TY - JOUR AU1 - Brendan, Meany, AU2 - Joshua, Berning, AU3 - Travis, Smith, AB - Abstract Current Blue laws are primarily concerned with limiting the sale of alcohol on Sunday. This presumably decreases adverse outcomes related to alcohol consumption. We examine whether the repeal of alcohol sales on Sunday in Georgia had an impact on teenage drinking, which is linked to a multitude of risky behaviors. We exploit the heterogeneous repeal across counties and municipalities. To account for potential endogeneity, we employ an instrumental variable approach. Across several model specifications, we find no effect of repeal on underage drinking. Concerns that repeal might contribute to increases in underage drinking appear to be unfounded in this case. Blue Laws, Sunday Sales, Underage Drinking Young people in the United States consume more alcohol than tobacco or other illegal drugs. In fact, 33% of high school students report having had at least one drink in the past 30 days, and 18% binge drank (Kann 2016). Underage drinking poses a serious threat to public health, with serious personal, social, and economic consequences. Alcohol consumption can lower inhibitions and affect a person’s ability to assess risks (Yoruk and Yoruk 2015). For example, underage drinking is linked to increased sexual activity in teens (Sen 2002; Miller et al. 2007; Yoruk and Yoruk 2015; Waddell 2012). Grossman and Markowitz (2005) find that alcohol consumption among teens lowers the probability of using birth control and condoms. Carpenter (2005) finds that youth-targeted drunk driving laws reduced the gonorrhea rate among 15–19-year-old white males. Chesson, Harrison, and Kassler (2000) identify a lower sexually -transmitted disease rate in the United States after an increase in alcohol taxes. Miller et al. (2007) link binge drinking, the most common pattern of alcohol consumption among high school youth, to dating violence. Other studies link youth consumption to increased tobacco use, (Bobo and Husten 2000; Grant et al. 2004; Miller et al. 2007; Dee 1999a), and illicit drug use (Miller et al. 2007). Koch and McGeary (2005) find that alcohol initiation at the age of 14 significantly reduces probability of completing high school by between 7% and 22%. Most notably, drinking contributes to the three leading causes of death among teens: unintentional injury, homicide, and suicide (Carpenter 2004; Miller et al. 2007). Annually, about 5,000 young people die from alcohol-related injuries (Office of the Surgeon General 2007). Carpenter and Dobkin (2009) find an increase in mortality rate of 9% following increased alcohol consumption at age 21. One form of regulation aimed at reducing alcohol availability and consumption is the enforcement of Blue laws, particularly the sale of alcohol on Sunday. While there is mixed evidence in the literature regarding the effect of Sunday sales restrictions on alcohol consumption and public health outcomes in general, to our knowledge, no one has examined the effect of Sunday sales bans on teenage alcohol use. Although alcohol is illegal for minors, this does not completely eliminate access for underage drinkers. An additional day of sales during the week creates more opportunities for teenagers to acquire alcohol through a variety of sources. As such, opponents of repealing Sunday sales laws cite underage drinking as a harmful consequence. This article examines the incidence of consumption and ease of access to alcohol for minors in the State of Georgia using the Georgia Student Health Survey from 2008–2014. We rely on the heterogeneous repeal of Sunday sales laws across municipalities in Georgia that occurred over 2011–2014 to examine its effect on teenage drinking. It may be that unobservable, fixed characteristics of municipalities (e.g., preferences for alcohol) are correlated with the decision to repeal. More importantly, the timing of the decision to repeal could be driven by unobserved factors that change during the referendum period (e.g., political pressures from outside the municipality). To control for potential endogeneity of repeal, we estimate a fixed -effects instrumental variables model. Across numerous specifications, our results indicate no effect of changes in Sunday sales laws on underage drinking at the school level. We also examine student -specific responses using the repeated cross-section of surveys. After controlling for potential endogeneity of repeal, we still find no significant effect on underage drinking. We do identify other significant factors, however. Most notably, minority neighborhoods harbor an environment that makes alcohol more accessible to youth. Additionally, various entertainment activities are associated with lower levels of teenage drinking. Our results provide insight into whether Sunday alcohol sales bans do indeed attenuate negative externalities associated with excess drinking, and thus can be used to guide legislatures into formulating laws that benefit society in the best way. More broadly, our results also contribute to the literature examining the efficacy of product sales bans and restrictions. Motivation Blue laws that restrict commerce on Sundays have been a point of contention since their inception. Although their origin is religious, Sunday closing laws have been enacted in connection with the state’s objective to preserve the health, safety, and welfare of its citizens (Dilloff 1979). Following a U.S. Supreme Court decision in 1961, a slew of legal challenges were levied on Blue laws contending they did not satisfy this objective of preservation (Theuman 2005). Over the years, the collection of classifications of commerce permissible on Sundays has exceedingly increased. While few restrictions on Sunday commerce persisted in 2015, laws targeting only the sale of alcohol remained intact in certain states. Currently, 38 states permit Sunday retail sales of alcohol products. Since 2002, 16 states have joined the list of states allowing Sunday sales. While 16 of these 38 states have enacted statewide repeals, 22 states allow a local option—cities and unincorporated counties are permitted to hold citizen referenda to allow retail stores to sell packaged beer, wine, and liquor on Sundays. In April 2011, Georgia legislation passed a bill approved by Governor Deal that would allow a local option in the form of citizen referenda to permit retail stores to sell beer, wine, and liquor on Sundays. The decision to repeal the statewide ban was contentious, however. During the course of political battle, two activist groups, the Georgia Christian Coalition and the Georgia Baptist Commission, which were heavily involved in anti-repeal lobbying, cited increased drunk driving, underage drinking, and domestic violence as their main arguments. At the same time, retail industry began to lobby for holding local referenda and approving such measures. After the first two years of referenda, retailers began to cite pressure from surrounding localities as the main influence to hold referenda. In particular, alcohol retailers complained that customers were traveling to nearby cities that had previously repealed to do their Sunday shopping. The decision to repeal Sunday sales has been contentious. Governor Deal’s predecessor, Governor Purdue, famously vowed to veto any legislation aimed at allowing Sunday sales, specifically citing a paper by Lapham and McMillan (2006), which linked a 42% increase in alcohol-related vehicle accidents with Blue law repeal in New Mexico in 1990. Subsequent studies have cast doubt, however, on the finding of Lapham and McMillan (2006)1. Overall, it is not clear what impact, if any, the repeal of Sunday sales bans has on society. For instance, Yoruk (2014) finds that the legalization of Sunday sales lead to a significant increase in alcohol consumption in 3 of 5 states that changed their laws2. However, Carpenter and Eisenburg (2009) find no increase in overall consumption in Ontario, Canada, following the repeal of Sunday alcohol sale bans. Rather, these authors find a 15% increase in consumption on Sunday, commensurate with a fall in consumption on Saturday. These authors further observe significant cross-border shopping, which accounts for at least 20% of the increase in sales. Stehr (2010) examines the effect of Sunday Sales repeals in fourteen states and finds small or statistically insignificant increases in all but one state, New Mexico. Similar studies show an increase in consumption and violent crime following a loosening of weekend alcohol sale restrictions in Sweden (Grönqvist and Niknami 2014), Brazil (Biderman et al. (2010), and Virginia (Heaton 2012). At the same time, using a more detailed data set across more states, Lee and Yoruk (2014) document a smoothing effect in crime, that is, the increase in crime on Sunday is commensurate with a decrease in crime on the other days. As a result, overall crime throughout the week does not increase significantly. Other authors have suggested potential health benefits from eliminating Blue laws. For example, the extra day of sales might smooth total drinking across more days, thus reducing binge drinking and its deleterious related reckless behavior (Carpenter and Eisenburg 2009). It is also arguable that Sunday sales may curb drunk-driving fatalities if their repeal induces individuals, who would otherwise consume alcohol at a bar, to consume alcohol from the safety of their homes (Lovenheim and Steefel 2011). In addition, Stehr (2007) questions the efficacy of Sunday sales bans, asking: “Why should public policy target traffic fatalities on Sundays and not other days? If the goal is to reduce reckless behavior… a ban on Friday or Saturday sales… would probably be more effective.” Teenage Drinking In summary, previous studies on the effect that Blue laws have on consumption and consumption -related externalities are mixed. More importantly for our research, previous studies have focused on the legal sales of alcohol to adults. It is not immediately clear how lifting Sunday sales bans would affect the teenage consumption of alcohol. A related body of literature shows that a lower legal drinking age leads to higher consumption and heavy episodic drinking (Dee 1999b), an elevated number of vehicle fatalities among the newly -legal cohort (Figlio 1995; Lovenheim and Slemrod 2010), higher nuisance and violent crime rates (Joksch and Jones 1993; Carpenter 2005; Carpenter and Dobkin 2010), and more hospitalizations (Conover and Scrimgeour 2013). In addition, Lee (2013) finds that a statewide repeal on Sunday sales leads to a reduction in years of education of about 0.13 years. To our knowledge, however, there have been no studies that explicitly investigate how the repeal of Sunday sales bans impacts teenage drinking. The Public health literature examining teenage drinking has provided substantial empirical evidence supporting the availability theory (Single 1988), which suggests that the availability of alcohol leads to greater consumption and alcohol-related problems (Dent, Grube, and Biglan 2005; Paschall et al. 2007; Friese, Grube, and Moore 2013), and that restricting access lessens underage drinking (Grube 2010) by increasing the “full price” of alcohol, which includes opportunity costs as well as acquisition costs. Friese, Grube, and Moore (2013) identify four primary ways that underage drinkers illegally obtain alcohol: provision without exchange (i.e., payment); provision with exchange; shoulder-tapping; and theft. The first tends to involve parents, relatives, or close friends. The second can also involve commercial sources. Shoulder-tapping involves soliciting an unknown adult to purchase alcohol. Teen surveys indicate that provision without exchange is the most common source of alcohol for teens, whereas the other types are reported to be rare (Friese, Grube, and Moore 2013). The National Survey on Drug Use and Health (2012) report approximately 8.7% of drinkers aged 12 to 20 purchased their own alcohol, but not necessarily from commercial sources. In the 2006 National Survey on Drug Use and Health (Pemberton et al. 2008), 9.3% of underage drinkers purchased alcohol themselves, but only 7.3% purchased alcohol from a commercial outlet including a store, restaurant, bar, club, or event. Based on these surveys, repealing Sunday sales laws would likely have a negligible impact on underage drinkers directly purchasing alcohol from commercial sources. At the same time, a greater number of sales days increase the overall number of opportunities to purchase alcohol and therefore the ability of underage drinkers to obtain alcohol, ceteris paribus. Although difficult and risky, a greater number of sales opportunities increase the probability of success with direct purchase by underage drinkers. If underage drinkers acquire alcohol from other social connections, Sunday sales provides an additional opportunity for others to purchase alcohol for them. Even the probability of successful theft increases with Sunday sales. Given the negative effects of teenage alcohol consumption, an examination of the connection between Blue laws and teenage drinking is warranted. Data We used local news reports throughout the state to gather information concerning the existence of a referendum in a particular municipality, the rationale behind the holding of the referendum, the date of the referendum, the outcome of the referendum, and the date of repeal, if applicable. When no local news was available, we made in-person phone calls to local officials to collect the above information. The first of the referenda in Georgia were permitted to occur in November 2011. The bulk of the referenda occurred immediately following the repeal of the statewide ban, with 128 counties and city municipalities voting, and 108 passing the referenda (see table 1). The first packaged alcohol Sunday sale happened in Winder, Georgia on November 13th of that year. For some municipalities, however, there was a lag between the referenda and the effective dates. For instance, in 2011, only 81 localities that passed the referendum in 2011 made the change effective that year. The number of referenda decreased each year until there was no city referenda in 2014, and only three county referenda, of which two passed. In total, 278 referenda have taken place out of 668 localities in Georgia, with 243 of them passing the referenda. In figure 1, we display the distribution of referenda before and after 2013. The bulk of the city and county referenda passed near Atlanta and Savannah (i.e., Georgia’s east coast). Fewer county referenda passed in southern Georgia, although many city referenda still passed. Table 1 Timing of Referenda and Effective Dates City County Total Year Referenda Passed Effective Referenda Passed Effective Effective 2011 118 101 77 10 7 4 81 2012 60 58 77 50 45 41 118 2013 32 26 23 5 4 9 32 2014 0 0 9 3 2 3 12 Total 210 185 186 68 58 57 243 City County Total Year Referenda Passed Effective Referenda Passed Effective Effective 2011 118 101 77 10 7 4 81 2012 60 58 77 50 45 41 118 2013 32 26 23 5 4 9 32 2014 0 0 9 3 2 3 12 Total 210 185 186 68 58 57 243 Source: Own tabulation. Table 1 Timing of Referenda and Effective Dates City County Total Year Referenda Passed Effective Referenda Passed Effective Effective 2011 118 101 77 10 7 4 81 2012 60 58 77 50 45 41 118 2013 32 26 23 5 4 9 32 2014 0 0 9 3 2 3 12 Total 210 185 186 68 58 57 243 City County Total Year Referenda Passed Effective Referenda Passed Effective Effective 2011 118 101 77 10 7 4 81 2012 60 58 77 50 45 41 118 2013 32 26 23 5 4 9 32 2014 0 0 9 3 2 3 12 Total 210 185 186 68 58 57 243 Source: Own tabulation. Figure 1 View largeDownload slide Repeal of ban on Sunday sales of alcohol in Georgia by city and county, 2013 Source: Own calculations. Figure 1 View largeDownload slide Repeal of ban on Sunday sales of alcohol in Georgia by city and county, 2013 Source: Own calculations. We obtained data on underage drinking behaviors from 2008–2013 from the Georgia Student Health Survey II, which is administered yearly by the Georgia Department of Education. The survey is an anonymous, self-reported statewide survey instrument developed to identify safety and health issues that have a negative impact on student achievement and school climate. The survey is offered commensurate with the requirements of No Child Left Behind, which specifies that data must be collected for categories including the incidence, prevalence, age of onset, perception of health risks, and perception of social disapproval of drug use and violence. From 2008–2010, 9th and 11th graders did not take the survey. Consequently, we also estimate models that only include 10th and 12th graders. Importantly, the health survey is administered every year from October–February. As such, we match repeals that occur prior to the health survey for each academic year. We utilize two responses from this survey: Past Use and Ease of Access. The first is a response to the statement: “Within the past 30 days, I have used alcohol…”, where the student selects the number of days they have used alcohol. The second variable is a response to the statement “It’s easy to get alcohol,” where the student selects one of four responses: “Strongly Agree (1), Somewhat Agree (2), Somewhat Disagree (3), and Strongly Disagree (4)”. We reverse the actual survey delineations for use in regression to make results more intuitive, that is, a larger ease of access value signifies an easier time accessing alcohol. We collect city-level demographic information from the U.S. Census, including: total population, median income, median age, number of females, number of blacks, and the number of senior citizens. These measures are often correlated with propensity to repeal, drinking attitudes of a location, or both (Desimone 2001; Stehr 2007). We also collect U.S. Census Bureau data on alcohol vendors as classified by the North American Industry Classification System (NAICS), including the number of supermarkets (NAICS 44511), convenience stores (44512), gas stations (44711) and package stores (44531). It is well -documented in the literature that alcohol outlets indeed impact drinking behaviors. For example, several studies suggest increased alcohol outlets present in a location heavily influence the amount of nuisance and violent crime through alcohol consumption (Scribner, MacKinnon, and Dwyer 1995; Scribner et al. 1999; Gruenewald et al. 2006). Presumably, then, areas with a greater availability of alcohol will be more likely to have underage drinkers. We combine all vendors of alcohol into an aggregate variable: Outlets. We also gathered data on various entertainment activities, including the number of amusement and recreational facilities, amusement arcades, amusement parks, marinas, bowling centers, golf courses, fitness centers, and theatres (NAICS 7131 and NAICS 7139). We combine all the data to construct an aggregate variable of all activities: Activities. These facilities provide substitute leisure opportunities for underage drinkers (Astone et al. 2014; Burney 2011). We also compose a measure of religiosity by adding up the number of religious institutions per capita (NAICS 81311). It is commonly cited in the literature that religiousness is inversely related to alcohol use and related problems (Patock-Peckham et al. 1998; Menagi, Harrell, and June 2008; Brechting and Carlson 2015). From our own experience gathering Sunday sales data, and as we will show below, religiosity may also be negatively correlated with the propensity to repeal the sale ban in Georgia. Before 2010, the demographic data was made available once every five years. Similarly, demographic and industry data is not available for the year 2014, and in some instances, 2013. For this reason, we linearly interpolate missing variables for 2013, and linearly extrapolate missing variables for 2014. To conduct our analysis, we aggregate the data to the school -district level for the following several reasons. First, the voting occurred at the city or county level, so it is possible that two contiguous cities have different Sunday sales laws, or that an entire county has repealed the law, but cities within the county have not repealed. As a result, we cannot always classify one entire county as dry on Sunday. Second, in some cases one city feeds into multiple schools or multiple cities feed into a single school, making it infeasible to match a city or county with a particular school. Finally, the student surveys are anonymous, hence students are not indexed in the survey data and we are precluded from panel analysis at the individual student level. To aggregate the repeal data at the school district level, we identify the geographic regions (city or county) that send students to a particular school district. Then we calculate the population -weighted average of the geographic regions that repealed Sunday sales. Therefore, the repeal variable is calculated as the proportion of students within the boundaries of a school district that live in an area that has voted to repeal. Intuitively, if only 10% of a school district’s students lived in a city that repealed Sunday sales, then we would expect a small increase in the opportunity to access alcohol for those students, but not zero. If 90% of the students lived in a city that voted to repeal, we would expect a relatively greater opportunity to access alcohol for those students. Although the vote was binary, the impact on students is continuous between 0 and 1. With the student survey responses, we calculate the average value of responses by school district, grade level (9th through 12th), and year. Our dependent variables, Past Use and Ease of Access, which are originally discrete numbers, become continuous once aggregated. This makes them more amenable to linear regression. The average values per school district are summarized in table 2. Table 2 Summary Statistics for Key Variables Variable Mean Std. Dev. Min Max Past Use 2.36 0.94 0.00 9.89 Log of Past Use 0.79 0.41 −2.08 2.29 Ease of Access 2.21 0.28 1.00 4.00 Past Use* 0.37 0.18 0.00 1.93 Log of Past Use* −1.10 0.48 −3.69 0.66 Ease of Access* 2.28 0.28 1.00 4.00 Repeal 0.11 0.29 0.00 1.00 Total Population 60,887 112,661 786 914,464 Median Income 38,992 10,911 17,226 91,591 Median Age 37.98 3.80 28.30 56.25 Fraction Female 0.51 0.05 0.23 0.79 Fraction Black 0.27 0.18 0.00 0.77 Fraction Seniors 0.13 0.04 0.05 0.29 Fraction Educated 0.11 0.06 0.02 0.46 Outlets 52.18 84.93 1.00 601.00 Religiosity 0.00 0.00 0.00 0.00 Activities 15.74 38.31 0.00 292.00 Tourist Lodges 12.62 28.73 0.00 272.00 Variable Mean Std. Dev. Min Max Past Use 2.36 0.94 0.00 9.89 Log of Past Use 0.79 0.41 −2.08 2.29 Ease of Access 2.21 0.28 1.00 4.00 Past Use* 0.37 0.18 0.00 1.93 Log of Past Use* −1.10 0.48 −3.69 0.66 Ease of Access* 2.28 0.28 1.00 4.00 Repeal 0.11 0.29 0.00 1.00 Total Population 60,887 112,661 786 914,464 Median Income 38,992 10,911 17,226 91,591 Median Age 37.98 3.80 28.30 56.25 Fraction Female 0.51 0.05 0.23 0.79 Fraction Black 0.27 0.18 0.00 0.77 Fraction Seniors 0.13 0.04 0.05 0.29 Fraction Educated 0.11 0.06 0.02 0.46 Outlets 52.18 84.93 1.00 601.00 Religiosity 0.00 0.00 0.00 0.00 Activities 15.74 38.31 0.00 292.00 Tourist Lodges 12.62 28.73 0.00 272.00 Note: * signifies the variable has been rid of outliers Source: Own tabulation and US Census Outlets includes business classified using North American Industry Classification System (NAICS) as supermarkets (NAICS 44511), con-venience stores (44512), gas stations (44711) and package stores (44531) Religiosity includes businesses classified as NAICS 81311 Activities includes business classified as NAICS 7131 and 7139 Tourist Lodges includes business classified as NAICS 72111 Table 2 Summary Statistics for Key Variables Variable Mean Std. Dev. Min Max Past Use 2.36 0.94 0.00 9.89 Log of Past Use 0.79 0.41 −2.08 2.29 Ease of Access 2.21 0.28 1.00 4.00 Past Use* 0.37 0.18 0.00 1.93 Log of Past Use* −1.10 0.48 −3.69 0.66 Ease of Access* 2.28 0.28 1.00 4.00 Repeal 0.11 0.29 0.00 1.00 Total Population 60,887 112,661 786 914,464 Median Income 38,992 10,911 17,226 91,591 Median Age 37.98 3.80 28.30 56.25 Fraction Female 0.51 0.05 0.23 0.79 Fraction Black 0.27 0.18 0.00 0.77 Fraction Seniors 0.13 0.04 0.05 0.29 Fraction Educated 0.11 0.06 0.02 0.46 Outlets 52.18 84.93 1.00 601.00 Religiosity 0.00 0.00 0.00 0.00 Activities 15.74 38.31 0.00 292.00 Tourist Lodges 12.62 28.73 0.00 272.00 Variable Mean Std. Dev. Min Max Past Use 2.36 0.94 0.00 9.89 Log of Past Use 0.79 0.41 −2.08 2.29 Ease of Access 2.21 0.28 1.00 4.00 Past Use* 0.37 0.18 0.00 1.93 Log of Past Use* −1.10 0.48 −3.69 0.66 Ease of Access* 2.28 0.28 1.00 4.00 Repeal 0.11 0.29 0.00 1.00 Total Population 60,887 112,661 786 914,464 Median Income 38,992 10,911 17,226 91,591 Median Age 37.98 3.80 28.30 56.25 Fraction Female 0.51 0.05 0.23 0.79 Fraction Black 0.27 0.18 0.00 0.77 Fraction Seniors 0.13 0.04 0.05 0.29 Fraction Educated 0.11 0.06 0.02 0.46 Outlets 52.18 84.93 1.00 601.00 Religiosity 0.00 0.00 0.00 0.00 Activities 15.74 38.31 0.00 292.00 Tourist Lodges 12.62 28.73 0.00 272.00 Note: * signifies the variable has been rid of outliers Source: Own tabulation and US Census Outlets includes business classified using North American Industry Classification System (NAICS) as supermarkets (NAICS 44511), con-venience stores (44512), gas stations (44711) and package stores (44531) Religiosity includes businesses classified as NAICS 81311 Activities includes business classified as NAICS 7131 and 7139 Tourist Lodges includes business classified as NAICS 72111 We found that certain students responded with potentially excessive responses to the past use question. Looking at table 2, the maximum past use is more than nine days in the past month for a school district. As this is a maximum average for the school districts, certain students report in excess of nine days. There are certain practical and health reasons to think these responses are due to reporting error. First, underage drinkers may have a hard time acquiring sufficient alcohol to drink so frequently. Secondly, students drinking so frequently seem likely to perform poorly in school and face other health issues. We eliminated potential reporting errors based on the Substance Abuse and Mental Health Services Administration’s definition of heavy drinking, which is characterized as five or more drinking occasions in the past 30 days (NIAAA). Specifically, we deleted any student response where the student reported five or more drinks per month. Although this could introduce bias, our rationale is that if the data are accurate, these students are heavy drinkers and likely have steady access to alcohol. It follows then that the relaxing of a Sunday stricture would not influence their drinking patterns. That is, they are unlikely to drink more after the bans are lifted when they are already heavy drinkers. Further, by eliminating these suspected misreported data, our analysis focuses on how the Sunday sales laws might impact those students who currently do not drink so excessively to begin with. The revised data are followed by an asterisk in table 2. Importantly, the differences in means of past use and ease of access in the revised data versus the full dataset are highly significant. This suggests that students defined as heavy drinkers do in fact shift the distribution of responses. To examine how potential misreporting could impact our findings, we also estimate models including all of the observations. All of the census and business data are normalized by school district population. The final panel dataset reports student drinking behaviors for the majority of school districts in Georgia, before and after repeal effective dates. Methods To examine the effect that a repeal of Sunday sales bans has on teenage alcohol use, we start with a linear fixed effects model: Yit=αi+Xit'β+γRit+δt+ɛit (1) where Yit is the log of average Past Use responses for school district i in year t, or Ease of Access responses for school district i in year t; αi are school district fixed effects, Xitis a vector of time -varying control variables, Ritis the fraction of school district i that has experienced repeal in year t, δt is a year fixed effect, and εit is an error term. We exploit the pseudo-natural experiment that has arisen from the heterogeneity in packaged Sunday alcohol sales status across localities in Georgia3. Variation in the timing of the repeal allows us to disentangle policy effects from spurious trends. That is, the rollout in timing allows us to control for unobserved shocks to underage drinking that may be correlated with repeal (Jensen 2007; Lovenheim and Steefel 2011; Hoynes and Schanzenbach 2012; Lee 2013). One concern is that the incidence of repeal of Sunday bans is endogenous to underage drinking behaviors. Specifically, unobservable characteristics in each voting area may be correlated with the referendum and underage drinking behavior. For example, it is plausible that repeals are introduced and passed in cities/counties where attitudes toward drinking are more conducive to facilitating underage drinking. While school district and time -fixed effects are used to account for potential endogeneity, we use several other methods to ensure identification. First, we run an event study -type regression predicting the percentage of a school district that repeals as a function of school district -specific demographic, economic, and industrial variables that may have an important impact on the decision to repeal at the city and county level, using a fixed effect approach. We identify significant explanatory variables in the event study and incorporate those in our main model specification (Besley and Case 2000, Gruber and Hungerman 2008; Hoynes and Schanzenbach 2012; Lee 2013). Second, we employ an instrumental variable specification to attenuate the influence of endogeneity on estimation (Besley and Case 2000). Specifically, we use the number or tourist lodges per capita in a school district as our instrumental variable. These are classified in NAICS as Hotels and Motels (NAICS code 72111) and provided by the U.S. Census Bureau. We specify the first -stage instrumental variable model as Rit=αi+Xit'θ+ηZit+δt+μit (2) where the instrumental variable Zit is the number of tourist lodges in school district i, normalized by population of school district i, in year t , and μit is an error term. The number of tourist lodging establishments measures the extent of tourism and visitors to a given area. Since alcohol demand is positively correlated to tourism activity, tourist lodges will want to encourage greater tourism visits by making alcohol more widely available. Hence, localities with the potential for greater tourist revenues via alcohol sales will be more inclined to vote for Sunday sales. As such, tourist lodging is a positive predictor of the decision to repeal Sunday Sales laws (Beasley 2011). Our next concern is excludability. Tourist lodging can (but does not necessarily) provide food or beverages. If they do provide alcoholic beverages, they are for on-premise and in-lodging consumption, that is, for hotel visitors, which makes them unlikely sources for providing underage drinkers with alcohol. Establishments located at tourist lodging that serve alcohol (i.e., bars and cocktail lounges) are considered separate establishments according to NAICS. An increase in tourist lodging does not necessarily result in an increase in establishments serving alcohol. In this sense, tourist lodging is excludable from the regression on teenage alcohol usage or ease of access. Finally, we consider whether our instrumental variable is uncorrelated with unobservable variables in our primary equation (1) such that covZ,ɛ=0 even after controlling for school district and year fixed effects and observed time -varying control variables. If there are any unobserved time-varying characteristics in the school district environment that are correlated with both teenage alcohol use and the presence of tourist lodging, then covZ,ɛ≠0. As we are controlling for a large number of demographic and market characteristics that are changing over time (table 2), we expect that we are capturing any time-varying unobservable variation in our estimation procedure. Further, it is infeasible that there is reverse causality, where underage drinking would facilitate more tourism. Finally, to examine the impact of the Sunday sales repeal on teenage consumption and access, we estimate the second -stage model Yit=αi+Zit'β+γRit̂+δt+ɛit (3) where Rit̂ is the predicted value of Rit from equation (2). Results We first estimate a linear fixed effects model to identify specific school district characteristics that predict the percentage of the school district that has experienced repeal (table 3). Across all models, the percentage of the population that is black has a positive effect on repealing the sales ban. While this is not a causal effect, the correlation does bring into question why the relationship is significant. Interestingly, education, which measures the percentage of the population with a bachelor’s degree, has a positive effect on repeal. This could indicate that people with more education are more likely to view Blue laws as outdated or unnecessary. Both median income and total population, which are related to total demand, are also positively correlated with repeal, as would be expected. The level of religiosity is consistently negative, affirming previous social motivations for promoting Blue laws and Sunday sales bans. Finally, the number of tourist lodges is positively related to repeal, suggesting that the measure performs well as an instrumental variable. In particular, this indicates that school districts with more of an emphasis on tourism are more likely to repeal the sales bans. Table 3. Estimates of the Repeal of Sunday Sales Bans Dependent Variable: Fraction SD Repeal Female −5.298*** −0.488 −0.329 −0.332 Seniors 10.73*** 0.8 1.085 0.943 Black 4.724*** 1.965** 2.056** 1.948** Age 0.0693*** −0.0181 −0.018 −0.015 Educated – 3.793** 3.952** 4.215** Income – 4.581*** 4.466*** 4.529*** Population – 0.940*** 0.912*** 0.879*** Religiosity – −1.575** −1.488** −1.648** Outlets – – −0.177 −0.233 Tourist Lodges – – – 0.469*** Year Fixed Effects – ✓ ✓ ✓ School District Fixed Effects ✓ ✓ ✓ ✓ Observations 1014 1014 1014 1014 R-Squared 0.142 0.482 0.482 0.487 Dependent Variable: Fraction SD Repeal Female −5.298*** −0.488 −0.329 −0.332 Seniors 10.73*** 0.8 1.085 0.943 Black 4.724*** 1.965** 2.056** 1.948** Age 0.0693*** −0.0181 −0.018 −0.015 Educated – 3.793** 3.952** 4.215** Income – 4.581*** 4.466*** 4.529*** Population – 0.940*** 0.912*** 0.879*** Religiosity – −1.575** −1.488** −1.648** Outlets – – −0.177 −0.233 Tourist Lodges – – – 0.469*** Year Fixed Effects – ✓ ✓ ✓ School District Fixed Effects ✓ ✓ ✓ ✓ Observations 1014 1014 1014 1014 R-Squared 0.142 0.482 0.482 0.487 Calculated using robust standard errors *** p < 0.01, **p < 0.05, *p < 0.1 Table 3. Estimates of the Repeal of Sunday Sales Bans Dependent Variable: Fraction SD Repeal Female −5.298*** −0.488 −0.329 −0.332 Seniors 10.73*** 0.8 1.085 0.943 Black 4.724*** 1.965** 2.056** 1.948** Age 0.0693*** −0.0181 −0.018 −0.015 Educated – 3.793** 3.952** 4.215** Income – 4.581*** 4.466*** 4.529*** Population – 0.940*** 0.912*** 0.879*** Religiosity – −1.575** −1.488** −1.648** Outlets – – −0.177 −0.233 Tourist Lodges – – – 0.469*** Year Fixed Effects – ✓ ✓ ✓ School District Fixed Effects ✓ ✓ ✓ ✓ Observations 1014 1014 1014 1014 R-Squared 0.142 0.482 0.482 0.487 Dependent Variable: Fraction SD Repeal Female −5.298*** −0.488 −0.329 −0.332 Seniors 10.73*** 0.8 1.085 0.943 Black 4.724*** 1.965** 2.056** 1.948** Age 0.0693*** −0.0181 −0.018 −0.015 Educated – 3.793** 3.952** 4.215** Income – 4.581*** 4.466*** 4.529*** Population – 0.940*** 0.912*** 0.879*** Religiosity – −1.575** −1.488** −1.648** Outlets – – −0.177 −0.233 Tourist Lodges – – – 0.469*** Year Fixed Effects – ✓ ✓ ✓ School District Fixed Effects ✓ ✓ ✓ ✓ Observations 1014 1014 1014 1014 R-Squared 0.142 0.482 0.482 0.487 Calculated using robust standard errors *** p < 0.01, **p < 0.05, *p < 0.1 Past Use We estimate equations (2) and (3) with log of Past Use as the dependent variable, and responses from all grades (table 4, columns 1–2) and just the 10th and 12th grade only using the Generalized Method of Moments (GMM) estimator (table 4, columns 3–4). The standard errors are clustered by school district. The results of the first stage (columns 1 and 3) provide results similar to the fixed effects models showing that tourist lodging is significantly related to the fraction of school districts that repeal Sunday sales laws. The Angrist-Pischke F-test suggests that our instrumental variable is not under -identified or weakly identified in either model. Table 4 GMM Regression Coefficients for Log of Past Use All Grades 10th and 12th grade First Stage Second Stage First Stage Second Stage Female −0.054 0.341 −0.109 1.380 (0.886) (2.454) (0.897) (2.135) Black 1.961* 2.145 2.065* 2.279 (1.132) (2.721) (1.161) (2.609) Educated 5.097* 0.299 5.124* 1.290 (2.908) (6.315) (2.937) (6.231) Age −0.010 0.009 −0.010 −0.003 (0.017) (0.030) (0.018) (0.029) Income 4.333*** −1.807 4.328*** −1.484 (1.136) (4.464) (1.139) (4.367) Population 0.861*** 0.468 0.846*** 0.644 (0.289) (0.811) (0.290) (0.832) Outlets −0.141 0.475 −0.158 0.552 (0.342) (0.636) (0.346) (0.529) Religiosity −1.579 −1.197 −1.678 −1.365 (1.143) (2.779) (1.155) (2.984) Activities −1.426 −3.240 −1.501 −5.467** (1.711) (2.454) (1.765) (2.353) Tourist Lodges 0.546** − 0.550** − (0.252) (0.252) Repeal − −0.377 − −0.464 (0.883) (0.867) N 923 923 917 917 R-Squared 0.499 0.113 0.501 0.012 AP F-test 7.2*** 7.26*** All Grades 10th and 12th grade First Stage Second Stage First Stage Second Stage Female −0.054 0.341 −0.109 1.380 (0.886) (2.454) (0.897) (2.135) Black 1.961* 2.145 2.065* 2.279 (1.132) (2.721) (1.161) (2.609) Educated 5.097* 0.299 5.124* 1.290 (2.908) (6.315) (2.937) (6.231) Age −0.010 0.009 −0.010 −0.003 (0.017) (0.030) (0.018) (0.029) Income 4.333*** −1.807 4.328*** −1.484 (1.136) (4.464) (1.139) (4.367) Population 0.861*** 0.468 0.846*** 0.644 (0.289) (0.811) (0.290) (0.832) Outlets −0.141 0.475 −0.158 0.552 (0.342) (0.636) (0.346) (0.529) Religiosity −1.579 −1.197 −1.678 −1.365 (1.143) (2.779) (1.155) (2.984) Activities −1.426 −3.240 −1.501 −5.467** (1.711) (2.454) (1.765) (2.353) Tourist Lodges 0.546** − 0.550** − (0.252) (0.252) Repeal − −0.377 − −0.464 (0.883) (0.867) N 923 923 917 917 R-Squared 0.499 0.113 0.501 0.012 AP F-test 7.2*** 7.26*** Standard errors in parantheses are clustered by school district *** p < 0.01, **p < 0.05, *p < 0.1 AP: Angrist -Pischke F-test Table 4 GMM Regression Coefficients for Log of Past Use All Grades 10th and 12th grade First Stage Second Stage First Stage Second Stage Female −0.054 0.341 −0.109 1.380 (0.886) (2.454) (0.897) (2.135) Black 1.961* 2.145 2.065* 2.279 (1.132) (2.721) (1.161) (2.609) Educated 5.097* 0.299 5.124* 1.290 (2.908) (6.315) (2.937) (6.231) Age −0.010 0.009 −0.010 −0.003 (0.017) (0.030) (0.018) (0.029) Income 4.333*** −1.807 4.328*** −1.484 (1.136) (4.464) (1.139) (4.367) Population 0.861*** 0.468 0.846*** 0.644 (0.289) (0.811) (0.290) (0.832) Outlets −0.141 0.475 −0.158 0.552 (0.342) (0.636) (0.346) (0.529) Religiosity −1.579 −1.197 −1.678 −1.365 (1.143) (2.779) (1.155) (2.984) Activities −1.426 −3.240 −1.501 −5.467** (1.711) (2.454) (1.765) (2.353) Tourist Lodges 0.546** − 0.550** − (0.252) (0.252) Repeal − −0.377 − −0.464 (0.883) (0.867) N 923 923 917 917 R-Squared 0.499 0.113 0.501 0.012 AP F-test 7.2*** 7.26*** All Grades 10th and 12th grade First Stage Second Stage First Stage Second Stage Female −0.054 0.341 −0.109 1.380 (0.886) (2.454) (0.897) (2.135) Black 1.961* 2.145 2.065* 2.279 (1.132) (2.721) (1.161) (2.609) Educated 5.097* 0.299 5.124* 1.290 (2.908) (6.315) (2.937) (6.231) Age −0.010 0.009 −0.010 −0.003 (0.017) (0.030) (0.018) (0.029) Income 4.333*** −1.807 4.328*** −1.484 (1.136) (4.464) (1.139) (4.367) Population 0.861*** 0.468 0.846*** 0.644 (0.289) (0.811) (0.290) (0.832) Outlets −0.141 0.475 −0.158 0.552 (0.342) (0.636) (0.346) (0.529) Religiosity −1.579 −1.197 −1.678 −1.365 (1.143) (2.779) (1.155) (2.984) Activities −1.426 −3.240 −1.501 −5.467** (1.711) (2.454) (1.765) (2.353) Tourist Lodges 0.546** − 0.550** − (0.252) (0.252) Repeal − −0.377 − −0.464 (0.883) (0.867) N 923 923 917 917 R-Squared 0.499 0.113 0.501 0.012 AP F-test 7.2*** 7.26*** Standard errors in parantheses are clustered by school district *** p < 0.01, **p < 0.05, *p < 0.1 AP: Angrist -Pischke F-test In the second stage of the model with all grades included (column 2), none of the explanatory variables are significant. Further, the results indicate that as communities in the school district repeal their Sunday sales bans on alcohol sales, past use does not change, even at a much more relaxed significance level. This suggests that increased access to alcohol does not spill over to affect teenage alcohol use. Looking at the results of the model including only 10th and 12th grade, we find that as activities per capita increase in a school district, Past Use decreases, indicating substitution of leisure, as might be expected. If there are more activities in which to partake on the weekends, for example, movies or sporting events, there is a higher opportunity cost for youths to drink. Again, the repeal of Sunday sales laws does not have a statistically significant effect at any meaningful level. We estimate additional models where the potential outliers due to misreporting are included and where we do not take the log of the dependent variable. All of the primary findings are similar. Finally, we estimate a simple model with only the Repeal variable included, and find it is not significant4. Ease of Access We next estimate equations (2) and (3) with Ease of Access as the dependent variable with all grades (table 5, columns 1–2) and just the 10th and 12th grade only using a GMM estimator (table 5, columns 3–4). Again, the results here are for the dataset that excludes outliers and the standard errors are clustered by school district. The Angrist-Pischke F-test suggests the instrumental variable is not under -identified or weakly identified. Table 5 GMM Regression Coefficients for Ease of Access All Grades 10th and 12th grade First Stage Second Stage First Stage Second Stage Female −0.003 −2.626*** −0.140 −1.964** (0.887) (0.993) (0.892) (0.922) Black 1.842 2.882** 2.074* 2.599** (1.125) (1.320) (1.153) (1.292) Educated 4.786* 2.367 5.046* 1.257 (2.821) (3.451) (2.908) (3.521) Age −0.008 0.0353* −0.009 0.022 (0.017) (0.019) (0.017) (0.021) Income 4.405*** 0.066 4.303*** 0.915 (1.130) (2.118) (1.136) (2.246) Population 0.840*** 0.275 0.853*** 0.315 (0.283) (0.372) (0.291) (0.433) Outlets −0.107 0.700* −0.138 0.516 (0.347) (0.419) (0.350) (0.491) Religiosity −1.587 −0.069 −1.610 0.387 (1.128) (1.179) (1.139) (1.372) Activities −1.441 −1.868 −1.549 −1.971 (1.721) (1.220) (1.766) (1.536) Tourist Lodges 0.544** − 0.548** − (0.251) (0.251) Repeal − −0.267 − −0.393 (0.422) (0.464) N 934 934 929 929 R-Squared 0.496 0.282 0.499 0.082 F-test 7.18*** 7.20*** All Grades 10th and 12th grade First Stage Second Stage First Stage Second Stage Female −0.003 −2.626*** −0.140 −1.964** (0.887) (0.993) (0.892) (0.922) Black 1.842 2.882** 2.074* 2.599** (1.125) (1.320) (1.153) (1.292) Educated 4.786* 2.367 5.046* 1.257 (2.821) (3.451) (2.908) (3.521) Age −0.008 0.0353* −0.009 0.022 (0.017) (0.019) (0.017) (0.021) Income 4.405*** 0.066 4.303*** 0.915 (1.130) (2.118) (1.136) (2.246) Population 0.840*** 0.275 0.853*** 0.315 (0.283) (0.372) (0.291) (0.433) Outlets −0.107 0.700* −0.138 0.516 (0.347) (0.419) (0.350) (0.491) Religiosity −1.587 −0.069 −1.610 0.387 (1.128) (1.179) (1.139) (1.372) Activities −1.441 −1.868 −1.549 −1.971 (1.721) (1.220) (1.766) (1.536) Tourist Lodges 0.544** − 0.548** − (0.251) (0.251) Repeal − −0.267 − −0.393 (0.422) (0.464) N 934 934 929 929 R-Squared 0.496 0.282 0.499 0.082 F-test 7.18*** 7.20*** Standard errors in parantheses are clustered by school district *** p < 0.01, **p < 0.05, *p < 0.1 AP: Angrist -Pischke F-test Table 5 GMM Regression Coefficients for Ease of Access All Grades 10th and 12th grade First Stage Second Stage First Stage Second Stage Female −0.003 −2.626*** −0.140 −1.964** (0.887) (0.993) (0.892) (0.922) Black 1.842 2.882** 2.074* 2.599** (1.125) (1.320) (1.153) (1.292) Educated 4.786* 2.367 5.046* 1.257 (2.821) (3.451) (2.908) (3.521) Age −0.008 0.0353* −0.009 0.022 (0.017) (0.019) (0.017) (0.021) Income 4.405*** 0.066 4.303*** 0.915 (1.130) (2.118) (1.136) (2.246) Population 0.840*** 0.275 0.853*** 0.315 (0.283) (0.372) (0.291) (0.433) Outlets −0.107 0.700* −0.138 0.516 (0.347) (0.419) (0.350) (0.491) Religiosity −1.587 −0.069 −1.610 0.387 (1.128) (1.179) (1.139) (1.372) Activities −1.441 −1.868 −1.549 −1.971 (1.721) (1.220) (1.766) (1.536) Tourist Lodges 0.544** − 0.548** − (0.251) (0.251) Repeal − −0.267 − −0.393 (0.422) (0.464) N 934 934 929 929 R-Squared 0.496 0.282 0.499 0.082 F-test 7.18*** 7.20*** All Grades 10th and 12th grade First Stage Second Stage First Stage Second Stage Female −0.003 −2.626*** −0.140 −1.964** (0.887) (0.993) (0.892) (0.922) Black 1.842 2.882** 2.074* 2.599** (1.125) (1.320) (1.153) (1.292) Educated 4.786* 2.367 5.046* 1.257 (2.821) (3.451) (2.908) (3.521) Age −0.008 0.0353* −0.009 0.022 (0.017) (0.019) (0.017) (0.021) Income 4.405*** 0.066 4.303*** 0.915 (1.130) (2.118) (1.136) (2.246) Population 0.840*** 0.275 0.853*** 0.315 (0.283) (0.372) (0.291) (0.433) Outlets −0.107 0.700* −0.138 0.516 (0.347) (0.419) (0.350) (0.491) Religiosity −1.587 −0.069 −1.610 0.387 (1.128) (1.179) (1.139) (1.372) Activities −1.441 −1.868 −1.549 −1.971 (1.721) (1.220) (1.766) (1.536) Tourist Lodges 0.544** − 0.548** − (0.251) (0.251) Repeal − −0.267 − −0.393 (0.422) (0.464) N 934 934 929 929 R-Squared 0.496 0.282 0.499 0.082 F-test 7.18*** 7.20*** Standard errors in parantheses are clustered by school district *** p < 0.01, **p < 0.05, *p < 0.1 AP: Angrist -Pischke F-test In the second stage, more covariates are statistically significant. As the fraction of females making up the population of a school district decreases, so does the ease of access. This is consistent with previous findings showing that women use alcohol less frequently and encounter fewer problems due to their drinking than men (Cooper et al. 1992; Nolen-Hoeksema 2004). Alternatively, an increase in the fraction of blacks in the population of the school district increases the ease of access. Clearly this is not causal, but indicates that minors who reside in black neighborhoods find it easier to acquire alcohol. For the model estimated using all grades (column 2), an increase in median age of the school district is associated with an increase in the ease of access, suggesting that overall older populations provide an easier outlet for underage access. Finally, the repeal of Sunday sales laws has no effect on ease of access across all school districts. Again, we estimate models where the potential outliers due to misreporting are included and all of the primary findings are similar5. Robustness: Extensive and Intensive Margin It is plausible that students reporting zero drinks in the previous 30 days were dampening the effect of repeal on those that may be more experienced drinkers since the data is averaged at the school -district level. Put simply, the repeal might have no effect on people who do not already drink, but still could have had an effect on those that do drink. We consider the effect of the ban on the intensive margin by including in the sample only those who reported any drinking at all in the previous 30 days. Again, across multiple model specifications, we find the effect of the repeal is not significantly different from zero. Another possibility is that, although the number of drinks taken in the overall sample did not increase, the number of individuals who had any drinks at all in the previous 30 days increased as a result of repeal. For this reason, we consider the extensive margin by calculating the percentage of individuals who report having taken any drink at all in the previous 30 days. Again, the effect of repeal is not significantly different from zero for any of the model specifications. We exclude these results here, but they are available upon request. Finally, if the decision to repeal is not endogenous in equation (1), then instrumental variable estimation will be inefficient. We therefore estimate equation (1) using just fixed effects estimation to see if the effect of repeal has a significant, albeit biased, effect on past use or ease of access. In all model specifications including different covariates, the effect of repeal is not statistically significant. Robustness: Student Specific Responses Since our data are repeated cross-sections and we do not identify student characteristics, our primary approach was to aggregate the data to the school district level. Given our null findings, we further explore the student -specific responses. We start by estimating a count model of Past Use or Ease of Access using the following specification: Yijt=Xit'ρ+τRit+α+δt+ωijt (4) where j indexes students in school district i at time t, α is a vector of dummy variables for each school district and ωjt is an error term. We estimate equation (4) using a negative binomial specification. In addition, we estimate alternative specifications to control for potential endogeneity of Sunday sales in equation (4). Given that the primary equation is a count model, we follow Wooldridge (2010) and estimate a control function model. First, we estimate the fraction of repeal as a function of our instrumental variable using ordinary least squares, Rit=ϕZit+α+δt+υit (5) then we take the predicted residuals from equation (5) given as υ^it, insert them into equation (4) and estimate Yijt=Xit'ρ+τRit+α+δt+υ^it+ωijt. (6) According to Wooldridge, fewer assumptions are required to estimate equation (6) using a Poisson specification rather than a negative binomial and there are no concerns regarding the overdispersion parameter. As such, we only estimate equation (6) using a Poisson specification. With the large number of student observations, we only estimate the 10th and 12th grade respondents as two separate groups using several specifications (table 6). We estimate models as follows: excluding and including our covariates (Outlets, Religiosity and Activities); including and excluding non-drinkers (Intensive) to evaluate the intensive margin of students that are identified as drinkers; estimating models with and without the previously described control function methodology. Finally, in all of our specifications, we include the heavy drinkers that we excluded from our previous models. Table 6 Analysis of Past Use and Ease of Access with Individual Student Sample 10th grade Dependent variable: Past Use Dependent Variable: Ease of Access Repeal −0.155*** −0.139** −0.0389 −0.0397 0.911 0.958 0.472 0.647 0.0205* 0.0218* 0.642 0.447 (0.0440) (0.0436) (0.0256) (0.0258) (0.8070) (0.8850) (0.5590) (0.5990) (0.0096) (0.0104) (0.4650) (0.4140) Model NB NB NB NB P P P P NB NB P P Covariates ✓ ✓ ✓ ✓ ✓ ✓ Intensive ✓ ✓ ✓ ✓ Control function ✓ ✓ ✓ ✓ ✓ ✓ Observations 322,463 322,463 68,562 68,562 322,463 322,463 68,562 68,562 248,967 248,967 248,967 248,967 10th grade Dependent variable: Past Use Dependent Variable: Ease of Access Repeal −0.155*** −0.139** −0.0389 −0.0397 0.911 0.958 0.472 0.647 0.0205* 0.0218* 0.642 0.447 (0.0440) (0.0436) (0.0256) (0.0258) (0.8070) (0.8850) (0.5590) (0.5990) (0.0096) (0.0104) (0.4650) (0.4140) Model NB NB NB NB P P P P NB NB P P Covariates ✓ ✓ ✓ ✓ ✓ ✓ Intensive ✓ ✓ ✓ ✓ Control function ✓ ✓ ✓ ✓ ✓ ✓ Observations 322,463 322,463 68,562 68,562 322,463 322,463 68,562 68,562 248,967 248,967 248,967 248,967 12th grade Dependent variable: Past Use Dependent Variable: Ease of Access Repeal 0.016 0.0241 0.0424 0.0467 0.231 −0.0352 −0.104 −0.363 0.00895 0.00862 0.0495 −0.0324 (0.0549) (0.0553) (0.0261) (0.0268) (0.9350) (0.9020) (0.4910) (0.5000) (0.0119) (0.0121) −0.268 −0.27 Model NB NB NB NB P P P P NB NB P P Covariates ✓ ✓ ✓ ✓ ✓ ✓ Intensive ✓ ✓ ✓ ✓ Control function ✓ ✓ ✓ ✓ ✓ ✓ Observations 261,428 261,428 76,327 76,327 261,428 261,428 76,327 76,327 200,407 200,407 200,407 200,407 12th grade Dependent variable: Past Use Dependent Variable: Ease of Access Repeal 0.016 0.0241 0.0424 0.0467 0.231 −0.0352 −0.104 −0.363 0.00895 0.00862 0.0495 −0.0324 (0.0549) (0.0553) (0.0261) (0.0268) (0.9350) (0.9020) (0.4910) (0.5000) (0.0119) (0.0121) −0.268 −0.27 Model NB NB NB NB P P P P NB NB P P Covariates ✓ ✓ ✓ ✓ ✓ ✓ Intensive ✓ ✓ ✓ ✓ Control function ✓ ✓ ✓ ✓ ✓ ✓ Observations 261,428 261,428 76,327 76,327 261,428 261,428 76,327 76,327 200,407 200,407 200,407 200,407 Standard errors in parantheses are clustered by school district; *p < 0.05, **p < 0.01, ***p < 0.001 Model: NB = Negative Binomial; P = Poisson Covariates includes: Outlets, Religiosity, and Activities Intensive excludes students with 0 drinks in past 30 days Control function indicates control function model Table 6 Analysis of Past Use and Ease of Access with Individual Student Sample 10th grade Dependent variable: Past Use Dependent Variable: Ease of Access Repeal −0.155*** −0.139** −0.0389 −0.0397 0.911 0.958 0.472 0.647 0.0205* 0.0218* 0.642 0.447 (0.0440) (0.0436) (0.0256) (0.0258) (0.8070) (0.8850) (0.5590) (0.5990) (0.0096) (0.0104) (0.4650) (0.4140) Model NB NB NB NB P P P P NB NB P P Covariates ✓ ✓ ✓ ✓ ✓ ✓ Intensive ✓ ✓ ✓ ✓ Control function ✓ ✓ ✓ ✓ ✓ ✓ Observations 322,463 322,463 68,562 68,562 322,463 322,463 68,562 68,562 248,967 248,967 248,967 248,967 10th grade Dependent variable: Past Use Dependent Variable: Ease of Access Repeal −0.155*** −0.139** −0.0389 −0.0397 0.911 0.958 0.472 0.647 0.0205* 0.0218* 0.642 0.447 (0.0440) (0.0436) (0.0256) (0.0258) (0.8070) (0.8850) (0.5590) (0.5990) (0.0096) (0.0104) (0.4650) (0.4140) Model NB NB NB NB P P P P NB NB P P Covariates ✓ ✓ ✓ ✓ ✓ ✓ Intensive ✓ ✓ ✓ ✓ Control function ✓ ✓ ✓ ✓ ✓ ✓ Observations 322,463 322,463 68,562 68,562 322,463 322,463 68,562 68,562 248,967 248,967 248,967 248,967 12th grade Dependent variable: Past Use Dependent Variable: Ease of Access Repeal 0.016 0.0241 0.0424 0.0467 0.231 −0.0352 −0.104 −0.363 0.00895 0.00862 0.0495 −0.0324 (0.0549) (0.0553) (0.0261) (0.0268) (0.9350) (0.9020) (0.4910) (0.5000) (0.0119) (0.0121) −0.268 −0.27 Model NB NB NB NB P P P P NB NB P P Covariates ✓ ✓ ✓ ✓ ✓ ✓ Intensive ✓ ✓ ✓ ✓ Control function ✓ ✓ ✓ ✓ ✓ ✓ Observations 261,428 261,428 76,327 76,327 261,428 261,428 76,327 76,327 200,407 200,407 200,407 200,407 12th grade Dependent variable: Past Use Dependent Variable: Ease of Access Repeal 0.016 0.0241 0.0424 0.0467 0.231 −0.0352 −0.104 −0.363 0.00895 0.00862 0.0495 −0.0324 (0.0549) (0.0553) (0.0261) (0.0268) (0.9350) (0.9020) (0.4910) (0.5000) (0.0119) (0.0121) −0.268 −0.27 Model NB NB NB NB P P P P NB NB P P Covariates ✓ ✓ ✓ ✓ ✓ ✓ Intensive ✓ ✓ ✓ ✓ Control function ✓ ✓ ✓ ✓ ✓ ✓ Observations 261,428 261,428 76,327 76,327 261,428 261,428 76,327 76,327 200,407 200,407 200,407 200,407 Standard errors in parantheses are clustered by school district; *p < 0.05, **p < 0.01, ***p < 0.001 Model: NB = Negative Binomial; P = Poisson Covariates includes: Outlets, Religiosity, and Activities Intensive excludes students with 0 drinks in past 30 days Control function indicates control function model With the 10th grade sample, we find that Repeal is negatively correlated with Past Use for two of our model specifications. This is counter to claims that repealing Sunday sales laws will lead to greater drinking in teenagers. In our other specifications, however, the effect of Repeal is insignificant. Controlling for the endogeneity of repeal in these models could explain the insignificant finding if communities with a lower prevalence of teenage drinking are more likely to repeal Sunday sales bans. However, this effect is not persistent. The Ease of Access increases for 10th graders following repeal in two of the model specifications. This effect is economically insignificant, however, as the increase in access is only 0.02. Further, the effect is insignificant in the control function models, suggesting that controlling for endogeneity is important in these models as well. Across all the specifications for the 12th grade sample, we find that the decision to repeal Sunday sales laws is not significantly correlated with either Past Use or Ease of Access. Overall, it appears that our null finding is also consistent with student -level data. Discussion Using data on underage drinking patterns and perceptions combined with heterogeneous repeal of Sunday alcohol sales bans, we analyze whether alcohol-related Blue laws provide a secular public health benefit in ameliorating teen drinking behaviors within the state of Georgia. We account for the endogeneity of repeals using fixed effects and including variables found to be correlated with propensity to repeal. Because the timing of repeal at the city and county levels is staggered, the percentage of a school district that has experienced repeal changes yearly. This allows us to control for shocks occurring uniformly across school districts that may affect student drinking behaviors. Further, we expand upon the literature by employing an instrumental variable specification. Across several specifications, our results indicate that there is no evidence that the repeal of Sunday alcohol sales bans increases underage alcohol use, nor do they effect the perception of access to alcohol for teens. These findings are consistent with previous findings suggesting that consumption may be smoothed over the week following the removal of a sales ban. In this case, any illegal drinking by teenagers does not increase. If the removal of Sunday sales bans allows for more opportunities to drink, then teenage consumption may be smoothed over the week as well. The results of this research have policy implications for the remaining twelve states that defend the efficacy of Blue laws targeting alcohol. The rationale of imposing a Sunday sales ban is becoming difficult to defend on economic grounds. We further demonstrate that the repeal of such laws does not increase teens’ self-reported drinking behaviors, either. Nevertheless, it is not certain if our findings would translate to other states. Our findings contribute to the literature on the efficacy of Sunday prohibition laws by showing that these laws do not achieve their objective, but rather serve as an inconvenience to consumers, and also result in the forfeiture of potential tax revenue from lost sales. Importantly, there are also potential benefits from repealing Sunday sales laws, including increased tax revenues and consumer welfare due to reduced dead weight loss (Lovenheim and Steefel 2011) and increased employment (Gradus 1996). Furthermore, there is a rift in logic in banning sales for only one day. If a benefit of Sunday bans were to be identified, it may be the case that a ban pertaining to more or all days of the week is optimal. In this light, a Sunday ban is rather arbitrary. Our results do show other significant determinants of underage alcohol use and perceptions that are of importance to policy makers. In particular, it appears that black neighborhoods harbor an environment that provides easier access to alcohol for youths. Racial inequality is and has been a pressing issue facing society. Further, such inequality negatively affects economic opportunity, crime, and education, as well as underage alcohol access. The fact that teens living in black communities are more able to procure alcohol presents itself as another symptom of inequality. Of course, policies aimed at any one of these issues are likely palliative. Inequality needs to be attacked at its core; this subject, however, is beyond the scope of this article. The effect of activities on teen alcohol consumption is corroborated with research related to positive effects of youths’ engagement in the community. There are a broad range of activities, such as sports, art, music, and politics that provide positive outcomes for youth, one of which is decrease in the rate of substance abuse (Burney 2011; Astone et al. 2014). Immersing youth into their communities offers an unconventional method of reducing underage drinking. Policy makers may attempt to stem the underage consumption of alcohol by encouraging after -school activities, attracting performance venues, and expanding parks and outdoor spaces. The authors would like to thank the Georgia Department of Education for providing the student survey data. The authors received valuable comments from Greg Colson, attendees at the Beeronomics Conference in Seattle, WA, three anonymous reviewers, and the editor, Roderick Rejesus. All errors are the authors’ responsibility. Footnotes 1 Lovenheim and Steefel (2011) deem the methods of Lapham and McMillan (2006) problematic: the study entails only one state and only one policy change, making it very difficult to disentangle the effects of policy from trends or secular shocks. They expand the past work on vehicle fatalities by including fixed effects in their model, which allows them to control for omitted variables that are correlated with repeal and vehicle accidents. They find no effect of repeal on vehicle fatalities. Stehr (2010) looks at fourteen states and only finds an effect on driving fatalities in New Mexico. Further, he finds similar increases in non-alcohol related fatalities as well, casting doubt on causality. 2 Yoruk also cites mixed findings in the literature on the effect of Sunday sales on fatalities. 3 While there is some cross city/county variation in on-premise alcohol sale laws, we ignore this variation because underage drinkers typically do not seek to acquire alcohol from such establishments. 4 The results of these additional models are available upon request. 5 The results of these additional models are available upon request. References Astone N. , Blum R.W. , Brahmbhatt H. , Delany-Moretlwe S. , Jejeebhoy S. , Olumide A. , Wang Z. . 2014 . Social Capital and Vulnerable Urban Youth in Five Global Cities . Journal of Adolescent Health 55 6 : S21 – 30 . Google Scholar Crossref Search ADS PubMed Beasley D. 2011 . Georgia Voters Approve Alcohol Retail Sales on Sundays. Available at: http://www.reuters.com/article/us-election-georgia-alcohol-idUSTRE7A86B720111109. Besley T. , Case A. . 2000 . Unnatural Experiments? Estimating The Incidence of Endogenous Policies . Economic Journal 110 467 : 672 – 94 . Google Scholar Crossref Search ADS Biderman C. , De Mello J.M.P. , Schneider A. . 2010 . Dry Laws and Homicides: Evidence from the São Paulo Metropolitan Area . Economic Journal 120 543 : 157 – 82 . Google Scholar Crossref Search ADS Bobo J.K. , Husten C. . 2000 . Sociocultural Influences on Smoking and Drinking . Alcohol Research and Health 24 4 : 225 – 32 . Google Scholar PubMed Brechting E.H. , Carlson C.R. . 2015 . Religiousness and Alcohol Use in College Students: Examining Descriptive Drinking Norms as Mediators . Journal of Child and Adolescent Substance Abuse 24 1 : 1 – 11 . Google Scholar Crossref Search ADS Burney L N. 2011 . Community-Directed Engagement and Positive Youth Development . Children and Youth Services Review 33 (S1) : S23 – 8 . Carpenter C. 2004 . Heavy Alcohol Use and Youth Suicide: Evidence from Tougher Drunk Driving Laws . Journal of Policy Analysis and Management 2004 4 : 831 – 42 . Google Scholar Crossref Search ADS Carpenter C. 2005 . Youth Alcohol Use and Risky Sexual Behavior: Evidence from Underage Drunk Driving Laws . Journal of Health Economics 24 3 : 613 – 28 . Google Scholar Crossref Search ADS PubMed Carpenter C. , Dobkin C. . 2009 . The Effect of Alcohol Consumption on Mortality: Regression Discontinuity Evidence from the Minimum Drinking Age . American Economic Journal: Applied Economics 1 1 : 164 – 82 . Google Scholar Crossref Search ADS PubMed Carpenter C. , Dobkin C. 2010 . The Drinking Age, Alcohol Consumption, and Crime. University of California. Working Paper. Carpenter C.S. , Eisenberg D. . 2009 . Effects of Sunday Sales Restrictions on Overall and Day-specific Alcohol Consumption: Evidence from Canada . Journal of Studies on Alcohol and Drugs 70 1 : 126 – 33 . Google Scholar Crossref Search ADS PubMed Chesson H. , Harrison P. , Kassler W.J. . 2000 . Sex under the Influence: The Effect of Alcohol Policy on Sexually Transmitted Disease Rates in the United States . The Journal of Law and Economics 43 1 : 215 – 38 . Google Scholar Crossref Search ADS PubMed Conover E. , Scrimgeour D. . 2013 . Health Consequences of Easier Access to Alcohol: New Zealand Evidence . Journal Of Health Economics 32 3 : 570 – 85 . Google Scholar Crossref Search ADS PubMed Cooper M.L. , Damaris R.J. , Russell M. , Skinner J.B. , Frone M.R. , Mudar P. . 1992 . Stress and Alcohol Use: Moderating Effects of Gender, Coping, and Alcohol Expectancies . Journal of Abnormal Psychology 101 1 : 139 – 52 . Google Scholar Crossref Search ADS PubMed Dee T.S. 1999a . The Complementarity of Teen Smoking and Drinking . Journal of Health Economics 18 6 : 767 – 93 . Google Scholar Crossref Search ADS Dee T.S. 1999b . State Alcohol Policies, Teen Drinking and Traffic Fatalities . Journal of Public Economics 72 2 : 289 – 315 . Google Scholar Crossref Search ADS Dent C.W. , Grube J.W. , Biglan A. . 2005 . Community Level Alcohol Availability and Enforcement of Possession Laws as Predictors of Youth Drinking . Preventive Medicine 40 3 : 355 – 62 . Google Scholar Crossref Search ADS PubMed Desimone J. 2001 . The Effect of Cocaine Prices on Crime . Economic Inquiry 39 4 : 627 – 43 . Google Scholar Crossref Search ADS Dilloff N.J. 1979 . Never on Sunday: The Blue Laws Controversy . Maryland Law Review 39 : 679 – 714 . Figlio D.N. 1995 . The Effect of Drinking Age Laws and Alcohol-Related Crashes: Time-Series Evidence from Wisconsin . Journal of Policy Analysis and Management 14 4 : 555 – 66 . Google Scholar Crossref Search ADS Friese B. , Grube J.W. , Moore R.S. . 2013 . Youth Acquisition of Alcohol and Drinking Contexts: An In-depth Look . Journal of Drug Education 43 4 : 385 – 403 . Google Scholar Crossref Search ADS PubMed Gradus R. 1996 . The Economic Effects of Extending Shop Opening Hours . Journal of Economics 64 3 : 247 – 63 . Google Scholar Crossref Search ADS Grant B.F. , Hasin D.S. , Chou S.P. , Stinson F.S. , Dawson D.A. . 2004 . Nicotine Dependence and Psychiatric Disorders in the United States: Results from the National Epidemiologic Survey on Alcohol and Related Conditions . Archives of General Psychiatry 61 11 : 1107 – 15 . Google Scholar Crossref Search ADS PubMed Grönqvist H. , Niknami S. . 2014 . Alcohol Availability and Crime: Lessons from Liberalized Weekend Sales Restrictions . Journal Of Urban Economics 81 : 77 – 84 . Google Scholar Crossref Search ADS Grossman M. , Markowitz S. . 2005 . I Did What Last Night? Adolescent Risky Sexual Behaviors and Substance Use . Eastern Economic Journal 31 3 : 383 – 405 . Grube J.W. 2010 . Environmental Approaches to Preventing Adolescent Drinking. In Handbook of Drug Use Etiology: Theory, Methods, and Empirical Findings , ed. Scheier L. , 493 – 509 . Washington DC : American Psychological Association . Gruber J. , Hungerman D. . 2008 . The Church Versus the Mall: What Happens When Religion Faces Increased Secular Competition? Quarterly Journal Of Economics 123 2 : 831 – 62 . Google Scholar Crossref Search ADS Gruenewald P.J. , Freisthler B. , Remer L. , LaScala E.A. , Treno A. . 2006 . Ecological Models of Alcohol Outlets and Violent Assaults: Crime Potentials and Geospatial Analysis . Addiction 101 5 : 666 – 77 . Google Scholar Crossref Search ADS PubMed Heaton P. 2012 . Sunday Liquor Laws and Crime . Journal of Public Economics 96 1 : 42 – 52 . Google Scholar Crossref Search ADS PubMed Hoynes H.W. , Schanzenbach D.W. . 2012 . Work Incentives and the Food Stamp Program . Journal Of Public Economics 96 ( 1): 151 – 62 . Jensen R. 2007 . The Digital Provide: Information (Technology), Market Performance, and Welfare in the South Indian Fisheries Sector . The Quarterly Journal of Economics 122 3 : 879 – 924 . Google Scholar Crossref Search ADS Joksch H. C. , Jones R.K. . 1993 . Changes in the Drinking Age and Crime . Journal of Criminal Justice 21 3 : 209 – 21 . Google Scholar Crossref Search ADS Kann L. 2016 . Youth Risk Behavior Surveillance—United States, 2015. MMWR. Surveillance Summaries, 65. Koch S. F. , McGeary K.A. . 2005 . The Effect of Youth Alcohol Initiation on High School Completion . Economic Inquiry 43 4 : 750 – 65 . Google Scholar Crossref Search ADS Lapham S. , McMillan G. . 2006 . Effectiveness of Bans and Laws in Reducing Traffic Deaths . American Journal of Public Health 96 11 : 1944 – 8 . Google Scholar Crossref Search ADS PubMed Lee D.N. 2013 . The Impact of Repealing Sunday Closing Laws on Educational Attainment . Journal of Human Resources 48 2 : 286 – 310 . Google Scholar Crossref Search ADS Lee J. , Yörük B.K. . 2014 . Does Legalization of Sunday Alcohol Sales Increase Crime? (No. 5065). CESifo Group Munich. Lovenheim M. F. , Slemrod J. . 2010 . The Fatal Toll of Driving to Drink: The Effect of Minimum Legal Drinking Age Evasion on Traffic Fatalities . Journal of Health Economics 29 1 : 62 – 77 . Google Scholar Crossref Search ADS PubMed Lovenheim M.F. , Steefel D.P. . 2011 . Do Blue Laws Save Lives? The Effect of Sunday Alcohol Sales Bans on Fatal Vehicle Accidents . Journal of Policy Analysis and Management 30 4 : 798 – 820 . Google Scholar Crossref Search ADS Menagi F. S. , Harrell Z. , June L. . 2008 . Religiousness and College Student Alcohol Use: Examining the Role of Social Support . Journal of Religion and Health 47 2 : 217 – 26 . Google Scholar Crossref Search ADS PubMed Miller J. W. , Naimi T.S. , Brewer R.D. , Jones S.E. . 2007 . Binge Drinking and Associated Health Risk Behaviors among High School Students . Pediatrics 119 1 : 76 – 85 . Google Scholar Crossref Search ADS PubMed The National Survey on Drug Use and Health Report : State Estimates of Underage Alcohol Use and Self-Purchase of Alcohol: 2008 to 2010. Rockville, MD.Substance Abuse and Mental Health Services Administration, Center for Behavioral Health Statistics and Quality. (November 20, 2012 ). National Institute on Alcohol Abuse and Alcoholism . 2014 . Drinking Levels Defined. Available at: http://www.niaaa.nih.gov/alcohol-health/overview-alcohol-consumption/moderate-binge-drinking. Nolen-Hoeksema S. 2004 . Gender Differences in Risk Factors and Consequences for Alcohol Use and Problems . Clinical Psychology Review 24 8 : 981 – 1010 . Google Scholar Crossref Search ADS PubMed Office of the Surgeon General (US) , National Institute on Alcohol Abuse and Alcoholism (US), Substance Abuse and Mental Health Services Administration (US), The Surgeon General's Call to Action To Prevent and Reduce Underage Drinking. Rockville (MD), Office of the Surgeon General (US). 2007. Section 1: Underage Drinking in America: Scope of the Problem. Available from: https://www.ncbi.nlm.nih.gov/books/NBK44364/. Paschall M.J. , Grube J.W. , Black C. , Ringwalt C.L. . 2007 . Is Commercial Alcohol Availability Related to Adolescent Alcohol Sources and Alcohol Use? Findings from a Multi-level Study . Journal of Adolescent Health 41 2 : 168 – 74 . Google Scholar Crossref Search ADS PubMed Patock-Peckham J.A. , Hutchinson G.T. , Cheong J. , Nagoshi C.T. . 1998 . Effect of Religion and Religiosity on Alcohol Use in a College Student Sample . Drug and Alcohol Dependence 49 2 : 81 – 8 . Google Scholar Crossref Search ADS PubMed Pemberton M. R. , Colliver J.D. , Robbins T.M. , Gfroerer J.C. . 2008 . Underage Alcohol Use: Findings from the 2002–2006 National Surveys on Drug Use and Health (DHHS Publication No. SMA 08–4333, Analytic Series A-30). Rockville, MD: Substance Abuse and Mental Health Services Administration, Office of Applied Studies. Sen B. 2002 . Does Alcohol-use Increase the Risk of Sexual Intercourse among Adolescents? Evidence from the NLSY97 . Journal of Health Economics 21 6 : 1085 – 93 . Google Scholar Crossref Search ADS PubMed Scribner R. , Cohen D. , Kaplan S. , Allen S. . 1999 . Alcohol Availability and Homicide in New Orleans: Conceptual Considerations for Small Area Analysis of the Effect of Alcohol Outlet Density . Journal Of Studies On Alcohol 60 3 : 310 – 6 . Google Scholar Crossref Search ADS PubMed Scribner R. , MacKinnon D. , Dwyer J. . 1995 . The Risk of Assaultive Violence and Alcohol Availability in Los Angeles County . American Journal of Public Health 85 3 : 335 – 40 . Google Scholar Crossref Search ADS PubMed Single E.W. 1988 . The availability theory of alcohol related problems. In Theories on alcoholism, ed C.D. Chaudron, & D.A. Wilkkinson, 325–351. Toronto: Addiction Research Foundation. Stehr M. 2007 . The Effect of Sunday Sales Bans and Excise Taxes on Drinking and Cross-border Shopping for Alcoholic Beverages . National Tax Journal 60 : 85 – 103 . Google Scholar Crossref Search ADS Stehr M. 2010 The Effect of Sunday Sales of Alcohol on Highway Crash Fatalities . B. E. Journal of Economic Analysis and Policy 10 1 : 1 – 20 . Google Scholar Crossref Search ADS Theuman J. 2005 . Validity, Construction, and Effect of “Sunday Closing” or “Blue” Laws—Modern Status . American Law Reports ALR4th: 246 – 332 . Waddell G. R. 2012 . Gender and the Influence of Peer Alcohol Consumption on Adolescent Sexual Activity . Economic Inquiry 50 1 : 248 – 63 . Google Scholar Crossref Search ADS PubMed Wooldridge J.M. 2010 . Econometric Analysis of Cross Section and Panel Data . Boston : MIT Press . Yörük B.K . 2014 . Legalization of Sunday alcohol sales and alcohol consumption in the United States . Addiction 109 1 : 55 – 61 . Google Scholar Crossref Search ADS PubMed Yörük C.E. , Yörük B.K. . 2015 . Alcohol Consumption and Risky Sexual Behavior among Young Adults: Evidence from Minimum Legal Drinking Age Laws . Journal of Population Economics 28 1 : 133 – 57 . Google Scholar Crossref Search ADS © The Author 2017. Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) TI - The Effect of Sunday Alcohol Sales Bans on Teen Drinking in Georgia JF - Applied Economic Perspectives and Policy DO - 10.1093/aepp/ppx046 DA - 2018-09-01 UR - https://www.deepdyve.com/lp/oxford-university-press/the-effect-of-sunday-alcohol-sales-bans-on-teen-drinking-in-georgia-hY5Ll0FsAP SP - 461 VL - 40 IS - 3 DP - DeepDyve ER -