Empty Discarded Pack Data and the Prevalence of Illicit Trade in Cigarettes in California

Pepperdine University Pepperdine University

Pepperdine Digital Commons Pepperdine Digital Commons

School of Public Policy Working Papers School of Public Policy

1-22-2019

Empty Discarded Pack Data and the Prevalence of Illicit Trade in Empty Discarded Pack Data and the Prevalence of Illicit Trade in

Cigarettes in California Cigarettes in California

James Prieger

Pepperdine University

, [email protected]

Follow this and additional works at: https://digitalcommons.pepperdine.edu/sppworkingpapers

Part of the Econometrics Commons, Health Economics Commons, Political Economy Commons, and

the Public Affairs, Public Policy and Public Administration Commons

Recommended Citation Recommended Citation

Prieger, James, "Empty Discarded Pack Data and the Prevalence of Illicit Trade in Cigarettes in California"

(2019). Pepperdine University,

School of Public Policy Working Papers.

Paper 75.

https://digitalcommons.pepperdine.edu/sppworkingpapers/75

This Article is brought to you for free and open access by the School of Public Policy at Pepperdine Digital

Commons. It has been accepted for inclusion in School of Public Policy Working Papers by an authorized

administrator of Pepperdine Digital Commons. For more information, please contact bailey.berr[email protected].

School of Public Policy

24255 Pacific Coast Highway

Malibu, CA 90263- 7490

Jonathan Kulick

JDK Analysis

30 Regent #318

Jersey City, NJ 07302

jd[email protected]

James E. Prieger

Professor

Pepperdine University

School of Public Policy

24255 Pacific Coast Highway

Malibu, CA 90263-7490

James.Prieger@

pepperdine.edu

January 22,

2019

Empty Discarded

Pack Data and the

Prevalence of Illicit

Trade in Cigarettes

in California

PEPPERDINE UNIVERSITY

Abstract

Illicit trade in tobacco products (ITTP) creates many harms including reduced tax revenues; damages to

the economic interests of legitimate actors; funding for organized-crime and terrorist groups; negative

effects of participation in illicit markets, such as violence and incarceration; and reduced effectiveness of

smoking-reduction policies, leading to increased damage to health. To study the prevalence of tax

avoidance and ITTP, we analyze a large, novel set of data from empty discarded pack (EDP) studies. In

EDP studies, teams of researchers collect all cigarette packs discarded in publicly accessible spaces of

selected neighborhoods. Packs are examined for the absence of local tax stamps, signs of non-authentic

packaging or stamps, and other indications of potential tax evasion or counterfeit product. We describe

the data and analyze the prevalence of ITTP in three California metropolitan areas. Data from 2011 to

2015 are available, yielding 32,000 observations. Each observation includes dozens of variables covering

the brand, location to the ZIP code level, tax status, counterfeit status, and other information about the

pack. There is modest evidence of tax avoidance (up to 19.8% of packs in San Diego) and illicit trade (no

more than 10% in Los Angeles, 17% in San Francisco, or 20% in San Diego under the broadest

assumptions), which includes bootlegging, counterfeits, cigarettes produced for illicit-market sales, and

cigarettes without any tax stamps. California increased its excise tax on cigarettes by $2 per pack (to

$2.87) on April 1, 2017; these data will inform future studies of the effect of this increase on ITTP.

In our econometric investigation, we explore the determinants of ITTP. Prices in other states matter a

lot: A dollar increase per hour of driving time in the price differential with other states is associated with

a 36 to 49 percentage point higher probability that tax was not paid for a pack. Tax avoidance also rises

with the proximity of licensed cigarette retailers (at least where they are most common). Other parts of

the variation in ITTP are due to the differing demographic makeup of the areas. Income, at least in some

ranges, was found to have a negative impact on tax avoidance. The fraction of the population that is

Black has a negative effect on tax avoidance, compared to the omitted race/ethnicity category of

Whites. The median age of the area has an inverted U-shaped impact on avoidance.

Acknowledgments

This study’s authors were funded by BOTEC Analysis, which is under contract to Altria Client Services.

Altria had no role in the writing of the paper and did not exercise any editorial control.

JEL Codes

C83 (Survey Methods; Sampling Methods); K42 (Illegal Behavior and the Enforcement of Law); C81

(Methodology for Collecting, Estimating, and Organizing Microeconomic Data; Data Access)

Keywords

Black markets, contraband, counterfeit cigarettes, excise taxes, garbology, interstate smuggling, ITTP,

smoking, state and local taxation, tobacco.

I. Introduction

The tobacco market is highly regulated because of harms to health due to smoking and use of oral

tobacco. High taxes and limits on the sale of tobacco products, intended to reduce tobacco

consumption, are adopted worldwide (WHO, 2015). As is often the case, however, regulation may

influence behavior in ways not in accord with the intentions of policymakers (Benham, 2008). Tobacco

taxes and regulation may induce or promote the growth of illicit markets for tobacco products (Prieger

and Kulick, 2018a). In the case of cigarettes, large differences among state excise taxes exacerbate the

problem by making it highly profitable for smugglers to acquire cigarettes in low-tax states and resell

them illicitly in higher-tax states (Pelfrey, 2015). In turn, illicit trade in tobacco products (hereafter

“ITTP”) creates its own detrimental impacts on public welfare, including reduced tax revenues; damage

to the economic interests of legitimate actors; funding for organized-crime and terrorist groups;

negative effects of participation in black markets, such as violence and incarceration; and reduced

effectiveness of smoking-reduction policies, leading to increased harms to health from smoking

(Joossens et al., 2000; State Department, 2015; Kulick, Prieger, and Kleiman, 2016).

Policymakers need to learn about the current state of illicit trade both to evaluate the impacts of

existing policies and to analyze prospectively the outcomes that would result from proposed policy

changes. This requires data on the scale, geographic distribution, and composition of ITTP. However,

despite the manifold criminal, social, and economic consequences of illicit trade, reliable data from illicit

markets generally and ITTP specifically are scarce (Calderoni, 2014). Consequently, tobacco-policy

researchers, health economists, criminologists, and others interested in illicit markets and the

underground economy more generally find themselves with limited data for analysis, for hypothesis

testing, and for offering policy recommendations (Allen, 2014).

We have assembled a large, novel set of data from empty discarded pack (EDP) surveys. The data are

provided by an industry source, Altria Client Services LLC (ALCS).

ALCS has conducted EDP surveys since

2010 in select U.S. markets to estimate the percentage of untaxed and illicit cigarettes consumed in

those markets. Market Survey Intelligence (MSI), a third-party private company that specializes in

ALCS is a subsidiary of Altria Group, Inc., which also holds the major tobacco company Philip Morris USA Inc. (PM

USA). The most popular brand of cigarettes produced by PM USA is Marlboro; other PM USA brands (listed in order

of prevalence in the data) include Parliament, L&M, Benson & Hedges, Virginia Slims, Basic, and Merit. Outside the

U.S., these brands are also produced by Philip Morris International (PMI); Altria Group spun PMI off in 2008,

retaining no ownership.

market research and EDP surveys, executed the collection on behalf of ALCS and analyzed the discarded

packs jointly with ALCS. The purposes of this paper are to describe the data to interested researchers,

analyze the prevalence of ITTP in three surveyed markets in California, and investigate factors (both

causal and correlational) related to the prevalence of ITTP. A more complete description of the

methodology and findings in nine other MSAs is in Aziani et al. (2017).

Given the interest in ITTP in

economics, criminology, and tobacco control, we expect that these new findings will be of great interest

to researchers. The results here are also useful in that they help establish a baseline for ITTP in California

before the recent state excise-tax increase in April 2017.

EDP studies have at least two advantages over other methods of studying ITTP. EDP analysis attempts

to measure the prevalence of illicit activity based directly on actual consumption rather than relying on

reports from individuals on their own illegal behavior. Moreover, since the study area can be tightly

defined, ITTP prevalence can be associated with local neighborhood characteristics (Merriman, 2010;

Davis et al., 2014). This allows the exploration of how economic and social factors affect the supply and

demand of illicit product. EDP studies are also subject to some limitations, most notably the difficulty of

determining the licit or illicit status of some discarded packs.

We use EDP to estimate the level of tax avoidance and a minimum verified level of ITTP, which does

not include inter-jurisdictional bootlegging of taxed cigarettes, product intended for foreign markets, or

brands suspected of being manufactured for illicit sale (i.e., “cheap whites”) if no other evidence of

illegality is found. This approach provides the most conservative estimates of ITTP. Then, by adapting

the methodology developed by Davis et al. (2014), which relies on more liberal assumptions on the

flows of illicit cigarettes within the United States, we provide broader and upper-bound estimates of

ITTP in the San Diego, Los Angeles, and San Francisco markets in the data. We find modest evidence of

tax evasion and avoidance, compared with some other MSAs in Aziani et al. (2017). Verified ITTP is

highest in San Diego (7.4% of packs analyzed) and lowest in Los Angeles (under 1% in all years); San

Francisco is intermediate in all measures. Under broader yet still conservative assumptions, the

prevalence of ITTP rises to 10.7% in San Diego and a maximum of 3.1% in Los Angeles. Under more

aggressive assumptions, we estimate ITTP to be 13.1% in San Diego but still no more than 5.2% in Los

Angeles.

Academic researchers wishing to access the data for some of the markets studied here and for all the MSAs

studied in Aziani et al. (2017) can do so from illicittobaccoinfo.com.

For investigation of ITTP before and after the California tax increase using other data, see Prieger and Kulick

(2018 b, c; 2019).

This paper consists of seven sections. After this introduction, the next section introduces the concept

of ITTP and explains how its study can inform smoking-reduction policies. The third section discusses the

main methods researchers have adopted to study ITTP; it briefly reviews previous studies that have

exploited EDP analysis and discusses strengths and weaknesses of this method. The fourth section

describes the methodology behind the data collection. The fifth section presents estimates of the level

of tax avoidance and ITTP in three California metropolitan areas, with insights about the prevalence and

the composition of ITTP in the different MSAs. The sixth section discusses a regression analysis of local

characteristics associated with the incidence of non-California-tax-paid cigarettes. The final section

discusses the results of the EDP analysis and options for further studies.

II. The Illicit Trade in Tobacco Products

In 2009, Congress enacted the Family Smoking Prevention and Tobacco Control Act, which grants the

U.S. Food and Drug Administration (FDA) the authority to regulate tobacco products to reduce their use.

Such regulation includes restrictions on the manufacturing, marketing, sale, and distribution of tobacco

products (Reuter and Majmundar, 2015); for example, flavored cigarettes other than menthol are

banned (Jo, Williams, and Ribisl, 2015). Tobacco taxation, passed on to consumers in the form of higher

cigarette prices,

is recognized as one of the most effective strategies to deter smoking and, together

with market regulations, is a key component of the smoking-reduction policies of most governments

worldwide (WHO, 2008; Chaloupka, Straif, and Leon, 2010; Bader, Boisclair, and Ferrence, 2011). Taxes

are also intended to raise revenue, both to fund public-health measures that mitigate the harms from

tobacco consumption and other expenditures. Consequently, cigarettes are among the commodities

with the highest tax value by weight (Joossens and Raw 1998; Calderoni, 2014). Taxes, on average,

account for about 44% of cigarette retail price in the United States and exceed 75% of retail in most

European countries (von Lampe, 2011; WHO, 2013; Orzechowski and Walker, 2014).

These restrictions generate criminal opportunities as one of their unintended consequences (Reuter

and Majmundar, 2015). The market for tobacco products is a “dual market,” in which legal and illegal

transactions coexist and interact (Calderoni, Savona, and Solmi, 2012). In such markets, restricting

access to a product through regulation is often met by evasion of the law by consumers, reducing the

efficacy of the regulations and giving birth to other harms (Benham, 2008). In the case of tobacco,

Tobacco tax pass-through to consumers has been measured at slightly more than dollar for dollar (Keeler et al.,

1996; Sullivan and Dutkowsky, 2012; Prieger and Kulick, 2018a).

product bans, high taxes, and large differentials among the tax rates of the U.S. states

spur illicit

markets for banned products and tax evasion (Baltagi and Levin, 1986; Becker, Grossman, and Murphy,

1994; Stehr, 2005; DeCicca, Kenkel, and Liu, 2013; Kleiman, Prieger, and Kulick, 2016; Kulick, Prieger, and

Kleiman, 2016).

ITTP reduces the effectiveness of taxation in protecting health by making available cheaper cigarettes

(Joossens et al., 2000; Stehr, 2005). In some cases involving counterfeit cigarettes, concentrations of

toxins are higher than in licitly produced cigarettes (Stephens, Calder, and Newton, 2005). ITTP also

creates significant losses in tax revenue: estimates on the order of $40 billion globally and between $3.0

and $6.9 billion in the United States are widely accepted (Joossens and Raw, 2008; 2012; Reuter and

Majmundar, 2015). Because ITTP tends to be a low-risk, high-reward criminal activity, traffickers can

make large profits, with less risk of detection or harsh punishments than for other illicit activities with

similar revenue potential (GAO, 2011; Allen, 2014; Pelfrey, 2015). Consequently, organized crime groups

may exploit ITTP for revenues to fund other illicit activities (Joossens et al., 2000; Joossens and Raw,

2008; 2012; OECD, 2016; State Department, 2015). ITTP also damages the economic interests of

legitimate actors in the tobacco industry, distorting the incentives to hire labor and invest in capital, and

reducing state and federal taxes on profits. Like any illicit traffic, it can generate disorder and violence,

enforcement expenditures, and the public costs and private suffering associated with arrest,

prosecution, and incarceration (Joossens et al., 2000; Kleiman, 2010; Green, 2015; Kulick, Prieger, and

Kleiman, 2015). ITTP may also damage the tobacco-control effort indirectly by discouraging decision-

makers from raising taxes and tightening regulations.

ITTP products, modi operandi, and actors vary significantly, depending on criminal opportunities

(Transcrime, 2015). Still, it is possible to identify three main schemes: 1) cigarettes legally manufactured

and sold but then smuggled from lower-tax jurisdictions to higher-tax jurisdictions, whether by large-

scale operators or by casual bootlegging;

2) cheap whites, also known as illicit whites, which are brands

State excise taxes range from $4.50 per pack in Washington, DC, to $0.17 per pack in Missouri (as of January 15,

2019). Local taxes can increase the differential between nearby areas; state and local taxes are $6.16 in Chicago

and $5.85 in New York City, but only $0.995 and $2.60 in nearby Indiana and Pennsylvania, respectively. The

federal excise tax is $1.01 in all locations.

Large-scale smuggling occurs when cigarettes are sold without the payment of any taxes or duties, even in the

jurisdiction of their origin; large-scale smuggling refers to the modus operandi by which it occurs and not the actual

scale of the evasion activity (Joossens et al., 2000; Reuter and Majmundar, 2015). In large-scale smuggling

schemes, cigarettes are usually obtained directly from the manufacturer at factory price (Reuter and Majmundar

2015). Bootlegging indicates the legal purchase of tobacco products in a low-tax jurisdiction and their illegal retail

sale in a high-tax jurisdiction. Therefore, while in large-scale smuggling, no taxes or fees are paid, bootleggers take

produced primarily for illicit markets;

and 3) illegal production, in the form of counterfeit products and

illegal manufacturing

(Joossens et al., 2000; Reuter and Majmundar, 2015; Transcrime, 2015).

In the United States, ITTP mostly consists of bootlegging from low-tax states (and Indian

reservations)

to high-tax states. The dominance of bootlegging has been explained by a relatively

effective external border and customs control, significant interstate tax differentials, and the

preferences of U.S. consumers for domestic brands (DeCicca, Kenkel, and Liu, 2013; Reuter and

Majmundar, 2015). Another factor is the ease of the practice; the organization required to bootleg tends

to be low (Antonopoulos, 2007; Joossens et al., 2009; Calderoni et al., 2014).

The volume of cigarettes smuggled from lower-tax to higher-tax jurisdictions in the United States is

impressive. Recent studies estimate that ITTP accounts for between 8.5% and 21.0% of national

consumption, or between 1.24 and 2.91 billion packs of cigarettes per year (Reuter and Majmundar,

2015). A large empirical literature indicates that price differentials are fundamental in predicting

tobacco smuggling flows in the United States (Baltagi and Levin 1986; Becker, Grossman, and Murphy

1994; Saba et al., 1995; Galbraith and Kaiserman, 1997; Thursby and Thursby, 2000; Stehr, 2005; Chiou

and Muehlegger, 2008; Lovenheim, 2008; DeCicca, Kenkel, and Liu, 2013). Indeed, in high-tax states

such as New York, Arizona, Washington, and New Mexico, ITTP is estimated to have market share as

much as double the national average (Drenkard and Henchman, 2015). Data limitations create

substantial uncertainties about the scale of illicit trade in the United States and elsewhere (Khetrapal

Singh, 2015).

advantage of tax differentials. Bootlegging usually concerns individuals or small groups who smuggle smaller

quantities of cigarettes but it may also entail operations trading truckloads of cigarettes (Hornsby and Hobbs,

2007; Allen, 2014; KPMG, 2014; Reuter and Majmundar, 2015).

Cheap whites are cigarettes legally manufactured in one country, but normally intended for smuggling into

countries where the manufacturer does not hold the permission to sell them. Export from manufacturing countries

may occur legally, and taxes in production countries are normally paid. Import into destination countries, instead,

takes the form of smuggling (Joossens and Raw 2012; Transcrime 2015). The exact definition of cheap whites can

differ in the literature and by law- or tax-enforcement agency (Ross et al. 2015; 2016). Later in the paper we

distinguish between cheap whites and illicit whites, since not all of the former are illicit.

Illegal manufacturing indicates the unlicensed or underreported production of tobacco products, while

counterfeiting indicates the production of branded cigarettes without the permission of the trademark owner

(Allen, 2014; Reuter and Majmundar, 2015).

An Indian reservation is a legal designation in the U.S. for land managed by a federally recognized Native

American tribe. By this designation we loosely also include rancherias, which are small Native American managed

areas in California that are treated similarly to reservations under the law.

III. Approaches to Estimating ITTP

A. Available methods

Several approaches to estimating ITTP are in common use (Kleiman, Prieger, and Kulick, 2015).

Population surveys are conceptually straightforward, but run into difficulties: market participants may

try to hide, may not be available for interview, and have reasons not to be frank in responding to

questions. Indeed, even in anonymous surveys, consumers may be unwilling to disclose their illegal

behavior and interview subjects tend to underreport even their legal purchases of cigarettes and

alcohol, especially when they are socially undesirable.

Moreover, in some cases smokers might not be

sure if all applicable taxes were paid on the cigarettes they bought at legal retail (Merriman, 2002).

Finally, even a survey that provides a representative sample of people may not provide a representative

sample of cigarettes consumed (the appropriate target population to determine the market share of

ITTP).

Gap analysis examines discrepancies between consumption (as reported from surveys) and licit sales;

gaps between the two are attributed to illicit-market sales (see HM Customs & Excise and HM Treasury,

2000 and the supplementary analysis in Prieger and Kulick, 2018). While this method can be

comprehensive, providing estimates for any jurisdiction for which survey and sales data are available, in

the end the estimate reduces to calculating a residual and ascribing it to the illicit market. Of course, the

computed discrepancies between stated consumption and licit sales reflect not only illicit trade but also

survey errors (including sampling error and reporting error), accounting inaccuracies, and any mistakes

in the statistical modeling.

The pack observation/swap survey approach interviews smokers regarding their smoking habits but,

at the end of the interview, the smokers are asked to show their pack of cigarettes or to exchange it for

another. By doing this, researchers are able to collect information on both the smokers and their

actually consumed cigarettes (Reuter and Majmundar, 2015). Some researchers favor this method (GfK

Group, 2006; Gallus et al., 2012; Fix, 2013; Stoklosa and Ross, 2013; Joossens et al., 2014), despite the

This section and the following draw heavily on Aziani et al. (2018).

This phenomenon is an example of social-desirability bias in survey responses (Krumpal, 2013). However, see

Prieger and Kulick (2018b) for survey methods designed to mitigate social-desirability bias regarding ITTP and for

evidence that many California smokers admit to tax avoidance and evasion.

This may be due to inattention or counterfeit tax stamps on the product.

difficulty they face in engaging large and representative samples of smokers (Reuter and Majmundar,

2015).

Other approaches include econometric modeling (Becker, Grossman, and Murphy 1994; Merriman,

Yürekli, and Chaloupka, 2000; Yürekli and Sayginsoy, 2010); analysis of data on product seizures

(Calderoni et al., 2013); estimation of trade gaps (Bhagwati, 1974; Joossens 1998); and expert opinion.

All these methods can lead to different estimates of the size of the illicit tobacco market, at least in part

because they capture different combinations of tax avoidance and evasion (Reuter and Majmundar,

2015).

The International Agency for Research on Cancer (2008) and Reuter and Majmundar (2015)

have analyzed these methods.

There is a growing interest in sewage epidemiology, which is the analysis of wastewater to determine

the consumption of drugs at the community level (Banta-Green and Field, 2011; van Nuijs et al., 2011;

Prichard et al., 2014; Kilmer, Reuter, and Giommoni, 2015). Researchers measure substance and

metabolite concentrations, and can deduce the quantities consumed by the population served by the

sewage-treatment plants (Castiglioni et al., 2006; Zuccato and Castiglioni, 2012; Castiglioni et al., 2014).

Sewage tests might become a powerful instrument to collect epidemiologic information also in the field

of tobacco studies, and a few studies have demonstrated the feasibility of measuring tobacco

consumption this way (Castiglioni et al., 2015; Tscharke, White, and Gerber, 2016). However, no studies

have applied this technique to estimating the prevalence of ITTP, although (coupled with data on local

licit sales) it could enable a gap analysis.

These methods are complementary, but all have shortcomings, some of which can be addressed via

EDP methods, which yield a geographically broad yet granular view of illicit-cigarette consumption. In

EDP studies, teams of researchers are sent out to find all discarded cigarette packs in a defined

geographic area in order to assess the market shares of manufacturers and brands and to measure the

prevalence of ITTP (both non-locally taxed genuine products and counterfeit products). Some studies

For the latter method, see Joossens et al. (2010) for discussion and Prieger and Kulick (2018a) for an application

of using such data.

Tax avoidance involves legal methods of circumventing tobacco taxes, while tax evasion relates to illegal

methods. Tax avoidance is mostly due to individual tobacco users and includes some cross-border, tourist, and

duty-free shopping. For example, in New York state up to 400 cigarettes can be legally brought into the state for

personal consumption without use tax being required in lieu of the excise tax on cigarettes (refer to

tobaccopolicycenter.org/tobacco-control/new-york-state-law/new-york-state-tax-laws-related-to-tobacco-

products). Internet sales to avoid excise taxes are not legal anywhere in the United States. Tax evasion is the

purchase of smuggled and illicitly manufactured tobacco products, in both small and large quantities, and is more

likely to involve criminal offenders (IARC, 2011).

limit collection to littered packs for the sake of convenience, while others (including the present study)

are more comprehensive. Once collected, the packs are analyzed to determine which tax stamps are

present and whether the packs bear other potential indicators of contraband status (e.g., absence of tax

stamp, absence of obligatory health warnings, or cheap-white brands). See Figure 1 for an example of a

California state cigarette-tax stamp. Counterfeit product can be determined by examining packaging or

with sophisticated laboratory analyses. The data described below are from a large set of EDP surveys in

California.

B. Brief review of EDP studies

EDP studies have gained broad currency in the last decade in the United States and Europe since the

seminal study of Lakhdar (2008). Most studies have focused on limited areas with short time frames.

Merriman (2010) analyzed littered packs collected in Chicago and found evidence that proximity to

lower-tax jurisdictions is an important determinant of the share of packs that are missing their required

tax stamps. Collectors gathered cigarette packs from each identified area on a single occasion between

mid-May and mid-June 2007. Almost 2,400 packs were collected, 47.7% of which had cellophane

attached so that tax stamps could be identified if present.

Merriman and Chernick (2011) estimated the prevalence in New York City of packs lacking a New

York State tax stamp before and after the 2009 federal excise-tax increase. They found that packs

without the tax stamp rose from 15% to 24%. Two additional waves of data collection in subsequent

months suggested that the tax-avoidance rate stabilized. In total, four collection waves gathered 1,662

packs (61.1% with cellophane). Kurti, von Lampe, and Thompkins (2012) focused on the illicit market for

cigarette in the Bronx. Their study constitutes one of the first attempts to conduct EDP analysis at a local

level in an area generally known to be rife with ITTP. The study is based on a sample of 497 packs of

cigarettes collected in 30 randomly selected census tracts.

Davis et al. (2014) employed a broader

geographical approach and collected 1,439 discarded packs from a random sample of census tracts in

five northeastern U.S. cities. They found that 58.7% of packs did not have the proper tax stamps, and

In the United States, tax stamps are applied on the cellophane wrapper, so that if a pack has no cellophane it is

generally impossible to determine if state or local excise taxes have been paid (Kurti, von Lampe, and Johnson

2015). In the case of counterfeit cigarettes, it is reasonable to assume that taxes most likely have not been paid,

regardless of the cellophane (although we find a small number of counterfeit packs with apparently genuine

stamps, as discussed below).

U.S. census tracts typically contain 1,200–8,000 people.

that 30.5%–42.1% were ascribable to ITTP, reflecting as much as $700 million annually in lost tax

revenues.

Kurti, von Lampe, and Johnson (2015) conducted a study in the South Bronx to investigate the impact

of a change in New York State tax law on the numbers of untaxed cigarettes bootlegged from Indian

reservations. The study collected 1,737 packs over three waves, 1,111 of which still had their cellophane

wrapping. Researchers found that the

tax amendment introduced in June 2010 in New York State to

limit the sale of untaxed cigarettes by wholesalers to Indian reservations, which was intended to reduce

illicit supply from reservations, affected the distribution of contraband types.

After its introduction,

packs originating from reservations almost disappeared from the sample, while the prevalence of packs

without any tax stamps rose from 18.3% to 66.3%. In related work, Kurti et al. (2018) analyzed a sample

of packs discarded in New York City to find that 23% of the tax stamps were counterfeit and that almost

two-thirds of the counterfeits mimicked the hidden ultraviolet watermark of a genuine stamp.

Consroe et al. (2016) studied a college campus in New York City after spring break, during which

many students were presumed to have traveled out of state (in particular, to warm southern states,

most of which have far lower excise taxes than New York). They refer to their EDP collection method as

“garbology, an archaeological method that reconstructs patterns of human behavior from discarded

materials.” They found that 72.4% of the cigarette packs collected in 2012 and 2013 on the campus

lacked either the state or local tax stamp.

Aziani et al. (2017) analyzed EDP data from ten U.S. metro areas in several waves from 2010 to 2014,

with a total of 106,500 EDPs, of which 72.2% had their cellophane wrapping. The study found a wide

variation in the incidence of packs not fully state-tax-paid in the jurisdiction of collection, from 62.6% in

New York City (of which about one-third were from Virginia) to 4.4% in Minneapolis. In Buffalo, where

50.7% were not New York state-tax paid, about half of those were Native American brands without a tax

stamp and another one-sixth had a Native American tax stamp. The study yielded a range of estimates

The state tax law as amended requires state-licensed stamping agents (i.e., wholesalers) to prepay the cigarette

excise tax and affix tax stamps on all cigarette packs, including those intended for resale to tax-exempt Native

Americans. To account for tribal tax immunity, taxable and tax-free cigarettes sold to tribes or reservation retailers

are distinguished. The tax applies to all cigarettes sold on a reservation to non-members of the Native American

tribe. Thus, when purchasing inventory of taxable cigarettes, tribe or reservation retailers must prepay the tax to

wholesalers. Because the tax does not apply to cigarettes sold to Native Americans for their own use, tribes or

reservation retailers may purchase a limited quantity of cigarettes without prepaying the tax to wholesalers. To

prevent tax evasion, the quantity of untaxed cigarettes wholesaled in such manner is limited to the tribe’s

“probable demand” [paraphrased from the decision in Oneida Nation of New York v. Cuomo, 645 F.3d 154 (2011)].

of the extent of ITTP, with various assumptions on inferring illicit status from limited information. With

the strictest assumptions, ITTP ranged from 31.4% in Buffalo to .3% in Chicago. With the most-

aggressive assumptions, ITTP ranged from 61.1% in New York City in 2011 to below 5% in several years

in Los Angeles, Miami, Minneapolis, and Oklahoma City. The composition of ITTP, under these

assumptions, varied substantially across the markets; overall, in roughly descending order: interstate

bootlegging, no tax stamp (likely also interstate bootlegging), foreign-market, illicit and cheap whites,

counterfeit stamps, and counterfeit product.

There are also several EDP studies conducted in other parts of the world. KPMG (2011 through 2018)

analyzed EDPs collected by the tobacco industry to estimate national-level ITTP in the European Union.

Researchers used ad hoc surveys,

border crossings, and sales data to disentangle tax avoidance and tax

evasion. These results are widely cited but remain controversial; Gilmore et al. (2013) reviewed these

estimates and, by comparing them with data from independent sources, concluded that KPMG’s analysis

inflated the estimates of ITTP. Calderoni (2014) and Transcrime (2015) used KPMG EU data to estimate

ITTP at the subnational level. Rijo and Ross (2017) employed a novel approach to EDP collection in India.

They collected empty packs from retailers selling single cigarettes (which is legal in India, unlike in most

developed countries) in exchange for a small cash reward. They conclude that the incidence of ITTP in

India is lower than suggested by tobacco industry-related estimates.

In Canada, research has focused on discarded butts (unlike in the United States, individual cigarettes

in Canada have distinctive markings that can indicate licit status). The Canadian Convenience Stores

Association (2007; 2008) gathered 11,267 butts near high schools in 2007 and 26,210 in 2008. Butts

were classified as illegal if the butt had no brand, a foreign brand, or an untaxed/native brand; legal if

the brand was considered to be legitimate in the Canadian market; unknown if the butt had no

identifiable marking. According to these studies, ITTP accounts for about the 30% of the consumption of

young Canadians. In 2009, another butt analysis was conducted in Canada to assess young adults’ use of

tobacco from First Nations reserves (native Canadian reservations, with tax-exempt status similar to

Indian reservations in the United States). Discarded cigarette butts were collected from smoking

locations at 12 universities and 13 colleges. Of 36,355 butts collected, 14% were from First Nations

reserves (Barkans and Lawrance, 2013).

C. Strengths and weaknesses of EDP analysis

Unlike other approaches, EDPs allow direct measure of the prevalence of illicit activity (Merriman,

2010) and thus avoid the underreporting of illegal behavior in surveys (Davis et al., 2014). In addition,

since the study area typically is tightly defined, prevalence of ITTP can be linked to local demographic,

social, and cultural characteristics. This allows exploration of how such factors are related to the supply

and demand of illicit tobacco products, thus permitting the design of more focused and effective

counter-ITTP policies (Transcrime, 2015). EDP studies can also identify the location of the last licit

transaction in the sales chain from the tax stamps or other features of the pack, such as the language of

health-warning labels or industry production codes printed on the pack (Merriman, 2010).

EDP surveys also allow for identifying different kinds of untaxed and illicit products, such as foreign

and duty-free cigarettes, counterfeits, and cheap whites. This permits researchers to differentiate

organized ITTP from simple tax avoidance by consumers. Although all types of tax avoidance and evasion

affect revenues and tobacco control, they affect these to different extents and through different

channels. To know the absolute and relative importance of different illicit products, it is crucial to design

better policies and to allocate resources in the most effective way (Stehr, 2005; Transcrime, 2015).

EDP

approaches also yield replicable figures based on a constant methodology, allowing direct comparison

across location and time (Calderoni et al., 2014).

EDP studies are not, however, without limitations (Merriman and Chernick 2011). Some biases might

emerge from the fact that, because of feasibility and cost constraints, the surveys focus on

manufactured cigarettes and exclude other products such as roll-your-own (RYO) tobacco products or

cigars (Calderoni, 2014). Actual biases may emerge depending on which aspect of the market is being

estimated. If the focus of inquiry is illicit retail trade in cigarettes, then illicit activity in the markets for

these other products is of no concern. However, often a broader subset of ITTP is of interest, for

example cigarettes of any kind and close substitutes such as cigarillos and little cigars. If so, then

whenever the incidence of ITTP among manufactured cigarettes and the other products differ, estimates

of the prevalence of overall ITTP from a discarded pack study will suffer from bias. For example,

Joossens et al. (2014) found that, especially in the United Kingdom, the share of illicit products is higher

for RYO tobacco than it is for manufactured cigarettes. In this case, EDP analysis would lead to

underestimation of the level of the ITTP.

For the present data, such biases are likely to be minor. In the United States, RYO tobacco has a

minuscule market share—less than 1% in 2014 and projected to fall further in the next few years. Thus,

it may appear that any biases caused by lack of information on ITTP in this market segment would

necessarily be minimal. However, due to higher taxes on RYO tobacco than pipe tobacco in some states,

consumption of product sold as pipe tobacco may be replacing product explicitly designated RYO

tobacco. Therefore the worst-case scenario, from the standpoint of an EDP study intending to reflect the

market for all cigarettes, would be to consider all RYO and pipe tobacco as ending up in RYO cigarettes.

Two factors may provide opportunities for the smuggling of non-cigarette tobacco products in the

United States. First, Figure 2 shows that the consumption of loose tobacco and cigars was increasing

from 2000 through at least 2013 (see also Centers for Disease Control and Prevention, 2012; Agaku and

Alpert, 2015); second, state taxes on these products vary widely from state to state (O’Connor 2012;

Boonn, 2019). However, only six states explicitly require excise-tax stamps on products other than

cigarettes, thus making it difficult to measure the size of their illicit markets (Chriqui et al., 2015; Reuter

and Majmundar, 2015).

EDP analysis also misses consumption in private residences by concentrating on packs discarded

outside the home. A related issue afflicts studies (unlike the present data

) that examine littered packs

only, because the behavior of litterers with regard to ITTP may systematically differ from that of non-

litterers. However, Merriman (2010) argues that no evidence suggests that littering is limited to any

specific sociodemographic groups. An Australian study found that almost a quarter of the population

littered, with little correlation with gender, age, background or access to trash bins (Williams, Curnow,

and Streker, 1997). Schultz et al. (2013) state that, although the widely accepted conclusions from

observational studies are that littering is more common among men, young people, and in rural

communities, the associations are “far from conclusive.” In their own observational study of general

littering in 130 locations in the United States, however, Schultz et al. (2013) found statistically significant

evidence that younger people littered more than older people and that men littered more than

women.

The association between gender or age and the propensity to litter, even if confirmed, may

not introduce biases into studies of ITTP since these have been shown to be insignificant predictors of

purchasing illicit cigarettes (at least in Europe; Joossens et al., 2014).

In his study, Merriman (2010) compared his main sample of littered packs with a smaller sample

collected from trash bins and found no evidence of systematic discrepancies. He did, however, find that

packs from Newport brand cigarettes (a menthol cigarette disproportionately consumed by African

Americans) are littered out of proportion to their local market share as ascertained from point-of-sale

As discussed below, in the EDP surveys conducted by MSI, attempt was made to collect all discarded packs,

whether in trash containers or on the ground.

Bator, Bryan, and Schultz (2011) also found that younger people and men littered more than older people and

women, respectively, in an observational study of littering in 14 locations in eight states.

scanner data. One less-examined sociodemographic determinant of littering is income. If lower-income

people are both more likely to litter and more likely to engage in ITTP, then the prevalence of ITTP as

calculated from simple proportions in the EDP samples would be overestimated. However, these

correlations have been neither confirmed nor rejected, mainly due to lack of information on the income

of people in observational studies of littering and because of the paucity of individual-level studies of

engaging in ITTP.

Because collection occurs at the local level, the extrapolation of estimates based on EDPs to the

aggregate level is more difficult and may lead to biases (Fix et al., 2013). In particular, EDP analysis may

lead to inflated estimates whenever packs collected in urban areas are used to estimate the level of ITTP

for a larger jurisdiction, which include both urban and rural parts. This is because the concentration of

illicit packs is likely to be higher in larger cities (Gilmore et al., 2013). In the present study, random,

representative sampling from the target geographic area (an MSA or a city) helps avoid this potential

bias. See Aziani et al. (2017), Appendix B, for a careful discussion of the sampling technique and the

resulting representativeness of the samples.

The factors mentioned above that might bias estimates of the prevalence of ITTP would not

necessarily bias the estimates of regression coefficients in structural or analytic studies examining the

determinants of ITTP. With a proper set of control variables and appropriate specification of the

regression function (for example, modeling the scale of ITTP in logs instead of levels, so that the impacts

are in percentage terms), the estimated marginal effects of the regressors on the conditional mean of

the dependent variable could be unbiased even when the scale of ITTP is mismeasured or the

demographic characteristics of the sample do not coincide with those of the population.

A more difficult challenge for EDP and some other studies—one that the present data cannot avoid—

is to distinguish tax avoidance from tax evasion (Merriman, 2010; Reuter and Majmundar, 2015).

Examination of EDPs reveals the prevalence of products lacking state or local tax stamps, which

inevitably include cigarettes legitimately purchased by commuters, tourists, or nationals traveling

abroad. For this reason, the share of products lacking the mandatory stamp should not be considered as

a direct estimate of ITTP, especially in states with high cigarette prices and in regions bordering lower-

Further discussion of these issues regarding survey data, sampling, weighting, structural versus descriptive

modeling, and the bias and consistency of weighted versus unweighted regression is beyond the scope of the

present work. Interested readers are referred to Pfeffermann (1993), section 17.8 of Wooldridge (2002), section

24.4 of Cameron and Trivedi (2005), section 7.2 of Heeringa, West, and Berglund (2010), and Appendix B of Aziani

et al. (2017).

price jurisdictions (Gilmore et al., 2013; Calderoni, 2014). Moreover, in some locations EDP analysis

cannot identify “gray-market” contraband consisting of genuine cigarettes diverted from the legitimate

supply chain and resold within the same jurisdiction (Calderoni, 2014). In this case, smuggled packs may

appear similar to legitimate ones and not be ascribed to ITTP.

The largest disadvantage of the method is its cost. EDP studies are highly time- and labor-intensive,

both to collect the samples and then to conduct careful analysis of the specimens in a lab. This

disadvantage of such studies in general becomes an advantage of the present data set, however: the

costly collection and analysis has already been performed.

Most large-scale EDP surveys, including the present one, are subject to the criticism that cigarette

manufacturers funded the collection of the packs and analysis of the data.

This criticism arises in the

tobacco-control literature, for example, concerning the Project Star and Project Sun reports (KPMG

2011–2015) performed in Europe (Stoklosa and Ross, 2013). Gilmore et al. (2013) assert that the

tobacco industry has an interest in inflating the relevance of the ITTP—counterfeiting, in particular

—to

argue against tobacco-control measures such as high taxation and plain packaging. Such criticisms,

however, commingle distrust of the sampling methodology (which is often not well described) with

suspicion about the subsequent analysis of the data (which is typically complex and often draws on

other data sources and expert opinion). Unlike those reports, however, as part of the present project we

thoroughly describe the sampling plan, discuss its strengths and weaknesses, make available the raw

data, and carefully describe our methodology for arriving at our estimates of the incidence of tax

avoidance and ITTP (see Aziani et al. (2017)). By making the data available to researchers, we therefore

Note that this shortcoming is a greater concern in Europe than in the United States, where any legitimate

product manufactured for export and exempt from excise taxes is required to have a notice on the package

(typically, “U.S. Tax-exempt. For use outside the U.S.”). See ttb.gov/tobacco/tobacco-faqs.shtml. In the present

data, the production codes on the packs have also been examined to determine the intended retail markets for the

product.

We note that the distinction between industry-funded data collection and analysis and that performed by

tobacco-control researchers (inside or outside academia) is not always that of an interested party versus

dispassionate, disinterested investigators. Given the goals of tobacco control, some researchers have an interest in

minimizing the importance of ITTP and particularly its relationship to tobacco taxes (Prieger and Kulick, 2018a).

Pecuniary conflicts of interest may also be found on both sides of tobacco control policy. A U.S. court found that

three anti-tobacco members of an advisory committee to the FDA had financial conflicts of interest (Lorillard Inc.

et al. v. United States Food and Drug Administration, No. 11-440, July 21, 2014 [later vacated on appeal for an

unrelated reason]).

It is worth noting here in passing that the incidence of counterfeit cigarettes in the data examined here, as will

be shown in section V, is relatively low.

allow those who wish to use alternative methods of data analysis to arrive at their own estimates of ITTP

to do so.

When data collection is funded by industry, external validation would be desirable but rarely is

possible (Gilmore et al., 2013). Independent EDP collections may be more transparent, but they usually

rely on smaller samples (Calderoni, 2014), limiting the precision of the estimates. Moreover, the

participation of manufacturers is crucial in the analysis of counterfeit packs, especially in the

identification of the proprietary, hidden security features in the packs (Transcrime, 2015).

However, analysis of EDPs does not suffer from many of the disadvantages of other methods.

Nonresponse or underreporting by subjects is not an issue, as it is with surveys of individuals. Unlike gap

analysis, no demand modeling is required. In contrast to expert opinion, objective and replicable

procedures are employed. Identification of the location of consumption and the category of illicit

product are straightforward. EDP analyses have thus proven to be a fruitful method to study ITTP at the

local level and across different areas (Merriman, 2010; Davis et al., 2014). In the absence of large-scale,

independent data collections, there is limited alternative to the use of industry-sponsored EDP studies

(Transcrime, 2015).

IV. The Empty Discarded Pack Surveys

Altria Client Services has conducted EDP surveys since 2010 in select markets for its own purposes of

assessing brand integrity and the incidence of ITTP. The present authors arranged under a consulting

agreement with ALCS to examine and analyze the data. ALCS’s EDP surveys are executed by a third

party, Market Survey Intelligence (MSI), which specializes in market research and the execution of EDP

surveys.

Table 1 provides a list of EDP surveys commissioned by ALCS in California through 2015. All

surveys rely on a sampling plan intended to collect a representative sample of empty discarded cigarette

packs from the target area and timeframe. Some of the data are available to interested academic

researchers online at illicittobaccoinfo.com

, where users will also find more detailed information about

the variables in the dataset.

MSI, a private company headquartered in Geneva, Switzerland, has executed over 850 EDP surveys in over 70

countries using survey methods developed by its research and development team. See msintelligence.com.

In brief, all packs from public spaces and trash receptacles within the surveyed areas are collected.

Packs are analyzed to determine which tax stamps are present,

whether the package or tax stamps are

counterfeit, where the packs were intended for sale, and whether the packs bear other potential

indicators of contraband status (e.g., foreign or cheap-white cigarettes). The EDPs are also analyzed to

determine whether applicable state tax was paid, when possible; we supplemented the analysis by also

examining whether local excise taxes were paid. Details on the sampling, collection, pack categorization,

and analysis follow in the remainder of this section and in Aziani et al. (2017).

A. Sampling of neighborhoods

ALCS selects markets for EDP surveys based on a number of factors, including historical or recent

contraband activity, upcoming or recent changes in law, and exploratory purposes. Since markets are

not randomly selected, there is no claim to representativeness at anything larger than the market area.

In particular, these data cannot be used to estimate the national incidence of tax avoidance or ITTP in

the United States.

Markets correspond either to the largest city in a Metropolitan Statistical Area

(MSA) or several such cities within an MSA.

The set of markets used for this study includes Los Angeles,

San Diego, and San Francisco. The seven total EDP surveys have been conducted in Los Angeles each

year in 2011–2015, and in San Diego and San Francisco in 2014 (see Table 1). For each selected market,

MSI develops a sampling and collection plan that, while not a true probability sample,

is intended to

result in samples that are representative of the cities composing the market area. The neighborhoods

from which discarded packs are collected are selected at random, with a probability of selection

proportional to the population, and mirror the socioeconomic features of the market to be sampled.

47 U.S. states require tax stamps on cigarettes; in North Carolina, South Carolina, and North Dakota wholesalers

are not required to attach the local tax stamps to cigarette packs before distributing them to retailers to indicate

that taxes have been paid. Similarly, local governments with high cigarette excise taxes, including Chicago and

Cook County, Illinois, and New York City require tax stamps (Reuter and Majmundar 2015; Boonn 2018a).

At least, that is, without model-based extrapolation to non-surveyed markets, which we do not attempt here.

An MSA is a collection of counties composing a metropolitan area with a high population density at its core and

close economic ties throughout the area. Some MSAs contain a single large city that wields substantial influence

over the region (e.g., San Diego), while others contain more than one large city with no single municipality holding

a substantially dominant position (e.g., the Dallas–Fort Worth metroplex). MSAs are defined by the Office of

Management and Budget (OMB) for statistical purposes.

As discussed in Aziani et al. (2017), Appendix B, probability sampling (i.e., a sampling plan resulting in a known

inclusion probability for each observation in the sample) is impossible for EDPs, since there is no frame (i.e., a list

of population units from which to sample) available.

Depending on the MSA, MSI selects the largest city or a collection of the largest cities, as noted in

Table 1. Each city is divided into five mutually exclusive and collectively exhaustive sectors.

The sectors

are arbitrarily defined but designed to have approximately equal population. Sectors may be missing

due to irregular borders or topographical features of the city. As many non-overlapping circular areas of

radius 250 meters as possible are defined within each sector; these are termed neighborhoods.

Neighborhoods are the ultimate sampling cluster and are sampled to create the set of areas to which to

send the collection teams. The number of neighborhoods sampled within each sector depends on its

population. One neighborhood is selected for every 100,000 people in the sector population.

MSI

classifies each neighborhood as residential, industrial, or commercial; this will be referred to below as

the neighborhood’s type. If the initial random selection of neighborhoods does not reflect the

proportion of neighborhood types in the sector, resampling is performed to bring the final set of

neighborhoods more in line with the sector proportions.

The sampling of neighborhoods is completed

once during the market’s initial assessment; once selected, the same neighborhoods are used for all

subsequent collections.

The total number of EDPs to be collected and analyzed in market i at time t, N

, is determined by

budgetary constraints and the desire to have no fewer than about thirty observations per neighborhood

(while respecting the approximate guideline of one neighborhood per 100,000 people). The total

number of EDPs collected in each neighborhood is determined by first apportioning N

to the cities in

the MSA in proportion to population (when the market contains multiple cities). Then the apportioned

number of EDPs for the city is divided proportional to the population of the sector. Finally, the sector’s

allotment of packs is divided by the number of neighborhoods in the sector. An additional small number

of “buffer” packs is also collected to replace packs that do not meet the quality criteria for data

analysis.

Given the sampling scheme, the various areas of the market are represented in approximate

All the cities are subdivided into MSI-defined sectors labeled North, East, South, West, and Center.

Thus if the sector’s population is less than 100,000, collection takes place in one neighborhood, if the sector’s

population is 100,000–200,000, collection takes place in two, and so on. Neither the population counts used by

MSI in these calculations nor the precise boundaries of the neighborhoods are available in the data (although the

ZIP code into which the neighborhood falls is recorded).

For example, if one third of the neighborhoods are in commercial areas in the sector but none happen to be

chosen in the initial random sample, then some randomly chosen residential and industrial neighborhoods in the

initial sample will be discarded and replaced with an equal number of randomly selected commercial

neighborhoods.

A pack must be free of mold, legible, and intact. If a pack does not satisfy these three conditions it is replaced

with a pack from the buffer. Wherever possible, replacements are made using a random selection from among

buffer packs with the same brand. If the same brand is not available, a replacement is made from among buffer

proportion to their population and type of neighborhood. In the statistics presented below, therefore,

the data are not weighted.

Not every pack within a neighborhood is collected. MSI precisely defines two random routes the

collectors are to take and prepares detailed collection instructions for each sampled neighborhood.

Within each neighborhood, the routes correspond to two circuits; the first one covers the center of the

area (roughly the inner, centered circle of diameter 250 meters), while the second traverses a peripheral

route through the remainder of the neighborhood.

The second, outer route may not be used if enough

packs are collected from the inner route (as explained in the next subsection). For some example routes,

see Figure 3. While the details of the creation of the “random walks” are proprietary, our examination of

the neighborhood route maps revealed that in most cases there are only a few alternative routes at

most in any event, given the tight boundaries of the inner and outer parts of the neighborhoods.

Geographic coverage within the neighborhood is therefore typically fairly comprehensive and randomly

determined.

Finally, the specific neighborhoods, dates, and times of collections are checked to ensure that people

visiting the area do not unduly distort the distribution of packs. In particular, collection locations do not

include airports, sporting venues, or major tourist attractions; neighborhoods that include such

extraordinary features are replaced with neighborhoods of the same type during the sampling process.

If the survey schedule would coincide with an extraordinary event in the neighborhood (e.g., parades,

street fairs, or planned demonstrations), the collection is postponed until after the event.

B. Collection of packs

At least two collectors are assigned to each neighborhood.

Collectors are instructed to complete

their entire route and to pick up all discarded cigarette packs observed on streets and readily accessible

public trash receptacles on the route, irrespective of the brand, country of origin, or whether the

cellophane is intact. Collection takes place any time of day, including weekends, and in all weather

conditions. Collectors move along the predefined routes, beginning with the inner route in the

packs within the same brand family. If there are no buffer packs in the same brand family, a replacement is made

from among buffer packs from the same manufacturer. If there are no buffer packs from the same manufacturer,

then a randomly chosen buffer pack is used.

I.e., through the annulus with inner diameter 250 meters and outer diameter 500 meters.

MSI extensively trains the collectors and sends the same teams to the various markets. Before each collection

survey, a pilot is performed with the collectors during which the supervisor demonstrates all of the steps in the

collection process and ensures that the collectors understand the protocol.

neighborhood. If at the end of the first route the supervisors determine that the collectors have not

gathered the minimum number of packs desired from that neighborhood, the collectors walk through

the outer route. The collectors do not know the objective of the collection nor the quotas; only the

supervisors know this information. Collectors always pick up all packs along an entire route.

In the

(rare) event that they have not gathered enough packs after completing the second circuit, they restart

with the first circuit on another day.

Collectors place packs within a labeled bag for the neighborhood. After collection from a route is

complete, the neighborhood bag is turned in to the supervisors. The supervisors randomly pull packs

from the bag, until the quota and buffer-pack requirements for the neighborhood are met with the

specified number of adequate-quality packs. Pack quality is deemed inadequate if the pack is severely

damaged (e.g., large portions of the pack are missing), illegible, or excessively dirty, wet, or moldy. If the

quota is not met, the collectors return to walk the next route in the neighborhood as described above.

Once the quota is met, the supervisor places the complete set of EDPs in the neighborhood bag,

discarding any surplus packs except those in the buffer, and seals and labels the bag. The neighborhood

bags from each market are then delivered for cleaning and analysis.

Each pack is cleaned,

flattened, and placed into a “pack bag.” A sticker is placed on each pack bag

with a barcode that links the sample to the date, city, sector, and neighborhood of collection. Buffer

packs collected during the survey are maintained separately as replacements for packs that fail to meet

the acceptance criteria during the cleaning, data entry, and final pack screening processes.

C. Analysis of packs

The physical analysis of the packs is performed by MSI, with supplemental analysis by ALCS as

described below. The manufacturer, brand, and presence of the cellophane wrapper on collected packs

are recorded. Packs with cellophane are then examined for the current or former presence and

This avoids the potential problem of collectors “cherry-picking” packs that are easy to find because they know

they need to collect only a few more packs to make quota for the neighborhood. Such selective collection may

introduce unknown biases into the sample since littered packs would be easier to spot than packs discarded in

trash receptacles (refer to the discussion of the behavior of litterers in section III.C above).

The cleaning process involves first removing the cellophane wrapper (to which any tax stamp is affixed) and

removing any dirt or mold on the pack. Doing so prevents the pack from deteriorating while in storage or transit

prior to categorization or analysis. Because moisture can degrade quality, any wet packs that are collected are

dried as part of the cleaning process. The cellophane wrapper is placed back on the pack after cleaning.

A pack could fail in the final screening for counterfeit packs and stamps if, for example, moisture created mildew

on the pack while it was stored awaiting analysis.

jurisdiction of tax stamps. Stamps from most states have a chemical marker in them that allows

researchers to determine if there was previously a stamp on the pack even when it is physically missing.

If analysis of the cellophane detected the chemical marker from a stamp, the status is recorded as if a

stamp were present.

The next phase of analysis involves counterfeit tax stamps and cigarette packs. Screening is

performed on all tax stamps. MSI performs an initial analysis, based on training provided by ALCS, using

confidential and non-confidential authentication techniques to determine if the stamp is counterfeit.

California has “high-tech” counterfeit-resistant tax stamps with features such as holographic or

encrypted images, color-shifting dyes, tamper-evident surface cuts, and unique serial numbers (CDC,

2016). Other evidence of counterfeit stamps includes stamps with inconsistent coloring, poor imaging,

crooked placement on the pack, or that were affixed with adhesive tape (NJ OCI, 2014).

Unlike the screening of tax stamps, inspection to detect counterfeit manufacture is performed only

on packs that are Philip Morris USA (PM USA) or Philip Morris International (PMI)

brands (whether or

not the wrapper is present).

Per ALCS, genuine packs contain multiple features used to detect

counterfeits.

Some of the testing procedures are available only for PM USA brands.

Methods known

from the literature to be used by industry to identify counterfeits include visual assessment, ultraviolet

irradiation of the pack, and chemical analysis of the packaging (Kurti, He, von Lampe, and Li, 2017).

The chemical marker typically does not identify which state a stamp was from. When the state cannot be

identified, the state is recorded with code 99 in variable TAX_STAMP_STATE. Per staff at ALCS, often times there is

enough residual stamp remaining once the presence of the chemical marker has been found to allow identification

of the state from which the stamp originated, and in such cases the actual state is recorded.

As with many efforts to combat counterfeiting and fraud more generally, some methods of detection remain

confidential to law enforcement and industry actors so that counterfeiters do not make their products harder to

detect.

See footnote 1.

There are a few exceptions in cases where the counterfeiting was obvious. In the 2012 Los Angeles market, two

packs of brand Parliament purportedly produced for a non-domestic market by PMI (marked for sale in duty-free

zones in Korea) were marked as counterfeit, for reasons unknown to us.

The comment in footnote 39 applies here as well.

This means that ALCS does not test Marlboro (or other branded) packs produced by PMI as thoroughly as

Marlboro packs produced by PM USA. In particular, the test for the presence of a chemical taggant in genuine

packaging (mentioned below) is performed only on packs purportedly produced by PM USA.

Visual inspection looks for low-quality printing on the pack, generally due to the use of offset printing instead of

the more expensive and higher-quality gravure printing. The quality differences may be difficult to detect with the

naked eye but show up readily under a microscope. Fluorescence under UV radiation signals counterfeit status

because counterfeit packaging often contains optical brightening agents (OBAs) to conceal the use of low-quality

paper; legitimate packages manufactured by PM USA and some (but not all) other manufacturers have no OBAs.

OBAs fluoresce a blue light. Industry sources inform us that the lack of OBAs has become a less reliable indicator of

MSI recorded its preliminary determination of counterfeit status and submitted its data and all collected

packs to ALCS, which performs its own laboratory analyses to verify the authenticity of the packs.

Other data elements for each pack are also recorded and linked to the pack by scanning the barcode

label on the bag containing the pack. Data entered include: market, sector, ZIP code, neighborhood

location, date of collection, UPC, pack production code, tax-stamp serial number, fluorescence status,

and original-design market (although not all of these are necessarily available for any given pack). See

Aziani et al. (2017), Appendix A for a list of the variables MSI and ALCS entered for each pack; see also

the codebook available online at illicittobaccoinfo.com

D. Categorization of packs

Based on the analysis of the packs, the status of each pack can be categorized regarding taxes not paid,

counterfeit product, and contraband. MSI and ALCS categorized packs in various ways in the different

waves of the survey. In this section we describe their schema; our own categorization and analysis is

presented in section V. Their main grouping, recorded in the variable CF_GROUP,

has the following

categories: 1) no cellophane (so that the tax status cannot be determined from the presence or absence

of tax stamps); 2) correct state tax stamp (local tax status is ignored);

3) incorrect state tax stamp;

counterfeit tax stamp; 5) cellophane intact but no tax stamp; 6) counterfeit pack; and 7) product

intended for a non-domestic market (including U.S. duty free sales).

The absence of an applicable state tax stamp (categories 3 and 5) does not necessarily indicate illicit

trade, because the presence of a discarded pack in a location does not mean it was sold in that

jurisdiction, nor does it indicate how it was transported into the collection area (Ross, 2015).

The tax-

paid status of packs without cellophane cannot be determined,

although if the pack were intended for

a foreign market it can be assumed federal, state, and local taxes were not paid even if the film is

missing from the pack.

genuine product over the years, since OBAs are now sometimes absent from higher-quality counterfeit packages.

In addition, incorrect production codes (or codes known by industry to be copied by counterfeiters) printed on

packs signal counterfeit manufacture (Kurti, He, von Lampe, and Li, 2015).

Variable names refer to variables in the Excel and Stata files containing the EDP data.

See DeLong et al. (2016) for details on state-tribal compacts and intergovernmental agreements regarding

tobacco excise taxes.

Packs with tribal or Native American stamps other than those mentioned in the previous footnote fall into this

category.

MSI makes no determination regarding whether brands are certified for sale in the states of collection pursuant

to the Master Settlement Agreement (in effect in 46 states) and associated state laws.

But see footnote 15.

Since the measures used to protect the brand integrity of PM USA products are a trade secret,

ALCS

does not make available the exact relationships between certain indicators of counterfeits found in the

data and the placement into category 6 (counterfeit pack) of CF_GROUP. However, the relationships can

be examined from the data presented for domestic-market Marlboro packs in Table 2. The data show

that whenever a chemical taggant incorporated into genuine PM USA product packaging was detected,

the pack was deemed genuine regardless of the other indicators.

Whenever a Marlboro pack

fluoresces under UV radiation,

it was deemed to be counterfeit. In the remaining cases (i.e., the pack

did not fluoresce and the taggant either was not detected or the test results are not available) the

counterfeit status was completely determined by whether the PM USA production code printed on the

pack is one known by ALCS to be copied by counterfeiters.

While these results give some insight into

how counterfeit packs may be detected, we emphasize that these apparent “rules” are merely inferred

from the data, and that in any event ALCS informed us that it may conduct additional tests for brand

integrity that are not captured by any of the indicators.

Special care must be taken when using these data to investigate the prevalence of counterfeiting in

the California tobacco market. Because analysis to detect counterfeits is conducted only on PM USA and

PMI brands, the simple prevalence of counterfeits out of the entire sample is a downward-biased

estimate of the share of all counterfeit product in the cigarette market. Conversely, the counterfeiting

rate for PM USA or PMI brands cannot be simply extrapolated to other brands. Marlboro brands, which

are produced by PM USA and PMI, are the world’s most counterfeited (WCO, 2014). In 2014, Marlboro

alone accounted for the 45% of the counterfeit cigarettes seized worldwide (WCO, 2015), while

accounting for only 7.2% of world market share (Euromonitor, 2016). Seizures made worldwide testify to

the counterfeiting of other legitimate brands and even of cheap whites (WCO, 2013). In the absence of

further information to adjust the estimates for the United States, calculating the incidence of

See footnote 39.

In two cases not included in Table 2 (because some of the indicators of counterfeit status were missing), field

TAGGANT_STATUS has value “failed” but ALCS changed MSI’s categorization from “counterfeit domestic” to

“correct tax stamp” in variable CF_GROUP, from “contraband” to “applicable tax paid” in variable CATEGORY, and

from “counterfeit” to “genuine” in variable PACK_STATUS. In such cases, additional testing by ALCS verified that

these two packs were genuine and that the taggant field reflected a false negative. Detection of the taggant on the

packaging can be impaired if the pack is wet, mutilated, etc.

See footnote 46.

There are several cases where a positive taggant presence trumps a production code on the counterfeit list,

leading to a designation as not counterfeit.

counterfeits for the inspected manufacturer would likely lead to overestimating the incidence of

counterfeit packs in the market as a whole.

Packs of cheap whites are identified by MSI only in the 2011 EDP samples.

We supplemented the

data with an expanded list of brands of cheap whites from Ross et al. (2015b). Although cheap whites

are produced mainly for purposes of illicit sale, in some cases in the sample a cheap white pack has a

valid tax stamp. We therefore distinguish between cheap whites and illicit whites hereafter, with the

former depending only on the brand. In our determination, two conditions have to hold simultaneously

to label a pack of cigarettes as illicit whites: 1) the brand is identified with cheap whites and 2) the

wrapper is intact but there are no tax stamps. Given the contentious nature of the definition of cheap

white brands in the literature (Ross et al., 2015, 2016),

we separate cheap whites based on whether

the brand was determined as such by multiple sources (which must include an academic source) or by

industry sources alone.

The list of brands designated as cheap whites is in Table 3.

MSI and ALCS created a summary field with their final determination of contraband status. The field

CATEGORY takes values 1) applicable tax paid, 2) applicable tax not paid (ignoring local taxes, if any are

applicable), 3) contraband (counterfeit product or counterfeit tax stamp), 4) non-domestic,

and 5)

undetermined due to lack of cellophane.

The correspondence between the categorical variables

CF_GROUP and CATEGORY is shown in Table 4.

In summary, the categorization of packs by MSI and ALCS with fields CF_GROUP and CATEGORY

provides a useful starting point for statistical analysis. There are a few possible shortcomings with the

categorizations. First, the dimensions of tax status and counterfeiting are collapsed to a single

dimension in these variables, at the cost of some potentially useful information. There are a few cases in

The brands identified as cheap whites by MSI are Chunghwa, Double Happiness, Esse, and Hatamen. Regarding

these, see footnote 70 in section V.B below. We identified several other brands based on other sources, as

described below.

Ross et al. (2016) suggest that the traditional tobacco companies have incentive to label cheap-white brands as

illicit even when they are not since they create lower-price competition for traditional brands. Ross et al. (2016)

find that the illicit status of cheap-white brands is variable, sometimes even within the same country of sale,

although the preponderance of brands examined had neither the appropriate tax stamp nor the required health

warnings in at least one country of sale.

The relevant variables in the dataset are CHEAP_WHITE (MSI’s determination), A_CHEAP_WHITE_BRAND (our

determination based on the list of brands in Table 3), and A_CHEAP_WHITE_SOURCE (the source of the

designation of the brand as a cheap white).

This category includes product intended for U.S. and worldwide duty-free sales.

See footnote 15.

which the pack is counterfeit but bears a valid tax stamp appropriate to the jurisdiction.

Second, there

are unexplained discrepancies in the categorizations as noted above, in table notes, and in footnotes,

although such discrepancies are very rare. Most importantly, however, the category contraband

undoubtedly misses a huge amount of activity involving large-scale bootlegging of genuine product,

which should be considered contraband as well.

The next section contains statistical analysis of the

data based on our own categorizations of the packs. Future researchers can use these data to explore

alternative categorizations as desired.

E. How representative are the surveys?

The potential for the sample of packs collected through EDP surveys to differ from the actual population

of packs smoked was discussed above in section III.C. If the brands or types of packs collected differed

substantially from those consumed in the population, then estimates of ITTP based on the surveyed

packs could be biased. For example, as mentioned above, Marlboro is the most counterfeited brand

(WCO, 2014), and if the surveys had an unduly high proportion of Marlboro packs then the prevalence of

counterfeiting in the survey could be higher than the actual prevalence in the local markets.

To assess how representative is the distribution of cigarette brands in the surveys, we compared the

brand shares with external data from the same markets, albeit from a slightly later date. A large-scale

survey of California smokers conducted in early 2017 provides estimates of brand shares for the markets

found in the EDP surveys. The reference survey data are weighted to reflect the packs consumed by the

population of California adult smokers.

The comparisons are in Table 5.

Across the three markets, the brand shares of Marlboro do not differ (at the 5% significance level)

between the EDP and reference surveys. Thus, the prevalence of this important brand appears to be

accurate in the EDP surveys. There are some statistically significant differences in shares for other

brands, and the overall equivalence of the EDP and reference survey brand share distribution is rejected

It may be that counterfeiters reuse valid tax stamps from previously sold packs (which would be illegal).

In addition, we cannot rule out the possibility of false-negative results: counterfeit packs that pass scrutiny and

are classified as genuine.

In particular, the target population of the external survey was California smokers aged 18 to 74 who are literate

in English. For details on the survey, see Prieger and Kulick (2018b). The person-level weights in the reference

survey are adjusted using the self-reported number of cigarettes smoked each day so that the estimates pertain to

brand shares.

The standard errors for the differences in the proportions are calculated as follows. The EDP surveys are treated

as stratified by MSA, clustered by neighborhood, and each pack is equally weighted. The observations from the

external reference survey are weighted but otherwise treated as unstratified and unclustered. For discussion of

inference involving the EDP data, which is problematic, see Appendix B of Aziani et al. (2017).

in each MSA. However, the differences appear to be idiosyncratic; among brands with at least five

percent market share in any survey, only Parliament has a consistent direction of the shares differences

across the cities (that brand always has more market share in the EDP survey than in the reference

survey). Thus, the comparison reveals no obviously troubling sign of consistent overrepresentation of

important targets of counterfeiting in the EDP surveys.

V. Descriptive Analysis

A strength of this new database is that it allows for a rigorous and geographically broad study of ITTP.

Packs have been collected between one and five times each in ten different metropolitan areas, using a

constant methodology. Overall, the database includes 32,000 discarded packs—5,000 per wave in Los

Angeles and 3,500 in San Diego and San Francisco—of which 72.3% have cellophane.

The analysis presented here is based on study of the EDPs. First, we estimate the combined

prevalence of tax avoidance and tax evasion, where the latter is defined as packs lacking genuine

California tax stamps. Then, we examine ITTP by enumerating packs whose provenance is nearly

certainly illicit (counterfeits, illicit whites, and U.S. packs without any tax stamps

) and those of

questionable legality (U.S. packs with non-California tax stamps and foreign packs).

While counterfeits, illicit whites, and U.S. packs without any tax stamp are nearly certainly part of

ITTP, it is not possible to confirm the illicit origin of other packs lacking tax stamps without further

information or assumptions. Our first estimate of ITTP in this study does not include packs with the

wrong U.S. tax stamp or foreign packs. Consequently, it is a conservative, lower-bound estimate.

To account for bootlegging and other forms of ITTP, we also relax the previous assumptions to

provide broader estimates. In addition to counterfeits, illicit whites, and packs with no stamp, some

packs from nonadjacent states are included in these estimates (as proposed by Davis et al. (2014)), as

well as some foreign packs, intrastate bootlegging to evade local taxes (e.g., packs with an Illinois stamp

but lacking the Cook County/Chicago stamp), and cheap whites.

Packs with no stamp may originate from the three U.S. states that do not use tax stamps (i.e., North Carolina,

North Dakota, and South Carolina), Internet sales (illegal since 2010 in the U.S.) or Indian reservations (Davis et al.,

2014). We follow the convention in the literature of ascribing any U.S. pack without a tax stamp to ITTP (e.g., Davis

et al., 2013; Wang et al., 2018), despite the fact that packs coming from North Carolina, South Carolina, and North

Dakota do not have stamps even if licitly purchased.

All averages in this section are at the market level. We do not calculate averages across the entire

sample because the sample does not reflect any well defined target population. In particular, averages

across the entire sample would not yield unbiased estimates of ITTP for the nation as a whole.

A. The prevalence of tax avoidance and tax evasion

The database supports estimation of the proportion of cigarettes consumed in each area that are

missing the required tax stamps. For packs with cellophane, we deem that the California excise tax was

not paid if any of the following are true: 1) there is no stamp, 2) there is a counterfeit stamp, or 3) the

stamp is for other than CA.

Additionally, packs lacking the cellophane wrapper but satisfying one of the

following conditions are also deemed tax unpaid: 4) the intended market is non-domestic,

or 5) the

pack is counterfeit.

Our determination is recorded in variable A_STATE_TAX_PAID, and all statistics in

this section regarding California tax compliance are for the subsample of packs for which it can be

determined whether the California tax was paid. In this subsection we make no distinction between

avoidance and evasion of taxes.

Overall, the data confirm that the proportion of cigarettes for which appropriate state excise taxes

were not paid varies greatly across markets (Table 6) and years (Table 6). San Diego has the highest

prevalence of non-California-taxed cigarettes in the panel; 19.6% of packs lack a valid California tax

stamp; the same holds for 8.1% of the packs collected in Los Angeles and 15.6% in San Francisco.

The

difference in the prevalence of tax avoidance and evasion among the California markets is modest,

compared with differences nationwide (in Aziani et al. (2017), for instance, non-state-taxed cigarettes

account for 62.9% in New York City, 16.1% in Chicago, and 4.7% in Minneapolis

We follow MSI’s convention regarding the tribal stamps discussed in footnote 46. We treat other tribal stamps as

if they were stamps from an incorrect state.

We limit this condition to packs without cellophane because in a few cases packs intended for non-domestic

markets have a genuine, appropriate tax stamp. These may be gray-market packs that were reimported to avoid

federal excise taxes. Non-domestic packs with cellophane otherwise fall under condition 1.

We limit this condition to packs without cellophane because in a few cases counterfeit packs had genuine,

appropriate tax stamps. Counterfeit packs with cellophane otherwise fall under conditions 1 or 2.

The share of packs on which state tax has been paid is calculated by dividing their number by the number of

packs for which it was possible to determine if taxes were paid. The set of years over which the averages here are

calculated differs by market; refer to Table 1 for the years each market was surveyed.

Minnesota has the third-highest average price per pack of cigarettes ($8.40) after New York State ($10.45) and

Massachusetts ($9.10); however, it is among the states with the smallest increase in taxes in recent years (+1.0%

between 2006 and 2013) (Drenkard and Henchman, 2015; Boonn, 2016b).

Where do untaxed out-of-state cigarettes in the CA markets come from? Figure 4 shows the source

locations for packs that are not properly state taxed (based on the tax stamp). Of the untaxed packs

collected in Los Angeles, 3% come from Nevada (where the excise tax is $1.80 since 2015, when it was

raised from $0.80, compared with California’s tax of $0.87, until the raise to $2.87 on April1, 2017);

about 10% of packs were intended for domestic markets but lack a genuine stamp and so are of

unknown origin,

or come from other states or tribal areas. A full 82% come from abroad (5% from

China and 4% from Vietnam).

The picture is quite different in San Diego, also shown in Figure 4. 24% of untaxed packs come from

Mexico, 28% are non-U.S. duty free, 41% are from other or unknown states, and 2% are from a nearby

Indian reservation. And San Francisco differs substantially, as well. Only 17% of untaxed packs are

domestic, with 19% from China, 9% from Armenia, 7% from Mexico, 3% from Vietnam, and 42% non-

U.S. duty free.

In Los Angeles, the MSI EDP surveys have been conducted in multiple years, thus allowing for

assessment of changes in the consumption of non-taxed cigarettes. In Los Angeles, where tax

compliance was already relatively high in the first survey, the non-state-taxed proportion decreased by

12% from 2011 to 2015, although not monotonically. Aziani et al. (2017) found both increases and

decreases in five other MSAs surveyed, with no clear secular trends.

B. The prevalence of cheap whites

The data show that cheap whites are relatively rare in California (and in the other U.S. MSAs cited in

Aziani et al. (2017)), as opposed to in Europe where they are more prevalent (Joossens and Raw, 2012;

Ross et al., 2016). Table 7 shows that only a tiny fraction of packs are cheap whites as defined by

industry. Furthermore, almost all packs so designated are Asian brands that, while inexpensive, do not

appear on lists of cheap-white brands composed by researchers outside of the tobacco industry.

Only

in San Diego do cheap whites compose one percent of the sample. Such small numbers of cheap whites,

See footnote 63 regarding unstamped packs.

Over four-fifths of the packs designated as cheap whites in the sample are Chunghwa, Double Happiness, or Esse

brand. Chunghwa and Double Happiness are produced by the Shanghai Tobacco Group, a subsidiary of China

National Tobacco Corporation (the state tobacco monopoly in China) and Esse is produced by the KT&G, the

leading tobacco company in South Korea. These brands are not manufactured with the sole purpose of being sold

illicitly abroad, and therefore would not seem to satisfy the common definition of cheap whites. However, Scollo et

al. (2014) note that the list prices for these brands are very low and that they may be being sold illicitly in Australia.

Note that designation as a cheap white brand is not enough to be included in our category of verified ITTP in the

next table.

and their non-negligible presence only in MSAs close to international borders (San Diego) and the

second-most visited U.S. city by foreigners (Los Angeles),

suggest that most cheap whites may be

carried into the country as the personal property of travelers from abroad. Such small-scale importation

of cheap whites from abroad is not necessarily illicit, since travelers can bring up to 200 cigarettes (10

packs) into the United States duty free if they are properly declared to customs officials and are

intended for personal use.

The fraction of cheap-white brands identified only from industry varies greatly between the MSAs

(see columns three and four of Table 7). Given the minuscule presence of cheap whites in the sample,

the distinction does not appear to be important in CA (or in the MSAs studied in Aziani et al. (2017)).

C. The prevalence of illicit tobacco products

Understanding the consumption of non-taxed cigarettes provides information to improve tobacco-

control policies and estimates of revenue losses due to cross-border purchases. A broader

understanding of ITTP dynamics involves law-enforcement costs, violence related to illicit markets,

funding of criminal and terrorist organizations, etc. (Joossens et al., 2000; Stehr, 2005; Shelley and

Melzer, 2008; Perri and Brody, 2011; FATF, 2012; Prieger and Kulick, 2014).

Given the indeterminacy of the licit status of some of the packs, we present a range of estimates of

the incidence of ITTP. Our most conservative estimate of ITTP, which we refer to as verified ITTP,

includes counterfeits and domestically produced cigarettes lacking a genuine tax stamp. This estimate

also includes all cheap whites without a genuine tax stamp (i.e., illicit whites) as ITTP, regardless of

where they were produced or their intended market.

This methodology does not, however, include in

ITTP any pack with genuine tax stamps differing from the jurisdiction where the pack has been collected

or foreign cigarettes (apart from illicit whites). While verified ITTP is a very conservative estimate of ITTP

and probably greatly undercounts the scale of illicit activity (as will be shown below), the measure is not

Thirty percent of all visitors to the United States list New York as their destination city, while Los Angeles has

about a 12% share (NTTO, 2015).

Refer to the U.S. Customs and Border Protection information available at help.cbp.gov/app/answers/detail/a_id/

53/~/traveler-bringing-tobacco-products-(cigarettes,-cigars,-bidis)-to-the-u.s.-for.

Thus illicit whites are the only foreign-market packs included in verified ITTP.

necessarily a lower bound on ITTP.

The results of this definition are stored in variable A_ITTP1 in the

database.

The first columns of Table 8 show the proportion of verified ITTP out of determinable packs. The

same information as in Table 8 is also depicted graphically in Figure 5.

Comparison of the verified ITTP shares shows a modest variation among the MSAs, at 7.4% in San

Diego, 3.0% in San Francisco, and less than 1% in Los Angeles in all years (in Aziani et al. (2017), ITTP

accounts for at least 31.4% of the market in Buffalo and at least 13.7% in New York City, with all the

other MSAs under 2.5%).

The definition of verified ITTP necessarily misses much smuggling of genuine product. Much

contraband product will bear a valid stamp from a lower-tax state. Tax evasion composes an unknown

fraction of the figures in the next three columns of Table 8: interstate cross-border purchases and

bootlegging, intrastate cross-local-border purchases and bootlegging, and product intended for foreign

markets. These three categories mix tax evasion and tax avoidance, the latter of which breaks no laws.

The final column in these tables show the fraction of fully tax-paid product intended for the domestic

market, which is the only column certain not to contain any ITTP. After accounting for verified ITTP, tax

avoidance, and tax evasion, 80.2% of packs in San Diego, 83.8% in SF, and 91.8% in Los Angeles are fully

tax-paid genuine product (compared with 30.5% in Chicago, 34.3% in New York City, and 49.3% in

Buffalo (Aziani et al. (2017)).

Genuine cigarettes originating from low-tax states (an indicator of bootlegging) contribute minimally

to the potentially illicit category in California, at 1.3% in San Diego, .7% in Los Angeles, and .7% in San

Francisco. Foreign cigarettes make 6.9% of the market in Los Angeles, 11.1% in San Diego, and 12.5% in

San Francisco.

In particular, stamps may have been removed from or fallen off the wrapper and a pack designated as illicit

whites may have been purchased legally in a foreign location for personal use and either properly declared to U.S.

Customs and Border Protection or included under the 200-cigarette personal-use exemption. The former appears

to be a rare occurrence; in 2014 MSI added an indicator for the presence of a remnant of a tax stamp, and less

than .5% of packs fell into that category. Regarding illicit whites, the small number of cheap whites in the sample

implies that the method of dividing cheap whites into licit and illicit categories affects the statistics on ITTP very

little. Finally, see also footnote 63 on unstamped packs.

But recall that brands other than PM USA’s brands destined for domestic markets are not checked for

counterfeiting. Thus, there may be some counterfeit packs of other brands in the fully tax paid domestic category

in the table.

We now explore broader—and more realistic—definitions of ITTP, with results shown in Figure 6 and

Table 9. The first column shows the prevalence of verified ITTP as in the previous tables. The second

column of results in the tables contains the more conservative of the two broader measures of ITTP.

Under the conservative assumptions, all verified ITTP, one-quarter of packs coming from bordering

states,

three-quarters of packs coming from nonadjacent states, and one-quarter of foreign-market

packs are treated as illicit.

With these conservative assumptions, ITTP is measured at 10.7% in San

Diego, 6.6% in San Francisco, and 2.7% in Los Angeles (see Table 9).

Under more aggressive assumptions the definition of ITTP can be further broadened. In the third

column of results in Table 9, it is assumed that all verified ITTP, three-quarters of packs coming from

bordering states,

all packs coming from nonadjacent states, half of packs coming from within the state

but not locally taxed, one-half of foreign-market cigarettes, and all cheap whites are illicit. The impact of

broadening the definition of ITTP is similar in the three MSAs, rising to 13.7% in San Diego, 10.0% in San

Francisco, and 4.8% in Los Angeles (compared with from 61.1% in New York City to under 5% in Miami,

Minneapolis, and Oklahoma City in Aziani et al. (2017)).

Table 9 shows the upper bound for ITTP. We calculate the upper bound as including all verified ITTP

and all other packs except for those bearing genuine state and local stamps.

The upper bound, so

calculated, is unrealistically high as an estimate of ITTP since it assumes that no out-of-state or foreign-

market packs are licit. I.e., the upper bound assumes there is no licit tax avoidance but only illicit tax

evasion. The upper bound on ITTP is 19.9% in San Diego, 16.2% in San Francisco, and 8.2% in Los

Angeles.

This assumption is the low-range assumption adopted by Davis et al. (2014).

A recent (unscientific) survey found that 22% of Americans try to sneak goods through customs (Agence France-

Presse, 2015). A recent (unscientific but weighted to be demographically representative) poll of UK travelers found

that about half (46%) of those who bootlegged brought cigarettes into the country (Gocompare.com, 2014; we

could not find any comparable poll for the U.S. market). Multiplying these two figures yields an estimate of

approximately 10% of people bootlegging cigarettes through customs, which would imply that roughly half of

smokers bootleg (surveys from Eurobarometer and the Centers for Disease Control, respectively, indicate that the

prevalence of smoking was about 22% in the UK and 17% in the US in 2014). If bootleggers bring in more foreign

cigarettes on average than those who declare them at customs, then the share of illicit foreign-market cigarettes

would be even higher than one-half. On the other hand, some smuggled cigarettes may just as well have been

declared (since they were under the personal-use importation limit), and while technically illicit no duty would

have been imposed on them anyway.

This assumption is the high-range assumption adopted by Davis et al. (2014).

Despite our term for this category, it is not a true upper bound for ITTP since the base for the proportions is

packs with intact wrappers. If packs lacking wrappers are more likely to be illicit, the share of illicit packs could be

even higher than the figures in our “upper bound.”

D. The composition of ITTP

In this section, ITTP is split into its constituent components to see which forms of illicit behavior are

most prevalent in the markets for cigarettes. Table 10 shows the composition of verified ITTP. In San

Diego, the overwhelming majority (97.0%) of verified ITTP takes the form of packs lacking any tax stamp.

In San Francisco, these account for about half (50.7%), with illicit whites at 34.3% and counterfeit packs

at 15.1%. In Los Angeles, the distributions vary hugely by survey year; overall, illicit whites account for

42.1% of verified ITTP, packs lacking any tax stamp at 33.3%, and counterfeit packs at 23.8%. (In most

MSAs in Aziani et al. (2017), the majority of verified ITTP takes the form of packs lacking any tax stamp,

illicit whites compose no more than 8% anywhere except in the Texas markets, and counterfeit product

makes up no more than 6% in any market.) Counterfeit tax stamps placed on otherwise genuine product

make up .8% of verified ITTP in Los Angeles, and zero percent in San Diego and SF (compared with 21%

in Chicago and 59% in New York City (Aziani et al. (2017)). Counterfeiting tax stamps has a dual purpose;

it reduces the risk of being identified by law-enforcement agencies and it may allow smugglers to resell

their products at a higher price since they appear to be legitimate (at least to the casual observer). In

some cases, packs with counterfeit stamps are sold at the regular price in retail shops to unaware

consumers who do not get the benefit of an “illicit discount” (Campbell, 2015; Silver et al., 2016).

The elements of the broader measures of ITTP are presented in Table 11. The three CA markets are

each distinctive. In Los Angeles, foreign-market illicit packs dominate, with interstate bootlegging

following; in San Diego, packs without a stamp (which is mostly likely another form of interstate

bootlegging) dominated, with foreign-market illicit packs following; and in San Francisco, foreign-market

illicit packs dominate, with packs without a stamp following. (In Aziani et al. (2017) the rough order of

importance of the various methods of ITTP across the markets is interstate bootlegging (in first place by

a large margin), packs without a stamp, foreign-market illicit packs, illicit and cheap whites, counterfeit

stamps, and counterfeit product.)

All three California MSAs surveyed are the most diverse illicit markets for cigarettes in the United

States (compared with the markets in Aziani et al. (2017)), as measured with the Shannon diversity

index.

Compared to other markets, counterfeit product, illicit whites and cheap whites, and foreign

illicit are more prevalent in the California MSAs. Together with the long distance from the eastern states

The Shannon index of diversity (or entropy) is calculated as −

∑







(





)

, where 



is the market share of an

individual component of ITTP.

with the cheapest cigarettes, trade and travel between California and China might explain the

concentration of these products.

Indeed, more than 80% of illicit whites collected in Los Angeles and

San Francisco are manufactured in China. Relative proximity to China might be part of the explanation

for the greater prevalence of counterfeits. China is the largest source country of counterfeit cigarettes

and in the United States counterfeit cigarettes have been traced back, in particular, to China, North

Korea, and Paraguay (Shen, Antonopoulos, and von Lampe, 2010; WCO, 2013). Even so, the California

MSAs have low overall levels of ITTP, even when potentially bootlegged cigarettes are included in the

calculations.

VI. Regression Analysis of Tax-Paid Status

We turn now to logit regressions of whether the California excise tax was not paid for a pack.

For

convenience the dependent variable will be referred to as “tax avoidance,” although as defined it also

includes evasion. The unit of observation is a single pack, and only the 23,131 packs for which a

determination can be made are included in the regressions. Summary statistics on the variable used in

the regressions are in Table 12. The logit coefficients (i.e., the log odds ratios) are reported in Table 13

and the average marginal effects in the sample are in Table 14. Estimation 1 includes pack-specific

regressors only. Brand-indicator variables for the top three brands and for Philip Morris brands other

than Marlboro are included. The results show that, compared to brands other than these, tax avoidance

is between 15 to 21 percentage points less likely for Marlboro, Camel, and Newport packs; tax

avoidance is also less likely for other Philip Morris brands (by 6 p.p.). This is because some minor brands

have relatively high proportions of tax-avoiding packs.

Tax avoidance is much more likely for Native

American brands (13 p.p.) and especially cheap-white brands (32 p.p.).

In Estimation 2, a regressor MaxPrDiffPerHr is added to examine the impact of the price difference

between average cigarette prices in the MSA and the lowest average price in neighboring states,

Despite the fact that the Los Angeles MSA covers only about 4% of the nation’s population, almost 12% of U.S.

residents traveling to China in 2012 by air were from Los Angeles (OTTI, 2013).

With less than 2% of the sample being verified ITTP, we do not similarly investigate the determinants of ITTP via

regression, given the well known bias that can attend logit regression (or other standard methods for binary

dependent variables) of rare events (King and Zeng, 2001).

For example, among brands represented by at least 100 packs in the sample, the following brands have a tax-

paid incidence of less than 75%: Craven A, Davidoff, Sheriff, and USA Gold. The first three are non-domestic

brands.

Average prices for California MSAs are calculated from prices paid for last pack smoked in the Tobacco Use

Supplement of the Current Population Survey, 2010–11 and 2014–15 waves. Prices for other years were

calculated per hour of driving time.

As found in many other studies (e.g., Baltagi and Levin, 1992;

Chiou and Muehlegger, 2008; Coats, 1995; DeCicca, Kenkel, and Liu, 2013; Goel, 2008; Lovenheim, 2008;

Stehr, 2005; Sung, Teh-Wei, and Keeler, 1994), prices in other states matter a lot. A dollar increase per

hour of driving time in the price differential is estimated to increase tax avoidance by 49 p.p. (although

such a large increase is a large extrapolation; the actual range of MaxPrDiffPerHr in the data is only

about 14 cents/hour).

We also explore whether the local prevalence of licensed cigarette retailers is related to tax

avoidance. Where access to fully taxed cigarettes is less convenient because retailers are farther away,

untaxed or illicit substitutes to fully taxed product may become relatively more attractive. If so, then

density of tobacco outlets would be negatively associated with intended tax evasion and ITTP. On the

other hand, if licensed retailers are also engaged in covert sales of untaxed product (as evidence from

other states indicates sometimes is the case),

density would be positively associated with ITTP. The

density of licensed cigarette retailers (CigRetailDen) in the ZIP code was found to have a nonlinear effect

on tax avoidance.

The results show that in most areas, CigRetailDen has no discernible effect on tax

avoidance. However, in the top 10% of areas with the most cigarette retailers, CigRetailDen is positively

and significantly associated with avoidance. These results may indicate evidence of illicit sales through

retail stores, although further investigation is warranted.

Note that we include local population density

interpolated. Prices for neighboring states are from Tax Burden on Tobacco reports (Weighted Average Price Per

Package, with generic brands included). Interpolation of TUS data for the other states is not appropriate since

excise taxes changed during this period (unlike in California). On the other hand, using only TBOT prices would not

reflect price differences among the MSAs in California. Since the TBOT prices were systematically higher than those

from the TUS, a conversion factor was applied to the TUS prices (calculated as the ratio of statewide average last-

pack prices from TUS to the TBOT weighted-average price). All prices are converted to real terms (January 2015

dollars) using the CPI-U.

MaxPrDiffPerHr = max

∈

{(price difference with out of state location i) ÷ (shortest driving time to location i)}. Set

L contains many addresses just inside the borders of contiguous states along major border crossings from

California. The driving times are the minimums over routes calculated from the ZIP code of the EDP collection area

to location i. Travel times were taken from Google Maps via Google’s Distance Matrix API.

There are many news reports in recent years of bodegas in New York City that are licensed to sell cigarettes but

also are caught selling untaxed tobacco (e.g., Kochman, 2015).

Data on licensed retailers were provided upon request from the California Board of Equalization, and pertain to

May 2017. Due to obligations to protect confidentiality, the BOE list does not include licenses issued to individuals.

Density is calculated as the number of licensed retailers in the ZIP code divided by the square mileage of the

associated census ZCTA.

We have evidence that tax-evaded sales occur through otherwise legitimate retail outlets from other research as

well. Our survey described in Prieger and Kulick (2018b) reveals that, in early 2017, 13.7% of smokers in California

bought untaxed cigarettes in the last month, and an additional 9.1% said that “I’m not sure, but I suspect that

some cigarettes I bought were not taxed.” Of those two groups composing 22.8% (95% CI = [75.8,78.5]) of the

population, 34.8% [31.9,38.0] said they bought the (potentially) untaxed cigarettes at “a mainstream store that

in the regression as well, to ensure that the retailer density does not merely serve as a proxy for

population.

Finally, we add several demographic variables known in the literature to be related to crime and tax

compliance at the individual level. These include income, race and ethnicity, education, and age. Since

we are unable to associate a pack with the demographics of the person who discarded it, we must

instead use characteristics of the ZIP code area in which the pack was collected. Note also that the

person discarding the pack may not live in the area. Perhaps it is for these reasons that only a few of the

demographic variables have significant coefficients, and we focus here on those (see Estimation 3 in

Table 13 and Table 14 for the full set of results). Extreme levels of income were found to exert great

influence on the estimated coefficient for income. Therefore a three-part linear spline was used to split

the impacts of the bottom and top 10% of the income distribution from the middle 80% of the

distribution. Setting the extremes of the distribution aside (neither coefficient is significant), the impact

on tax avoidance of log income in the middle of the distribution is strongly negative. In that range, a one

log point increase in income is associated with a 13 p.p. decrease in tax avoidance. The fraction of the

population that is non-Hispanic Black has a negative effect on tax avoidance, compared to the omitted

race/ethnicity category of non-Hispanic Whites. This result may be related to the fact that, at least in

some contexts, Blacks in California have been found to be more likely than others to support anti-

smoking policies (Unger et al., 1999). The median age of the area was found to have an inverted U-

shaped impact on avoidance, with the peak at an age of 40.0. None of the other demographic

coefficients, including those for education, have any significant effects.

In a final regression, Estimation 4 in Tables 13 and 14, all regressors are included. The results are

similar to those discussed above, except for the following. The impact of a dollar’s price difference with

the neighboring states, while still significant, falls to 36 p.p. per hour of travel. The impacts of retailer

density and income (in its main range) are also somewhat smaller than in the previous estimation. The

significance of the retailer and main income spline coefficients falls to the 10% level.

also sells fully taxed cigarettes.” These estimates are for the population of adult smokers in the state who are

literate in English.

VII. Discussion and Conclusions

This study describes the methodology of EDP surveys and illustrates their potential by presenting a

series of estimates of the level of tax avoidance and ITTP in three metropolitan areas in California. The

review of the literature suggests that EDP surveys have several advantages over alternative data sources

and methods, which generally do not allow for equally broad estimates, geographically and temporally,

and have other shortcomings (confounding factors complicate gap analysis, underreporting makes

consumer surveys suspect, etc.).

Our analysis confirms that the incidence of cigarette-tax avoidance and ITTP in three of California’s

largest MSAs is low compared with some other MSAs (especially so in Los Angeles), but distinctive in the

distribution of non-state-taxed types and with considerable variation among the three MSAs. Foreign

cigarettes, including illicit and cheap whites, duty free, and foreign-tax paid, are especially prevalent in

the California markets, as are counterfeits (likely of foreign origin). The variations within California, and

between California and the rest of the country, merit further investigation of these and other data. Our

regression analysis finds that the local incidence of tax avoidance is associated with smoking prevalence,

population density, and income inequality.

Future research will include econometric analysis of how ITTP relates to tobacco-control policies,

such as the elasticity of illicit consumption with respect to taxation rates, especially as California’s excise

tax increased by $2.00 per pack in 2017. Whether the relatively low incidence of tax avoidance, evasion,

and ITTP in California found in the present work will survive the tax increase will be of intense interest to

policymakers in the state and elsewhere. For now, such investigation awaits further EDP collection after

the tax increase.

References

Agaku, I. T., & Alpert, H. R. (2015). Trends in Annual Sales and Current Use of Cigarettes, Cigars, Roll-

Your-Own Tobacco, Pipes, and Smokeless Tobacco among US Adults, 2002–2012. Tobacco Control

25 (4): 451–457. doi:10.1136/tobaccocontrol-2014-052125

Agence France-Presse (AFP). (2015). Americans Most Likely to Steal Toiletries, Germans to Cheat on

Holiday: Survey. Retrieved January 9, 2019 from: msn.com/en-ca/travel/news/americans-most-

likely-to-steal-toiletries-germans-to-cheat-on-holiday-survey/ar-BBlbtnm

Allen, E. (2014). The Illicit Trade in Tobacco Products and How to Tackle It. 2nd Ed. International Tax and

Investment Center. Retrieved January 9, 2019 from: aph.gov.au/DocumentStore.ashx?id=6057af72-

80e2-47f5-99c5-afbfe4ca85c0&subId=512929

Antonopoulos, G. A. (2007). Cigarette Smugglers: A Note on Four “Unusual Suspects.” Global Crime 8 (4):

393–398. doi:10.1080/17440570701739769

Aziani, A., Kulick, J., Norman, N., & Prieger, J. E. (2017). Empty Discarded Pack Data and the Prevalence

of Illicit Trade in Cigarettes. SSRN Scholarly Paper ID 2906015. Retrieved from: ssrn.com/

abstract=2906015. doi:10.2139/ssrn.2906015

Bader, P., Boisclair, B., & Ferrence, R. (2011). Effects of Tobacco Taxation and Pricing on Smoking

Behavior in High Risk Populations: A Knowledge Synthesis. International Journal of Environmental

Research and Public Health 8 (11): 4118–4139. doi:10.3390/ijerph8114118

Baltagi, B. H., & Levin, D. (1986). Estimating Dynamic Demand for Cigarettes Using Panel Data: The

Effects of Bootlegging, Taxation and Advertising Reconsidered. The Review of Economics and

Statistics 68 (1): 148–155. doi:10.2307/1924938

Baltagi, B. H., & Levin, D. (1992). Cigarette Taxation: Raising Revenues and Reducing Consumption.

Structural Change and Economic Dynamics 3 (2): 321–335. doi:10.1016/0954-349X(92)90010-4

Banta-Green, C., & Field, J. (2011). City-Wide Drug Testing Using Municipal Wastewater. Significance 8

(2): 70–74. doi:10.1111/j.1740-9713.2011.00489.x

Barkans, M., & Lawrance, K. (2013). Contraband Tobacco on Post-Secondary Campuses in Ontario,

Canada: Analysis of Discarded Cigarette Butts. BMC Public Health 13 (335): 1–8. doi:10.1186/1471-

2458-13-335

Bator, R. J., Bryan, A. D., & Schultz, P. W. (2011). Who Gives a Hoot? Intercept Surveys of Litterers and

Disposers. Environment and Behavior 43 (3): 295–315. doi:10.1177/0013916509356884

Becker, G. S., Grossman, M., & Murphy, K. M. (1994). An Empirical Analysis of Cigarette Addiction. The

American Economic Review 84 (3): 396–418.

Benham, L. (2008). Licit and Illicit Responses to Regulation. In C. Menard and M. M. Shirley (Eds.),

Handbook of New Institutional Economics, 591–608. Berlin: Springer-Verlag.

Bhagwati, J. N. (1974). In J. Bhagwati (Ed.), Illegal Transactions in International Trade: Theory and

Measurement, 1: 138–147. Amsterdam: North-Holland.

Boonn, A. (2018a). State Excise and Sales Taxes per Pack of Cigarettes Total Amounts & State Rankings.

Campaign for Tobacco-Free Kids. Retrieved January 9, 2019 from: tobaccofreekids.org/research/

factsheets/pdf/0202.pdf

Boonn, A. (2018b). Top Combined State-Local Cigarette Tax Rates. Campaign for Tobacco-Free Kids.

Retrieved January 9, 2019 from: tobaccofreekids.org/research/factsheets/pdf/0267.pdf

Boonn, A. (2019). State Excise Tax Rates For Non-Cigarette Tobacco Products. Campaign for Tobacco-

Free Kids. Retrieved January 9, 2019 from: tobaccofreekids.org/research/factsheets/pdf/0169.pdf

Calderoni, F. (2014). A New Method for Estimating the Illicit Cigarette Market at the Subnational Level

and Its Application to Italy. Global Crime 15 (1-2): 51–76. doi:10.1080/17440572.2014.882777

Calderoni, F., Angelini, M., Aziani, A., De Simoni, M., Rotondi, M., & Vorraro, A. (2014). Lithuania. The

Factbook on the Illicit Trade in Tobacco Products, Issue 6. The Factbook. Trento: Transcrime -

Università degli Studi di Trento. Retrieved from: transcrime.it/wp-content/uploads/2014/05/

LITHUANIA_EN_v5.pdf

Calderoni, F., Aziani, A., & Favarin, S. (2013). Poland. The Factbook on the Illicit Trade in Tobacco

Products, Issue 4. Trento: Transcrime – Universita degli Studi di Trento. Retrieved January 9, 2019

from: transcrime.it/wp-content/uploads/2013/11/Factbook-poland_eng2.pdf

Calderoni, F., Savona, E. U., & Solmi, S. (2012). Crime Proofing the Policy Options for the Revision of the

Tobacco Products Directive: Proofing the Policy Options under Consideration for the Revision of EU

Directive 2001/37/EC against the Risks of Unintended Criminal Opportunities. Trento: Transcrime -

Joint Research Centre on Transnational Crime.

Cameron, A. C., & Trivedi, P. K. (2005). Microeconometrics. New York: Cambridge University.

Campbell, J. (2015). Smuggled, Untaxed Cigarettes Are Everywhere in New York City. Village Voice. April

7. Retrieved January 9, 2019 from: villagevoice.com/news/smuggled-untaxed-cigarettes-are-

everywhere-in-new-york-city-6717621

Canadian Convenience Stores Association. (2007). Cigarette Butt Study. Retrieved from:

stopcontrabandtobacco.ca/wp-content/themes/stopcontrabandtobacco/pdf/buttstudy2007.pdf

Canadian Convenience Stores Association. (2008). Youth Contraband Tobacco Study, 2008. Retrieved

January 9, 2019 from: studylib.net/doc/11558641/canadian-convenience-stores-association-youth-

contraband-...

Castiglioni, S., Borsotti, A., Riva, F., & Zuccato, E. (2014). Illicit Drug Consumption Estimated by

Wastewater Analysis in Different Districts of Milan: A Case Study. Drug and Alcohol Review 35 (2):

128–132. doi:10.1111/dar.12233

Castiglioni, S., Senta, I., Borsotti, A., Davoli, E., & Zuccato, E. (2015). A Novel Approach for Monitoring

Tobacco Use in Local Communities by Wastewater Analysis. Tobacco Control 24 (1): 38–42.

doi:10.1136/tobaccocontrol-2014-051553

Castiglioni, S., Zuccato, E., Crisci, E., Chiabrando, C., Fanelli, R., & Bagnati, R. (2006). Identification and

Measurement of Illicit Drugs and Their Metabolites in Urban Wastewater by Liquid

Chromatography-Tandem Mass Spectrometry. Analytical Chemistry 78 (24): 8421–8429.

doi:10.1021/ac061095b

Centers for Disease Control and Prevention (CDC). (2016). STATE System Tax Stamp Fact Sheet March

31. US Department of Health and Human Services. Retrieved January 9, 2019 from: data.cdc.gov/

download/gxi9-gfcw/application/pdf

Centers for Disease Control and Prevention (CDC). (2012). Consumption of Cigarettes and Combustible

Tobacco—United States, 2000–2011. Morbidity and Mortality Weekly Report 61 (30): 565–569.

Chaloupka, F. J., Straif, K. & Leon, M. E. (2010). Effectiveness of Tax and Price Policies in Tobacco Control.

Tobacco Control 20 (3): 235–238. doi:10.1136/tc.2010.039982

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.

doi:10.2202/1935-1682.2027

Chriqui, J., DeLong, H., Gourdet, C., Chaloupka, F., Edwards, S. M., Xu, X., & Promoff, G. (2015). Use of

Tobacco Tax Stamps to Prevent and Reduce Illicit Tobacco Trade—United States, 2014. Morbidity

and Mortality Weekly Report. 64 (20): 541–546.

Coats, R. M. (1995). A Note on Estimating Cross-Border Effects of State Cigarette Taxes. National Tax

Journal 48 (4): 573-584.

Consroe, K., Kurti, M., Merriman, D., & von Lampe, K. (2016). Spring Breaks and Cigarette Tax

Noncompliance: Evidence From a New York City College Sample. Nicotine & Tobacco Research 18

(8): 1773-1779. doi:10.1093/ntr/ntw087

Davis, K. C., Grimshaw, V., Merriman, D., Farrelly, M. C., Chernick, H., Coady, M. H., Campbell, K., &

Kansagra, S. (2014). Cigarette Trafficking in Five Northeastern US Cities. Tobacco Control 23 (e1):

e62–68. doi:10.1136/tobaccocontrol-2013-051244

DeCicca, P., Kenkel, D., & Liu, F. (2013). Excise Tax Avoidance: The Case of State Cigarette Taxes. Journal

of Health Economics 32 (6): 1130–1141. doi:10.1016/j.jhealeco.2013.08.005

DeLong, H., Chriqui, J., Leider, J., & Chaloupka, F. (2016). Common State Mechanisms Regulating Tribal

Tobacco Taxation and Sales, the USA, 2015. Tobacco Control 25 (Suppl 1): i32–i37. doi:10.1136/

tobaccocontrol-2016-053079

Drenkard, S., & Henchman, J. (2015). Cigarette Taxes and Cigarette Smuggling by State, 2013.

Washington: Tax Foundation. Retrieved January 9, 2019 from: taxfoundation.org/article/cigarette-

taxes-and-cigarette-smuggling-state-2013-0

Elgin, C., & Oyvat, C. (2013) Lurking in the Cities: Urbanization and the Informal Economy. Structural

Change and Economic Dynamics 27: 36–47. doi.org/10.1016/j.strueco.2013.06.003

Euromonitor International. (2012). RYO Tobacco and the New Age of Total Tobacco. Research report

retrieved February 10, 2016,from the Passport database.

Euromonitor International. (2016). Brand Shares (Global - Historical Owner), Historical, Retail Volume, %

breakdown. Data retrieved June 13, 2016 from the Passport database.

Financial Action Task Force (FATF). (2012). Illicit Tobacco Trade. Annual Report. FATF Guidance. Paris.

Retrieved January 9, 2019 from: fatf-

gafi.org/publications/methodsandtrends/documents/illicittobaccotrade.html

Fix, B. V., Hyland, A., O’Connor, R. J., Cummings, K. M., Fong, G. T., Chaloupka, F., & Licht, A. S. (2013). A

Novel Approach to Estimating the Prevalence of Untaxed Cigarettes in the USA: Findings from the

2009 and 2010 International Tobacco Control Surveys. Tobacco Control 23 (Supp. 1): i61–i66.

doi:10.1136/tobaccocontrol-2013-051038

Galbraith, J. W., & Kaiserman, M. Kaiserman. (1997). Taxation, Smuggling and Demand for Cigarettes in

Canada: Evidence from Time-Series Data. Journal of Health Economics 16 (3): 287–301.

doi:10.1016/s0167-6296(96)00525-5

Gallus, S., Ugo, A., La Vecchia, C., Boffetta, P., Chaloupka, F. J., Colombo, P., Currie, L, et al. (2012).

PPACTE, WP2: European Survey on Smoking. Dublin: PPACTE. Retrieved January 9, 2019 from: tri.ie/

uploads/3/1/3/6/31366051/

european_survey_on_economic_aspects_of_smoking_wp2_ppacte.pdf

Geocompare. (2014). 1 in 5 UK Holidaymakers Admit to Smuggling! [Press release]. Retrieved January 9,

2019 from gocompare.com/press-office/2014/05/smuggling-release

GfK Group. (2006). Tobacco Product Illicit Trade Phenomena: National Study for Imperial Tobacco

Canada. Mississauga, Ontario: GfK Group, Imperial Tobacco Canada. Retrieved January 9, 2019

from: ocat.org/pdf/ctmc2006study.pdf

Gilmore, A. B., Rowell, A., Gallus, S., Lugo, A., Joossens, L., & Sims, M. (2013). Towards a Greater

Understanding of the Illicit Tobacco Trade in Europe: A Review of the PMI Funded “Project Star”

Report. Tobacco Control 23 (1): e51–e61. doi:10.1136/tobaccocontrol-2013-051240

Green, F. (2015). Cigarette Trafficking Spawning Other Crimes and Possibly Violence. Richmond Times-

Dispatch. March 28. Retrieved January 9, 2019 from: richmond.com/news/local/crime/

article_e101477f-1c3d-5117-bcce-f8839f52485c.html

Heeringa, S. G., West, B. T., & Berglund, P. A. (2010). Applied Survey Data Analysis. Boca Raton, FL:

Chapman Hall/CRC.

HM Customs & Excise, and HM Treasury. (2000). Tackling Tobacco Smuggling. Retrieved January 9, 2019

from: mhlw.go.jp/shingi/2006/03/dl/s0302-3g4.pdf

Hornsby, R., & Hobbs, D. (2006). A Zone of Ambiguity. British Journal of Criminology 47 (4): 551–571.

doi:10.1093/bjc/azl089

International Agency for Research on Cancer (IARC). (2008). Methods for Evaluating Tobacco Control

Policies: Measures to Assess the Effectiveness of Tobacco Taxation. Lyon: International Agency for

Research on Cancer. Retrieved January 9, 2019 from: iarc.fr/wp-

content/uploads/2018/07/Tobacco_vol12.pdf

International Agency for Research on Cancer (IARC). (2011). Effectiveness of Tax and Price Policies for

Tobacco Control. 14. IARC Handbook of Cancer Prevention. Lyon, France: IARC. Retrieved January 9,

2019 from: iarc.fr/wp-content/uploads/2018/07/handbook14-0.pdf

Jo, C. L., Williams, R. S., & Ribisl, K. M. (2015). Tobacco Products Sold by Internet Vendors Following

Restrictions on Flavors and Light Descriptors. Nicotine & Tobacco Research 17 (3): 344–349.

doi:10.1093/ntr/ntu167

Joossens, L. (1998). Tobacco Smuggling: An Optimal Policy Approach. In I. Abedian, R. van der Merwe, N

Wilkins, and P. Jha (Eds.), The Economics of Tobacco Control: Towards an Optimal Policy Mix, 146–

154. Cape Town: Applied Fiscal Research Centre, University of Cape Town.

Joossens, L., & Raw, M. (1998). Cigarette Smuggling in Europe: Who Really Benefits? Tobacco Control 7

(1): 66–71. doi:10.1136/tc.7.1.66

Joossens, L., & Raw, M. (2008). Progress in Combating Cigarette Smuggling: Controlling the Supply Chain.

Tobacco Control 17 (6): 399–404. doi:10.1136/tc.2008.026567

Joossens, L., & Raw, M. (2012). From Cigarette Smuggling to Illicit Tobacco Trade. Tobacco Control 21

(2): 230–234. doi:10.1136/tobaccocontrol-2011-050205

Joossens, L., Chaloupka, F. J., Merriman, D., & Yürekli, A. (2000). Issues in the Smuggling of Tobacco

Products. In F.J. Chaloupka and P. Jha (Eds.), Tobacco Control in Developing Countries, 393–406.

Oxford: Oxford University.

Joossens, L., Lugo, A., La Vecchia, C., Gilmore, A. B., Clancy, L., & Gallus, S. (2014). Illicit Cigarettes and

Hand-Rolled Tobacco in 18 European Countries: A Cross-Sectional Survey. Tobacco Control 23: e17–

e23. doi:10.1136/tobaccocontrol-2012-050644

Joossens, L., Merriman, D. Ross, H., & Raw, M. (2009). How Eliminating the Global Illicit Cigarette Trade

Would Increase Tax Revenue and Save Lives. Paris: International Union Against Tuberculosis and

Lung Disease (The Union). Retrieved January 9, 2019 from:

tobaccofreekids.org/assets/global/pdfs/en/ILL_global_cig_trade_full_en.pdf

Joossens, L., Merriman, D., Ross, H., & Raw, M. (2010). The Impact of Eliminating the Global Illicit

Cigarette Trade on Health and Revenue. Addiction 105 (9): 1640–1649. doi:10.1111/j.1360-

0443.2010.03018.x

Keeler, T. E., Hu, T., Barnett, P. G., Manning, W. G., & Sung, H. (1996). Do Cigarette Producers Price-

Discriminate by State? An Empirical Analysis of Local Cigarette Pricing and Taxation. Journal of

Health Economics 15 (4): 499–512. doi:10.1016/s0167-6296(96)00498-5

Khetrapal Singh, P. (2015). Stop Illicit Trade of Tobacco Products. World Health Organization. Retrieved

January 9, 2019 from: searo.who.int/mediacentre/features/2015/stop-illicit-trade-of-tobacco-

products/en

Kilmer, B., Reuter, P. H., & Giommoni, L. (2015). What Can Be Learned from Cross-National Comparisons

of Data on Illegal Drugs? Crime and Justice 44 (1): 227–296. doi:10.1086/681552

King, G., & Zeng, L. (2001). Logistic Regression in Rare Events Data. Political Analysis 9 (2): 137–163.

Kleiman, M. A. R. (2010). When Brute Force Fails: How to Have Less Crime and Less Punishment.

Princeton: Princeton Univ.

Kleiman, M. A. R., Prieger, J. E., & Kulick, J. (2015). Illicit Trade as a Countervailing Effect: What the FDA

Would Have to Know to Evaluate Tobacco Regulations. Journal of Drug Policy Analysis 9(1): 1–30.

doi: 10.1515/jdpa-2015-0016

Kochman, B. (2015). Investigators bust Bronx crew that sold thousands of untaxed cigarettes smuggled

in from Virginia. New York Daily News. July 9. Retrieved January 9, 2019 from nydailynews.com/

new-york/nyc-crime/investigators-bust-bronx-crew-sold-untaxed-cigarettes-article-1.2287306 on

January 1, 2019.

KPMG. (2011). Project Star 2010 Results. KPMG. Retrieved January 9, 2019 from:

pmi.com/resources/docs/default-source/pmi-sustainability/star-report-

2010.pdf?sfvrsn=1f02b0b5_0

KPMG. (2012). Project Star 2011 Results. Project Star. KPMG. Retrieved January 9, 2019 from

yumpu.com/en/document/read/28927039/project-star-2011-results-philip-morris

KPMG. (2013). Project Star 2012 Results. Project Star. KPMG. Retrieved January 9, 2019 from:

yumpu.com/en/document/read/34237015/conducted-by-kpmg-philip-morris

KPMG. (2014). Project Sun. A Study of the Illicit Cigarette Market in the European Union 2013 Results.

KPMG. Retrieved January 9, 2019 from: pmi.com/resources/docs/default-source/pmi-

sustainability/sun-report-2013.pdf?sfvrsn=0

KPMG. (2015). Project Sun. A Study of the Illicit Cigarette Market in the European Union, Norway and

Switzerland 2014 Results. KPMG. Retrieved January 1, 2017 from:

kpmg.co.uk/creategraphics/2015/06_2015/CRT026736/files/assets/basic-html/index.html

KPMG. (2016). Project Sun. A Study of the Illicit Cigarette Market in the European Union, Norway and

Switzerland 2015 Results. KPMG. Retrieved January 9, 2019 from:

assets.kpmg/content/dam/kpmg/pdf/2016/06/project-sun-report.pdf

KPMG. (2017). Project Sun. A Study of the Illicit Cigarette Market in the European Union, Norway and

Switzerland 2016 Results. KPMG. Retrieved January 9, 2019 from: pmi.com/resources/docs/default-

source/pmi-sustainability/project-sun-2017-report.pdf?sfvrsn=cdf89ab5_6

KPMG. (2018). Project Sun. A Study of the Illicit Cigarette Market in the European Union, Norway and

Switzerland 2017 Results. KPMG. Retrieved January 9, 2019 from: pmi.com/resources/docs/default-

source/pmi-sustainability/sun-report-2017-executive-summary.pdf?sfvrsn=bfc59ab5_2

Krumpal, I. (2011). Determinants of Social Desirability Bias in Sensitive Surveys: A Literature Review.

Quality & Quantity 47 (4): 2025–2047. doi:10.1007/s11135-011-9640-9

Kulick, J., Prieger, J. E., & Kleiman, M. A. R. (2016). Unintended Consequences of Cigarette Prohibition,

Regulation, and Taxation. International Journal of Law, Crime and Justice. doi:10.1016/

j.ijlcj.2016.03.002

Kurti, M. K., He, Y., von Lampe, K., & Li, Y. (2017). Identifying Counterfeit Cigarette Packs Using

Ultraviolet Irradiation and Light Microscopy. Tobacco Control 26 (1): 29–33. doi:10.1136/

tobaccocontrol-2015-052555

Kurti, M. K., von Lampe, K., He, Delnevo, C., Qin, D. (2018). Innovations in Counterfeiting Tax Stamps: A

Study of Ultraviolet Watermarks in a Sample of Discarded New York City Packs. Tobacco Control

Published Online First: 03 September 2018. doi: 10.1136/tobaccocontrol-2018-054501

Kurti, M. K., von Lampe, K., & Thompkins, D. E. (2012). The Illegal Cigarette Market in a

Socioeconomically Deprived Inner-City Area: The Case of the South Bronx. Tobacco Control 22 (2):

138–140. doi:10.1136/tobaccocontrol-2011-050412

Kurti, M. K., von Lampe, K., Johnson, J. (2015). The Intended and Unintended Consequences of a Legal

Measure to Cut the Flow of Illegal Cigarettes into New York City: The Case of the South Bronx.

American Journal of Public Health 105 (4): 750–56. doi:10.2105/AJPH.2014.302340

Lakhdar, C. B. (2008). Quantitative and Qualitative Estimates of Cross-Border Tobacco Shopping and

Tobacco Smuggling in France. Tobacco Control 17: 12–16. doi:10.1136/tc.2007.020891

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. doi:10.17310/ntj.2008.1.01

Merriman, D. (2002). Understand, Measure, and Combat Tobacco Smuggling. Economics of Tobacco

Toolkit. Washington: The World Bank. Retrieved January 9, 2019 from:

siteresources.worldbank.org/INTPH/Resources/7Smuggling.pdf

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. doi:10.1257/pol.2.2.61

Merriman, D., & Chernick, H. (2010). Using Littered Pack Data to Estimate Cigarette Tax Avoidance in

NYC. SSRN Electronic Journal. doi:10.2139/ssrn.2192169

Merriman, D., Yürekli, A., & Chaloupka, F. J. (2000). How Big Is the Worldwide Cigarette-Smuggling

Problem? In F.J. Chaloupka and P. Jha (Eds). Tobacco Control in Developing Countries, Oxford:

Oxford University.

National Travel and Tourism Office (NTTO) (2015). Overseas Visitation Estimates for U.S. States, Cities,

and Census Regions: 2014. Undated PDF document with creation date May 27, 2015. Washington:

U.S. Department of Commerce, International Trade Administration, Industry & Analysis unit.

Retrieved October 4, 2018 from: travel.trade.gov/research/programs/ifs/documents/

1Q12_US_to_Overseas_Banner1_30Sep13.pdf

O’Connor, R. J. (2012). Non-Cigarette Tobacco Products: What Have We Learned and Where Are We

Headed? Tobacco Control 21 (2): 181–190. doi:10.1136/tobaccocontrol-2011-050281

Office of Travel and Tourism Industries (OTTI) (2013). Survey of International Air Travelers: U.S. Travelers

to Overseas, January-March 2012. Undated PDF document with creation date October 21, 2013.

Washington: U.S. Department of Commerce, International Trade Administration. Retrieved July 22,

2017 from: travel.trade.gov/research/programs/ifs/documents/

1Q12_US_to_Overseas_Banner1_30Sep13.pdf

Organization for Economic Cooperation and Development (OECD). (2016). Illicit Trade: Converging

Criminal Networks. Retrieved January 9, 2019 from: oecd.org/gov/risk/charting-illicit-trade-

9789264251847-en.htm

Orzechowski & Walker. (2014). The Tax Burden on Tobacco. Arlington, VA. Retrieved January 9, 2019

from: taxadmin.org/assets/docs/Tobacco/papers/tax_burden_2014.pdf

Pelfrey, Jr., W. V. (2015). Cigarette Trafficking, Smurfing, and Volume Buying: Policy, Investigation, and

Methodology Recommendations from a Case Study. Criminal Justice Policy Review 26 (7): 713–726.

doi.org/10.1177/0887403414540518

Perri, F. S., & Brody, R. G. (2011). The Dark Triad: Organized Crime, Terror and Fraud. Journal of Money

Laundering Control 14 (1): 44–59. doi:10.1108/13685201111098879

Pfeffermann, D. (1993). The Role of Sampling Weights When Modeling Survey Data. International

Statistical Review 61 (2): 317–337.

Prichard, J., Hall, W., de Voogt, P., & Zuccato, E. (2014). Sewage Epidemiology and Illicit Drug Research:

The Development of Ethical Research Guidelines. The Science of the Total Environment 472: 550–

555. doi:10.1016/j.scitotenv.2013.11.039

Prieger, J. E., & Kulick, J. (2014). Unintended Consequences of Enforcement in Illicit Markets. Economics

Letters 125 (2): 295–297. doi:10.1016/j.econlet.2014.09.025

Prieger, J. E., & Kulick J. (2018a). Cigarette Taxes and Illicit trade in Europe. Economic Inquiry 56 (3):

1706–1723. doi.org/10.1111/ecin.12564

Prieger, J. E., & Kulick, J. (2018b). Tax Evasion and Illicit Cigarettes in California: Part I—Survey Evidence

on Current Behavior. SSRN Scholarly Paper ID 3181586. doi:10.2139/ssrn.3181586

Prieger, J. E., & Kulick, J. (2018c). Tax Evasion and Illicit Cigarettes in California: Part II— Smokers’

Intended Responses to a Tax Increase. SSRN Scholarly Paper ID 3181617. doi:10.2139/ssrn.3181617

Prieger, J. E., & Kulick, J. (2019). Tax Evasion and Illicit Cigarettes in California: Part IV— Smokers’

Behavioral and Market Responses to a Tax Increase. SSRN Scholarly Paper ID (TBD). doi:10.2139/

ssrn.(TBD)

Reuter, P., & Majmundar, M. (2015). Understanding the U.S. Illicit Tobacco Market: Characteristics,

Policy Context, and Lessons from International Experiences. Washington: Committee on the Illicit

Tobacco Market: Collection and Analysis of the International Experience; National Research

Council. Retrieved January 9, 2019 from: nap.edu/catalog/19016/understanding-the-us-illicit-

tobacco-market-characteristics-policy-context-and

Rijo, M. J, & and Ross, H. (2018). Illicit Cigarette Sales in Indian Cities: Findings from a Retail Survey.

Tobacco Control 27 (6): 684–688. doi:10.1136/tobaccocontrol-2017-053999

Ross, H. (2015). Understanding and Measuring Tax Avoidance and Evasion: A Methodological Guide. The

Economics of Tobacco Control Project Report. doi:10.13140/RG.2.1.3420.0486

Ross, H., Vellios, N., Smith, K. C., Ferguson, J., & Cohen, J. E. (2015). A Closer look at “Cheap White”

Cigarettes: Data Supplement 1. Tobacco Control. Online appendices; Retrieved January 9, 2019

from: tobaccocontrol.bmj.com/content/tobaccocontrol/suppl/2015/09/28/tobaccocontrol-2015-

052540.DC1/tobaccocontrol-2015-052540supp.pdf

Ross, H., Vellios, N., Smith, K. C., Ferguson, J., & Cohen, J. E. (2016). A Closer Look at “Cheap White”

Cigarettes. Tobacco Control. 25 (5): 527–531. doi:10.1136/tobaccocontrol-2015-052540

Saba, R. R., Beard, T. R., Ekelund, R. B., & Ressler, R. W. (1995). The Demand for Cigarette Smuggling.

Economic Inquiry 33 (2): 189–202. doi:10.1111/j.1465-7295.1995.tb01856.x

Schultz, P. W., Bator, R. J., Large, L. B., Bruni, C. M., & Tabanico, J. J. (2013). Littering in Context: Personal

and Environmental Predictors of Littering Behavior. Environment and Behavior 45 (1): 35–59.

doi:10.1177/0013916511412179

Scollo, M., Zacher, M., Durkin, S., Wakefield, M. (2014) Early Evidence About the Predicted Unintended

Consequences of Standardised Packaging of Tobacco Products in Australia: A Cross-Sectional Study

of the Place of Purchase, Regular Brands and Use of Illicit Tobacco. BMJ Open 4:e005873.

doi:10.1136/bmjopen-2014-005873

Shelley, L. I., & Melzer, S. A. (2008). The Nexus of Organized Crime and Terrorism: Two Case Studies in

Cigarette Smuggling. International Journal of Comparative and Applied Criminal Justice 32 (1): 43–

63. doi:10.1080/01924036.2008.9678777

Shen, A., Antonopoulos, G. A., & von Lampe, K. (2010). “The Dragon Breathes Smoke.” British Journal of

Criminology 50 (2): 239–258.

doi: 10.1093/bjc/azp069

Silver, D., Giorgio, M. M., Bae, J. Y., Jimenez, G., & Macinko, J. (2016). Over-the-Counter Sales of out-of-

State and Counterfeit Tax Stamp Cigarettes in New York City. Tobacco Control 25 (5): 584–586.

doi:10.1136/tobaccocontrol-2015-052355

State of New Jersey Office of Criminal Investigation (NJ OCI). (2014). A Guide to the Enforcement of the

New Jersey Cigarette Tax Act. OCI-100 Rev. 4/14. Division of Taxation. Retrieved January 9, 2019

from: state.nj.us/treasury/taxation/pdf/pubs/misc/ocicigs.pdf

Stehr, M. (2005). Cigarette Tax Avoidance and Evasion. Journal of Health Economics 24 (2): 277–297.

doi:10.1016/j.jhealeco.2004.08.005

Stephens, W. E., Calder, A., & Newton, J. (2005). Source and Health Implications of High Toxic Metal

Concentrations in Illicit Tobacco Products. Environmental Science & Technology 39 (2): 479–488.

Stoklosa, M., & Ross, H. (2013). Contrasting Academic and Tobacco Industry Estimates of Illicit Cigarette

Trade: Evidence from Warsaw, Poland. Tobacco Control 23 (e1): e30–e34. doi:10.1136/

tobaccocontrol-2013-051099

Sullivan, R. S., & Dutkowsky, D. H. (2012). The Effect of Cigarette Taxation on Prices: An Empirical

Analysis Using Local-Level Data. Public Finance Review 40 (6): 687–711. doi:10.1177/

1091142112442742

Sung, H. Y., Teh-Wei, H. and Keeler, T. E. (1994) Cigarette Taxation and Demand: An Empirical Analysis,

Contemporary Economic Policy 12: 91–100. doi:10.1111/j.1465-7287.1994.tb00437.x

Thursby, J. G., & Thursby, M. C. (2000). Interstate Cigarette Bootlegging: Extent, Revenue Losses, and

Effects of Federal Intervention. National Tax Journal 53 (1): 59–78. doi:10.17310/ntj.2000.1.04

Torgler, B.; & Schneider, F. G. (2007). Shadow Economy, Tax Morale, Governance and Institutional

Quality: A Panel Analysis. CESifo working paper, No. 1923.

Transcrime. (2015). European Outlook on the Illicit Trade in Tobacco Products. Trento: Transcrime –

Università degli Studi di Trento. Retrieved January 9, 2019 from:

transcrime.it/pubblicazioni/european-outlook/

Tscharke, B. J., White, J. M., & Gerber, J. P. (2016). Estimates of Tobacco Use by Wastewater Analysis of

Anabasine and Anatabine. Drug Testing Analysis 8: 702–707.

Unger, J. B., Rohrbach, L. A., Howard, K. A., Boley Cruz, T., Johnson, C. A., & Chen, X. (1999). Attitudes

Toward Anti-Tobacco Policy Among California Youth: Associations With Smoking Status,

Psychosocial Variables And Advocacy Actions. Health Education Research 14 (6): 751–763.

U.S. Department of State. (2015). The Global Illicit Trade in Tobacco: A Threat to National Security.

Retrieved January 9, 2019 from: 2009-2017.state.gov/documents/organization/250513.pdf

U.S. Government Accountability Office (GAO). (2012). Illicit Tobacco: Various Schemes Are Used to

Evade Taxes and Fees. GAO-11-313. Retrieved January 9, 2019 from: gao.gov/assets/320/

316372.pdf

van Nuijs, A. L. N., Mougel, J., Tarcomnicu, I., Bervoets, L., Blust, R., Jorens, P. G., Neels, H., & Covaci, A.

(2011). Sewage Epidemiology—A Real-Time Approach to Estimate the Consumption of Illicit Drugs

in Brussels, Belgium. Environment International 37 (3): 612–621. doi:10.1016/j.envint.2010.12.006

von Lampe, K. (2011). The Illegal Cigarette Trade. In M. Natarajan (Ed.), International Crime and Justice,

148–154. Cambridge: Cambridge University.

Wang, S., Merriman, D., & Chaloupka, F. J. (2018). Relative Tax Rates, Proximity, and Cigarette Tax

Noncompliance: Evidence from a National Sample of Littered Cigarette Packs. Public Finance Review

Online First. doi:10.1177/1091142118803989

Weaver, T. (2015). Millions Up in Smoke: NY Has Nation’s Highest Cigarette Tax; Why Do So Few Pay It?

Syracuse.com. Retrieved January 9, 2019 from: syracuse.com/state/index.ssf/2015/12/

ny_losing_big_with_nations_highest_cigarette_tax.html

Williams, E., Curnow, R., & Streker, P. (1997). Understanding Littering Behaviour in Australia. Community

Change Consultants Report. Victoria, Australia: Beverage Industry Environment Council. Retrieved

January 9, 2019 from: kab.org.au/wp-content/uploads/2012/05/understanding-littering-behaviour-

lbs1.pdf

Wooldridge, J. M. (2002). Econometric Analysis of Cross Section and Panel Data. Cambridge, MA: MIT.

World Customs Organization (WCO). (2013). Illicit Trade Report 2012. Brussels: World Customs

Organization. Retrieved January 9, 2019 from: illicittrade.com/reports/downloads/WCO-Illicit-

Trade-Report-2012.pdf

World Customs Organization (WCO). (2014). Illicit Trade Report 2013. Brussels: World Customs

Organization. Retrieved January 9, 2019 from: wcoomd.org/-

/media/wco/public/global/pdf/topics/enforcement-and-compliance/activities-and-

programmes/illicit-trade-report/itr_2013_en.pdf?db=web

World Customs Organization (WCO). (2015). Illicit Trade Report 2014. Brussels: World Customs

Organization. Retrieved January 9, 2019 from:wcoomd.org/-

/media/wco/public/global/pdf/topics/enforcement-and-compliance/activities-and-

programmes/illicit-trade-report/itr_2014_en.pdf?db=web

World Health Organization (WHO). (2008). WHO Report on the Global Tobacco Epidemic, 2008: The

mPOWER Package. Geneva: World Health Organization. Retrieved January 9, 2019 from: who.int/

tobacco/mpower/mpower_report_full_2008.pdf

World Health Organization (WHO). (2013). WHO Report on the Global Tobacco Epidemic, 2013.

Enforcing Bans on Tobacco Advertising, Promotion and Sponsorship. Geneva: World Health

Organization. Retrieved January 9, 2019 from: apps.who.int/iris/bitstream/10665/85380/1/

9789241505871_eng.pdf?ua=1

World Health Organization (WHO). (2015). WHO Report on the Global Tobacco Epidemic, 2015. Raising

Taxes on Tobacco. Geneva: World Health Organization. Retrieved January 9, 2019 from: who.int/

tobacco/global_report/2015/report/en/

Yürekli, A., & Sayginsoy, O. (2010). Worldwide Organized Cigarette Smuggling: An Empirical Analysis.

Applied Economics 42 (5): 545–561. doi:10.1080/00036840701720721

Zuccato, E., & Castiglioni, S. (2012). Consumi Di Sostanze Stupefacenti Nelle Città Europee [Consumption

of Illicit Drugs in European Cities]. Ricerca E Pratica, 28 (6): 252–260.

doi:10.1707/1191.13221

Tables

Table 1 – Collected samples by market and year

Market/MSA Cities Included Collection Date Packs Collected

Los Angeles

Anaheim, Long Beach, Los Angeles,

Riverside, and Santa Ana

October–November 2011

5,000

Los Angeles

(same as above)

September–October 2012

5,000

Los Angeles

(same as above)

February–March 2013

5,000

Los Angeles

(same as above)

February 2014

5,000

Los Angeles

(same as above)

April 2015

5,000

San Diego

September–October 2014

3,500

San Francisco

September–October 2014

3,500

Total

32,000

Table 2 – Relationships among counterfeit indicators and final determination of counterfeit status for domestic-

market Marlboro packs

Pack code is on the counterfeit list Pack code is not on the counterfeit list

Taggant

detected

Pack

fluoresces

Pack does

not fluoresce

Pack

fluoresces

Pack does

not fluoresce

No 13 counterfeit

13 counterfeit,

2 genuine

9 counterfeit 185 genuine

Yes

5 genuine

10,463 genuine

N/A

4 counterfeit

368 genuine

Notes: The subsample analyzed here are the 11,052 Marlboro brand packs intended for the domestic market

from survey waves in which all three indicators (TAGGANT_STATUS, PACK_FLUORESCES, AND

COUNTERFEIT_LIST) were recorded. Data for Taggant detected are from variable A_TAGGANT_DETECTED (a

recoding of TAGGANT_STATUS). A value of N/A refers to packs not tested for taggant. Refer to the codebook

for details on the recoding.

Table 3 – List of cheap-white brands found in the data

Brand

Source of designation as

cheap white

Chunghwa

MSI

Double Happiness

MSI

Esse

MSI

Golden Bridge

multiple

Hatamen

multiple

VIP

industry

Yesmoke

multiple

Yun Yan

industry

Note: Brands with source industry or multiple are designated as cheap whites in Ross et al. (2015b); brands with source

MSI were so designated in our data by MSI. Refer to text for discussion.

Table 4 – Correspondence between the MSI/ALCS fields CF_GROUP and CATEGORY

CF_GROUP CATEGORY Number of cases

No cello

Undetermined

8,855

Correct tax stamp

Applicable tax paid

13,529

Applicable state stamp

Applicable tax paid

7,281

Wrong tax stamp

Applicable tax not paid

Different state stamp

Applicable tax not paid

Counterfeit tax stamp

Contraband

No tax stamp

Applicable tax not paid

243

Counterfeit

Contraband

Counterfeit domestic

Contraband

Counterfeit non-domestic

Contraband

Non-domestic

1,875

Total: 32,000

Table 5 – Brand shares in the EDP and reference surveys

Los Angeles

San Diego

San Francisco

Brand

EDP

surveys

Reference

survey

Differ-

ence

EDP

surveys

Reference

survey

Differ-

ence

EDP

surveys

Reference

survey

Differ-

ence

Marlboro 44.6 45.2 -0.6 41.8 39.5 2.3 44.0 45.8 -1.8

Camel 12.7 14.7 -1.9** 16.4 11.4 4.9** 15.6 10.7 4.9**

Newport 11.0 10.3 0.8* 9.7 5.4 4.3*** 9.6 12.8 -3.2*

American Spirit 5.5 3.3 2.2*** 3.5 4.1 -0.7 5.1 4.8 0.3

Parliament 5.2 1.7 3.5*** 2.4 0.5 1.9*** 2.8 0.8 1.9***

Pall Mall 2.2 3.8 -1.7** 4.9 8.4 -3.5** 2.3 3.5 -1.3*

Rave 2.0 1.6 0.3* 0.4 0.2 0.2 0.9 0.0 0.9***

Maverick 1.9 1.5 0.3* 5.2 3.8 1.4* 1.5 4.3 -2.7*

Kool 1.5 1.1 0.4* 0.6 0.0 0.6*** 1.7 1.7 0.0

Timeless 0.7 1.6 -0.9** 0.6 0.7 0.0 1.4 0.4 1.0***

Other 12.7 15.1 -2.4** 14.5 26.0 -11.5*** 15.1 15.1 0.0

P-value for Pearson chi-square

test of overall differences

<0.001 <0.001 0.003

* Significant difference at the 10% level. ** Significant difference at the 5% level. *** Significant difference at the 1% level.

Notes: The reference survey for brand shares of California smokers is that of Prieger and Kulick (2018b); see text for details. For computation of the

standard errors, see footnote 62. The Pearson chi-square test is for the null hypothesis that the brand shares are identical in the two populations

(i.e., the target populations of the EDP and reference surveys; these tests are performed individually by city). The “other” category includes “no

usual brand preferred” in the reference survey.

Table 6 – Tax-paid status of collected samples by market and year

Share of

all packs

Share of

determinable packs

State Tax

Share of all packs

Share of determinable packs

Market

Year

Wrapper

intact, %

Tax status

determined, %

Applicable

tax paid, %

Applicable

tax not paid, %

Los Angeles

2011

71.3

74.3

90.9

9.2

Los Angeles

2012

62.0

63.7

92.0

8.0

Los Angeles

2013

77.8

80.3

91.3

8.7

Los Angeles

2014

74.4

76.3

93.4

6.6

Los Angeles

2015

73.2

73.9

91.9

8.1

Los Angeles

total

71.7

73.7

91.9

8.1

San Diego

2014

62.1

65.1

80.4

19.6

San Francisco

2014

66.4

69.6

84.4

15.6

Note: Tax status determined refers to the proportion of all packs that have intact cellophane, are intended for

non-domestic markets, or are counterfeit and lacking cellophane. Figures include domestic and foreign

product.

Table 7 – Cheap whites in the sample by market

Market

Out of all packs

Share of packs with brands

identified as cheap whites

Cheap white

brands, number

of packs

Cheap white

brand,

Brand identified only

from an industry

source, %

Brand identified

from multiple

sources, %

Los Angeles

0.3

92.9

7.1

San Francisco

0.1

100.0

0.0

San Diego

1.0

97.1

2.9

Note: see Ross et al. (2015b) for brand list and source. The brands mentioned in footnote 54 are added to the

list of brands identified by industry. The base for the calculations is all packs, with or without cellophane. See

also footnote 70.

Table 8 – ITTP status of collected samples, as percentage of determinable packs

Verified

ITTP

Packs other than verified ITTP

Total

Market

Year

Interstate tax

avoidance

and evasion

Foreign

Fully tax paid

domestic

Los Angeles

2011

0.6

0.4

8.3

90.7

100

Los Angeles

2012

0.6

0.4

7.1

92.0

100

Los Angeles

2013

1.0

0.7

7.3

91.1

100

Los Angeles

2014

0.4

0.8

5.5

93.3

100

Los Angeles

2015

0.9

1.0

6.3

91.8

100

Los Angeles

total

0.7

6.9

91.8

100

San Diego

2014

7.4

1.3

11.1

80.2

100

San Francisco

2014

3.0

0.7

12.5

83.8

100

Notes: The base for the percentages is the number of packs with determinable tax status: those with intact wrappers,

intended for foreign markets, or counterfeit. Verified ITTP includes counterfeits and domestically produced cigarettes

lacking a genuine tax stamp. Interstate tax avoidance and evasion includes interstate licit cross-border purchases and

(illicit) bootlegging, which cannot be distinguished from each other in the data; the same comment applies to the category

for Intrastate tax avoidance and evasion. Figures in the Fully tax-paid domestic column may be lower than the figures for

applicable state-tax paid in Table 6 because the latter does not exclude ITTP and a small number of counterfeits have

apparently valid tax stamps. Figures may not sum to 100 due to rounding.

Table 9 – ITTP status of collected samples (with broader assumptions), as percentage of determinable packs

Verified

ITTP

Broader Measures

of ITTP

Upper Bound

for ITTP

Market

Year

Conservative

Assumptions

Aggressive

Assumptions

Verified ITTP + all

other packs except

for fully tax paid

Los Angeles

2011

0.6

2.8

5.2

9.3

Los Angeles

2012

0.6

2.5

4.5

8.0

Los Angeles

2013

1.0

3.1

5.2

8.9

Los Angeles

2014

0.4

2.3

4.0

6.7

Los Angeles

2015

0.9

3.0

5.0

8.2

Los Angeles

Total

0.7

2.7

4.8

8.2

San Diego

2014

7.4

10.7

13.1

19.9

San Francisco

2014

3.0

6.6

10.2

16.2

Notes: The Conservative Assumptions are that ITTP includes 100% of verified ITTP packs, 25% of packs from bordering

states, 75% of packs from non-bordering states and other domestic locations (e.g., packs bearing a tribal stamp), and 25%

of foreign-market packs. The Aggressive Assumptions are that ITTP includes 100% of verified ITTP packs, 75% of packs

from bordering states, 100% of packs from non-bordering states and other domestic locations (e.g., packs bearing a tribal

stamp), and 50% of foreign-market packs. The base for the percentages is the number of packs with intact wrappers.

Table 10 –Composition of verified ITTP by market, as percentage of ITTP

Market

Year

Counterfeit

packs

No tax

stamp

Counterfeit

tax stamp

Illicit

whites

Verified

ITTP

Los Angeles

2011

19.0

52.4

0.0

28.6

100

Los Angeles

2012

61.1

11.1

0.0

27.8

100

Los Angeles

2013

29.0

42.1

0.0

29.0

100

Los Angeles

2014

18.8

43.8

0.0

37.5

100

Los Angeles

2015

3.0

18.2

3.0

75.8

100

Los Angeles

total

23.8

33.3

0.8

42.1

100

San Diego

2014

0.0

97.0

0.0

3.0

100

San Francisco

2014

15.1

50.7

0.0

34.3

100

Notes: If a pack is counterfeit and also bears a counterfeit tax stamp, it is placed in the Counterfeit pack

category. If a pack has no stamp or a counterfeit stamp and is a cheap-white brand, it is placed in the Illicit

whites category. Figures may not sum to 100 due to rounding.

Table 11 – Composition of broader measures of ITTP by market, as percentage of ITTP

Verified ITTP

Other components of ITTP

Total

ITTP

Assumptions and market

Counterfeit

packs

No tax

stamp

Counterfeit

tax stamp

Illicit

whites

Interstate

bootlegging

Foreign

illicit

Cheap

whites

Conservative assumptions

100

Los Angeles

6.0

8.3

0.2

10.5

12.1

62.9

0.0

100

San Diego

0.0

67.4

0.0

2.1

4.5

26.0

0.0

100

San Francisco

6.9

23.2

0.0

15.7

6.7

47.6

0.0

100

Aggressive assumptions

Los Angeles

3.4

4.7

0.1

6.0

12.2

70.0

3.6

100

San Diego

0.0

52.5

0.0

1.6

5.4

40.5

0.0

100

San Francisco

4.4

14.9

0.0

10.1

6.5

60.1

4.0

100

Notes: Figures may not sum to 100 due to rounding. See notes to Table 9 for definition of the assumptions. See also notes to Table 10.

Table 12 – Summary statistics of variables used in the regressions

Variable Mean Std. Dev. Minimum Maximum

State cigarette tax not paid

0.100

0.301

0.000

1.000

Pack brand: Marlboro

0.443

0.497

0.000

1.000

Pack brand: another Philip Morris brand

0.067

0.250

0.000

1.000

Pack brand: Camel

0.134

0.341

0.000

1.000

Pack brand: Newport

0.107

0.310

0.000

1.000

Pack brand: Native American

0.003

0.059

0.000

1.000

Pack brand: cheap white

0.004

0.062

0.000

1.000

Max price difference per hour of driving time (maxPrDiffPerHr)

0.061

0.029

-0.043

0.099

Cigarette retailer log density (CigRetailDen)

1.686

0.894

-1.519

4.636

Population density, log

8.140

0.718

5.954

9.917

Median income, log

4.755

0.157

4.299

5.220

Race/ethnicity in area: Black

0.072

0.100

0.000

0.685

Race/ethnicity in area: Asian

0.143

0.118

0.002

0.583

Race/ethnicity in area: Hispanic

0.414

0.250

0.039

0.966

Race/ethnicity in area: another race

0.072

0.100

0.000

0.685

Area educational attainment: % without HS degree

0.218

0.150

0.004

0.592

Area educational attainment: % with more than HS degree

0.592

0.199

0.169

0.972

Area median age

35.211

4.944

23.000

49.200

Table 13 – Logit regression of tax paid status (coefficients)

Y = cigarette tax not paid Estimation 1 Estimation 2 Estimation 3 Estimation 4

Pack brand: Marlboro

-1.446***

-1.439***

-1.436***

(0.071)

Pack brand: another PM brand

-0.892***

-0.898***

-0.881***

-0.890***

(0.110)

(0.111)

Pack brand: Camel

-2.386***

-2.379***

-2.391***

-2.386***

(0.133)

(0.135)

Pack brand: Newport

-3.541***

-3.539***

-3.532***

-3.533***

(0.247)

(0.246)

(0.247)

Pack brand: Native American

1.092***

1.133***

1.109***

1.137***

(0.276)

(0.286)

(0.284)

(0.291)

Pack brand: cheap white

2.198***

2.221***

2.197***

2.228***

(0.293)

(0.298)

(0.296)

(0.301)

Max price difference per hour of driving

6.126***

4.476**

time (maxPrDiffPerHr)

(1.703)

(2.048)

Cigarette retailer log density, spline

0.049

0.009

(up to 90th percentile)

(0.077)

(0.087)

Cigarette retailer spline

0.494***

0.376*

(over 90th percentile)

(0.152)

(0.205)

Population density, log

-0.051

0.109

(popDenLn)

(0.090)

(0.110)

Area median log income, spline

1.446

3.063**

(up to 10th percentile)

(1.237)

(1.454)

Income spline

-1.599***

-0.884*

(10th to 90th percentiles)

(0.507)

(0.519)

Income spline

0.611

1.401

(over 90th percentile)

(1.412)

(1.452)

Race/ethnicity: Black

-1.045**

-0.913*

(0.492)

(0.491)

Race/ethnicity: Asian

0.495

0.454

(0.508)

(0.525)

Race/ethnicity: Hispanic

0.695

0.460

(0.602)

(0.618)

Race/ethnicity: another race

0.620

1.445

(3.763)

(3.799)

Area educational attainment:

-0.635

-1.661

% without HS degree

(1.311)

(1.521)

Area educational attainment:

0.524

-0.736

% with more than HS degree

(1.112)

(1.340)

Area median age

0.350***

0.287***

(0.108)

(0.103)

Age squared

-0.004***

-0.004**

1.446

3.063**

Observations

23,141

23,109

Pseudo R

0.143

0.146

0.148

0.149

statistic [d.o.f.] (p-value)

1100 [12] 0.0

1209 [16] 0.0

1247 [23] 0.0

1419 [27] 0.0

Log likelihood

-6461

-6444

-6396

-6387

*** p<0.01, ** p<0.05, * p<0.1

Table notes: all estimations include MSA and year fixed effects. Robust standard errors (accounting for clustering on

112 ZIP codes) are in parentheses. The linear spline coefficients show the slope of the linear predictor in the relevant

range.

Table 14 – Logit regression of tax paid status (marginal effects)

Y = cigarette tax not paid Estimation 1 Estimation 2 Estimation 3 Estimation 4

Pack brand: Marlboro

-0.151***

-0.150***

-0.149***

-0.148***

(0.008)

Pack brand: another PM brand

-0.057***

-0.056***

(0.006)

Pack brand: Camel

-0.188***

-0.187***

-0.186***

(0.009)

(0.008)

Pack brand: Newport

-0.206***

-0.205***

-0.204***

-0.203***

(0.008)

(0.007)

Pack brand: Native American

0.122***

0.127***

0.123***

0.127***

(0.039)

(0.041)

(0.042)

Pack brand: cheap white

0.314***

0.317***

0.310***

0.315***

(0.057)

(0.058)

(0.057)

(0.058)

Max price difference per hour of driving

0.489***

0.355**

time (maxPrDiffPerHr)

(0.137)

(0.161)

Cigarette retailer log density, spline

0.004

0.001

(CigRetailDen, up to 90th percentile)

(0.006)

(0.007)

Cigarette retailer spline

0.039***

0.030*

(over 90th percentile)

(0.012)

(0.016)

Population density, log

-0.004

0.009

(0.007)

(0.009)

Area median log income, spline

0.115

0.243**

(up to 10th percentile)

(0.098)

(0.115)

Income spline

-0.127***

-0.070*

(10th to 90th percentiles)

(0.040)

(0.041)

Income spline

0.048

0.111

(over 90th percentile)

(0.112)

(0.115)

Race/ethnicity: Black

-0.083**

-0.072*

(0.039)

Race/ethnicity: Asian

0.039

0.036

(0.040)

(0.042)

Race/ethnicity: Hispanic

0.055

0.036

(0.048)

(0.049)

Race/ethnicity: another race

0.049

0.115

(0.299)

(0.301)

Area educational attainment:

-0.050

-0.132

% without HS degree

(0.104)

(0.121)

Area educational attainment:

0.042

-0.058

% with more than HS degree

(0.088)

(0.106)

Area median age

0.003***

(0.001)

Observations

23,141

23,109

*** p<0.01, ** p<0.05, * p<0.1

Table notes: marginal effects are calculated for each observation and then averaged. See also notes to previous table.

Figures

Figure 1 – California state cigarette excise-tax stamp.

Figure 2 – Sales of cigarettes, cigars, and smoking tobacco in the United States.

100

150

200

250

300

350

400

450

2000 2002 2004 2006 2008 2010 2012 2014 2016 2018

Billion sticks (cigarettes)

Billion units (cigars) or

stick equivalents (loose tobacco)

Source: Euromonitor International

Cigarettes

Smoking tobacco

(RYO & pipe)

Cigars

Figure 3 – Examples of collection routes.

Figure 4 – Sources of state-untaxed packs in California markets

Notes: Percentages are out of all packs for which state taxes have not been paid. Category Domestic unknown includes packs without a stamp

and packs with counterfeit stamps. Category Sycuan in San Diego is for packs bearing stamps from the Sycuan Band of the Kumeyaay Nation, a

Native American tribe whose reservation and casino is within 20 miles of San Diego.

10%

63%

Nevada

Other

domestic

China

US duty free

Other

duty free

Vietnam

Other non-domestic

Los Angeles

37%

24%

28%

Domestic

known

Domestic

unknown

Mexico

Other

non-

domestic

China

Sycuan

Non-US duty free

San Diego

13%

19%

42%

Domestic

known

Domestic

unknown

Mexico

Other

non-domestic

China

Armenia

Vietnam

Non-US duty free

San Francisco

Figure 5– ITTP status of collected samples by market and year, as percentage of determinable packs

Notes: See also notes to Table 8, which contains the same figures.

0 20 40 60 80 100

percentage of determinable packs

San Francisco

San Diego

Los Angeles

Tax compliance

Interstate tax avoidance

& evasion

Foreign Verified ITTP

Figure 6– ITTP status of collected samples (with broader assumptions) by market, as percentage of determinable packs

Notes: See also notes to Table 9, which contains the same figures.

0 5 10 15

percentage of determinable packs

San Francisco

San Diego

Los Angeles

Verified ITTP

ITTP (conservative

assumptions)

ITTP (aggressive

assumptions)

Upper bound

on ITTP