Some have argued that Michigan differs from other control states because Michigan government is not a retailer of alcohol products and because it acts as a wholesaler for only one type of alcohol: spirituous liquor. In light of this less intrusive regime, Michigan is said to have struck a balance between the need to control alcohol and the need to allow market competition and business opportunities.[*]
Does Michigan’s intermediate alcohol control regime lead to different safety results? We assess that view by revisiting the data and subdividing states into four categories of state alcohol control: “heavy control,” “moderate control,” “light control” and license states.
Each of the categories has a specific definition. In a heavy-control state, state government sells at least two of the three major types of alcohol (beer, wine or spirits) at the retail level and also sells one or more of these at the wholesale level. In a moderate-control state, state government sells only one of the three major types of alcohol at the retail level, but still sells one or more at the wholesale level.
In a light-control state, such as Michigan, state government sells no alcohol at the retail level, but sells at least one or more of the three alcohols at the wholesale level (in Michigan’s case, hard liquor). As before, a license state simply licenses private retailers and wholesalers of the three types of alcohol.[7]
Note that these four terms are relative. It is difficult to review Michigan’s extensive controls and consider them “light,” even if state government’s marketplace intervention is less intrusive than in some other states.
Graphic 1 shows total alcohol-attributable deaths per 100,000 residents in 48 states during the period from 2001 through 2005, the most recent years for which data are available from the Centers for Disease Control and Prevention.[†] The data include deaths of both adults and children. The states in Graphic 1 are grouped from left to right by the four degrees of liquor control: heavy-control, moderate-control, light-control and license states.
The four groups are essentially indistinguishable. If state alcohol controls worked in proportion to their scope, the bars would tend to rise like stair steps from left to right across the graphic. Instead, to take just one example, the average alcohol-attributable fatality rate is lower in the license group than in the low-control group (28.46 vs. 29.95 deaths per 100,000 people, respectively). The same holds true for the under-21 fatality rate, where the average in license states is 1.70 and the average in light-control states is 1.84.[8] Statistical tests do not indicate that a state’s alcohol control regime affects average alcohol-attributable death rates.[‡]
Note that of the 10 states with the lowest fatality rates, eight are license states. The two others are a light-control state, Iowa (eighth), and a moderate-control state, New Hampshire (10th); none of the top 10 is a high-control state.
Graphic 1: Annual Average Alcohol-Attributable Deaths per 100,000 People by State, 2001-2005
Source: Authors’ calculations based on “Alcohol and Public Health: Alcohol-Related Disease Impact (ARDI),” (Centers for Disease Control and Prevention, 2008), Alcohol-Related Disease Impact (ARDI) Alcohol-Attributable Deaths, http://goo.gl/U5GsA (accessed April 20, 2012); “Population Estimates: State Intercensal Estimates (2000-2010),” (United States Census Bureau, 2011), State Intercensal Estimates (2000-2010): Annual Population Estimates, http://goo.gl/TJv2M (accessed April 20, 2012). Fatality rates include deaths from 54 chronic and acute causes and also include both adults and children. Average populations for 2001 through 2005 were used. Utah is excluded from the analysis because it changed classifications during the time period. Maryland is also excluded, because the degree of alcohol control is not constant among its counties.
It is worth noting here that some other unmeasured factors may be unduly influencing reported alcohol-attributable deaths. Proponents of strict alcohol regulation might argue that statistically controlling, for example, for a state’s unemployment rate or proportion of heavy drinkers, the bars would surely line up stepwise.
However, since atypical factors specific to individual states will tend to cancel out in the averages, the unmeasured factor would have to be related to the level of control so that it affected control states differently than it affected noncontrol states. The fact that some excluded factor differs across states is not enough, in itself, to produce the lack of evidence in Graphic 1 for the efficacy of state alcohol controls.
[*] For instance, in a news release calling for “strong alcohol laws,” the executive director of Michigan Alcohol Policy Promoting Health & Safety stated, “For many years, Michigan’s alcohol laws have served to promote public health, moderation and safety, while balancing the needs of businesses to pursue new opportunities and growth in the beer, wine and liquor sector.” Tobias and Hansen, “Dozens of Groups Sign Letter Supporting Strong Alcohol Laws and Public Health as Controversial Committee Recommends Changes,” (Michigan Alcohol Policy Promoting Health & Safety, 2011), http://goo.gl/wSwnY (accessed February 2, 2012). The president of the Michigan Beer and Wine Wholesalers Association has likewise stated: “Alcohol is not like other consumer products. It must be regulated at a higher standard. Here in Michigan, we have found a balance that emphasizes accountability yet promotes competition and wide consumer choice, and that’s why Michigan is a national model for alcohol regulations.” “National Study: Americans support meaningful alcohol regulations,” Today’s Wholesaler Volume 35, no. 2 (2011) http://goo.gl/3P5NX (accessed March 27, 2012).
[†] “Alcohol and Public Health: Alcohol-Related Disease Impact (ARDI),” (Centers for Disease Control and Prevention, 2008), Alcohol-Related Disease Impact (ARDI) Alcohol-Attributable Deaths, http://goo.gl/U5GsA (accessed April 20, 2012). Alcohol-attributable deaths include 54 acute and chronic causes in which alcohol played a direct or indirect role; see ibid.; “Alcohol and Public Health: Alcohol-Related Disease Impact (ARDI): Methods,” (Centers for Disease Control and Prevention, 2008), http://goo.gl/qhd0U (accessed April 22, 2012). The state population figures used to generate the rates were averages from 2001 through 2005 drawn from “Population Estimates: State Intercensal Estimates (2000-2010),” (United States Census Bureau, 2011), http://goo.gl/TJv2M (accessed April 20, 2012). Utah is excluded from the analysis because it changed classifications during the time period. Maryland is also excluded, since the degree of alcohol control varies among its counties.
[‡] P-values do not suggest a link between the average alcohol-attributable fatality rates and the level of alcohol control. The following results are generated for tests of the null hypothesis that there is no difference in the population means:
- Joint difference among heavy, moderate, light and license: p-value of 0.85 for ANOVA test
- Difference between heavy/moderate/light and license: p-value of 0.97 for difference of means test
- Difference between heavy/moderate and light/license: p-value of 0.55 for difference of means test.
P-values produce similar conclusions for the under-21 alcohol-attributable fatality rate:
- Joint difference among heavy, moderate, light and license: p-value of 0.52 for ANOVA test
- Difference between heavy/moderate/light and license: p-value of 0.51 for difference of means test
- Difference between heavy/moderate and light/license: p-value of 0.15 for difference of means test.
Note that other difference-of-means tests are problematic due to the small number of observations.