Survey Estimates of Changes in Alcohol Use Patterns Following the 2012 Privatization of the Washington Liquor Monopoly

Abstract

Aims

The US state of Washington’s 333 state-run liquor stores were privatized on 1 June 2012 and purchases began in ~1500 licensed stores of a variety of types. A regime of taxes and fees was implemented to replace the revenues generated by the state stores and, 1 year later, the beer tax was reduced by two thirds. This study evaluates the impact of these changes on total alcohol and spirits consumption in a retrospective pre-test design.

Methods

The study sample consists of 2289 adults recruited in three cross-sectional surveys during 2014 and 2015. Retrospective typical past month quantity–frequency measures for before privatization drinking and current past month quantity–frequency measures were compared within subjects, for all alcohol and for spirits only.

Results

No change in alcohol volume was seen across privatization while spirits volume was found to decrease, suggesting a shift from spirits to beer. This decline in spirits volume came from a reduction in drinking days while overall drinking days were found to increase. This was offset by a reduction in drinks per drinking day and in heavy occasions.

Conclusions

These findings accurately mirror the overall flat trend in per capita alcohol sales but seem to exaggerate the very small shift towards beer seen in sales data. Effects of increased spirits availability appear to have been countered by increased spirits prices and a decreased beer tax, leading to a shift to beer consumption.

Short summary

Survey-based analyses of alcohol use across Washington’s spirits privatization, beer tax reduction and marijuana legalization found no change in alcohol volume, a reduction in spirits volume and a shift to more moderate drinking patterns. Reductions in drinking occurred among marijuana users and those with lower educational attainment reduced spirits volume.

INTRODUCTION

Following the passage of Initiative 1183, Washington’s 333 state-run liquor stores were privatized in June of 2012, when purchases began in ~1500 licensed stores of a variety of types including supermarkets, drugstores and wholesale clubs. These changes increase expanded hours of sale and placed liquor in the same stores as beer, groceries and other products. Based on I-1183, a regime of taxes and fees was implemented to replace the revenues generated by the state stores. These taxes included fees equal to 10% of the wholesale price (paid by the distributor) plus 17% of the retail price (paid by the retailer), as well as taxes on top of the retail price of 20.5% plus $2.83 per 750 mL bottle. The impact of these was that prices rose by 15.5% for a six brand index of 750 mL containers and 5.5% for a five brand index of 1.75 L containers (Kerr et al., 2015), with wide variation by brand and store type, resulting in a price range from an over 25% reduction to an over 60% increase. The stability of liquor prices in the neighboring states of Oregon and Idaho suggest that these changes were due to the privatization. While there is strong evidence that higher prices are associated with reduced alcohol consumption and related outcomes, they also spur efforts by drinkers to avoid reducing consumption such as quality downgrading, context shifting and cross-border purchasing (Fitzgerald and Mulford, 1993; Ruhm et al., 2011; Xu and Chaloupka, 2011). Our analysis of cross-border purchasing following privatization found that while significantly increased purchasing could be detected in both Oregon and Idaho, the volume of these was very small, ~0.25% of yearly spirits sales in Washington (Ye and Kerr, 2016). Meta-analyses from 112 studies confirm that there are negative relationships between alcoholic beverage prices and alcohol consumption, including heavy drinking (Wagenaar et al., 2009). That study’s findings include a higher price elasticity estimate for spirits, −0.80, while beer is less responsive at −0.46. This could suggest that spirits drinkers are more willing or able to reduce consumption or shift to other beverage types.

Evaluation of the effects of the privatization on alcohol consumption patterns is further complicated by two other significant policy changes in Washington. First, a temporary beer tax increase from $8.08 to $23.58 per 31-gallon barrel that had been enacted on 1 July 2010 expired on 30 June 2103, a year after spirits privatization. Second, voter-approved Initiative 502 legalized marijuana for recreational use in the November 2012 election, leading to retail marijuana stores opening in July of 2014. While the beer tax reduction in 2013 would be expected to increase demand for beer and potentially shift sales from spirits to beer, the influence of the legalization of marijuana on spirits is less clear. Although most marijuana users are also drinkers, the literature regarding relationships between marijuana and alcohol is mixed, with findings of both substitution and complementarity (Subbaraman, 2016).

Per capita apparent consumption based on official statistics for 2009–2015 are presented in Fig. 1 (Haughwout et al., 2016; Haughwout and Slater, 2017). Reduced beer consumption in 2010 and 2011, following the beer tax increase, was the main source of the initial decline in alcohol sales. Beer intake then rose from following the 2013 beer tax reduction, while wine intake was generally flat. Spirits consumption also showed little change between 2012 and 2015. This pattern of rising per capita consumption with a rising share of beer in Washington contrasts with a pattern of consumption shifting from beer to spirits for the USA as a whole during these years. Trends in per capita consumption for the bordering states of Oregon and Idaho also show rising consumption from 2010 to 2014 with a shift toward wine and spirits.

Fig. 1.

The trends in the per capita apparent consumption of beer, spirits, wine and total ethanol for the 2009–2015 period and a timeline of key alcohol and marijuana policy changes and the implementation of these as well as the fielding dates of the three surveys utilized in the analyses are shown.

This study considers the effects of the liquor privatization and other policy changes utilizing self-reported alcohol measures from general population surveys of Washington conducted in 2014 and 2015. Surveys offer an alternative view of changes in drinking that enables the identification of changes in drinking patterns and across socio-demographic and behavioral sub-groups. Retrospective measures of drinking in 2012 prior to the 1 July privatization are compared to the same measures of current drinking both group-wise and within subjects utilizing a retrospective pre-test design (Pratt et al., 2000; Hill and Betz, 2005; Sibthorp et al., 2007). Privatization effects on measures of all alcohol and spirits specifically are tested and changes in these were modeled to evaluate associations with drinker characteristics including income, education and marijuana use status.

METHODS

Sample

The study sample consists of 2289 adult (aged 18 and over) residents of Washington State. Respondents were recruited in three cross-sectional surveys (part of a series of six surveys) using random digit dial sampling, with more than 40% of cases acquired from cell phone exchanges. Here, we utilize data from Wave II (August–October 2014), Wave III (March–June 2015) and Wave IV (August–October 2015). ICF Macro, Inc., of Burlington, Vermont conducted the surveys with the following AAPOR2 cooperation rates for landlines and cell phones, respectively: Wave II (45.8%, 62.4%), Wave III (43.7%, 61.5%) and Wave IV (41.7%, 59.6%). Respondents received a $10 Amazon gift card upon completion of the survey. The Public Health Institute’s Institutional Review Board (I13-010) approved this study.

Measures

Current alcohol and spirits use

Drinking status (lifetime abstainer, past drinker, current drinker) was determined based on a series of questions regarding past year drinking frequency follow-up questions clarifying lifetime abstention or former drinker status. Respondents reporting at least one drink within the past year were classified as current drinkers. Current alcohol frequency (0–30 days) was determined by the question: ‘Considering all types of alcoholic beverages, on how many days during the past month, that is the past 30 days, did you have at least one drink of any alcoholic beverage?’ Current usual quantity (0–70 drinks/day) was determined using the question: ‘On the days that you drank in the past 30 days, how many drinks did you drink on average?’ Additionally, we assessed frequency of heavy drinking (5+ drinks on an occasion). Current spirits drinkers (yes/no) were determined using the following question: ‘How often do you usually have drinks containing liquor, including scotch, bourbon, gin, vodka, rum and so on?’ Respondents reporting such drinking at any time in the past year were classified as spirits drinkers. Similarly to the measures for current alcohol use, frequency (0–30 days) and quantity (0–70 drinks/day) of spirits use within the past 30 days was assessed for spirits drinkers.

Retrospective alcohol and spirits use

Retrospective alcohol frequency (0–30 days), before privatization, was determined using the question: ‘Thinking about a typical month around this time of year 2 years ago, before the privatization of liquor stores; considering all types of alcoholic beverages, on how many days during a typical month did you have at least one drink of any alcoholic beverage?’ We determined retrospective alcohol quantity (0–70 drinks/day) for respondents reporting one or more drinking days: ‘On the days that you drank, how many drinks did you drink on average?’ Retrospective heavy episodic drinking days (0–30 days) was determined with the question: ‘Considering all types of alcoholic beverages, on how many days in a typical month did you have 5 or more drinks on an occasion?’ Similarly, retrospective frequency of spirits use was determined with the following question: ‘Now, thinking of liquor drinking only. On how many days in a typical month did you have at least one drink of liquor?’ We determined retrospective spirits quantity (0–70 drinks/day) for respondents with one or more retrospective spirits drinking days: ‘On the days that you drank liquor, how many liquor drinks did you drink on average?’

Demographics

We used the following personal characteristics to examine and predict change in alcohol and spirits consumption: sex, age (18–29, 30–39, 40–49, 50–64, 65+), race/ethnicity (White, Black, Hispanic, Other), education (<high school, high school, some college, college, graduate school) and household income (<$10,000, $10,001–$20,000, $20,001–$40,000, $40,001–$60,000, $60,001–$80,000, $80,001 $100,000, $100,001+). Additional explanatory variables included marijuana use and voting on the privatization initiative. Current marijuana use (yes/no) was determined using the question ‘How often have you used marijuana, hash or pot during the last 12 months?’ Current users were classified as weekly users (ever day or nearly every day, and about once a week), monthly users (once every 2 or 3 weeks and once every month or 2) and yearly users (less often than that). Voting on privatization was determined using: ‘Did you vote for or against proposition 1183 privatizing liquor sales in 2012?’ Response options included not voting.

Analyses

Bivariate Chi-squared- and t-tests as appropriate were conducted to test for differences in demographics, marijuana use, support of privatization and drinking status (drinkers, spirits drinkers and abstainer). We calculated alcohol and spirits volume by multiplying the frequency by quantity measures for both current (after privatization) and retrospective (before privatization) periods. To evaluate changes across privatization, we excluded respondents missing on either current or retrospective volume for alcohol or spirits, resulting in a total sample of 2153 respondents (94%).

Wave II and Wave IV surveys, respectively, were administered during the fall months of 2014 and 2015 (August–October), at which point respondents were asked to recall their alcohol and spirits in a typical month around the same time of year before privatization (implemented on 1 June 2012). Although respondents were instructed to report alcohol and spirits use before privatization, the ‘around the same time of year’, if taken literally, referred to a period after privatization. These conflicting instructions could have led some respondents to report post-privatization drinking. For this reason, differences in mean alcohol and spirits volumes, before and after privatization, were examined separately for Wave III (March to June, 2015), for which the instructions did not conflict, in order to determine whether the results for this survey were consistent with those from analyses that pooled Waves II and IV. For the combined surveys and Wave III only, we also examined differences in mean volume across demographics, marijuana use and voting; as well as mean number of drinking days (frequency) and drinks per day (quantity) for both total and spirits, and mean number of heavy drinking days (5+ drinks per day).

Noting generally similar results for Wave III only, we focus on the combined survey results and utilize the combined surveys in linear regressions to identify characteristics associated with changes in alcohol and spirits volume. The dependent variables in these models was alcohol (spirits) volume in the past month minus typical monthly alcohol (spirits) volume prior to privatization so that positive values represent increases and negative values represent decreases. These variables were not normally distributed due to large numbers of zeros and small change values but were otherwise not overly skewed or influenced by outliers and model results should be robust (Hayes, 2013). Data were weighted using sampling weights to adjust for probability of selection introduced through the sampling design. All analyses were performed using STATA version 14.

RESULTS

Demographics, marijuana use, and support of privatization for each wave are presented in Table 1. Also included are bivariate tests of differences across survey waves and drinking status. Results indicated no significant differences between surveys on any of the variables. Analyses focus on drinkers and significant differences in demographic measures, marijuana use and support for privatization were found between drinkers and non-drinkers (including both lifetime-abstainers and past drinkers). Table 2 presents the mean alcohol and spirits volume before and after privatization. A small non-significant decrease in average alcohol volume from 12.49 before privatization to 12.17 drinks per month was observed. The only group showing a significant decrease in alcohol volume were current marijuana users. There were also significant changes in alcohol intake patterns across privatization, with a decrease in drinks per day and heavy drinking days, and an increase in the number of drinking days. These changes indicate a shift to more moderate drinking patterns. Spirits volume was found to have decreased significantly overall from 6.43 to 4.10 drinks per month (P < 0.01). Significant reductions across privatization were found for both genders, the younger 18–39 age group, for the White race/ethnicity group and for both marijuana users and non-users. Changes in the pattern of spirits use were also found: Frequency of use (drinking days) decreased, whereas quantity of use (drinks per day) increased. Overall, spirits volume decreased from 6.42 before privatization to 4.10 drinks after privatization (P < 0.01). Results for Wave III only are highly consistent in terms of the direction and magnitude of changes across privatization but the level of drinking in both time periods is lower and changes were not significant, with the exception of a reduction in total volume in the 18–39 age group. The lower consumption in Wave III (Spring) compared to Waves II and IV (late Summer–Fall) may be attributed to seasonal variation in drinking.

Table 1.

Demographic and drinking status of Washington states residents

	Total sample			Drinkers (N = 1603) (%)	Non-drinkers (N = 686) (%)
	Wave II (N = 804) (%)	Wave III (N = 823) (%)	Wave IV (N = 662) (%)
Sex
Female	50.4	50.4	50.4	49.4	52.9
Male	49.6	49.6	49.6	50.7	47.1
Age
18–29	22.4	20.7	21.8	22.8	18.8
30–39	16.2	19.5	16.5	18.9	13.6a
40–49	17.1	13.7	17.3	16.0	16.1
50–59	18.1	18.7	15.3	17.9	15.9
60+	24.7	25.3	26.8	22.4	33.4a
Missing	1.6	2.1	2.4	2.0	2.2
Race/ethnicity
White	75.1	75.5	74.4	75.9	72.8
Black	4.8	4.8	3.4	4.0	5.2
Hispanic	9.5	9.5	9.7	9.9	8.8
Other	10.7	10.2	12.5	10.3	13.2
Education
<High school	10.5	9.6	9.3	8.8	12.0
High school	24.8	23.5	25.2	21.9	30.8a
Some college	33.7	35.4	33.5	34.8	32.7
College	16.0	16.4	17.2	18.1	12.7a
Graduate school	14.8	14.9	14.5	16.2	11.1a
Missing	0.3	0.2	0.5	0.2	0.8
Any marijuana use
No	72.7	75.8	71.1	67.1	87.6a
Yes	26.8	23.8	28.4	32.3	12.2a
Missing	0.5	0.5	0.5	0.6	0.2
Vote 1183
For	32.5	32.2	33.1	37.7	20.4a
Against	25.3	28.9	25.7	25.9	28.5
Did not vote	31.8	30.0	30.5	28.1	37.0a
Missing	10.4	8.9	10.8	8.3	14.1a

^aSignificant differences between drinkers and non-drinkers at the P < 0.05 level.

Table 2.

Mean alcohol and spirits volume (drinks per month) before (retrospective) and after (current) privatization in Washington

Mean (Std err)	Alcohol volume				Spirits volume
	Retrospective		Current		Retrospective		Current
Total sample (N = 2153)
Total	12.49	(1.03)	12.17	(1.21)	6.42	(0.91)	4.10a	(0.44)
Sex
Female	8.58	(0.80)	7.34	(0.59)	3.92	(0.57)	2.26a	(0.26)
Male	16.50	(1.89)	17.10	(2.35)	8.98	(1.73)	5.98a	(0.84)
Age
18–39	17.01	(2.31)	14.33	(2.52)	9.94	(2.09)	5.47a	(1.04)
40+	9.80	(0.81)	10.97	(1.20)	4.24	(0.66)	3.25	(0.30)
Race/ethnicity
White	11.77	(0.81)	11.01	(0.95)	5.15	(0.46)	3.48a	(0.28)
Non-White	14.73	(3.37)	15.77	(3.99)	10.35	(3.42)	6.00	(1.57)
Marijuana use
No	8.01	(0.64)	9.05	(0.99)	3.82	(0.52)	2.72a	(0.27)
Yes	25.18	(3.35)	20.87a	(3.65)	13.82	(3.10)	7.77a	(1.46)
Drinking days	4.68	(0.22)	5.04a	(0.22)	2.29	(0.15)	1.67a	(0.10)
Drinks per day	1.30	(0.06)	1.19a	(0.06)	0.95	(0.05)	1.09a	(0.06)
5+ Drinks per day	0.72	(0.11)	0.46a	(0.07)
Wave III sample (N = 783)
Total	9.68	(1.09)	9.38	(0.86)	4.29	(0.53)	3.47	(0.48)
Sex
Female	5.36	(0.69)	5.83	(0.64)	2.32	(0.33)	2.43	(0.55)
Male	14.14	(2.04)	13.05	(1.57)	6.33	(1.01)	4.55	(0.80)
Age
18–39	12.32	(2.30)	8.63a	(1.39)	5.32	(0.97)	3.41	(0.89)
40+	8.00	(0.99)	9.93	(1.13)	3.55	(0.62)	3.46	(0.55)
Race/ethnicity
White	10.33	(1.28)	9.62	(0.89)	4.18	(0.54)	3.61	(0.59)
Non-White	7.50	(1.98)	8.57	(2.27)	4.68	(1.45)	2.99	(0.67)
Marijuana use
No	6.51	(0.96)	7.50	(0.86)	2.79	(0.39)	2.73	(0.51)
Yes	19.74	(3.20)	15.08	(2.28)	8.99	(1.81)	5.77	(1.23)
Drinking days	3.87	(0.31)	4.54	(0.32)	1.83	(0.17)	1.50	(0.15)
Drinks per day	1.10	(0.07)	1.12	(0.09)	0.84	(0.06)	1.04	(0.07)
5+ Drinks per day	0.65	(0.11)	0.32	(0.07)

^aSignificant differences between retrospective and current volume at the P < 0.05 level.

The results of the regression models predicting individual characteristics of the degree of change in (separately) alcohol and spirits volume are presented in Table 3. In interpreting the presented coefficients, it should be noted that both models had negative intercepts so that positive coefficients may identify groups that did not reduce alcohol or spirits volume across privatization, rather than indicating an increase. Regarding changes in alcohol volume, education (graduate school vs. high school), very low and high incomes (compared to those with a household income of $40,000–$60,000) were positively associated with alcohol intake after privatization. Any past year marijuana use was significantly negatively associated with alcohol volume (β = −4.91, P < 0.05) compared to non-users (model not shown). However, marijuana use frequency categories were not significantly associated when classified in three groups, nor was there a clear pattern across these. Models predicting changes in spirits volume revealed that education (some college or more vs. high school) was positively associated with change in spirits intake after privatization, as was having not voted on Initiative-1183.

Table 3.

Results of regression analyses on changes in alcohol and spirits volume (current drinks per month minus pre-privatization drinks per month) after privatization in Washington

	Alcohol			Spirits
	b	95% CI	P-value	b	95% CI	P-value
Sex (Ref female)
Male	2.05	(−0.75, 4.84)	0.15	−0.69	(−3.33, 1.95)	0.61
Age (Ref 18–29)
30–39	0.70	(−5.44, 6.84)	0.82	−3.49	(−10.06, 3.09)	0.30
40–49	4.29	(−2.44, 11.03)	0.21	2.41	(−1.52, 6.35)	0.23
50–59	4.49	(−0.36, 9.35)	0.07	0.79	(−3.88, 5.47)	0.74
60+	4.64	(−0.04, 9.32)	0.05	1.60	(−2.16, 5.36)	0.40
Missing	4.54	(−0.42, 9.50)	0.07	1.60	(−2.43, 5.63)	0.44
Race/ethnicity (Ref White)
Black	−5.17	(−14.08, 3.74)	0.26	−4.44	(−13.21, 4.34)	0.32
Hispanic	5.13	(−2.08, 12.35)	0.16	1.12	(−7.51, 9.75)	0.80
Other	2.46	(−2.34, 7.26)	0.32	−4.31	(−11.20, 2.58)	0.22
Education (Ref HS)
<High school	6.07	(−5.13, 17.27)	0.29	0.59	(−7.61, 8.79)	0.89
Some college	3.50	(−0.37, 7.36)	0.08	4.86	(0.58, 9.14)	0.03
College	2.90	(−1.78, 7.57)	0.23	5.02	(0.34, 9.70)	0.04
Graduate school	5.18	(0.75, 9.62)	0.02	6.62	(1.70, 11.53)	0.01
Income (Ref $40–60k)
<$10k	7.27	(0.92, 13.61)	0.03	−2.76	(−10.84, 5.33)	0.50
$10,001–20k	1.08	(−3.65, 5.82)	0.65	2.66	(−1.05, 6.38)	0.16
$20,001–40k	5.77	(−1.36, 12.90)	0.11	1.25	(−2.76, 5.25)	0.54
$60,001–80k	3.47	(−1.12, 8.06)	0.14	0.69	(−2.79, 4.18)	0.70
$80,001–100k	4.39	(0.46, 8.33)	0.03	2.25	(−0.55, 5.05)	0.11
$100k+	2.82	(−1.84, 7.48)	0.24	−1.89	(−6.99, 3.21)	0.47
Marijuana use (Ref never)
Yearly	−4.45	(−10.77, 1.86)	0.17	−2.24	(−6.19, 1.72)	0.27
Monthly	−2.31	(−7.11, 2.49)	0.35	−0.34	(−4.07, 3.38)	0.86
Weekly	−6.13	(−13.10, 0.84)	0.09	−6.09	(−13.55, 1.36)	0.11
Vote 1183 (Ref for)
Against	−3.47	(−7.12, 0.19)	0.06	0.93	(−2.20, 4.07)	0.56
Did not vote	−0.12	(−5.72, 5.49)	0.97	4.16	(0.44, 7.88)	0.03
Time (Ref Wave II)
Wave III	−0.96	(−4.59, 2.67)	0.60	1.34	(−1.57, 4.25)	0.37
Wave IV	−2.10	(−6.19, 1.98)	0.31	−2.06	(−6.10, 1.97)	0.32
Intercept	−7.69	(−15.01, −0.38)	0.04	−6.41	(−14.73, 1.90)	0.13

Note: Bold indicates significance at the P < 0.05 level.

DISCUSSION

The effect of spirits privatization on alcohol consumption patterns is a significant public health policy issue on which relatively little data has been available. Evaluations of the few previous retail privatizations and studies of increases in alcohol availability strongly indicate a likelihood of increased alcohol use and related problems (Hahn et al., 2012). However, each situation has unique characteristics, and evaluation of the Washington privatization is complicated by other significant policy changes occurring in the following year. Our survey-based retrospective pre-test analyses found no change in alcohol volume across privatization. Spirits volume was found to decrease, suggesting a shift from spirits to beer, although beer use was not directly measured. This decline in spirits volume came from a reduction in spirits drinking days, while a small increase was seen in spirits drinks per drinking day. In contrast, overall drinking days were found to increase and this was offset by a reduction in drinks per drinking day. Importantly, a reduction in heavy drinking days was also found. Together, these results suggest a shift from spirits drinking occasions to beer and wine drinking occasions and, possibly, a removal of spirits from multiple beverage drinking days, resulting in fewer days having five or more drinks. Given the increase in spirits prices starting with privatization in 2012 and the decrease in beer prices with the tax reduction in 2013, the shift from spirits to beer is consistent with these price changes, despite the substantial expansion of spirits availability. These findings accurately mirror the overall flat trend in per capita alcohol sales but seem to exaggerate the very small shift towards beer seen in sales data.

Details of the characteristics of drinkers who may have been differentially impacted by the privatization and beer tax reduction can only be seen in survey data. Model results indicate that the reduction in spirits volume was largest in the lowest educational attainment groups, with a high school education or less. These groups did not show reductions in alcohol volume, indicating a shift from spirits to beer in response to relative and absolute price changes. Another important policy change occurring in Washington between privatization and the surveys was recreational marijuana legalization. Washington has one of the highest rates of past year marijuana use in the US, behind only Colorado, Vermont and Rhode Island (Hughes et al., 2016), and Table 1 shows that nearly one third of drinkers were past year users. These marijuana users were found to reduce both alcohol volume and spirits volume across privatization (and legalization). Models also found significantly reduced alcohol volume among users in relation to non-users controlling for other characteristics. These results suggest that there may have been a substitution relationship between alcohol and marijuana at the population level in Washington across legalization and privatization. Reviews of the substitution/complementarity literature support that more liberal marijuana policies may lead to less alcohol use in the general population, though the relationship complex and likely affected by demographics and other factors (Guttmannova et al., 2016; Subbaraman, 2016).

While these analyses did not find that liquor privatization was associated with increased alcohol use or heavy drinking, there still could have been increases in certain alcohol-related problems associated with privatization and increased spirits availability. An analysis of alcohol outlet locations and violent crime in Seattle over the 2010–2013 period, across privatization, found that the risk of aggravated assault increase by 8% with each additional off-premise outlet (Tab et al., 2016). While this did not include the expansion of spirits sales in existing outlets associated with privatization, it raises significant concerns regarding alcohol availability in Washington. A study utilizing a different survey from the same series utilized here considered the opinions of Washington voters regarding privatization (Subbaraman and Kerr, 2016). Results indicate that enough of those who supported privatization would in retrospect have changed their votes, and Initiative 1183 would have failed based on their current knowledge of privatizations effects. Voter regret was found to be associated with endorsing a reduction in the number of store selling liquor, suggesting some dissatisfaction with expanded availability.

Limitations of these analyses, in addition to the multiple policy changes, include the general tendency towards under-reporting of alcohol volume in surveys and reporting issues specific to retrospective recall of alcohol use from 2 or 3 years prior to interview (Kerr and Greenfield, 2007; Greenfield and Kerr, 2008). The issue of conflicting instructions regarding the recall time period for Waves II and IV also complicate this limitation. While retrospective recall is a potentially significant issue, we note that results are generally in line with per capita consumption figures and that the salience of the privatization for spirits purchasing could aid in recall of spirits use in the prior period. If retrospective reports were biased downwards declining drinking would be suggested.

Although the combination of major policy changes occurring between the reference period in early 2012 and the follow-up measures from 2014 to 2015 complicates the interpretation of our results, the analyses represent an opportunity to consider the impacts of multiple policy changes taken together. Importantly, differential trends in per capita sales of spirits and beer in Washington, as compared to national and neighboring state trends, suggest that the price effects from tax changes by beverage type shifted drinkers from spirits to beer despite greatly increased availability of spirits. This was in marked contrast to the national trend away from beer toward spirits and wine over the same period. Our results also show this beverage type shift in the context of no change in total alcohol volume, also indicating that lower educational attainment and income drinkers were more responsive to these price factors. We were also able to identify changes in alcohol use among marijuana users and non-users over across a period of increased marijuana availability involving recreational legalization and the opening of retail stores. Results suggest a substitution relationship between marijuana and alcohol volume in the context of recreational legalization. Finally, these results suggest that price factors played a more significant role in alcohol use changes that availability factors in this case, with the large beer tax reduction seeming to counteract reduced spirits drinking responding to increased prices induced by a privatization policy designed to maintain state revenues. Privatization of spirits elicited widespread public health concerns; however, in the event, price effects may have dominated and largely canceled out potential increases in spirits drinking from availability changes.

ACKNOWLEDGEMENTS

This research was supported by R01 AA021742 and National Alcohol Research Center Grant (P50AA005595) from the U.S. National Institute on Alcohol Abuse and Alcoholism and was approved by the Public Health Institute (IRB #I13-010).

FINANCIAL DISCLOSURE

Drs Kerr, Greenfield and Subbaraman and Ms Williams have received funding through contracts and travel support from the National Alcohol Beverage Control Association.

CONFLICTS OF INTEREST STATEMENT

None declared.