9596 0.7523 -0.0173 dtotalexp3 0.0754 2.6390 1.1614 0.0754 dtotalexp4 0.2648 2.9232 * 1.5204 0.2648 dtotalexp5 0.3637 3.5090 1.5906 0.3637 dtotalexp6 -0.0792 (omitted) 0.2832 -0.0792 cpec 0.4903 *** 0.5083 *** 1.5987 *** 1.0854 *** 0.4903 *** eduh -0.0455 -0.0457 -0.1096 -0.1371 -0.0455 eduw 0.2839 *** 0.2909 *** 0.2883 0.6514 *** 0.2839 *** englishh_0 0.1231 0.0964 -0.9516 0.1391 0.1231 englishw_0 -0.4324 ** -0.4309 ** -0.5446 -0.9070 ** -0.4324 ** soeh_0 0.0452 0.0410 -0.2945 0.1520 0.0452 soew_0 -0.1284 -0.1321 -0.6157 -0.4450 -0.1284 wifeworking -0.2180 -0.2419 -0.2230 -0.4132 -0.2180 genderhh -0.0295 0.0203 -0.4308 0.0124 -0.0295 ageh -0.0034 -0.0019 -0.0545 -0.0308 -0.0034 agew 0.0428 ** 0.0410 ** 0.1826 *** 0.1140 *** 0.0428 ** local 0.0098 -0.0118 0.4860 0.1003 0.0098 ownh 0.2786 ** 0.2777 ** 0.8505 * 0.5253 * 0.2786 **
( * ): Statistic significance at 10% level, ( ** ): Statistic significance at 5% level, ( *** ): Statistic significance at 1 % level h
Table 15 – Correlation between household’s total income and total expenditure and current payment for English education cpe 0.4299 0.4521 1.0000 totalexp 0.8064 1.0000 income 1.0000 income totalexp cpe
Table 16 – OLS regression result of all explanatory variables
The analysis reveals several key findings regarding various factors affecting the outcome The variable 'cons' shows a coefficient of -1.325324, indicating a potential negative impact, though it approaches significance with a p-value of 0.075 The number of children ('Nrofchildren') has a negligible effect with a coefficient of -0.0310962 and a p-value of 0.744 In contrast, 'ownh' demonstrates a significant positive influence with a coefficient of 0.2785548 and a p-value of 0.042 Age-related variables, 'agew' and 'ageh', show varying effects, with 'agew' being significant (p = 0.023) and 'ageh' not significant (p = 0.849) Gender of the household head ('genderhh') and whether the wife is working ('wifeworking') do not yield significant results Notably, 'englishw_0' has a significant negative coefficient of -0.432435 (p = 0.021), while 'eduw' shows a strong positive effect (p < 0.0001) The variable 'cpec' also indicates a substantial positive correlation (p < 0.0001) Income variables, particularly 'dincome5', reveal a significant positive relationship (p = 0.008), while other income levels do not show significant effects Overall, the results highlight the importance of specific factors such as education, household characteristics, and income in influencing the outcome.
Total 408.114068 262 1.55768728 Root MSE = 93625 Adj R-squared = 0.4373 Residual 209.49897 239 876564727 R-squared = 0.4867 Model 198.615099 23 8.63543908 Prob > F = 0.0000 F( 23, 239) = 9.85 Source SS df MS Number of obs = 263 h
Table 17 - The results of the models (OLS, Logit, Tobit)
Current payment for English education
Coefficient Coefficient Marginal effect (dy/dx)
Over 30 millions VND 1.275 *** (omitted) (omitted) 6.654 *** 6.654 ***
Current payment for extra class
Dummy English var of husband 0.096 -1.078 -0.080 -0.729 * -0.729 *
Dummy English var of wife -0.431 ** -0.486 -0.034 -0.242 -0.242
Dummy Public sector of husband
Dummy Public sector of wife
( * ): Statistic significance at 10% level, ( ** ): Statistic significance at 5% level, ( *** ): Statistic significance at 1 % level
Table 18 - The results of the models (OLS, Logit, Tobit)
Current payment for English education
Coefficient Coefficient Marginal effect (dy/dx)
Coefficient Marginal effect (dy/dx)
Current payment for extra class
(0.00) *** (0.46) (0.47) (0.69) (0.69) Dummy English var of husband
Dummy English var of wife -0.364 -0.307 -0.021 -0.165 -0.165
(0.06) * (0.66) (0.68) (0.71) (0.71) Dummy Public sector of husband
-0.023 -0.300 -0.020 -0.055 -0.055 (0.54) (0.62) (0.63) (0.47) (0.47) Dummy Public sector of wife
( * ): Statistic significance at 10% level, ( ** ): Statistic significance at 5% level, ( *** ): Statistic significance at 1 % level h
M Please tell us the full name of your family's members, start from the household's lead
Gender Relationship to the household's head Year of
How long has [name] been in HCMc?
(Write the name of district)
B The family's member is the one who live and eat together with your family at least 6/12 months and shares the expenses
E Mark * on the right of household's head name
Grandchildrens…… 6 Divorced…… 4 Nr Of Nr Of
Others……… 7 year Legal separation 5 Years Months
The central areas of the city encompass Districts 1, 3, 4, 5, 6, 7, 8, 10, 11, Binh Thanh, Phu Nhuan, Tan Binh, Tan Phu, and parts of District 2 In contrast, the suburban regions include Districts 2, 9, 12, Binh Tan, Binh Chanh, Hoc Mon, and Cu Chi.
Please let us know the education information of your household's members
The highest education level of [name] attained
What is the type of the school?
Certificate / Grade Self-assessment the
E No certificate 0 Public school… 1 [name] Does the [name] take part in any English class within last 12 months?
M Primary 1 Charter school…2 attain (if any)
B Junior high school 2 Private school… 3 Don't know………… 0
E Senior high school 3 International school………… 4 Pre-intermediate (basic communication)…… 1
O Vocational colleges 7 Advance (fluent in 4 skills (Listen, speak, read, write………… 3
PhD……… 11 At school At English center Other (specific )… 12
[name]… Expense for education of …[name] per month (1000 VND)
Expenditure for food Expenditure for Health's care a
Extra English class at school c
Extra learning English at English center d
Table 20 – Price of English class of beginner and pre-intermediate of English centers and school in Ho Chi Minh City
Class name Price Divided Rank of CPE
ILA Jumpstart (below 6 years old) 2,752,000
VND/month Beginnner and Pre-intermediate level (from 6-11 years old)
For preschool children (below 6 years old)
Superkids For primary school children (from 6-11 years old)
Apollo Kindy (below 6 years old) 1,204,000
VND/month HCM University of
VND/month English at primary school