The effect of marijuana decriminalization on alcohol consumption : evidence from the United States, 1970-2014

(1)

The Effect of Marijuana Decriminalization on Alcohol

Consumption: Evidence from the United States, 1970-2014

Name: Rebecca Blathras Student number: 11373245

Thesis supervisor: Prof. Hessel Oosterbeek Date: 15-08-2017

Word count: 6716

MSc Economics, Public Policy

(2)

STATEMENT OF ORIGINALITY

This document is written by Rebecca Blathras, who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

ABSTRACT

This study uses decriminalization status of marijuana by state from 1970-2014 to estimate the causal effect of increased marijuana toleration on alcohol use. I find that consumption of alcohol does not change significantly when a state decriminalizes marijuana. The results do not support or refute the hypothesis that alcohol and marijuana are either complements or substitutes.

KEY WORDS: Alcohol, marijuana, decriminalization, public policy

Acknowledgements: I would like to thank Gerry McCourt for giving me the initial spark to write about this topic and Maria Pia Iocco Barias for helping me tremendously throughout the process.

(4)

TABLE OF CONTENTS 1 INTRODUCTION ... 4 2 LITERATURE REVIEW ... 6 3 CONTEXT ... 9 4 DATA ... 11 5 EMPIRICAL METHOD ... 17 6 RESULTS ... 19 7 CONCLUSION ... 23 8 REFERENCES ... 24 9 APPENDIX ... 27

(5)

1. INTRODUCTION

Policies regarding the decriminalization of marijuana are at the forefront of many lively and unsettled discussions today in the United States. The idea that the lawful penalties for using this drug should be decreased began during the 1970s and has resurfaced with quite some tenacity in the last decade (Drug Policy Alliance 2017a). There is significant debate regarding the advantages and disadvantages to decriminalization. Some of the key arguments of both sides rely on how decriminalization impacts the usage of alcohol among other drugs. In this paper, I look specifically at whether increased legal toleration of marijuana does affect alcohol consumption. My question specifically is, ‘what is the effect of marijuana decriminalization on alcohol consumption?’.

Answering this question is essential in order to engage in the policy sphere that disputes whether marijuana should be decriminalized. Where both sides of this contested issue rely on how reducing the penalties of using marijuana may impact the usage of other drugs (notably alcohol), it is important to empirically assess what the actual impact is.1 Advocates of decriminalization commonly argue that there will be fewer arrests of otherwise law abiding marijuana users, noting that many of these ‘criminals’ are young citizens. Relatedly, they reason that there will be reduced enforcement and social costs if arrests and prosecution of marijuana possession fall. Next, they argue that even though decriminalization warrants increased toleration of marijuana, it would have at most a small positive effect on the frequency of use. Most importantly, advocates view marijuana and alcohol as alternatives for getting high (substitutes), so decriminalizing marijuana should reduce alcohol consumption which could be considered a positive externality of the policy. On the other hand, those in opposition simply acknowledge that there will be increased usage of marijuana and that some harm always comes from an increased usage of drugs. As a group, they believe marijuana itself is dangerous and that it is a gateway drug, so it would inevitably lead to the usage of other dangerous drugs. They believe an increased legal toleration regarding this drug will promote disrespect for the law and lead to a mass of drug using citizens. Lastly, as those in opposition believe marijuana is a gateway drug,

1_{Thies and Register (1993) nicely highlight the viewpoints of both sides in the debate,}

(6)

they argue that it is a complementary good with other drugs such as alcohol; so, decriminalization would increase alcohol usage.

According to the Drug Policy Alliance, the United States spends more than $51,000,000,000 annually on the war on drugs (2017b). In 2015, there were about 1,488,707 arrests for drug law violations. Approximately 643,000 people were arrested for a marijuana law violation in 2015 and 89% of those were arrested for possession only. Furthermore, it is important to note that the U.S. has the highest incarceration rate in the world (1 out of every 11 adults).2 Grossman, Chaloupka and Shim (2002) highlight that almost all drug offenders in U.S. prisons commit nonviolent crimes; this brings into question whether decriminalizing marijuana would be a good financial investment for the federal government.

Most previous studies on the economic relationship between marijuana and alcohol find evidence for either complementarity or substitution, but there are a few that have inconclusive results as this study does. These prior studies focus on the relationship between alcohol and marijuana primarily but some include data on cigarettes, heroin and/or cocaine in order to control for any influence these other drugs may have on the demands. Several studies find that lowering the availability of alcohol (using the minimum legal drinking age, MLDA) led to a subsequent increase in marijuana usage of adults, suggesting the goods are substitutes (Dinardo and Lemieux 1992, Chaloupka and Laixuthai 1994, Crost and Guerrero 2012). Others find that although alcohol availability and usage may be limited by the MLDA, so is the consumption of marijuana, arguing that the goods are instead complements (Pacula 1998a, Pacula 1998b, Saffer and Chaloupka 1999).

This paper differs significantly from previous literature as it employs entirely state level data rather than solely micro or micro and state level data combined. This paper does not study anything at the individual level. Almost all of the relevant literature on this topic uses survey data for the consumption of marijuana and alcohol as well as individual characteristics such as age, and often estimates demand equations for the two goods, while this study relies on sales data and is constricted from estimating demands. Relatedly, many past papers focus on youth or a

2_{These statistics were taken from the Drug Policy Alliance, a New York City based non-profit}

(7)

specific age group while this paper does not. It is worth noting that survey data has the flaw that the results depend on truthfulness and memory of the respondents, especially regarding the use of drugs where it may be illegal.

This study shows that marijuana decriminalization has no statistically significant effect on alcohol consumption. Consequently, this analysis neither supports nor refutes previous findings that the two goods are either complements or substitutes. The remainder of this paper is organized as follows: Section 2 provides a review of the relevant literature. Section 3 provides background information on the history of marijuana regulation in the United States, which states have decriminalized the drug, and what exactly it means to decriminalize marijuana. Section 4 describes the data used in this study, how it was appended and its limitations. Section 5 is an explanation of my research method and justifications for my technique. Section 6 is an overview and interpretation of my empirical results. Lastly, section 7 is a discussion of the implications of my findings.

2. LITERATURE REVIEW

During the past few decades, there have been several studies attempting to explore the relationship between the demands for alcohol and marijuana in the general and youth population. The empirical findings are quite mixed. Previous studies have most likely produced varied results as they are based on several different data sets (different age groups, states, time periods), use different proxies for alcohol and marijuana price, and overall use a slew of empirical methods. Those studies that focused on youths and young adults typically concluded that alcohol and marijuana are economic substitutes. Subsequent literature, which frequently extended by including additional proxies for the price of marijuana and/or alcohol or used data involving more recent cohorts generally found evidence of complementarity. There are a few exceptions that find inconclusive results or no relationship between the two goods. Such contrasting results may very well be adding to the continued dispute about whether marijuana should be decriminalized in the United States.

Dinardo and Lemieux (1992) were the first to examine the relationship between the demands for alcohol and marijuana. They used 1980-1989 data from Monitoring the Future (MTF), an annual survey of high school seniors in the United States. The authors evaluated the effect of a

(8)

specific government intervention, the MLDA, on the consumption of both drugs in a simple demand theoretic framework. They used state-aggregated data to estimate a bivariate probit specification of alcohol and marijuana use. The prevalence equations they estimated included the price of alcohol, MLDA, and marijuana decriminalization state status. The authors found that marijuana decriminalization had a significant and negative effect on the prevalence of alcohol use by high school seniors, concluding from this finding that alcohol and marijuana are economic substitutes for youth. The authors inferred that increases in the minimum legal drinking age during 1980-89, caused adolescents to increase their consumption of marijuana.

Chaloupka and Laixuthai (1997) also used data from the 1980-1989 MTF and confirmed Dinardo and Lemieux’s earlier finding. They used the price of beer as the measure of alcoholic beverage prices as well as the MLDA for an additional proxy of the full price of alcohol. Furthermore, they used two variables to capture the full price of marijuana- one being an indicator for whether marijuana is decriminalized in the state the respondent resides in, and the second being a measure of the money price of the drug (data obtained from the Drug Enforcement Agency of the United States). As with Dinardo and Lemieux, they found statistically significant results supporting a substitution effect between marijuana and alcohol. They estimated an ordered probit model of the frequency of alcohol use. They utilized a decriminalization dummy variable to represent the legal risk of consuming marijuana. With these models for both the frequency of drinking as well as the probability of heavy drinking, they found that both were negatively related to beer prices and state decriminalization status. Such results suggested again that, the two goods are economic substitutes for youth. In a more recent study, Crost and Guerrero (2012) used data from the National Survey of Drug Use and Health (NSDUH). They employed a regression discontinuity created by the minimum legal drinking age and compared those respondents just about to turn 21 to those who just turned 21 to observe the impact of increased alcohol availability on marijuana usage. In doing so, they assumed that people are similar in characteristics that determine substance abuse, apart from the ability to legal purchase alcohol Again, this study found that marijuana and alcohol are substitute goods.

Pacula (1998b) used micro level data from the National Longitudinal Survey of Youth (NLSY) to estimate individual demand equations for both alcohol and marijuana. The NLSY used in her study is a national survey conducted in the United States in 1979 on a sample of

(9)

individuals between 14-21 years of age, who were interviewed annually until 1984. Pacula estimated individual demand equations for alcohol and marijuana and included proxies for the monetary prices of alcohol, marijuana and cigarettes. She found that state regulations designed to reduce the consumption of alcohol (beer taxes and the MLDA) had a negative and significant effect on the probability of using marijuana, suggesting the two goods are complements. Pacula’s “Adolescent Alcohol and Marijuana Consumption: Is There Really A Gateway Effect?” (1998a) extended by exploring ‘is there really gateway effect?’ between adolescent alcohol and marijuana consumption. Again, the NLSY was used but only from 1983-1984. She found that there is indeed a contemporaneous relation between the consumption of legal and illicit substances, including alcohol and marijuana, further adding to the complement story. In addition, she found that previous experience with alcohol significantly increases the current probability of using marijuana, illustrating the gateway theory- a comprehensive catchphrase for the phenomenon that using certain drugs increases the probability of the use of further drugs.

Saffer and Chaloupka (1999) conducted arguably the most comprehensive analysis of cross-price effects. They used the National Household Survey of Drug Abuse (NHSDA, renamed NSDUH in 2002); the NHSDA surveys the general population, not specifically youth, however the authors do include a control for youth in their regression. They estimated annual prevalence equations for marijuana, cocaine and heroin. Additionally, they measured the number of days alcohol was used in the past month for individuals at a micro level (taking note of ethnic, gender and age subgroups). They utilized county level alcohol prices and state level decriminalization laws to measure the price of both marijuana and alcohol. They also included state level prices for cocaine and heroin to control for any relationship alcohol and marijuana demands have with cocaine and heroin. They found strong evidence of complementarity between alcohol and marijuana for white males and African Americans, as county level alcohol prices were negatively correlated with marijuana usage. But for Native Americans and Hispanics, their findings indicated that the two goods are substitutes. They found no significant cross-price effects for Asians, women or youth. Farrelly et al. (1999) also used data from the NHSDA, and found that increases in beer prices at the state level are negatively correlated with marijuana usage (so marijuana and alcohol are complements) for those aged 12-20, but not for those aged 21-30. Finally, Williams et al. (2004) focused specifically on the demand for the two goods among

(10)

college students, and takes data from Harvard School of Public Health’s College Alcohol Study (CAS). They found that alcohol and marijuana are economic complements and that policies that increase the full price of alcohol decrease participation in marijuana use.

Lastly, and most notably for the findings of this paper, there is some literature that finds no statistically significant relationship between alcohol and marijuana consumption. Thies and Register (1993) used the 1984 and 1988 NLSY surveys for eleven states and when observing the causal effect of marijuana decriminalization on different drug usage (alcohol, marijuana, cocaine), controlled for a slew of variables thought to impact drug usage at the individual level. They were the first to use individual level data to estimate both the prevalence and quantity of marijuana consumed. They found inconclusive evidence regarding the relationship between marijuana and alcohol. They used logit specifications to estimate the probability of using alcohol, being a heavy drinker, using marijuana and cocaine. State decriminalization status and the MLDA for alcohol were included as measures of state controls of drugs; no proxies were used for the price of alcohol. They found that state decriminalization status only significantly influenced the probability of using alcohol heavily. Most importantly, coefficients from their Tobit specifications of the quantity of each drug consumed revealed that state decriminalization status had no significant effect on the quantity of alcohol consumed. With this finding, they concluded that there is insufficient evidence to infer that alcohol and marijuana are substitutes. Good (2015) focused solely on Colorado as a treatment state, and the six states that share its border as controls. The survey data used was provided by adults aged 18-39, taken from the Center for Disease Controls Behavior Risk Surveillance Study. Using a difference-in-difference model Good analyzed the impact of marijuana legalization (not the same as decriminalization but still measures an increase legal tolerance of marijuana) from 2008-2013 on alcohol consumption. None of the causal effects are found to be statistically significant, so just as Thies And Register, Good’s study could not support or refute the hypothesis that alcohol and marijuana are either complements or substitutes.

3. CONTEXT

In the early 20th century, the United States began to develop laws regulating drugs and dangerous products. In 1906, the Pure Food and Drug Act was passed in the United States, paving the way

(11)

for the creation of the Food and Drug Administration. This was the first comprehensive food and drug law in the U.S., giving new regulatory powers to the federal government. Alcohol, morphine, opium and marijuana were all labeled as addictive and/or dangerous and appropriate regulations were assembled for their usage. In 1934, the Uniform State Narcotic Drug Act was passed to create uniform laws across the U.S. in regard to controlling the sale and use of narcotic drugs; included were state laws regulating the sale of marijuana. Next, came the Marihuana Tax Act of 1937, essentially making the possession or transfer of marijuana illegal in the U.S. In 1956, the Narcotics Control Act was passed, making a first-time marijuana possession offense a crime warranting a minimum of two to ten years in jail and fines up to $20,000. The Controlled Substances Act (CSA) of 1970 was then put in place, repealing the Marihuana Tax Act of 1937. Under the presidency of Richard Nixon, this CSA established a federal drug policy that classified substances into five categories. It was then that marijuana was classified as a Schedule I substance-the most restrictive category for drugs. In 1971, President Nixon declared a ‘war on drugs’. During this time, there was a strong initiative to create policies discouraging any kind of usage, distribution, or production of seemingly dangerous drugs. Arguably a political mask for quelling the mass growth of hippies, known for protesting the Vietnam War, this was likely a strategic move made by the government; they tried to associate hippies (those who frequently protested the War) with marijuana usage to make citizens believe that using marijuana was bad.

Nevertheless, from 1971 to 1979, twelve states decriminalized marijuana and this wave of attitude change was further reflected as President Jimmy Carter was elected in 1977 on a campaign platform that included marijuana decriminalization.6 7_{However, attitudes transformed}

in the following years as marijuana became caught up in the broader cultural backlash of the apparent permissiveness of the 1970s and put the movement at a standstill. It was not until the 21st century that decriminalization proposals began surfacing at the state level again; Connecticut (2011), Delaware (2015)*, Illinois (2016)*, Washington D.C. (2014)*, Maryland (2014)*,

6_{Alaska, California, Colorado, Maine, Minnesota, Mississippi, Nebraska, New York (possession}

in public still a misdemeanor), North Carolina, Ohio, Oregon, Washington

7_{Alaska formally decriminalized marijuana in 1982, recriminalized in 1990, decriminalized}

again in 2003 and in 2014 legalized it. Prior, the state observed Ravin v. State (1975), which protected an adult’s ability to possess and use a small amount of marijuana in their home. Most literature assumes this to be Alaska decriminalizing marijuana starting in 1975, and so in my study I do as well.

(12)

Massachusetts (2009), Missouri (2017)*, Nevada (2001), New Hampshire (2017)*, Rhode Island (2013) and Vermont (2013) then decriminalized marijuana.8_{Since then, Alaska (2014),}

California (2016), Colorado (2012), Washington, D.C. (2015), Maine (2016), Massachusetts (2016), Nevada (2017), Oregon (2014), and Washington (2012) have legalized recreational marijuana usage. From 1970-2014, even though decriminalization is significantly distinguishable from legalization of recreational marijuana, states legalizing marijuana were given a dummy value of 1 to indicate such increased lawful tolerance towards the drugs.

This study focuses on states specifically where the recreational use and possession of small amounts of marijuana was decriminalized. Typically, decriminalization means no arrest, prison time, or criminal record for the first-time possession of a small amount of marijuana for personal consumption. It does not observe anything regarding medical marijuana or the legalization of marijuana. It is crucial to define decriminalization in this context as the reduction of a penalty of possessing a small amount of marijuana to a fine rather than imprisonment; therefore, possessing marijuana is by no means legal, but there is increased lawful toleration. Many argue that decriminalization thus reduces the full price of marijuana use. Relatedly, economic theory suggests that when the cost of consuming some good increases, people will consume more of its substitutes and less of its complements. One can typically not be arrested, ticketed, convicted, etc. for using small amounts of marijuana if they follow the law regarding age, place and amount for consumption.

4. DATA

Data for per capita alcohol consumption by state and type of alcoholic beverage for the years 1970-2014 was taken from the National Institute on Alcohol Abuse and Alcoholism (NIAAA).10 The data is based upon alcoholic beverage sales and such information was collected by the Alcohol Epidemiologic Data System (AEDS), the National Alcohol Beverage Control Association (NABCA) and from the states themselves. The report provides consumption data by state and year for beer, wine and distilled spirits as well as for all alcoholic beverages combined.

8_{* These states were not included as having a decriminalization status in the empirics, as they}

decriminalized after 2014, or the law went into effect more than half way through 2014.

10_{For a complete description of the methods involved in compiling the dataset, see the AEDS}

(13)

Beverage sales data is utilized to construct consumption information. This data is reported in gallons, so the alcohol consumption per capita results are reported as gallons consumed per capita. The NIAAA justifies using beverage sales data because “sales data more accurately reflect[s] actual consumption of alcoholic beverages than do production and shipments data from beverage industry sources” (Haughwout, LaVallee, and Castle, 2015). The sales data is converted to gallons consumed; the amounts sold are reported by volume or tax revenue, which the AEDS converts to gallons consumed utilizing state tax rates.

Ethanol conversion coefficients (ECCs) were used to convert gallons consumed to gallons of ethanol consumed. The ECCs are standard proportions of pure alcohol (ethanol) in each kind of alcohol. The standard ECC measurements were reformed for wine and spirits twice over time to best capture the average alcohol content.11Once gallons consumed are converted to gallons of ethanol consumed for beer, wine, and spirits, they are summed to produce data for gallons of ethanol consumed of all beverages. Finally, the NIAAA obtained population data from the U.S. Census Bureau. This data is used to calculate the per capita rates, accomplished by dividing gallons of ethanol consumed by state and year, by the corresponding population. Both 14+ and 21+ populations are used to calculate separate per capita consumption values. Although the legal age to buy and consume alcohol is 21, most self-reported surveys indicate that many people aged 14 drink alcoholic beverages. For example, Chen et al. (2015) found the average age of initiation of alcohol usage to be 14, when using data from the 2011-2013 National Survey on Drug Use and Health. Furthermore, data from the NIAAA 2001-2002 National Epidemiologic Survey on Alcohol and Related Conditions indicate that 12.2% of U.S. drinkers aged 18+ reported having drank for the first time around the age of 15 (AEDS 2004).

With this panel data from 1970-2014, I use 51 states as entities and have a unit of observation as one year. Thus with 51 entities across 45 years, there are 2295 observations. When I control for beverage price by using taxes, all states have an appropriate excise tax for beer, so the controlled regression for beer remains with 2295 observations. The controlled regression for wine drops 7 states without appropriate wine excise taxes, so the number of

11_{Ethanol conversion coefficients: Beer 0.045 (1960-2014), Wine 0.16 (1970-71), 0.145}

(14)

observations becomes 44 entities across 45 years, 1980.12 When controlling for the price of spirits, 15 states are dropped, leaving 36 entities across 45 years, 1620 observations.13_Lastly,

when running regressions for all beverages combined and implementing a price control, any states without appropriate wine or spirit taxes are dropped (15), which is the same as the controlled spirits regression, again leaving 36 states, 45 years, for 1620 observations. Although a substantial amount of observations are dropped when controlling for alcohol price, specifically for spirits and all beverages, I report estimates before these controls, so considerable weight should be put on those as no observations were dropped and the estimates are then not biased.

Consumption reports from the NIAAA are issued annually. In 2014, the per capita consumption of ethanol from all alcoholic beverages was 2.32 gallons, down 0.4% from 2.33 in 2013 (Haughwout, LaVallee, and Castle 2015). For interpretation, the NIAAA, defines a “standard drink” in the United States as one that contains 0.6 fluid ounces (0.005 gallons) of ethanol. Tables 1 and 2 contain summary statistics of yearly per capita consumption of alcoholic beverages for both 14+ and 21+ populations. When I control for alcohol price with excise taxes, states without appropriate taxes are dropped. Therefore, the summary statistic including ‘tax’ is indicative of the per capita consumption using only the states with a relevant tax. There is no statistic for “beer tax” because all states have a suitable beer tax and so the statistics are the same as all states and years are used in both the uncontrolled and controlled regressions for beer consumption.

12_{Alaska, Hawaii, Iowa, Montana, New Hampshire, Pennsylvania, Utah}

13_{Alaska, Hawaii, Idaho, Iowa, Maine, Michigan, Montana, New Hampshire, North Carolina,}

(15)

Table 1. Summary statistics

Gallons of ethanol consumed per capita, U.S. population 14+ 1

Mean Std. Dev. N Beer 1.29 0.24 2295 Wine 0.33 0.18 2295 Wine Tax 2 0.33 0.18 1980 Spirits 0.91 0.45 2295 Spirits Tax 3 _0.94 _0.45 ₁₆₂₀ All 2.52 0.72 2295 All Tax 3 2.57 0.70 1620 1_{Data from 1970-2014}

2_{Excluding AK, HI, IA, MT, NH, PA, UT}

3_{Excluding AK, HI, ID, IA, ME, MI, MT, NH, NC, OR, PA, UT, VT, VA,}

WV

Table 2. Summary statistics

Gallons of ethanol consumed per capita, U.S. population 21+ 1

Mean Std. Dev. N Beer 1.50 0.29 2295 Wine 0.38 0.21 2295 Wine Tax 2 0.39 0.21 1980 Spirits 1.06 0.55 2295 Spirits Tax 3 1.10 0.55 1620 All 2.94 0.86 2295 All Tax 3 2.99 0.85 1620 1_{Data from 1970-2014}

2_{Excluding AK, HI, IA, MT, NH, PA, UT}

3_{Excluding AK, HI, ID, IA, ME, MI, MT, NH, NC, OR, PA, UT, VT, VA,}

(16)

Data on legal information regarding marijuana laws by state was taken from The National Organization for the Reform of Marijuana Laws (NORML), an American non-profit organization. I also consulted The Marijuana Policy Project (MPP), another non-profit organization. The MPP is the largest organization in the United States working solely on marijuana policy reform. Some state official websites were also consulted for clarification and cross checking. As mentioned, decriminalization is a term regarding policy that is not very straightforward; the definition seems to be misconstrued by many. This study, and most legitimate sources acknowledge it as a policy where the state has enacted a law that imposes penalties other than jail time for possession of marijuana, at a minimum, for a first offense. There is sometimes a delay in when a law decriminalizing marijuana was passed and when it went into effect, and some discrepancies in information on specific years regarding the eleven states that decriminalized in the 1970s. Careful attention was paid to such issues when consulting the appropriate sources.

In this study, 17 states are observed as ever decriminalizing the possession of recreational marijuana by the first half of 2014. States that decriminalized after or more than halfway into 2014 (Delaware, Illinois, Maryland, Missouri, New Hampshire, Washington D.C.) were not given a dummy value of 1 (decriminalization status). There is a table in the appendix outlining the legal consequences for each relevant state regarding their decriminalization policies.

Data on state beer, wine and spirit excise taxes was taken from the Distilled Spirits Council of the U.S., Inc., compiled by the group Economic and Strategic Analysis (ESA). The ESA consulted the U.S. Department of the Treasury (Alcohol and Tobacco Tax and Trade Bureau) as well as the U.S. Department of Commerce (U.S. Census Bureau). They obtained state and local alcohol beverage revenue data and tax rates from various state and local agencies, as well as several published sources. This “History of Beverage Alcohol Tax Changes, 2015” lists each state and the change in its alcohol excise taxes for beer, wine and spirits from 1965-2015. If the tax change occurred after the middle of a year (as the information is quite specific and typically reports the month and year of the tax change), I implemented the tax change for the following year to keep consistency.

(17)

Lastly, it is crucial to highlight the limitations of the data I utilized. Firstly, in regards to alcohol consumption information, most available and appropriate data for this kind of study makes identifiable variables such as geographic location confidential and so, not accessible in the public use files available for my access. State level information is crucial for this study to utilize decriminalization status and the price of alcohol. Furthermore, when considering using survey results detailing consumption of marijuana and alcohol, again individual level data on marijuana and other drug consumption is unavailable in the public use files for most data sets. Thus, I was strained to use state level data, and not able to focus on individual characteristics. Regarding my empirical approach, these constraints prevented me from estimating demand equations or cross-price elasticities between the two goods. Another shortcoming of using cross-sectional/between state data is that there is variation in prices of alcohol and marijuana, the MLDA, alcohol taxes, and laws that decriminalize marijuana. Since variation in such crucial data is likely correlated with unobserved characteristics among the population living in the United States, it can sometimes make it difficult to infer causality from such cross sectional comparisons.

The data regarding alcohol consumption per capita is taken from the National Institute on Alcohol Abuse and Alcoholism (NIAAA). Much of the data are only estimates (beverage sales data, shipment data from industry sources) and are such, subject to error. Further, the ethanol conversion coefficients (ECCs) may result in inaccuracies of estimates of per capita alcohol consumption as light beers and wines contain less ethanol on average than the ECCs were constructed to indicate. Next, the estimates of alcohol consumption derived from sales data may be skewed due to cross border sales (individuals from one state purchasing alcohol from another), tourist alcohol consumption, unrecorded/unlawful alcohol production, variation in how States report alcoholic beverage sales, and possible delays in actual consumption and State records. The NIAAA does purposely put the word “apparent” in the title of the data for the above-mentioned reasons and most importantly, since the data is not actual measures of alcohol consumption, but based on reported beverage sales.17

Then, when considering how to control for the price of alcohol in regressions, there are several options but not without complications. Much of the relevant literature uses beer taxes or

17_{The official name of the report is “Apparent per capita alcohol consumption: national, state,}

(18)

prices as a proxy for the price of alcohol but there are studies that find beer taxes to be poor predictors of alcohol prices (Young, Bieliska-Kwapisz, 2001). For example, Chaloupka and Laixuthai (1997) justify using beer price as their proxy because beer is the most heavily consumed alcoholic beverage and is the beverage of choice among youths. I chose to utilize state excise taxes for beer, wine and spirits instead. An excise tax is an indirect type of taxation, included in the price of a product. It is typically levied on cigarettes, alcohol, gambling, etc. There are two types of excise taxes: specific and ad valorem. A specific excise tax is based on quantity, for example a fixed amount per gallon of an alcoholic beverage. An ad valorem tax, (ad valorem literally translates to ‘according to value’) is levied based on a value, in this context a percentage of the sale price of the beverages. In this study, only those states with specific excise taxes were kept in the data set when running regressions controlling for alcohol price using the proxy of excise tax, so this method is not without limitations either. My assumption is that the taxes are passed onto consumers in the form of a higher alcohol price. However, there are studies that conclude when controlling for state and period fixed effects, that excise taxes appear to be over shifted; this means that the price of the alcoholic beverage increases more than the tax amount (Young, Bieliska-Kwapisz 2002). There are also some findings that state alcohol taxes are not good indicators of beverage prices.

There are also limitations of using decriminalization to measure increased legal tolerance for marijuana.18 The term decriminalization is defined by the National Commission on Marijuana and Drug Abuse as “those policies in which possession of marijuana for personal use or casual distribution of small amounts for no remuneration [is] not considered a criminal offense” (1972). Pacula, Chriqui and King (2003) find that many states that have decriminalized marijuana have employed policies that do not meet this narrow definition. For example, the description of what constitutes a small amount of marijuana varies widely across the state statues.19

5. EMPIRICAL METHOD

In this study I employed a linear, ordinary least squares (OLS) regression with panel data. The aim is to estimate the causal effect of marijuana decriminalization on alcohol consumption. The

18_{Such issues are explained at length in Pacula, Chriqui, King (2003).} 19_{See Table 6 in the appendix for summary information of state statutes.}

(19)

initial assumption is that marijuana decriminalization is a good measurement of increased legal toleration of marijuana. Marijuana usage is a government intervention used to decipher the relationship between marijuana and alcohol. I assume that decriminalization reduces the price of marijuana usage. However, as mentioned previously, there is some evidence that this assumption about decriminalization measuring increased legal tolerance for marijuana is not without possible issues. The next assumption is that the relationship between decriminalization status of a state and alcohol consumption per capita is linear. I also assume that my data is a random sample, representative of the entire U.S; therefore I do not have a biased estimate of the causal effect on alcohol consumption in the United States since I am using data from every state and every year as well.

I use the panel data with 51 states measured over time (45 years). Panel data contains observations of multiple phenomena recorded for the same entity over multiple time periods. With this type of data, it is common to use fixed effects, a useful tool for removing omitted variable bias. Using fixed effects controls for unobserved heterogeneity when such heterogeneity is constant over time. I include fixed effects by creating state and year dummies in my regressions. In the context of this study, using state fixed effects eliminates the state specific characteristics that are constant over time. For example, if people in Massachusetts have a certain social attitude towards marijuana usage from 1970-2014 that differs from the attitudes in New Hampshire, there may be a bias on the estimates. By using state fixed effects, the impact this may have on the estimates is eliminated. By using year dummies, I implement time fixed effects, which eliminates trends that have changed over the years, but affected all states simultaneously. My assumption here is that these unobservable factors that differ between states are time invariant, and so by using fixed effects, I eliminate omitted variable bias. The linear regression employed has gallons of ethanol consumed per capita as the dependent variable and decriminalization status as the binary, independent variable. State dummies and year dummies are used, as well as a proxy for alcohol price. The following is my general regression model, where “Decrim. Status” takes a value of only 0 or 1:

!"#$. &'ℎ")*# ,-_./

(20)

I clustered the standard errors by state to avoid misleading small standard errors, and therefore falsely narrow confidence intervals, large t-statistics and low p-values. The assumption here is that standard errors are correlated by state and uncorrelated between different states. Clustering errors is based on the idea that there are some phenomena affecting groups of observations uniformly within the data distribution.

6. RESULTS

The estimated effects of marijuana decriminalization in the United States on alcohol consumption per capita are reported in Tables 3 and 4. The treatment variable had an effect that was not statistically significant from zero in any case. From an economic perspective, the results imply that the price of marijuana does not fall by as much as previously expected, or do not fall enough to cause a change in alcohol consumption. If the sign on the coefficients for decriminalization were negative, the interpretation would be that when a state decriminalized marijuana, alcohol consumption decreases, inferring the two goods are substitutes. Relatedly, if the sign was positive, this would imply that when a state decriminalizes marijuana, alcohol consumption increases, suggesting the goods are complements. In this study, the coefficients are mostly negative, however none of these coefficients are statistically significant from zero at the 0.05 significance level. The conclusion therefore is that although there is no discernable relationship between the two goods, I cannot prove or refute that they are neither complements nor substitutes. I simply conclude that decriminalization policy does not appear to impact alcohol consumption.

The results are reported in Tables 4 and 5. For interpretation, a standard drink (defined by the NIAAA) contains 0.005 gallons of ethanol. Keeping this in mind, when observing the decriminalization coefficient in Table 4, column 3, when a state decriminalizes marijuana, alcohol consumption of all beer per capita decreases by 0.045 gallons. This translates to about 9 standard drinks per year per person. The decrease is arguably small; it is maybe not surprising that the estimate then is statistically insignificant. I can exclude with 95% that the effect of decriminalization is larger than the top confidence interval and smaller than the bottom confidence internal for each coefficient, and it is important to note that all estimates contain 0 in their confidence interval. See Table 3 for the 95% confidence intervals for each coefficient.

(21)

These intervals contain the true value of the causal effects with 0.95 probabilities, and one can observe that the values within these intervals are all quite small. My conclusion of no significant causal effect here is based primarily on the fact that the estimates are small, rather than that the standard errors (reported in Tables 4 and 5) are large. With all coefficients statistically insignificant, I conclude that decriminalizing marijuana does not impact alcohol consumption in the United States.

Tax Tax 21+ population [-0.076, 0.029] [-0.065, 0.039] [-0.090, 0.035] [-0.076, 0.049] [-0.34, 0.13] [-0.41, 0.14] [-0.40, 0.17] [-0.48, 0.18] [-0.53, 0.19] [-0.60, 0.23] [-0.15, 0.056] [-0.63, 0.24] [-0.69, 0.29] [-0.13, 0.042] [-0.16, 0.071] [-0.14, 0.055] 14 + population

Table 3. 95% Confidence Bounds for Decriminalization Coefficients

Intervals containing change in gallons of ethanol consumed per capita with 95% confidence

(22)

Table 4. Effects of marijuana decriminalization on alcohol consumption per capita, U.S. population, age 14+ 1

Gallons of ethanol consumed per capita

All Beverages Beer Wine Spirits

(1) (2) (3) (4) (5) (6) (7) (8) Decriminalization -0.17 -0.18 -0.045 -0.044 -0.023 -0.013 -0.10 -0.14 (0.18) (0.20) (0.050) (0.043) (0.026) (0.026) (0.12) (0.14) All Tax -0.035*** (0.0099) Beer Tax -0.19*** (0.038) Wine Tax 0.014** (0.0067) Spirits Tax -0.14 (0.0091) Constant 2.78*** 3.26*** 1.338*** 1.52*** 0.13*** 0.38*** 1.31*** 1.43*** (0.066) (0.014) (0.023) (0.089) (0.013) (0.034) (0.054) (0.010) Observations 2295 1620 2295 2295 2295 1980 2295 1620

Standard errors in parentheses robust to clustering at a state level. State and year fixed effects included for all regressions.

*** p<0.01, ** p<0.05, * p<0.1

1_{Data from 1970-2014}

(23)

Table 5. Effects of marijuana decriminalization on alcohol consumption per capita, U.S. population, age 21+

1

Gallons of ethanol consumed per capita

All Beverages Beer Wine Spirits

(1) (2) (3) (4) (5) (6) (7) (8) Decriminalization -0.19 -0.20 -0.047 -0.045 -0.028 -0.13 -0.12 -0.15 (0.21) (0.24) (0.059) (0.050) (0.031) (0.031) (0.14) (0.16) All Tax -0.044*** (0.12) Beer Tax -0.23*** (0.043) Wine Tax -0.020*** (0.0080) Spirits Tax -0.017 (0.011) Constant 3.42*** 4.03*** 1.64*** 1.87*** 0.17*** 0.46*** 1.60*** 1.76*** (0.081) (0.17) (0.029) (0.099) (126.3) (0.041) (0.066) (0.13) Observations 2295 1620 2295 2295 2295 1980 2295 1620

Standard errors in parentheses robust to clustering at a state level. State and year fixed effects included for all regressions.

*** p<0.01, ** p<0.05, * p<0.1

1_{Data from 1970-2014}

(24)

7. CONCLUSION

Many previous studies find that marijuana and alcohol are economically related, as substitutes or complements. Most of these studies also use data at an individual level and rely on self-reported used of drugs such as marijuana and alcohol. This study takes a different approach by using state level data and examining the impact of a marijuana policy on alcohol consumption. It does not find strong results that imply any kind of economic relationship between the two goods, although there may exist some complementarity or substitution.

It appears that marijuana policy has two main aims: to minimize the social costs and individual consequences of using the drug and to minimize health and safety hazards associated with use. If the goal of decriminalizing marijuana is to also reduce alcohol consumption, the results from this study imply the policy may fail. In addition, arguments about marijuana being a gateway drug specifically for alcohol are called into question by this study. Overall, understanding the relationship between alcohol and marijuana is important in the context of public policy. If lawmakers utilize the MLDA to decrease youth alcohol consumption for example, and the goods are truly substitutes, this may unknowingly cause a spike in youth marijuana usage, which could have important consequences. On the other hand, if the goods are complements, raising the MLDA could be beneficial and quell marijuana usage. The nature of the outcome of certain policies regarding the two drugs is very much reliant on their true, economic relationship to be most effective. In the end, I have not found evidence of the substitution effects hypothesized by advocates of decriminalization, nor have I found evidence of the gateway effects hypothesized by those opposed to the policy, however, neither theory can be safely refuted by the results of this study.

(25)

8. REFERENCES

Alcohol Epidemiologic Data System (2004). Unpublished data from the 2001–2002 National Epidemiologic Survey on Alcohol and Related Conditions, National Institute on Alcohol Abuse

and Alcoholism.

Chaloupka, F.J. & Laixuthai, A. (1997). Do Youths Substitute Alcohol and Marijuana? Some Econometric Evidence. Eastern Economic Journal, [Online] 23(3), 253-276. Available at: http://www.jstor.org/stable/40325781 [Accessed 23 April 2017].

Chen, C.M., Yi, H. & Faden, V.B. (2015). Surveillance Report #101: Trends in Underage Drinking in the United States, 1991–2013. NIAAA, Division of Epidemiology and Prevention

Research, Alcohol Epidemiologic Data System.

Crost, B. & Guerrero, S. (2012). The effect of alcohol availability on marijuana use: Evidence from the minimum legal drinking age. Journal of Health Economics, [Online] 31(1), 112-121. Available at: https://doi.org/10.1016/j.jhealeco.2011.12.005 [Accessed 25 April 2017].

DiNardo, J. & and Lemieux, T. (1992). Alcohol, Marijuana, and American Youth: The

Unintended Effects of Government Regulation, National Bureau of Economic Research Working

Paper No. W4212, [Online] Available at: https://ssrn.com/abstract=227070 [Accessed 20 April

2017].

Drug Policy Alliance (2017a). A Brief History of the Drug War, [Online] Available at:

http://www.drugpolicy.org/facts/new-solutions-drug-policy/brief-history-drug-war-0 [Accessed 5 July 2017].

Drug Policy Alliance (2017b). Drug War Statistics, [Online] Available at: http://www.drugpolicy.org/drug-war-statistics [Accessed 5 July, 2017].

Economic and Strategic Analysis (2015). History of Beverage Alcohol Taxes, 2015, Distilled

Spirits Council of the U.S.,Inc.

Farrelly, M. C., Bray, J. W., Zarkin, G. A., Wendling, B. W., & Pacula, R. L. (1999). The effects of prices and policies on the demand for marijuana: Evidence from the national household surveys on drug abuse. National Bureau of Economic Research Working Paper No. 6940, [Online] Available at: http://dx.doi.org/10.3386/w6940 [Accessed 25 April 2017].

Good, A. (2015). The Effect of Marijuana Availability on Alcohol Use: Evidence from Marijuana Legalization. University of Notre Dame, Department of Economics, [Online]

Available at: https://economics.nd.edu/assets/165128/alex_good_research_paper1.pdf [Accessed 5 May 2017].

Grossman, M., Chaloupka, F.J. & Shim, K. (2002). Illegal Drug Use And Public Policy, Health

Affairs 21, [Online] 21(2),134-145. Available at: http://content.healthaffairs.org/content/21/2/134

(26)

Haughwout, S.P., LaVallee, R.A. & Castle, I-J.P. (2016). Surveillance Report #104: Apparent Per Capita Alcohol Consumption: National, State, and Regional Trends, 1977–2014. NIAAA,

Division of Epidemiology and Prevention Research, Alcohol Epidemiologic Data System.

Marijuana Policy Project (2017). State Policy, [Online] Available at: https://www.mpp.org.states/ [Accessed 5 June 2017].

National Commission on Marihuana and Drug Abuse (1972). Marihuana: A Signal of

Misunderstanding. First Report of the National Commission on Marihuana and Drug Abuse.

U.S. Government Printing Office.

Nephew, T.M., Yi, H., Williams, G.D., Stinson, F.S., & Dufour, M.C. (2004). U.S. Apparent Consumption of Alcoholic Beverages Based on State Sales, Taxation, or Receipt Data, U.S.

Alcohol Epidemiologic Data Reference Manual, 1(4). NIAAA, Alcohol Epidemiologic Data

System.

Organization for Economic Co-operation and Development, Consumer Price Index: Total All Items for the United States© [CPALTT01USA661S], Federal Reserve Bank of St. Louis, [Online] Available at: https://fred.stlouisfed.org/series/CPALTT01USA661S, July 22, 2017 [Accessed 20 May 2017].

Pacula, R. (1998a). Adolescent Alcohol and Marijuana Consumption: Is There Really A Gateway Effect? National Bureau of Economic Research Working Paper No. 6348, [Online] Available at: http://www.nber.org/papers/w6348 [Accessed 18 May 2017].

Pacula, R. (1998b). Can increasing the beer tax reduce marijuana consumption? Journal of

Health Economics, [Online] 17(5) 557-585. Available at:

https://doi.org/10.1016/S0167-6296(97)00039-8 [Accessed 2 May 2017].

Pacula, R., Chriqui, J. & King, J. (2003). Marijuana Decriminalization: What Does it Mean in the United States? National Bureau of Economic Research Working Paper No. 9690, [Online] Available at: http://www.nber.org/papers/w9690 [Accessed 25 April 2017].

Saffer, H. and Chaloupka, F.J. (1999). Demographic Differentials in the Demand for Alcohol and Illicit Drugs. In: F.J. Chaloupka, M. Grossman, W. Bickel and H. Saffer, ed., The Economic

Analysis of Substance Use and Abuse: An Integration of Econometrics and Behavioral Economic Research, [Online] 187-212. Available at: http://www.nber.org/chapters/c11160 [Accessed 30

April 2017].

The National Organization for the Reform of Marijuana (2017). State Laws, [Online] Available at: http://norml.org.laws [Accessed 15 June 2017].

Thies, C. & Register, C. (1993). Decriminalization of Marijuana and the Demand for Alcohol, Marijuana and Cocaine. The Social Science Journal, [Online] 30(4), 385-99. Available at: https://doi.org/10.1016/0362-3319(93)90016-O [Accessed 20 April 2017].

(27)

U.S. Food and Drug Administration (2014). Significant Dates in U.S. Food and Drug Law

History, [Online] Available at:

https://www.fda.gov/AboutFDA/WhatWeDo/History/Milestones/ucm128305.htm [Accessed 19 July 2017].

Williams, J., Pacula, R., Chaloupka, F.J. (2004). Alcohol and marijuana use among college students: economic complements or substitutes? Health Economics, [Online] 13(9), 825–843. Available at: http://dx.doi.org/10.1002/hec.859 [Accessed 5 June 2017].

Young, D. & Bieliska-Kwapisz, A. (2002). Alcohol Taxes and Beverage Prices. National Tax

Journal, [Online] LV(1), 57-73. Available at:

(28)

9. APPENDIX

State Amount Penalty Jail Time Max Fine

0 - 1 oz. No penalty None $0

1 - 4 oz. in your residence Not classified N/A $0

1+ oz. Misdemeanor 6 months $500

0 - 1 oz.a _{No penalty} _None _$100

More than 1 - 2 oz. Petty offense None $100

Less than 1/2 oz. (First Offense) Civil penalty N/A $150

Less than 1/2 oz. (Subsequent Offense) Civil penalty N/A $500

More than 1/2 oz. Misdemeanor 1 year $2,000

Up to 1 oz. Civil penalty None $100

1 oz. - less than 175 g Misdemeanor 3 months $575

10 g or less Civil violation None $200

More than 10 - 30 g (First Offense) Misdemeanor 1 year $2,500

More than 10 - 30 g (Subsequent Offense)Felony 1*-6 years $25,000

2.5 oz. or less No penalty None $0

More than 2.5 - 8 oz. Crime 6 months $1,000

Less than 10 g Civil offense None $100

10 g - less than 50 lbs Misdemeanor 1 year $1,000

1+ oz. (First Offense) Misdemeanor 6 months $500

1+ oz. (Subsequent Offense) Misdemeanor 2 years $2,000

0 - 1.5 oz. b _Misdemeanor _N/A _$200

1.5 oz. - 10 kg Felony 5 years $10,000

0 - 1 oz. (First Offense) N/A N/A $250

0 - 1 oz. (Subsequent Offense) Misdemeanor 5*-60 days $250

1 - 8.8oz. Felony 1-3 years $1,000

Up to 10 g (First Offense) Misdemeanor None $500

Up to 10 g (Subsequent Offense) Misdemeanor 1 year $2,000

0 - 1 oz. (First Offense) Infraction N/A $300

0 - 1 oz. (Subsequent Offense) Misdemeanor 5 days $500

More than 1 oz. - 1 lb. Misdemeanor 3 months $500

1+ oz. Misdemeanor N/A $600

0 - 25 g (First Offense) Not classified N/A $100

0 - 25 g (Subsequent Offense) Not classified N/A $200

More than 25 g - 2 oz. Misdemeanor 3 months $500

0 - 1/2 oz. Misdemeanor N/A $200

1/2 - 1 1/2 oz. Misdemeanor 1-45 days $1,000

0 - 100 g Misdemeanor N/A $150

100 - 200 g Misdemeanor 30 days $250

More than 1 - 2 oz. Violation N/A $650

Less than 1 oz. Civil violation None $150

1 oz. - 1 kg Misdemeanor 1 year $500

0 - 1 oz. (First Offense)c _{Civil violation} _None _$200

0 - 1 oz. (Subsequent Offense)a _{Civil violation} _None _$300

1 - 2 oz. (First Offense) Misdemeanor 6 months $500

0 - 1 oz. (Private possession/consumption)No criminal penalty None $0

0 - 1 oz. (Public consumption) Civil penalty None $100

1oz. - 40 g Misdemeanor 24 hours*- 90 days $1,000

0 - 2 oz.a _{No penalty} _None _$0

2+ oz. Misdemeanor 6 months $1,000

* Mandatory minimum sentence a_{For 21 years +}

b _{(A conditional discharge is possible for first time offenders. There is a possible drug education course requirement.)}

c_{Initiative 71, which took effect on 2/26/15, permits adults 21+ years of age to possess up to 2 oz. of marijuana in one's primary residence without penalty.}

Source: National Organization for the Reform of Marijuana Laws (NORML)

South Dakota also enacted a decriminalization provision in 1977, but it was immediately repealed. Only two states have changed their status since these original laws were passed. In 1990, Alaska modified its statute and recriminalized possession of marijuana. In 1996, Arizona decriminalized possession of small amounts of marijuana.

California Alaska Connecticut Illinois Missouri Rhode Island Delaware Maryland Nebraska Mississippi Minnesota Massachusetts Maine Colorado

Table 6. Summary of Decriminilization Statutes, Penalties for Small Amounts of Personal Use

Washington D.C. Washington Vermont Oregon Ohio North Carolina New York Nevada

(29)

Decriminalized state trends in ethanol consumption of all beverages 0.00 1.00 2.00 3.00 4.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Gal. ethanol consumed PC Decriminalization policy

Alaska 14+ 0.00 2.00 4.00 6.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Alaska 21+ 0.00 1.00 2.00 3.00 4.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

California 14+ 0.00 1.00 2.00 3.00 4.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

California 21+ 0.00 1.00 2.00 3.00 4.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Colorado 14+ 0.00 1.00 2.00 3.00 4.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

(30)

0.00 1.00 2.00 3.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Connecticut 14+ 0.00 1.00 2.00 3.00 4.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Connecticut 21+ 0.00 1.00 2.00 3.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Maine 14+ 0.00 1.00 2.00 3.00 4.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Maine 21+ 0.00 1.00 2.00 3.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Massachusetts 14+ 0.00 1.00 2.00 3.00 4.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

(31)

0.00 1.00 2.00 3.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Minnesota 14+ 0.00 1.00 2.00 3.00 4.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Minnesota 21+ 0.00 0.50 1.00 1.50 2.00 2.50 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Mississippi 14+ 0.00 1.00 2.00 3.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Mississippi 21+ 0.00 0.50 1.00 1.50 2.00 2.50 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Nebraska 14+ 0.00 1.00 2.00 3.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

(32)

0.00 2.00 4.00 6.00 8.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Nevada 14+ 0.00 2.00 4.00 6.00 8.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Nevada 21+ 0.00 1.00 2.00 3.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

New York 14+ 0.00 1.00 2.00 3.00 4.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

New York 21+ 0.00 0.50 1.00 1.50 2.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

North Carolina 14+ 0.00 0.50 1.00 1.50 2.00 2.50 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

(33)

0.00 0.50 1.00 1.50 2.00 2.50 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Ohio 14+ 0.00 1.00 2.00 3.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Ohio 21+ 0.00 1.00 2.00 3.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Oregon 14+ 0.00 1.00 2.00 3.00 4.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Oregon 21+ 0.00 1.00 2.00 3.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Rhode Island 14+ 0.00 1.00 2.00 3.00 4.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

(34)

0.00 1.00 2.00 3.00 4.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Vermont 14+ 0.00 1.00 2.00 3.00 4.00 5.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Vermont 21+ 0.00 1.00 2.00 3.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years

Washington 14+ 0.00 1.00 2.00 3.00 4.00 G a llo n s o f e th a n o l 1970 1980 1990 2000 2010 Years