THE IMPACT OF THE CRAFT BEER MARKET BOOM ON FATALITIES

A THESIS

Presented to

The Faculty of the Department of Economics and Business

The Colorado College

In Partial Fulfillment of the Requirements for the Degree

Bachelor of Arts

By

Bradley Hale

March 2017 THE IMPACT OF THE CRAFT BEER MARKET BOOM ON DRUNK DRIVING FATALITIES

Bradley Hale

March 2017

Economics

Abstract

Drunk driving fatalities in the United States have decreased significantly in the last several decades. To explain this trend, current literature examines possible reasons for the reduction and points to correlations between legislation, police enforcement levels, education levels, and abuse programs. However, there are limited studies analyzing the direct relationship between drunk driving fatalities and craft breweries. This paper studies this relationship at a nationwide level across the United States from 1994-2014 to better understand the complexities of the craft beer industry revolution and attempt to explain the decline in drunk driving deaths. To represent craft breweries, total breweries are analyzed while controlling for craft beer with a variety of demographic variables and other control variables. This study concludes that craft breweries are negatively correlated with drunk driving fatalities. Although the explanation to this discovery is challenging, the results provide insight to policy makers on the relationship between the rapid expansion of the craft beer market and drunk driving. These conclusions may lead to continued strategic efforts to eliminate drunk driving from our society.

KEYWORDS: (craft beer, beer, drunk driving, fatalities) JEL CODES: (L66, K14, J10) ON MY HONOR, I HAVE NEITHER GIVEN NOR RECEIVED UNAUTHORIZED AID ON THIS THESIS

Signature Acknowledgements

First and foremost, I would like to thank Dr. Kristina Lybecker for being my research mentor and for providing strategic guidance throughout this project.

Thank you to Dr. Kevin Rask, Colorado College Economics Paraprofessional Jackie Dugan and McKinley Sielaff for assisting in technical data problems.

Additionally, I would like to thank Bart Watson and Sarah Wolf from the Brewers Association, Bill Campbell, John Mann, Scott Whitley, Matt Kruger, and James Harrison for proving help and insight. TABLE OF CONTENTS ABSTRACT ...... ii

ACKNOWLEDGEMENTS ...... iv

1 INTRODUCTION AND EXPLORATION OF THE PROBLEM ...... 1 Introduction ...... 1 Literature Review...... 3

2 THEORY ...... 12 Base Model ...... 12 Modifications ...... 13 Empirical Model ...... 14

3 DATA ...... 16 Dataset...... 16 Dependent Variable ...... 17 Independent Variables ...... 18 Summary Statistics...... 20 Advantages and Limitations ...... 21

4 INTERPRETATIONS AND LIMITATIONS OF THE ANALYSIS ...... 23 Results and Analysis ...... 23

5 CONCLUSION AND IMPLICATIONS FOR FUTURE RESEARCH ...... 29

WORKS CONSULTED ...... 32 SECTION I

INTRODUCTION AND EXPLORATION OF THE PROBLEM

Introduction

During the past 20 years, the craft beer market has shown tremendous growth.

From 1994 with 537 craft breweries in the United States to 4,225 craft breweries in

20141, the boom is widespread and continuing to expand. Craft beer is different from general macro beers in a variety of ways. A craft brewery is a small, independently run operation that uses traditional ingredients from malted barley, not diluted with corn or rice. According to the Brewers Association (2017)2, which works to protect craft beers and the community of brewing enthusiasts, small refers to brewing less than six million barrels per year. Independent means less than 25% of the brewery is owned by a macro or non-craft brewer. Although craft breweries are small, Bart Watson, Chief Economist at the Brewers Association, calculates the craft beer market to represent 21% of the overall beer market in terms of dollar sales volume.3 Furthermore, craft breweries typically focus on the surrounding communities including philanthropy, product donations, volunteerism and sponsorship of events. Craft brewers maintain their independence by establishing connections with customers.

Additionally, another primary component to craft beer is the alcohol percentage.

Craft beer is disrupting the beer industry and palates of beer consumers. Although craft

1 The number of breweries on a nationwide scale according to the Brewers Association based in Boulder, CO. 2 The Brewers Association defines an American craft brewer as small, independent, and traditional with specific statistical metrics for each of these three variables. Hence, the following statistics are all retrieved from the Brewers Association website. 3 The overall beer market represents $105.9 billion in sales within the United States, while the craft beer market represents $22.3 billion in sales.

1 beers typically have an increased ABV (alcohol by volume), with an average of 5.9% while the average domestic beer typically has an ABV of 4.5%, the consumers tend to have a greater understanding and appreciation of their drink of choice. The recent growth in popularity of the craft beer segment derives from a change in consumer demand attributed to the increased buying power of the millennial generation; individuals born between 1980-2000, who demand authentic and customizable product options. Further, millennial customers want to collaborate and represent certain brands that align with their values, especially in the age of social media where social-decision making influences consumption habits. The target consumer base for craft beer is full of beer enthusiasts not simply interested in the effects of alcohol, but exploring different places, trends, and lifestyles. Nonetheless, an important question is the increased alcohol content having detrimental effects on society.

Since the end of , policy makers have been working to curb the social problem of drinking alcoholic beverages and driving under the influence. This research study seeks to understand the relationship between drunk driving fatalities and the number of craft breweries across the United States. Since 1982, drunk driving accidents have decreased by 51% in the United States4, while the amount of beers with an ABV of

6.5% of greater increased by 319% in the U.S from 2011 to 2014. However, many other factors can contribute to BAC () levels and ultimately DUI deaths, such as the amount of police enforcement and education levels of consumers. Therefore, an examination of the influences of craft beer within each state, which varies on

4 According to the Foundation for Advancing Alcohol Responsibility, drunk driving fatalities in America have decreased 51% and total traffic fatalities have decreased by 20% since 1982. Nonetheless, drunk drivers still account for 69% of the nation’s drunk driving fatalities, in which there is a known alcohol test result for the driver.

2 population and percentage of a younger millennial generation is necessary to accurately determine the relationship between drunk driving fatalities and craft breweries.

This study uses data for the number of DUI fatalities issued by the National

Highway Traffic Safety Administration (NHTSA) in each state from 1994-2014. In order to comprehend the factors influencing the craft beer expansion and the declining rate of drunk driving deaths, this study utilizes a random effect GLS regression model to correlate the number of drunk driving arrests in a state to its demographics, including income, number of young adults, population, vehicle miles traveled, licensed drivers, education levels, as well as binary variables for marijuana legalization and BAC drunk driving laws. These variables will be defined in more detail later in this study.

Understanding the consumer demand of craft beer will provide relevant insights into the emerging focus of localized and specialized products and the correlations between drunk driving. This study expands on contributions to current literature, but as the first study to examine a connection between drunk driving and the increase of craft breweries, at a nationwide level, this model implements modifications to previous examples and predicts the craft beer revolution to show a significant negative correlation between the reduction in drunk driving fatalities.

Literature Review

Various theories examine the influx of craft brewers and more importantly the future consequences in the market. Although the literature encompasses a wide variety of theories, this review will focus on drunk driving history within the United State and the possible reasons for the decline in drunk driving rates. Studies within the craft beer market are limited because of the recent emergence and popularity; therefore the

3 following research investigations are dissimilar from this particular study. However, they will provide conceptual context into the relationship between craft breweries and drunk driving as well as a better understanding of the complexities within the industry. This study uses parts of models of recent literature with the main adaptation to the craft beer market measured by number of breweries in each state.

According to the NHTSA, there was an alcohol-impaired traffic fatality every 52 minutes in 2014. Alcohol-impaired fatalities are defined as crashes with at least one drive with a BAC at or above .08% and account for about 30% of all crash deaths while driving.5 Further, the definition of drunk driving is common across the United States; currently all states and the District of Columbia have the same BAC law of .08%.

However, it was not until 2000 that Congress required states to have this law. Although drunk driving fatalities across the U.S. have decreased by 51% since 1982, the commitment amongst organizations such as Mothers Against Drunk Driving (MADD) and Students Against Destructive Decisions (SADD) is to eliminate drunk driving.

A study conducted by Kathryn Stewart, Robert Voas, and James Fell of the

NHTSA (1996) addresses possible reasons for the decline in drunk driving in the United

States. The main focus of this paper is safety program initiatives that have reduced alcohol related highway deaths. After analyzing possible variables such as vehicle miles traveled, population, licensed drivers, trends in alcohol consumption and economic conditions, safety improvements in vehicles and roadways between 1982 and 1994, the authors concluded that their model was not best fit. They attributed the reduction to a wide variety of factors including deterrence, including enforcement practices,

5 Retrieved from the NHTSA and the United States Department of Transportation research and data page.

4 administrative license revocation, lower BAC limits, raising of the drinking age to 21, and increased public awareness and activism.

However, the authors could not proportionately determine which individual factors contributed to the decline in drunk driving fatalities. In summary, the study indicated that regions that incorporated checkpoints had significant decreases in drunk driving fatalities, but the amount is not known. Further, the study concluded that loss of license arrest laws reduced fatal nighttime crashes involving alcohol. In working with the NHTSA, the authors determined that before Congress adopted the .08 BAC requirement, 11 states reduced their BAC limit from .10 to .08 and saw statistically significant decreases in drunk driving fatalities. The increase in drinking age to 21 occurred in 1975 and most certainly attributed to the reduction according to this study.

Additionally, the number of drunk driving activist groups grew dramatically in the 1980s and led to increased legislation with enforcement and prosecution penalties.

A research study performed by Emily Miller (2016) examines the impact of public service announcements (PSAs) on drunk driving fatalities in specific states from

1995-2010. The effectiveness of PSAs is measured using an OLS regression and yields results where the timing of the announcement is more predictive of greater reductions in drunk driving fatalities. The data is retrieved from the Fatality Analysis Reporting

System (FARS) through the NHTSA and only records accidents involving fatalities, not all injuries. Primetime PSAs (7:00PM-10:59PM) result in the largest predictors for both adult and underage drunk drivers. As the announcements got farther away from

Primetime, the effects got closer to zero. PSAs are an important tool for policy makers

5 and community leaders to decrease drunk driving and reveal some apparent effectiveness

in this study.

The Brewers Association and their “Savor the Flavor Responsibly” campaign

aims to protect the integrity of beer producers as well as the right of consumers to enjoy

different beer styles. Craft beer companies are aware of the potential problems with their

products and consumers behind the wheel. According to Bellware (2014), firms are

embracing a less is more attitude as breweries like Founder’s (Grand Rapids, Michigan),

Deschutes Brewery Inc. (Bend, Oregon) and Lagunitas Brewing Co. (Petaluma,

California) are focusing on session beers with ABV levels close to 5.5% opposed to

hoppy bombers with double-digit ABV levels. The research suggests the craft beer

market is trying to widen its audience and offer a greater selection to all beer drinkers.

According to The Journal of Consumer Marketing (2014), the market is maturing as more consumers gain exposure with the craft segment growing as the overall beer market slows. Although mass market brews will always exist, consumers are demanding more flavor and need to be aware of the possible driving consequences.

Furthermore, the presence of alcohol does not simply imply that alcohol caused the crash. Reed (2016) conducts numerous studies that compare BAC levels of drivers in crashes with BAC levels of drivers not involved in a crash. The data reveals that 24% of the accidents would not have occurred if alcohol was not involved, but not majority of crashes. For example, over 50% of all fatal highway crashes involving two or more cars are alcohol related.6 Reed also examined the effects of increased drunk driving enforcement on the number of accidents. The National Research Council (2016) panel

6 These statistics were retrieved from Fathers Against Drunk Driving and complied by the United States Department of Transportation and Department of Safety.

6 confirmed the findings, in that effectively enforced laws deter driving under the influence of alcohol.

Further research examines the effect drinking age laws have on traffic accidents.

Given that the laws apply to all drivers within each state, the crashes can be compared

before and after accidents. The National Minimum Drinking Age Act passed in 1984

restricting the purchasing and publicly possessing of alcoholic beverages to persons

under 21 years of age in every state across the U.S. The United States General

Accounting Office (1996) examined results from 49 studies that adopted the minimum

of 21 and confirmed that raising the drinking age has a direct effect on

reducing alcohol related traffic accidents among young adults affected by the laws

(Toomey, Rosenfeld, and Wagenaar, 1996). These laws reduced youth drinking by

eliminating some of the alcohol availability and by developing a punishment threat for

breaking the law. The reason for raising the drinking age to 21 has been successful in

reducing traffic crashes.

Furthermore, zero tolerance laws are a combination of the minimum legal

drinking age law and per se laws that restrict driving to certain BAC levels. The majority

of the population supports zero tolerance laws. Hingson (1995) studied 12 states that

lowered BAC limits for younger drivers and found that 22% of their nighttime crashes

were eliminated in states with a 0.00 BAC limit. Additionally, Voas (1999) discovered

that zero tolerance laws reduced underage drunk driving accidents by 24%. Zero

tolerance laws most likely influenced the behavior associated with minimum legal

drinking age laws where drinking underage is somewhat accepted in society, as drinking

and driving is not.

7 The FBI (1995) studied the relationship between the number of DWI arrests and liquor law violation arrests for individuals under 21 in each state annually from 1989-

1995. The model showed the number of youth drinking drivers involved in accidents as a function of DWI arrests, liquor law arrests, and year and state. The regression analysis demonstrated significant values between enforcement measures and youth drinking and driving.

Successful law enforcement works to increase the public’s perception of

enforcement levels without focusing solely on arrest levels. For example, driving

checkpoints usually result in few arrests, but create publicity for the public to believe an

effect has occurred. The majority of these studies have focused on governmental

measures motivated to fear the population of the legal consequences. School programs,

such as MADD and SADD have also worked to educate the youth on crash and injury

risks while driving impaired and promote activities that do not involve alcohol.

However, throughout the research there is no direct evidence supporting these community

programs on the reduction on drinking and driving. Nonetheless, the combination of

information, skills, and education likely influences the public’s perception of decreasing

the rate of drunk driving across America.

Additionally, there are other external factors that might be reasons for the

reduction in drunk driving over time, including economic and travel changes and youth

behavior changes. Ulmer (2000) conducted a study to show the effects of travel,

population, and employment changes on overall alcohol related crashes. The results

concluded that vehicles miles of travel annually, the amount of employment and

unemployment are all related to the fatalities.

8 Moreover, strong economic conditions tend to produce more alcohol related crashes. This study also introduced another variable as a function of fatalities, per capita beer consumption. The linear constructed model suggests that as overall employment and income increase, so does drinking and driving because there are more funds to spend on gasoline, vehicle maintenance, and alcohol. Youth behavior changes due to the increased use of media, internet, cell phones, and other technology are also believed to have a relationship with drunk driving accidents. However, no study has examined these issues while addressing drunk driving specific to craft beer; scholars associate the media age with more well informed consumers and therefore more aware drivers of the severe

DUI consequences.

Although most research on drunk driving is focused on legislation and community

groups, there are some that, like this study, develop a model to reveal other underlying

relationships that effect drunk driving rates. For example, a study using cross-sectional

data from Mississippi found that as gas prices increase, drunk driving rates decrease (Chi,

2011). This study also found demographic variables of gender and race to be

significantly correlated to drunk driving accidents.

Another recent discussion within literature is social driving services. There has

been debate about how the ride sharing services effect alcohol related traffic fatalities.

Although there are 121 million drunk driving accidents in the U.S. every year, only 1.1

million individuals are arrested. In the American Journal of Epidemiology, Kirk and

Moon (2016) analyzed drunk driving statistics in 100 of the most populated cities in the

U.S. from 2009-2014 and found no correspondence between the rise in Uber usage to the

decrease in crashes. The study concludes that these findings will need to be revisited

9 down the road to adjust for the greater presence of Uber and Lyft in the marketplace.

Perhaps these new generation apps will develop a way to better appeal to drunk drivers with discounts to establish a real presence in the art of saving lives, as they often inaccurately claim.

A study developed by McMakin (2016) which will be used as this study’s base model examines the rise in the craft brewing industry and its effect on drunk driving in

North Carolina. McMakin uses the number of breweries opening per year to account for craft breweries apart from macro brewers. This statistic has its weaknesses because some counties do not have any breweries, but could still see sales of distributed craft beer. The drunk driving measurements are from the University of North Carolina Safety Research

Center and the North Carolina Governor’s Highway Safety Program for alcohol-related accidents. The main limitation with this data and any attempt to measure drunk driving crashes is that they are not beer specific; the alcohol consumption mix problem is a limitation of McMakin’s study and this research paper as well.

Furthermore, demographic variables are added as a set of controls and the results show that an increase of one brewery in a county in a year leads to about six more alcohol-related accidents. The interpretations of the author lead to more breweries causing more drinking and therefore more opportunities for drinking and driving.

However, this study does not take into account the craft beer consumer differences opposed to traditional beer drinkers. For example, craft beers are not advertised as a cheap way to get drunk, but rather as a delicacy that requires the right taste and food pairing. Therefore, a study conducted at the national level will provide greater insight into the relationship between craft beer and drunk driving.

10 Summary

The various studies reveal valuable insights into the craft beer market and more importantly provide context for further investigation. Though the limited literature implements focused analysis through the relationships between drunk driving and possible reasons for the decline, additional investigation is required to develop an extensive analysis of the dynamic between the craft beer revolution and DUI fatalities.

More specifically, provide statistically significant evidence that these two variables are linked at a nationwide level. This study will utilize various parts of models constructed in previous research, but modify them to fit the hypothesis. The next section seeks to develop an extended theoretical model to test the strength of this relationship.

11 SECTION II

THEORY

Although there is no current literature focusing on the relationship between the number of brewery permits and arrests for driving while intoxicated in regions surrounding new breweries, there is data for the number of brewery permits provided by the Alcohol and Tobacco Tax and Trade Bureau (TTB). To accurately understand the relationship between craft beer and drunk driving in the United States, this study utilizes a model of previous drunk driving research with the main modification of breweries at a state level.

Base Model

This study uses a random effects regression that correlates the number of DUI fatalities in a state for a given year to demographic variables, number of breweries, marijuana legalization and BAC control variables. Nonetheless, this study builds upon the work of Miller (2016), Stewart, Voas, and Fell (1996), and McMakin (2016). The base model of this study focuses on that of McMakin (2016) because of the implementation of craft breweries, although at county-level, and with the addition of demographic characteristics. McMakin uses a fixed effects pooled OLS model to capture unobserved heterogeneity across individuals. Although the research only analyzes craft breweries at a county-level across North Carolina, McMakin’s model accounts for many correlates of drunk driving identified in the existing literature.

McMakin uses the following model:

Cit= αi + θBit + β’Dit + ψ’Eit + εit

12 which reflects the number of alcohol fatalities in county i during a specific year t as a

function of the number of operating breweries Bit, demographic variables controlling for craft beer specifically Dit, a vector of the macroeconomic variables including unemployment rate and income levels Eit, and αi as an average of the county fixed effects

for each county. Demographic variables included are population, percent of the

population that is male, percent that is not white, and median age.

Modifications

For this study, some variables are implemented into the model to more accurately

represent the craft beer market at a national level and control for external factors that

could affect drunk driving crashes. This study adds demographic variables including

percent of the population between 20 and 29 to more accurately reflect the high

concentration of craft breweries across the United States and the positive correlation to a

state’s income level and population of young adults. Further, adding variables associated

with licensed drivers helps identify which states have a larger population driving

vehicles. An education variable is added to account for the increase in general awareness

and responsibility amongst the population in each state.

Additionally, more variables are utilized to account for other factors contributing

to drunk driving. Beyond adding demographic variables, a police enforcement variable is

used to control for the number of police officers within each state. Further, a vehicle

miles traveled (VMT) variable is added to account for the amount of driving in each state.

Finally, binary variables including marijuana legalization and BAC legal limits are added

to control for the influence of marijuana in specific states from 1994-2014 and the

adopted Congress action in 2000 requiring all states to abide by a .08 BAC legal limit.

13 These variables are extremely important to the integrity of the model because they control for possible reasons for the DUI reduction across the United States.

Empirical Model

The empirical model used in this study is as follows:

DUIit = β1Incomeit + β2YoungAdultsit + β3Populationit + β4Breweriesit + β5VMTit +

β6Policeit + β7LicensedDriversit + β8Educationit + β9Marijuanait + β10BACit

 DUI, representing the number of drunk driving fatalities from 1994-2014.

 Income, representing a state’s median annual household income from 1994-2014.

 YoungAdults, representing a state’s population of young adults age 20 to 29 from

1994-2014.

 Population, representing a state’s population from 1994-2014.

 Breweries, representing the number of breweries per state from 1994-2014.

 VMT, representing the amount of vehicle miles traveled per state from 1994-

2014.

 Police, representing the number of police officers per state from 1994-2014.

 LicensedDrivers, representing the number of licensed drivers per state from

1994-2014.

 Education, representing the percentage of the state’s total population 25 years

and over with a Bachelor's Degree or higher from 1994-2014.

 BAC, binary variable representing a BAC national level enforcement of .08 in

2000.

 Marijuana, binary variable representing marijuana legalization for specific states

from 1994-2014.

14 This section outlines the theoretical model to be tested. The desire to use this model approach is based on the intention to more accurately determine if the craft beer industry, away from the larger macro brewers in association with the millennial generation and the craft beer revolution is a contributing factor to the decrease in DUI fatalities. The following section will describe these variables and the purpose of this investigation.

15 SECTION III

DATA

This section examines the data sources and their use in the model, as well as the limitations and advantages of the model. This study uses data from the National

Highway Traffic Safety Administration (NHTSA), the Alcohol and Tobacco Tax and

Trade Bureau (TTB), the United States Census Bureau, the Brewers Association, the

United States Department of Justice and the United States Department of Transportation from 1994-2014.

Dataset

Variables were selected on previously identified predictors and control variables

for craft beer including the number of breweries in a state and the area’s demographic

makeup, including income, the number of young adults, population, education levels,

licensed drivers, police enforcement numbers and vehicle miles traveled. Further, binary

variables for marijuana legalization and BAC laws were added.

Although the statistics from the Alcohol and Tobacco Tax and Trade Bureau are

not exclusive to the craft beer industry, they provide insight into the number of firms

entering the beer industry. Furthermore, larger macro brewers rely exclusively on

economies of scale for low cost production and accounted for less than 2% of the total

number of breweries in the United States in 2005 (Elzinga, 2013). This provides a higher

level of confidence when examining the number of craft breweries within the market,

where the focus is on smaller production scales in local communities; where demand is

high and distribution is more expensive while ensuring the quality of the product.

16 Data on DUI statistics, specifically the number of fatalities in crashes involving an alcohol-impaired driver was retrieved from the United States Department of

Transportation; fatalities are defined as death by accident for these purposes. These data are from 1994-2014. Police enforcement data was retrieved from the United States

Department of Justice and the Bureau of Justice Statistics Uniform Crime Reporting

Program. These data have extremely small fluctuations over time across every state; therefore, the data act as a placeholder for each state. Binary variables were established to account for marijuana legalization and BAC law changes. Between 1994 and 2014, 17 states passed laws legalizing medical or recreational cannabis. More specifically,

Colorado and Washington legalized recreational marijuana in 2012, while Alaska and

Oregon legalized recreational marijuana in 2014. Nonetheless, the calculations were made only from legalization of cannabis, not CBD oils or decriminalization. Congress adopted 0.08 BAC as the national illegal limit for impaired driving in 2000. These resources helped to successfully generate independent variables for the model.

Dependent Variable

The dependent variable in this study is the number of DUI fatalities across each state within the United States, recorded by year from 1994-2014. As discussed in the literature review, very few studies have examined the craft beer industry and they usually focus on a marketing perspective in the factors influencing the craft beer market boom. This study seeks to expand on the existing literature to develop a more complete model and predict the dependent variable.

 DUI: the amount of drunk driving fatalities from 1994-2014 by state.

17 Independent Variables

This model uses a representation of the entire beer industry, while controlling for

craft beer using demographic characteristics as well as vehicle miles traveled and police

enforcement to model DUI fatalities in each state. The variables and their definitions are

as follows:

 Income: the amount of annual household income in the state.

 YoungAdults: the number of young adults age 20 to 29 in the state.

 Population: the number of people residing in the state.

 Breweries: the number of breweries operating in the state.

 VMT: the amount of vehicle miles traveled in the state.

 Police: the number of full-time officers reported to the Uniform Crime Reporting

Program, by sworn status in the state.7

 LicensedDrivers: the number of licensed drivers in the state.

 Education: the number of young adults between the age of 20 and 29 with a

Bachelor's Degree in the state.

7 Note: Totals provided for Uniform Crime Reporting (UCR) Program law enforcement employee data are the counts reported by the convenience sample of agencies that reported at least one employee to the FBI that year and do not represent a national estimate. All rates are calculated from state-level Census population statistics for the reference year. Because reporting to the UCR is voluntary and information on the number of sworn officers is based on a convenience sample of the agencies that reported, the number and rate of sworn officers in this report are likely to be lower than the actual number and rate of officers nationally.

Sources: Bureau of Justice Statistics, based on data from the FBI, Uniform Crime Reporting (UCR) Program police employee data, 1992–2012; and U.S. Census Bureau, Intercensal Estimates of the Resident Population for the United States, 1992–2012.

18  BAC: either 1 or 0; 1 accounting for the year 2000 and up for when Congress

adopted national law.

 Marijuana: either 1 or 0; 1 accounting for years where states have legalized

medical and/or recreational marijuana.

Demographic data was retrieved from the United States Census Bureau including median household income, population estimates, percentage of individuals age 20 and older across the state, licensed drivers and education levels. Median household income data was only available from 1994-2013. Although the legal drinking age for adults in the

United States is 21, difficulties gathering the respective data resulted in population estimations of 20 years of age and older. As mentioned in the literature review, the number of young adults in each state is a critical component to the rapid expansion of craft breweries in specific regions. The United States Census Bureau only provides data segregated between age ranges; therefore, data was retrieved for population estimates between the ages of 20 and 25 in addition of estimates for individual state’s population between the ages of 26 and 29. The licensed drivers’ data was calculated based upon the population estimates provided by the Census Bureau. According to the United States

Department of Transportation, from 1960 to the present, roughly half of the population is registered as a licensed driver.8 Due to the difficulties obtaining this data, this method was used as a proxy approach to accurately show the data. Additionally, the education data was gathered from the United States Census Bureau. The education variable is measured as the percent of the total population 25 years and over with a Bachelor's

Degree or higher; these calculations were made using the young adult population data.

8 Information retrieved from http://www.fhwa.dot.gov and used to calculate the number of licensed drivers based on population estimates from United States Census Bureau.

19 Despite the data limitations, the sample size is still sufficiently large across each state for

analysis to be considered relevant and illustrative of the market conditions.

Summary Statistics

Summary statistics for the dependent variable and independent variables are

presented in Table 1. Drunk driving fatalities by state range from a minimum of three to

a maximum of 1569, for District of Columbia and Texas respectively. The number of

breweries in a given state ranges from three to 654, for Alaska and California

respectively. States range in population size from the smallest with 327,539 people in

Wyoming, to the largest with 28,600,000 in California.

TABLE 1

SUMMARY STATISTICS

Standard Variable Count Mean Minimum Maximum Deviation State_code9 1071 26 14.72648 1 51 Year 1071 2004 6.05813 1994 2014 DUI 1071 240.4024 253.7874 3 1569 VMT 1071 55300000000 57300000000 33100000000 3330000000000 Police 1071 14110.36 16347.91 1000 75000 Income 1020 44111.93 9202.865 23564 71836 YoungAdults 1071 798158.9 921848.4 58659 5846747 Population 1071 4187999 4585738 327539 28600000 Breweries 1048 42.8979 58.6386 3 654 LicensedDrivers 1071 2791999 3057159 218359.3 19100000 Education 1071 186360 231384.3 1027.89 1555235 Marijuana 1071 0.1540616 0.3611765 0 1 BAC 1071 0.7142857 0.451965 0 1 Source: author’s calculation

9 The variable State_code was created to establish the use of panel data in the GLS regression analysis.

20 Advantages and Limitations

This study has several important advantages. First and foremost, research studies within the craft beer industry are extremely rare because of the recent rapid emergence.

According to individuals in the craft beer business, the opportunities to conduct research are extremely limited because of the access to data sources. Most craft breweries are private businesses, which makes acquiring data rather difficult.

Nonetheless, because the study is one of the first of its kind, limitations exist.

Although this model works to demonstrate a craft beer market by controlling for certain variables, perhaps craft beer consumer behavior still remains unrepresented. The difficulty in trying to reveal a correlation between DUI fatalities and craft breweries is that many other variables could play a role. According to Bart Watson, Chief Economist at the Brewers Association, data has shown crime decreases associated with more breweries coming into the market. However, it’s necessary to question if this results from gentrification. For example, DUI fatalities in Colorado have decreased in recent years as the number of craft breweries and population have grown exponentially.

The biggest limitations and challenges of this study are accounting for every variable that may play a role in drunk driving fatalities, such as anti-DUI advertising, closing times of bars and breweries, and the effect educational programs have on drunk driving. Further, the legalization of marijuana could contribute to the decrease in DUI fatalities which is difficult to measure accurately. It is also challenging to quantify the beer and alcohol consumption mix in blood levels of drunk drivers; for example, a drunk driver could have had no beer at all, let alone craft beer during or prior to their accident.

It’s also critical to remember that this is solely the individuals that are caught and

21 arrested. The fact that alcohol spirits are usually consumed with beers makes the argument more difficult to prove because more breweries doesn’t necessarily equal more beer drinking.

22 SECTION IV

INTERPRETATIONS AND LIMITATIONS OF THE ANALYSIS

This section presents the results of the Generalized Least Squares regression analysis of drunk driving fatalities at a nationwide level. This section also evaluates the methods and modifications to the model that were implemented to yield the empirical model and results. Overall, the model is able to capture the relationship between craft beer components of drunk driving.

The main focus of this study is understanding the role of craft beer in predicting drunk driving fatality rates across the United States. The coefficients for breweries is expected to be statistically significant and negative. This would indicate that states with more craft breweries would have a lower number of drunk driving fatalities. This is the expected relationship given the research and current consumer market behavior of craft beer consumers, as well as accounting for other possible reasons for the decline in DUI fatality rates.

Results and Analysis

The diagnostic results from the regression model shown in Table 2 illustrate that the coefficient sign of the breweries variable matches the anticipated results hypothesized from the review of relevant literature. According to the calculated results, for every unit change in breweries, there is a 0.04303355 change in DUI fatalities, in the opposite direction. In other words, as the number of breweries increases, there is a significant negative correlation to DUI fatalities. Therefore, it is reasonable to assume that attempting to control for various variables and factors contributing to DUI fatalities, the

23 emergence of the craft beer market has a relationship with consumers and driving under the influence of alcohol.

In addition to confirming the hypothesis, variables that are statistically significant with at least 95% confidence levels include: VMT, Police, Income, Population, and

Breweries. With the exception of VMT and Police, all of the aforementioned variables have a negative correlation to DUI fatalities. Therefore, unit increases for these variables correlate with a decrease in DUI fatalities. Although the results are statistically significant, the coefficients for these variable are relatively small, indicating a weak relationship to DUI fatalities. The variables of Income and Population yield interesting results. The significant negative correlation reveals that with increases in median income, there is a slight drop in DUI deaths. While Population experiences the same relationship, it cannot be predicted with the same level of confidence. The other two variables of VMT and Police suggest just the opposite. That is, unit increases in these variables would lead to an increase in DUI deaths. For example, for every unit increase in police, there is a 0.0148242 increase in DUI deaths. This makes intuitive sense because with more police and vehicle miles traveled, the opportunity to experience a drunk driving accident increases. Further, the variables of YoungAdults, Education,

Marijuana, and BAC yield no statistically significant results. Due to the low levels of confidence, no correlations or relationships can be determined.

24 TABLE 2

RESULTS FOR IMPACT OF DRUNK DRIVING FATALITIES

VARIABLE MODEL VMT 4.88e-09**** (6.84e-10) Police 0.0148242**** (0.0040375) Income -0.0012328**** (0.0003618) YoungAdults 0.0001702* (0.0001197) Population -0.0000937*** (0.0000331) Breweries -0.04303355*** (0.195469) Education -0.0000765* (0.0003209) Marijuana -4.950254* (9.619141) BAC 2.099825* (10.28059) Observations 997 R-squared 0.8199 Standard errors in parentheses **** p<0.01, *** p<0.05, ** p<0.1, * p<0.9 10 Source: author’s calculations

10 LicensedDrivers variable removed due to multicollinearity with Population variable. As mentioned in the Data Section of this study, LicensedDrivers is a function of Population.

25 Interpreting the results, it is important to remember the limitations of the study.

The lack of consistent significant correlations is likely due to the complexities of the

issue and model. There are many components to drunk driving which makes it difficult

to analyze accurately and efficiently. The analysis could change if the model can adapt to

the craft beer industry as it continues to grow. For example, as data becomes more

readily available in all regions of individual states and the country, data can be heavily

influenced by craft beer sales and consumption. If the model were different and more

accurately represented by craft beer producers, the regression analysis could account for

sales and consumption. Perhaps when individuals drink more craft beer in densely

populated cities, they are less inclined to drive. While remote communities and cities

offer more incentives to drive home after drinking. Further, results focusing on the changes on a year basis or state basis might provide insight into whether marijuana or

BAC law changes have an impact of drunk driving opposed to a nationwide outlook like this study.

Nonetheless, this study reveals interesting policy perspectives and interpretations.

Based on the findings of this study, the craft beer market boom is a possible reason for the decline in drunk driving fatalities. More than 50% of Americans live in the largest

144 counties and 75% live within ten miles of a craft brewery.11 The majority of craft breweries are located closely to where individuals live and perhaps that encourages walking to and from bars opposed to drunk driving. Craft beer is advertised in a variety of ways that focus on communities, trends and seasonal styles. According to Mintel’s

11 According to U.S. Census data from 2014, 50.03% of the United States population lives in the 144 most densely populated areas, while the Brewers Association confirms the number of Americans that live near craft breweries in 2016.

26 2013 consumer and market analysis report, the millennial generation is currently the largest purchaser of not only craft beer, but alcohol as a whole. The millennial generation looks for a premium beer that offers quality; the entire process is important to the consumer, from manufacturing and production to the way the beer is served. Craft breweries are selling more than just a beer, but an experience.

Take for example, Big Bend Brewing Company of Texas. Co-founder, Matt

Kruger wished to establish a unique craft beer experience in a specific region. Their

focus is specifically on the mountainous region of West Texas with small batch craft

beer. Kruger states, “We wanted to create a brewery that made world-class beer and

proudly represent the region and far West Texas with a great product…I just thought that

the area was so amazing and cool, and that if we made great beer, the rest of Texas and

beyond would probably buy it” (Klepper, 2013). The brewery offers full tours of their

operation throughout the week and exclusively uses cans instead of bottles to converse

space and decrease their carbon footprint on the environment. Big Bend, like many other

craft breweries across the country, is working to establish a competitive advantage by

attracting customers to their unique product and passion. Consumers will support

businesses that they believe can make a difference in the world and will stay loyal to that

brand because of what it stands for.

The results of this study highlight a potential reason for the reduction in drunk

driving fatalities from 1994-2014 in the United States. Many consumers choose craft

beer over domestic beer for the brand behind the beer. Imagine craft beer as a type of

, where consumers don’t necessarily drink to become intoxicated, but to taste a

variety of flavors and pair with certain meals. Many craft breweries offer tours of their

27 facilities to educate customers of the brewing process and encourage questions about any part of the production, such as quality control measures. With a greater understanding and appreciation, educated craft beer consumers can perhaps make more informed choices about driving drunk after consuming craft beers.

28 SECTION V

CONCLUSION AND IMPLICATIONS FOR FUTURE RESEARCH

This model expands on current research by examining drunk driving at a nationwide level and accounting for more control variables to isolate the effect of craft beer. Moreover, the results fill a gap in literature, where there is not national data measuring the effect of craft breweries on drunk driving. The model more accurately captures the interaction between the craft beer environment and a DUI fatality, yielding more representative results than previous studies.

Based on the interpretations on this research study, the decrease in DUI fatalities is due to a variety of complex variables. Although the craft beer market boom is likely due to the ability of brand marketing and advertising that especially attracts the millennial generation with new products, there is evidence that it has a significant negative correlation to drunk driving fatality rates in the United States from 1994-2014. Despite significant correlations and findings while performing a regression analysis at a nationwide level, it is challenging to trust the data with the limitations of the study.

Further, although the vast majority of the adult beverage market is dominated by spirits and lower alcoholic domestic beers, drunk driving is still an issue in the United States.

This study implements a strategic landscape to examine possible reasons for the decline in DUI fatalities over time. The policy results of this study can hopefully lead to more analysis attempting to discover reasons for the decrease in DUI deaths. Additional research can create more efficient alcohol programs and policies by drawing reliable conclusions from the data.

29 Although the number of breweries within each state offers some insight into measuring the popularity of craft beer, it is not a perfect metric without accounting for consumption and sales. Further, the challenge in correlating the amount of millennial individuals with breweries per state in this dataset is that the generation was not of legal drinking age until 2002. Therefore, only a percentage of the population would have the opportunity to influence the beer market and consequently the drunk driving market.

Further research studies that investigate the relationship between the craft beer market and drunk driving should examine not only the number of breweries, but the sales and consumption levels of beer produced in each state or region. Another possibility for future research could provide analysis into the characteristics and demographics of craft beer consumers opposed to the limited variables in this study. This could develop a more in depth understanding of the issues causing the craft beer revolution and subsequently a narrower scope of drunk driving. With unlimited funds and resources, the possibilities grow, but it would be interesting to see a study track DUI deaths in a state to determine the root cause of the issue; by understanding the biopsy reports, a study could determine the exact liquids inside the victim during the accident. Similar studies could also adventure into other industries that have seen recent changes in consumer demands due to more localized products provided by smaller firms, such as specialty foods, clothing, artwork, or local sports leagues. Another focus could analyze the social and economic impact of the craft beer market in surrounding regions or neighborhoods; examining the impacts the supplier and producer have on the individuals within the community. A study could determine the number of DUIs or arrests following the opening of a new brewery within a certain mile radius as well as income and employment levels. This

30 approach could be propitious for policy makers adjusting community plans or laws changing an industry’s regulations.

A final thought for future research could construct a model entirely built upon the comparison to the marijuana industry. A study could attempt to discover the consequences on sales growth of beer. Social issues exist within both industries and by investigating the nuances of them together could provide policy makers with advice or confidence in adjusting state and federal jurisdiction, or in the private sector with companies like Uber or Lyft.

31 WORKS CONSULTED

Ashenfelter, O., Hosken, D., & Weinberg, M. (2013). Efficiencies Brewed: Pricing and Consolidation in the U.S. beer industry. NBER Working Paper no. 19353,

Bellware, K. (2014). Craft Beer Industry’s Game Plan: Decrease the Booze to Increase the Fun. Retrieved from http://www.huffingtonpost.com/2014/08/15/craft-beers- alcohol_n_5679911.html

Bray, J. W., Loomis, B. R., & Engelen, M. (2009). You save money when you buy in bulk: Does volume‐based pricing cause people to buy more beer? Health Economics, 18(5), 607-618.

Brewers Association. (2011). Brewers Association: A Passionate Voice for Craft Brewer.

Brewers Association. (2011). Craft Brewing Statistics. Retrieved from: http://www.brewersassociation.org/pages/business-tools/craft-brewing- statistics/facts

Brewers Association. (2014). Craft Brewer Defined. Retrieved from: https://www.brewersassociation.org/statistics/craft-brewer-defined/

Brewers Association. (2014). National Beer Sales and Production Data. Retrieved from: http://www.Brewersassociation.org/statistics/national-Beer-Sales-Production-Data/

Broda, C., & Weinstein, D. E. (2004). Variety growth and world welfare. The American Economic Review, 94(2), 139-144.

Cameron, O. (2013). Reducing dui-related collisions: Evaluating the effectiveness of enhanced dui enforcement, 8-28.

Carpenter, A. J., Darlington, W., Frick, D., Garver, T., Kane, H., Mimken, R., & Serwitz, K. (2013). Craft beer and consumer behavior, 25.

Chen, X. Y. (2013). Boston beer company: The rise of craft beer. Retrieved from: http://marketrealist.com/2013/12/boston-beer-company-introduction-better-beer- business/

Crowell, C. (2013). Craft Beer Consumer Stats: How will they affect your business plan?. Retrieved from: https://www.craftbrewingbusiness.com/business-marketing/craft- beer-consumer-stats-how-will-affect-your-business-plan/

Dole, E. H., Gerstein, D. R., Olson, S., & National Research Council. (1985). Alcohol in 32America: Taking action to prevent abuse. National Academies Press.

32 Drunk driving. Retrieved from: http://www.iii.org/fact-statistic/drunk-driving

Drunk driving fatalities. (2016). Retrieved from: http://responsibility.org/get-the- facts/research/statistics/drunk-driving-fatalities/

Editorial Staff. (2013). The history of drunk driving laws in the U.S. Retrieved from: https://www.lifesafer.com/blog/the-history-of-drunk-driving-laws-in-the-u-s/

Elzinga, K., Tremblay, C. & Tremblay, V. (2015). Craft beer in the united states: History, numbers, and geography. Retrieved from: http://www.wine-economics.org/aawe/wp- content/uploads/2015/12/Vol.10-Issue03-Craft-Beer-in-the-United-States-History- Numbers-and-Geography-by-Kenneth-G.-Elzinga.pdf

Enforcement, S., & Fagin, M. D. (2014). Insurance Institute for Highway Safety.

FADD. (2015). National stats - total fatalities / fatality rates. Retrieved from: http://www.faddintl.org/pr/p5.html

Fromm, J. The millennial consumer craves craft beer. Retrieved from: http://www.millennialmarketing.com/2014/01/the-millennial-consumer-craves-craft- beer/

Harrison, J. (2016). Malt, Hops, and a New Demographic: A Study of the Forces behind the Craft Brew Revolution.

Klein, T. M. (1989). Changes in Alcohol-Involved Fatal Crashes Associated with Tougher State Alcohol Legislation.

Klepper, D. (2015). The Beer from Out Here. Retrieved from: http://www.texashighways.com/food-drink/item/7738-the-beer-from-out-here-big- bend-brewing-company-alpine

McMakin, K. (2016). A Panel Data Analysis of the Craft Brewing Industry and North Carolina Drunk Driving Rates.

Miller, E. (2016). The effectiveness of anti-alcohol PSAs in reducing drunk-driving fatalities: 1995-2010.

MPL Law Firm. (2015). Craft beer and the changing face of DUI. Retrieved from: http://mpl-law.com/craft-beer-dui/

Murray, A. (2012). Factors influencing Brand Loyalty to Craft Breweries in North Carolina.

National Highway Traffic Safety Administration. (2006). NHTSA. Traffic Safety Facts 2000: Pedestrians. (Report DOT HS 809311),

33 O'Malley, P. M., & Wagenaar, A. C. (1991). Effects of minimum drinking age laws on alcohol use, related behaviors and traffic crash involvement among American youth: 1976-1987. Journal of Studies on Alcohol, 52(5), 478-491.

Panetta, F., Panetta, D., & Panetta, F. (2003). Are mergers beneficial to consumers? evidence from the market for bank deposits American Economic Review, 93(4), 1152.

Stewart, K. G., Voas, R. B., & Fell, J. (1995). The Nature of and Reasons for the Decline in Drinking and Driving in the United States: An Update. Proceedings of the 13th International Conference on Alcohol, Drugs and Traffic Safety, 1

Timeline of Cannabis Laws in the United States. (2017). Retrieved from: Timeline of cannabis laws in the United States.

Toro-González, D., McCluskey, J. J., & Mittelhammer, R. C. (2014). Beer snobs do exist: Estimation of beer demand by type. Journal of Agricultural and Resource Economics, 39(2), 174-187.

U.S. Department of Justice. (2016). National sources of law enforcement employment data. Retrieved from: https://www.bjs.gov/content/pub/pdf/nsleed.pdf

U.S. Department of Transportation. (2017). Licensed drivers, vehicle and populations. Retrieved from: https://www.fhwa.dot.gov/ohim/hs00/dlchrt.htm

United States Census Bureau. (2017). Data. Retrieved from: https://www.census.gov/data/tables/2000/dec/phc-t-41.html

34