Historical Differences in State Tax Policy
Transcript of Historical Differences in State Tax Policy
Variances in State Sales Tax Rates:A Study of the Source of Sales Tax Differences in the States
Sean Connell
Will Monkowski
Brian Prewitt
Political Science 425
Dr. Zachary Baumann
April 13, 2012
What causes variances in state sales tax rates?
What seems like a few cents at the cash register can really add up. Sales tax plays a role
in the fiscal lives of most -- but not all -- Americans nearly every day of their lives. Whether it is
the 2.9% tax added on to transactions in Colorado or the 7.25% tax in California, the rate of
sales tax levied by the states across the United States varies greatly. There are even 5 states that
don’t have a sales tax at all. While some areas make up for smaller sales tax by instituting a
higher local sales tax, theses variances have implications in a number of areas. One of the most
prevalent aspects of state government that sales tax affects is the budgeting process. Some states
earn up to 60% of their revenue from sales tax, while others earn 0% (Book of States, 2010). In
fact, sales tax accounts for one-third of the entire total of state revenue in the U.S. (Fletcher &
Murray, 2006). Since so much of state revenue is derived from sales tax, the state’s budget will
inevitably be impacted in some manner by the sales tax rate. As the graph below (A) shows, the
most common sales tax rate in the nation is around 6%, with most other states within 1% of this
figure. Through our research, we aimed to find what can account for these differences in sales
tax rates and in the end, we found that we’ve only scratched the surface of sourcing these
variances, but we did find at least one factor that has an impact.
Graph A: State Sales Tax Rate Distribution
Overview of Literature
In order to fully understand what might cause a state to adopt a sales tax at a particular
rate, it is important to understand why a state would adopt a sales tax in the first place. A natural
first place to look within the literature was at the history of sales tax, specifically the first use of
it and the reasoning behind the implementation. Howe and Reeb surveyed the evolution of state
and local tax systems over the time period beginning with the colonial era and extending to the
present day tax system. They focused on how governments have found ways to access the
wallets of their tax base by implementing alternative taxes, one of which was the sales tax. They
pointed out that although various excise and sin taxes had existed since 1919, when the first
motor fuel tax was implemented, the first retail sales tax was created in 1932 in Mississippi at a
meager 2% rate. Howe and Reeb added that within 6 years, 20 states had implemented the tax
due to its success. Since this was during the Great Depression, local and state governments
weren’t able to raise the classical taxes -- politically speaking -- such as property and income
taxes. Sales taxes provided a way to generate tax revenue a little at a time from consumers, so
they wouldn’t be as outraged by the increase in taxation since they spike was less noticeable
(Howe & Reeb, 1997). Based off of this study, we discovered that sales tax seems to have a close
relationship with the classical taxes, as it was originally implemented as an alternative source of
income from those. They made a point to emphasize the close relationship sales tax shares with
income tax, which seemed to be a common point that emerged amongst the rest of the literature.
Numerous studies have shown that sales taxes have also been used as a means of enacting
various types of social and health changes. Peterson, Zeger, Remington, and Anderson found that
over the 33-year span preceding 1988, 249 increases in the tax rates on cigarettes were associated
with a decrease by 3.0 packs per capita in the amount of cigarettes consumed by Americans.
Over the same time period, when there were no tax increases implemented, cigarette
consumption actually rose by 0.6 packs per capita (Peterson, Zeger, Remington & Anderson,
1992). In recent years, studies have been conducted regarding potential associations in taxes on
soft drinks and the rates of obesity in the United States. In 2009, Fletcher, Frisvold, and Tefft
conducted a study measuring taxes on soft drinks and the national average of body mass index
(BMI). They actually found that a 1% increase in these taxes led to a .001 decrease in national
BMI. While these figures are relatively miniscule, this is a relatively new area of research since
the taxes themselves are relatively new (Fletcher, Frisvold & Tefft, 2009). Regardless, the
intention seems to be there in terms of enacting social change in the form of a sales tax. Mikesell
put it best when he called the sales tax a “tax on consumption” (Mikesell, 1997). All of these
studies point at the idea that taxation can have implications beyond just fiscal use, therefore
factors other than just monetary reasons could come into play in determining what a state’s sales
tax rate will be.
With the origins of sales tax in mind, next it was important to take a look at the literature
regarding what explains the differences in sales tax rates across the states. One study by Jason
Fletcher and Matthew Murray that stood out among the rest since it used a very similar method
to the one being applied in this study. Fletcher and Murray were looking to uncover what causes
states to choose their sales tax bases, as well as researching the effect that competition among
neighboring bodies on a state and local sales tax rate. It’s worth noting that one line from their
paper revealed the difficulty in finding relevant research when it comes to state sales tax rates, as
they wrote, “There has been little research undertaken to date to explain these variations in the
structure of sales taxation across the states.” Fortunately, this study provides plenty of valuable
knowledge. They use empirical analysis to study 21 different independent variables including
demographic factors, income tax rates, and various sales tax exemptions among many others.
Ultimately, they discovered that income tax rates and demographic factors played a role in how a
state’s sales tax base was determined. This study uses both of these variables that they found to
be significant in our own research. They also found that geographic competition did not actually
cause competition in sales tax rates, but rather it tended to induce mimicking of sales tax levels
(Fletcher & Murray, 2006). Some of their data applied only to the local level, so that would only
prove partially valuable in this study, but their findings of those significant factors proved
invaluable. In this study, significant variables from a variety of studies were run in a single
model, and Fletcher and Murray provided a few that seemed likely to have an impact on the state
sales tax rate variations.
As is evident at this point, the idea of income tax playing a role in a state’s sales tax level
seemed to recur across the literature that was found. John Mikesell conducted a study that
surveyed states’ reliance on sales tax as a source of revenue since its implementation. He found
that by the 1930’s, sales tax had become the highest revenue producer among state taxes. By the
1990’s, however, income tax had equaled the status of sales tax. He ultimately concluded that
states will not be doing away with sales tax anytime soon, despite the emergence of income tax
as a strong source of revenue. In the paper, he writes that, “Disappointing sales tax
performance...bring major fiscal distress to the many states depending on its revenue -- and
absolute terror to the handful of states without a broad individual income tax” (Mikesell, 1992).
Ultimately, the relevance in this article is that sales tax and income tax are very closely linked
and the income tax rate in a state should have some bearing in the state’s sales tax rate. This is a
vital finding when it comes to this studies specific research.
Another study revealed that the partisanship of a state could also play a substantial role in
setting a sales tax rate. David Nice performed a study regarding whether party ideology actually
affects party policy once in office. While his ultimate goal wasn’t to discover what effects party
I.D. might have on a state’s sales tax rate, he use sales tax reliance as one of his variables to
discover whether parties affect policy. Using two different measures of party ideology --
Costain’s method and McGregor’s method -- Nice performed a cross-sectional analysis on a
number of independent variables, one of which was sales tax reliance. He added more controls
each time he ran the data and ultimately found that even when controlling for social and
economic conditions of a state, higher levels of Republicanism in a state was associated with
higher reliance on income tax and lower reliance on sales tax (Nice, 1985). Based on this
finding, it was crucial that state party I.D. be tested in this research.
Hypotheses
Based on the literature and the collective knowledge of the researchers, a set of hypotheses
were formulated as to what effects certain independent variables will have on the sales tax and
why that might be so.
1.) States with a higher income tax rate will have a lower sales tax rate.
Throughout the afore-mentioned literature, a common theme that emerged was a strong
relationship between income tax and sales tax. Mikesell’s quote regarding poor sales tax
performance being a “terror” to states without a broad income tax summed up the relationship
nicely (Mikesell, 1992). In fact, sales taxes were instituted in the first place to combat the
inability of income and property tax to generate adequate funds during the Great Depression
(Howe & Reeb, 1997). The rationale here is that the two taxes compensate for each other as the
main sources of revenue for a state. If a state has a high income tax, they wouldn’t need to have a
high sales tax since they were making sufficient revenue from the income tax.
2.) States with a higher GDP and higher budget will have a higher sales tax rate.
States with bigger economies and higher budget expenditures will need as much revenue as
they can get in order to pay for their complex fiscal system. One way to generate this revenue
could be through the use of a higher sales tax, as it would provide a stable source of money
throughout the fiscal year. Other taxes, like an income tax, are only collected in one huge chunk
once-a-year, so sales taxes provide a consistent flow of revenue for the state governments,
3.) States that generally lean Republican will have lower sales taxes.
The David Nice study regarding party ideology and party policy implementation found that
Republican state governments were more reliant on income taxes and less reliant on sales taxes
(Nice, 1985). He found that party policy is more apt to affect social issues, specifically welfare
spending, so we believe that since Republicans are less likely to spend more on welfare, they will
have less need of the sales tax to back that spending.
4.) States with a higher median age will have a higher sales tax.
Since states that have an older population will have more retired citizens, the states ability to
generate revenue through income tax will be greatly diminished. Similar to our hypothesis
regarding income tax, we believe that since the state will have a greater need for revenue from
non-income-tax sources, sales tax would be one of the places they would make up for the
difference.
5.) States with a larger amount of debt will have a higher sales tax.
Similar to the GDP/budget hypothesis, a state in need of more money will need to generate
revenue in as many ways as possible. Obviously, a state with a higher debt will be in need of
more money. Sales tax would allow the state to not only get more money from the tax-paying
citizens, but it would allow them to collect from tourists or any other non-citizens that would be
purchasing within the state.
6.) States with a higher non-white population will have a higher sales tax.
States with higher minority populations receive more grocery tax exemptions (Bahl &
Hawkins, 1998). In order to make up for the lack of ability to tax the sales on such a vital part of
the state economy, the state will need to increase the state’s overall sales tax rate in order to
make up for the lost revenue from groceries.
Research Design and Data Collection
When doing any type of research it is important to explain the methodology that is used
to analyze the data. Looking at the conclusions that were presented in the literature that was
previously analyzed, it is clear that outside of the Fletcher and Murray piece that most of the
findings presented showed only the significance of one variable in relationship to state sales tax
percentage (Fletcher & Murray, 2006). A handful of those variables and others from other
studies were taken and run through a multivariate analysis to see if the affects of the variables
discussed in the literature hold up when run in a model with other variables. This will help
further pinpoint what has an impact on state sales tax rates.
The data used in this study comes from many different sources. The first and most
important source of our data was the 2010 United States Census. Being that this is the most
recent census and holds the most accurate data possible, we found it necessary to gather our data
from the year of 2010 (if not the closest we could get) across all sources. Another important data
set we used came from the Book of States, which provides data from each of the 56 United
States and territories on a wide range of topics. Also used were data from the Bureau of
Economic Analysis, The National Association of State Budget officers, and
USGovernmentSpending.com. Though not main sources, these three still contributed important
numbers to the research.
Data Analysis and Interpretation
The variables were run in both a bivariate and multivariate model. A bivariate analysis
shows simply the relationships between one independent variable and the dependent variable,
and for the purposes of our study it is unwise to draw serious conclusions from these
relationships. That being said the bivariate analyses reveal very interesting results. Below are
coefficients, significances, graphs, and other statistics that will reveal a very interesting pattern
produced by the bivariate analysis.
Table 1: Bivariate Analysis
Coefficient Std. Error P > T Significance State GDP 0.00000152 0.000000802 0.065 *Partisanship 3.171 2.4631 0.204Budget 0.00001 0.00000803 0.097 *Age -0.0743 0.126 0.558Income -0.2129 0.1356 0.123Debt -0.0873 0.0738 0.243Minority Population 0.0202 0.0153 0.192GDP Per Capita -60.5832 30.511 0.053 *Budget Per Capita -369.8486 118.5193 0.003 ***
Graph B.) Bivariate Analysis of State Sales Tax Rates vs. State GDP (in millions)
Graph C.) Bivariate Analysis of State Sales Tax Rates vs State Partisanship
Graph D.) Bivariate Analysis of State Sales Tax Rates vs State Budget (in millions)
Graph E.)Bivariate Analysis of State Sales Tax Rates vs. Median Age
Graph F.) Bivariate Analysis of State Sales Tax Rates vs % of Personal Revenue from
Income Taxes
Graph G.) Bivariate Analysis of State Sales Tax Rates vs State Debt (% of GDP)
Graph G.) Bivariate Analysis of State Sales Tax Rates vs Minority Population
Graph H). Bivariate Analysis of State Sales Tax Rates vs GDP Per Capita
Graph I.) Bivariate Analysis of State Sales Tax Rates vs Budget Per Capita
When looking at the bivariate data and graphs, it is quite easy to see a theme: there is a very
intense, very dense grouping of states, containing a few outliers. Many of these graphs (most
severely B, D, F, and H) show a “lollipop” effect, where states are grouped very heavily towards
one end of the graph, with a few outliers scattered throughout. This occurs because when looked
at more intensely and when one thinks about it, the variances in state sales tax rates are actually
not very high; at least not as high as one might imagine. It is again important to note that while
some variables are statistically significant (Table 1) there are no conclusions about the relevancy
of our hypothesis being drawn. The bivariate analyses are being used simply to determine what
relationships may have an impact on the multivariate analysis.
Next is the important analysis: the multivariate analysis. A multivariate analysis
essentially tests the impact an independent variable has on a dependent variable in the presence
of other independent variables. This is the test that we can use to show just how much of an
impact the variables the literature put forth have in a more advanced situation where they must
interact with other variables, opposed to a more naïve bivariate model. Looking at the table of
coefficients and significance is eye opening as an interesting result is revealed.
Table 2: Multivariate Analysis
Coef. Std. Err. P>t Significance
GDP Per Capita -39.80247 36.06547 0.276Budget Per Capita -357.6437 183.7427 0.058 *Partisanship 3.476236 2.810184 0.223Age -0.109702 0.1576638 0.49Income Tax 0.1589727 0.1989525 0.429Debt -.0116314 0.0834845 0.89Minority Populatipon 0.0093503 0.0158096 0.557Constant 4.222406 6.172629 0.498
Significance: *=0.10; **=0.05; ***=0.01
The statistical analysis (Table 2) shows that all of the variables except for one had no statistically
significant impact. The only one that had a statistically significant impact was the budget per
capita variable which was significant at the .10 level (94.2%). As far as significance goes, this is
not a high significance at all. The correlation says that sales tax rates actually go up as budget per
capita goes down ( for every .01 unit increase in budget per capita there is a 3.576 unit decrease
in sales tax rates). This is a shocking result and undoubtedly the most important, as none of the
variables from the previous literature seem to have an impact on state sales tax rates when run
through and multivariate model.
Another interesting statistic to look at is the R-squared value (Table 3). The R-squared
value is a calculation of how much of the variance is explained by the variables used in the
study. The R-squared value was .2438, meaning that only 24 percent of the variance was
explained by the variables that were included in the analysis. This is a very low number because
it says that 75 percent of the variance in states sales tax rates was not explained in our model.
This may be because of other variables not used in the study that may have had an impact.
So far this study has shown that a.) the multivariate analysis shows that the variables in
the previous literature do not hold up in the presence of other variables b.) only budget per capita
is significant when it comes to state sales tax variance and c.) our model only explained 24% of
the variance. At this point it is extremely important to discuss why the results might happen.
They were very unexpected and are very odd when one looks at the previous literature. Next it
will be important to rationalize why the findings came out the way they did, and look deeper into
what may actually be causing variances in state sales tax rates.
Conclusions
The study’s findings revealed only one statistically significant multivariate variable. This
finding that budget per capita was a negatively correlated relationship was actually the opposite
of what was hypothesized. This combined with the other variables being insignificant lends to
the discovery that variations in sales tax rates are much more complex than the original theories
suggested. It is now easy to conclude that there are many factors that contribute to the variation
of sales tax rates among states.
While discovering the lack of significance of the variables, a few reasons developed as to
why this study’s results differ from the results observed in the literature and what was initially
hypothesized. In examining the data one of the things that sticks out is that the data are not as
diverse as initially observed. Since nearly one third of states have a sales tax about equal to 6%
the data tends to get grouped together around that number. This grouping is one the main reasons
that the variables did not have the predicted effect.
The lack of significance of the variables was especially surprising given the previous
literature suggesting that many of the variables that were used would have some degree of
significance. Upon further analysis of the data and that of the studies cited in the literature
review, it is logical to draw the conclusion that these results differ because many of the studies
cited only examined the significance of one variable’s effect on the sales tax rate of a state. Our
study used a multivariate analysis of many variables at once; when performed in this research
these other theories fell apart.
The lack of significance of the variables in the study led to a search for other explanations
for the variation that is seen in sales tax rates among states. A possible explanation is that the
right variables were not chosen for this study. When looking at the states with high sales tax rates
it seems that many of these states are states which raise a large portion of their revenue from
tourism. In a future study it would be ensured that percentage of revenue raised from tourism is
one of the variables because given what has been observed this variable seems like it would have
some significance. It is also likely that since this tax affects everyone there are many more
factors at play than few that what was chosen to study.
Another factor that was not researched was the effect that local sales tax have on the
statewide rate. Some states allow their localities to tax the sale of goods separately from the
statewide rate; this could have an effect on rates that lawmakers choose to charge across the
entire state. In future studies the affect of local sales taxes on the statewide rate would certainly
be examined.
The main take away from this study is that given majority of the states have sales tax
rates between 5-7%, and that although no clear evidence was found in the research about what
causes this variance, there must be some sort of significance of these numbers. It is likely that
there is some political or historical reason that most states have settled on a rate in this range. A
number like 6% seems to be a happy medium that politicians select that will raise enough capital
for the tax to be effective, while not being so high that citizens are upset by the amount they are
being charged on their purchases. Fletcher and Murray’s paper states that counties are more
likely to mimic each other’s tax policies rather than compete with each other (Fletcher &
Murray, 2006). This seems to be supported by our data given that many states have decided to
adopt similar tax rates rather than undercutting their neighbors in order to attract consumers. This
mimicking is the best explanation that that could be formulated to explain not only why many
states have similar tax rates, but also why there is a lack of variation between most states.
Though this study only provided one significant variable, but upon further analysis and
conclusions showed that there are many more things influencing the sales tax rates other than the
short list of variables that were tested. Most notably of these are the effects of a political or
historical significance of having a rate in the 5-7% range. Future studies will examine the
significance of tourism and local sales tax rates also. Though the research may have not been
able to confirm our original hypotheses or the previous literature, the study did provide valuable
insight into the variation of sales tax rates and why these variables were not as significant as
originally hypothesized.
Works Cited
Bahl, Roy & Hawkins, Richard. (1998). A Georgia Sales Tax for the 21st Century.
Atlanta, GA: Georgia State University.
Fletcher, J. M., Frisvold, D., & Tefft, M. (2009). Can soft drink taxes reduce population
weight?.Contemporary Economic Policy, 28(1), 23-35. doi: 10.1111/j.1465-7287.2009.00182.x
Fletcher, J. M., & Murray, M. N. (2006). Competition over the tax base in the state sales
tax. Public Finance Review, 34(3), 258-280. doi: 10.1177/1091142105285571
Howe, E. T., & Reeb, D. J. (1997). The historical evolution of state and local tax systems.
Social Science Quarterly, 78(1), 109-121. doi: 0038-4941
Mikesell, J. L. (1992). State sales tax policy in a changing economy: Balancing political
and economic logic against revenue needs. Public Budgeting & Finance, 83-91. doi:
10.1111/1540-5850.00931
Mikesell, J. L. (1997). The American retail sales tax: considerations on their structure,
operations, and potential as a foundation for a federal sales tax. National Tax Journal, 50(1), doi:
0028-0283
Nice, D. C. (1985). State party ideology and policy making.Policy Studies Journal, 13(4),
780-796. doi: 10.1111/j.1541-0072.1985.tb01618.x
Peterson, D. E., Zeger, S. L., Remington, P. L., & Anderson, H. A. (1992). The effect of
state cigarette tax increases on cigarette sales, 1955 to 1988. American Journal of Public Health,
82(1), 94-96. doi: 0090-0036, 01/1992