Who pays the bar tab? Beer consumption and economic growth in the united states.
Cesur, Resul ; Kelly, Inas Rashad
What drug provides Americans with the greatest pleasure and the
greatest pain? The answer, hands down, is alcohol. The pain comes not
only from drunk driving and lost lives but also addiction, family
strife, crime, violence, poor health, and squandered human potential.
Young and old, drinkers and abstainers alike, all are affected. Every
American is paying for alcohol abuse.
--Phillip J. Cook, Paying the Tab: The Costs and Benefits of
Alcohol Control (2007)
I. INTRODUCTION
Alcohol consumption has severe social and economic consequences:
lost productivity, disability, early death, crime, family neglect,
personality deterioration, to name a few (Cook 2007). At the individual
level, the effects of alcohol consumption have been explored
considerably by researchers in different fields, including economics,
medicine, public health, criminology, and sociology. However, few
studies have examined the economic impact of alcohol use at the
aggregate level.
Probably, due to the research deficiency regarding its aggregate
effects, some disagreement exists about the overall net benefits of
alcohol use. On the one hand, scholars who have done extensive research
on individual outcomes claim that the overall costs of alcohol use
outweigh its benefits. For example, Cook and Moore (2002, 120) state in
their review article, "The production and sale of alcoholic
beverages account for a small share of national product in the United
States and in other advanced economies. However, the deleterious effects
of alcohol consumption on health and safety constitute a substantial
economic burden, reducing our overall standard of living."
On the other hand, organizations formed by the producers and
distributors of alcoholic beverages claim that the alcoholic beverage
industry contributes significantly to the U.S. economy through job
creation and tax revenues. According to Beer Serves America (2011), an
organization sponsored by the Beer Institute and the National Beer
Wholesalers Association (NBWA):
Directly and indirectly, the beer industry employs
approximately 1.8 million Americans, paying them
over $71 billion in wages and benefits. The industry
pays over $44 billion in business, personal and consumption
taxes, including $5.3 billion in excise taxes
and $5.8 billion in sales, gross receipts, and other
taxes. (1)
Similarly, the Distilled Spirits Council of the United States
(DISCUS 1999), the national trade association representing
America's leading distillers and about 70% of all distilled spirits
brands sold in the United States, states that the alcohol beverage
industry is a major contributor to the U.S. economy through job
creation, wages, and taxes. (2)
Room and Jernigan (2000) point out that the benefits reaped from
the alcohol industry in terms of economic development should be weighed
against the negative impacts on economic development in addition to
public health. They go on to say that alcohol's contribution to
total production can vary tremendously from country to country
(generating 3% of total manufacturing value added in Argentina or the
Netherlands, but as little as 0.07% in Iran). Aside from the
aforementioned contributions of the alcohol industry to the economy
often cited by the industry itself, supply-side efforts also play a role
in the consumption of alcohol. This can be done through influencing
government regulation of the market for alcohol and selling cheap
alcohol in bulk, which has "a devastating public health
impact" (Stockwell and Crosbie 2001). Indeed, the economic burden
of excessive alcohol consumption has been found to be substantial,
recently estimated at $223.5 billion (Bouchery et al. 2011).
Although a few studies have examined the impact of macroeconomic
conditions on drinking prevalence and patterns, such as those by Ruhm
and Black (2002) and Ruhm (1995), we are not aware of any studies
exploring the impact of alcohol consumption on economic growth. This
study fills that void by using annual, state-level U.S. data for the
period 1971-2007 to examine the effect of alcohol use on per capita
personal income growth and per capita gross domestic product (GDP)
growth (hereafter, economic growth) in a general equilibrium framework.
Although incorporating all alcoholic beverages into the analysis would
be ideal, our focus is exclusively on beer consumption for the following
reasons. Beer is the major source of alcohol consumption in the United
States, making up approximately 55% of total ethanol consumption
(LaVallee and Yi 2011; Ruhm et al. 2011). (3) It is the beverage of
choice among Americans, particularly among youths (Chaloupka, Grossman,
and Saffer 2002), making it more policy-relevant for tax changes as
youths are more price-sensitive and are more likely to binge-drink
(Centers for Disease Control and Prevention 2012). The potential
endogeneity of beer consumption can be addressed by using the excise
beer tax rate as an instrumental variable. As wine and liquor are sold
through state stores in many states, it is difficult to determine tax
rates for wine and liquor. (4) Therefore, in line with previous studies
and following the convention in the health economics literature, we use
per capita beer sales as a proxy for alcohol consumption at the
state-level. (5)
We refer to the relevant literature, examining the effects of
alcohol consumption on a variety of outcomes, to identify the
microeconomic channels through which drinking may affect economic
growth. In a regression of economic growth on beer consumption, we
address the potential endogeneity of per capita beer consumption by
utilizing the levels of beer excise tax rates as instrumental variables.
We show that, once the potential endogeneity is accounted for, per
capita ethanol consumption taken through beer is negatively associated
with short-run economic growth measures. These results are robust to
various specification checks, including reduced form relationships
between economic growth variables and different types of excise taxes.
(6)
Our study contributes to the economics literature in several
important respects. First, we document that, parallel to a large body of
literature in different disciplines, the aggregate costs of the
"bar tab" are indeed borne by society as a whole. Second, to
our knowledge this is the first study providing evidence on the
potential relationship between consumption-related negative
externalities and aggregate economic well-being in a general equilibrium
framework. The findings of this study suggest that Pigouvian taxes may
serve as externality-correcting instruments, leading to increased
efficiency in terms of aggregate economic growth. Finally, our results
suggest that exploring the relationships between health behaviors and
other types of aggregate economic outcomes may produce valuable
information for policy makers (i.e., exploring the aggregate effect of
beer consumption on health care expenditures).
The remainder of the paper is organized as follows. Section II
reviews a broad range of literature to discuss the various microeconomic
channels through which drinking may affect worker productivity and
national savings and, thus, economic growth. Section III introduces the
econometric model. Section IV presents the data. Section V presents the
ordinary least squares (OLS) results. (7) Section VI shows the
Instrumental Variables (IV) estimates. Section VII examines the reduced
form relationships between beer excise tax rates and economic growth to
explore the validity of the findings in Section VI. Section VIII
performs additional robustness checks. Conclusions are given in Section
IX.
II. BACKGROUND AND CONCEPTUAL FRAMEWORK
It is well documented in the economics literature that labor
productivity, population growth, human capital accumulation, and
physical capital accumulation are among the more important determinants
of economic growth. (8) Alcohol consumption can affect economic growth
through its effect on various microeconomic sources of economic growth,
as aggregate-level relationships are based on microeconomic foundations.
Drinking-related lost workdays, which represent a direct negative
shock to the labor supply of a worker, can result from sickness
absences, driving under the influence (DUI) arrests, alcohol-related
injury work leaves, incarcerations due to alcohol-induced criminal
activity, and alcohol-dependence-related withdrawals from the labor
force. Several studies show a direct association between alcohol use and
reduced workplace performance measured through absenteeism, poor
relations with coworkers, and accidents. (9) Various survey-based
studies point to the fact that overindulgence in alcohol impedes
productivity at work, leading to reduced earnings and fringe benefits,
reduced on-the-job performance, and reduced employment opportunities.
(10)
The relationship between alcohol use and morbidity-mortality is
also well documented. In about two-thirds of all injury deaths, the
victim or other involved party had been drinking enough to be
intoxicated by the usual standard (Centers for Disease Control and
Prevention 2004; Smith, Branas, and Miller 1999). Alcohol consumption is
linked to car accidents, disease burden around the world, and sexually
transmitted infections. (11) Morbidity reduces the returns to human
capital by lowering labor productivity as well as increasing lost
workdays. Injury-related spending shifts funds from productive
investments to injury treatment and related rebuilding expenditures.
(12) Therefore, morbidity is likely to have a negative impact on
economic growth through lost workdays and the reallocation of savings
from productive uses to injury prevention and damage-replacement
spending.
Macroeconomic studies examining the impact of health on economic
growth often use life expectancy at birth as a proxy for health at the
aggregate level. Acemoglu and Johnson (2007) employ an instrumental
variables strategy to estimate the causal effect of life expectancy on
income growth. (13) They find that increases in life expectancy (and the
associated increases in population) appear to reduce income per capita.
On the other hand, some studies find that lower mortality directly
affects economic growth. Specifically, Kalemli-Ozcan, Ryder, and Weil
(2000) and Kalemli-Ozcan (2002) report that increased life expectancy
leads to greater long-run economic growth since longevity increases
investment in human capital through education, while fertility decreases
in response to increased longevity; thus, in the long run, life
expectancy has a positive impact on economic growth. Ashraf, Lester, and
Weil (2009) find that the impact of positive health shocks on income
growth depends on which part of the population is most affected by the
disease. For example, in the short run, eradicating tuberculosis raises
per capita income, whereas eradicating malaria lowers it. The
differential effects of eradicating these diseases on income arise
largely because tuberculosis strikes mostly prime-aged workers, while
malaria affects mainly young children (Ashraf, Lester, and Weil 2009).
When it comes to evaluating the impact of mortality on per capita
income growth, one must consider the employment status of those who die.
(14) A person's death decreases population, which has a positive
impact on per capita income if that person is not in the labor force. If
the person is employed, however, the loss of productive labor has a
negative impact on per capita income. The aggregate effect depends on
the mix of individuals in and out of the labor force. Deaths of
prime-aged workers who are in the labor force and deaths of people who
are not in the labor force do not have the same effect on per capita
income growth. Consequently, the net effect of alcohol-induced deaths on
economic growth depends on the employment status of the deceased.
Various studies undertaken in different parts of the world show
that alcohol use is costly to societies. According to the U.S.
Department of Health and Human Services (2000) report, the economic cost
of drinking in the United States for 1992 was about $148 billion. This
report makes its calculations based on medical expenses, lost present
and future earnings due to alcohol-related illnesses, deaths and crimes,
costs of crashes, fires, and criminal justice expenses. Alcohol-related
hospital charges in 1998 in New Mexico were $51 million in comparison to
$35 million collected as alcohol taxes. In Canada, the economic costs of
alcohol amount to approximately US$18.4 billion, representing 2.7% of
the GDP (World Health Organization 2001). Studies in other countries
have estimated the cost of alcohol-related problems to be around 1% of
the GDP (Collins and Lapsley 1996; Rice et al. 1990). Alcohol-related
externalities also impose large costs to the economies in the European
Union. Total tangible costs of alcohol were estimated to be 125 billion
[euro] in the EU in 2003. These costs are those associated with the
criminal justice system, health care, and lost workdays. Barker (2002)
finds that in New Zealand in 1999, alcohol excise taxes only account for
half of alcohol-related direct costs.
Different studies advocate increases in alcohol excise taxes,
arguing that they would produce favorable outcomes, particularly through
reducing the negative externalities associated with alcohol consumption.
Manning et al. (1989) show that alcohol excise taxes cover only about
half the costs imposed on others. According to Parry, West, and
Laxminarayan (2009), the
optimal excise taxes in the United States should be three to ten
times more than current levels.
On the other hand, despite the fact that heavy drinking lowers
productivity at work, several studies examining the impact of moderate
drinking on wages find a positive relationship between the two. (15)
Stated differently, these studies provide evidence in favor of the
hypothesis that there is a drinker's bonus for moderate drinkers.
Similarly to the drinker's bonus hypothesis, there are studies
focusing on the potential health benefits of alcohol consumption.
Hoffmeister et al. (1999) find that light alcohol consumption reduces
the risk of cardiovascular disease and total mortality risk. Additional
studies showing the beneficial effects of alcohol consumption on
cardiovascular health include Gronbaek (2001), Klatsky (2002), Rehm et
at. (2003), and Thun et al. (1997). (16)
III. EMPIRICAL IMPLEMENTATION
Although the existing literature is suggestive of the potential
effects of alcohol use, they do not necessarily represent the causal
effect of alcohol consumption. Drinking and low economic productivity
may be due to unobservable personal characteristics of those who drink
heavily and have low on-the-job productivity. Similarly, the positive
association between moderate drinking and improved health/higher wages
may be driven by unobservable individual traits. On the basis of the
literature discussed above, the net impact of alcohol consumption on
economic growth depends on the relative magnitudes of opposing factors
and remains an empirical question.
The following OLS model is estimated:
(1) [Y.sub.s,t] = [alpha] + [[beta].sub.1] [A.sup.s,t] +
[[beta].sub.2][Z.sub.s,t] + [[eta].sub.s] + [[tau].sub.t] +
[[epsilon].sub.s,t]
where subscripts s and t refer to state and year, respectively.
Depending on the model estimated, [Y.sub.s,t] is either the yearly
state-level per capita personal income (PCPI) or per capita GDP growth.
A refers to the variable of interest, state-level per capita ethanol
consumption taken through beer. Z represents a vector of control
variables. State-fixed effects, year dummies, and the idiosyncratic
error term are denoted by [[eta].sub.s], [[tau].sub.t], and
[[epsilon].sub.s,t], respectively. State-fixed effects control for
timeinvariant state characteristics, such as culture surrounding
drinking. Year dummies control for common time trends and time shocks,
such as changes in the prevalence of drinking at the national level.
(17)
The vector Z includes lagged values for PCPI (or per capita GDP) to
control for the impact of convergence on economic growth as well as the
potential effect of state-level PCPI (or per capita GDP) on alcohol
sales. (18) The ratio of total public expenditures to personal income
(GDP) accounts for the effect of state government size on economic
growth. The average total tax rate controls for the effect of taxes on
productive activities. Percent of population under 18 and above 64
crudely controls for the impact of labor force participation on economic
growth. Individuals under 18 and above 64 are expected to drink less and
produce less. Hence, omitting this variable would lead to an
underestimation of the potential negative effect of alcohol consumption
on income growth. The percent of the population older than 17 and
younger than 25 is specified in the estimation equation to capture the
joint effect of changes in the percentage of this age group on income
growth and alcohol use. People in the 18-24 age group engage in more
episodes of binge drinking (Centers for Disease Control and Prevention
2012), which is why we control for this age group separately.
State-level education variables, which are percent with a high school
degree, percent with some college education, and percent with a college
degree, are also specified to account for the impact of past
human-capital accumulation on income growth. Controlling for state-level
education variables also helps to account for omitted variables if
education is associated with both alcohol consumption patterns and
economic growth. Variables pertaining to ethnic composition (percent
Black and percent other race) are specified to reflect the effect, if
any, of ethnic composition on income growth and alcohol consumption. For
instance, controlling for population characteristics would account for
the potential correlation between drinking patterns and worker
productivity.
OLS estimates may be biased for several reasons. Research stressing
the psychological responses to economic conditions predicts that alcohol
use will rise during economic downturns as a tool for stress control.
(19) Even though overall per capita alcohol consumption can fall during
economic downturns due to an income effect, alcohol consumption may
nevertheless go up as people self-medicate to ease their stresses. Using
BRFSS data from 1984 to 1995, Dee (2001) finds that overall drinking and
the probability of having 60 or more drinks falls during economic
downturns; however, the probability of binge drinking increases. Thus,
Dee concludes that the prevalence of binging is countercyclical.
Nevertheless, various studies provide evidence against this view.
Studying the impact of job stress, research in psychology implies that
drinking may increase with the intensity of employment. (20)
Furthermore, other studies find that alcohol consumption is positively
related to income (Skog 1986; Sloan, Reilly, and Schenzler 1995). Ettner
(1997), using the 1988 National Health Interview Survey (NHIS), shows
that nonemployment significantly reduces both alcohol consumption and
dependence symptoms with the exception of mixed evidence with respect to
involuntary employment. She finds that job loss increases the
consumption of alcohol in the overall sample, but reduces dependence
symptoms among individual drinkers. Ruhm and Black (2002) and Ruhm
(1995) find that intake of hard liquor declines in bad economic times,
leading to a decrease in alcohol abuse as well. Thus, stress-induced
drinking during economically depressed periods might be partially or
fully offset by reductions due to decreased earnings (Ruhm and Black
2002). Using individual-level data from the BRFSS for the period
1987-1999, Ruhm and Black (2002) find that the decrease in per capita
alcohol consumption measures in bad economic times is concentrated among
heavy drinkers, while light drinking increases. Based on these studies,
whether alcohol consumption is procyclical or countercyclical is
ambiguous.
Depending on the magnitudes of the opposing factors, OLS estimates
of the effect of beer sales on economic growth may be biased in either
direction. To address the potential endogeneity of beer sales in this
context, we undertake an instrumental variables (IV) methodology,
described as follows.
A. Instrumental Variables
Our IV method employs the levels of beer excise taxes as
instruments. (21) More specifically, the following first-stage
relationship between per capita alcohol consumption measures and alcohol
excise taxes is postulated:
(2) [A.sub.s,t] = [alpha] + [[beta].sub.1] [T.sub.s,t] +
[[beta].sub.2][Z.sub.s,t] + [[eta].sub.s] + [[tau].sub.t] +
[[epsilon].sub.s,t]
where T is the per gallon beer excise tax rate. The levels of beer
excise tax rates are used as instruments for the corresponding per
capita alcoholic beverage type with the assumption that these alcoholic
beverages are purchased in national markets, so alcohol excise taxes
proxy for alcohol prices. (22) In addition to a national market
assumption, the validity of alcohol excise tax rates as instruments
depends on whether the following two conditions hold. First, there has
to be a statistically significant relationship between beer excise tax
rates and per capita beer consumption. Second, the beer excise tax rates
should not otherwise influence income growth beyond their indirect
effect through alcohol consumption. Stated differently, the following
exclusion restriction should hold: Cov([T.sub.s,t], [[epsilon].sub.s,t])
= 0, where [[epsilon].sub.s,t] represents the error term in the
second-stage equation.
The evidence on the effectiveness of beer excise tax rates in
predicting alcohol sales is mixed. Other factors may cast doubt on the
power of beer excise taxes in serving as proxies for beer prices. Much
of the variation in excise tax rates over time is induced through
changes in inflation instead of changes in the levels of nominal values
of excise tax rates, in addition to the fact that state tax rates
usually comprise a small share of the alcoholic beverage prices (Young
and Bielinka-Kwapisz 2002). Using American Chamber of Commerce
Researchers Association (ACCRA) price data, Young and Bielinka-Kwapisz
(2002) document that beer excise taxes are poor predictors of alcohol
prices; nevertheless, in a follow-up study, Young and Bielinka-Kwapisz
(2003) find that ACCRA price data are not reliable due to substantial
measurement error, but employing state and federal tax rates as
instruments mitigates the problem. By using detailed cross-sectional
barcode data, Ruhm et al. (2011) also show that ACCRA price data are not
reliable indicators of alcohol prices. According to Ruhm et al., in a
cross-sectional setting, the total tax elasticities of beer consumption
are indistinguishable from zero (both in terms of magnitude and
statistical significance), while higher tax rates seem to have a small
inverse effect on beer drinking participation. It is worth mentioning
that, even though the findings of Ruhm et al. (2011) are very
informative, in our framework their findings are of minimal concern
because the variation in beer tax rates in our study is generated not
only by the cross-sectional variation but also by the longitudinal
variation in beer tax rates. Furthermore, Ruhm et al. data are for the
period 2000-2004. With the exception of the 1991 federal alcohol tax
increase, alcohol excise tax rates exhibit strong downward trends in the
period 1971-2007. During the 2000s, on average, the real values of beer
excise tax rates are lower than the real values of beer tax rates during
the 1970s and 1980s. Hence, in our framework, we expect that beer excise
tax rates exhibit enough variation to serve as instruments in predicting
per capita alcohol sales measures. In the final analysis, the
statistical significance of the relationship between beer excise tax
rates and beer consumption can be empirically tested.
A legitimate concern related to the exclusion restriction is the
possibility that alcohol excise tax rates may affect economic growth
through other channels. For example, if alcohol excise tax rates affect
economic growth through their effect on state budgets, then the validity
of alcohol excise tax rates as instrumental variables could be called
into question. Similarly, if state-level economic growth determines
state alcohol excise tax rates, and/or state policies on economic growth
are linked to alcohol excise tax rates, alcohol excise taxes may not
serve as suitable instrumental variables in estimating the effect of
alcohol consumption on economic growth. By estimating the reduced form
relationships between economic growth and the different types of excise
tax rates, one can (imperfectly) test the validity of these concerns. If
the reduced form models provide uniform results (similar effects on
economic growth measures) for the different types of alcohol excise tax
rates regardless of the effects of excise taxes on drinking behavior
(i.e., beer excise taxes and alcohol use), then utilizing alcohol excise
tax rates as instrumental variables would be questionable. To address
these concerns, we estimate the effects of different types of excise tax
rates on economic growth in Section VII. If the various types of alcohol
excise tax rates have differential effects on economic growth and the
suggested relationship is likely through the effect of excise taxes on
consumption behavior, the findings would provide evidence in favor of
the hypothesis that alcohol excise tax rates do not violate the
exclusion restriction; nevertheless, the opposite would cast doubt on
the legitimacy of excise tax rates as suitable instruments.
Another concern is that if beer excise tax rates have an impact on
income growth through their correlations with policies taking place at
the national level within the same year, the exclusion restriction may
be violated. To mitigate this concern, we control for year dummies to
account for the effect of annual changes at the national level. Hence,
this potential problem is minimized in our framework. Finally, if
alcohol excise tax rates affect income growth through their potential
effect on business activity, the exclusion restriction, discussed above,
may be violated. The validity of this concern is also examined in
Section VII when we provide reduced form estimates of economic growth on
different alcohol excise tax rates. If the estimated coefficients on the
alcohol excise tax rates are negative and statistically significant in
the reduced form specifications, then the validity of alcohol excise tax
rates can be at risk because an inverse relationship between excise tax
rates and income growth would suggest deadweight losses induced by
inefficient taxation. However, if the estimated effects are positive and
statistically significant, the validity of alcohol excise tax rates as
instruments will be much less questionable since a reduced form positive
relationship suggests that the influence of alcohol excise tax rates is
through their effect on behavior. (23)
IV. DATA
Descriptive statistics for the data are shown in Table 1. Data on
yearly state-level per capita alcohol sales measures for the population
14 years and older for the period 1971-2007 come from the National
Institute on Alcohol Abuse and Alcoholism (NIAAA 2010). Per capita beer,
wine, and liquor consumption reflect the gallons of per capita pure
ethanol calculated through the sales of beer, wine, and liquor,
respectively.
Data on state-level PCPI and per capita GDP come from the Bureau of
Economic Analysis (BEA). The regional CPI data are obtained from the
Bureau of Labor Statistics (1982-1984= 100). Economic growth measures,
PCPI growth, and per capita GDP growth are calculated by using the
regional CPI-adjusted values of PCPI and per capita GDP.
Data on state-level average total tax to personal income (GDP)
ratio and average total expenditure to personal income (GDP) ratio come
from the various issues of the Statistical Abstract of the United
States. The population age composition data are obtained from the U.S.
Census Bureau (2011).
Data on state-level beer excise taxes are taken from the U.S.
Brewers Association's Brewers' Almanac for the period
1971-1999. Data on state-level excise liquor and wine taxes for the
period 1971-1999 are taken from DISCUS. For the period 2000-2007, data
on alcohol excise taxes come from the Tax Foundation (2011). Taxes used
in the analysis are adjusted using the regional CPI. In all cases,
alcohol excise taxes are calculated by summing up the federal and state
excise taxes. (24) Data on the education and ethnic composition
variables are obtained from the March Current Population Survey.
V. OLS RESULTS
In all models estimated in this article, standard errors are robust
to arbitrary heteroskedasticity and serial correlation at the state
level. OLS estimates of PCPI growth are presented in Panels A and B of
Table 2. In Panel A, column 1 controls for the lagged PCPI, state total
tax to personal income ratio, state total expenditure to personal income
ratio, regional inflation, and percent of population under 18 and above
64. As we move from column 2 to 4, the following variables are
sequentially added to the models estimated: percent of population in the
age group 18-24, educational attainment indicators, and race composition
indicators. Although generally statistically insignificant, results show
an inverse association between per capita beer sales and PCPI growth. In
column 1, a one-gallon increase in per capita beer consumption is linked
to a 0.71 percentage point decrease in PCPI growth. In column 4, the
model with the most comprehensive control variables, a one-gallon
increase in per capita beer consumption is associated with a 0.48
percentage point lower
PCPI growth. Based on column 4, if OLS estimates were to represent
the causal relationship between per capita beer consumption and PCPI
growth, at the mean level, a 10% increase in per capita beer consumption
would cause a 0.062 percentage point decrease in PCPI growth.
Panel B replicates the same exercise for GDP growth. OLS estimates
of GDP growth on per capita beer consumption do not suggest a meaningful
measurable relationship between the two.
Some of the results associated with the control variables merit
discussion. In column 4, the coefficients on lagged PCPI suggest that a
thousand dollar increase in PCPI (corresponding to a 7.14% increase in
PCPI at the mean level) is associated with a 0.72 percentage point lower
PCPI growth rate. The sign of the estimated coefficient on lagged PCPI
supports the convergence hypothesis, which predicts that states with
higher per capita PCPI are expected to have slower PCPI growth rates.
State total tax collections to state personal income ratio and state
total expenditures to state personal income ratio are negatively linked
to PCPI growth with statistically significant coefficients at less than
the 5% level. The coefficient on regional inflation is negative and
statistically significant at less than 1% level in all cases. The
coefficients on education and race indicators do not seem to be strongly
linked to PCPI growth. It should be noted that the estimated
coefficients on the control variables just discussed, as well as the
ones on the variable of interest (per capita beer consumption), do not
always reflect their causal effect on PCPI growth.
VI. INSTRUMENTAL VARIABLES RESULTS
A. First-Stage Results
Table 3 presents the first-stage estimation of per capita beer
sales on the excise beer tax rate. According to the results, a
one-dollar increase in the per gallon beer excise tax rate is associated
with a 0.365-gallon decrease in per capita beer sales, and the estimated
coefficient is statistically significant at less than the 1% level. At
the mean level, the implied beer excise tax elasticity of per capita
beer sales is -0.163. (25) That is, a 10% increase in the beer excise
tax rate leads to a 1.63% decrease in per capita beer sales. This
finding runs parallel to that of previous studies (Ruhm et al. 2011).
B. Second-Stage Results
Table 4 presents the IV estimates for the economic growth measures.
Panel A shows the results for the estimates of PCPI growth, whereas
Panel B presents the results for the estimates of per capita GDP growth.
The first stage F-statistics values (> 100 in Panel A, and >50 in
Panel B) indicate a strong first-stage relationship between excise beer
tax and beer sales.
The estimated coefficient in Panel A, column 1 suggests that,
conditional on other controls, a one-gallon increase in per capita beer
consumption leads to a 3.5 percentage point decrease in state-level PCPI
growth. Controlling for the percent of population (> 17 and <25)
reduces the estimated coefficient from -0.035 to -0.030 in magnitude. In
column 3, controlling for state-level education variables increases the
estimated coefficient from -0.030 to -0.032 in magnitude. In column 4,
results from the model with the most comprehensive controls are
presented; a one-gallon increase in per capita beer consumption is
associated with a 3.15 percentage point decrease in state PCPI growth.
That is, at the mean level, a 10% increase in per capita beer sales
leads to a 0.409 percentage point decrease in state PCPI growth. The
magnitude and significance of the coefficients on beer consumption
remain virtually unaltered across the columns, increasing our confidence
in the estimates.
Similar to the results shown in Panel A, the IV estimates of per
capita GDP growth on per capita beer sales shown in Panel B suggest an
inverse and statistically significant relationship between per capita
GDP growth and per capita beer consumption.
VII. REDUCED FORM ESTIMATES OF INCOME GROWTH ON EXCISE TAXES
Examining the reduced form relationships between alcohol and
cigarette excise taxes and economic growth measures may produce valuable
information, particularly in exploring the validity of beer excise taxes
as instruments in our framework.
Panels A and B in Table 5 present the regression results for PCPI
growth and per capita GDP growth on alcohol and cigarette excise tax
rates, respectively. (26) The results for the excise beer, wine, liquor,
and cigarette tax rates, separately and then jointly, are presented in
columns i through 5, respectively. The results focusing on the beer
excise tax rate are reported in column I. According to the findings, the
level of the beer excise tax rate has a positive and statistically
significant effect on PCPI growth and per capita GDP growth. On the
basis of the estimated coefficients in column 5, a 25% increase in the
beer excise tax rate leads to a 0.242 percentage point increase in PCPI
growth and a 0.346 percentage point increase in per capita GDP.
The estimated coefficients on the wine, liquor, and cigarette
excise tax rates are not statistically significant in any of the models
estimated. In column 5, when all four types of excise taxes are jointly
estimated, the results are similar to those obtained in columns 1 to 4.
The differential relationships between the various types of excise
taxes and the measures of economic growth presented in this section
provide evidence supporting using excise tax rates as instruments. If
the relationship between the beer excise tax rate and economic growth
variables were through state tax revenues or due to other state
characteristics affecting economic growth and taxation simultaneously,
such relationships between economic growth measures and the different
types of excise tax rates would be very similar. Furthermore, it is
informative that even though the impacts of wine/liquor and cigarette
excise taxes on behavior are different (i.e., excise cigarette taxes
seem to affect cigarette sales while excise wine/liquor taxes do not
seem to affect wine/liquor sales), neither type seems to have a
measurable effect on economic growth in an annual state-level framework
for different reasons. That is, even though wine and liquor consumption
would have a negative effect on aggregate productivity, wine/liquor
excise tax rates do not seem to be related to economic growth, possibly
because their power to predict per capita wine/liquor consumption is
very limited. On the other hand, although the cigarette excise tax rate
strongly predicts per capita cigarette consumption, it does not seem to
be significantly related to economic growth (results available upon
request), potentially for the reason that smoking-induced costs are
expected to have a much smaller effect on economic productivity in the
short run in comparison to that induced by drinking.
Overall, these findings suggest that excise tax rates can
potentially affect aggregate productivity to the degree that they can
affect the consumption behavior in our framework. Hence, these findings
suggest that our instrumental variables specifications do not violate
the exclusion restriction.
VIII. ADDITIONAL ROBUSTNESS CHECKS
To examine the robustness of our findings we perform additional
robustness checks in Tables A1 and A2.
Owing to complementarities between beer drinking and nonbeer
alcoholic beverages, the estimates of the effect of per capita beer
consumption on economic growth may be misleading if increases/decreases
in beer drinking are compensated by corresponding changes in per capita
wine and liquor consumption. Therefore, in Table A1, Panel A, economic
growth measures are estimated on per capita beer consumption while
controlling for per capita wine and liquor consumption. The estimates
are robust to controlling for per capita wine and liquor consumption and
therefore rule out the possibility that the coefficient on per capita
beer consumption is contaminated by simultaneous changes in beer and
liquor consumption.
NIAAA data also include per capita alcohol consumption measures for
the population 21 years and older. (27) In Panel B, we present IV
estimates of economic growth measures on per capita beer sales 21 and
older. Results are very similar to those with per capita beer sales 14
and older.
To examine whether outliers in the data may be driving the results,
in Panel C we regress PCPI and per capita GDP growth on per capita beer
sales by excluding the state-year observations at the top and bottom 10%
of the distribution, respectively. Results, which are similar to the
main estimates, suggest that our findings are not driven by outliers in
the data. In Panel D, we present the results for the 5-year averages for
the period 1971-2005. The estimates are similar to the main results.
To further investigate the robustness of our findings, in Table A2,
Panel A, we present estimates of income growth measures on per capita
total ethanol/alcohol sales. By doing so, we also provide a test of
whether the estimated coefficient on per capita beer consumption is
representative of the changes in per capita total ethanol change.
Results run parallel to the ones for per capita beer sales
specifications. In Panels B, C, and D of Table A2, we perform robustness
checks similar to the ones performed in the corresponding panels of
Table A1. The estimates of economic growth measures on per capita total
ethanol sales are similar to those on per capita beer use measures.
Therefore, these results are suggestive that changes in per capita
ethanol consumption through beer are representative of changes in per
capita total ethanol consumption. Nevertheless, it should be noted that
we cannot dismiss the possibility that the coefficient estimate on per
capita total ethanol consumption may be driven by per capita beer
consumption.
IX. DISCUSSION
A vast body of literature on alcohol use documents that beer
consumption generates negative externalities that are large in
magnitude, and increases in excise beer taxes would reduce such
externalities. This article provides evidence of an inverse relationship
between drinking and state-level income growth. Our results suggest that
the cost of drinking may indeed be borne by society as a whole.
In particular, our preferred estimates suggest that a standard
deviation increase in per capita beer consumption (of 0.241 gallons)
from the mean (of 1.299 gallons) would lead to a 0.76 percentage point
increase in PCPI growth and a 1.54 percentage point increase in per
capita GDP growth. Nevertheless, we should note that the aim of this
study was not to predict economic growth in the long run, but rather to
estimate the potential effect that drinking might have on economic
growth in the short run (i.e., in a state-level yearly framework). Since
per capita beer sales have remained relatively stable over time, they
cannot explain a high portion of the variation in income growth over the
entire period (1971 to 2007). (28) Another naive interpretation of these
results would be to claim that prohibition would increase prosperity.
Prohibition may affect business activity through multiple channels
including loss in tourism revenues, increases in illegal activity, and
changes in market structure, in addition to its potential effect on
drinking. Even though this study provides evidence that alcohol use has
a small but statistically significant effect on income growth, our
findings cannot be used to draw conclusions regarding prohibition as a
policy tool.
There are several limitations to this study. In this article, we
are able to explore net effects on the economy, but we do not
empirically test mechanisms. While we argue as to reasons for using beer
consumption as a proxy for alcohol abuse, we do not directly test for
abuse. Moreover, while we use several methods to address the potential
endogeneity of beer consumption, establishing causality remains a
difficult task.
Our results do, contrary to the claims of the Distilled Spirits
Council and Beer Serves America, suggest that alcohol excise taxes do
not harm economic activity at the aggregate level. If anything, our
results show that beer excise tax rates are directly linked to economic
growth. As a policy prescription, our reduced form estimates suggest
that an increase in the real beer excise tax of 26 cents (one standard
deviation) over the current mean of 31 cents (based on 2011 data) would
potentially increase PCPI growth by 0.43 percentage points, or 27% from
the mean, and lower the economic costs associated with alcohol
consumption. While no prior studies to our knowledge have examined the
effects of excise beer taxes on economic growth, many analyze the
effects of taxes in general (particularly income taxes) on growth, and
the magnitudes we find are reasonable. (29) Owing to the negative
externalities associated with excessive alcohol consumption, analyzing
the effects of these types of excise taxes on economic growth in more
detail should be high on economists' research agendas.
ABBREVIATIONS
ACCRA: American Chamber of Commerce Researchers Association
BEA: Bureau of Economic Analysis
CPI: Consumer Price Index
DISCUS: Distilled Spirits Council of the United States
DUI: Driving Under the Influence
GDP: Gross Domestic Product
IV: Instrumental Variables
NBWA: National Beer Wholesalers Association
NIAAA: National Institute on Alcohol Abuse and Alcoholism
OLS: Ordinary Least Squares
PCPI: Per Capita Personal Income
doi: 10.1111/ecin.12048
APPENDIX
TABLE A1
IV Estimates of PCPI and GDP Growth on Per Capita Beer Sales,
Robustness Estimates
(1) (2)
Dependent Variable PCPI Growth GDP Growth
Panel A: Controlling for per capita wine and liquor sales
Per capita beer sales 14 and older -0.0465 ** -0.0664 *
(0.0228) (0.0348)
Observations 1.776 1,776
[R.sup.2] 0.562 0.523
First stage F test 43.86 55.04
First stage F test p value 0.00 0.00
Panel B: Estimates of growth on per capita beer sales 21 and older
Per capita beer sales 21 and older -0.0370 ** -0.0520 **
(0.0169) (0.0263)
Observations 1,776 1,776
[R.sup.2] 0.556 0.516
First stage F test 54.50 61.95
First stage F test p value 0.00 0.00
Panel C: Estimates of growth measures excluding potential outliers
Excluding Top 10%
Per capita beer sales 14 and older -0.0458 ** -0.0727 *
(0.0229) (0.0373)
Observations 1,598 1,598
[R.sup.2] 0.563 0.543
First stage F test 44.30 50.22
First stage F test p value 0.00 0.00
Panel D: Estimates of growth measures, 5-year averages sample
Per capita beer sales 14 and older -0.0343 * -0.0512
(0.0192) (0.0381)
Observations 336 336
[R.sup.2] 0.482 0.292
First stage F test 23.95 29.74
First stage F test p value 0.00 0.00
(3) (4)
Dependent Variable PCPI Growth GDP Growth
Panel A: Controlling for per capita wine and liquor sales
Per capita beer sales 14 and older
Observations
[R.sup.2]
First stage F test
First stage F test p value
Panel B: Estimates of growth on per capita beer sales 21 and older
Per capita beer sales 21 and older
Observations
[R.sup.2]
First stage F test
First stage F test p value
Panel C: Estimates of growth measures excluding potential outliers
Excluding Bottom 10%
Per capita beer sales 14 and older -0.0581 * -0.0814 *
(0.0343) (0.0492)
Observations 1,599 1,599
[R.sup.2] 0.528 0.493
First stage F test 15.27 21.13
First stage F test p value 0.00 0.00
Panel D: Estimates of growth measures, 5-year averages sample
Per capita beer sales 14 and older
Observations
[R.sup.2]
First stage F test
First stage F test p value
Notes: Robust standard errors corrected for clustering on the state
are in parentheses. All models control for full set of controls used
in column 4 of Table 4 as well as state- and year-fixed effects.
*** 17 < .01, ** p < .05, * p < .1.
TABLE A2
IV Estimates of PCPI and GDP Growth on Per Capita Total
Ethanol/Alcohol Sales
(1) (2)
Dependent Variable PCPI Growth GDP Growth
Panel A: Estimates of growth on per capita total ethanol/alcohol sales
Per capita total ethanol/alcohol -0.0314 ** -0.0442 *
sales 14 and older (0.0142) (0.0228)
Observations 1.776 1.776
[R.sup.2] 0.531 0.487
First stage F test 12.31 13.96
First stage F test p value 0.00 0.00
Panel B: Estimates of growth on per capita ethanol/alcohol sales 21
and older
Per capita total ethanol/alcohol -0.0247 * -0.0349 **
sales 21 and older (0.0110) (0.0177)
Observations 1.776 1,776
[R.sup.2] 0.526 0.481
First stage F-test 12.23 13.42
First stage F test p value 0.00 0.00
Panel C: Estimates of growth measures
Excluding Top 10%
Per capita total ethanol/alcohol -0.0371 ** -0.0581 *
sales 14 and older (0.0178) (0.0312)
Observations 1.598 1.598
[R.sup.2] 0.550 0.510
First stage F test 20.12 19.68
First stage F test p value 0.00 0.00
Panel D: Estimates of growth measures, 5-Year averages sample
Per capita total ethanol/alcohol -0.0216 * -0.0333
sales 14 and older (0.0121) (0.0254)
Observations 336 336
[R.sup.2] 0.431 0.201
First stage F test 6.99 7.60
First stage F test p value 0.01 0.00
(3) (4)
Dependent Variable PCPI Growth GDP Growth
Panel A: Estimates of growth on per capita total ethanol/alcohol sales
Per capita total ethanol/alcohol
sales 14 and older
Observations
[R.sup.2]
First stage F test
First stage F test p value
Panel B: Estimates of growth on per capita ethanol/alcohol sales 21
and older
Per capita total ethanol/alcohol
sales 21 and older
Observations
[R.sup.2]
First stage F-test
First stage F test p value
Panel C: Estimates of growth measures excluding potential outliers
Excluding Bottom 10%
Per capita total ethanol/alcohol -0.0286 * -0.0330
sales 14 and older (0.0150) (0.0225)
Observations 1.599 1.599
[R.sup.2] 0.525 0.508
First stage F test 11.55 14.24
First stage F test p value 0.00 0.00
Panel D: Estimates of growth measures, 5-Year averages sample
Per capita total ethanol/alcohol
sales 14 and older
Observations
[R.sup.2]
First stage F test
First stage F test p value
Notes: Robust standard errors corrected for clustering on the state
are in parentheses. All models control for full set of controls used
in column 4 of Table 4 as well as state- and year-fixed effects.
*** p < .01, ** p < .05. * p < .1.
REFERENCES
Acemoglu, D., and S. Johnson. "Disease and Development: The
Effect of Life Expectancy on Economic Growth." Journal of Political
Economy, 115(6), 2007, 925-85.
Ashraf, Q. H., A. Lester, and D. N. Weil. "When Does Improving
Health Raise GDP?" in NBER Macroeconomics Annual 2008, Vol. 23,
edited by D. Acemoglu, K. Rogoff, and M. Woodford. Chicago: University
of Chicago Press, 2009, 157-204.
Baker, D. B. "The Study of Stress at Work." Annual Review
of Public Health, 6, 1985, 367-81.
Barker, F. "'Consumption Externalities and the Role of
Government: The Case of Alcohol." New Zealand Treasury Working
Paper 02/25, 2002.
Beer Serves America. "Economic Impact." 2011. Accessed
January 14, 2012. http://www.beerservesamerica.org/
economic/default.aspx.
Berger. M.C., and J.P. Leigh. "The Effect of Alcohol Use on
Wages." Applied Economics, 20, 1988, 1343-51.
Blum, T. C., P.M. Roman, and J. K. Martin. "Alcohol
Consumption and Work Performance." Journal of Studies on Alcohol.
54(1), 1993, 61-70.
Bouchery, E. E., H. J. Harwood, J. J. Sacks, C. J. Simon, and R. D.
Brewer. "Economic Costs of Excessive Alcohol Consumption in the
U.S." American Journal of Preventive Medicine, 41(5), 2006, 516-24.
Bray, J. W. "Alcohol Use, Human Capital, and Wages."
Journal of Labor Economics, 23(2), 2005, 279-312.
Brenner, M. H., and A. Mooney. "Unemployment and Health in the
Context of Economic Change." Social Science Medicine, 17(16), 1983,
1125-38.
Centers for Disease Control and Prevention. "WISQARS Fatal
Injuries: Mortality Reports." 2004. Accessed December 15, 2011.
http://webappa.cdc.gov/sasweb/ ncipc/mortrate.html.
--. "Binge Drinking: Nationwide Problem, Local
Solutions." 2012. Accessed December 15, 2011.
http://www.cdc.gov/vitalsigns/BingeDrinking/#.
Chaloupka F. J., M. Grossman, and H. Saffer. "The Effects of
Price on Alcohol Consumption and Alcohol-Related Problems." Alcohol
Research and Health, 26(1), 2002, 22-34.
Chaloupka F. J., H. Saffer, and M. Grossman "Alcohol-Control
Policies and Motor Vehicle Fatalities." Journal of Legal Studies,
22(1), 1993, 161-86.
Chatterji, P., and J. DeSimone. "High School Alcohol Use and
Young Adult Labor Market Outcomes." National Bureau of Economic
Research Working Paper No. 12529. 2006.
Chesson, H., P. Harrison, and W. J. Kassler. "Sex Under the
Influence: The Effect of Alcohol Policy on Sexually Transmitted Disease
Rates in the United States." Journal of Law and Economics, 43(1),
2000, 215-38.
Collins, D. J., and H. M. Lapsley. "The Social Costs of Drug
Abuse in Australia in 1988 and 1992." National Drug Strategy
Monograph Series, 30, 1996.
Cook, P. J. Paying the Tab: The Costs and Benefits of Alcohol
Control. Princeton, NJ: Princeton University Press, 2007.
Cook, P. J., and M. J. Moore. "This Tax's for You: The
Case for Higher Beer Taxes." National Tax Journal, 47(3), 1994,
559-73.
--. "The Economics of Alcohol Abuse and Alcohol Control
Policies." Health Affairs, 21(2), 2002, 120-33.
Dee, T. S. "Alcohol Abuse and Economic Conditions: Evidence
from Repeated Cross-Sections of Individual-Level Data." Health
Economics, 10(3), 2001, 257-70.
Distilled Spirits Council of the United States (DISCUS). History of
Beverage Alcohol Tax Changes. Washington, DC: DISCUS, 1999.
Ettner, S. L. "Measuring the Human Cost of a Weak Economy:
Does Unemployment Lead to Alcohol Abuse?" Social Science Medicine,
44(2), 1997, 251-60.
Fenwick, R., and M. Tausig. "The Macroeconomic Context of Job
Stress." Journal of Health and Social Behavior, 35(3), 1994,
266-82.
Freeman, D. G. "Alternative Panel Estimates of Alcohol Demand,
Taxation and the Business Cycle." Southern Economic Journal, 67(2),
2000, 325-44.
French, M. T., and G. A. Zarkin. "Is Moderate Alcohol Use
Related to Wages? Evidence from Four Worksites." Journal of Health
Economics, 14(3), 1995, 319-44.
Greenfield, T. K., and J. D. Rogers. "Who Drinks Most of the
Alcohol in the U.S.? The Policy Implications." Journal of Studies
on Alcohol, 60(1), 1999, 78-89.
Gronbaek, M. "Factors Influencing the Relation between Alcohol
and Mortality With Focus on Wine." Journal of Internal Medicine,
250, 2001, 291-308.
Gwartney, J. D., and R. A. Lawson. "The Impact of Tax Policy
on Economic Growth, Income Distribution, and Allocation of Taxes."
Social Philosophy and Policy, 23, 2006, 28-52.
Hamilton, V., and B. Hamilton. "Alcohol and Earnings: Does
Drinking Yield a Wage Premium?" Canadian Journal of Economics,
30(1), 1997, 135-51.
Harwood, H. J., D. M. Napolitano, P. L. Kristiansen, and J. J.
Collins, eds. Economic Costs to Society of Alcohol and Drug Abuse and
Mental Illness. Research Triangle Park, NC: Research Triangle Institute,
1984.
Hilton, M. E., and W. B. Clark. "Changes in American Drinking
Patterns and Problems, 1967-1984." Journal of Studies on Alcohol,
48(6), 1987, 515-22.
Hingson, R., and M. Winter. Epidemiology and Consequences of
Drinking and Driving. Bethesda, MD: National Institute on Alcohol Abuse
and Alcoholism of the National Institute of Health, 2003.
Hoftmeister, H., F. P. Schelp, G. B. Mensink, E. Dietz, and D.
Bohning. "The Relationship Between Alcohol Consumption, Health
Indicators and Mortality in the German Population." International
Journal of Epidemiology, 28(6), 1999, 1066-72.
Johansson, E., P. Bockerman, and A. Uutela. "Alcohol
Consumption and Sickness Absence, Evidence from Micro-Data."
European Journal of Public Health, 19(1). 2009, 19-22.
Kalemli-Ozcan, S. "Does the Mortality Decline Promote Economic
Growth?" Journal of Economic Growth, 7(4), 2002, 311-49.
Kalemli-Ozcan, S., H. E. Ryder, and D. Weft. "Mortality
Decline, Human Capital Investment, and Economic Growth." Journal of
Development Economics, 62(1), 2000, 1-23.
Karasek, R. A., and T. Theorell. Healthy Work: Stress,
Productivity, and the Reconstruction of Working Life. New York: Basic
Books, 1990.
Kenkel. D.S., and P. Wang. "Are Alcoholics in Bad Jobs?"
in The Economic Analysis of Substance Use and Abuse, edited by F. J.
Chaloupka, M. Grossman, W. K. Bickel, and H. Saffer. Chicago: University
of Chicago Press, 1999, 251-78.
Klatsky, A. L. "Alcohol and Cardiovascular Diseases: A
Historical Overview." Annals of the New York Academy of Sciences,
957, 2002, 7-15.
LaVallee, R. A., and H. Yi. "Apparent Per Capita Alcohol
Consumption: National, State, and Regional Trends, 1977-2009."
National Institute on Alcohol Abuse and Alcoholism, 2011. Accessed
January 20, 2012. http://pubs.niaaa.nih.gov/publications/Surveillance92/
CONS09.pdf.
Lehman, W. E. K., and D. D. Simpson. "Employee Substance Use
and On-the-Job Behaviors." Journal of Applied Psychology, 77(3),
1992, 309-21.
Levitt, S. D., and J. Porter. "How Dangerous Are Drinking
Drivers?" Journal of Political Economy, 109(61), 2001, 1198-237.
Manning, W.G., E.B. Keeler, J.P. Newhouse, E. M. Sloss, and J.
Wasserman. "The Taxes of Sin: Do Smokers and Drinkers Pay Their
Way?" Journal of the American Medical Association, 261(11), 1989,
1604-9.
Markowitz, S. "The Effectiveness of Cigarette Regulations in
Reducing Cases of Sudden Infant Death Syndrome." Journal of Health
Economics, 27(1), 2008, 106-33.
Markowitz, S., R. Kaestner, and M. Grossman. "An Investigation
of the Effects of Alcohol Consumption and Alcohol Policies on Youth
Risky Sexual Behaviors." American Economic Review, 95(2), 2005,
263-66.
McDonald, Z., and M. A. Shields. "The Impact of Alcohol
Consumption on Occupational Attainment in England." Economica, 68,
2001, 455-63.
Murray, C. J. L., and A. D. Lopez. The Global Burden of Disease: A
Comprehensive Assessment of Mortality and Disability from Diseases,
Injuries and Risk Factors in 1990 and Projected to 2020. Global Burden
of Disease and Injury Series, Vol. I. Cambridge, MA: Harvard School of
Public Health on behalf of the World Health Organization and the World
Bank, 1996.
National Institute on Alcohol Abuse and Alcoholism. Apparent Per
Capita Ethanol Consumption for the United States, 1970-2007. 2010.
Accessed September 20, 2011.
http://www.niaaa.nih.gov/Resources/DatabaseResources/QuickFacts/AlcoholSales/cons um01.htm.
Norstrom T. "Per Capita Alcohol Consumption and Sickness
Absence." Addiction, 101(10), 2006, 1421-27.
Orzechowski, W., and R. C. Walker. The Tax Burden on Tobacco, Vol.
40. Arlington, VA: Orzechowski & Walker, 2005.
Parry, I. W. H., S. E. West, and R. Laxminarayan. "Fiscal and
Externality Rationales for Alcohol Policies." B.E. Journal of
Economic Analysis & Policy, 9(1), 2009, 1-48.
Pierce, R. S., M. R. Frone, M. Russell, and M. L. Cooper.
"Relationship of Financial Strain and Psychosocial Resources to
Alcohol Use and Abuse: The Mediating Role of Negative Affect and
Drinking Motives." Journal of Health and Social Behavior, 35(4),
1994, 291-308.
Rashad, I., and R. Kaestner. "Teenage Sex, Drugs and Alcohol
Use: Problems Identifying the Cause of Risky Behaviors." Journal of
Health Economics, 23(3), 2004, 493-503.
Rehm, J., G. Gmel, C.T. Sempos, and M. Trevisan.
"Alcohol-Related Morbidity and Mortality." Alcohol Research
and Health, 27(1), 2003, 39-51.
Rice, D. P., S. Kelman, L. S. Miller, and S. Dunmeyer. The Economic
Costs of Alcohol and Drug Abuse and Mental Illness: 1985. Report
submitted to the Office of Financing and Coverage Policy of the Alcohol,
Drug Abuse, and Mental Health Administration (US Department of Health
and Human Services, Institute for Health & Aging, University of
California, San Francisco), 1990.
Room, R., and D. Jernigan. "The Ambiguous Role of Alcohol in
Economic and Social Development." Addiction, 95(12s4), 2000,
523-35.
Rosenbaum, D.P., S.F. Bennett, B.D. Lindsay. D. L. Wilkinson, B.D.
Davis, C. Taranowski, and P. J. Lavrakas. Executive Summary: The
Community, Responses to Drug Abuse National Demonstration Program Final
Process Report. Chicago, IL: Center for Research in Law and Justice,
University of Illinois, 1992.
Ruhm, C. J. "Economic Conditions and Alcohol Problems."
Journal of Health Economics, 14(5), 1995, 583-603.
--. "Alcohol Policies and Highway Vehicle Fatalities."
Journal of Health Economics, 15(4), 1996, 435-54.
Ruhm, C.J., and W. E. Black. "Does Drinking Really Decrease in
Bad Times?" Journal of Health Economics, 21, 2002, 659-78.
Ruhm, C.J., A.S. Jones, W.C. Kerr, T.K. Greenfield, J. V. Terza, R.
S. Pandian, and K. McGeary. "What U.S. Data Should Be Used to
Measure the Price Elasticity of Demand for Alcohol." National
Bureau of Economic Research Working Paper No. 17578, 2011.
Skog, O.-J. "An Analysis of Divergent Trends in Alcohol
Consumption and Economic Development." Journal of Studies on
Alcohol, 47(1), 1986, 19-25.
Sloan, F. A., B.A. Reilly, and C. Schenzler. "Effects of Tort
Liability and Insurance on Heavy Drinking and Drinking and
Driving." Journal of Law and Economics, 38(1), 1995, 49-78.
Smith G. S., C. C. Branas, and T. R. Miller. "Fatal
Non-traffic Injuries Involving Alcohol: A Meta-Analysis." Annals of
Emergency Medicine, 33(5), 1999, 659-68.
Sokejima, S., and S. Kagamimori. "Working Hours as a Risk
Factor for Acute Myocardial Infarction in Japan: A Case-Control
Study." British Medical Journal, 317(19), 1998, 775-80.
Stockwell, T., and D. Crosbie. "Supply and Demand for Alcohol
in Australia: Relationships Between Industry Structures, Regulation and
the Marketplace." International Journal of Drug Policy, 12(2),
2001, 139-52.
Tax Foundation. Tax Data. 2011. Accessed November 18, 2011.
http://www.taxfoundation.org/news/show/245. html.
Tekin, E. "Employment, Wages, and Alcohol Consumption in
Russia." Southern Economic Journal, 71(2), 2004, 397-417.
Thun, M.J., R. Peto, A. D. Lopez, J. H. Monaco, S. J. Henley, C. W.
Heath Jr., and R. Doll. "Alcohol Consumption and Mortality among
Middle-Aged and Elderly U.S. Adults." New England Journal of
Medicine, 337(24), 1997, 1705-14.
Tomljanovich, M. "The Role of Fiscal Policy in State Economic
Growth." Contemporary Economic Policy, 22 (3), 2004, 318-30.
Traynor, T. L. "The Impact of Driver Alcohol Use on Crash
Severity: A Crash Specific Analysis." Transportation Research: Part
E: Logistics and Transportation Review, 41(5), 2005, 421-37.
United State Census Bureau. The Statistical Abstract (various
years). Accessed December 14, 2011. http://www.
census.gov/compendia/statab/.
U.S. Department of Health and Human Services Report. The Economic
Costs of Alcohol Abuse in the United States. 2000. Accessed July 17,
2008. http://pubs.niaaa. nih.gov/publications/economic-2000/index.htm.
West, R., J. Wilding, D. French, R. Kemp, and A. Irving.
"Effect of Low and Moderate Doses of Alcohol on Driving Hazard
Perception Latency and Driving Speed." Addiction, 88(4), 1993,
527-32.
Winton, M., N. Heather, and I. Robertson. "Effects of
Unemployment on Drinking Behavior: A Review of the Relevant
Evidence." International Journal of Addictions, 21(12), 1986,
1261-83.
World Health Organization. "Mental Health: New Understanding,
New Hope." World Health Report 2001. 2001.
Young, D. J., and A. Bielinska-Kwapisz. "Alcohol Taxes and
Beverage Prices." National Tax Journal, 55(1), 2002, 57-73.
--. "Alcohol Consumption, Beverage Prices and Measurement
Error." Journal of Studies on Alcohol, 64(2), 2003, 235-38.
* We thank Rexford Santerre, Jorge Martinez, Sally Wallace, Neven
Valev, Angela Snyder, Dhaval Dave, Robert E. Rosenman, two anonymous
reviewers, and participants at the 2009 Eastern Economic Association
meetings for their valuable comments and suggestions. We are grateful to
Sara Markowitz for allowing us to use her data on cigarette sales and
taxes from Markowitz (2008). We are grateful to Marc Tomljanovich for
allowing us to use his data from Tomljanovich (2004) in the earlier
versions of this paper. R.C. thanks the Center for Real Estate and Urban
Economic Studies of the Business School at the University of Connecticut
for the statistical software purchase.
Cesur: Assistant Professor, Department of Finance, School of
Business, University of Connecticut, Storrs, CT 06269. Phone
1-860-486-6315, Fax 1-860 486-0634, E-mail
[email protected]
Kelly: Associate Professor, Department of Economics, Queens College
of the City University of New York, Flushing, NY 11367. Phone
1-718-997-5440, Fax 1-718997-5466, E-mail
[email protected]
(1.) See http://www.beerservesamerica.com/about.aspx.
(2.) See http://www.discus.org/about/.
(3.) This statistic takes into account ethanol content and has
already made the conversion; since beer has a relatively low ethanol
content compared to other alcoholic beverages, if raw sales not taking
ethanol into account were looked at, this percentage would be much
higher. The Alcohol Epidemiologic Data System uses an estimate of
average ethanol content in the alcoholic beverages to convert the
gallons of sold or shipped beer, wine, and spirits into gallons of
ethanol--pure alcohol--before calculating per capita consumption
estimates. The ethanol conversion coefficients are 0.045 for beer, 0.129
for wine, and 0.411 for spirits (LaValee and Yi 2011).
(4.) There are 18 such "control" states for liquor and
five such states for wine (Ruhm et al. 2011).
(5.) See Chaloupka, Saffer, and Grossman (1993), Cook (2007), Cook
and Moore (1994, 2000), Freeman (2000), Markowitz, Kaestner, and
Grossman (2005), Ruhm (1996), and Young and Bielinska-Kwapisz (2002,
2003). Note that we also estimate income growth measures on per capita
beer consumption controlling for per capita wine and liquor consumption
as well as total ethanol consumption in Section VIII as a robustness
check. Results from the robustness check exercises, presented in Tables
A1 and A2, have similar implications.
(6.) We also conduct a falsification exercise using cigarette
consumption and cigarette excise taxes similar to that conducted by
Rashad and Kaestner (2004). This exercise reveals that cigarette
consumption has no significant effect on per capita GDP or per capita
personal income (results available upon request). Since smoking is not
without its negative externalities, this exercise is comforting in that
it may arguably suggest that the results we obtain are conservative.
Data for this exercise on cigarette excise taxes and per capita
cigarette sales come from Markowitz (2008), who obtained data from
Orzechowski and Walker (2005).
(7.) In all specifications, we control for state-fixed effects.
Therefore, by OLS we refer to an OLS with state-fixed effects
methodology. The term OLS is used for simplicity.
(8.) A host of other factors can be listed as the determinants of
economic growth. For the sake of providing a clear argument, we restrict
the discussion to the factors more relevant to our study.
(9.) See French and Zarkin (1995), Johansson, Bockerman, and Uutela
(2009), Lehman and Simpson (1992), Norstrrm (2006), and Rosenbaum et al.
(1992).
(10.) See Blum, Roman, and Martin (1993), Harwood et al. (1984),
Hilton and Clark (1987), Kenkel and Wang (1999), and Rice et al. (1990).
(11.) See Chesson, Harrison, and Kassler (2000), Hingson and Winter
(2003), Levitt and Porter (2001), Markowitz, Kaesmer, and Grossman
(2005), Murray and Lopez (1996), Traynor (2005), and West et al. (1993).
(12.) While one can argue that money is being put into circulation
through injury treatment and rebuilding expenditures, this money could
be better put to use productively through investment in new business
enterprises. We thank an anonymous reviewer for pointing this out.
(13.) They instrument for life expectancy using predicted
mortality, which is created by interacting baseline mortality rates with
exogenous worldwide health shocks.
(14.) As discussed above, the impact of mortality and longevity may
have differential effects in the short and long run. The focus of this
study is on short-run income growth.
(15.) See Berger and Leigh (1988), Bray (2005), Chatterji and
DeSimone (2006), French and Zarkin (1995), Hamilton and Hamilton (1997),
McDonald and Shields (2001), and Tekin (2004).
(16.) Our aggregate models do not distinguish between alcohol use
and alcohol abuse. However, our measures are a good reflection of abuse,
as those who abuse alcohol are responsible for most of its consumption
(Greenfield and Rogers 1999). Our 2007 data indicate that there is a
strong, positive correlation between per capita beer consumption and
binge drinking (males having five or more drinks on one occasion or
females having four or more drinks on one occasion, obtained from the
Behavioral Risk Factor Surveillance System), significant at the 1%
level.
(17.) In addition to the state and year-fixed effects,
state-specific time trends could be specified. However, beer excise
taxes exhibit strong long-term downward trends since the 1970s, with the
exception of the 1990 federal excise tax increases. In particular, the
state and time-fixed effects along with individual state time trends and
time-fixed effects explain 97.2% of the variation in the beer tax.
Therefore, because state and regional trends are strongly collinear with
the instrument utilized, we do not control for state-specific time
trends in our models.
(18.) The current value of the state-level PCPI (or per capita GDP)
is jointly determined with the growth rate; therefore, lagged values of
PCPI (or per capita GDP) are utilized in the specifications to account
for the effect of convergence as well as the effect of PCPI on alcohol
use. An alternative is to use initial per capita personal income.
Because state-fixed effects are controlled for, the initial per capita
personal income cannot be used. A common practice in aggregate level
income growth estimations is to average the data over 5-year intervals
to account for the effect of business cycles, which last about 4 to 5
years. In such specifications, the initial per capita income for each
5-year period is controlled for to account for the convergence effect.
Because we restrict our analysis to yearly data to capture the changes
in beer use by utilizing annual state-level per capita beer consumption,
we do not use 5-year averages in our main estimates; nevertheless, in
Appendix Tables A1 and A2 we present results from regression models
using 5-year averages for the period 1971-2005. The results obtained
from these specifications are similar to our main results.
(19.) See, for example, Brenner and Mooney (1983), Pierce et al.
(1994), and Winton, Heather, and Robertson (1986).
(20.) See Baker (1985), Fenwick and Tausing (1994), Karasek and
Theorell (1990), and Sokejima and Kagamimori (1998).
(21.) Alternative instruments, such as dry county laws, were
considered. However, dry county laws are not valid instruments in our
framework because they may have a direct impact on business activity,
since they automatically crowd out some businesses out of the county.
People from dry counties may go to bars and restaurants in neighboring
counties or states; thus, dry county laws may diminish business activity
in addition to reducing alcohol use in the county and/or state.
Moreover, because time variation in such laws is relatively limited,
state-fixed effects generally capture the impact of such laws.
(22.) Instead of alcohol excise taxes, alcohol prices may be
employed as instruments. In our framework, prices of alcoholic beverages
are not suitable instruments since it is likely that beverage prices may
be influenced by economic growth (or the business cycle). Because the
potential relationship between beverage prices and income growth
violates the exclusion restriction, alcohol prices are not used as
instruments. In addition, the most commonly used price data, which come
from the ACCRA, are subject to significant measurement error (Ruhm et
al. 2011; Young and Bielinka-Kwapisz 2002).
(23.) Table 5 presents the reduced form regressions of income
growth on alcohol excise tax rates. The results, presented and discussed
in Section VII, provide evidence in favor of the view that alcohol
excise tax rates are suitable instruments in our framework.
(24.) This is done to present the taxes that individuals face, for
the purpose of showing accurate means. However, since federal excise
taxes are constant across states, all variations are due to the
state-year variation in state excise taxes. Results are robust to using
state-level beer excise tax rates excluding the federal beer excise tax
rate.
(25.) Beer excise tax elasticity of per capita beer consumption =
[(-0.3561 / 1.299) * 100]/[(1/0.581) * 100] = -0.0163.
(26.) We also include cigarette excise tax rates in this model for
the sake of completeness, as we used these taxes in the falsification
exercise outlined in footnote 6.
(27.) NIAAA data contain information on gallons of alcoholic
beverage (beer, wine, and liquor) sales per capita as well. Estimates
based on per capita alcoholic beverage sales measures are very similar
to those obtained by utilizing per capita ethanol sales measures.
(28.) A one standard deviation increase in per capita beer sales,
which corresponds to 18.5%, is a very large increase given that annual
changes are relatively small; the annual period average is 0.33% and in
most years, in absolute value, it is less than 2%.
(29.) Gwartney and Lawson (2006), for example, estimate that a 10
percentage point reduction in a country's top marginal tax rate
(mainly for marginal tax rates in excess of 50%) will increase the
economic growth rate by approximately three-tenths of a percentage
point.
TABLE 1
Descriptive Statistics
Variable All Years 1970s 1980s
Per capita beer sales (14 plus), in 1.299 1.279 1.363
gallons (0.241) (0.277) (0.247)
Per capita wine sales (14 plus), in 0.296 0.272 0.320
gallons (0.154) (0.155) (0.163)
Per capita liquor sales (14 plus), 0.872 1.132 0.929
in gallons (0.403) (0.531) (0.362)
Per capita ethanol/Alcohol sales 2.468 2.683 2.611
(14 plus), in gallons (0.668) (0.854) (0.671)
CPI-adjusted beer excise tax per 0.581 0.897 0.470
gallon (levels, 1982-1984 dollars) (0.276) (0.355) (0.161)
CPI-adjusted wine excise tax per 0.963 1.107 0.698
gallon (levels, 1982-1984 dollars) (0.526) (0.755) (0.432)
CPI-adjusted liquor excise tax per 13.836 23.406 12.731
gallon (levels, 1982-1984 dollars) (6.322) (5.200) (1.810)
CPI-adjusted cigarette excise tax 3.820 3.837 2.825
per 10 packs (levels, 1982-1984 (1.774) (1.034) (0.655)
dollars)
Per capita personal income in 13.984 11.235 12.700
thousands (1982-1984 dollars) (3.047) (1.548) (1.984)
Per capita personal income growth 0.017 0.022 0.016
(0.028) (0.038) (0.029)
Total tax to state personal income 0.064 0.062 0.063
ratio (0.012) (0.011) (0.013)
Total public expenditure to state 0.103 0.082 0.079
personal income ratio (0.037) (0.017) (0.018)
Per capita gross domestic product 16.648 13.737 15.067
(GDP) in thousands (1982-1984 (3.833) (2.133) (2.771)
dollars)
Per capita GDP growth 0.016 0.022 0.012
(0.036) (0.044) (0.044)
Total tax to state GDP ratio 0.054 0.051 0.053
(0.010) (0.009) (0.009)
Total public expenditure to state 0.087 0.067 0.067
GDP (0.032) (0.013) (0.014)
Regional inflation 0.045 0.070 0.053
(0.028) (0.026) (0.032)
Percent of population (<18 and >64) 0.395 0.423 0.391
(0.025) (0.021) (0.021)
Percent of population (>17 and <25) 0.113 0.129 0.122
(0.016) (0.009) (0.012)
Percent with less than high school 0.232 0.355 0.263
degree (0.105) (0.076 (0.073)
Percent with a high school degree 0.355 0.343 0.364
(0.042) (0.036) (0.041)
Percent with some college or 0.210 0.153 0.188
vocational degree (0.060) (0.042) (0.044)
Percent with a college degree 0.202 0.143 0.185
(0.061) (0.032) (0.041)
Percent White 0.842 0.907 0.866
(0.116) (0.067) (0.095)
Percent Black 0.096 0.081 0.090
(0.092) (0.069) (0.088)
Percent other race 0.062 0.012 0.044
(0.080) (0.020) (0.053)
Observations 1776 432 480
Variable 1990s 2000s
Per capita beer sales (14 plus), in 1.283 1.263
gallons (0.212) (0.209)
Per capita wine sales (14 plus), in 0.274 0.324
gallons (0.132) (0.159)
Per capita liquor sales (14 plus), 0.703 0.721
in gallons (0.254) (0.229)
Per capita ethanol/Alcohol sales 2.260 2.308
(14 plus), in gallons (0.497) (0.465)
CPI-adjusted beer excise tax per 0.522 0.440
gallon (levels, 1982-1984 dollars) (0.120) (0.090)
CPI-adjusted wine excise tax per 1.107 0.954
gallon (levels, 1982-1984 dollars) (0.385) (0.288)
CPI-adjusted liquor excise tax per 10.559 8.548
gallon (levels, 1982-1984 dollars) (1.516) (1.227)
CPI-adjusted cigarette excise tax 3.435 5.525
per 10 packs (levels, 1982-1984 (1.074) (2.658)
dollars)
Per capita personal income in 14.855 17.591
thousands (1982-1984 dollars) (2.077) (2.330)
Per capita personal income growth 0.016 0.016
(0.020) (0.019)
Total tax to state personal income 0.066 0.064
ratio (0.012) (0.012)
Total public expenditure to state 0.111 0.147
personal income ratio (0.033) (0.030)
Per capita gross domestic product 17.678 20.612
(GDP) in thousands (1982-1984 (2.934) (3.582)
dollars)
Per capita GDP growth 0.017 0.013
(0.026) (0.023)
Total tax to state GDP ratio 0.056 0.055
(0.010) (0.011)
Total public expenditure to state 0.093 0.127
GDP (0.028) (0.028)
Regional inflation 0.029 0.027
(0.010) (0.007)
Percent of population (<18 and >64) 0.389 0.376
(0.018) (0.012)
Percent of population (>17 and <25) 0.099 0.101
(0.009) (0.010)
Percent with less than high school 0.175 0.126
degree (0.053) (0.037)
Percent with a high school degree 0.364 0.345
(0.042) (0.042)
Percent with some college or 0.239 0.264
vocational degree (0.044) (0.040)
Percent with a college degree 0.222 0.265
(0.045) (0.051)
Percent White 0.826 0.760
(0.109) (0.135)
Percent Black 0.103 0.114
(0.100) (0.104)
Percent other race 0.071 0.127
(0.071) (0.110)
Observations 480 384
Note: Values in parentheses are standard deviations.
TABLE 2
OLS Estimates of State-Level PCPI Growth on Alcohol Consumption
Sales Measures
(1) (2)
Panel A: Estimates of PCPI Growth on Per Capita Beer Sales
Per capita beer sales -0.0071 -0.0079 *
(0.0047) (0.0046)
Lagged per capita personal income -0.0067 ** -0.0063 **
(0.0026) (0.0026)
State total tax PI ratio -0.4743 ** -0.4884 **
(0.1927) (0.1840)
State total expenditure to PI ratio -0.2187 ** -0.2025 **
(0.0961) (0.0975)
Regional inflation -0.7546 *** -0.7968 ***
(0.1182) (0.1148)
Percent of population (<18 and >64) -0.1955 *** -0.0747
(0.0722) (0.0979)
Percent of population (> 17 and <25) 0.2936 **
(0.1360)
Percent with a high school degree
Percent with some college or
vocational degree
Percent with a college degree
Percent Black
Percent other race
[R.sup.2] 0.565 0.567
Panel B: Dependent Variable Is Per Capita GDP Growth
Per capita beer sales 0.0048 0.0031
(0.0101) (0.0097)
[R.sup.2] 0.537 0.540
Observations 1.776 1,776
Number of states 48 48
(3) (4)
Panel A: Estimates of PCPI Growth on Per Capita Beer Sales
Per capita beer sales -0.0034 -0.0048
(0.0050) (0.0050)
Lagged per capita personal income -0.0069 ** -0.0072 **
(0.0029) (0.0030)
State total tax PI ratio -0.4976 *** -0.4716 **
(0.1807) (0.1780)
State total expenditure to PI ratio -0.1974 ** -0.1982 **
(0.0949) (0.0952)
Regional inflation -0.7910 *** -0.7693 ***
(0.1135) (0.1084)
Percent of population (<18 and >64) -0.1059 -0.1097
(0.1039) (0.1050)
Percent of population (> 17 and <25) 0.2871 ** 0.2995 **
(0.1403) (0.1412)
Percent with a high school degree -0.0169 -0.0200
(0.0227) (0.0223)
Percent with some college or -0.0725 -0.0701
vocational degree (0.0465) (0.0493)
Percent with a college degree 0.0216 0.0186
(0.0393) (0.0380)
Percent Black 0.0350
(0.0278)
Percent other race 0.0014
(0.0149)
[R.sup.2] 0.569 0.570
Panel B: Dependent Variable Is Per Capita GDP Growth
Per capita beer sales 0.0085 0.0072
(0.0098) (0.0098)
[R.sup.2] 0.542 0.542
Observations 1,776 1,776
Number of states 48 48
Notes: Robust standard errors corrected for clustering on the state
are in parentheses. All models control for state- and year-fixed
effects. In Panel B, each column includes the same control
variables that are specified in the corresponding column shown in
Panel A.
*** p < .01, * p < .05. * P < .1.
TABLE 3
First-Stage Estimates of Per Capita Beer Sales
on Excise Taxes
Variables Beer
Real beer excise tax -0.3651 ***
(0.0519)
Lagged per capita personal income 0.0019
(0.0079)
State total tax PI ratio 0.5815
(0.8395)
State total expenditure to PI ratio -0.1764
(0.4731)
Regional inflation -0.5616
(0.5383)
Percent of population (<18 and >64) -0.5647
(1.1658)
Percent of population (> 17 and <25) 3.1545 **
(1.3623)
Percent with a high school degree 0.9040 ***
(0.2168)
Percent with some college or vocational degree 0.9452 ***
(0.2568)
Percent with a college degree 0.0917
(0.2317)
Percent Black 0.1719
(0.2973)
Percent other race -0.0978
(0.2013)
Observations 1,776
[R.sup.2] 0.565
Number of states 48
Notes: Robust standard errors corrected for clustering on
the state are in parentheses. All models control for state- and
year-fixed effects.
*** p < .01, ** p < .05, * p < .1.
TABLE 4
Instrumental Variable Estimates of Per Capita Income Growth and
GDP Growth on Beer Sales
(1) (2)
Panel A: PCPI Growth
Per capita beer sales -0.0352 ** -0.0303 **
(0.0142) (0.0142)
[R.sup.2] 0.551 0.555
First stage F test 105.962 106.104
First stage F test P value 0.000 0.000
Panel B: Per Capita GDP Growth
Per capita beer sales -0.0494 ** -0.0395 *
(0.0209) (0.0209)
[R.sup.2] 0.523 0.532
First stage F test 91.669 100.190
First stage F test p value 0.000 0.000
Observations 1.776 1.776
Number of states 48 48
Controls for
Lagged PCPI Yes Yes
Total tax to personal income ratio Yes Yes
Total expenditure to personal income ratio Yes Yes
Regional inflation Yes Yes
Percent of population (< 18 and >64) Yes Yes
Percent of population (> 17 and <25) No Yes
Education controls No No
Race controls No No
(3) (4)
Panel A: PCPI Growth
Per capita beer sales -0.0322 ** -0.0315 **
(0.0141) (0.0143)
[R.sup.2] 0.565 0.565
First stage F test 105.544 100.175
First stage F test P value 0.000 0.000
Panel B: Per Capita GDP Growth
Per capita beer sales -0.0508 * -0.0640 *
(0.0298) (0.0329)
[R.sup.2] 0.528 0.522
First stage F test 72.914 56.772
First stage F test p value 0.000 0.000
Observations 1,776 1,776
Number of states 48 48
Controls for
Lagged PCPI Yes Yes
Total tax to personal income ratio Yes Yes
Total expenditure to personal income ratio Yes Yes
Regional inflation Yes Yes
Percent of population (< 18 and >64) Yes Yes
Percent of population (> 17 and <25) Yes Yes
Education controls Yes Yes
Race controls No Yes
Notes: Robust standard errors corrected for clustering on the state are
in parentheses. All models control for state- and year-fixed effects.
*** p < .01. ** p < .05. * P < .1.
TABLE 5
Reduced Form Estimates of PCPI Growth and Per Capita GDP Growth on
Excise Taxes
(1) (2) (3)
Panel A: Estimates of PCPI Growth
Real beer excise tax 0.0167 **
(0.0081)
Real wine excise tax 0.0016
(0.0015)
Real liquor excise tax 0.0002
(0.0007)
Real cigarette excise tax
[R.sup.2] 0.573 0.570 0.570
Panel B: Estimates of GDP Growth
Real beer excise tax 0.0238 *
(0.0125)
Real wine excise tax 0.0001
(0.0025)
Real liquor excise tax 0.0014
(0.0011)
Real cigarette excise tax
[R.sup.2] 0.545 0.542 0.543
Observations 1,776 1,776 1,776
Number of states 48 48 48
(4) (5)
Panel A: Estimates of PCPI Growth
Real beer excise tax 0.0158 *
(0.0082)
Real wine excise tax 0.0010
(0.0012)
Real liquor excise tax -0.0003
(0.0008)
Real cigarette excise tax 0.0007 0.0005
(0.0005) (0.0005)
[R.sup.2] 0.571 0.573
Panel B: Estimates of GDP Growth
Real beer excise tax 0.0227 *
(0.0124)
Real wine excise tax -0.0028
(0.0021)
Real liquor excise tax 0.0014
(0.0011)
Real cigarette excise tax 0.0010 0.0007
(0.0008) (0.0008)
[R.sup.2] 0.542 0.546
Observations 1,776 1,776
Number of states 48 48
Notes: Robust standard errors corrected for clustering on the state
are in parentheses. All models control for full set of controls used
in column 4 of Table 4 as well as state- and year-fixed effects.
*** p < .01, ** p < .05, * p < .1.
