Data and Variable Definitions

ASIAN ECONOMIC JOURNAL 4 © 2009 The Authors Journal compilation © 2009 East Asian Economic Association and Blackwell Publishing Ltd. Hoffman and Kassouf, 2005. Yen and Rosinski 2008 show that such an approximation can lead to substantial errors. They also present the conditional mean of the dependent variable and marginal effect formulas for a log-transformed sample selection model. In essence, the conditional mean of the dependent variable y is Yen and Rosinski, 2008: E y | y = exp x ′β + σ 2 2 Φ z ′α + ρσ Φ z ′α . 4 Because the marginal probability of a positive observation is: Pr y = Φz′α 5 the unconditional mean of y is: E y = expx′β + σ 2 2 Φz′α + ρσ. 6 Differentiating Equations 4, 5 and 6 gives the marginal effects on probability, conditional mean and unconditional mean of a common element of x and z say x j = z j : ∂Pr y 0∂x j = φz′αα j 7 ∂Ey|y 0∂x j = [Φz′α] −2 expx ′β + σ 2 2{[ Φz′αφz′α + ρσ − φz′αΦz′α + ρσ]α j + Φz′αΦz′α + ρσβ j } 8 ∂Ey∂x j = expx′β + σ 2 2 [ φz′α + ρσα j + Φz′α + ρσβ j ]. 9 These marginal effects can be evaluated at data points of interest, such as the sample means of explanatory variables.

III. Data and Variable Definitions

III.1 The survey The dataset used in the present study is from the Malaysian Household Expenditure Survey 20042005 collected by the Department of Statistics Malaysia 2005. This dataset is the most recent of the national household expenditure surveys. The sample was designed using a stratified multi-stage, area probability sampling method, thus ensuring that socio-economic and geographical considerations are taken into account to reflect the Malaysian population. In the survey, respondents were asked to record their monthly expenditure on tobacco products. In addition, socio-economic and other demographic charac- teristics of the respondents were also recorded. Although a total number of 14 084 households responded to this survey, 2 of those households were found to have incomplete information. Therefore, a total of 14 082 observations were subsequently retained for analysis. From this sample, a total of 5291 households ROLE OF EDUCATION IN CIGARETTE SMOKING 5 © 2009 The Authors Journal compilation © 2009 East Asian Economic Association and Blackwell Publishing Ltd. 37.57 percent reported tobacco expenditures during the survey period Table 1. III.2 The variables Given the lack of empirical research on Malaysian tobacco products, the selection of variables likely to affect household purchase decisions and expenditure levels relies on previous cigarettes and tobacco studies by Wasserman et al. 1991, Blaylock and Blisard 1992a,b, Douglas and Hariharan 1994, Jones and Labeaga 2003, Ross and Chaloupka 2004 and Yen 2005a,b. The following socio-demographic variables are therefore hypothesized to influence the probability and amount of expenditure on tobacco products: i ethnicityrace; ii education level; iii occupation type; iv location of residence; v household size; vi gender of household head; vii gross monthly household income; and viii age of household head Table 1. Studies by Blaylock and Blisard 1992a,b, Douglas and Hariharan 1994, Ross and Chaloupka 2004 and Yen 2005a,b suggest that ethnicity be included to allow for the possibility of cultural and taste differences influencing tobacco expenditure. As most of these previous studies compare whites, blacks, Hispanics Table 1 Variable definitions and sample statistics Variable Definition Mean Standard deviation EXPENDITURE Household expenditure on tobacco RMmonth dependent variable 27.84 57.45 Consuming households n = 5291 74.09 73.20 EDUC Number of years of formal education of household head 8.51 4.81 HHSZ Total number of family members in household 4.35 2.22 INCOME Gross monthly household income in RM 2732.03 2669.17 AGE Age of household head years 46.85 14.06 Binary variables yes = 1; no = 0 MALAY Malay household head reference 0.57 0.50 CHINESE Chinese household head 0.23 0.42 INDIAN Indian household head 0.06 0.23 OTHER Household head of other race 0.15 0.36 WHITE Household head is a white-collar worker 0.25 0.43 URBAN Household resides in an urban area 0.67 0.47 MALE Household head is male 0.84 0.37 AGE 15–30 Household head is between 15–30 years old 0.12 0.33 AGE 31–56 Household head is between 31–56 years old reference 0.64 0.48 AGE ≥57 Household head is ≥57 years old 0.24 0.43 Source: Compiled using the Malaysian Household Expenditure Survey 20042005 Department of Statistics Malaysia, 2005. ASIAN ECONOMIC JOURNAL 6 © 2009 The Authors Journal compilation © 2009 East Asian Economic Association and Blackwell Publishing Ltd. and Asians in their ethnicityrace categories, the unique racial composition in Malaysia allows a comparison of four distinct races Malay, Chinese, Indian and a small proportion of various other races. In the current study, respondents are segregated into Malay base group, Chinese, Indian, and other races to allow for the possibility of cultural, ethnic and religious differences to influence tobacco consumption patterns amongst Malaysians. Prior health economics studies have consistently emphasized the importance of education or schooling Jones, 1989; Chaloupka, 1991; Sander, 1995; Dorsett, 1999; Nayga, 1999; Jones and Labeaga, 2003; Yen, 2005a. The various theories that arise include the contention that education is a form of human capital investment that improves cognitive skills and enhances the understanding of the relationship between health behavior and health Berger and Leigh, 1989; Kenkel, 1991. Others have hypothesized that the reverse might be true, as having healthier individuals could result in an increase in the demand for education Currie and Hyson, 1999. Education could also be a proxy for unobservables, such as time preference, that are related to health. In such instances, self-selection processes might be evident Becker and Murphy 1988; Chaloupka, 1991; Sander, 1995. In the current study, the number of years of formal education EDUC possessed by the household head is used, with the range varying from those without any formal education 0 to those with tertiary education 17. Occupation type WHITE consisting of white-collar and blue-collar households is included to reflect the social status of respondents Yen, 2005a; Yen and Jones, 1996. Location of residence URBAN is included given that peer or work pressure in urban surroundings or more intense media advertising in metropolitan areas might determine tobacco consumption patterns Blaylock and Blisard, 1992a,b; Jones and Labeaga, 2003; Yen, 2005a. Blaylock and Blisard 1992a provide evidence to suggest that household size HHSZ , represented by the number of individuals living in the household, might determine decisions made in relation to tobacco consumption expenditure. A gender variable MALE is included in the present study as Yen 2005a,b and Douglas and Hariharan 1994 determine that gender might account for preferences among consumers of tobacco products. Total monthly household INCOME RM is considered to account for household spending patterns as previous researchers have found that wealth might have positive Douglas and Hariharan, 1994; Goel and Morey, 1995, negative Wasserman et al., 1991 or even no Blaylock and Blisard, 1992b; Dorsett, 1999 statistical effects on participation decisions or expenditure levels for tobacco consumption. Finally, AGE of the household head in years and age groups denoting younger AGE15 –30, middle age AGE31–56 and retiree AGE ≥ 57 house- holds are used in the current model with the assumption that differences in age and life-cycle patterns lead to variations in preferences and expenditure patterns for tobacco products Jones, 1989; Blaylock and Blisard, 1992a,b; Jones and Labeaga, 2003; Ross and Chaloupka, 2004; Yen, 2005a,b. ROLE OF EDUCATION IN CIGARETTE SMOKING 7 © 2009 The Authors Journal compilation © 2009 East Asian Economic Association and Blackwell Publishing Ltd. III.3 Characteristics of survey respondents Descriptive statistics for the sample are presented in Table 1. 2 The average household expenditures on tobacco amount to RM27.84 for the overall sample compared to RM74.09 amongst consuming households. In the entire sample, the average educational attainment of the household head is slightly more than 8 years of formal education at least high school education and the average household size is approximately 4. The ethnic composition of the sample reflects the population of Malaysia, whereby Malays 57 percent base group, Chinese 23 percent and Indians 6 percent form the main ethnic groups, and approximately 15 percent belong to various other races. With 25 percent of the total sample classified as having white-collar occupations, these households have an average monthly income of approximately RM2732. Approximately 67 percent of the sample resides in urban areas, with 84 percent of households being headed by men. The average age of the household head is approximately 47 years. Approximately 12 percent of the sample is in the younger age group between 15 –30 years old; 64 percent in the middle age group between 31–56 years old base group; and 24 percent are retirees in the 57 years and above age group Table 1.

IV. Estimation results