International Advances in Economic Research (2005)11:275–290 DOI: 10.1007/s11294-005-6657-7
* IAES 2005
Factors Affecting Cigarette Demand ELENI RAPTOU*, KONSTADINOS MATTAS*, EFTHIMIA TSAKIRIDOU**, AND CONSTANTINOS KATRAKILIDIS***
Abstract This paper addresses the impact of smoking restrictions in workplaces and educational establishments, cigarette price measures and various psychosocial indicators on cigarette demand, controlling for demographic and socio-economic factors. The data used for the analysis are collected via questionnaire that was administered in personal-in home interviews. A two-part model of cigarette demand [Cragg, J. G. BSome Statistical Models for Limited Dependent Variables with Application to the Demand for Durable Goods,^ Econometrica, 39, 5, 1971, pp. 829 Y 44.] is estimated. According to the estimations, cigarette price measures do not influence cigarette demand. On the contrary, smoking restrictions in workplaces and educational establishments and most of the psychosocial variables are found to affect cigarette demand considerably. (JEL D12, I00, M31)
Introduction Consumers make rational and informed choices after weighting the costs and benefits of purchases, and compromising all the costs of choice according to the principle of consumer sovereignty. Thus, smokers perceive benefits from smoking, such as pleasure and the avoidance of withdrawal, that outweigh the perceived costs. However, the decision to smoke differs from the decision concerning the consumption of other goods. Many smokers are not fully informed or underestimate the high risks of disease and premature death, which their choice entails. There is a delay between smoking initiation, smoking addiction and tobacco related diseases, and tobacco industry disguises the information on the health risks of smoking. Most smokers become addicted to tobacco in adolescence or early adulthood and even when are informed on smoking risks, they do not always have the capacity to make sound decisions. In addition, smoking imposes costs on non-smokers, such as direct physical costs (health impacts due to passive smoking), financial costs (tobacco related public health care costs) and Fcaring externalities_ (emotional suffering of non-smokers due to the illness and death of smokers) [Pan American Health Organization, 2000; Wilkins et al., 2001]. Therefore, Bthe existence of internal costs (in the form of harm to smokers themselves) and external costs (in the form of harm to others) justifies both government intervention and research on the effects and benefits of alternative policies to limit demand of addictive substances such as tobacco [ Wilkins et al., 2001].^ The aim of the present study is to investigate the impact of smoking restrictions in workplaces and educational establishments, cigarette price measures and psychosocial
* Aristotle University of Thessaloniki, ** Democritus University of Thrace, and *** Aristotle University of Thessaloniki — Greece.
275
276
ELENI RAPTOU ET AL.
factors on cigarette demand controlling for various demographic and socio-economic variables. The main reason to analyze cigarette demand is to examine how cigarette demand is determined, which factors affect it and the direction of their influence. This study comprises four sections, and it is structured as it follows. The first section briefly presents the main outcomes of other researches on cigarette demand. The next section includes the methods employed for the analysis, and more specifically describes the data and settings, the variables used in the econometric analysis, and the model applied. The following section elucidates the factors that are found to affect smoking behavior, and the last section comments on these estimations and compares them with the outcomes of other researches. Previous Studies Smoking related issues have been of great interest for econometric analyses for many years. An increasing number of studies on cigarette demand use data from individual level surveys and most of these studies deal with price responsiveness of cigarette smoking. Chaloupka and Warner [1999] give an analytical report of the price elasticities of demand, which are estimated in the most recent studies. With respect to price responsiveness among different age groups, adolescents are found to be more sensitive to cigarette price than adults concerning mainly smoking participation rather than cigarette consumption [Lewit et al., 1981; Grossman et al., 1983; Chaloupka and Grossman, 1996; Chaloupka and Wechsler, 1997; Tauras and Chalpoupka, 1999; Hersch, 2000; Emery et al., 2001; Ross and Chaloupka, 2003]. Regarding the decision to smoke, it was found that young adults are more price-sensitive than older ones [Lewit and Coate, 1982]. In addition, Farrelly and his colleagues [1998] estimated that young adults are more price-sensitive than the full sample of young and older adults. In opposition to the above-mentioned researches, Chaloupka in his 1991, 1992 publications presumed that cigarette price increase reduces cigarette consumption but this increase is understated if addiction is ignored. Chaloupka found young adults to be less price-sensitive than older ones supporting the results of Wasserman and his colleagues [1991] who calculated insignificant effects of cigarette prices on cigarette consumption by adolescents. Another research topic on smoking concerns of whether or not smoking restrictions in public places and in private workplaces have negative effects on cigarette consumption. Such studies deduced that these restrictions have a negative impact on adult cigarette consumption [Chaloupka, 1992; Chaloupka and Wechsler, 1997; Evans et al., 1999; Tauras and Chalpoupka, 1999]. With respect to youth smoking, Chaloupka and Grossman [1996] asserted that strong restrictions on smoking in public places would reduce the smoking prevalence among adolescents. Other studies support that smoking restrictions in public places and in private worksites do not affect youth cigarette demand [Emery et al., 2001; Ross and Chaloupka, 2003]. While most of the econometric studies analyze the effect of cigarette price and antismoking regulations on cigarette demand, only a few of them take into account psychosocial variables. Emery et al. [2001] examined adolescents’ smoking patterns and concluded that the psychosocial variables are strongly associated with cigarette smoking in the expected directions. Thus, poor school performance, rebelliousness, exposure, depression, and belief that cigarettes are easy to get are positively associated with smoking behavior. On the other hand, participating in sports and strong parental bond are negatively associated with smoking behavior.
277
CIGARETTE DEMAND
TABLE 1 Variables Used in the Econometric Analysis Variable Dependent variables Smoking participation: SMOKER Average daily cigarette consumption: CONSUME Independent variables Demographic variables Gender: FEM Age: AGE1 Age: AGE2
Age: AGE3 Age: AGE4 Age: AGE5 Marital status: SINGLE
Under age members: CHILDEX
Region of growth: BCITY
Region of growth: BTOWN
Region of growth: BVILL
Socioeconomic variables Occupational status: CIVSERV Occupational status: PRIVEMPL
Definition Dichotomous indicator equal to one for smokers and zero for non-smokers The average cigarette quantity (number of cigarettes) that a smoker consumes daily
Dichotomous indicator equal to one for women and zero for men Dichotomous indicator equal to one for adolescents (15Y18 years old) and zero otherwise Dichotomous indicator equal to one for young adults (19Y29 years old) and zero otherwise-omitted variable Dichotomous indicator equal to one for adults 30Y40 years old and zero otherwise Dichotomous indicator equal to one for adults 41Y51 years old and zero otherwise Dichotomous indicator equal to one for adults older than 51 years old and zero otherwise Dichotomous indicator equal to one if the respondent is single (including unmarried, separated, divorced and widowed) and zero if he/she is married Dichotomous indicator equal to one if at least one person less than 18 years old lives permanently in the respondent’s household (including the respondent if he/she is under age), and zero otherwise Dichotomous indicator, which takes on a value of one if the respondent grew up in a big city and zero otherwise Dichotomous indicator, which takes on a value of one if the respondent grew up in a town and zero otherwise Dichotomous indicator, which takes on a value of one if the respondent grew up in a village and zero otherwise-omitted variable Dichotomous indicator that takes on a value of one if the respondent is a civil servant and zero otherwise Dichotomous indicator that takes on a value of one if the respondent is an employee in the private sector and zero otherwise-omitted variable
278
ELENI RAPTOU ET AL.
TABLE 1 Continued Variable Occupational status: SELFEMPL Occupational status: MANUAL Occupational status: RETIR Occupational status: DEPEN
Lower educational level: HIGHSCHO Moderate educational level: TEIEK
High educational level: UNIVERST
Attachment to religion: RELIG
Disposable income: INC
Psychosocial variables Family smoking: FAMSMOK
Friends smoking: FRSMOKER Cigarette advertising influence: ADINFLU Participation in cigarette sales promotion events: SALPROM Influence by cigarette offerings: SMOKINF
Definition Dichotomous indicator that takes on a value of one if the respondent is an employer in the private sector and zero otherwise Dichotomous indicator that takes on a value of one if the respondent is a manual worker/farmer and zero otherwise Dichotomous indicator that takes on a value of one if the respondent is retired and zero otherwise Dichotomous indicator that takes on the values of one if the respondent depends economically on others and zero otherwise Dichotomous indicator equal to one for educational level less than high school and zero otherwise Dichotomous indicator that takes on a value of one for respondents who are students of a technological institute or have a degree from a technological institute and a value of zero otherwise-omitted variable Dichotomous indicator equal to one for respondents who have at least a university degree or they are university students at the time that this research took place, and zero otherwise Dichotomous indicator, which takes on a value of one if the respondent attends religious services at least once a week (frequent attendance) and a value of zero otherwise (infrequent attendance) A continuous variable that represents the average monthly disposable income (in Euro) from all sources (employment, allowances and other) Dichotomous indicator equal to one for respondents whose most of their family members are smokers and zero otherwise Dichotomous indicator equal to one for respondents whose most of their friends are smokers and zero otherwise Dichotomous indicator equal to one for respondents who are influenced by cigarette advertising and zero otherwise Dichotomous indicator equal to one for respondents who participate in sales promotion events and zero otherwise Dichotomous indicator equal to one for respondents who feel more receptive to smoking whenever they are offered a cigarette and zero otherwise
CIGARETTE DEMAND
279
TABLE 1 Continued Variable Family control: FCONTROL
Definition Dichotomous indicator equal to one for respondents who feel that their families control their lives and zero otherwise Lack of family bond: LOSS Dichotomous indicator equal to one for respondents FAMB who do not have a strong bond with their families and zero otherwise Socializing: SOCBEH Dichotomous indicator equal to one for respondents who are sociable and zero otherwise Stress for Bwhat other people Dichotomous indicator equal to one for respondents think of me^: STRAPEAR who feel anxious of what other people think of them and zero otherwise Risky behavior: RISKB Dichotomous indicator equal to one for respondents who appear to have risky behavior and zero otherwise Satisfaction from work / Dichotomous indicator equal to one for respondents studies: WSATISF who feel satisfied from their work/studies and zero otherwise Stress for work /studies: Dichotomous indicator equal to one for respondents WSTRESS who feel stressed for their work/studies and zero otherwise Healthy lifestyle: Dichotomous indicator equal to one for respondents HCONCERN who have a healthy lifestyle and zero otherwise Perceived negative Dichotomous indicator equal to one for respondents consequences of smoking: who support that smoking has negative NEGCON consequences and zero otherwise Perceived positive Dichotomous indicator equal to one for respondents consequences of smoking: who support that smoking has positive POSCON consequences and zero otherwise Smoking restrictions in workplaces and educational establishments Total ban: WORKT Dichotomous indicator that takes on the value of one if there is total smoking ban in the respondent’s workplace/educational establishment and zero otherwise Smoking restrictions: Dichotomous indicator equal to one if there are WORKR smoking restrictions (smoking is permitted in specific areas) in the respondent_s workplace/ educational establishment and zero otherwise No smoking restriction: Dichotomous indicator equal to one if smoking is WORKNO free in the respondent_s workplace/educational establishment and zero otherwise-omitted variable Cigarette price measure Evaluation of cigarette price: Dichotomous indicator equal to one if the PRICOPIN respondent evaluates cigarette price as high and zero otherwise
280
ELENI RAPTOU ET AL.
Even though smoking rates seem to decrease in most of the developed countries [ World Bank Report, 1999], smoking consumption in Greece increased from 2,864 cigarettes per capita in 1990 to 3,020 in 1997 [Eurostat Yearbook, 2001]. The first antismoking policies were enforced during 1979Y1982 and have been improved since then to focus on the consequences of smoking in public health and its insurance. According to the Government Brochure 1001, since 2002 smoking is totally banned in public places, in health services and in transportation means. Researches on cigarette demand conducted in Greece concluded that price increases through taxes increases do not reduce cigarette consumption substantially [Stavrinos, 1987; Hondroyiannis and Papapetrou, 1997; Nikolaou and Velentzas, 2001; Michalopoulos and Demoussis, 2002; Dritsakis, 2003]. All these researches employed time series or aggregate data for the econometric analyses except Michalopoulos and Demoussis [2002] who used individual level data. Methods Setting and Data This research is based on primary data, collected via questionnaire. The questionnaire construction and design was based on the results of a qualitative research (eight focusgroup discussions, N = 50). The formal questionnaire was administered in personal-in home interviews to the residents of Thessaloniki, an urban area of one million people in northern Greece (between May and September 2003). Finally, 680 questionnaires were collected, and constitute a representative sample of the total population. Variables Description All the variables, constructed from the data, are defined analytically on Table 1. Conventional cigarette price measures, as published cigarette price data by the government or the tobacco industry for different municipalities, are not applied to the present research because there is no price or tax variation across Greece. The description of the sample is derived through Table 2, where all the variables are introduced in percentages. In addition, the main descriptive statistics (means and standard deviations) for the variables used in the econometric analysis are presented on Table 3. Model Specification A two-part model estimated by Cragg [1971] is applied for data analysis. This model presumes that the decision concerning the consumption of a product is disparate to the decision concerning the quantity consumed. According to the context of this research, the first part uses a probability model (probit) to estimate an individual’s decision to smoke as a function of the demographic, socio-economic and psychosocial variables, the smoking restrictions indicators and the cigarette price measures that were reported on Table 1. Therefore, the first part of the model (smoking participation equation) produces: Z Xi 1 1 pffiffiffiffiffiffi exp z 2 dz PrðY ¼ 1Þ ¼ ðXiÞ ¼ ð1Þ 2 2 1 PrðY ¼ 0Þ ¼ 1 ðXi Þ EðY j X Þ ¼ 0xð1 ðXi ÞÞ þ 1 xðXi Þ ¼ ðXi Þ
ð2Þ ð3Þ
where X = k 1 vector of independent variables, = k 1 vector of coefficients and F refers to the standard normal cumulative distribution function.
281
CIGARETTE DEMAND
TABLE 2 Means of the Variables Used between Smoking Groups (in Percentages) Smoking participation (%) Aver. daily cig. consum. FEM (%) SINGLE (%) AGE1 (%) AGE2 (%) AGE3 (%) AGE4 (%) AGE5 (%) CHILDEX (%) HIGHSCHO (%) TEIEK (%) UNIVERST (%) CIVSERV (%) PRIVEMPL (%) SELFEMPL (%) MANUAL (%) RETIR (%) DEPEN (%) BCITY (%) BTOWN (%) BVILL (%) RELIG (%) PRICOPIN (%) WORKT (%) WORKR (%) WORKNO (%) FCONTROL (%) HCONCERN (%) RISKB (%) SOCBEH (%) STRAPEAR (%) LOSSFAMB (%) FAMSMOK (%) FRSMOKER (%) WSTRESS (%) WSATISF (%) SMOKINF (%) SALPROM (%) NEGCON (%) POSCON (%) ADINFLU (%) INC (in Euro)
Non-Smoker 40.6 0 24.7 28.8 4.6 19.1 10.3 4.4 2.2 18.5 11.8 11.9 16.9 6.5 11.2 4.3 1.6 0.1 16.9 25.9 6.9 7.8 12.1 28.8 7.6 17.6 15.7 6.0 17.2 1.5 22.2 21.8 4.3 25.9 26.8 22.8 23.8 4.3 2.5 39.7 7.4 4.3 585.96
Smoker 59.4 16.6 33.7 40.6 4.1 25.3 19.1 7.5 3.4 23.8 15.9 22.4 21.2 7.8 19.3 7.9 5.3 1.5 17.6 38.5 11.3 9.6 4.0 43.1 5.0 26.4 27.6 24.3 13.7 13.2 34.6 28.0 16.8 49.3 52.4 33.7 23.5 16.5 13.7 52.1 29.3 22.5 786.09
282
ELENI RAPTOU ET AL.
TABLE 3 Descriptive Statistics for the Dependent and the Independent Variables (Means and Standard Deviations) Variable Dependent variables Smoking participation: SMOKER Average daily cigarette consumption: CONSUME Independent variables Demographic variables Gender: FEM Age: AGE1 Age: AGE2 Age: AGE3 Age: AGE4 Age: AGE5 Marital status: SINGLE Under age members: CHILDEX Region of growth: BCITY Region of growth: BTOWN Region of growth: BVILL Socioeconomic variables Occupational status: CIVSERV Occupational status: PRIVEMPL Occupational status: SELFEMPL Occupational status: MANUAL Occupational status: RETIR Occupational status: DEPEN Lower educational level: HIGHSCHO Moderate educational level: TEIEK High educational level: UNIVERST Attachment to religion: RELIG Disposable income: INC
Full Sample
Non-Smoker
Smoker
Y
16.625 (12.150)
0.59 (0.491) 9.865 (12.509)
0.58 0.09 0.44 0.29 0.12 0.06 0.69 0.42
(0.493) (0.282) (0.497) (0.456) (0.324) (0.230) (0.461) (0.494)
0.60 0.12 0.48 0.26 0.10 0.05 0.72 0.46
(0.490) (0.320) (0.500) (0.438) (0.302) (0.215) (0.452) (0.500)
0.55 0.07 0.44 0.33 0.13 0.04 0.70 0.40
(0.498) (0.260) (0.496) (0.470) (0.333) (0.187) (0.459) (0.491)
0.64 (0.479)
0.64 (0.481)
0.65 (0.477)
0.18 (0.386)
0.17 (0.378)
0.19 (0.391)
0.17 (0.379)
0.19 (0.393)
0.16 (0.365)
0.14 (0.350)
0.16 (0.368)
0.13 (0.339)
0.30 (0.460)
0.28 (0.452)
0.33 (0.471)
0.12 (0.328)
0.11 (0.316)
0.13 (0.339)
0.09 (0.254)
0.04 (0.199)
0.12 (0.330)
0.02 (0.126)
Y
0.01 (0.113)
0.34 (0.476)
0.40 (0.491)
0.27 (0.443)
0.28 (0.448)
0.28 (0.448)
0.27 (0.446)
0.34 (0.475)
0.30 (0.460)
0.37 (0.484)
0.38 (0.486)
0.42 (0.495)
0.35 (0.479)
0.16 (0.367)
0.30 (0.458)
0.05 (0.222)
712.27 (635.787)
585.96 (489.616)
786.09 (799.585)
283
CIGARETTE DEMAND
TABLE 3 Continued Variable Psychosocial variables Family smoking: FAMSMOK Friends smoking: FRSMOKER Cigarette advertising influence: ADINFLU Participation in cigarette sales promotion events: SALPROM Influence by cigarette offerings: SMOKINF Family control: FCONTROL Lack of family bond: LOSS FAMB Socializing: SOCBEH Stress for Bwhat other people think of me^: STRAPEAR Risky behavior: RISKB Satisfaction from work / studies: WSATISF Stress for work /studies: WSTRESS Healthy lifestyle: HCONCERN Perceived negative consequences of smoking: NEGCON Perceived positive consequences of smoking: POSCON Smoking restrictions in workplaces Total ban: WORKT Smoking restrictions: WORKR No smoking restriction: WORKNO Cigarette price measure Evaluation of cigarette price: PRICOPIN
Full Sample
Non-Smoker
Smoker
0.75 (0.432)
0.64 (0.480)
0.83 (0.379)
0.79 (0.407)
0.65 (0.477)
0.88 (0.324)
0.27 (0.443)
0.11 (0.311)
0.39 (0.487)
0.16 (0.369)
0.06 (0.237)
0.23 (0.423)
0.21 (0.406)
0.10 (0.306)
0.28 (0.448)
0.30 (0.460)
0.15 (0.361)
0.42 (0.452)
0.21 (0.408)
0.11 (0.311)
0.28 (0.023)
0.57 (0.496) 0.50 (0.500)
0.55 (0.499) 0.53 (0.500)
0.58 (0.494) 0.47 (0.500)
0.15 (0.354) 0.47 (0.500)
0.04 (0.190) 0.59 (0.493)
0.23 (0.418) 0.39 (0.489)
0.56 (0.496)
0.57 (0.496)
0.56 (0.497)
0.31 (0.462)
0.43 (0.496)
0.22 (0.413)
0.92 (0.275)
0.98 (0.148)
0.88 (0.330)
0.37 (0.482)
0.18 (0.387)
0.51 (0.501)
and educational establishments 0.13 (0.333) 0.19 (0.390) 0.44 (0.497) 0.43 (0.496)
0.09 (0.280) 0.45 (0.498)
0.43 (0.496)
0.38 (0.487)
0.47 (0.500)
0.72 (0.450)
0.71 (0.453)
0.73 (0.447)
Notes: Standard deviations are in parentheses. a Retired is constant when SMOKER = 0. It has been omitted. b Daily cigarette consumption is constant when SMOKER = 0. It has been omitted.
284
ELENI RAPTOU ET AL.
The second part uses least square techniques to estimate average daily cigarette consumption by smokers (only part of the sample), where the dependent variable is the natural logarithm of the average daily cigarette consumption variable. Both parts of the Cragg model include the same set of independent variables. The conditional demand equation (second part of the model) obtains the general form: Log Q ¼ 0 þ Xi i þ "
ð4Þ
Where: " = error term, Log Q = logarithm of average daily cigarette consumption, X = k 1 vector of independent variables, = k 1 vector of coefficients.
Results Table 4 provides the results of the two-part model estimation. The cigarette price measure (PRICOPIN) is statistically insignificant in both equations of the Cragg model. Even though consumers evaluate cigarette prices as being high, these do not dissuade them from smoking. Total smoking bans in workplaces and educational establishments ( WORKT) discourage cigarette consumption, even though they do not affect smoking participation. Eventually, in such case, employees and students who are either established or experimenters smokers, they are forced to limit the quantity consumed or quit smoking. With respect to the demographic characteristics, the most important conclusions, accrued from the statistically significant variables, are summarized as it follows. Significant gender related differences are observed in cigarette smoking. Consequently females ( FEM ) are found to be much more likely to smoke (as it derives from the smoking participation equation) but to smoke less in average than males (as it derives from the conditional demand equation). Consumer’s age does not affect smoking participation. With respect to cigarette consumption, adolescents (AGE1) are more likely to consume fewer cigarettes than adults. Most of the socio-economic variables are found to be statistically insignificant in both equations and only a few of them affect smoking behavior. Both low and high educational levels (HIGHSCHO, UNIVERST) are negatively associated with smoking participation. Therefore, individuals with low or high educational level are less likely to smoke. The variable that describes attachment to religion ( RELIG) is statistically significant in the smoking participation equation. Thus, individuals with a stronger attachment to religion are less likely to smoke. The variable INC, which corresponds to the monthly average disposable income, is statistically significant and positively associated with smoking participation. On the other hand, it does not affect the quantity of cigarettes consumed. The effect of income is expressed as income elasticity. Three types of income elasticity are computed from the two-part model: income elasticity of participation, income elasticity of consumption and total income elasticity of demand. The income elasticity of participation is computed by using the average of the partial derivatives from the Probit model, and represents the percentage change in the smoking participation at the personal level caused by 1 percent change in disposable income. The income elasticity of participation is estimated to be 0.21, which means that if income increases per 1 percent smoking participation increases per 0.21 percent. The income elasticity of consumption results from the OLS logYlinear estimates and measures the percentage change in the average number of cigarettes consumed at the personal level by smokers after a 1 percent change in their disposable income. For the present study, it is estimated to be 0.07. The total income elasticity of demand is the sum
CIGARETTE DEMAND
285
TABLE 4 Two-Part (Cragg) Model Estimation
Variable INTERCEPT FEM SINGLE AGE1 AGE3 AGE4 AGE5 CHILDEX HIGHSCHO UNIVERST CIVSERV SELFEMPL MANUAL RETIR DEPEN BCITY BTOWN RELIG PRICOPIN WORKT WORKR FCONTROL HCONCERN RISKB SOCBEH STRAPEAR LOSSFAMB FAMSMOK FRSMOKER WSTRESS WSATISF SMOKINF SALPROM NEGCON POSCON ADINFLU INC
Probit Model Estimation Direction Coefficient P-Value of Influence j0.413 0.435 0.480 0.001 + 0.023 0.913 j0.037 0.885 0.027 0.888 j0.092 0.759 j0.592 0.179 j0.153 0.305 j0.343 0.064 j j0.327 0.036 j j0.053 0.804 j0.081 0.720 0.213 0.480 8.091 0.999 j0.155 0.438 0.086 0.618 0.227 0.294 j1.195 0.000 j 0.019 0.894 j0.339 0.112 j0.009 0.946 0.505 0.001 + j0.394 0.004 j 0.908 0.000 + 0.145 0.285 j0.331 0.012 j 0.301 0.089 + 0.416 0.003 + 0.402 0.008 + 0.197 0.148 j0.166 0.207 0.247 0.167 0.575 0.007 + j0.792 0.024 j 0.288 0.051 + 0.722 0.000 + 0.006ej01 0.000 +
Least Squares Estimates Direction Coefficient P-Value of Influence 1.368 0.000 + j0.228 0.053 j 0.035 0.827 j0.589 0.012 j 0.055 0.698 0.117 0.595 0.191 0.607 j0.036 0.764 0.095 0.529 j0.027 0.831 0.064 0.716 0.170 0.344 j0.021 0.920 j0.628 0.203 j0.179 0.270 0.028 0.855 j0.113 0.526 j0.320 0.176 0.027 0.818 j0.472 0.022 j j0.117 0.302 0.192 0.086 + j0.320 0.015 j 0.368 0.005 + 0.054 0.609 j0.188 0.073 j 0.219 0.079 + 0.377 0.006 + 0.578 0.000 + 0.137 0.197 j0.008 0.940 j0.411 0.001 j 0.232 0.074 + 0.176 0.284 0.310 0.007 + 0.044 0.724 1.003ej04 0.923
(Note: e + nn or e j nn means multiply by 10 to + or j nn power).
286
ELENI RAPTOU ET AL.
of the income elasticity of participation and the income elasticity of consumption and equals to 0.28. Concerning the psychosocial variables, the great majority of them are statistically significant in both smoking participation and conditional demand equation. More specifically, individuals receptive to intense family control ( FCONTROL ) are much more likely to smoke, and to smoke much more in average than individuals with lower extent of family control. In addition, individuals who are not attached to their families (LOSSFAMB) are much more likely to smoke as well, and to smoke much more in average than individuals who have a strong bond with their families. The adaptation of a healthy lifestyle ( HCONCERN ) is negatively associated with smoking behavior as it derives from both equations. Therefore, individuals who follow a healthy lifestyle and are interested in health issues are less likely to smoke and less likely to smoke on average than individuals who are less concerned with such matters. On the other hand, Table 4 shows that individuals who appear to have a risky behavior (RISKB ) are more vulnerable to smoking, and smoke much more in average than individuals who lead a quiet life. Concerning the stress for Bwhat other people think of me^ ( STRAPEAR ), it is found to affect considerably both smoking participation and cigarette consumption. Namely, individuals who are less anguished of the opinion that others have on them present a higher level of smoking prevalence and consumption intensity, compared to individuals who are more interested on what other people think of them. Family and friends’ smoking behavior ( FAMSMOK, FRSMOKER) are positively associated with individual’s smoking patterns. More specifically, individuals whose most of their friends and most of their family members are smokers, are much more likely to smoke and to smoke more in average than individuals whose social or family environment is against smoking. Regarding the SALPROM variable, it is found to be positively associated with smoking behavior in both equations. Hence, individuals who participate in competitions or other promotion events and own sales promotion items are more disposed to smoke and consume more cigarettes in average than individuals who do not take part in such events. With respect to SMOKINF variable, it seems that individuals who feel more receptive to smoking whenever they are offered cigarettes, smoke less in average, as most of the times they are light smokers or experimenters. On the other hand, cigarette advertising ( ADINFLU ) is positively associated with the decision to smoke but does not affect the quantity of cigarettes consumed. Therefore, individuals who are more receptive to cigarette advertising appear a higher prevalence of smoking compared to individuals who are not susceptible to cigarette advertising. Perceived positive consequences of smoking (POSCON ) are positively associated with smoking patterns. Thus, individuals who believe that smoking has positive consequences are more vulnerable to smoke and consume more cigarettes in average than individuals who think that smoking behavior does not result to advantages. On the other hand, perceived negative consequences of smoking ( NEGCON) are negatively associated with smoking participation but do not affect the quantity of cigarettes consumed. Discussion Given the results presented above, the most important determinants of smoking behavior are the psychosocial factors and the smoking restrictions enforced in the workplaces and educational establishments. According to consumers’ opinion, cigarette prices do not affect their smoking patterns. Recent studies, conducted in Greece by using
CIGARETTE DEMAND
287
conventional cigarette price measures, concluded that price elasticity of demand for cigarettes is inelastic, hence, a change in price causes a proportionately smaller change in quantity demanded [Stavrinos, 1987; Hondroyiannis and Papapetrou, 1997; Nikolaou and Velentzas, 2001; Michalopoulos and Demoussis, 2002; Dritsakis, 2003]. The total income elasticity of demand is inelastic (0.28), indicating that cigarettes are normal goods. Similar results have also been derived from most recent studies in Greece [Michalopoulos and Demoussis, 2002; Dritsakis, 2003]. Comparing the income elasticity of participation (0.21) and the income elasticity of consumption (0.07), it seems that income affects smoking behavior mostly in determining smoking propensity rather than smoking intensity. Concerning smoking restrictions in workplaces and educational establishments, the estimations of both equations applied (smoking participation equation and conditional demand equation) demonstrate that the enforcement of such regulation decreases smoking intensity. The results of the present work are in line with the estimates of other authors [Chaloupka, 1992; Chaloupka and Grossman, 1996; Chaloupka and Wechsler, 1997; Evans et al., 1999; Tauras and Chalpoupka, 1999], concluding that smoking restrictions in public places and private worksites result to decreasing smoking rates. With respect to the psychosocial factors, they seem to be the most important determinants of cigarette demand. Family and friends’ smoking behavior influence significantly individual’s smoking patterns. These results are very crucial for adolescents and young adults who are more receptive to messages received from their social and family environment. In addition, studies have shown that parental smoking affects significantly adolescent’s smoking initiation [Fleming et al., 2002; Den Exter Blokland et al., 2004]. On the other hand, Unger and her colleagues [2002] identified peer influence as one of the strongest determinants of adolescent smoking. Adolescents with smokers in their social network may be especially susceptible to smoking because peers may give them the impression that smoking is more normative and prevalent than it actually is [Unger et al., 2002]. Perceived negative consequences of smoking are negatively associated with smoking participation. Given that the majority of adolescents and adults agree that smoking is addictive and is harmful for health, Arnett [2000] concluded that adolescent and adult smokers were more likely than non-smokers to doubt that they will have to face health problems because of cigarette smoking. Therefore, smokers appear to keep optimistic bias regarding smoking risks and addiction compared to non-smokers. Perceived positive consequences of smoking affect substantially both smoking intensity and propensity, especially for youth. Johnson and his colleagues [2003] concluded that adolescents and young adults use cigarette smoking to Bmanage social situations, relieve feelings of anxiety and feel confident.^ This is in accordance with the estimations of this work where smokers were found to be more confident and less anxious of Bwhat other people think of them.^ Generally, literature supports that smoking initiation and experimentation are positively associated with the perceived positive consequences of smoking, especially for adolescents who believe that can gain group membership with smoking [Rugkasa et al., 2001]. In addition, young girls and women believe that smoking helps them to remain thin and not gain weight [Lowry et al., 2002]. Cigarette advertising and participation in cigarette sales promotion events influence smoking behavior and this is the case in other studies as well. Cigarette advertising and sales promotion events affect most of all adolescents and young adults although only young adults seem to increase their purchases due to advertising influence [Redmond, 1999; Sargent et al., 2000; MacFadyen et al., 2001]. In addition, receptivity to cigarette
288
ELENI RAPTOU ET AL.
sales promotion activities is an important factor in progressing from experimentation to established smoking for youth [Choi et al., 2002]. Intensive family control and lack of family bond were found to be positively associated with smoking demand. Family control and strong family bond are connected because individuals, who are close to their families, have developed a relationship of mutual support and trust and they are not receptive to intensive family control. A recent study argues that many adolescents and young adults use smoking to rebel against parental attempts to control their behavior [Johnson et al., 2003]. These results are more important for adolescents and young adults who are economically dependent on their parents and cannot react to intensive family supervision. The estimates for risky behavior conform to Juon’s study [2002], supporting that smokers appear extreme behaviors, such as drug use, more frequently than non-smokers. Smoking itself consists a form of risky behavior because its consequences on health are well known. Therefore, it is more likely individuals who Bhave adapted^ a risky habit such as cigarette smoking to have also adapted other risky habits or behaviors. On the other side, it is expected that healthy lifestyle is negatively associated with smoking behavior. Results presented and discussed above, strongly confirm that smoking restrictions in workplaces and educational establishments lead to decreasing smoking rates if they are enforced with accuracy. Furthermore, estimations prove that cigarette advertising and sales promotion affect smoking behavior, hence their total ban would probably result to decreasing smoking rates. Findings can contribute to policy makers and professionals in developing appropriate tobacco control regulations and informative anti-smoking programs. Moreover, as psychosocial factors affect substantially smoking patterns of youths, who have not yet build up sufficient addictive capital to make future cigarette consumption a serious consideration, this signals the importance of being seriously taken into consideration by educators and instructors. References Arnett, J. J. BBrief ReportVOptimistic Bias in Adolescent and Adult Smokers and Nonsmokers,^ Addictive Behaviors, 25, 4, 2000, pp. 625Y32. Chaloupka, F. J. BRational Addictive Behavior and Cigarette Smoking,^ Journal of Political Economy, 99, 4, 1991, pp. 722Y42. V. BClean Indoor Air Laws, Addiction and Cigarette Smoking,^ Applied Economics, 24, 2, 1992, pp. 193Y205. Chaloupka, F. J.; Grossman, M. BPrice, Tobacco Control Policies and Youth Smoking,^ NBER Working Paper No. 5740, 1996. Chaloupka, F. J.; Warner K. E. The Economics of Smoking, NBER Working Paper No. 7047, 1999. Chaloupka, F. J.; Wechsler, H. BPrice, Tobacco Control Policies and Smoking Among Young Adults,^ Journal of Health Economics, 1997, 16, pp. 359Y73. Choi, W. S.; Ahluwalia, J. S.; Harris, K. J.; Okuyemi, K. BProgression to Established SmokingVThe Influence of Tobacco Marketing,^ American Journal of Preventive Medicine, 22, 4, 2002, pp. 228Y33. Cragg, J. G. BSome Statistical Models for Limited Dependent Variables with Application to the Demand for Durable Goods,^ Econometrica, 39, 5, 1971, pp. 829 Y 44. Den Exter Blokland, E. A. W.; Engels, R. C. M. E.; Hale, W. W. III; Meeus, W.; Williemsen, M. C. BLifetime Parental Smoking History and Cessation and Early Adolescent Smoking Behavior,^ Preventive Medicine, 38, 2004, pp. 359Y68. Dritsakis, N. BForecasting Cigarette Consumption in Greece: An Empirical Investigation with Cointegration Analysis,^ Agricultural Economics Review, 4, 2, 2003, pp. 47Y56. Emery, S.; White, M. M.; Pierce, J. BDoes Cigarette Price Influence Adolescent Experimentation?,^ Journal of Health Economics, 20, 2001, pp. 261Y70.
CIGARETTE DEMAND
289
Eurostat Yearbook 2001. Health and Safety, Eurostat, 2001. Evans, W. N.; Farrelly, M. C.; Montgomery, E. BDo Workplace Smoking Bans Reduce Smoking?,^ The American Economic Review, 89, 4, 1999, pp. 728Y47. Farrelly, M. C; Bray, J. W.; Office on Smoking and Health. BResponse to Increases in Cigarette Prices by Race / Ethnicity, Income and Age GroupsVUnited States 1976Y1993,^ Morbidity and Mortality Weekly Report, 47, 29, 1998, pp. 605Y9. Fleming, C. B.; Hyoshin, K.; Harachi, T. W.; Catalano, R. F. BFamily Processes for Children in Early Elementary School as Predictors of Smoking Initiation,^ Journal of Adolescent Health, 20, 2002, pp. 184 Y 9. Grossman, M.; Coate, D.; Lewit, E. M.; Shakotko, R. A. Economic and Other Factors in Youth Smoking, Washington: National Science Foundation, 1983. Hersch, J. BGender, Income Levels and the Demand for Cigarettes,^ Journal of Risk and Uncertainty, 21, 2/3, 2000, pp. 263Y82. Hondroyiannis, G.; Papapetrou E. BCigarette Consumption in Greece: Empirical Evidence from Cointegration Analysis,^ Applied Economics Letters, 4, 1997, pp. 571Y 4. Johnson, J. L.; Bottorff, J. L.; Moffat, B.; Ratner, P. A.; Shoveller, P. A.; Lovato, C. Y. BTobacco Dependence: Adolescents’ Perspectives on the Need to Smoke,^ Social Science & Medicine, 56, 7, 2003 pp. 1481Y92. Juon, H.; Ensminger, M. A.; Sydnor, K. D. BA Longitudinal Study of Developmental Trajectories to Young Adult Cigarette Smoking,^ Drug and Alcohol Dependency, 66, 2002, pp. 303Y14. Lewit, E. M.; Coate, D. BThe Potential for Using Excise Taxes to Reduce Smoking,^ Journal of Health Economics, 1, 2, 1982, pp. 121Y 45. Lewit, E. M.; Coate, D.; Grossman, M. BThe Effects of Government Regulation on Teenage Smoking,^ Journal of Law and Economics, 24, 3, 1981, pp. 545Y69. Lowry, R.; Galuska, D. A.; Fulton, J. A.; Wechsler, H.; Kann, L. BWeight Management Goals, and Practices among U.S. High School Students: Association with Physical Activity, Diet and Smoking,^ Journal of Adolescent Health, 31, 2, 2002, pp. 133Y44. MacFadyen, L.; Hastings, G.; MacKintosh, A. BCross Sectional Study of Young People’s Awareness of and Involvement with Tobacco Marketing,^ BMJ, 322, 2001, pp. 513Y7. Michalopoulos, V.; Demoussis, M. BCigarette Demand in Greece: A Microeconomic Approach,^ Paper presented at the 6th Panhellenic Conference of Agricultural EconomicsVProceedings, Conference Proceedings, Stamoulis Publishing, 2002, pp. 345Y56. Nikolaou, A.; Velentzas, K. BEstimating the Demand for Cigarettes in Greece: An Error Correction Model,^ Agricultural Economics Review, 2, 1, 2001, pp. 22Y9. Pan American Health Organization. BNew Findings on the Health and Economic Impact of Tobacco Control Policies: Curbing the Epidemic: Governments and the Economics of Tobacco Control,^ May 31, 2000. Redmond, W. H. BEffects of Sales Promotion on Smoking among U.S. Ninth Graders,^ Preventive Medicine, 28, 1999, pp. 343Y50. Ross, H.; Chaloupka, F. J. BThe Effect of Cigarette Prices on Youth Smoking,^ Journal of Health Economics, 12, 2003, pp. 217Y30. Rugkasa, J.; Knox, B.; Sittlington, J.; Kennedy, O.; Treacy, M. P.; Santos Abaunza, P. BAnxious Adults vs. Cool Children: Children’s Views on Smoking and Addiction,^ Social Science & Medicine, 53, 2001, pp. 593Y602. Sargent, J. D.; Dalton, M.; Beach, M.; Bernhardt, A.; Heatherton, T.; Stevens, M. BEffect of Cigarette Promotion on Smoking Uptake among Adolescents,^ Preventive Medicine, 30, 2000, pp. 320Y7. Stavrinos V. G. BThe Effects of an Anti-Smoking Campaign on Cigarette Consumption: Empirical Evidence from Greece,^ Applied Economics, 19, 1987, pp. 323Y9. Tauras, J. A.; Chalpoupka, F. J. BPrice, Clean Indoor Air, and Cigarette Smoking: Evidence from the Longitudinal Data for Young Adults,^ NBER Working Paper No. 6937, 1999. Unger, J. B.; Rohrbach, L. BWhy Do Adolescents Overestimate Their Peers’ Smoking Prevalence? Correlates of Prevalence Estimates among California 8th-Grade Students,^ Journal of Youth and Adolescence, 31, 2, 2002, pp. 147Y53.
290
ELENI RAPTOU ET AL.
Unger, J. B.; Li Yan, M. D.; Shakib, S.; Rohrbach, L. N.; Chen, X.; Quian, G.; Chou, C.; Jianguo, S.; Azen, S.; Zheng, H.; Johnson, A. C. BPeer Influences and Access to Cigarettes as Correlates of Adolescent Smoking: A Cross-Cultural Comparison of Wuhan, China and California,^ Preventive Medicine, 34, 2002, pp. 476Y84. Wasserman, J.; Manning, W. G.; Newhouse, J. P.; Winkler, J. D. BThe Effects of Excise Taxes and Regulations on Cigarette Smoking,^ Journal of Health Economics, 10, 1, 1991, pp. 43Y64. Wilkins, N.; Yurekli, A.; Hu, T. BEconomic Analysis of Tobacco Demand, Tool 3. Demand Analysis,^ in World Bank, Economics of Tobacco Toolkit; Ayda Yurekli, Joy de Beyer (eds.), 2001. http://www1.worldbank.org/tobacco/toolkit.asp, retrieved at 20/3/2002. World Bank Report on Worldwide Tobacco Use (05/18-5). Action on Smoking and Health, Excerpts from Curbing the Epidemic: Governments and the Economics of Tobacco Control, The World Bank, Washington DC, May 1999, http://www.globalink.org/tobacco/wb.