Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, ECONOMICS AND FINANCE ( (c) Oxford University Press USA, 2020. All Rights Reserved. Personal use only; commercial use is strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 17 January 2020

The Economics of Smoking Prevention

Summary and Keywords

Smoking prevention has been a key component of health policy in developed nations for over half a century. Public policies to reduce the physical harm attributed to cigarette smoking, both externally and to the smoker, include cigarette taxation, smoking bans, and anti-smoking campaigns, among other publicly conceived strategies to reduce smoking initiation among the young and increase smoking cessation among current smokers. Despite the policy intensity of the past two decades, there remains debate regarding whether, and to what extent, the observed reductions in smoking are due to such policies. Indeed, while smoking rates in developed countries have fallen substantially over the past half century, it is difficult to separate secular trends toward greater investment in health from actual policy impacts. In other words, smoking rates might have declined in the absence of these anti-smoking policies, consistent with trends toward other healthy behaviors. These trends also may reflect longer-run responses to policies enacted many years ago, which also poses challenges for identification of causal policy effects. While smoking rates fell dramatically over this period, the gradient in smoking prevalence has become tilted toward lower socioeconomic status (SES) individuals. That is, cigarette smoking exhibited a relatively flat SES gradient 50 years ago, but today that gradient is much steeper: relatively less-educated and lower-income individuals are many times more likely to be cigarette smokers than their more highly educated and higher-income counterparts. Over time, consumers also have become less price-responsive, which has rendered cigarette taxation a less effective policy tool with which to reduce smoking. The emergence of tax avoidance strategies such as casual cigarette smuggling (e.g., cross-tax border purchasing) and purchasing from tax-free outlets (e.g., Native reservations in Canada and the United States) have likely contributed to reduced price sensitivity. Such behaviors have been of particular interest in the last decade as cigarette taxation has roughly doubled cigarette prices in many developed nations, creating often large incentives to avoid taxation for those who continue to smoke. Perhaps due to the perception that traditional policy has been ineffective, recent anti-smoking policy has focused more on the direct regulation of cigarettes and smoking behavior. The main non-price-based policy has been the rise of smoke-free air laws, which restrict smoking behavior in workplaces, restaurants, and bars. These regulations can reduce smoking prevalence and exposure to secondhand smoke among nonsmokers. However, they may also shift the location of smoking in ways that increase secondhand smoke exposure, particularly among children. Other non-tax regulations focus on the packaging (e.g., the movement towards plain packaging), advertising, and product attributes of cigarettes (e.g., nicotine content, cigarette flavor, etc.), and most are attempts to reduce smoking by making it less desirable to the actual or potential smoker. Perhaps not surprisingly, research in the economics of smoking prevention has followed these policy developments, though strong interest remains in both the evaluation of price- and non-price policies as well as any offsetting responses among smokers that may undermine the effectiveness of these regulations. While the past two decades have provided fertile ground for research in the economics of smoking, we expect this to continue, as governments search for more innovative and effective ways to reduce smoking.

Keywords: cigarette smoking, tobacco control, prevention, health behaviors, taxation, regulation, health economics


Smoking prevention has been a key component of health policy in the developed world since the reports by the UK Royal College of Physicians and the US Surgeon General from roughly half a century ago. Public policies intended to reduce the harm from smoking include cigarette taxation, smoke-free air laws that prohibit smoking in areas ranging from bars and restaurants to workplaces, anti-smoking informational campaigns, and regulations designed to make tobacco products less desirable. In addition, this period has witnessed a dramatic increase in the availability of products that are intended to encourage quitting (e.g., nicotine replacement therapies) and those that might encourage quitting (e.g., electronic cigarettes). Despite recent policy intensity, there remains debate regarding whether, and to what extent, the observed reductions in smoking prevalence over time are due to such policies, or instead are due to other, unrelated factors such as a trend toward greater demand for health.

An extensive body of research examines the impact of anti-smoking regulations on smoking behavior. The research varies with respect to the data used, the time period studied, and the credibility of the research design. The sheer volume of estimates makes it difficult for policymakers (and often researchers) to glean consistent lessons from these studies. This article reviews evidence on the two most developed areas of the economics of smoking literature: cigarette excise taxes and smoke-free air laws. It then discusses emerging issues in the broader literature, reviewing what has been done and speculating on likely future directions. The goals are to highlight areas of consensus in findings among the most rigorous studies as well as to discuss issues on which there still remains disagreement in the research literature. Due to space constraints, the focus is on studies from developed nations. Moreover, several relevant and interesting issues in the economics of smoking prevention are necessarily omitted. Rather than provide a cursory overview of a wider range of topics, a smaller but prominent set of anti-smoking policies are discussed in more depth. It is left to future review articles to address these other important areas of research.

In order to describe briefly the context in which research in the economics of smoking takes place, the Background section discusses issues common to nearly all of the studies reviewed in this article. Next, we examine the cigarette excise tax literature, focusing separately on how taxes affect youth versus adult smoking as well as tax avoidance behaviors. After that, we review research on smoke-free air laws in terms of effects on smoking behavior, secondhand smoke exposure, and health effects. We then discuss emerging methodological issues in the economics of smoking prevention literature, such as identification and research design, inference, and smoking measurement. The last section summarizes and provides some thoughts on the future of smoking-related research in economics.


In general, there are two types of cigarette demand studies—those which use aggregate-level data (typically at the state or country level) and those which use individual-level data. The article focuses on the latter, since this type of data is now quite prevalent throughout the world and dominates the current literature. The availability of longitudinal data has grown tremendously over the past quarter of a century. Early studies that used individual-level smoking data were limited largely to single-year cross-sectional variation in policies of interest. Such studies are susceptible to omitted variables bias that occurs when there are unobserved factors that are correlated with the policy (e.g., cigarette tax rate) and that also exert influence on the outcome in question (e.g., smoking prevalence). The greatest concern in the economics of smoking literature is the existence of unobserved preferences or attributes that drive both anti-smoking policy and smoking behavior. These unobserved variables can typically be characterized as attitudes towards smoking, or, more generally, anti-smoking sentiment. They are relevant because anti-smoking policies are often set by governmental units below a nation’s federal government (e.g., US states dictate the majority of anti-smoking policies), and attitudes towards smoking can vary considerably by location. Because such unobserved sentiment is almost certainly correlated both with the smoking policy one wants to evaluate and smoking behavior, it is impossible to obtain unbiased policy impact estimates without accounting for these factors.

The literature has taken two general empirical approaches to deal with the existence of these unobserved differences. The first is to include controls that serve as a proxy for anti-smoking sentiment. Earlier studies include place-specific characteristics that are correlated with collective preferences towards smoking (cf., Wasserman et al., 1991), while more recent work has attempted to more accurately study anti-smoking sentiment using responses to attitudinal questions about smokers and smoking (cf., DeCicca et al., 2008). The main concern with respect to this approach is that the proxies for anti-smoking sentiment are weak, meaning they will not fully overcome the omitted-variables bias they were designed to address.

The place-based approach typically uses the growing availability of repeated observations on those within certain geographic areas (e.g., repeated cross-sectional or individual panel data) that allow researchers to include both geography and time-fixed effects. Such fixed effects account for unchanging characteristics of each geographic area along both observed and unobserved dimensions. For example, state-fixed effects control for time-invariant heterogeneity in anti-smoking sentiment at the state level. The value of this approach is that one does not have to actually measure anti-smoking sentiment but rather can control for it by eliminating all cross-sectional variation in policies and outcomes as long as anti-smoking sentiment is unchanged over the period of analysis. An added concern with using panel data in this setting is that anti-smoking sentiment is rising over time across geographic regions, and this is correlated with more stringent anti-smoking regulations. Year-fixed effects account for such aggregate changes over time in individuals’ attitudes towards smoking, but only if they are common to all geographic units.

Due to the ease of implementation and the ability to account for difficult-to-observe cross-sectional and time-series heterogeneity, the “two-way fixed effects” model has become the most prominent approach among researchers. In practice, there are many issues associated with this approach, however. Chief among them is the fact that the time invariance assumption is unlikely to be met. In particular, it needs to be the case that any changes in anti-smoking sentiment are uncorrelated with smoking policy changes. This is much less likely to hold over long periods of time, because place-specific sentiment toward smoking can change dramatically and can drive anti-smoking policy. Issues related to this model will be discussed in more detail when we eventually discuss methodological issues.

Another feature that nearly all economics of smoking studies share is the use of self-reported data on smoking behavior. It is well known that self-reported data, particularly related to outcomes that are stigmatized, are subject to measurement error. In linear models, classical measurement error in the dependent variable implies inflated standard errors and a greater likelihood of under-rejection.1 Since over-rejection of the null hypothesis of no policy effect on smoking is of greatest concern, measurement error is often viewed as secondary, particularly relative to the identification concerns outlined above. However, it is not clear whether the measurement error inherent in self-reported smoking behavior is classical in nature or more systematic. Indeed, when the dependent variable is a count variable (e.g., the number of cigarettes smoked per week) or is binary (e.g., whether one is a smoker), the measurement error cannot be classical. In such a setting, using self-reported data may result in biased estimates of policy impacts, not just a loss in statistical precision (Hausman, Abrevaya, & Scott-Morton, 1998; Kenkel, Lillard, & Mathios, 2004; Meyer & Mittag, 2014). Nonclassical errors seem particularly possible, as the stigma associated with cigarette smoking has increased tremendously over the past two decades; individuals in areas with more stringent anti-smoking policies may be more likely to misreport their smoking status and cigarette consumption in directions that indicate less smoking. Thus, the measurement error tends to occur in one direction and may be differentially large in areas with more aggressive anti-smoking policies. Newer data sources with more objective measures of cigarette consumption offer an opportunity to learn something useful about the extent to which measurement error in self-reported data produces biased estimates. There currently are few research studies that use such data, however; understanding the role of measurement error in driving anti-smoking policy estimates is a fruitful area for future research.

The Impact of Cigarette Excise Taxation on Smoking and Other Outcomes

This section reviews existing evidence on the impact of cigarette excise taxation on smoking behavior. It focuses first on youth and adult smoking and then discusses the growing tax avoidance literature. As discussed in the Background section, cigarette taxes tend to vary by location and time, which provides an opportunity to exploit within-place variation over time to estimate policy effects. Generally speaking, there are two dimensions to smoking: the extensive margin (i.e., smoking participation) and the intensive margin (i.e., the number of cigarettes smoked conditional on smoking). Historically, researchers have examined smoking participation and smoking intensity (or conditional demand) separately, often via “two-part” models that assume the independence of these two outcomes. The present discussion of smoking behavior primarily reviews the smoking participation literature, since reducing initiation and increasing cessation tend to be the focus among public health officials and policymakers. The literature on the impact of cigarette excise taxes on smoking is dominated by US work, due principally to data availability, substantial spatial variation in taxes, and the large increases in cigarette taxes implemented in the past two decades. The focus is therefore predominantly on US evidence, though several non-US studies are cited.

Cigarette Taxes and Smoking

Youth Smoking

The impact of cigarette taxation on youth smoking behavior has garnered relatively more interest than adult smoking over the past quarter-century. Perhaps the primary reason for this attention is that most lifetime smokers report starting regular smoking as teenagers (Chaloupka & Warner, 2000; Glied, 2002). The underlying logic implies that if policies can deter youth smoking, they will prevent lifetime smoking. Although limited causal evidence exists to support this hypothesis, it has intuitive appeal (Auld, 2005; Glied, 2002; Glied, 2003; Breslau & Peterson, 1996). Researchers have posited that youth should be more price-sensitive to cigarette taxation than their adult counterparts for two primary reasons (Chaloupka & Warner, 2000). First, the addictive nature of cigarettes implies that established adult smokers will be less price-sensitive than youth, who are presumably less addicted because they have consumed less nicotine in their lifetimes than their adult counterparts (or none at all). Second, youth are thought to have fewer monetary resources available, so that the income effect associated with any given tax-induced price increase will tend to be relatively larger than for adults. Conversely, youth may be less price-sensitive because they are typically not allowed to purchase cigarettes legally until they reach a certain age and thus must obtain their cigarettes through illegal channels.2

The earliest youth smoking studies primarily used cross-sectional data. As discussed in the Background section, such studies are particularly vulnerable to overestimating price effects because they confound causal policy impacts with anti-smoking sentiment. These early studies of youth smoking typically found price participation elasticity estimates near unity, implying that a 1% increase in cigarette prices would decrease the probability of youth smoking by 1% (cf., Lewit, Coate, & Grossman, 1981; Lewit & Coate, 1982; Chaloupka & Grossman, 1996; Chaloupka & Wechsler, 1997; Chaloupka & Pacula, 1999; Harris & Chan, 1999; Gruber & Zinman, 2001).3

More recent studies using longitudinal data on smoking behavior tend to find much smaller price participation elasticities for youths (cf., Hansen, Sabia, & Rees, 2017; Lillard, Molloy, & Sfekas, 2013; Nonnemaker & Farrelly, 2011; Sen, Ariizumi, & Driambe, 2010; Carpenter & Cook, 2008; DeCicca, Kenkel, & Liu, 2008; DeCicca, Kenkel, Liu, Shin, & Lim, 2008; Sloan & Trogdon, 2004; DeCicca, Kenkel, & Mathios, 2002; Dee, 1999). Many of these studies do not even find statistically significant relationships between cigarette prices/taxes and smoking participation. Even those that do find negative effects, however, indicate that the price-responsiveness of youth smoking participation is much smaller than suggested by earlier studies. For example, Carpenter and Cook (2008) find a statistically significant negative relationship between cigarette taxes and smoking initiation in the Youth Risk Behavior Surveillance System (YRBSS) data, but the implied price elasticities range between −0.25 and −0.50. Likewise, Lillard, Molloy, and Sfekas (2013) present participation elasticities ranging from −0.03 to −0.23, depending on the age of the analysis sample in question. A recent study by Hansen, Sabia, and Rees (2017) suggests that even the modest price elasticities found in Carpenter and Cook (2008) are growing smaller in absolute value over time: adding four more waves of YRBSS data from 2007–2013 significantly weakens the price sensitivity of youths to taxes. Indeed, in the 2007–2013 data, they find no evidence that youth smoking participation is affected by cigarette tax changes.

Taken together, this research indicates that youth smoking is not very sensitive to tax changes and that earlier cross-sectional work likely overstated any such effects. This body of work calls into question the efficacy of using cigarette taxes to reduce youth smoking.

Adult Smoking

Unlike the youth smoking literature, where some debate remains, it is well documented that adult smoking participation is at most modestly negatively affected by price. This is true for early studies of the price responsiveness of adult smoking behavior (cf., Lewitt & Coate, 1982; Mullahy, 1985; Jones, 1989; Wasserman et al., 1991; Evans & Farrelly, 1998; Jimenez-Martin et al., 1998; Labeaga, 1999) as well as later ones (cf., Farrelly et al., 2001; Gruber, Sen, & Stabile, 2003; Tauras, 2006; Chung et al., 2007; DeCicca & McLeod, 2008; Lovenheim, 2008; Aristei & Pieroni, 2009; Gospodinov & Irvine, 2009; Harding, Leibtag, & Lovenheim, 2012; Callison & Kaestner, 2014; Maclean, Sikora-Kessler, & Kenkel, 2016; Nesson, 2017). Indeed, given their timing, these early adult smoking studies formed the basis for the “consensus” elasticity estimates of −0.4 as cited by Chaloupka and Warner (2000) in their Handbook of Health Economics chapter. This range represents a total price elasticity of cigarette consumption, where “total” refers to the total consumption response that includes both the intensive and extensive margins. While it is not clear that the sum of these two conceptually different elasticities can be meaningfully interpreted, it is a widely reported metric in the literature. As the focus here is on smoking participation, it is notable that Chaloupka and Warner (2000) state that the range of consensus estimates is split evenly between the price participation elasticity and the conditional demand elasticity. This implies a price participation elasticity of −0.2, suggesting that adult cigarette smoking participation is at most modestly sensitive to price.4 This lack of price sensitivity may be due to the availability of lower-taxed alternatives, discussed in the “Cigarette Tax Avoidance” subsection.

While the consensus is that adult smoking participation is not very price responsive, recent work by DeCicca and Kenkel (2015) suggests that even the range of consensus estimates put forth in Chaloupka and Warner (2000) may be too large. DeCicca and Kenkel (2015) implement a dynamic population model, similar to that implemented by Warner and Mendez (2012), which relies on projecting smoking rates based on US population demographics from 1995 to 2010. This time period is used because it was characterized by numerous and often large cigarette tax increases. They show that changing population characteristics alone predict nearly all of the observed downward trend in adult smoking participation. When their model is extended to include the implied impact of price at levels consistent with the consensus participation elasticity estimate, however, they substantially underpredict adult smoking prevalence. This finding implies that the consensus estimate is too large in magnitude. For example, adding the impact of price via a price participation elasticity of −0.2 to the model suggests that the actual 2010 smoking rate should be about 10% lower than observed. Moreover, the model still underpredicts the true 2010 smoking rate at even a price participation elasticity of −0.1, but overpredicts it at zero. As a result, these simulation results suggest the true elasticity lies between zero and −0.1, which is considerably smaller than the corresponding consensus estimate.

The economics literature on adult smoking behavior generally treats smoking decisions as being made contemporaneously with cigarette taxes. This modeling assumption ignores potential dynamic effects of taxation driven by addiction. The earliest work in the economics of addiction focused on “myopic addiction models,” where consumption of an addictive good was seen as depending only on past consumption of that good (cf., Haavelmo, 1944; Houtakker & Taylor, 1970; Pollak, 1978). This research then gave rise to “fully rational addiction” models that allow not just past consumption decisions but future ones to affect current consumption decisions (Becker & Murphy, 1988). In particular, these authors recognized that the decision to smoke is dependent not only on past consumption or contemporaneous policy variables but also upon past and future realizations of these variables. Empirically, rational addiction models have proved difficult to estimate. One approach is to use past and future prices as instruments for past and future consumption (Becker, Grossman, & Murphy, 1991). Another approach is to estimate reduced form versions of the rational addiction specification, where past and future prices replace consumption levels. For both approaches, the high degree of serial correlation in cigarette price, which is driven primarily by changes in cigarette taxes, makes credible estimation of these models difficult, but reinforces the idea that longer-run price impacts may be larger in absolute value than short-run responses. This is particularly important in the context of the small contemporaneous effects that characterize the literature. Indeed, while empirically challenging, understanding the longer-run implications of cigarette taxation is a laudable future goal. Work by Auld and Grootendorst (2004) and, more recently, by Laporte, Dass, and Ferguson (2017) further elaborates on the difficulties of empirically estimating rational addiction models.

Finally, another line of work following Becker and Murphy (1988) focused on nonstandard preferences in modeling smoking behavior. Prominent examples include Gruber and Koszegi’s (2001) model of addiction with quasi-hyperbolic discounting/time inconsistency and Bernheim and Rangel’s (2004) model of cue-triggered addiction. Gruber and Koszegi (2001) estimate their model empirically and find that individuals are forward-looking with respect to cigarette price in a way that is indistinguishable from the rational addiction model. However, each model has starkly different normative implications. In the standard rational addiction model, there is no scope for government policy to improve the individual addict’s welfare, although a policy like cigarette taxation may be justified by externalities. In the model with hyperbolic discounting, the smoker’s time-inconsistent choices impose costs on his or her “future self.” Optimal cigarette taxation then reflects not only the harm addicts impose on others, but also the harm addicts impose on themselves. In the model of cue-triggered addiction, the optimal tax on the addictive good can even be negative (i.e., under some conditions it can be optimal to subsidize addictive consumption).

Cigarette Tax Avoidance

Although differences in cigarette taxes across European countries also create opportunities for tax avoidance, the literature on cigarette tax avoidance tends to be dominated by work using US data. Indeed, the large variation in cigarette taxes across states combined with the availability of tax-free cigarettes on Native American reservations and over the Internet provides much opportunity for cigarette tax avoidance. The effective doubling of state cigarette excise taxes in recent decades has not been evenly spread across states. To give an extreme example, Washington, DC, has a per-pack tax of $2.92, while neighboring Virginia has a per-pack tax of $0.30. Similarly, the tax difference between Illinois and Kentucky is $1.38 per pack. It is reasonable to expect those in Washington or Illinois who live close to these lower-tax borders to purchase cigarettes in the nearby cheaper locality. Those who live farther away could until recently purchase tax-free cigarettes online, the source of which was often Native American reservations. While economists have long recognized that the decentralized, state-by-state nature of cigarette taxation in United States creates opportunities for cross-state tax avoidance, the growth in empirical work on this topic was spurred by the rise in tax avoidance opportunities driven by recent tax hikes combined with a general perception that higher cigarette taxes were not reducing smoking to the extent implied by early econometric evidence. Growing tax avoidance is one plausible explanation for this pattern of results.

Related to this phenomenon, a sizeable literature examines the extent of cigarette tax avoidance and how such behavior alters the effectiveness of cigarette taxes in reducing smoking. The most prominent avoidance behavior examined is “casual cigarette smuggling,” which occurs when individuals travel across borders or go online to purchase cigarettes for personal consumption.5 Researchers have used various methods to measure cross-state tax avoidance. Some studies use indirect methods to infer the extent of tax avoidance: comparing changes in reported consumption versus administrative state sales (Coats, 1995; Stehr, 2005) and examining how tax responsiveness changes with home-border state tax differentials interacted with the distance to the border (Lakhdar, Vaillant, & Wolff, 2016; Chiou & Muehlegger, 2014; Lovenheim, 2008; Gruber, Sen, & Stabile, 2003). Other studies exploit data that provide direct measures of tax avoidance, including self-reported cigarette purchases from lower-tax jurisdictions (DeCicca, Kenkel, & Liu, 2015; Chiou & Muehlegger, 2008; DeCicca, Kenkel, & Liu, 2013a; DeCicca, Kenkel, & Liu, 2013b) and scanner data that includes the geographic location of the consumer and the purchase (Harding, Leibtag, & Lovenheim, 2012). Another approach collects packs discarded as litter and counts home versus border tax stamps on littered cigarette packs differentially by the distance to a lower-tax border (Chernick & Merriman, 2013; Merriman, 2010). The different approaches have different weaknesses. For example, in self-reported data people might understate the extent to which they avoid taxes, and in the littered-pack data it is unclear how the behavior of smokers who discard packs as litter compares to the general population of smokers. Put together, however, these studies universally find evidence of high levels of tax avoidance that significantly reduce the responsiveness of consumers to tax increases. This is especially the case among those living close to lower tax borders. Goolsbee, Lovenheim, and Slemrod (2010) further use cross-state differences in the diffusion of the Internet over time to show that tax responsiveness falls for a given tax increase when Internet penetration is higher. This is consistent with smokers purchasing tax-free cigarettes over the Internet to avoid local excise taxes, though it should be noted that this channel is no longer open to US smokers.

Tax avoidance opportunities can affect prices as well as quantity: closer to lower-tax borders, taxes may be shifted less to prices. Harding, Leibtag, and Lovenheim (2012) and Carpenter and Mathes (2016) both show evidence consistent with this hypothesis. In the former study, only about 50% of a tax increase is passed through to prices in areas very close to a lower-tax jurisdiction, while farther away prices rise almost one-for-one with taxes. This leads cigarette taxes to be even less effective in reducing smoking near lower-tax borders. Additionally, several studies have examined other margins of consumer response that may mute the effect of taxes on smoking, including purchasing less expensive generic cigarettes (cf., Espinosa & Evans, 2013), smoking cigarettes longer and/or harder (cf., Evans & Farrelly, 1998; Adda & Cornaglia, 2006; Abrevaya & Puzzello, 2012; Cotti, Nesson, & Tefft, 2016; Nesson, 2017) and purchasing carton vs. single packs (cf., Espinosa & Evans, 2013; DeCicca, Kenkel, & Liu, 2013a).

These broadly defined tax avoidance behaviors have the potential to attenuate the impact of cigarette tax increases and offer one set of explanations for why cigarette taxes do not appear to be very effective in reducing smoking. Lovenheim (2008) provides an illustrative calculation: consumers on the border with a lower-tax state do not change their smoking behavior when there is a tax increase, while for those sufficiently far away the extensive margin price elasticity is −0.23. These results suggest that increases in tax avoidance in response to larger tax differentials might cause cigarette tax elasticities to attenuate over time. An important question arising from these findings is whether health outcomes are strongly associated with cigarette avoidance opportunities. Answering this question requires examining the impact of tax avoidance on health outcomes thought to be linked to cigarette smoking using some of the strategies discussed in the “Background” section. No such research yet exists, but this literature will continue to grow rapidly as geographic disparities in cigarette taxes and improved data on avoidance behaviors grow. Finally, another important related area of research revolves around the impact of tax avoidance on tax regressivity (i.e., the idea that the burden of a tax is borne disproportionately by those with relatively lower incomes). Given that smokers tend to be lower-income relative to the general population, the lack of price sensitivity implies that cigarette taxes fall most heavily on lower-income groups. Opportunities for tax avoidance may increase or decrease the level of regressivity depending on which smokers are engaging in such tax avoidance activities. No research directly estimates the impact of tax avoidance on cigarette tax regressivity, which is an interesting area for future exploration.

The Impact of Smoke-Free Air Laws

Economists have devoted significant efforts to estimating the effects of smoke-free air laws in addition to cigarette taxes. The policy goals of these laws are twofold: to reduce individual smoking behavior by making smoking less convenient and thus raising the total cost of smoking, and to lower exposure to secondhand smoke among those in public spaces. It is common for smoke-free air laws to ban smoking in specific locations like workplaces, restaurants, and, more recently, bars and casinos. However, unlike taxes, smoke-free air laws bind only in certain locations, and so there is the potential that they will displace smoking from areas in which smoking is restricted to areas where it is unrestricted. Thus, these laws may affect secondhand smoke exposure and any resultant health impacts differentially as a function of the effects on both the level and location of cigarette consumption.

In this section, research on the impact of smoke-free air laws on smoking, secondhand smoke exposure, and health is reviewed. The general approach that researchers have taken mirrors that of the investigation of the effects of taxes: individual-level data on smoking behavior, secondhand smoke exposure, or health outcomes are connected to the smoke-free air laws of the geographic area in question, and two-way fixed effects models designed to control for unobserved time-invariant geographic characteristics and national secular trends in smoking are estimated. Similar to the research on cigarette taxes already discussed, concerns regarding whether there are time-varying unobserved factors correlated both with the passage of smoke-free air laws and the outcomes of interest are central to many of the recent advances in this area.

Impacts on Smoking Behavior

While there are similarities to research on taxes and smoking behavior, smoke-free air laws have a much shorter history. In developed nations, evidence on the effectiveness of smoke-free air laws on smoking behavior is mixed. Some early studies suggest that workplaces that banned smoking saw reductions in both smoking prevalence and cigarette consumption among remaining smokers (Evans, Farrelly, & Montgomery, 1999). Later studies have found inconsistent evidence on the effects of smoke-free air laws on smoking prevalence and intensity, with many papers finding that smoke-free air laws reduce smoking behavior (e.g., Pieroni et al., 2013; Bitler, Carpenter, & Zavodny, 2010; Cheng et al., 2017; Yurekli & Zhang, 2000; Tauras, 2006) and others finding less evidence of reductions in smoking (Nesson, 2017; Cotti, Nesson, & Tefft, 2016; Carpenter, Postolek, & Warman, 2011; Carpenter, 2009; Adda & Cornaglia, 2010; Jones et al., 2015). Though this body of research is less developed than the research on cigarette taxation, there is a lack of consensus across studies. Some of the variation in estimates may reflect heterogeneity in the specifics of the bans, including the areas to which the ban applies as well as differences across areas in enforcement. Further study of smoke-free air laws with an eye toward reconciling the literature is needed.

Impacts on Secondhand Smoke Exposure

Unlike smoking behavior, exposure to secondhand smoke raises the possibility of adverse external health outcomes. Measuring the effects of smoke-free air laws on secondhand smoke exposure faces many of the same challenges as examining impacts on smoking. However, such analyses also face additional challenges. The first major issue is data availability. In the United States, many large datasets contain information on smoking behavior, though only a few record measures of exposure to secondhand smoke. Only the Tobacco Use Supplements to the Current Population Surveys (TUS-CPS), a dataset which is available in certain years, and the National Health and Nutrition Examination Surveys (NHANES), a smaller dataset from the Centers for Disease Control and Prevention, contain information about secondhand smoke exposure. In Canada, the Canadian Tobacco Use Monitoring Survey (CTUMS) and the Canadian Youth Smoking Surveys (CYSS) both contain information about secondhand smoke exposure.

The second major issue surrounds the difficulty of measuring exposure. The most common measure of secondhand smoke exposure is through self-reports. While the Canadian datasets include more detail regarding areas of exposure to secondhand smoke, this information is much less available in US data sets. An unanswered but important question in this area of research is the accuracy of self-reported secondhand smoke exposure. Do errors in self-reported exposure induce classical measurement error, or are they related to smoke-free air laws, thus potentially biasing estimated policy effects? In addition to self-reports, some datasets contain biological markers of secondhand smoke exposure. In particular, the NHANES contains cotinine levels, which can be used to measure recent secondhand smoke exposure and reasonably distinguish between such exposure and smoking behavior. As more waves of relevant data become available, these questions are receiving more attention.

An additional core concern related to smoke-free air laws relates to the degree to which these laws bind, or put differently, who they might actually affect. Consider the case of workplace laws: if many workplaces have banned smoking prior to the passage of a workplace-specific smoke-free air law, or if workplaces do not comply with a new law, then econometric estimates may be internally valid but difficult to generalize. Bitler, Carpenter, and Zavodny (2010) examine these issues using state-level smoke-free air laws in 12 specific venues and data from the TUS-CPS, which includes information on employment, occupation, and workplace smoking policies as well as on smoking behavior. They find little evidence that workplace smoke-free air laws increase workers’ reports of smoking restrictions. The one exception is among bartenders, who report that smoke-free air laws in bars reduces cigarette consumption. Their findings suggest that workplace smoke-free air laws may not be binding because they either were already in place or were ignored.

Other studies, however, find stronger evidence that smoke-free air laws bind. Cheng et al. (2017) re-examine the findings in Bitler, Carpenter, and Zavodny (2010) using county-level changes in smoke-free air laws and find evidence that these laws increase workers’ reports of a workplace smoking policy and decrease secondhand smoke exposure among nonsmoking workers. Using data from Ontario, Canada, Carpenter (2009) examines workplace smoking restrictions, smoking behavior, and exposure to secondhand smoke. He finds evidence that local smoke-free air laws increased smoking restrictions in workplaces, with the largest increase in workplace restrictions for men and blue-collar workers. Although the laws did not reduce smoking behavior, they did reduce secondhand smoke exposure among blue-collar workers, who were the least likely to have been previously covered by smoke-free air laws prior to the passage of local legislation.

Smoke-free air laws also may have unintended consequences by displacing smoking from areas that are covered by the ban to uncovered areas. Adda and Cornaglia (2010) examine whether such displacement takes place using data from the American Time Use Surveys. They find that smoke-free air laws in bars and restaurants decrease the time smokers spend in these locations by about 20 minutes per week and increase time at home. Then, using data from the National Health and Nutrition Examination Surveys, the authors find that smoke-free air laws in restaurants and bars increase biological markers of exposure to secondhand smoke (i.e., cotinine levels) among children during the weekend. Similarly, smoke-free air laws in workplaces increase biological markers of smoking among children on weekdays. Other studies have found less evidence of displacement. Carpenter, Postolek, and Warman (2011) examine smoking restrictions in cities and provinces across Canada and self-reported exposure to secondhand smoke in different venues. They find that the smoke-free air laws reduced secondhand smoke exposure in many venues, particularly in and outside of bars and restaurants. Moreover, they do not find a displacement effect of increased secondhand smoke exposure in places not affected by the laws, such as cars or homes. Nguyen (2013) examines the effect of a ban in Canada on smoking in vehicles while children are present. Like Carpenter, Postolek, and Warman (2011), he finds that these bans reduced children’s self-reported exposure to secondhand smoke in cars but did not increase reported exposure to secondhand smoke in other venues such as restaurants, parks, or homes.

The Impact of Smoke-Free Indoor Air Laws on Health Outcomes

Both own-smoke and secondhand smoke can lead to a range of negative health outcomes. A small but growing literature examines the effect of smoke-free air laws on these smoking-related health outcomes. The most prevalent outcomes examined are the incidence of acute myocardial infarction (AMI, or heart attacks) and asthma. In principle, both could be affected by short-run changes in smoking behavior, as opposed to lung cancer and emphysema, which though strongly associated with cigarette smoking develop over relatively long time periods. Pell et al. (2008) find that a smoke-free air law implemented in Scotland in 2006 substantially reduced coronary-related hospitalizations. While an interesting finding, this study was restricted to a before-and-after approach, since the law was nationwide. As a result, it is susceptible to bias from secular trends in such hospitalizations. Other studies (Mackay et al., 2012, 2013) find evidence that this Scottish smoke-free air law improved pregnancy-related outcomes and reduced the incidence of stroke, but again they are limited to research designs that are pre/post in nature.

Shetty, DeLeire, White, and Bhattacharya (2011) examine the relationship between the passage of local smoke-free air laws and short-term health outcomes as measured by hospital admissions for various conditions in the United States. They find no impact on mortality, admissions for heart attacks, or chronic obstructive pulmonary disease (COPD). Interestingly, Shetty et al. (2011) do find some evidence of an increase in asthma admissions for minors after these laws are passed, perhaps consistent with the displacement effects found in Adda and Cornaglia (2010). More recently, Mazzonna and Salari (2015) exploit the rollout of smoke-free air laws across regions in Switzerland to estimate effects of these laws on AMI. The results show a large negative effect on AMI, which is most prominent for men and is largest for 50–65-year-olds. Related estimates suggest that secondhand smoke exposure is the main mechanism driving the effects. Overall, research credibly linking smoke-free air laws and health outcomes is scarce, due in part to their recency in the 2010s and the difficulty in obtaining local health outcomes. It is expected that this will change in the coming years as these laws proliferate and better data on secondhand smoke exposure and health become available.

Emerging Issues

In this section, some recent methodological issues in the economics of smoking that are likely to continue in importance for the foreseeable future are considered. In particular, the growing use (and potential misuse) of place-specific trends, synthetic control methods, relevant developments in statistical inference, and improved measurement of smoking behavior are discussed.

Attempts to Improve Estimation of Anti-Smoking Policy Effects

Because the familiar two-way fixed effects model exploits place and time variation, and often features a continuous treatment (e.g., cigarette taxes), it is considered a “generalized” difference-in-differences model. As such, it implicitly assumes “parallel trends” in treated and untreated areas in order to interpret estimated associations as causal. Recognition that the parallel trends assumption may not be met has led several researchers to augment their standard two-way fixed effects models with place-specific trends. Typically, these trends are linear in nature—that is, they are comprised of a linear time variable that is interacted with place-specific indicators and included in the model. While often there are good reasons to believe that policy-making jurisdictions may not share common counterfactual smoking trends, such a strategy is potentially problematic. Most simply, the inclusion of place-specific trends may soak up a great deal of policy variation, especially if treatment effects are time-varying. With time-varying treatment effects, linear trends thus will absorb some of the policy effect, thereby leading researchers to underestimate policy impacts (Wolfers, 2006). To test for this problem, one can estimate an auxiliary regression where the policy variable is regressed on the state-specific trends variable. A “high” R-squared may indicate that there is not enough variation to separately identify the effect of the policy variable from the effects of the trend variables. A high degree of collinearity also may imply ill-conditioned data, which can lead to biased point estimates and not just inflated variances (cf., Marquardt & Snee, 1975; Anderson & Wells, 2008). Practically, the degree of collinearity will depend on the amount of policy variation available to the researcher over the period of analysis.

Recent work in labor economics also cautions about the use of place-specific trends, particularly at the US state level, where such trends are relevant for US-based smoking studies. In response to work by Dube, Lester, and Reich (2010) and Allegretto, Dube, and Reich (2011), Neumark, Salas, and Wascher (2014), herein NSW, show that the inclusion of place-specific trends may actually discard relevant identifying information. While the two earlier papers found evidence of negative employment effects due to minimum wage increases with two-way fixed effect models, these effects disappeared once place-specific trends were included in their models. The crux of NSW’s argument is that place-specific trends may effectively reduce valid identifying variation—in their words, throwing the “baby” (i.e., valid identifying variation) out with the “bathwater” (i.e., unobserved differences correlated with the policy in question). While the NSW paper deals with the employment effects of the minimum wage, several recent papers in the economics of smoking add state-specific linear trends. Many researchers consider the use of place-specific trends as a robustness check for estimates from more basic models, rather than strictly preferring models that include them if they substantively impact treatment effect estimates. However, this approach implicitly assumes that it is the “baby” rather than the “bathwater” that is being thrown away when the place-specific trends are included. Instead, it seems more useful to compare estimates from models with and without place-specific trends, in order to gauge whether the coefficients of interest have changed in substantive ways. If they are similar in magnitude, it seems reasonable to conclude that the model without place-specific trends is not seriously biased by differential secular trends.

A more comprehensive approach is to estimate “event studies” that nonparametrically trace out pre-treatment relative trends and time-varying treatment effects. This method does not restrict secular trends to be linear, nor is it biased by time-varying treatment effects. However, such specifications tend to be underpowered, and they are difficult to implement in the presence of multiple or continuous treatments. Future research could usefully explore in more depth the tradeoffs between the various approaches to dealing with time trends.

A further approach that has been used to estimate the impact of anti-smoking policies is the synthetic control method, introduced by Abadie and Gardeazabal (2003). This method was designed to evaluate a policy that takes place in a single location (or very few locations) and lacks an obvious control group, beyond that provided by a pre/post-research design. Briefly, the method constructs a control group/counterfactual based on a weighted average from untreated units that most closely matches the treated area in terms of trends in the outcomes of interest prior to policy implementation. This weighted average is called a synthetic cohort, and the relevant policy impact is computed by simple post-policy differences in the outcome of interest. In essence, the method combines elements of statistical matching with those of difference-in-differences. The method also is useful for conducting statistical inference in a setting with very few treated observations.

The most prominent paper in the economics of smoking that uses synthetic control methods is Abadie, Diamond, and Hainmueller (2010), which examines the impact of California’s 1988 Proposition 99 tobacco control program. This program levied a 25-cent-per-pack tax on cigarettes, imposed restrictions on vending machine cigarette sales in public places frequented by youth, and prohibited individual cigarette sales, sometimes referred to as “loosies.” Though in existence for some time, synthetic control methods have not yet been widely used in the economics of smoking. This could be due to a lack of awareness, or perhaps to some of the shortcomings of synthetic control methods, which include the use of only a limited number of controls to match untreated units to the treated one. Like all empirical methods, SCM may be sensitive to model choices, including the controls that determine the weighted average of untreated units (i.e., the synthetic cohort) and the lag length of these pre-period variables (McClelland & Gault, 2017). However, synthetic control methods offer an alternative tool for researchers in the economics of smoking, particularly for highly localized or unique policies without an obvious control group (i.e., policy evaluations that might otherwise be treated as case studies).

Attempts to Improve Inference

Thus far, the focus of this review has been on causal estimation of policy parameters—that is, issues related to obtaining an unbiased estimate of a particular anti-smoking policy. Of equal importance are empirical challenges related to statistical inference, which has received little attention in the prior economics of smoking literature. In this literature, the inference issue that has received the most attention over the past decade is the clustering of standard errors. Briefly, the place-specific nature of anti-smoking policies means that all individuals residing in the same policy area, or cluster, are assigned the same value of the policy variable. This creates potential nonindependence of observations within a given cluster, where a “cluster” is typically synonymous with the unit that generates the policy variation in question. In US smoking studies, for example, the relevant clusters are states, since they are the origin of most policy variation in anti-smoking policies. In other words, the smoking decisions of two individuals in the same state are likely to be more correlated with one another than are the decisions of individuals living in different states, because those in the same state face the same tobacco regulatory environment. Failure to account for this within-cluster error dependence leads to an over-rejection of null hypotheses: policy effects spuriously appear statistically significant.

While the issue has been recognized in economics at least since Moulton (1990), more recent research investigates inference concerns more deeply. Indeed, much recent work finds that even standard clustering can over-reject the null hypothesis. In general, this research suggests that the number of treated clusters and the number of observations within clusters are important in determining whether or not valid variance estimates can be computed. For example, Conley and Taber (2011) show that standard cluster procedures are invalid when the number of policy-generating clusters is small.

Practically, this feature creates difficulties in areas with relatively few policy-generating units and is one reason why research on the United States (with 51 such units) is so prominent in the literature. For example, Canada taxes cigarettes similarly to the United States, but has only 10 provinces rather than 51 taxing units. In such a situation, standard clustering may result in biased standard errors, particularly ones that are too small and lead to over-rejection of the null hypothesis. Relatively recent work suggests techniques like the Wild Cluster Bootstrap can produce better variance computations for jurisdictions with relatively small numbers of treated clusters (Cameron, Gelbach, & Miller, 2008). However, other research suggests that the Wild Cluster Bootstrap may not solve the problem in the context of few treated clusters and different-sized clusters and can either under- or over-reject (MacKinnon & Webb, 2017). These issues have led applied econometricians to explore methods based on Randomized Inference, which, according to MacKinnon and Webb (2016), tend to perform better in simulations than Conley-Taber methods.

Another relevant but emerging issue deals with the number of observations in different clusters (i.e., different cluster size). While standard clustering seems to have become the norm in smoking-related studies by economists, many of these innovations have yet to be widely implemented. It is to be expected that researchers studying the economics of smoking will begin to consider these more advanced methods for inference more seriously in order to bolster the credibility of the evidence created.

Attempts to Improve the Measurement of Smoking

As noted in the Background section, the vast amount of smoking-related data comes from individual self-reports. While potentially more objective sources will be discussed briefly, the self-reported nature of smoking behavior lends itself to measurement error. The classical treatment of measurement error in the dependent variable tells us that random errors will tend to result in variance estimates that are biased upwards. There are two main reasons measurement error in smoking participation is unlikely to be classical. First, smoking participation is a dichotomous variable, and this implies issues beyond inflated variances (Hausman, Abrevaya, & Scott-Morton, 1998; Kenkel, Lillard, & Mathios, 2004; Meyer & Mittag, 2014). The second reason classical measurement error is unlikely to apply is that such errors may not be random. In particular, self-reports may be related to an area’s anti-smoking sentiment or perhaps even the very smoking policies examined, if, for example, anti-smoking stigma increases with policy implementation. Admittedly, however, relatively little is known about this possibility.

While not much is known about the nature of misreported smoking behavior, it has been descriptively quantified, and it is particularly prevalent in groups where cigarette smoking is most socially taboo, like pregnant women (cf., Hall, Wexelblatt, & Greenberg, 2016). Systematic misreporting is especially concerning if people who live in areas that implement (or augment) anti-smoking policies are less likely to admit their smoking behavior, perhaps due to growing taboos in their local environment; such systematic misreporting would lead to an overestimate of the policy in question, since there is no way to verify true behavioral responses. While large-scale data is limited, sources such as the National Health and Nutrition Examination Survey and the Health Survey for England, which contain more objective smoking information in the form of biological markers, can help address these important questions. It is likely that more data of this type will be gathered in the future, and that this will further improve the measurement of smoking behavior as well as open other useful research doors.


Over the past quarter of a century, developed nations have substantially increased their regulation of cigarette smoking, most along several policy dimensions: cigarette excise taxes have increased dramatically even in places with historically low rates; most nations, even those where public smoking is part of the social fabric, now impose smoke-free air laws that prohibit smoking in public places; large-scale information campaigns warning against the health impacts of cigarette smoking are commonplace; and nearly all countries tightly regulate the availability, composition, and marketing of cigarettes. Indeed, the past 25 years have witnessed a full-scale change both in private attitudes toward smoking and its policy treatment by governments.

The first half of this article reviewed evidence on the impact of the two most prominent anti-smoking policy areas: cigarette excise taxation and smoke-free air laws. Regarding the former, the prevailing literature suggests that smoking behavior is largely insensitive to price—particularly for adults, but also for youth. Especially in the case of adults, this may be due in part to the availability of lower-taxed or lower-priced alternatives which we have documented. While these conclusions seem warranted, it is likely that research on cigarette taxation will continue full-bore but will focus even more on areas beyond cigarette consumption—for example, improved estimation of tax avoidance behaviors and the potential impact of cigarette taxes on health outcomes. This article also reviewed evidence on smoke-free air laws and showed that the literature is substantially mixed in terms of whether such laws reduce smoking or displace it, and whether they have appreciable impacts on health. This is not surprising, since these laws are, at least relative to cigarette taxation, newer and more heterogeneous. The last part of the article discussed some emerging issues in the economics of smoking research, particularly those relevant to valid identification and estimation of policy effects. It is highly likely that future work will more fully address these issues and hence produce improved estimates both of the policies discussed and those that were not discussed here due to space constraints.

While much has been learned over the past quarter-century, much remains unknown. The next generation of smoking research in economics is likely to improve on our knowledge of the policies covered, as well as those omitted due to space constraints, though the Further Reading section below notes work in some omitted areas. Moreover, it will certainly focus on emerging products and cigarette alternatives, including recent innovations like electronic cigarettes and heat-not-burn cigarettes and further developments in nicotine replacement therapies (NRT). As it does so, researchers will build upon the lessons of the past 25 years as they break new ground.


The authors thank an anonymous referee for very helpful comments. Research reported in this article was supported by the National Institutes of Health under award number R01DA042064. We acknowledge that the research by DeCicca, Kenkel, and Lovenheim was partially funded by this grant. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Further Reading

Adams, S., & Cotti, C. (2008). Drunk driving after the passage of smoking bans in bars. Journal of Public Economics, 92(5–6), 1288–1305.Find this resource:

Bedard, K., & Deschênes, O. (2006). The long-term impact of military service on health: Evidence from World War II and Korean War veterans. The American Economic Review, 96(1), 176–194Find this resource:

Colman, G. J., & Remler, D. K. (2008). Vertical equity consequences of very high cigarette taxes: If the poor are the ones smoking, how could cigarette tax increases be progressive? Journal of Policy Analysis and Management, 27(2), 376–400.Find this resource:

Cotti, C., Nesson, E., & Tefft, N. (2016). The effects of tobacco control policies on tobacco products, tar, and nicotine consumption: Evidence from household panel data. American Economic Journal: Economic Policy, 8(4), 103–123.Find this resource:

Courtemanche, C. (2009). Rising cigarette prices and rising obesity: Coincidence or unintended consequence? Journal of Health Economics, 28(4), 781–798.Find this resource:

Darden, M., Gilleskie, D. B., & Strumpf, K. (2015). Smoking and mortality: New evidence from a long panel. Working paper.Find this resource:

DeCicca, P., Kenkel, D. S., Liu, F., & Wang, H. (2017). Behavioral welfare economics and FDA tobacco regulation. In K. Bolin, B. Lindgren, M. Grossman, D. Gyrd-Hansen, T. Iversen, R. Kaestner, & J. Sindelar (Eds.), Advances in Health Economics and Health Services Research: Human capital and health behavior (pp. 143–179). Bingley, UK: Emerald Publishing.Find this resource:

Evans, W. N., & Ringel, J. S. (1999). Can higher cigarette taxes improve birth outcomes? Journal of Public Economics, 72, 135–154.Find this resource:

Evans, W. N., Farrelly, M., & Montgomery, E. (1999). Do workplace smoking bans reduce smoking? American Economic Review, 89(4), 728–747.Find this resource:

Goldin, J., & Homonoff, T. (2013). Smoke gets in your eyes: Cigarette tax salience and regressivity. American Economic Journal: Economic Policy, 5(1), 302–336.Find this resource:

Gospodinov, N., & Irvine, I. (2009). Tobacco taxes and regressivity. Journal of Health Economics, 28(2), 375–384.Find this resource:

Haavio, M., & Kotakorpi, K. (2011). The political economy of sin taxes. European Economic Review, 55(4), 575–594.Find this resource:

Levy, H., Norton, E. C., & Smith, J. (2016). Tobacco-regulation and benefit-cost analysis: How should we value foregone consumer surplus? NBER Working Paper #22471.Find this resource:

Markowitz, S. (2008). The effectiveness of cigarette regulations in reducing cases of Sudden Infant Death Syndrome. Journal of Health Economics, 27(1), 106–133.Find this resource:

O’Donoghue, T., & Rabin, M. (2006). Optimal sin taxes. Journal of Public Economics, 90(10–11), 1825–1849.Find this resource:

Simon, D. (2016). Does early life exposure to cigarette smoke permanently harm childhood welfare? Evidence from cigarette tax hikes. American Economic Journal: Applied Economics, 8(4), 128–159.Find this resource:


Abadie, A., Diamond, A., & Hainmueller, J. (2010). Synthetic control methods for comparative case studies: Estimating the effect of California’s tobacco control program. Journal of the American Statistical Association, 105, 493–505.Find this resource:

Abadie, A., & Gardeazabal, J. (2003). The economic costs of conflict: A case-control study for the Basque Country. American Economic Review, 93(1), 113–132.Find this resource:

Abrevaya, J., & Puzzello, L. (2012). Taxes, cigarette consumption and smoking: Comment. American Economic Review, 102(4), 1751–1763.Find this resource:

Adda, J., & Cornaglia, F. (2006). Taxes, cigarette consumption and smoking intensity. American Economic Review, 96(4), 1013–1028.Find this resource:

Adda, J., & Cornaglia, F. (2010). The effect of bans and taxes on passive smoking. American Economic Journal: Economic Policy, 2(1), 1–32.Find this resource:

Allegretto, S., Dube, A., & Reich, M. (2011). Do minimum wages really reduce teen employment? Accounting for heterogeneity and selectivity in panel data. Industrial Relations, 50(2), 205–240.Find this resource:

Anderson, W., & Wells, M. (2008). Numerical analysis in least squares regression with an application to the abortion-crime debate. Journal of Empirical Legal Studies, 5(4), 647–681.Find this resource:

Aristei, D., & Pieroni, L. (2009). Addiction, social interaction and gender differences in cigarette consumption. Empirica, 36(3), 245–272.Find this resource:

Auld, M. C. (2005). Causal effect of early initiation on adolescent smoking patterns. Canadian Journal of Economics, 38(3), 709–734.Find this resource:

Auld, M. C., & Grootendorst, P. (2004). An empirical analysis of milk addiction. Journal of Health Economics, 23(6), 1117–1133.Find this resource:

Becker, G., Grossman, M., & Murphy, K. M. (1991). Rational addiction and the effect of price on consumption. American Economic Review, 81(2), 237–241.Find this resource:

Becker, G., & Murphy, K. M. (1988). A theory of rational addiction. Journal of Political Economy, 96(4), 675–700.Find this resource:

Bernheim, B. D., & Rangel, A. (2004). Addiction and cue-triggered decision processes. American Economic Review, 94(5), 1558–1590.Find this resource:

Bitler, M., Carpenter, C., & Zavodny, M. (2010). Effects of venue-state indoor clean air laws on smoking related outcomes. Health Economics, 19(2), 1425–1440.Find this resource:

Breslau, N., & Peterson, E. L. (1996). Smoking cessation in young adults: Age at initiation of cigarette smoking and other suspected influences. American Journal of Public Health, 86(2), 214–220.Find this resource:

Callison, K., & Kaestner, R. (2014). Do higher tobacco taxes reduce adult smoking? New evidence of the effect of recent cigarette tax increases on adult smoking. Economic Inquiry, 52(1), 155–172.Find this resource:

Cameron, A. C., Gelbach, J., & Miller, D. L. (2008). Bootstrap-based improvements for inference with clustered errors. Review of Economics and Statistics, 90(3), 414–427.Find this resource:

Carpenter, C. (2009). The effects of local workplace smoking laws on smoking restrictions and exposure to smoke at work. Journal of Human Resources, 44(4), 1023–1046.Find this resource:

Carpenter, C., & Cook, P. (2008). Cigarette taxes and youth smoking: New evidence from national, state, and local youth risk behavior surveys. Journal of Health Economics, 27(2), 287–299.Find this resource:

Carpenter, C., & Mathes, M. (2016). New evidence on the price effects of cigarette competition. Public Finance Review, 44(3), 291–310.Find this resource:

Carpenter, C., Postolek, S., & Warman, C. (2011). Public place smoking laws and exposure to environmental tobacco smoke. American Economic Journal: Economic Policy, 3(3), 35–61.Find this resource:

Cawley, J., Markowtiz, S., & Tauras, J. (2004). Lighting up and slimming down: The effect of body weight and cigarette prices on adolescent smoking initiation. Journal of Health Economics, 23(2), 293–311.Find this resource:

Chaloupka, F., & Liccardo-Pacula, R. (1999). Sex and race differences in young people’s responsiveness to price and tobacco control policies. Tobacco Control, 8(4), 373–377.Find this resource:

Chaloupka, F., & Grossman, M. (1996). Price, tobacco control policies and youth smoking. NBER Working Paper No. 5740.Find this resource:

Chaloupka, F., & Warner, K. (2000). The economics of smoking. In A. J. Culyer & J. P. Newhouse (Eds.), The handbook of health economics (Vol. 1B, pp. 1539–1627). Amsterdam: Elsevier Science.Find this resource:

Chaloupka, F., & Wechsler, W. (1997). Price, tobacco control policies and smoking among young adults. Journal of Health Economics, 16(3), 359–373.Find this resource:

Cheng, K-W., Liu, F., Gonzalez, M., & Glantz, S. (2017). The effects of workplace clean air law coverage on workers’ smoking-related outcomes. Health Economics, 26(2), 226–242.Find this resource:

Chernick, H., & Merriman, D. (2013). Using littered pack data to estimate cigarette tax avoidance in NYC. National Tax Journal, 66(3), 635.Find this resource:

Chiou, L., & Muehlegger, E. (2008). Crossing the line: Direct estimation of cross-border cigarette sales and the effect on tax revenue. The B.E. Journal of Economic Analysis & Policy, 8(1), 1–41.Find this resource:

Chiou, L., & Muehlegger, E. (2014). Consumer response to cigarette excise tax changes. National Tax Journal, 67(3), 621–650.Find this resource:

Chung, W., Lim, S., Lee, S., Choi, S., Shin, K., & Cho, K. (2007). The effect of cigarette price on smoking behavior in Korea. Journal of Preventive Medicine and Public Health, 40(5), 371–380.Find this resource:

Coats, R. M. (1995). A note on estimating cross-border effects of state cigarette taxes. National Tax Journal, 48(4), 573–584.Find this resource:

Conley, T., & Taber, C. (2011). Inference with “difference in differences” with a small number of policy changes. Review of Economics and Statistics, 93(1), 113–125.Find this resource:

Cotti, C., Nesson, E., & Tefft, N. (2016). The effects of tobacco-control policies on tobacco products, tar and nicotine purchases among adults: Evidence from household panel data. American Economic Journal: Economic Policy, 8(4), 103–123.Find this resource:

DeCicca, P., & Kenkel, D. (2015). Synthesizing econometric evidence: The case of demand elasticities. Risk Analysis, 35(6), 1073–1985.Find this resource:

DeCicca, P., Kenkel, D., & Liu, F. (2013a). Who pays cigarette taxes? The impact of consumer price search. Review of Economics and Statistics, 95(2), 516–529.Find this resource:

DeCicca, P., Kenkel, D., & Liu, F. (2013b). Excise tax avoidance: The case of state cigarette taxes. Journal of Health Economics, 32, 1130–1141.Find this resource:

DeCicca, P., Kenkel, D., & Liu, F. (2015). Reservation prices: An economic analysis of cigarette purchases on Indian reservations. National Tax Journal, 68(1), 93–120.Find this resource:

DeCicca, P., Kenkel, D., & Mathios, A. (2002). Putting out the fires: Will higher taxes reduce the onset of youth smoking? Journal of Political Economy, 110(1), 144–169.Find this resource:

DeCicca, P., Kenkel, D., & Mathios, A. (2008). Cigarette taxes and the transition from youth to adult smoking: Smoking initiation, cessation and participation. Journal of Health Economics, 27(4), 904–917.Find this resource:

DeCicca, P., Kenkel, D., Mathios, A., Shin, Y.-J., & Lim, J.-Y. (2008). Youth smoking, cigarette prices, and anti-smoking sentiment. Health Economics, 17(6), 733–749.Find this resource:

DeCicca, P., & McLeod, L. (2008). Cigarette taxes and older adult smoking: Evidence from recent large tax increases. Journal of Health Economics, 27(4), 918–929.Find this resource:

Dee, T. (1999). The complementarity of teen smoking and drinking. Journal of Health Economics, 18(6), 769–793.Find this resource:

Dube, A., Lester, W., & Reich, M. (2010). Minimum wage effects across state borders: Estimating using contiguous counties. Review of Economics and Statistics, 92(4), 945–964.Find this resource:

Evans, W., & Farrelly, M. (1998). The compensating behavior of smokers: Taxes, tar, and nicotine. The Rand Journal of Economics, 29(3), 578–595.Find this resource:

Espinosa, J., & Evans, W. (2013). Excise taxes, tax incidence, and the flight to quality: Evidence from scanner data. Public Finance Review, 41(2), 147–176.Find this resource:

Farrell, M. (1952). Irreversible demand functions. Econometrica, 20(2), 171–186.Find this resource:

Farrelly, M. C., Bray, J., Paycheck, T., & Woollery, T. (2001). Responses by adults to increases in cigarette prices by sociodemographic characteristics. Southern Economic Journal, 68(1), 156–165.Find this resource:

Gospodinov, N., & Irvine, I. (2009). Tobacco taxes and regressivity. Journal of Health Economics, 28(2), 375–384.Find this resource:

Glied, S. (2002). Youth tobacco control: Reconciling theory and empirical evidence. Journal of Health Economics, 21(1), 117–135.Find this resource:

Glied, S. (2003). Is smoking delayed smoking averted? American Journal of Public Health, 93(3), 412–416.Find this resource:

Goolsbee, A., Lovenheim, M. F., & Slemrod, J. (2010). Playing with fire: Cigarettes, taxes, and competition from the internet. American Economic Journal: Economic Policy, 2(1), 131–154.Find this resource:

Gruber, J., & Koszegi, B. (2001). Is addiction “rational”? Theory and evidence. Quarterly Journal of Economics, 116(4), 1261–1303.Find this resource:

Gruber, J., Sen, A., & Stabile, M. (2003). Estimating price elasticities when there is smuggling: The sensitivity of smoking to price in Canada. Journal of Health Economics, 22(5), 821–842.Find this resource:

Gruber, J., & Zinman, J. (2001). Youth smoking in the United States: Evidence and implications. In J. Gruber (Ed.), Risky behavior among youths: An economic analysis (pp. 69–120). Chicago: University of Chicago Press.Find this resource:

Haavelmo, T. (1944). The probability approach in econometrics. Econometrica: Journal of the Econometric Society, 12, iii–115.Find this resource:

Hall, E. S., Wexelblatt, L., & Greenberg, J. M. (2016). Self-reported and laboratory evaluation of late pregnancy nicotine exposure and drugs of abuse. Journal of Perinatology, 36, 814–818.Find this resource:

Hansen, B., Sabia, J., & Rees, D. (2017). Have cigarette taxes lost their bite? New estimates of the relationship between cigarette taxes and youth smoking. American Journal of Health Economics, 3(1), 60–75.Find this resource:

Harding, M., Leibtag, E., & Lovenheim, M. (2012). The heterogeneous geographic and socioeconomic incidence of cigarette taxes: Evidence from Nielsen Homescan data. American Economic Journal: Economic Policy, 4(4), 169–198.Find this resource:

Harris, J., & Chan, S. (1999). The continuum‐of‐addiction: Cigarette smoking in relation to price among Americans aged 15–29. Health Economics, 8(1), 81–86.Find this resource:

Hausman, J. A., Abrevaya, J., & Scott-Morton, F. (1998). Misclassification of the dependent variable in a discrete response setting, Journal of Econometrics, 87, 239–269.Find this resource:

Houthakker, H. S., & Taylor, L. D. (1970). Consumer demand in the United States 1929–1970, analysis and projections (2nd ed.). Cambridge, MA: Harvard University Press.Find this resource:

Jimenez-Martin, S., Labeaga, J. M., & Lopez, A. (1998). Participation, heterogeneity and dynamics in tobacco consumption: Evidence from cohort data. Health Economics, 7(5), 401–414.Find this resource:

Jones, A. M., Laporte, A., Rice, N., & Zucchelli, E. (2015). Do public smoking bans have an impact on active smoking? Evidence from the UK. Health Economics, 24, 175–192.Find this resource:

Jones, A. M. (1989). A double‐hurdle model of cigarette consumption. Journal of Applied Econometrics, 4(1), 23–39.Find this resource:

Kenkel, D., Lillard, D. R., & Mathios, A. D. (2004). Accounting for misclassification error in retrospective smoking data. Health Economics, 13(10), 1031–1044.Find this resource:

Labeaga, J. (1999). A double-hurdle rational addiction model with heterogeneity: Estimating the demand for tobacco. Journal of Econometrics, 93(1), 49–72.Find this resource:

Lakhdar, C. B., Vaillant, N. G., & Wolff, F.-C. (2016). Does smoke cross the border? Cigarette tax avoidance in France. The European Journal of Health Economics, 17(9), 1073–1089.Find this resource:

Laporte, A., Dass, A. R., & Ferguson, B. (2017). Is the rational addiction model inherently impossible to estimate? Journal of Health Economics, 54, 161–175.Find this resource:

Lewit, E., & Coate, D. (1982). The potential for using excise taxes to reduce smoking. Journal of Health Economics, 1(2), 121–145.Find this resource:

Lewit, E., Coate, D., & Grossman, M. (1981). The effects of government regulation on teenage smoking. The Journal of Law and Economics, 24(3), 545–569.Find this resource:

Lillard, D., Molloy, E., & Sfekas, A. (2013). Smoking initiation and the iron law of demand. Journal of Health Economics, 32(1), 114–127.Find this resource:

Lovenheim, M. F. (2008). How far to the border? The extent and impact of cross-border casual cigarette smuggling. National Tax Journal, 61(1), 7–33.Find this resource:

Mackay, D., Nelson, S., Haw, S., & Pell, J. (2012). Impact of Scotland’s smoke-free legislation on pregnancy complications: Retrospective cohort study. PLoS Med, 9(3). Retrieved from this resource:

Mackay, D., Haw, S., Newby, D., Langhorne, P., Lloyd, S., McConnachie, A., & Pell, J. (2013). Impact of Scotland’s comprehensive, smoke-free legislation on stroke. PLoS ONE, 8(5). Retrieved from this resource:

MacKinnon, J., & Webb, M. (2016). Randomization inference for difference-in-differences with few treated clusters. Working Paper 1355, Queen’s University, Department of Economics.Find this resource:

MacKinnon, J., & Webb, M. (2017), Wild bootstrap inference for wildly different cluster sizes. Journal of Applied Econometrics, 32, 233–254.Find this resource:

Maclean, J. C., Sikora-Kessler, A., & Kenkel, D. (2016). Cigarette taxes and older adult smoking: Evidence from the Health and Retirement Study. Health Economics, 25(4), 424–438.Find this resource:

Marquardt, D., & Snee, R. (1975). Ridge regression in practice. The American Statistician, 29(1), 3–20.Find this resource:

Mazzonna, F. & Salari, P. (2015). Short term effect of public smoking bans on health. Unpublished manuscript.Find this resource:

McClelland, R. & Gault, S. (2017). The synthetic control method as a tool to understand state policy, Research Report of the Tax Policy Center of the Urban Institute, Washington, DC.Find this resource:

Merriman, D. (2010). The micro-geography of tax avoidance: Evidence from littered cigarette packs in Chicago. American Economic Journal: Economic Policy, 2(2), 61–84.Find this resource:

Meyer, B., & Mittag, N. (2014). Misclassification in binary choice models. NBER Working Paper #20509.Find this resource:

Moulton, B. R. (1990). An illustration of a pitfall in estimating the effects of aggregate variable on micro units. Review of Economics and Statistics, 72(2), 334–338.Find this resource:

Mullahy, J. (1985). Cigarette smoking: Habits, health concerns, and heterogeneous unobservables in a microeconometric analysis of consumer demand (Unpublished doctoral dissertation). University of Virginia, Charlottesville.Find this resource:

Nesson, E. (2017). Heterogeneity in smokers’ response to tobacco control policies. Health Economics, 26(2), 206–225.Find this resource:

Neumark, D., Salas, J. M. I., & Wascher, W. (2014). Revisiting the minimum wage-employment debate: Throwing out the baby with the bathwater? Industrial and Labor Relations Review, 67(Suppl.), 608–648.Find this resource:

Nguyen, V. H. (2013). Do smoke-free car laws work? Evidence from a quasi-experiment. Journal of Health Economics, 32(1), 138–148.Find this resource:

Nonnemaker, J., & Farrelly, M. (2011). Smoking initiation among youth: The role of cigarette excise taxes and prices by race/ethnicity and gender. Journal of Health Economics, 30(3), 550–567.Find this resource:

Pell, J., Haw, S., Cobbe, S., Newby, D. E., Pell, A. C. H., Fischbacher, C., . . . Borland, W. (2008). Smoke-free legislation and hospitalizations for acute coronary syndrome. New England Journal of Medicine, 395, 482–491.Find this resource:

Pieroni, L., Chiavarini, M., Minelli, L., & Salmasi, L. (2013). The role of anti-smoking legislation on cigarette and alcohol consumption habits in Italy. Health Policy, 111(2), 116–126.Find this resource:

Pollak, R. A. (1978). Endogenous tastes in demand and welfare analysis. American Economic Review, 68, 374–379.Find this resource:

Sen, A., Ariizumi, H., & Driambe, D. (2010). Do changes in cigarette taxes impact youth smoking? Evidence from Canadian provinces. Forum for Health Economics & Policy, 13(2).Find this resource:

Shetty, K., DeLeire, T., White, C., & Bhattacharya, J. (2011). Changes in U.S. hospitalization and mortality rates following smoking bans, Journal of Policy Analysis and Management, 30(1), 6–28.Find this resource:

Sloan, F. A., & Trogdon, J. (2004). The impact of the Master Settlement Agreement on cigarette consumption. Journal of Policy Analysis and Management, 23(4), 843–855.Find this resource:

Stehr, M. (2005). Cigarette tax avoidance and evasion. Journal of Health Economics, 24(2), 277–297.Find this resource:

Tauras, J. (2006). Smoke-free air laws, cigarette prices, and adult cigarette demand. Economic Inquiry, 44(2), 333–342.Find this resource:

Wasserman, J., Manning, W. G., Newhouse, J. P., & Winkler, J. D. (1991). The effects of excise taxes and regulations on cigarette smoking. Journal of Health Economics, 10(1), 43–64.Find this resource:

Warner, K., & Mendez, D. (2012). Accuracy and importance of projections from a dynamic simulation model of smoking prevalence in the United States. American Journal of Public Health, 102(11), 2045–2048.Find this resource:

Wolfers, J. (2006). Did unilateral divorce laws raise divorce rates? A reconciliation and new results. American Economic Review, 96(5), 1802–1820.Find this resource:

Yurekli, A., & Zhang, P. (2000). The impact of clean indoor air laws and cigarette smuggling on the demand for cigarettes. Health Economics, 9(2), 159–170.Find this resource:


(1.) Classical measurement error arises when a variable of interest is measured with error that is unrelated to the variable’s actual value or to that of other variables in the model (i.e., the error can be thought of as random noise, and hence uncorrelated with the true value of the variable itself, the other controls in the model, or the error term).

(2.) The age at which individuals can legally purchase cigarettes varies across and within nations, consistent with smoking laws more generally. In the United States, the legal purchasing age is 18 in all states, while in Canada it is 18 or 19 depending on province.

(3.) While some studies report tax elasticities, price elasticities are most commonly reported. A tax elasticity is converted to a price elasticity by multiplying the former by the ratio of average price to tax, as well as the rate at which cigarette taxes are passed through to cigarette price.

(4.) Marginal effects are reported here because price participation elasticities are highly sensitive to the level of smoking participation in the relevant sample. Nevertheless, participation elasticities are discussed, because they are most common in the literature. Researchers should report marginal tax effects as well as price participation elasticities.

(5.) Casual smuggling, which in general is legal, is in contrast to organized smuggling, which is typically conducted by organized crime groups and consists of obtaining a large volume of cigarettes and illegally selling them tax-free.