Determinants of computer ownership case of households in vietnam

AN OVERVIEW OF THE COMPUTER MARKET AND COMPUTER OWNERSHIP IN VIETNAM 3.1 An overview of computer ownership, computer market and empirical projects to enhance the computer ownership in Vi

VIETNAM- THE NETHERLANDS PROJECT FOR

M.A IN DEVELOPMENT ECONOMICS

DETERMINANTS OF COMPUTER OWNERSHIP:

CASE OF HOUSEHOLDS IN VIETNAM

By:

VO THI THANH NGA

B¢ GIAO ~UC VA DAO T~O

I would like to thank other professors, lecturers, classmates in the Project and other teachers and friends of mine for their help and friendship

As for my parents, I just cannot thank them enough for their unconditional love and support

ABSTRACT

Low computer ownership rate has been proved to lead to low quality of life of households, slow down the economic development and make government administration less efficient This study investigates the determinants of computer ownership for the case of households in Vietnam, a developing country with a very low computer ownership rate The author reviews the literature and employs the Logit model and descriptive statistics to model the computer ownership and use; the data from the Vietnam Living Standard Survey 2004 Results support the empirical evidences that household socio-demographic characteristics greatly influence that household ownership of computer, which identifies the digital divide between types of households and regions Based on the results and empirical projects that have been carried out to enhance the computer availability, policy recommendation is given

TABLE OF CONTENTS

Page Chapter 1 INTRODUCTION

1.1 Problem Statement !

1.2 Objective of Study 3

1.3 Methodology and Scope of Research 4

Chapter 2 LITERATURE REVIEW AND RESEARCH METHODOLOGY 2.1 Basic Terms and Concepts 5

2.2 Economic Theories of Consumer Behavior 7

2.2.1 Neoclassical quantitative models of consumer behavior 8

2.2.1.1 The Indifference Curve Model 8

2.2.1.2The Marginal Utility Model 9

2.2.2 Demand for durable goods and qualitative choice models 10

2.2.3 Justification ofthe models 11

2.3 Empirical Studies 13

2.3.1 Methods of empirical researches applied 13

2.3 1.1 Methods of Internet access estimation 13

2.3 1.2Methods of computer use, computer expenditure and computer ownership estimation 14

2.3.1.3Justification ofthe methods in empirical studies 16

2.3 2 Major findings in empirical studies 1 7 2.3 2.1 Determinants of internet access 17

2.3 2.2Determinants of computer use, computer expenditure and computer ownership 1 7 2.3.2.3Justification of the determinants in empirical findings and some suggestions in the thesis research model 19

2.4 Econometric Model Specification 21

Chapter 3 AN OVERVIEW OF THE COMPUTER MARKET AND COMPUTER OWNERSHIP IN VIETNAM

3.1 An overview of computer ownership, computer market and empirical

projects to enhance the computer ownership in Vietnam 28

3 1.1 An overview of computer ownership of households in Vietnam 28

3.1.2 Computer market in Vietnam and prospect after Vietnam joined WT0 32

3 1.3 Some projects that have been carried out to enhance the computer ownership of households 3 5 3.2 Descriptive analysis on factors affecting computer ownership of households in Vietnam in VLSS 2004 38

Chapter 4 ECONOMETRIC ANALYSIS 4.1 Logit model 45

4.2 The logit model in this study's estimation 46

4.3 Logistic regression results 47

4.4 Goodness-of-fit 49

4.5 Results interpretation 49

4.5.1 Coefficients and expected signs 50

4.5.2 Oddsratios 51

4.5.3 Probability estimated at different initial probability 54

Chapter 5 CONCLUSIONS AND POLICY IMPLICATION 56

REFERENCES 60

APPENDIX 1 Unrestricted model 63

APPENDIX 2 Restricted model 64

APPENDIX 3 Calculating P(computer)new from initial probability 65

LIST OF TABLES Page

Table 1 Number of computers of Vietnam in comparison with other countries in 2005 29

Table 2 The probability of computer ownership in Vietnam 30

Table 3 Hardware market value in Vietnam 2000-2005 (million USD) 33

Table 4 Mean of characteristics by groups of ownership and non-ownership 38

Table 5 Percentage of households owning a computer by income quintile 39

Table 6 Computer ownership by groups of households with head's education highest qualification 40

Table 7 Group of households by binary characteristics 41

Table 8 Groups of households by region and their computer ownership 43

Table 9 Unrestricted logit model 47

Table 10 Restricted logit model 48

Table 11 Coefficients and expected signs 50

Table 12 Interpreting in terms of odds ratios 52

Table 13 Interpreting probability changes at different initial probability 54

LIST OF FIGURES AND CHARTS Figure 1 Components oflnformation Society Index 28

Chart 1 Households with a home computer, 2000-2005, percentage of all households 30

Chart 2 Percent of households owning computer(s) by different education attainment of household head 40

Chart 3 Percentage of households owning a computer by geographical regions 43

LIST OF BOXES

Box 1 Success of Thanh Giong Project

Besides undeniable positive impacts of computer use in work places, many researchers believe that computers influence households profoundly A computer

in the home often serves three functions: (1) interpersonal communication; (2) information acquisition; and (3) entertainment (Papadakis 2001 ) More and more households own computers nowadays due to: (i) the availability of affordable computers for households; (ii) the user-friendliness of horne computers and; (iii) the expansion of home-based business (Wen Yin 2003) Significantly, in Vietnam, more than 50% of households are household enterprises (World Bank, 2004)

Computer ownership, therefore, plays an important role in improving the quality of life for every household

For the whole economy, the use of information and communication

technologies (ICT) by households, including the use of computer, undoubtedly has impacts on economic performance, according to a research by OECD (liS report 2005:47), because: (i) the demand for ICT goods (e.g computers) by households has stimulated the growth of ICT sector and ICT applications; (ii) this has stimulated demand for products in sectors which rely heavily on ICT (e.g media and entertainment), which leads to the growth in those industries as well; (iii) in the future, the wide diffusion of ICT and e-commerce can help firms to deliver their products to households more efficiently; (iv) ICT use by household can benefit companies that require experienced ICT users

For the government, computers to households and access to the Internet is a

way of education, access to information and policy propagation

In other words, computer use by households plays a big role in improving the quality of life for households, making profound impacts on the whole economy and helping the government administration more efficient

According to the Vietnam Development Report 2004 by the World Bank, the poverty rate has been reduced dramatically in the past 5 years: from 17.2% in

2001 to 8.3% in 2004 and 7% in 2005 On average, 340,000 poor families escape poverty each year However, to have a complete computer set, one must have at least 500 USD1 while the GDP per capita for Vietnamese in 2005 was only around

640 USD and around 715 USD in 2006 Low income makes computer unaffordable for many households Only 5% of households in Vietnam own a computer (VLSS 2004 ), a very low rate compared to that in 2004 of a developed Asian country like Korea (78%)2

Low computer ownership rate, as proved above, leads to low quality of life

of households, slows down the economic development and makes government administration less efficient Severely, the absence of computer in a household can cause social and economic disadvantages to that household, leading to social exclusion (Papadakis 2001) This creates the gap in ICT adoption between members in the society, called the "digital divide"

The extent to which the digital divide exists and why it exists, therefore, must be addressed In other words, the issue to be explored is the computer adoption of households in Vietnam "Given the scarcity of computers in low-income countries like Vietnam, an operational strategy of increasing computer availability and skills would seem to offer considerable hope for increasing the incomes for the poor" (World Bank 2003)

1

Hoan Long computer price list, Dec 2006

2 Korean National Statistical Office

Unfortunately, there has been no research on computer ownership of households in Vietnam so far Therefore, studying the determinants of computer ownership for the case of households in Vietnam is chosen for the thesis research

1.2 Objective of the Study

My study aims to answer the following questions:

What are the main determinants of computer ownership of households in Vietnam?

How is the current situation of the computer ownership of households and computer market in Vietnam today? How is the current digital divide in Vietnamese households?

What can be done to improve the probability of computer ownership for households in Vietnam?

The author aims to develop a conceptual model for estimating the demand for computer of households in Vietnam, and correspondingly suggest policy implications

For policy makers, the research aims to give an overview of the demand for

computers of households in Vietnam Based on the most significant determinants

of computer ownership and from that, policy makers can have positive policies to enhance the computer availability, particularly in rural and remote areas to minimize the risk of social exclusion

Besides, the government can estimate the demand for computer ownership and have a master plan for the development the computer industry together with the relevant plan for the development of the telecommunication system (internet broadband, telephone lines ) and supporting industries (other electronic production)

For producers, the estimation of computer ownership while other

determinants change can help computer producers to forecast the production and the targeted market This comes from the answer to the question: "Which types of

households tend to own computer most?" Facing the harsh competition after joining the WTO when the tariff on computer sets and computer components will gradually reduce to 0%, producers must have relevant strategies to provide products of higher quality at lower price to consumers

1.3 Methodology and Scope of Research

To answer the research questions, both econometric model and descriptive statistic methods are used to examine the relationship between the dependent variable computer ownership and its determinants Mainly, the Logit model is used to estimate the determinants of the computer ownership of households in Vietnam Besides, descriptive analysis about the situation of Vietnam computer purchases, computer use and computer manufacturing This research is mainly based on secondary data, taken from Vietnam Household Living Standards Survey

2004 (VLSS 2004 ) This survey is carried out by the General Statistics Office of Vietnam, with the support by the World Bank and the Swedish International Development Agency (SIDA) The data is collected throughout the country with the sample size of 45,900 households (36,720 households were surveyed on income, 9,180 households were surveyed on income and expenditure) of 3,063 communes, representing for the whole country with 8 regions, urban and rural areas, provinces and cities As the thesis mainly depends on the expenditure of the household rather than income, so the sample size will be around 9, 180 The reason for expenditure base will be explained later on

Particularly, the survey is well-designed, therefore, to take advantage of the available data, the author hope that the result should give an overview of the whole country

Descriptive analysis is also based on secondary data from VLSS 2004 and other domestic and international data sources

Chapter 2

LITERATURE REVIEW AND RESEARCH METHODOLOGY

The literature review and research methodology is organized into 4 sections

Section 1 defines some terms and concepts related to the objects of the research like computer, durable goods and household

Section 2 provides a review of economic theories on consumer behavior

On the one hand, two main models are focused in the Neoclassical models of consumer behavior: the Marginal Utility Model and the Indifference Curve Model The comments will point out the role of income, price and household's head in the purchase of a goods and the disadvantages of these quantitative models On the other hand, the characteristics of demand for durable goods and a qualitative choice model will be introduced as an alternative

Section 3 introduces different empirical studies on computer and other ICT goods ownership, which bring about a conclusion of the most significant variables that can influence the computer ownership

A justification of methodology and chosen variables are provided at the end

of each section Then, the econometric model specification and hypotheses are proposed in Section 4

2.1 BASIC TERMS AND CONCEPTS

First is the definition of computer In the VLSS 2004, the term "computer"

is general So, computer in this thesis is assumed to include both desktop personal computer and laptop It is appropriate to the definition of computer access in the Current Population Survey by the U.S Census Bureau in year 2000, where

computer access means the answer Yes to the question "Is there a personal computer or a laptop in this householcJ?"

Second is the definition of a durable good and whether a computer is a durable According to Koutsoyiannis (1979:57), durables include furniture, electrical appliances etc McEachern (2006: 49) defines durables as goods expected to last 3 or more years Importantly, in Consumer Expenditure Survey during 1986-1994 in the United States, durable holdings include cars, housing and personal computers (Wolak, 1994: 6) That is to say, computer is considered to be

an important durable nowadays

Third is the definition of household and household as the unit of analysis A household is defined as "a unit consisting of people with a common budget who organize some aspects of work and consumption together" (Himmelweit et al., 1998:149) Another definition ofhousehold by Messer is " that group of people, their relationships and activities, who acknowledge a common authority in domestic matters, a "budget unit" or a group who have a common fund of material and human resources and rules for practices and exchange within it" (Messer, 1990: 52, cited in Himmelweit et al, 1998: 149) Both definitions implies that household is a bounded unit that people make decisions together, in particular about the use of their budget of human and material resources (Himmelweit et al., 1998: 150)

New Household Economics (NHE) developed by Jacob Mincer and Gary Becker takes households rather than individual people as the rational decision-making units The reason for that is "individuals spend only their own income and spend it only on themselves But if income is shared in households, then the person who does the family shopping is neither spending just her own income nor buying just for herself' (Himmelweit et al., 1998: 151 ) Moreover, standard microeconomic theory considers household as the basic decision unit in consumer theory, assuming either a household utility function or a representative agent

preference structure (Chiuri, 1999:7) On the other hand, many marketers also recognize family as the major consumption unit within any society (Schiffman et

al, 2001 :31 0) Simultaneously, the data in VLSS is on households

Household, therefore, is chosen as the unit of analysis m this thesis research

2.2 ECONOMIC THEORIES OF CONSUMER BEHAVIOR

Demand by an individual:

According to the Neoclassical theory on consumer choice and demand decision, the demand by a single consumer depends on four elements: (a) the consumer's income, (b) the prices at which goods can be bought, (c) the consumer's tastes, ranking different bundles of goods by the satisfaction they yield, (d) the behavioral assumption that consumers pursue their own self interest (Begg et al., 2003:55) According to Zamagni (1993), determinants of demand are classified into two sets of variables: observed (quantity, income, price) and unobserved variables (taste)

Neoclassical economists classify goods as normal, inferior and neutral goods Demand for goods increases as income increases for normal goods and decreases as income increases for inferior goods and unchanged as income varies for neutral goods (Begg et al 2003) Computers, as stated above, are durable goods and manufactured goods, therefore they can be considered as normal goods The demand for computer, therefore, increases as income increases

Case of a household:

As stated above, household is a unit of analysis in this research, and a household is not an individual However, Paul A Samuelson looked at household decision making as only a special case when a collective of people with different individual preferences make decisions as if they had a single collective preference because "blood is thicker than water" A household utility function accordingly is

the result of "a consistent family consensus that represents a meeting of minds or a compromise between them" (Samuelson, 1956:9) He assumed the existence of a household welfare function to be maximized, subject to a given level of household income, besides the utility maximizing household members

In a different way, Gary Becker, another Nobel laureate and one of the founders of the New Household Economics, argued that a household consensus does not resolve the problem of how different preferences among household members His analysis got over this problem of reconciling different preferences

by showing how all members' actions could be determined by the preferences of just one member of the household, its 'head' (Himmelweit et al., 1998: 186) In this case, the head's preferences effectively became the preferences of the

household as a whole This theory recognizes the importance of the head of

household in the household's decision making The characteristics of the

household head, therefore, would be taken into account when considering determinants of computer ownership in each household

To conclude, the demand for a good by an individual depends on consumer's income, taste, and price of that good For a household, whether to look

at the household as a family consensus or as the dictatorship of a household head, the household is a unit of analysis and household head characteristics should be focused to study the demand of a household for some goods

2.2.1 Neoclassical quantitative models of consumer behavior

2.2.1.1 The Indifference Curve Model:

According to Neoclassical theory of demand and choice, the budget constraint describes different bundles that the consumer can afford The budget line shows the maximum combinations of two goods that the consumer can afford, given income and the prevailing price Points on the budget line means the consumer uses up his entire budget Points above the budget line are unaffordable and points inside the budget line would allow additional spending The slope of

the budget line is the trade-off of one good to get more of the other good (Begg et

al, 2003 :56)

Meanwhile, an indifference curve shows all the consumption bundles yielding a particular level of utility Every point on the indifference curve yields the same utility As the consumer prefers more to less, the indifference curve slope downwards and the slope of each indifference curve gets steadily flatter as the point moves to the right The slope of the indifference curve is the marginal rate of substitution of one good for the other good (Begg et al., 2003 :58)

To complete the model, it is assumed that the consumer chooses the affordable bundle that maximizes his utility Hence, the chosen bundle will be the point at which an indifference curve just touches the budget line or tangent to the budget line, where the slope of the budget line and the slope of the indifference curve coincide (Begg et al., 2003:61)

With different prices, we can find out the tangencies and quantities at which utility is maximized, and therefore, we can derive a demand curve, with a negative slope

2.2.1.2 The Marginal Utility Model

Marginal Utility, MU = dTU/dQ, is the ratio of the change in utility to the infinitely small change in consumption which brought it about

where TU is the total utility, Pi is the price of ith good and Qi is purchased quantity in the budget period, the condition to maximize the consumer's utility is:

~TU1/PI = L1TUz/Pz = = L1TUn1Pn

because at this equilibrium, no change m the combination of goods purchased can increase the total utility of the consumer

As L1 TU = MU * ~Q, setting all L1Qs equal to one, we derive the utility maximization condition,

MU1/P1 = MUz/Pz = MUn!Pn

By this we can see that the utility maxnntzmg consumer adjusts his purchases so that all items yield the same marginal utility per dollar Under this condition, if the price of one good increases, the quantity purchased of it should decrease and the consumption of all other products should increase

With the marginal utility of money constant, the marginal utility function could serve as the individual's demand function

2.2.2 Demand for durable goods and qualitative choice models

As defined earlier in the section Basic Terms and Concepts, durable goods are goods that last for 3 years or more, so they are often goods of high value like cars, house, electric appliances Buying a durable good, therefore, is considered

"adding to stock", in contrast to consumption, which refers to depletion or physical deterioration (Deaton and Muellbauer 1980:345) Durable goods are not purchased as frequently as consumable goods (such as clothes or rice) In expenditure studies, especially studies on durable goods, it is common that a large number of households do not have any purchases for a particular category during the survey period Consequently, the dependent variable will be zero for a substantial number of observations In econometrics, this is referred to as the

"limited dependent variable problem" (Chern and Senauer 1993) Thus, the Ordinary Least Square method is not relevant

Therefore, for many durable goods, the choice is between ownership and

non-ownership (discrete choice)

With x interpreted as income, E as threshold income level above which ownership occurs, hence S = 0 if x < E and S = 1 if x > E, Muellbauer (1980:369) showed that if we assume E to be distributed independently of x and that it has the lognormal distribution with parameters Jl and CJ2 g(x) is the probability of a randomly selected household owmng the durable conditional on household expenditure x, then:

g(x) ~ prob(S = 1; x) = f f(E; ,u, cr2

dE = A(x; ,u, cr2 -oo

where A(x; ,u, cr2

is the cumulative distribution function of the lognormal distribution

2.2.3 Justification of the models

Well-known Neoclassical quantitative models of consumer behavior like the Indifference Curve Model or the Marginal Utility Model indicate that income (budget) and price of a good are major determinants affecting the demand for that good They based on the assumption of goods substitutability However, these models themselves have problems, especially with durable goods

For example, the Indifference Curve Model has the problem of binary demand In a household, we either buy a house, car, washing machine, dishwasher (and a computer is a similar good) or we do not "What does such an indifference curve look like if, for example, one axis has one house or none and the other axis has all other things?" (Bumas 2002:1 09) Another problem is that the indifference curve analysis makes sense when a budget is spent on two goods, but infact we buy many more than two goods

The Marginal Utility Model has two main problems, they are the problem

of durable goods and the problem of lumpiness

Durable goods yield utilities beyond the current budget period, not just in the current budget period as presented in the model Alfred Marshall handled this problem by calculating the present value of the discounted flow of expected marginal utilities over the expected life of the good But not many consumers or even economists know how or bother to do such a calculation (Bumas, 2002)

Another problem is lumpiness problem The marginal utility per dollar of a car, camera, stereo or piano (and computer as well) is higher than that of other goods So we should buy more than one camera, stereo or piano But the purchase

of two yields a lower marginal utility per dollar than that of other goods So we should buy an additional fraction of such goods, but we cannot (Bumas 2002:97)

Furthermore, these two models all have problems with their assumptions of goods substitutability In fact, it is clear that not any good can be substituted by another good, so it is unrealistic

Durable goods creates discrete choice as presented in Section 2.3 as people cannot purchase durable goods constantly or even do not purchase that goods for a period of time Discrete choice causes difficult aggregation problems It would not make a great deal of sense to model the decision of a representative household to buy some fraction of a durable good This is the major disadvantage of quantitative models in estimating the demand for durables

Due to all of the above problems, quantitative models are not suitable to estimate computer demand (durable goods)

Qualitative Choice Models can be a good alternative because they can overcome the Limited dependent variable problem Importantly, Muellbauer (1980:369) remarked that, using cross-section microeconomic survey data on individual households, such models can be estimated directly by the principle of maximum likelihood

Of all qualitative choice models, the Logit model is most widely used The reason is the formula for logit choice probability is easy to interpret in terms of utility maximization and discrete choice and its parameters are easy to estimate (Train 1986: 15) Besides, according to Cramer, the Logit model is suitable to apply in examining the model of categorical phenomena and determination of categorical variables (Cramer 1991:1)

For all of the above reasons, the Logit model is employed to estimate the demand for computer (computer ownership) in my study

2.3 EMPIRICAL STUDIES

Unlike other traditional durables, computer is also an essential good of the information and communication products Therefore, the determinants of demand for computer partly includes some determinants of demand for durables, partly includes some determinants of demand for communication products for like postal service or internet

There have been limited amount of research on computer ownership due to the lack of information on home computers and the factors related to their purchase (Papadakis 2001 ) Prior researches had to base on theories of other similar goods as mentioned above as an alternative

In this section, besides some latest empirical studies on computer ownership and computer expenditure, I also would like to provide an overview of empirical studies on demand for some durables and ICT goods as a comparison, including:

1 Internet/ wireless internet access

2 Computer use, computer ownership and computer expenditure

2.3.1 Methods of empirical researches applied

2.3.1.1 Methods of Internet access estimation

The Logit model with Willingness To Pay method (WTP) was used by

Rappoport et al (2003) to estimate the demand for wireless Internet access with an omnibus survey of some 12,000 households in the U.S For a particular price of wireless internet access, rr*, say, access will be demanded for WTP's that are equal to or greater than rr*, while access will not be demanded for WTP's that are less than rr*

The Logit model was also used by Kridel et al ( 1999) and Chaudhuri et al

(2006) to estimate the determinants of internet access Kridel et al estimated

determinants of high-speed internet choice, using survey data from individuals that have residential internet access in the U S while Chaudhuri et al estimated the determinants of internet choice of households, not individuals like Kridel, and across all households, not simply focusing on those that already use the internet

Kridel et al used the model as follows:

Prob(High-speed AccesslxO = G(xi,(J)

where fJ is a vector of parameters to be estimated from the data that relate to geographic specific price effects, age, gender, ethnicity, employment status, income and educational attainment

With Chaudhuri et al, the model was:

Prob(Home AccesslxO= G(xi,(J)

where p is a vector of parameters that relate to geographic specific price effects, age, gender, ethnicity, employment status, income and educational attainment

2.3.2.2 Methods of computer use, computer expenditure and computer ownership estimation

(1) Chris N Sakellariou and Harry A Patrinos (2003) employed the

Ordinary Least Square (OLS) method to estimate the relation between computer use at work and wages, and employed the Probit model to estimate the determinants of computer use for the case of graduates in Vietnam The study is

on individuals, not households

(2) Descriptive statistics were used in the research by Microsoft and Forrester (2004) to find important determinants of computer use among working-age adults in the U.S Descriptive statistics were also used by John Stahura (2005) and Rappoport et al (2003 )

(3) John Stahura (2005) conducted a double-hurdle model to estimate

whether or not households spent on computer hardware and software and the amount spent on computer hardware of households using the 2000 Consumer Expenditure Survey in the United States This methodology is similar to that of Wen Yin (2003) They both based on neoclassical consumer theory to develop a conceptual model of computer expenditure including economic resources, consumer tastes and preferences

A double-hurdle model consists of two separate parts: a Probit model to estimate whether or not households spend on computer hardware and a Truncated

Regression model to estimate the amount spent on computer hardware and

software

Following this, the variables in Stahura's model are age, life cycle stage, gender, race, education, household income after tax, computer ownership, and region of residence and quarter of purchase (i.e seasonality) The variable of seasonality is only for the truncated regression model, it is excluded for the Probit model

(4) The heterogeneous probit model to assess residential computer choice

using a selection of choice factors was applied by Robertson et al (2007) Their research was on computer adoption (to have or not to have a computer) in 5,000 households in the United Kingdom, surveyed in year 2003 and 2005 The socio-demographics drivers of computer choice they applied were the presence of children, educational attainment and household disposable income

A contribution of this study by Robertson is the inclusion of the Technology Acceptance Model (TAM) framework TAM was applied to study how people perceive technology and how this information can lead to predictions

on how technology may be adopted/used That is the perceptions toward the

technology on its usefulness and ease of use

They estimated the probability of household computer adoption given the factor effects, Xn, by the following equation:

In this heterogeneous probit model <D is the standard notation for the cumulative standard normal distribution eaTAM defines the control for heteroscedasticity with a being the coefficient of dispersion of TAM In the data

collected by their survey, the variable TAM consists of three perceptions: computer ease of use, computer usefulness and computer enjoyment

2.3.1.3 Justification of the methods in empirical studies

WTP method is only often used to investigate people's response to newly introduced goods or services (like wireless internet demand), i.e when the purchase decision has not yet been made However, the WTP is not always the same as the actual expenditure when consumers make their purchase decision in reality In fact, the VLSS 2004 already has the data of computer ownership (the purchase decision has been made already), so the WTP is not appropriate

Ordinary Least Square or Truncated Regression Model (one part of the Double Hurdle Model) are not suitable as they are quantitative models and they have been proved to be irrelevant to estimate the computer ownership (discrete choice) in the theory review In fact, in these empirical studies, these models were not used to estimate the computer ownership but to estimate the wages and computer expenditure

Probit and Logit models are most preferred in estimating the choice of internet access or computer access (including internet access, computer use and ownership) Probit and Logit models are similar, but the Logit model is generally used in preference to Probit model due to convenience and ready availability of computer programs (Gujarati 1988)

The Logit model is decided to be employed for my study because the survey data in VLSS is across all households including the penetrating households and non-penetrating households, which can address the digital divide (similar to the study of Chaudhuri et al2006)

Descriptive analysis from VLSS 2004 and other data sources shall also be used in comparison with some other countries and between groups of households

in Vietnam to illustrate the digital divide because descriptive analysis is especially useful for comparison as shown in many studies above by Microsoft and Stahura 2.3.2 Major findings in empirical studies

2.3.2.1 Determinants of internet access

Demand for access to the Internet (home, high speed or wireless internet) is

elastic over the range of prices currently charged by service providers and

geographic specific price (Kridel et al 1999, Chaudhuri et al 2006, Rappoport et al 2003)

Household size was considered as the impact on internet adoption by Kridel

et al (1999) and he found it positively correlated with the ICT adoption

Age, gender, ethnicity, employment status, income and educational attainment are found to be determinants of home internet access by Chaudhuri et

al (2006) and high speed internet access by Kridel et al (1999, 2002) The penetration rates for Internet access display a strong positive relationship with income, education; a declining function of age Non demographic variables include indicators of current Internet use, presence of a mobile phone and presence

of a personal computer (complement goods) (Rappoport et al2003)

2.3.2.2 Determinants of computer use, computer expenditure and computer ownership

Earnings, speaking English, working in an economics/finance/accounting,

'

technology and administration related occupation are most significant

e¢ a1Ao nuc \~:~ oAo TAo

TRUClNO Df.-J KIN~1 TE TP.HCM

THt! '\~!¥N

17

determinants of computer use at work by individual graduates (note: not households) (Sakellariou and Patrinos 2003)

Household income, educational attainment have significant impact on

computer use, computer ownership and purchase of computer hardware and software (Microsoft and Forrester 2004, Robertson et al 2007, Stahura 2005) Adults with lower levels of education and household income are less likely to use computers at both home and work than are those with higher levels of education and household income

Employment status, age, gender, household size, region of residence have

significant impact on computer use, computer ownership and purchase of computer hardware and software Computer use rates are "highest for individuals

age 18-25 and lowest for individuals over age 60" (Microsoft and Forrester 2004)

For gender variable, Stahura (2005) said that female-headed households were

found to be less confident about information technology than male-headed households in regard to products that were related to technology Stahura's finding about gender factor was consistent with this for computer purchase and expenditure

Life cycle stage factor was also a significant factor affecting the purchase of

both hardware and software Household life cycle stage "has been utilized by consumer researchers and marketers as a way to classify family units into meaningful stages to examine their purchase and consumption behavior" (Schiffman et al, 2001 :317) The life cycle stages have frequently been identified

by age of the household head, marital status, and presence of children (Schiffman

et al, 2001 and Xiao 1996) Life cycle stage factor was significant for hardware expenditure; and gender, income and season were important predictors of software

expenditure (Stahura 2005) Presence of children, especially presence of children

at school-age is found to be a driver of computer choice in Robertson et al (2007) , Stahura (2005), Venkatesh et al (2003 )

TAM (Technology Acceptance Model) variable, which consists of three

perceptions: computer ease of use, computer usefulness and computer enjoyment, was found to possitively affect the home computer adoption (Robertson et al 2007)

2.3.2.3 Justification of the determinants in empirical findings and some suggestions in the thesis research model

It is generally agreed that household socio-demographics are maJor determinants of the purchase or use ofiCT goods and amount spent on them

In some empirical studies on internet access, the geographic specific price

effect is an important determinant However, the price shall not be taken into

account in the model of computer ownership estimation for the following reasons: (1) the survey in VLSS 2004 is cross-section data, not time-series data, so there are hardly any price differences; (2) the computer purchase price in the VLSS cannot be comparable, e.g the price of a computer purchased 3 years ago cannot

be compared to the price of a computer bought last month; (3) price of a computer set - an electronic good - is actually nearly the same between regions as given in some price list of major distributors, so it does not have the geographic specific price differences or price differences by providers like the internet service

Besides some households characteristics such as household mcome, household size, household residential place affecting the computer use and computer ownership, in this study, I shall assign some other characteristics like age, education, gender, employment status to the characteristics of the household head as the household head is considered to be the representative of the household decision maker, reconciling all the preferences of the household as proved in Becker's theory

Household life cycle stage factor, identified by (i) age of the household head, (ii) marital status, and (iii) presence of children used in the model by

Stahura, does have some pros and cons in the case of Vietnam The family life cycle is generally started with singles, on to marriage, to family growth (with the birth of children), then to family contraction (as grown and children leave the household) and ends with the dissolution of the unit (due to the death of one spouse) (Schiffman et al, 2001) In my study, besides the age of household head, the presence of school-age children shall be added to the model due to its positive effect on computer ownership of the household However, the life cycle stage concept is derived from Western societies, based on nuclear families In Vietnam, despite the fast growth of nuclear families, there are still a large number of traditional families with more than 2 generations together in a household So the stages mentioned may not be the same in Vietnam (e.g more than 2 couples, the cycle will not end with the death of the spouses but still continue throughout next generations) Therefore, marital status is difficult to identify in these households,

so I expect the marital status variable to be irrelevant in my study

TAM variable, despite its usefulness, is unfortunately not available in the VLSS 2004 data As VLSS is a general household living standard survey, it does not go into details in the perceptions of the interviewee toward computer This is

the limitation ofthe thesis research In fact, Gujarati D N (1988) also pointed out

that it is common experience ·in empirical analysis that the data we would like to have often are not available, and at that time the disturbance term may be used as a substitute for the omitted variables

Conclusions of the literature review and research methodology:

As we can see in these empirical studies, all of them are in common that determinants of the ownership and use of a computer or communication service (internet) are household mcome, household social and demographic characteristics However, after justifying the theories and empirical methods, I suggest that quantitative models are not suitable for the estimation of the computer ownership in Vietnam

I choose to employ a qualitative choice model, Logit model, for its relevant

to discrete choice estimation, especially for durable good ownership, proved by both theories and empirical evidences Moreover, it can address the digital divide between households in the society

Descriptive analysis is also used in the thesis research for its intuitiveness and comparability Descriptive analysis shall be carried out in Chapter 3, which provides an overview of computer ownership and computer market in Vietnam

The empirical studies also provided an overview of determinants of ICT goods use and ownership The most significant variables are justified and most suitable variables to the research are chosen They will be presented in details in the next section: Econometric Model Specification

2.4 ECONOMETRIC MODEL SPECIFICATION

As discussed above, the estimation for household ownership of computers

in this thesis based on the Logit model The dependent variable is whether or not a household owns a computer Independent variables, after justification, shall be household income, household social and demographic characteristics These variables are chosen from the literature review justification already mentioned

Following hereunder is to discuss how the proxies of these independent variables are chosen and their hypotheses

11 Log of household monthly expenditure per capita: The variable of household monthly income will be replaced by household monthly expenditure per capita, as according to Nguyen Trong Hoai et al (2005):

- Households tend to understate their actual income Richer households tend

to tell their income much lower

Complicated jobs make households hard to remember all of their income sources exactly

- Income from long-term plants or animals cannot be calculated exactly because the return is after several years, despite annual expenditures on these plants or animals

Some business in a household has constant fluctuations with gams and losses In a certain year, the household can suffer from loss (negative income), but we cannot expect the household to be a poor one In contrast, expenditure often depends on the assets owned by the household or depends on the expected income of the household Sudden high expenses, still, sometimes occurs, but only to non-poor households

- Expenditure is not only rarely understated than income but also is more stable through years, hence, we have enough evidence to use the expenditure to measure the living standard (Glewwe and Twum-Baah 1991, cited in Nguyen Trong Hoai et al 2005)

I take the log of household expenditure per capita to reduce the skewness and it is more comparable in the form of logarithm (expenditure increase by 1% makes more sense than increase by 1 unit of currency) Theories and empirical studies showed a positive relationship between income and computer ownership The hypothesis is that household with higher household expenditure per capita is more likely to own a computer The expected sign is positive

2/ Household size: is the number of residents in the household, not including the employees and temporary homestay persons Since all family members are potential computer technology users, it is clear that the larger the household, the greater the use of the computing technology Size specifies the upper limit of the number of users (Shih and Venkatesh 2003) The hypothesis is that the household size has positive effect on the household computer ownership The expected sign is positive

3/ Household residential location: is whether a household is in urban areas

or not This is a dummy variable, takes the value of 1 if the household resides in

urban areas and 0 otherwise Households in urban areas are believed to have better infrastructure to obtain computer utilities (e.g internet connection) and people tend to spend more on high-tech items than in rural areas because of their high income The expected sign is positive

4/ Presence of children at school age: 1s whether or not there is a child/children at school age (6-18) in the household Parents may utilize these technologies as part of their children's educational experience (Venkatesh et al 2003) and households with a school-age child are believed to have a higher computer ownership rate because school-age children use a computer at home (Stahura 2005) The presence of children at school-age, hence, is hypothesized to have positive effect on computer ownership of households The expected sign is positive

51 and 6/ Age and age square of household head respectively: is the age calculated from the year of birth of household head and the square of that age The expected sign of the age of household head is unknown, but the age square of the household head is hypothesized to be negative because young and mid-aged adults tend to have better computer skill to access to computers than too old adults (e.g

' over 60 years old)

71 Gender of household head: is a dummy variable (male= 1, female= 0) Female-headed households were found to be less confident about information technology than male-headed households in regard to products that were related to technology (Stahura 2005) The hypothesis is that a male household head is expected to have positive effect on household computer ownership The expected sign is positive

8/ Education of household head: is a dummy variable: based on the highest qualification the household head achieve: it takes 5 values: primary school, secondary school, high school, college and university degree Post graduate degree just includes 13 observations (0.1 %), so I exclude this group Dummy "No

qualification" is the base one, so it will not be included in the model (i.e m categories, so we have m-1 dummies) Educational attainment is found an important positive factor affecting the computer ownership in almost empirical studies, so the hypothesis is that the household with a higher educated household head will be more likely to own a computer

91 Occupation of household head: is determined as a dummy variable That

is whether the household head is a/an administrator/leader, a professional in science/technology/education/skilled work or an office clerk (taking value of 1) and otherwise (taking value of 0) Although the computer can be used by different persons in the household and meets various purposes, however, the purchase decision is made by the household head, which represents the preferences of the whole household Besides the effect of employment status (employed or not) in most empirical studies, the use of computer is positive to the occupation of the user m an economics/finance, technology and administration related field (Sakellariou and Patrinos 2003 ) So the hypothesis is that the fact that the household head 1s a/an administrator/leader, a professional m science/technology/education/skilled work or an office clerk will have positive effect on computer ownership

The model is specified as follows:

Li = ~0 + ~l(LNEXPPC) + ~2(LOCA) + ~3(SCH_CHILD) + ~4(HEADAGE) +

Ui

Particularly, the research is designed to answer the following questions:

- Does household income positively affect the computer ownership of the household?

- Does a household in urban area actually have a higher probability of owing

Determinants Variables Measurement thesized Related evidences

Hypo-sign

Venkatesh Presence of children at school age SCH CHILD Presence= 1 + Robertson, Stahura, Venkatesh

Household head characteristics

Kridel, Chaudhuri, Microsoft,

Stahura

-5 dummies: highest qualification

EDU2 secondary school = 1, ? Kridel, Chaudhuri, Rappoport, Educational attainment

-Chapter 3

AN OVERVIEW OF THE COMPUTER MARKET AND

COMPUTER OWNERSHIP IN VIETNAM

This Chapter includes two parts The first part provides an updated overview of computer market, computer ownership and projects that have been done to enhance the computer ownership in Vietnam The second part is the descriptive analysis on factors affecting the computer ownership by different groups of households in Vietnam, based on the VLSS 2004 statistics

3.1 An overview of computer ownership, computer market and empirical projects carried out to enhance the computer ownership in Vietnam

Vietnam

According to the Vietnam ICT Outlook 2006, the Information Society Index of Vietnam in 2005 dropped 1 step, ranking end of the listed countries (53/53) This index assessing the level of social information development, ranked by IDC and World Time, including 15 factors in 4 fields: computer, Internet, telecom and social infrastructures

Figure 1 Components of Information Society Index

Source: Vietnam ICT Outlook 2006- Ho Chi Minh City Computer Association

Of all the compositions of the lSI are some main items related to household access: personal computers per household, broadband households, and home internet users This means Vietnam is one of the worst countries listed in information social development, especially to households It gives an urge to boost the socialization of information technology and computer access to households in Vietnam

To see more ofthe situation, let us have a look at Table 1, 2 and Chart 1

Table 1 Number of computers of Vietnam in comparison with other countries in 2005

Country/ Region Total number of computers

(thousand units) China

1,044

226,640 17,450 296,491 236,116 16,119

792,816

Number of computers/

1000 people

4.08 1.54 1.36 54.15 54.49 19.16 4.46 52.78 5.83

···~·~···

1.26

6.51 2.24 34.73 30.21 50.73

13.30

Source: International Telecom Union (an organization of the United Nations)

statistics-cited in £-commerce Report 2006 by Ministry ofTrade of Vietnam (published 2007)

Source: Vietnam Living Standard Survey VLSS 2004 (author's calculation)

In the table, we can see that only 5.05% of households in Vietnam own a computer, a very low rate There's a very wide gap with the same statistics of some OECD countries

Chart 1 Households with a home computer, 2000-2005, percentage of all households

Source: OECD, ICT database and Eurostat, "Community Survey on ICT usage in

households and by individuals", September 2006

Mr Than Trong Phuc, CEO of Intel group - Indochina region, said that Vietnam has approximately 3

- 3 million households with annual income over 6,000 USD;

- 3 million households with annual income between 2,000 and 6,000 USD, and;

12 million households with annual income less than 2,000 USD

With the first group, computer ownership doesn't matter them However, with the second and third group, how can they get access to computers?

A research done by TNS group in Vietnam in 2005 showed that the purchase power for ICT items in Hanoi and Ho Chi Minh City had a growth rate

of 25% during 2003 and 2004 The research also showed that 50% households in the centers and suburbs of Hanoi and Ho Chi Minh city own mobile phones and 30% households in the centers and suburbs of Hanoi and Ho Chi Minh city own computers4 Another recent market research carried by the AC Nielsen in 2006 showed that the rate of Vietnamese owning computers is increasing rapidly, in

2006 that rates in Hanoi and Ho Chi Minh City respectively are 48% and 45% Of these people, 32% use internet regularly, mainly at the age of from 18 to 30

Internet use boom also boosts computer ownership in households Year

2006 saw a fast growth rate of internet and telecom in Vietnam The number of internet subscribers in 2006 increased by 38% compared to that of 2005 The number of internet users in 2006 increased by 36%, accounted for 17.5% of total population If 2005 marked Vietnam's reach at the average of internet user rate of Asia (9.78%), at the end of 2006, Vietnam passed the world's average internet user rate (16.8%) (E-commerce Report 2006) 5

According to Mr Chris Morley, CEO of AC Nielsen Vietnam, 60% of Vietnam population is under 30 years old6 More details, other statistics shows that 50% of the population is less than 25 years old7 with more than 23 million people

of Vietnam is at school-age "Vietnam is one of the youngest population in the

4 Ministry oflndustry: http://www.moi.gov.vn/News/Detail.asp?Sub=3l&id=i2936 dated May 30, 2005

5

Statistics in Vietnamnet Internet Center (VNNIC) Dec 2006 www.vnnic.net.vn and Internet World cited in E-commerce Report 2006- Ministry ofTrade of Vietnam (2007)

Stats-6 SaiGon Tiep Thi Magazine dated September 03,2005

7 McCool, a Reuters's correspondent, cited in http://www.moi.gov.vn/News/Detail.asp?Sub=31&id=12936

world, which is a very good market for computers and mobile phones", said Mr Morley During the first 6 months in 2006, the number of computers sold increased 24% compared to the same period of2005 Impressively, the number of computers sold in the last quarter of 2006 is more than the sum of 3 previous quarters 70% of computers sold are assembled domestically because of the advantage of price and distribution network Most of the computers purchased are low-price computers, mainly bought by households (especially for students), and small enterprises However, an alert is that of all computers sold in Vietnam, only 10% are brand-name computers, and the rest are computers assembled from unknown components with no warranty, so the quality is not ensured 8

3.1.2 Computer market in Vietnam and the prospect after Vietnam joined WTO

Computer industry includes hardware and software industry As we only concern about the computer itself in this study, not software, therefore, this section provides an overview of demand for hardware only (in which computer and computer components is a main contribution)

According to Vietnam ICT Outlook 2006 by HCA9 "IT market value" is

domestic IT spending, which includes (i) imports for domestic IT spending; (ii)

manufactures for domestic IT spending

IT market value includes hardware and software market value, here just mentions hardware market value only

Table 3 Hardware market value in Vietnam 2000-2005 (million USD)

Source: Vietnam ICT Outlook 2006, Ho Chi Minh City Computer Association

The table shows a fast growth through years in hardware demand and computer demand in particular in Vietnam

On the other hand, hardware industry (including hardware manufacture for domestic consumption and exports) in 2005 saw its first time passing 1 billion USD (from 760 million USD to 1.15 billion USD) and became one of the 7 industries having the largest exports In hardware manufacture, only 108 million USD was for domestic consumption However, most products were mainly produced by foreign invested companies (FDI companies) in Vietnam for export

to other countries 2006 saw the investment of some big computer corporations into Vietnam, such as the !-billion investment project to produce chips and semi-conduct components of Intel

Domestic companies - especially some Vietnamese leading computer manufacturers - after several fast-growing years, had little growth Two leading companies, FPT E-lead and CMS, in 2005 produced more computers than 2004; however, due to decreasing price, the revenue was only almost the same as 2004 Three biggest computer manufacturers are FPT Elead, CMS and Mekong Green respectively had the revenue of 13.4, 9.3 and 6.0 million USD in 2005

Low income makes households unaffordable for computers as we have discussed, or, computer price is still too high for households After Vietnam joined the WTO in November 2006, tariff bands are expected to lower to between 0% -