Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2011 Nov 1.
Published in final edited form as: Alcohol Clin Exp Res. 2010 Nov;34(11):1915–1921. doi: 10.1111/j.1530-0277.2010.01280.x

Effects of alcohol taxes on alcohol-related mortality in Florida: Time-series analyses from 1969–2004

Mildred M Maldonado-Molina 1,1, Alexander C Wagenaar 1
PMCID: PMC2965314  NIHMSID: NIHMS217022  PMID: 20659073

Abstract

Background

Over a hundred studies have established the effects of beverage alcohol taxes and prices on sales and drinking behaviors. Yet, relatively few studies have examined effects of alcohol taxes on alcohol-related mortality. We evaluated effects of multiple changes in alcohol tax rates in the State of Florida from 1969–2004 on disease (not injury) mortality.

Methods

A time-series quasi-experimental research design was used, including non-alcohol deaths within Florida and other states’ rates of alcohol-related mortality for comparison. A total of 432 monthly observations of mortality in Florida were examined over the 36-year period. Analyses included ARIMA, fixed-effects, and random effects models, including a noise model, tax independent variables, and structural covariates.

Results

We found significant reductions in mortality related to chronic heavy alcohol consumption following legislatively induced increases in alcohol taxes in Florida. The frequency of deaths (t=−2.73, p=.007) and the rate per population (t=−2.06, p=.04) declined significantly. The elasticity effect estimate is −0.22 (t=−1.88, p=.06), indicating a 10% increase in tax is associated with a 2.2% decline in deaths.

Conclusions

Increased alcohol taxes are associated with significant and sizable reductions in alcohol-attributable mortality in Florida. Results indicate that 600–800 lives per year could be saved if real tax rates were returned to 1983 levels (when the last tax increase occurred). Findings highlight the role of tax policy as an effective means for reducing deaths associated with chronic heavy alcohol use.

Keywords: Alcohol taxes, Florida, Time-series, Alcohol Policy, Mortality

In the United States, excessive alcohol consumption is the third leading preventable cause of death (Midanik et al., 2004) and alcohol use is an important contributor to violence, car crashes, unplanned pregnancy, school drop-out, reduced workplace productivity, increased risk for several types of cancer, cardiovascular disease, and stroke, as well as other alcohol-related mortality (Bagnardi et al., 2001; Boffetta and Hashibe, 2006; Bye, 2007; Cherpitel, 2007; Corrao et al., 2004; Corrao et al., 2000; English et al., 1995; Hingson et al., 2003; Inoue and Tsugane, 2005; Mangione et al., 1999; McCluskey et al., 2002; National Highway Traffic Safety Administration, 2005; Rehm et al., 2003; Room et al., 2005; Smith et al., 1999). Alcohol-related consequences represent a considerable expense, with the estimated cost exceeding $185 billion annually (NIAAA, 2006).

Effects of alcohol taxes and prices on drinking are very well established. A recent systematic review of the literature demonstrates the extensive evidence that beverage alcohol prices and taxes are inversely related to drinking and that public policies that increase the price of alcohol are an effective means to reduce alcohol consumption (Wagenaar et al., 2009b). Other studies have found that higher alcohol prices or taxes are associated with lower rates of alcohol dependence (Henderson et al., 2004; Skog and Melberg, 2006), liver cirrhosis (Cook, 1982; Heien and Pompelli, 1987; Ponicki and Gruenewald, 2006; Rush et al., 1986; Smart and Mann, 1998), and injuries (Birckmayer and Hemenway, 1999; Grossman and Markowitz, 1999; Markowitz, 2000; Markowitz, 2005; Markowitz et al., 2003; Markowitz and Grossman, 1998; Markowitz and Grossman, 2000; Yamasaki et al., 2005).

Only eleven published studies to date have examined effects of alcohol taxes on alcohol-related mortality. Of these, five studies were conducted in Finland, Denmark and Canada (Herttua et al., 2008; Koski et al., 2007; Rush et al., 1986; Skog and Melberg, 2006; Smart and Mann, 1998), and six studies were conducted in the United States (Cook and Tauchen, 1982; Heien and Pompelli, 1987; Rush et al., 1986; Schweitzer et al., 1983; Sloan et al., 1994; Wagenaar et al., 2009a). Of the six U.S. studies, only one study was completed in the last decade (Wagenaar et al., 2009a). This recent study of ours examined the effects of two sudden substantial tax increases (almost two decades apart) in Alaska. We found fairly large immediate and sustained reductions in alcohol-related disease mortality (ranging from 11% to 29%). Given the size of the observed effects, and concern that Alaska may not be typical of the lower 48 states, we designed the current study as a replication of the recent Alaska study in a state that is very different in many physical, demographic and social characteristics, and with tax rate changes somewhat smaller in magnitude than those in Alaska.

Florida is different from Alaska in many ways. First, according to a recent report fom the Centers for Disease Control and Prevention, the rate of alcohol-related mortality in Florida (8.7), although slightly higher than the U.S. rate (7.0), is much lower than the rate in Alaska (21.6); Alaska is the state with the highest rate of alcohol-induced deaths in the United States (Heron et al., 2009). Second, geographic location and demographic composition (e.g. population size, density, immigrant population) also differentiate Florida from Alaska. In Florida, population (15.9M in 2000) and density (296 persons per square mile) are significantly higher than Alaska (626K population and 1.1 persons per square mile; (U.S. Census Bureau, 2009). In Florida, the proportion of immigrants (17%) is higher than the US average (11.1%), and much higher than Alaska (6%). In particular, the proportion of Latino and Hispanic origin is higher in Florida (21%) compared to the US (15.4%) and Alaska (6.1%); while the proportion of American Indian or Alaska Native is much higher in Alaska (15.3%) when compared to the US (1.0%) and the state of Florida (0.5%) (U.S. Census Bureau, 2009).

Third, timing, frequency, and size of tax changes also differ between Florida and Alaska. In Alaska, two large tax changes were implemented almost 20 years apart (one in 1983, the other in 2002). In contrast, in Florida, no changes to beer, wine, or spirits tax rates have been implemented since 1983. Moreover, tax increases in Florida have been modest when compared to faily large sudden changes in Alaska. From 1969–2004, Florida changed the nominal alcohol tax rates on two separate occasions (Table 1). Effective July 1, 1977, the excise tax on beer increased from $0.32 to $0.40 per gallon, the tax on wine increased from $1.15 to $1.75 per gallon, and the tax on distilled spirits increased from $3.75 to $4.75 per gallon (Fla. Laws 1977, c. 77–407 -Fla. Stat. § 565.12 and Fla. Stat. § 563.05(1)). Effective September 1, 1983, the tax on beer increased to $0.48 per gallon, wine increased to $2.25 per gallon, and spirits increased to $6.50 per gallon (Fla. Laws 1983, c. 83–349 §12- Fla. Stat. § 563.05(1) and Fla. Stat. § 565.12). Two additional changes in tax legislation were implemented, but these changes did not really change the total rate of alcohol taxes. Effective July 1, 1988, the tax on distilled spirits was amended to $4.75 (Fla Laws, c. 88– 308 -Fla. Stat. § 565.12(1), but an additional $1.75 per gallon for distilled spirits imported into the state was implemented. Thus, for the bulk of the spirits market, the tax rate remained unchanged at $6.50 per gallon. After a suit from several alcohol distilleries (see 541 So. 2d 1129; 1989), the Florida Supreme Court determined the amended sections of Florida Statutes §565.12 were unconstitutional and therefore were invalidated on March 23, 1989. Due to this decision, the excise tax rate that existed the day before the invalidated law took effect were returned (i.e. rates implemented in 1983). Effective May 9, 1991, the Florida Legislature enacted Ch. 91–60 Laws of Florida cleaned up this detail by amending Florida Statutes §565.12 specifying the excise tax rate of distilled spirits and wine to $6.50 and $2.25 respectfully (identical rates of those implemented a decade earlier in 1983; 1991 Fla. Laws 60 or 1991 ALS 60, Fla. Stat. § 565.12(1)).

Table 1.

Nominal alcohol excise tax rates in Florida, 1969–2004

1969→ 1977→ 1983→
Quantity ($) per gallon
Beer 0.32 0.40 0.48
Wine 1.15 1.75 2.25
Spirits 3.75 4.75 6.50
Percent Change
Beer -- +25 +20
Wine -- +52 +29
Spirits -- +27 +37
Open in a new tab

State-specific analyses of alcohol tax effects are important for several reasons. First, alcohol tax policy in the U.S. is largely implemented at the state level. State policymakers seek evidence on policy effects in the jurisdiction over which they have decision authority. Data from other jurisdictions is deemed less relevant. This has been observed across many public health policy arenas. Changes in the legal drinking age, safety belt use laws, penalties for alcohol-impaired driving, regulation of young drivers, and many others were studied in specific early-adopter states, and over time additional state-specific studies finding beneficial policy effects facilitated the diffusion of the policy innovation to more states. Second, more in-depth state-specific analyses facilitate use of higher time-resolution data (monthly rather than simply annual), because the more complex error structures of higher resolution data are difficult to accurately capture in 50-state pooled analyses where simplifying assumptions regarding error structures are required. Third, Florida has a population of 18 million (U.S. Census Bureau, 2009), larger than 70% of all countries in the world, and replications in diverse settings is one of the key means of advancing the state of the science.

In summary, consistent with prior studies, we hypothesized that increases in alcohol taxes were associated with reductions to alcohol-related mortality in Florida, although we expected effects to be smaller than observed in Alaska because the tax changes were in smaller increments dispersed over time.

METHODS

Research Design

The current study used a time-series quasi-experimental research design. The objective was to identify effects of alcohol taxation on alcohol-related mortality while controlling for possible alternative explanations for an observed effect by the use of comparison data series. Use of comparison time-series controls for many factors that operate similarly across states or across outcome measures (e.g. economic conditions, national social trends), including many such potential confounds for which reliable measures are not available. Rates of alcohol excise taxes were collected separately for the three major alcohol beverage categories: beer, wine, and distilled spirits. The time-series included 102 baseline monthly observations before the first tax increase in 1977 on beer, wine and spirits; 74 monthly observations after the first but before the second tax increase in 1983 on beer, wine and spirits; with 256 follow-up monthly observations after the 1983 tax increase, producing a total time series of 432 repeated observations from January 1969 to December 2004.

Measures

Mortality

Alcohol-related mortality measures are based on data from death certificates recorded in the U.S. National Vital Statistics System of the National Center for Health Statistics. We obtained the complete annual dataset containing one record on each deceased person for each year from 1969 through 2005 from the National Bureau of Economic Research (www.nber.org/data/multicause.html). From these data we created monthly counts of deaths stratified by underlying cause of death, first for Florida and then for the other states as a group.

Alcohol-related mortality includes all deaths due to diseases where the alcohol-attributable fraction is greater than .35 (e.g., alcoholic liver disease, alcohol-induced chronic pancreatitis, alcohol psychoses, alcohol abuse, alcohol dependence syndrome, alcoholic polyneuropathy, alcoholic cardiomyopathy, alcoholic gastritis, acute alcohol poisoning; other cirrhosis; cholelithiasis; acute and chronic pancreatitis; malignant neoplasms of mouth, pharynx, esophagus, liver, breast; epilepsy; cardiovascular diseases including hypertensive, ischemic, arrhythmias, cerebrovascular, ischemic and hemorrhagic stroke). Note that unintentional and intentional injury deaths are purposely not included—this is a study specifically of tax effects on alcohol-related disease.

Alcohol-attributable fractions by ICD-code were based on English et al. (1995), Rehm et al. (2006) and Shultz et al. (1991). ICD-codes by outcome category and ICD version are shown in Table 2. We calculated death rates per 100,000 population age 15 and over using annual estimates of population from the U.S. Census Bureau (www.census.gov/popest/states/). Rates of alcohol-related mortality in Florida from 1969 to 2004 are shown in Figure 1 with vertical lines denoting months where legislative changes in excise tax rates took effect. The rates of alcohol-related mortality for the other 47 contiguous US states (including DC) are also shown for comparison.

Table 2.

Alcohol-related Mortality Measures by ICD code, Alcohol attributable fraction ≥ 0.35

ICD-8
(1976–1978)		 ICD-9
(1979–1998)		 ICD-10
(1999–2005)
291.0 291.1 291.2 291.3 291.0 291.1 291.2 291.3 F10.0 F10.1 F10.2 F10.3
291.9 303.0 303.1 303.2 291.4 291.5 291.8 291.9 303 F10.4 F10.5 F10.6 F10.7
303.9 571.0 980.0 E860 305.0 357.5 425.5 535.3 F10.8 F10.9 G62.1 I42.6
571.0 571.1 571.2 571.3 K29.2 K70.0 K70.1 K70.2
790.3 980.0 E860.0 E860.1 K70.3 K70.4 K70.9 G31.2
K86.0 R78.0 X45 Y15
T51.0 T51.9
140.0 140.1 140.2 140.9 140.0 140.1 140.3 140.4 C00.0 C00.1 C00.2 C00.3
141.0 141.1 141.2 141.3 140.5 140.6 140.8 140.9 C00.4 C00.5 C00.6 C00.8
141.9 142.0 142.8 142.9 141.0 141.1 141.2 141.3 C00.9
143.0 143.1 143.9 141.4 141.5 141.6 141.8 C01 C02.0 C02.1
144 145.0 145.1 141.9 142.0 142.1 142.2 C02.2 C02.3 C02.4 C02.8
145.8 145.9 146.0 146.8 142.8 142.9 143.0 143.1 C02.9 C03.0 C03.1 C03.9
146.9 147 143.8 143.9 144.0 144.1 C04.0 C04.1 C04.8 C04.9
148.0 148.1 148.8 148.9 149 144.8 144.9 145.0 145.1 C05.0 C05.1 C05.2 C05.8
150 161.0 161.8 161.9 571.8 145.2 145.3 145.4 145.5 C05.9 C06.0 C06.1 C06.2
571.9 577.0 577.1 145.6 145.8 145.9 146.0 C06.8 C06.9 C07 C08.0
146.1 146.2 146.3 146.4 C08.1 C08.8 C08.9 C09.0
146.5 146.6 146.7 146.8 C09.1 C09.8 C09.9 C10.0
146.9 147.0 147.1 147.2 C10.1 C10.2 C10.3 C10.4
147.3 147.8 147.9 148.0 C10.8 C10.9 C11.0 C11.1
148.1 148.2 148.3 148.8 C11.2 C11.3 C11.8 C11.9
148.9 149.0 149.1 149.8 C12
149.9 150.0 150.1 150.2 C13.0 C13.1 C13.2 C13.8
150.3 150.4 150.5 150.8 C13.9 C14.0 C14.2 C14.8
150.9 161.0 161.1 161.2 C15.0 C15.1 C15.2 C15.3
161.3 161.8 161.9 571.5 C15.4 C15.5 C15.8 C15.9
571.6 577.0 577.1 C32.0 C32.1 C32.2 C32.3
C32.8 C32.9 K74.0 K74.1
K74.2 K74.3 K74.4 K74.5
L74.6 K85 K86.1
Open in a new tab
Figure 1.

Figure 1

Open in a new tab

Rates of alcohol-related mortality per 100,000 population, 1969–2004

Alcohol Taxes

The Alcohol Policy Information System provides summaries of tax rates for multiple classes of alcoholic beverages, details on changes in rate, full legal citations and text from the relevant codified statutes, starting with calendar year 2003 (www.alcoholpolicy.niaaa.nih.gov). We collected data for the period from 1969 to 2003 using experienced research attorneys using standard legal research methods on the records of codified statutes in Westlaw (www.westlaw.com) and LexusNexis (www.lexisnexis.com) when available, and law library paper copy for earlier years.

Analytical Strategy

Time series analyses were conducted in phases. To maximize the analytic benefit of the large number of repeated observations (and to detect stationarity and seasonality), we used the Box-Jenkins method to fit autoregressive moving average (ARIMA) models (Brockwell and Davis, 2002; Chatfield, 2003). First, we examined a seasonal ARIMA model with structure (0,1,1)(0,1,1)12; and the final model is: (1 − B12)Yt = α + ωIt + βZt + ψ Xt + (1 − ΘB12)ut, where Yt is the outcome measure (i.e., alcohol-related mortatlity) by month from t=1 (January 1969) through t=432 (December 2004); α is a constant; ω is the estimated effect of alcohol taxation; It is the tax rate at month t; β is the estimated effect or shared variance due to Zt, the rate of alcohol-related disease in the comparison states; ψ is a vector of estimates that control for the effects of X, which are Florida-specific factors, such as economic conditions, included as covariates; Θ is the first order seasonal moving average parameter; ut is a random (white noise) error component, and B is the backshift operator such that B12(yt) equals yt-12. All models were estimated using SAS Proc ARIMA in SAS 9.1 (SAS Institute, 2004).

Just for completion, we also examined two additional sets of models. One set of models examined the effects of each separate alcohol tax change on alcohol-related mortality outcomes (first one in 1977, the second one in 1983) using an ARIMA model similar to the above. Another set of models used generalized linear mixed models. The general form of a linear mixed model for outcome variable Y (i.e. alcohol-related mortality) can be written as Yt = β0 + β1X1t + β2X2t + β2Zt + εt, where β0 is the intercept, β1 is the effect of implementation of first tax increase in 1977, X1t, is a equal to 0 before the 1977 tax change took effect and 1 after the change; β2 is the estimated effect of implementation of the 1983 alcohol tax increase; X2t is equal to 0 before the 1983 tax change took effect and 1 after the change, β2 is the estimated effects due to Zt, covariates in the model (e.g. rate of alcohol-related disease in the comparison states); and εt is the error term. We estimated a model with a first-order AR(1) autoregressive covariance structure (to account for annual repeated measures of alcohol-related mortality) plus seasonality as a fixed effect (by including month as a fixed effect in the model). This model is also known as a Marvok process (Gottman, 1981). We also estimated a model where seasonality was included as a random effect (by specifying a Toeplitz or banded covariance structure, which can be interpreted as a moving-average structure with order equal to q-1, where q=13). A Toeplitz structure is similar to the first-order autoregressive structure (where correlation decreases exponentially across the lags of the timepoints), but the Toeplitz structure is not as rigid as AR(1) for diminishing correlations across the lags (Hedeker and Gibbons, 2006).

Three outcome models were estimated: (1) raw frequency (including the comparison states, population, and other covariates), (2) rate per 100,000 population (including the comparison states and other covariate), and (3) natural logs of rate per 100,000 population (including the comparison states and other covariates). Rates per population were examined to assess whether observed differences were due to population changes over the 36-year period. Natural logs were modeled to ensure estimates were not affected by heteroscedasticity—changing variance in death rates over time. The comparison states covariate was included to ensure any estimated effects for Florida are not due to any of many other developmental, healthcare, and societal changes over time across states. Third, we adjusted the level of alcohol taxes by inflation using the Consumer Price Index for all urban consumers (CPI-U), obtained from the U.S. Department of Labor, Bureau of Labor Statistics (ftp://ftp.bls.gov/pub/special.requests/cpi/cpiai.txt). Figure 2 illustrates for beer the combined effect of legislated tax changes and inflation on real beer tax rates in Florida.

Figure 2.

Figure 2

Open in a new tab

Nominal and real beer excise tax rate in Florida, 1969– 2004

Because timing of the implementation of changes in tax rates on beer, wine, and spirits were identical (for the 1977 and 1983 changes), tax rates for beer, wine, and spirits are very highly correlated (ranging from 0.96 and 0.98). Given high colinearity, we compared model selection criteria for seven competing models, including different combinations of alcohol excise tax indicators: (1) beer, wine, and spirits, (2) beer and wine, (3) beer and spirits, (4) wine and spirits, (5) beer only, (6) wine only, and (7) spirits only. Using model selection indexes, Bayesian information criterion (Raftery, 1986) and Akaike information criterion (Akaike, 1974), we selected the beer model as the most parsimonious best-fitting model (BIC=−10006.03, SBC=−981.80, R2=.71). In addition, beer is the beverage category accounting for the highest proportion of ethanol consumed (1.30 gallons pure ethanol per person in Florida in 2007, vs. wine 0.49 gallons and spirits 0.93; (LaVallee et al., 2010). Thus, we first report results for models using real beer tax rates as the independent variable (and not including wine and spirits taxes). While the models technically used beer taxes as the independent variable, it is important to note that the magnitude of observed effects includes effects of collinear/simultaneous changes in wine and spirits taxes.

We included the corresponding rate of mortality in other US states as a right-hand side variable to account for many factors (even many such factors for which reliable measures are not available) operating similarly across states. Specifically, we aggregated the 47 other US contiguous states (plus DC) as a single comparison group. We also tested for two federal alcohol tax changes (e.g. changes in 1991 and 1995). These changes were not statistically significant (p=.86 and p=.30) after adjusting for rates of alcohol-related mortality for all other states. In other words, the other states covariate already accounts for the effects of the federal changes applying across all the states; thus, the federal taxes were not included in the final model. In addition to controlling for all variance in common across states with the other-states covariate, we included several Florida-specific covariates to account for measurement, economic, and quality of care factors that could be related to Florida alcohol-related mortality over time (e.g. non-alcohol related deaths, personal income). We tested for outliers, using a significance criterion of p<.001 for outlier detection; no significant outliers were detected. We also checked for possible measurement system effects on alcohol-related mortality over time by ICD code edition (8th vs. 9th vs. 10th edition coding schemes). The ICD coding edition did not have a significant effect and therefore was not included in the final model (p-values ranged between .48 and .99). We then tested for the Florida-specific effects of economic conditions using personal income and unemployment rates. Because unemployment rate was not significant (p=.68 for spirits, p=.86 for beer, and p=.97 for wine models), only personal income was retained to control for Florida-specific economic effects (the correlation between unemployment and personal income was −0.28, p<.001). Data on personal income was obtained from the U.S. Bureau of Economic Analysis (www.bea.gov/regional/sqpi/default.cfm). To account for factors related to health care, social conditions, and overall quality of life, we included the rate of intra-Florida non-alcohol-related deaths as a covariate. We obtained the counts of non-alcohol-related deaths by substracting from total deaths the deaths related to alcohol. Factors influencing overall quality of life and improvements in health care are relevant across diseases (and not exclusive to alcohol-related mortality).

RESULTS

Results show clearly significant reductions in mortality related to alcohol consumption following legislatively induced increases in alcohol taxes in Florida (Table 3). The frequency of alcohol-related mortality (i.e. causes of death where the alcohol-attributable fraction is greater than 0.35) declines by an estimated 69 deaths per month for each unit (dollar per gallon) increase in tax (t=−2.73, p=.007). Based on the mean monthly number of deaths over the 1969–2004 period of 303, this represents a 23% reduction in deaths. The rate per 100,000 population of such deaths declines by .77 (t=−2.06, p=.04), a 23% reduction from the mean rate of 3.3. The most straightforwardly interpretable estimates of the magnitude of this effect come from the double-log models, which provide direct elasticity estimates. The elasticity effect estimate is −0.22 (t=−1.88, p=.06), indicating a 10% increase in tax is related to a 2.2% decline in deaths, and a doubling of the tax is related to a 22% decline in deaths. Elasticity estimates from the ARIMA model are similar to estimates of the effects of real tax rate from fixed-effects and random-effects models--doubling of the tax is associated with a 17% to 18% reduction in alcohol-related mortality from the latter models (Table 4).

Table 3.

Effects of Alcohol Excise Tax Rate on Alcohol-related Mortality, Florida, 1969–2004, ARIMA (0,1,1)(0,1,1)12 models*

Estimate Std Error t p
Raw Frequency
MA lag 1 0.890 0.024 37.60 <.0001
MA lag 12 0.800 0.031 25.92 <.0001
Real Alcohol Tax Rate −69.280 25.369 −2.73 0.007
Personal Income 126.698 58.279 2.17 0.0303
FL Non-AR mortality −0.0003 0.0002 −1.56 0.1201
Other States AR mortality 0.036 0.009 3.76 0.0002
Rate per 100K population
MA lag 1 0.841 0.028 30.54 <.0001
MA lag 12 0.699 0.036 19.46 <.0001
Real Alcohol Tax Rate −0.771 0.373 −2.06 0.0396
Personal Income 1.624 0.904 1.80 0.0731
FL Non-AR mortality −0.0003 0.0002 −1.31 0.1924
Other States AR mortality 0.713 0.202 3.53 0.0005
Elasticity (log-log model)
MA lag 1 0.863 0.026 33.57 <.0001
MA lag 12 0.714 0.035 20.18 <.0001
Real Alcohol Tax Rate −0.217 0.115 −1.88 0.0609
Personal Income 0.433 0.242 1.79 0.0744
FL Non-AR mortality −0.182 0.243 − 0.75 0.4544
Other States AR mortality 0.535 0.149 3.60 0.0004
Open in a new tab
*

AR=Alcohol-related; Real Alcohol Tax Rate ($/per gallon, adjusted for inflation); Personal Income (Log of personal income, adjusted for inflation).

Table 4.

Elasticity estimates of real alcohol tax rate: Comparing results from ARIMA, fixed effects, and random effects models*

Estimate se t p
ARIMA (0,1,1)(0,1,1)12 −0.217 0.115 −1.88 0.061
Fixed-effects model −0.168 0.047 −3.54 0.0004
Random-effects model −0.180 0.052 −3.44 0.0006
Open in a new tab
*

Controlling for personal income, non-alcohol-related mortality, and other states alcohol-related mortality.

We also tested the separate effects of each alcohol tax (Table 5). The first legislative change, effective in 1977 (which increased beer, wine, and spirits tax rates), was associated with reductions of 4.7% in alcohol-related mortality (t=1.31, p=.19, cohen’s d=−0.35), although not statistically significant (according to the ARIMA model). Autoregresive models, where seasonality was modeled as a fixed effect, showed effects of similar magnitude, a 4.6% reduction, and statistically significant (t=−2.09, p=.04, cohen’s d=−.034). The second legislative change, effective in 1983 (which increased beer, wine, and spirits taxes), was also associated with reductions in alcohol-related mortality (t=−1.24, p=.22), representing a 4.4% reduction (Cohen’s d = −0.34). Similarly, autoregressive models, where seasonality was modeled as a fixed effect, showed statistically significant reductions in alcohol-related mortality, ranging between 9% and 10%, p<0.0001).

Table 5.

Effects of each separate alcohol tax change on alcohol-related mortality, Florida, 1969–2004*

Estimate Std Error t p % change Cohen’s d
1977 Tax
ARIMA (0,1,1)(0,1,1)12 −0.047 0.036 −1.31 0.1895 −4.6 −0.35
Fixed-effects model −0.046 0.022 −2.09 0.038 −4.4 −0.34
Random-effects model −0.025 0.024 −1.05 0.296 −2.5 −0.18
1983 Tax
ARIMA (0,1,1)(0,1,1)12 −0.044 0.036 −1.24 0.2158 −4.3 −0.34
Fixed-effects model −0.103 0.014 −7.45 <.0001 −9.8 −0.76
Random-effects model −0.090 0.015 −5.83 <.0001 −8.6 −0.66
Open in a new tab
*

Controlling for income, non-alcohol-related mortality, other states alcohol-related mortality, and inflation (consumer price index, CPI-U).

DISCUSSION

The current study finds significant and substantial effects of beverage alcohol taxes on disease deaths related to chronic heavy alcohol use, replicating findings from our previous study in Alaska (Wagenaar et al., 2009a), but in a state that implemented much smaller increases in excise tax rates. Although the size of the tax increases in Alaska were atypical, the effect of tax increases in reducing deaths clearly was not. A recent systematic review identified 11 previous studies of effects of tax or price changes on alcohol disease mortality; the meta-estimate elasticity of effect across all the studies is −0.347, p=<.001; (Wagenaar et al., in press). In terms of two of the most recent studies, tax increases in Alaska were associated with reductions ranging from −11 percent to −29 percent in alcohol disease mortality (Wagenaar et al., 2009a). In Finland, large 2004 alcohol tax and price reductions (resulting from European Union open borders and degregulation policies) were associated with increases of 16 to 31 percent in alcohol-related mortality (Herttua et al., 2008; Koski et al., 2007). These two recent studies demontrate the effects of tax policy in both directions—a tax increase reduces mortality and a tax decrease increases mortality. In summary, the current study is consistent with and adds to the previous literature finding significant effects of alcohol tax policy on mortality.

Note a key limitation of the current study. We examined the effects of alcohol taxes on deaths due to alcohol-related disease (e.g., cirrhosis, some cancers) which result from chronic heavy drinking (Table 2). Injuries resulting from acute situational heavy drinking are not included in the current study. Thus, the total number of deaths averted by increased alcohol taxes are much higher than the numbers reported here. In addition, we evaluated the overall effect of alcohol taxes on mortality over time, but did not test whether alcohol taxes had a differential effect during different time periods over the four decade period (i.e. alcohol taxes as a time-varying effect). Future studies should combine long time-series over many decades; and include state-specific results from multiple states to better examine whether population characteristics, such as demographics, poverty level or cultural characteristics interact with or moderate effects of alcohol taxes on mortality outcomes.

Our study has several strengths. The quasi-experimental design, incorporating comparison states and comparison (non-alcohol) death rates, along with statistical controls for Florida-specific economic conditions, increases confidence the observed effects are causally attributable to the alcohol tax changes, and not other factors. Additionally, the current study included higher time-resolution monthly observations over a lengthy a four-decade period, and rates of alcohol taxes were adjusted to account for general inflation.

Florida, just as every other U.S. state and the federal government, does not automatically adjust alcohol excise taxes for inflation. As a result, actual alcohol tax rates effectively decrease continuously with inflation—in the case of Florida, current beer tax rates in real (inflation-adjusted) dollars are only one-quarter the rate in the 1960s. Importantly, Florida has not increased alcohol taxes since 1983. Simply returning the real tax rates to the 1983 level by adjusting for the erosion due to inflation (achieved by doubling the current tax rates) would likely save at least 600 and perhaps as many as 800 lives per year in Florida alone (not including injury deaths averted). Returning to the real tax rates of the 1970s would likely double the number of lives saved annually. The public health significance of these findings are obvious.

Acknowledgements

This study was supported by the Robert Wood Johnson Foundation under grant 058005 to Alexander C. Wagenaar (Principal Investigator) and the National Institute on Alcohol Abuse and Alcoholism (AA017480) to Mildred M. Maldonado-Molina (Principal Investigator). The authors thank Emilia Sykes and Bradley H. Wagenaar for assistance with data collection and Dandan Xu for database design and management. Findings and conclusions are solely the authors’ and do not necessarily represent the views of the Robert Wood Johnson Foundation or the University of Florida.

References

  1. Akaike H. A new look at the statistical model identification. IEEE Transactions on Automatic Control. 1974;19(6):716–723. [Google Scholar]
  2. Bagnardi V, Blangiardo M, Corrao G, La Vecchia C. Alcohol consumption and the risk of cancer: a meta-analysis. Alcohol Research & Health. 2001;25:263–270. [PMC free article] [PubMed] [Google Scholar]
  3. Birckmayer J, Hemenway D. Minimum-age drinking laws and youth suicide, 1970–1990. American Journal of Public Health. 1999;89(9):1365–1368. doi: 10.2105/ajph.89.9.1365. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Boffetta P, Hashibe M. Alcohol and cancer. The Lancet Oncology. 2006;7(2):149–156. doi: 10.1016/S1470-2045(06)70577-0. [DOI] [PubMed] [Google Scholar]
  5. Brockwell PJ, Davis RA. Introduction to Time Series and Forecasting. 2nd ed. New York: Springer; 2002. [Google Scholar]
  6. Bye EK. Alcohol and violence: use of possible confounders in a time-series analysis. Addiction. 2007;102(3):369–376. doi: 10.1111/j.1360-0443.2006.01701.x. [DOI] [PubMed] [Google Scholar]
  7. Chatfield C. The Analysis of Time Series: An Introduction 6th ed. New York: Chapman & Hall/CRC; 2003. [Google Scholar]
  8. Cherpitel CJ. Alcohol and injuries: a review of international emergency room studies since 1995. Drug and Alcohol Review. 2007;26(2):201–214. doi: 10.1080/09595230601146686. [DOI] [PubMed] [Google Scholar]
  9. Cook PJ. Alcohol taxes as a public-health measure. British Journal of Addiction. 1982;77(3):245–250. doi: 10.1111/j.1360-0443.1982.tb02453.x. [DOI] [PubMed] [Google Scholar]
  10. Cook PJ, Tauchen G. The effect of liquor taxes on heavy drinking. Bell Journal of Economics. 1982;13(2):379–390. [Google Scholar]
  11. Corrao G, Bagnardi V, Zambon A, La Vecchia C. A meta-analysis of alcohol consumption and the risk of 15 diseases. Preventive Medicine. 2004;38:613–619. doi: 10.1016/j.ypmed.2003.11.027. [DOI] [PubMed] [Google Scholar]
  12. Corrao G, Rubbiati L, Bagnardi V, Zambon A, Poikolainen K. Alcohol and coronary heart disease: a meta-analysis. Addiction. 2000;95:1505–1523. doi: 10.1046/j.1360-0443.2000.951015056.x. [DOI] [PubMed] [Google Scholar]
  13. English DR, Holman CDJ, Milne E, Winter MG, Hulse GK, Codde JP, Bower CI, Corti B, de Klerk N, Knuiman MW, Kurinczuk JJ, Lewin GF, Ryan GA. Commonwealth Department of Human Services and Health. Australia: Canberra; 1995. The quantification of drug caused morbidity and mortality in Australia. [Google Scholar]
  14. Gottman JM. Time-series Analysis: A comprehensive Introduction for Social Scientists. New York: Cambridge University Press; 1981. [Google Scholar]
  15. Grossman M, Markowitz S. Alcohol Regulation and Violence on College Campuses. Cambridge, MA: National Bureau of Economic Research; 1999. (Report No. Working Paper 7129) [Google Scholar]
  16. Hedeker D, Gibbons RD. Longitudinal Data Analysis. New Jersey: John Wiley & Sons, Inc; 2006. [Google Scholar]
  17. Heien D, Pompelli G. Stress, ethnic and distribution factors in a dichotomous response model of alcohol-abuse. Journal of Studies on Alcohol. 1987;48(5):450–455. doi: 10.15288/jsa.1987.48.450. [DOI] [PubMed] [Google Scholar]
  18. Henderson C, Liu XH, Roux AVD, Link BG, Hasin D. The effects of US state income inequality and alcohol policies on symptoms of depression and alcohol dependence. Social Science & Medicine. 2004;58(3):565–575. doi: 10.1016/s0277-9536(03)00228-4. [DOI] [PubMed] [Google Scholar]
  19. Heron M, Hoyert DL, Murphy SL, Xu J, Kochanek KD, Tejada-Vera B. U.S. Department of Health and Human Services, Center for Disease Control and Prevention; Deaths: Final Data for 2006. 2009;vol 57(14) (Report NVS) [PubMed]
  20. Herttua K, Makela P, Martikainen P. Changes in Alcohol-Related Mortality and its Socioeconomic Differences After a Large Reduction in Alcohol Prices: A Natural Experiment Based on Regular Data. American Journal of Epidemiology. 2008;168(10):1110–1118. doi: 10.1093/aje/kwn216. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Hingson R, Heeren T, Winter MR, Wechsler H. Early age of first drunkenness as a factor in college students' unplanned and unprotected sex attributable to drinking. Pediatrics. 2003;111(1):34–41. doi: 10.1542/peds.111.1.34. [DOI] [PubMed] [Google Scholar]
  22. Inoue M, Tsugane S. Impact of alcohol drinking on total cancer risk: data from a large-scale population-based cohort study in Japan. British Journal of Cancer. 2005;92(1):182–187. doi: 10.1038/sj.bjc.6602277. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Koski A, Reino S, Vuori E, Poikolainen K. Alcohol tax cuts and increase in alcohol-positive sudden deaths- a time-series intervention analysis. Addiction. 2007;102:362–368. doi: 10.1111/j.1360-0443.2006.01715.x. [DOI] [PubMed] [Google Scholar]
  24. LaVallee RA, Williams GD, Yi H. [Accessed January 21, 2010];Apparent Per Capita Alcohol Consumption: National, State, and Regional Trends, 1970–2007. Available at: http://www.niaaa.nih.gov/Resources/DatabaseResources/QuickFacts/AlcoholSales/
  25. Mangione TW, Howland J, Amick B, Cote J, Lee M, Bell N, Levine S. Employee drinking practices and work performance. Journal of Studies on Alcohol. 1999;60:261–270. doi: 10.15288/jsa.1999.60.261. [DOI] [PubMed] [Google Scholar]
  26. Markowitz S. The price of alcohol, wife abuse, and husband abuse. Southern Economic Journal. 2000;67(2):279–303. [Google Scholar]
  27. Markowitz S. Alcohol, drugs and violent crime. International Review of Law and Economics. 2005;25(1):20–44. [Google Scholar]
  28. Markowitz S, Chatterji P, Kaestner R. Estimating the impact of alcohol policies on youth suicides. The Journal of Mental Health Policy and Economics. 2003;6:37–46. [PubMed] [Google Scholar]
  29. Markowitz S, Grossman M. Alcohol regulation and domestic violence towards children. Contemporary Economic Policy. 1998;16(3):309–320. [Google Scholar]
  30. Markowitz S, Grossman M. The effects of beer taxes on physical child abuse. Journal of Health Economics. 2000;19(2):271–282. doi: 10.1016/s0167-6296(99)00025-9. [DOI] [PubMed] [Google Scholar]
  31. McCluskey CP, Krohn MD, Lizotte AJ, Rodriguez ML. Early substance use and school achievement: an examination of Latino, White and African American youth. Journal of Drug Issues. 2002;32(3):921–944. [Google Scholar]
  32. Midanik LT, Chaloupka F, Saitz R, Toomey TL, Fellows JL, Landen VM, et al. Alcohol-attributable deaths and years of potential life lost-United States, 2001. MMWR. 2004;53:866–870. [PubMed] [Google Scholar]
  33. National Highway Traffic Safety Administration. Traffic Safety Facts 2005 Data. Washington, D.C: 2005. [Google Scholar]
  34. NIAAA. [Accessed January 25, 2010];Alcohol across the Lifespan. 2006 Available at: http://pubs.niaaa.nih.gov/publications/StrategicPlan/NIAAASTRATEGICPLAN.htm.
  35. Ponicki WR, Gruenewald PJ. The impact of alcohol taxation on liver cirrhosis mortality. Journal of Studies on Alcohol. 2006;67(6):934–938. doi: 10.15288/jsa.2006.67.934. [DOI] [PubMed] [Google Scholar]
  36. Raftery AE. A note on Bayes factors for log-linear contingency table models with vague prior information. Journal of the Royal Statistical Society Series B (Methodological) 1986;48(2):249–250. [Google Scholar]
  37. Rehm J, Gmel G, Sempos CT, Trevisan M. Alcohol-related morbidity and mortality. Alcohol Research & Health. 2003;27:39–51. [PMC free article] [PubMed] [Google Scholar]
  38. Room R, Babor T, Rehm J. Alcohol and public health. The Lancet. 2005;365:519–530. doi: 10.1016/S0140-6736(05)17870-2. [DOI] [PubMed] [Google Scholar]
  39. Rush B, Steinberg M, Brook R. Relationships among alcohol availability, alcohol consumption and alcohol-related damage in the province of Ontario and the State of Michigan. Advances in Substance Abuse. 1986;5(4):33–45. doi: 10.1300/J251v05n04_03. [DOI] [PubMed] [Google Scholar]
  40. SAS Institute. SAS/STAT 9.1 User's Guide. Cary, NC: SAS Institute, Inc.; 2004. [Google Scholar]
  41. Schweitzer SO, Intriligator MD, Salehi H. Alcoholism: an econometric model of its causes, its effects, and its control. In: Grant M, Williams A, Plant M, editors. Economics and Alcohol. London: Croom Helm; 1983. pp. 107–127. [Google Scholar]
  42. Skog OJ, Melberg HO. Becker's rational addiction theory: an empirical test with price elasticities for distilled spirits in Denmark 1911–31. Addiction. 2006;101(10):1444–1450. doi: 10.1111/j.1360-0443.2006.01551.x. [DOI] [PubMed] [Google Scholar]
  43. Sloan FA, Reilly BA, Schenzler C. Effects of prices, civil and criminal sanctions, and law-enforcement on alcohol-related mortality. Journal of Studies on Alcohol. 1994;55(4):454–465. doi: 10.15288/jsa.1994.55.454. [DOI] [PubMed] [Google Scholar]
  44. Smart RG, Mann RE. Treatment, Alcoholics Anonymous and alcohol controls during the decrease in alcohol problems in Alberta: 1975–1993. Alcohol and Alcoholism. 1998;33(3):265–272. doi: 10.1093/oxfordjournals.alcalc.a008390. [DOI] [PubMed] [Google Scholar]
  45. Smith G, Branas C, Miller T. Fatal nontraffic injuries involving alcohol: a meta-analysis. Annals of Emergency Medicine. 1999;33:659–668. [PubMed] [Google Scholar]
  46. U.S. Census Bureau. [Accessed December 19, 2009];State and County Quick Facts. 2009 Available at: http://quickfacts.census.gov/qfd/states/
  47. Wagenaar AC, Maldonado-Molina MM, Wagenaar BH. Effects of alcohol tax increases on disease mortality in Alaska: Time-series analyses from 1976 to 2004. American Journal of Public Health. 2009a;99:1–8. doi: 10.2105/AJPH.2007.131326. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Wagenaar AC, Salois MJ, Komro KA. Effects of beverage alcohol price and tax levels on drinking: a meta-analysis of 1003 estimates from 112 studies. Addiction. 2009b;104(2):179–190. doi: 10.1111/j.1360-0443.2008.02438.x. [DOI] [PubMed] [Google Scholar]
  49. Wagenaar AC, Tobler A, Komro K. Effects of alcohol tax and price policies on morbidity and mortality: A systematic review. American Journal of Public Health. doi: 10.2105/AJPH.2009.186007. (in press) [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Yamasaki A, Chinami M, Suzuki M, Kaneko Y, Fujita D, Shirakawa T. Tobacco and alcohol tax relationships with suicide in Switzerland. Psychological Reports. 2005;97(1):213–216. doi: 10.2466/pr0.97.1.213-216. [DOI] [PubMed] [Google Scholar]

RESOURCES