Better to Own or to Regulate? The Case of Alcohol Distribution and Sales

Abstract

Alcohol-related income generation is compared across monopoly and license off-premise alcohol regulatory models in U.S. states, 1977-2010. An optimum organizational-ownership mix is found when states directly own alcohol wholesale and employ a network of state-owned retailers serving urban regions and private agents serving less-populated regions. Per capita alcohol-related income in US dollars for these optimal systems was $58.82 compared with $26.72 with license systems. This disparity held after controlling for alcohol sales and retail hours of operation. The findings challenge the wisdom of asset divestment as a response to fiscal stress and contradict a central tenet of New Public Management (NPM) Theory.

Keywords

alcohol regulation privatization new public management public ownership asset divestment

New Public Management (NPM) Theory advocates governmental entrepreneurialism, defined as the use of markets (or quasi-markets) rather than bureaucracies for the provision of goods and services. Optimal public value is achieved by subjecting superiorly incentivized private enterprises to competitive conditions. In industries with sufficient private markets,government was to regulate and adjudicate. If the private market was incapable of producing enough of a desired good, necessitating public funding, government was instructed to practice “managed competition” by requiring multiple private bidders to vie for public contracts (Gansler, 2006; Lane, 2000; Lynn, 2006; Organisation for Economic Co-Operation and Development, 2005). Either way, the idea is for government to exit from the direct delivery of goods and services by transferring economic activity to private hands, that is, to steer the ship and let the private sector row (Osborne & Gaebler, 1992). Governmental entrepreneurialism thus became synonymous with privatization. This research tests the NPM conjecture that a reduction in state control over an industry increases public value.

The industry under study is alcohol distribution and sales. Ethyl alcohol is an addictive substance associated with numerous social harms, which is a reason why governments regulate alcohol traffic (Babor et al., 2010; Cook, 2007). Yet, although liquor law policy and enforcement are arguably inherently governmental functions, the same cannot be said of product distribution and sales. Verily, resilient private markets for alcohol doomed Prohibition. Nonetheless, since Prohibition, one method for marketing packaged products for off-premise consumption involves the public ownership of sales and distribution where, in the most comprehensive version, liquor laws are directly administered through purchases made in publicly owned retail outlets. Models where government owns all or part of an alcohol delivery system are referred to as “monopoly” or “control” systems. An alternative is to allow alcohol sales and distribution to occur through a regulated private market. In these latter models, referred to as “license” systems, the state role is limited to licensing and liquor law enforcement. These two models coexist in the U.S. economy, allowing for a natural test of whether public value is enhanced when government favors markets over direct public ownership.

In this research, “value” is measured by the per capita generation of income for public use. Other dimensions of value have been examined in the literature. For the consumer, alcohol system privatization does not appear to greatly affect product price (Siegel et al., 2012; Swidler, 1986; Trolldal & Ponicki, 2005; Zardkoohi & Sheer, 1984); however, access to ethyl alcohol becomes more convenient in license systems due to a greater number of retail outlets, less restrictive hours of retail operation, and more credit options to make purchases (Grubesic et al., 2012; Gruenewald, Madden, & Janes, 1992; Seim & Waldfogel, 2013; Trolldal & Ponicki, 2005). Monopoly systems, in contrast, feature larger stores stocked with a wider variety of products (Seim & Waldfogel, 2013), which might appeal to certain consumers.

At the societal level, alcohol system privatization is typically associated with negative outcomes that are caused by factors related to enhanced consumer access to ethyl alcohol. A large body of evidence associates privatization, and subsequent elevated alcohol consumption, with alcohol-related disease and illness (e.g., Campbell et al., 2009; Hahn et al., 2012; Her, Giesbrecht, Room, & Rehm, 1999; Norström et al., 2010; Stockwell et al., 2009). Although not directly assessed against system privatization, greater numbers of off-premise alcohol outlets are associated with higher crime rates (e.g., Gyimah-Brempon, 2001; Livingston, 2008; White, Gainey, & Triplett, 2012) and other undesirable behaviors (Wechsler, Lee, Hall, Wagenaar, & Lee, 2002).

Research into the effect of alcohol system privatization on state budgets is comparatively scarce despite the assertion that alcohol revenue may motivate state ownership (Room, 1987, 1993). Analyzing 1961 administrative data, Simon (1966) estimates the mean per capita revenue for monopoly states at US$6.55 versus US$3.22 for license states, for a gain of 103% attributed to state ownership. Using an alternative measure, revenue per gallon consumed, Simon estimates a revenue premium from state ownership of 152% (US$5.93 per gallon for monopolies versus US$2.35 per gallon for license systems). Mosher and Beauchamp (1983) report that monopoly states tax at higher percentage rates. Spellman and Jorgenson (1982) analyze 1976 data and calculate a per capita alcohol income of US$22.71 for monopoly states and US$19.70 for license states, but it is unclear whether these figures are limited to tax revenues and therefore exclude revenue from licenses and alcohol sales.

This research offers an updated test of the effect of alcohol system privatization on state budgets. Two questions are addressed. First, does the strength of state ownership over alcohol distribution and sales affect alcohol-related state income? If so, what is the optimal model of state ownership in terms of income generation? Our time-series cross-section analysis is based on statistics from the U.S. Census of Governments, 1977 to 2010.

Data and Measures

Alcohol Monopoly

Alcohol monopoly systems defy neat classification. Each state has a historical trajectory shaped by influential events, resulting in a diverse mix of policies that can only be fully appreciated by painting each state’s unique portrait. A careful review shows that not only does the scope of government control over alcohol distribution and sales vary but important financial and regulatory policies, such as licensing, taxes, hours of operation, level of local control, and so forth, lack uniformity as well. Even the term monopoly is imprecise, because all states allow for the off-premise sale of alcohol products by licensed retailers.¹ Any empirical analysis seeking to compare license versus monopoly states will undoubtedly have to settle for generalities that sacrifice this rich mosaic of state and local policy.

Taking that limitation in mind, the degree of state ownership was operationalized along two dimensions: first, by type of product under state control (i.e., spirits only or spirits plus wine); second, by the level of organizational ownership. Any amount of direct state ownership of product or organization classifies a state as monopoly. With this definition, there are 18 monopoly states in the United States during the time period of our analysis, 1977 to 2010. Data on the history of state control were collected through interviews with state agency representatives, an examination of legislative records, and compilations of law published by the Distilled Spirits Council of the United States (DISCUS; various years).

Product type

Consistent with the temperance legacy, all monopoly states control the most potent class of alcohol product, spirits. Divestment since 1977 has primarily involved wine, but as of 2010, five states still controlled wine, and one controlled spirits, wine, and beer. State variation exists over the classification of alcohol products. A few specify wine or spirits, but often the classification is by alcohol content, which depending on the statutory level may group fortified wine with distilled spirits. In this study, where the statutory alcohol percentage cutoff was greater than 14%, we assumed this was meant to include fortified wine and was coded as control over distilled spirits only.

Another variable coding issue was instances when a state divested from products gradually over time. For instance, Idaho switched to licensed wine sales in 1978, yet continued to carry popular wine products afterward in state stores as a courtesy to customers. The drawdown in wine from the shelves of Idaho state stores occurred over decades, until 2002, when Idaho removed all wine products from state store shelves except for Idaho-produced wine.² In such cases, the variable was coded as divestment beginning in the year when the state allowed licensed product sales (i.e., when the state monopoly of off-premise alcohol product ended).

Organizational ownership

A key operational distinction within monopoly states is whether control extends to the retail sector and, further, whether the retail operations are state stores or privately managed agency stores. All monopoly states control at the wholesale level,³ and as of 2010, 12 states exert control over retail, either by directly owning stores or through contracts with private agents. In a retail agency arrangement, the state sets the price of the product and owns the inventory, whereas private agents incur most overhead expenses and earn a commission on sales. Financially, the state rids itself of store operational expenses (e.g., labor, management, lease payments, and so forth) but pays a price by forgoing a share of gross revenue. In this study, agency retail is defined as an operation managed by a private concern, yet where the state either owns the inventory or sets product price.⁴ Figure 1 below plots the organizational trends for the state monopoly systems from 1997 to 2010.

Figure 1.

Organizational models of monopoly states, 1977-2010.

As Figure 1 illustrates, state ownership of wholesale and retail was the dominant monopoly model in the late 1970s. This model experienced dramatic change since then, dropping from 30% to 12% of states from 1977 to 2000. Movement away from retail ownership was roughly split between states that fully divested from retail and states that opted for retail agents. Much of the change happened during the early 1980s, when the recession led states to search for ways to fill budget gaps. Selling off the state stores or creating agency stores was perceived as a method of raising revenue or reducing state expenses. State monopoly models for 1977 and 2010 are outlined in the appendix.

Monopoly strength

A composite measure was created to assess the effect of the strength of the alcohol monopoly on alcohol-related revenues economic based on ownership of organization. For both spirits and wine, wholesale control is classified into three types—State Only, Agent, and License—whereas retail is classified into four types—State Only, State and Agency, Agency Only, and License. Values were assigned based on the matrix in Table 1.

Table 1.

Monopoly Strength Measure.

Retail ownership and control	Wholesale ownership and control
Retail ownership and control	State only	Private agent	Private license
State only	5 (290)^a [46]	4 (N/O) [N/O]	3 (N/O) [8]
State and private agent	4 (115) [34]	3 (N/O) [N/O]	2 (N/O) [N/O]
Private agent only	3 (74)^b [5]	2 (7) [N/O]	1 (N/O) [36]^c
Private license	2 (126) [118]^d	1 (N/O) [N/O]	0 (1,081) [1,454]

Note. For each cell, the top number is the assigned variable score for monopoly strength, the second number (in parentheses) is the number of observations for spirits, and the third number [in brackets] is the number of observations for wine. N/O indicates no observations.

Includes 34 observations from the North Carolina county stores.

Includes 5 observations where a state was transitioning from public into private license, and therefore had both types of retail stores.

All 36 observations are where the state retail stores held wine inventory during a transition into private license.

Includes 45 observations where the state stores carried a limited number of wine products as they phased out wine inventory.

The composite measures, Spirits Monopoly and Wine Monopoly, each have six possible values, from 0 to 5, where 0 is a license state and 5 represents state-owned wholesale and retail. Measurement values are in Table 1 along with observation counts for spirits and wine. Evident from the counts is that retail was the main area of state divestment illustrated in Figure 1.

State Finances

Alcohol-related income comes primarily from three sources: (a) the sale of alcohol products, (b) alcohol taxes, and (c) alcohol beverage licenses.⁵ Monopoly states earn income from all three sources; license states earn income from the second and third. Our source for finance data is the annual Census of Governments (COG), state-level data, years 1977 through 2010. The COG record revenues and expenses for alcoholic beverage distribution facilities and retail outlets are owned and operated by state governments using standard definitions across states and over time. Moreover, the COG collects information on state income from alcohol sales taxes and alcohol licensing. Income from the state operations, alcohol taxes, and alcohol licenses were combined into one measure to compare total alcohol-related income across monopoly and license states. Every state is included in the analysis.⁶

In the COG, net income measures income earned through the distribution and sales of alcohol products through state-controlled wholesale and retail outlets. Net income⁷ is defined as gross profit on sales⁸ minus operating expenses⁹ plus other income minus non-operating expenses.¹⁰ Standardization is achieved by dividing net income by the population of adults in a state that are older than 18 years. All figures are adjusted for inflation in 2010 dollars.

It is important to note that the COG figures are computed based on store operations only; excluded from the revenue side are sales and license taxes on alcoholic beverages collected through state stores and any state store profits that are earmarked for local governments. On the expense side, the COG excludes liquor law enforcement, the regulation of private on- and off-premise retail establishments, the collection of alcohol taxes and licenses, and any distribution of earnings to local governments. States with monopoly systems often include these revenues and expenses in financial statements to account for the wide range of roles performed by alcohol control boards. For our purposes, their removal from the calculation of store income improves the accuracy of the measure and allows for a comparison with license states that must administer tax collection, licensing, and law enforcement through other state departments.

Two other major alcohol-related sources of income from the COG statistics are alcohol taxes and alcohol beverage licensing.¹¹ To standardize the statistics across large and small states, alcohol tax and license income are expressed as a per capita statistics: alcohol taxes per capita, and alcohol license fees per capita, where per capita standardization is again the population of adults above 18 years of age. All figures were adjusted for inflation in 2010 dollars.

Three measures, net income per capita, alcohol taxes per capita, and alcohol license fees per capita, are combined to produce a composite statistic for comparing the income generating capacity of license and monopoly systems. The measure, alcohol income per capita, overcomes the problem of having to distinguish between product markup and alcohol tax for monopoly systems, because net income encompasses product markup. Similarly, an aggregate measure neutralizes differences in tax rates and license fees, as well as the potential inverse relationship between these two revenue sources. Figure 2 provides trends for the inflation-adjusted (2010 dollars) averages for the measures across the states.

Figure 2.

Alcohol-related income for all states, 1977-2010.

The top line in Figure 2, alcohol income per capita, is the sum of alcohol licenses, store income and alcohol taxes. In adjusted terms, total alcohol revenue per capita declined from US$70.49 to US$33.26 over the 1977 to 2010 period, with much of the decline happening during the inflationary era of the late 1970s. The decline in alcohol income per capita continued through the early 1980s recession period and stabilized around the year 2000. As illustrated, alcohol taxes comprise the largest share of total alcohol income across the states, but this share dropped from 75.7% to 70.5% over the period. In contrast, store net income as a share of total alcohol income grew from 19.6% to 24.0% for the period. License income also grew proportionately from 4.6% to 5.5% of total alcohol income. Relevant for the privatization debate, license income is a comparatively small fraction of total alcohol income and would, on average, have to increase by a factor of eight to offset the lost revenue from state divestment in the industry.

Separate statistics for monopoly and license states provides a more accurate understanding of the relative contributions made by store net income, alcohol taxes, and alcohol licenses. In 2010, store net income in monopoly jurisdictions comprised 46.3%, taxes were 47.8%, and alcohol licenses were 5.7% of the total alcohol income. For license states, nearly all the alcohol income is derived from taxes. Indeed, the two regulatory models are quite comparable in terms of alcohol tax and alcohol license income. In 2010, the per capita alcohol income from taxes was US$23.79 for license states and US$22.85 for monopoly states. That same year, license states averaged US$1.34 per capita from alcohol licenses whereas monopoly states averaged US$2.71. Alcohol taxes and licenses combined brought in per capita amounts of US$25.13 versus US$25.56 for license and monopoly states, respectively. Hence, the difference in alcohol income between the two regulatory models can almost be fully explained by whether the state owns the wholesale and retail industry.

Alcohol Sales

A factor that is expected to affect the magnitude of state alcohol-related revenue is alcohol sales. Alcohol sales data for spirits, wine, and beer were obtained from the Alcohol Policy Information System (APIS)¹² for years dating from 1977 to 2004. Data for the years 2004 to 2010 were provided by the Beverage Information Group and appended to the APIS data to complete the 1977 to 2010 data range. Measures for spirits, wine, and beer sales are standardized as gallons per capita for each combination of state and year, and expressed in natural log form.

Prohibited Hours and Days of Sale

One method of regulating alcohol traffic is to limit the times when it can be legally sold, either in on- or off-premise locations.¹³ Consistent with the APIS operationalization, we assess the level of restriction by the number of prohibited hours for Sunday and non-Sunday sales. Data on prohibited hours and days of sale were obtained from APIS for the years 1977 through 2005 and upgraded using the original APIS source, the state compilations published by the DISCUS. Remaining data gaps were filled by referencing the appropriate state statute.

Categorical variables were created for Sunday and non-Sunday off-premise retail policies using “low,” “medium,” and “high” designations. For Sunday sales, a state had “low” restrictions if the prohibited hours were 8 or less, “medium” restrictions if the prohibited hours were greater than 8 but less than 18, and “high” restrictions if the prohibited hours were 18 or greater. For non-Sunday sales, a “low” level of restrictions was 24 or less, “medium” was greater than 24 to 36, and “high” was greater than 36 prohibited hours. A final categorical variable was created for states that allow local governments the option of setting retail hours. Such options are usually exercised by urban municipalities seeking to liberalize store hours. Variables, definitions, and descriptive statistics are listed in Table 2.

Table 2.

Variables, Definitions, and Descriptive Statistics (N = 1,659).

Variable	Definition	M	SD
Alcohol income per capita (natural log)	Inflation-adjusted sum of store net income, alcohol taxes, and license income divided by the state adult population.	3.541	0.612
Spirits monopoly	6-point scale (0 to 5) for the degree of state spirits control.	1.356	1.954
Wine monopoly	Six point scale (0 to 5) for the degree of state wine control.	0.386	1.086
Spirits sales per capita (natural log)	Annual gallons of liquor sold divided by the state adult population.	0.715	0.338
Wine sales per capita (natural log)	Annual gallons of wine sold divided by the state adult population.	0.827	0.508
Sunday hours off-premise high	Indicator variable (yes = 1) if prohibited Sunday hours are 18 or greater.	0.605	0.489
Sunday hours off-premise medium	Indicator variable (yes = 1) if prohibited Sunday hours were greater than 8 but less than 18	0.209	0.406
Sunday hours off-premise low	Indicator variable (yes = 1) if prohibited Sunday hours were 8 or less.	0.186	0.389
Non-Sunday hours off-premise high	Indicator variable (yes = 1) if prohibited non-Sunday hours was greater than 36.	0.411	0.492
Non-Sunday hours off-premise medium	Indicator variable (yes = 1) if prohibited non-Sunday hours was greater than 24 to 36.	0.313	0.467
Non-Sunday hours off-premise low	Indicator variable (yes = 1) if prohibited non-Sunday hours was 24 or less.	0.267	0.443
Local hours option	Indicator variable (yes = 1) if municipalities can establish retail hours.	0.420	0.494
Year	Ordinal year count, 1977 = 0	16.556	9.794
Year-squared/100	Year count squared divided by 100	3.700	3.350

Alcohol Income Estimates

The association between measures of alcohol monopoly and the financial measures are analyzed using multi-level regression based on growth curve modeling as described by Rabe-Hesketh and Skrondal (2008). A main advantage of this approach is the ability to model within state random error to control for unmeasured factors that explain variation in the dependent variable. The general formula is as follows:

\begin{array}{l} Y_{i t} = β_{0} + β_{1} (Regulation) + β_{2} (Year) + β_{3} ({Year}^{2}) \\ + μ_{i 1} + μ_{i 2} (Year) + μ_{i 3} ({Year}^{2}) + ε_{i t}, \end{array}

where Y_it is the dependent variable of interest, alcohol income per capita, expressed in natural log form for state i and year t; β₀ is the sample intercept; β₁ is the coefficient for state monopoly measure or any other alcohol-related state regulation (Regulation) tested; β₂ and β₃ are coefficients for linear (Year) and curvilinear (Year-Squared) trends, respectively. The time trend factors out national change in the dependent variable depicted in Figure 2. The µ_i symbols are random components for the states, and symbol ϵ_it is the unexplained sample error.

By itself, µ_i1 are random state intercepts; as coefficients for time trend variables Year (µ_i2) and Year-Squared (µ_i3), they are random linear and quadratic state slopes, respectively. By including state intercepts and curvilinear state slopes, the intent is to factor out unmeasured state attributes or policy changes that might alternatively explain variation in the dependent variable. Variable intercepts adjust the findings for state traits that are stable over the time period, whereas variable slopes adjust for the trajectories of the dependent variable for each state.

Table 3 provides the regression results for net alcohol income per capita.Models 1 through 4 progressively introduce independent variables. Model 1 is a base that includes only the time trends. Consistent with Figure 2, alcohol-related income in real terms has declined over the period, but the rate of decline has slowed. Alcohol-related income for all states reached an average nadir at around 2005 and has since stabilized.

Table 3.

Regression Estimates for Alcohol Income Per Capita, 1977 to 2011.

Independent variables	Model 1	Model 2	Model 3	Model 4
Spirits monopoly		0.381*** (0.049)	0.380*** (0.050)	0.380*** (0.049)
Spirits monopoly squared		−0.046*** (0.007)	−0.046*** (0.007)	−0.046*** (0.007)
Wine monopoly		0.013 (0.014)	0.017 (0.014)	0.018 (0.014)
Spirits sales			−0.058 (0.061)	−0.051 (0.060)
Wine sales			0.035 (0.032)	0.036 (0.032)
Sunday hours off-premise high				0.005 (0.029)
Sunday hours off-premise medium				−0.039 (0.030)
Non-Sunday hours off-premise high				0.136*** (0.038)
Non-Sunday hours off-premise medium				0.125*** (0.035)
Local hours option				−0.030 (0.019)
Year	−0.058*** (0.006)	−0.057*** (0.006)	−0.060*** (0.006)	−0.060*** (0.006)
Year-squared/100	0.104*** (0.016)	0.104*** (0.015)	0.109*** (0.016)	0.108*** (0.016)
Constant	4.119*** (0.073)	3.846*** (0.066)	3.882*** (0.091)	3.804*** (0.097)
Random effects parameters
State	0.509 (0.052)	0.406 (0.042)	0.410 (0.043)	0.409 (0.043)
Year	0.039 (0.004)	0.038 (0.004)	0.038 (0.004)	0.037 (0.004)
Year-squared/100	0.107 (0.011)	0.102 (0.011)	0.103 (0.011)	0.099 (0.011)
Log likelihood	831.88	864.67	865.40	876.56
-BIC	1,589.41	1,632.70	1,619.28	1,604.45
N (states/observations)	50/1,693	50/1,693	50/1,693	50/1,693

Note. Standard errors are in parentheses. BIC = Baysian Information Criterion.

p < .05. **p < .01. ***p < .001.

Model 2 introduces the monopoly strength variables. The coefficient for wine monopoly is positive but does not breach standard levels of statistical significance. According to the Baysian Information Criterion (BIC) an optimal fit was found with a curvilinear spirits monopoly measure. The comparatively large positive coefficient for the linear term of spirits monopoly (β = .381) matched with the negative coefficient for the quadratic term (β = −.046) indicates a concave functional relationship. Alcohol-related income increases with greater state control over distribution and sales of spirits, but at a decreasing rate. To visualize the relationship, predicted values bounded by a 95% confidence interval (CI) based on Model 2 are plotted in Figure 3.

Figure 3.

Alcohol monopoly strength and net alcohol income (USD).

The curvilinear pattern in Figure 3 illustrates two findings. The steep upward slope for lower levels of monopoly control is capturing the substantive alcohol-related incomes when the state owns wholesale operations only. Recall that all state alcohol monopolies are involved in wholesale, so variation in monopoly strength is determined by retail control. The curvilinear pattern indicates that state control of retail brings a modest net income gain; the lion’s share of alcohol income arrives through wholesale. This conclusion can be discussed in relation to the divestment trends illustrated in Figure 1. States that sold off retail from the late 1970s to early 2000s managed to retain a large share of the pre-divestment alcohol-related incomes by keeping wholesale.¹⁴ Nonetheless, divesting from retail did not make the states better off financially, and sensitivity tests suggest that the states that retained ownership of retail recovered faster from the income slump depicted in Figure 1.¹⁵

There are at least two reasons why the gain from retail ownership is substantively smaller than for wholesale. First is the nature of the industry: Wholesale enterprises are generally more profitable than retail, in large part, because per unit operating expenses are lower. Second, and related, is competition: All states allow sales through licensed private retailers (on- and off-premise), which will draw income away from state stores. Wholesale operations, in contrast, can better resemble true monopolies for administering markups and taxes.

The second finding depicted in Figure 3 is that net incomes increase with ownership but up to the point just prior to where the state completely owns retail stores. In terms of generating income, the optimal organizational arrangement is where the state owns wholesale but uses a mix of state-owned and agency retail contracts. The states that have this model locate state-owned retail stores in densely populated areas and have a limited number of agency contracts in less populated areas (see Table A1 of the appendix). The apparent advantage of this arrangement is that the state secures a presence in the most lucrative retail markets and, through agency contracts, is able to extend the network to regions where the product market is not strong enough to justify the expense of a stand-alone state store. Using Model 2 to generate point estimates, the average per capita income from a license state was US$26.72 (95% CI [US$25.57, US$27.88]), whereas the average per capita income from this optimal monopoly system was US$58.82 (95% CI [US$57.63, US$60.02]), for a gain of approximately 120%.

Models 3 and 4 of Table 3 introduce variables that compete to explain variation in alcohol-related incomes. Major policy change rarely occurs in isolation. In the case of alcohol system privatization, factors such as sales and promotion law or hours of operation might be contemporaneously modified to further liberalize the industry, or perhaps the opposite, to impose new regulations in an effort to minimize foreseeable problems. Model 3 includes controls for the per capita sales of spirits and wine. Coefficients for these alcohol sales variables are statistically indistinguishable from zero. This counter-intuitive result can be attributed to the growth curve model. Random slopes in the model control for state alcohol sales trends; remove the random slopes and the coefficient for spirit sales becomes positive and statistically significant.

A main result is that state ownership is a robust predictor. Coefficients for spirits and wine monopolies change little with expanded models. This conclusion is buttressed by Model 4, which includes all the explanatory variables. The coefficients for state ownership hardly budge even after including measurements for restrictions on retail sales for Sundays and non-Sundays, and for statutes that allow local municipalities to set retail hours.

One unexpected result was the positive relationship between restrictions on non-Sunday hours and net income. States with an off-premise retail hours policy classified as having “high” restrictions (sales are banned more than 36 hr from Monday through Saturday) have 13.6% greater net incomes than the omitted group, states classified as having “low” restrictions (sales are banned less than 24 hr from Monday through Saturday). Similarly, states classified as having “medium” restrictions (banned sales 24 to 36 hr from Monday to Saturday) enjoy net incomes that are 12.5%greater than states with low restrictions. The liberalization of retail hours appears to reduce alcohol-related income.

Better to Own or to Regulate?

In the United States, the State of Washington converted from a public monopoly to a license model in 2012. Pennsylvania and Virginia have debated the merits of following the same path. Global trends, moreover, favor the privatization of state alcohol systems. Alberta Canada fully privatized alcohol liquor sales and distribution in 1993. Poland eliminated their liquor monopoly in 1989. Under pressure from the European Union, the Nordic countries have begun to reduce state control on alcohol; Sweden liberalized the private licensing process in 2011. Often, system privatization is incremental, beginning with products having lower alcohol content. Quebec allowed grocery stores to sell wine in 1978, ending the wine monopoly for provincial stores. Another form of incremental privatization is to have private and public retailers to coexist, which occurred in British Columbia beginning in 2002.

Scholars from a variety of disciplines have weighed in on the social consequences of alcohol monopoly privatization. This analysis exploits the incremental divestment of alcohol systems that took place in the United States from the 1970s to 1990s to test whether privatization affects alcohol-related income earned by states. Our analysis corroborates and expands on earlier literature. State alcohol monopolies contribute to state budgets, with greater intensity of state ownership equating with higher alcohol-related income. Our findings indicate that the optimal income generation model is where the state owns wholesale distribution, yet uses a mix of state outlets and private agency contracts for off-premise retail sales. Importantly, states with this optimal model do not attempt to induce competition between public and agency retail stores. Rather, state stores are placed in geographic areas with enough product demand to justify the costs of a stand-alone state store, while agency contracts are relegated to less-populated areas. Alcohol income optimization is strategically achieved by improving customer access in spatial terms. The results provide no evidence that state alcohol income increases when customer access is made more convenient by expanding retail hours of operation.

State monopoly income estimates can be discussed in relation to the contemporary problem of fiscal stress in the U.S. states. To gain an appreciation for the revenue side of the fiscal stress problem, Table 4 provides COG estimates for the state average per capita revenue from various tax sources over the recession period.

Table 4.

Per Capita Tax Revenue Averages for the U.S. States, 2007 to 2010.

Tax source	Year
Tax source	2007	2008	2009	2010
Individual income tax	$1,128	$1,119	$996	$921
General sales tax	$1,041	$1,018	$969	$906
Corporate income tax	$257	$239	$179	$162
Property tax	$113	$107	$114	$118

Source. Census of Governments (1977-2010).

Note. Figures are in 2010 US dollars. Per capita levels are based on the state population above 18 years.

With the exception of property tax, which dipped in 2008 and recovered a year later, there is a downward trend in these revenue sources. Concomitantly, in any recession, there is a surge in demand for state services to respond to the effects of rising unemployment. In an environment where revenues are dropping and the demand for public services are rising, it becomes tempting to sell public assets to remedy budget deficits without having to raise taxes or engage in service cuts.

One way to frame the value of alcohol monopolies is to examine the gain from state ownership with respect to other tax revenue sources. Using Table 3, Model 2, the estimated per capita alcohol-related net income difference between license states and an optimal state alcohol monopoly is US$32.10. Comparing this marginal gain with the 2010 per capita tax figures in Table 4, the alcohol-related income boost from state ownership was 3.5% of the average revenue from individual income tax, 3.6% of sales tax, 19.9% of corporate income tax, and 27.3% of property tax. For alcohol monopoly states, these percentages can be thought of as the average tax increase necessary, by category, to offset the loss of state income caused by full industry divestment.

If alcohol system divestment over the past three decades is judged by the financial benefit to the states, then, at best, the divesting states stayed even after eliminating state-owned retail stores. Most notably, Iowa managed to retain a large share of alcohol-related income after retail divestment by maintaining state ownership over wholesale and levying new taxes and fees. On average, however, divestment has reduced long-term per capita income, shifting the burden of financing state services to alternative income sources.

Our findings respond to the call for critical tests of the NPM Theory (e.g., Barzelay, 2001; Siltala, 2013). Here, the income generating capacity of alcohol monopolies challenges the NPM concept of governmental entrepreneurship. Viable private markets for alcohol wholesale and retail exist, in which case, the NPM prescription is for the state to externally regulate and leave the business operation in private hands, that is, to steer, not row. However, would such a tactic, from the perspective of the state, be entrepreneurial? The irreducible counter-question is whether a hypothetical private entrepreneur in possession of a similar asset would divest. And with few exceptions, the answer is no. Entrepreneurs with profitable monopolies hold and protect such assets, hence a logical inconsistency; the NPM theoretical foundation implores public managers to mimic the strategies and practices of private entrepreneurs, yet these same public managers are advised to divest from lucrative divisions of state enterprise. The illogic can be traced to a commitment to the ideal of minimal government or to an unconditional faith in the superiority of market economic systems. Regardless, the NPM ship runs aground when the state proves to be a capable oarsman.

In the case of alcohol monopoles, our findings indicate that optimal public value is achieved through state ownership and management, combined with restricted and limited use of private agents. Where public–private arrangements do exist, the goal is not to create quasi-market pressure but instead to efficiently serve non-urban populations. At least in terms of income generation, state ownership of alcohol distribution and sales is preferable over a private market model featuring third-party state regulation. These findings contribute to the literature cautioning against the divestment of income-generating assets as a solution to state fiscal stress (e.g., Dannin, 2011; Dense, 2009; Gilmour, 2012). More generally, it seems a reconceptualization of state entrepreneurialism is in order that contemplates the potential for growing public value through direct public ownership and management. In the case of alcohol distribution and sales, public monopolies apparently helped states weather periods of fiscal stress, sustain essential services, and meet the promise of tax restraint. This is entrepreneurialism on behalf of the electorate and for the long haul.

Limitations

Our analysis has two limitations that relate to the use of state-level COG data. First, this is a topic that affects both state and local governments. Montgomery County Maryland, for example, operates public liquor stores. Rather than break out Montgomery or any other counties that operate public liquor dispensaries, this analysis codes Maryland as a license state. In doing so, the chief assumption is that income collected at the local government level is used for local services and is not transferred to the state. Further refinement can be achieved by incorporating local government data in the analysis. Second, we acknowledge sensitivity in the estimates depending on the classification of each state. For instance, policies regarding agency stores vary, with some agency stores operating in many respects such as license stores, that is, where store owners purchase the inventory and have minimal restrictions on product pricing. To simplify the monopoly measure, “weak” agency stores (e.g., Montana) were treated the same as “strong” agency stores (e.g., Vermont). Additional gradation in the monopoly measure that captures this variation might change the magnitude of the point estimates. Correcting these abstractions would likely change the monopoly income estimates but not the overall conclusion.

Footnotes

Appendix

Table A1.

Monopoly State Product and Organizational Control, 1977 and 2010.

State	Status in 1977				Status in 2010
	Wholesale		Retail		Wholesale		Retail
	Public	Agent	Public	Agent	Public	Agent	Public	Agent
Alabama	S, W		S		S		S
Idaho	S, W		S, W		S		S	S
Iowa	S, W		S, W		S
Maine	S		S			S		S
Michigan	S		S		S
Mississippi	S, W				S, W
Montana	S, W		S, W		S			S
New Hampshire	S, W		S, W		S, W		S, W
North Carolina^a	S		S		S		S
Ohio	S		S		S			S
Oregon	S		S		S			S
Pennsylvania	S, W		S, W		S, W		S, W
Utah	S, W		S, W	S, W	S, W		S, W	S, W
Vermont	S		S		S			S
Virginia	S		S		S		S
Washington	S		S		S		S	S
West Virginia	S		S		S
Wyoming	S, W				S, W

Note. S = distilled spirits; W = wine.

North Carolina has county stores.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was made possible through financial support from the National Alcohol Beverage Control Association.

Notes

Author Biography

Roland Zullo is an Assistant Research Scientist at the University of Michigan where he researches topics related to labor policy, privatization, politics, and critical pedagogy. Zullo thanks Xi Bi, Yu Xiaohan, and Zehra Siddiqui for invaluable research assistance leading to this manuscript.

References

Babor

Caetano

Casswell

Edwards

Giesbrecht

Graham

. . .Rossow

(2010). Alcohol: No ordinary commodity (2nd ed.). Oxford, UK: Oxford University Press.

Barzelay

(2001). The new public management: Improving research and policy dialogue. Berkeley: University of California Press.

Campbell

C. A.

Hahn

R. A.

Elder

Brewer

Chattopadhyay

Fielding

, . . . Task Force on Community Preventive Services. (2009). The effectiveness of limiting alcohol outlet density as a means of reducing excessive alcohol consumption and alcohol-related harms. American Journal of Preventative Medicine, 37, 556-569.

Cook

P. J.

(2007). Paying the tab: The economics of alcohol policy. Princeton, NJ: Princeton University Press.

Dannin

(2011). Crumbling infrastructure, crumbling democracy: Infrastructure privatization contracts and their effects on state and local governance. Northwestern Journal of Law and Social Policy, 6, 47-105.

Dense

(2009). Wither state lotteries? Gaming Law Review and Economics, 13, 404-414.

Gansler

J. S.

(2006). Moving toward market-based government: The changing role of government as a provider. In Kamensky

J. M.

Morales

(Eds.), Competition, choice and incentives in government programs (pp. 37-120). Lanham, MD: Rowman & Littlefield.

Gilmour

J. B.

(2012). The Indiana toll road lease as an intergenerational cash transfer. Public Administration Review, 72, 856-864.

Grubesic

T. H.

Murray

A. T.

Pridemore

W. A.

Tabb

L. P.

Liu

Wei

(2012). Alcohol beverage control, privatization, and the geographic distribution of alcohol outlets. BMC Public Health, 12, Article 1015, 1-10.

10.

Gruenewald

P. J.

Madden

Janes

(1992). Alcohol availability and the formal power and resources of state alcohol beverage control agencies. Alcoholism: Clinical & Experimental Research, 16, 591-597.

11.

Gyimah-Brempon

(2001). Alcohol availability and crime: Evidence from census tract data. Southern Economic Journal, 68, 2-21.

12.

Hahn

R. A.

Cook

J. M.

Elder

Brewer

Fielding

Naimi

T. S.

, . . . Community Preventive Services Task Force. (2012). Effects of alcohol retail privatization on excessive alcohol consumption and related harms: A community guide systematic review. American Journal of Preventative Medicine, 42, 418-427.

13.

Her

Giesbrecht

Room

Rehm

(1999). Privatizing alcohol sales and alcohol consumption: Evidence and implications. Addiction, 94, 1125-1139.

14.

Lane

(2000). New public management. London, England: Routledge.

15.

Livingston

(2008). A longitudinal analysis of alcohol outlet density and assault. Alcoholism: Clinical & Experimental Research, 32, 1074-1079.

16.

Lynn

L. E.

(2006). Public management: Old and new. New York, NY: Routledge.

17.

Mosher

J. F.

Beauchamp

D. E.

(1983). Justifying alcohol taxes to public officials. Journal of Public Health Policy, 4, 422-439.

18.

Norström

Miller

Holder

Österberg

Ramstedt

Rossow

Stockwell

(2010). Potential consequences of replacing a retail alcohol monopoly with a private license system: Results from Sweden. Addiction, 105, 2113-2119.

19.

Organisation for Economic Co-Operation and Development. (2005). Modernising government: The way forward. Paris, France: Author.

20.

Osborne

Gaebler

(1992). Reinventing government. Reading, MA: Addison-Wesley.

21.

Rabe-Hesketh

Skrondal

(2008). Multilevel and longitudinal modeling using Stata. College Station, TX: Stata Press.

22.

Room

(1987). Alcohol monopolies in the U.S.: Challenges and opportunities. Journal of Public Health Policy, 8, 509-530.

23.

Room

(1993). Evolution of alcohol monopolies and their relevance for public health. Contemporary Drug Problems, 20, 169-187.

24.

Seim

Waldfogel

(2013). Public monopoly and economic efficiency: Evidence from the Pennsylvania Liquor Control Board’s entry decisions. American Economic Review, 103, 831-862.

25.

Siegel

DeJong

Albers

A. B.

Naimi

T. S.

Jernigan

D.H.

(2012). Differences in liquor prices between control state-operated and license-state retail outlets in the United States, Addiction, 108, 339–347.

26.

Siltala

(2013). New public management: The evidence-based worst practice? Administration & Society, 45, 468-493.

27.

Simon

J. L.

(1966). The economic effects of state monopoly of packaged-liquor retailing. Journal of Political Economy, 74, 188-194.

28.

Spellman

W. E.

Jorgenson

M. R.

(1982). The social and revenue effects of state alcoholic beverage control. American Journal of Economics and Sociology, 41, 77-83.

29.

Stockwell

Zhao

Macdonald

Pakula

Gruenewald

Holder

(2009). Changes in per capita alcohol sales during the partial privatization of British Columbia’s retail alcohol monopoly 2003–2008: A multi-level local area analysis. Addiction, 104, 1827-1836.

30.

Swidler

(1986). A reexamination of liquor price and consumption differences between public and private ownership states: Comment. Southern Economic Journal, 53, 259-264.

31.

Trolldal

Ponicki

(2005). Alcohol price elasticities in control and license states in the United States, 1982-99. Addiction, 100, 1158-1165.

32.

Wechsler

Lee

J. E.

Hall

Wagenaar

A. C.

Lee

(2002). Secondhand effects of student alcohol use reported by neighbors of colleges: The role of alcohol outlets. Social Science & Medicine, 55, 425-435.

33.

White

G. F.

Gainey

R. R.

Triplett

R. A.

(2012). Alcohol outlets and neighborhood crime: A longitudinal analysis. Crime & Delinquency. Advance online publication. doi: 10.1177/0011128712466386

34.

Zardkoohi

Sheer

(1984). Public versus private liquor retailing: An investigation into the behavior of the state governments. Southern Economic Journal, 50, 1058-1076.