Elsevier

Journal of Policy Modeling

Volume 35, Issue 1, January–February 2013, Pages 1-15
Journal of Policy Modeling

Tobacco spending patterns and their health-related implications in Turkey

https://doi.org/10.1016/j.jpolmod.2012.09.004Get rights and content

Abstract

A variety of censored dependent variable models are used to identify the impacts of social and economic variables on the decision to consume and the amount of expenditure on tobacco in Turkey. Using data from the Turkish Household Expenditure Survey, the odds of smoking and spending level on tobacco products are investigated jointly. Statistical tests identify the inverse hyperbolic sine double-hurdle model as the preferred model. Socio-demographic factors are found to have differentiated effects on participation and level decisions. Findings can inform deliberations of policy measures to curtail tobacco consumption and reduce the costs of the public healthcare system.

Introduction

The adverse social, economic, and health effects of tobacco consumption have been considered a hazard that developed countries take into account in their fiscal agenda (Zhao & Harris, 2004). Tobacco causes more than 100,000 deaths each year representing a quarter of all deaths in Turkey, and the rate is expected to reach 240,000 by 2030 (Bilir, Çakır, Daglı, Ergüder, & Önder, 2009). The death rate is twice the death rate of mothers, infants, and children combined and 37 times greater than the annual number of 3000 traffic fatalities in the country (TAPDK, 2002, WHO, 2002). During the 1980s and 1990s, liberalization of the domestic market to foreign-brand cigarettes increased cigarette smoking by 80% implying that one third of the population became smokers. Soon after, the increased number of smokers brought a tremendous increase in lung cancer cases which makes Turkey the sixth largest spender on cancer treatment among all European countries (THSD, 2010).

Daily consumption is estimated to be 15 million packs of cigarettes, or 76.1 packs per person per year (TAPDK, 2010). Smokers in Turkey annually spend almost as much for tobacco products (approximately 18 billion TL, or about 12.8 billion USD at the time of the study) as the government allocates to investment in hospitals, schools, roads, electricity, and water supply, among others (approximately 22 billion TL are allocated for such projects in 2010). Annual damage to the economy due to smoking is estimated to have reached 15 billion TL, which translates into 2.72 billion TL in annual GDP loss caused by all tobacco-related diseases (TAPDK, 2002). The contribution of tobacco to the economy is, on the other hand, remarkable. Turkey is a large tobacco-producing and exporting country. Before 1980, Turkey ranked fifth in the world with 4–5% of the world tobacco production share. This share has declined to 1.7% since then and the country now ranks seventh in world tobacco production and sixth in export (FAO, 2010).

Turkey became a party to the WHO Framework Convention for Tobacco Control (FCTC) in April 2004 and ratified the treaty in December of the same year. Despite fierce resistance from the tobacco industry and the lack of public support, Turkey enacted several public smoking bans in the last few years. The law prohibited smoking in public places by making all public buildings, public transportation, taxis, and the inside and outside of all schools, health care facilities, and sport facilities including soccer stadiums 100% smoke-free in April 2008 (Yürekli et al., 2010). The ban was expanded on July 19, 2009 to include hospitability sectors such as restaurants, coffeehouses, cafeterias and bars and to ban sponsorship by tobacco companies. The law also required that all public and private television channels broadcast at least 90 min per month of educational programs “on the health hazards of smoking”. With that legislation, Turkey became one of six countries, together with the United Kingdom, Ireland, New Zealand, Uruguay, and Bermuda, in the world with a powerful tobacco control law that marks the beginning of “100 percent smoke-free air”. Moreover, in 2010, the government regulated a “health tax” as earmarking taxes, which authorized the Social Security Institution (SSI) to receive 1–2% of the revenue from each pack of cigarettes or bottle of alcohol sold to support the treatment of nicotine and alcohol dependence and prevention programs such as lung and throat cancers.

With the growing public awareness of the harmful effects of cigarette smoking and the emergence of legislation to curb the consumption of tobacco in Turkey, there is a need to fill the gap in studies that examine the determinants of both the likelihood of purchasing and the amount spent on tobacco products in the country. Although many studies have addressed this issue for other countries, there exists a dearth of such analysis in Turkey to inform deliberations of policies to curb the use of tobacco. This study attempts to fill this void by applying a variety of generalized censored dependent variable models (the inverse hyperbolic sine sample selection (IHS-SSM), sample selection (SSM), IHS-double hurdle (IHS-DH), and double-hurdle (DH) models) to identify the impact of social and economic variables on the decision to spend and the amount of expenditures on tobacco products.

The next section presents the IHS-DH model. Section 3 briefly describes the data and variables. Results are discussed in Section 4. The final section presents conclusions and implications for future research.

Section snippets

Econometric method

Microdata typically feature zero observations in the expenditure variables. Statistical analyses not accounting for such “censoring” produce inconsistent empirical estimates. Censoring in the dependent variables can be accommodated with a limited dependent variable model(s) such as the sample selection model (SSM) and double-hurdle (DH) model. Existing applications of these models are predominantly based on the bivariate normality distribution for the error terms. However, limited dependent

Data

Data are drawn from the 2008 Turkish Household Expenditure Survey (THES) conducted by the Turkish Statistics Institute. The initial sample contains 8549 observations obtained between 1 January and 31 December, 2008. Households were replaced monthly by one with similar characteristics to minimize seasonal influences on consumption. The final sample includes 8069 households after deletion of incomplete responses and a few outliers.

Table 1 presents definitions and sample statistics of variables

Results and discussion

Four model specification tests are carried out and the results are presented in Table 2. The first set of tests relates to statistical significance of the transformation parameter in IHS-DH and IHS-SSM to justify the transformation. The second set of tests verifies the preferred model among all three competing specifications based on Vuong's (1989) standard normal statistic for non-nested tests. A third test is applied to the preferred models whether the tobacco spending decision and its

Conclusions and implications

This study examines determinants of household spending on tobacco using generalized forms of sample selection and double-hurdle models that inherent flexible parameterization and distributional assumptions and a large and recent Turkish Household Expenditure Survey dataset. Knowledge of the relevant factors is essential to reducing family consumption of tobacco. The current study provides insights for policy makers on important determinants of household tobacco participation decision and

References (33)

  • J.R. Blaylock et al.

    U. S. cigarette consumption: The case of low income women

    American Journal of Agricultural Economics

    (1992)
  • J.B. Burbidge et al.

    Alternative transformations to handle extreme values of the dependent variable

    Journal of American Statistical Association

    (1988)
  • S. Chatterjee et al.

    Regression analysis by example

    (2006)
  • FAO Corporate Document Repository

    Tobacco in Turkey

    (2010)
  • J. Garcia et al.

    Alternative approaches to modelling zero expenditure: An application to Spanish demand for tobacco

    Oxford Bulletin of Economics and Statistics

    (1996)
  • R.K. Goel et al.

    The effectiveness of anti-smoking legislation: A review

    Journal of Economic Surveys

    (2006)
  • Cited by (0)

    View full text