The effect of comprehensive smoking bans in European Workplaces Federica Origo (Università di Bergamo)
Claudio Lucifora (Università Cattolica, IZA)
The effect of comprehensive smoking bans in European Workplaces Federica Origo (Università di Bergamo)
Claudio Lucifora (Università Cattolica, IZA)
Motivation Tobacco smoke is the prime cause of avoidable death in industrialized countries
quite recently, more concern for passive smoking (as dangerous as active one): in the EU-25 passive smoking is the prime cause of death for more than 79 thousands adults each year and almost 9% of them die for exposure to tobacco smoke at work.
According to the World Bank (2003), the optimal policy mix to fight tobacco smoke should include a wide set of measures:
- bans and restrictions on smoking in public places and workplaces;
- cigarette taxation;
- public information campaigns;
- bans on the advertising and promotion of tobacco products;
- health warnings on tobacco product packaging;
- treatment to help quitting.
In its 2007 Green Paper “Towards a Europe free from tobacco smoke: policy at the EU level”, the European Commission has further emphasized the role of comprehensive smoke-free regulation in reducing exposure to tobacco smoke, with positive effects on health of both active and passive smokers.
Motivation Following the EU recommendations, in the last five years most EU Members States have implemented such type of comprehensive smoking bans, albeit at different dates and with different degree of enforcement.
The first EU-15 country moving in this direction was Ireland (March 2004), immediately followed by Italy (January 2005) and Sweden (June 2005). Most of the other countries did the same between 2006 and 2008, with the remaining ones planned to do so by the beginning of 2009.
Despite of the intense public debate and the high expectations following these reforms, very limited (and all country-specific) research has been carried out in order to evaluate the impact of these comprehensive smoking bans on health. Even less attention has been paid to their effects within the workplace.
Aim of this paper In this paper we try to fill the gap by using a quasi-experimental approach to evaluate the impact of national comprehensive smoking bans on perceived workers’ health for a large number of European countries on the basis of comparable micro-data.
Definition of comprehensive smoking bans They ban smoking in all the workplaces (both public and private)…
…and in all public places (including bars and restaurants)
Definitions Literature review Setting & facts Data & strategy Results Conclusions
Literature review Privately initiated workplace smoking bans significantly reduce both smoking prevalence and daily cigarette consumption among workers at the workplace (Evans et al 1999; Fitchtemberg and Glantz 2002; Levy and Friend 2003)
Public smoke-free policies reduce cigarette demand both for the young and the adult, especially in the case of males (Waaserman et al. 1991; Chaloupka 1992; Keeler et al. 1993; Yurkely and Zhang 2004; Tauras 2005).
Local smoking bans, other than reducing cigarette consumption in the population, push firms to introduce such bans also at the workplace, thus reducing exposure to tobacco smoke among workers (Moskovitz et al. 2000; Stephens et al. 1997; Carpenter 2007).
However, such policies are likely to produce displacement/ multiplier effects (Adda and Cornaglia 2006; Cutler and Glaeser 2007) and other unintended effects (Adams and Cotti 2003)
Furthermore, according to medical literature their effect on health in the short run are ambiguous (Sargent et al. 2004; Juster et al. 2007; Shetty et al. 2009).
Definitions Literature review Setting & facts Data & strategy Results Conclusions
Our approach
We focus on the effects of a specific type of public smoking control policy -- the so called “comprehensive” smoke-free law -- on workers’ health within workplaces.
We use comparable micro-data for a large number of (European) countries, and we study the effect of smoking control policies not only on exposure to smoke, but also on direct measures of (perceived) workers health (such as the presence of respiratory problems).
We exploit the timing and design of smoking control policies, as implemented by various countries, to assess the causal effect of comprehensive smoking regulations on workers perceived health using a quasi-experimental approach (i.e. a ‘Diff-in-Diff’ estimator). Definitions Literature review Setting & facts Data & strategy Results Conclusions
How to measure (and compare) tobacco control policies in Europe? To compare and quantify the implementation of tobacco control policies across European countries, a specific “smoking scale” (Tobacco Control Scale, TCS) has been created by a group of international experts (Jossens and Raw, 2006).
The scale is aimed at measure the six types of intervention recommended by the World Bank (2003).
For each policy, a score was assigned by national experts according to both quantitative data and subjective evaluation based on a common questionnaire.
In the case of smoking bans, the maximum score was equal to 22 (the maximum score for the overall TCS was 100) and it was the result of the subjective evaluation on three different aspects:
bans in cafes and restaurants
bans in other workplaces
bans on public transport and in other public places (such as educational, health, government and cultural places Definitions Literature review Setting & facts Data & strategy Results Conclusions
Comprehensive smoking bans & the TCS Definitions Literature review Setting & facts Data & strategy Results Conclusions
Smoking prevalence Definitions Literature review Setting & facts Data & strategy Results Conclusions
Incidence of passive smoking at work Definitions Literature review Setting & facts Data & strategy Results Conclusions
Data Empirical analysis based on different waves of the European Working Conditions Survey (EWCS) by the European Foundation for the Improvement of Living and Working Conditions
Representative sample of workers in all the EU Member States, with the aim to investigate the main characteristics and evolution of working conditions across Europe.
detailed information on a wide range of work-related issues, including work organisation, risk exposure, job satisfaction and work-related health.
So far, data available for 1990, 1995, 2000 and 2005
given institutional setting, core of our empirical analysis based on last two waves; other waves used for robustness check
Given the identification strategy, we matched this data with several (relevant) country-level controls (log population, GDP per capita, U rate, life expectancy at birth, % smokers, total taxes on cigarettes, indicator of OSH regulation)
Definitions Literature review Setting & facts Data & strategy Results Conclusions
Empirical strategy Diff-in-diff estimator:
Diff-in-diff estimator with heterogeneous effects:
Definitions Literature review Setting & facts Data & strategy Results Conclusions
Variables list Definitions Literature review Setting & facts Data & strategy Results Conclusions
Effect on probability of exposure to smoke and fumes Definitions Literature review Setting & facts Data & strategy Results Conclusions
Effect on probability of work-related respiratory problems Definitions Literature review Setting & facts Data & strategy Results Conclusions
Robustness check Change in the control group (by excluding workers of one country at a time)
Use a more general dependent variable, not directly affected by smoking bans
dummy equal to one for workers declaring that their jobs affected (negatively) their health
“Placebo” test with 1995 and 2000 EWCS waves Definitions Literature review Setting & facts Data & strategy Results Conclusions
Robustness check 1 NOTE: Estimates obtained by excluding workers of one control country at a time. The excluded country is reported on the X axis; countries are ranked in ascending order according to their overall score of the Tobacco Control Scale in 2005. 95% confidence intervals are also reported. Definitions Literature review Setting & facts Data & strategy Results Conclusions
Robustness check 2: effect on a more general work-related health measure Definitions Literature review Setting & facts Data & strategy Results Conclusions
Robustness check 3: placebo test Treatment has been fictitiously assigned to the treated countries in 2000. Diff-in-diff is the estimated difference between treated and controls in the 1995-2000 change. Definitions Literature review Setting & facts Data & strategy Results Conclusions
Comments