1. Farming Subsidies and the Vote on
Amendment 1, Right to Farm Bill
By: Nicole Eversman
2. I. Introduction
On August 5, 2014 Missouri voters made their way to the polling stations to vote on a
variety of legislation, including Amendment 1, the “Right to Farm” bill. After counting and then
recounting the votes, Amendment 1 barely passed with a fifty-one percent majority. This bill
passed a law stating that Missouri farmers and ranchers would forever be able to engage in
farming and ranching. County ordinances were unaffected by this bill and all of the rights stated
in it are subject to regulations. In this paper I look at the effects that farm subsidies, per capita
income, and education levels had on the proportion of people who voted yes on the Right to
Farm bill.
Using census data from American Fact Finder and crop subsidy data from NASDA, I
hypothesize that factors such as education, domestic migration, and the percent of people
employed in the agriculture industry will affect the proportion of people in a county who vote in
favor of the bill. The data are collected by the federal government’s Census Bureau every ten
years and include information on the population, housing, economic and geographic data at the
county level across the United States. From the Environmental Working Group, I gather
information on the total dollar amount of farming subsidies, and the amount of each of the top
five types of subsidies. This organization has a database that includes information on various
farming subsidies, broken down by state and county, as well as by the crop specific subsidies. It
also shows the specific farm or group that is receiving these subsidies. The data are available
from 1995 to 2012. All of this data will be examined at the county level.
My hypothesis is that counties which experience higher in-migration, have lower
education levels, a higher percentage of workers in agriculture, and receive higher amounts of
3. subsidies are more likely to have a higher percentage of the county vote yes for Amendment 1.
These statistics are indicators of a more rural lifestyle. I think these more rural counties would be
more likely to support a bill that promotes agricultural development and sustainment relative to
their urban counterparts.
II. Literature Review
There has been some analysis on the attitudes towards similar issues. Concerning genetically
modified organisms there is a lot of controversy on the long-term health affects to consumes
while suppliers are pushing to use GMOs to increase output.
A study by Michael Burton et al (2001) has indicated that consumers towards genetically
modified organisms (referred to GMOs from here throughout) rely on several things. This study
took place in Manchester, UK in 2001, and had a total of two hundred and twenty eight
responses out of two thousand surveys handed out. Gender and current consumption of organic
foods were two of the most significant indicators of consumers who were less likely to support
legislation supporting GMOs in the future. This study was able to indicate that there are
consumers who feel very strongly against GMOs and would go to greater lengths to prevent
consuming these types of foods. This is not surprising because of the lack of trust towards GMOs
that is often seen throughout Europe. Overall Burton’s examination of consumer’s willingness to
pay for foods without GMO technology may or may not be significant because the actual
monetary affects that would take place in a real market were not present.
It has been shown in research by Barrows et al. (2014) that farms in the United States are
becoming larger and that their numbers are decreasing due to high costs of labor and capital.
Much of this is also closely linked to the history of farm subsidies. Barrows et al. confirm that
4. the increasing size of farms is also causing an increase in the wealth of farmers which can be
attributed to recent farming policies. “Program Crops,” which are what both Barrows et al and
this study focus on are where most of the money for farming subsidies is distributed. These crops
largely include but are not limited to grains and cotton. The growth of farming subsidies in
recent years has been tremendous, going from $9.4 billion in 2011 to $15.8 billion in 2012. The
study predicted that these subsidies would impact what kinds of crops are being produced, but it
is difficult to measure the change in supply of various crops. Similarly, the authors found it
difficult to find a relation between subsidies and change in farm size because of the industries
complexity. Ultimately though, this study finds that farming subsidies have negligible effects on
the growing size of farms in the United States.
Wilson et al. (2003) looked at the impact of genetically modified wheat on a global scale.
Since wheat is a more complex organism, the genetic modification of the grain has not been done
as extensively as other crops, but it seems that producers would be willing to accept GM wheat if
it was available to them. More importantly though is consumer’s willingness to accept this
product into their diet. In the United States it was found that only 10% of customers surveyed by
The American Bakers Association were concerned about the use of GM wheat being substituted
into products. Their sources also claim that consumers are willing to pay on average 29-44%
more for foods that are not genetically modified. Wilson found data confirming that there was a
large increase in the market for organic foods between 1998 and 2003 when this paper was
written. This increase in demand is accredited to consumers’ aversion to buying GM foods and
buying organic foods as substitutes. The writers predict that the controversy and resistance to
GM foods will have a positive relationship with the increase in the production of them.
5. A similar study has looked at how labels affect markets for GMOs. Asymmetric information
about GMOs are crucial to examine when looking at voluntary and mandatory labeling
regulation (Caswell, 1998). When it comes to organic foods copious amounts of labels are used
to distinguish them from their GMO counterparts. When this article was published new
regulation was being considered to keep fraudulent claims of organic foods from reaching
consumers, as to avoid this issue of asymmetrical information. Consumers that are willing to pay
the premium for organic foods do so because they believe that the production method for them is
both safer and more environmentally sustainable so it is important to have proper labels on both
organic and GMO foods on the shelves. According to Caswell, some companies see tighter
regulations on food labels as an intrusion on their ability and rights to market their products.
Although there are also limitations on regulating labels such as most consumers unwillingness to
take the time to read through food labels before purchasing them. Caswell concludes though that
labeling programs are able to improve the markets as well as the quality of food available to
consumers.
III. Data Analysis
Table 1 reports descriptive statistics of the data which allow us to examine the variables
that potentially have an effect on the outcome of the vote on Amendment 1. The mean
percent of people who voted yes in each county was just shy of 63% of the population. This
is consistent with the result of the elections in that Amendment 1 did indeed get passed. The
average percent of each county that works in the agriculture industry was about 5%, but
ranged from 0.3% up to 26%.
6. The maximum subsidy as a proportion of agricultural income was over 100%, meaning
some counties received more money in farming subsidies than workers in agriculture made
during 2012. Net domestic migration shows that there was a slight decrease in the population
in most countries across the state with a negative value for the mean, but urban counties did
increase in population by a significant amount.
Not surprisingly, the proportion of each county with regards to education level varied
quite a bit. Some counties had over 20% of the population with a graduate degree, while in
some counties only 50% had a high school diploma. All of these factors may have had an
influence on how citizens of Missouri voted on Amendment 1 in August 2014.
Table 1. Descriptive Statistics
Variable Mean Minimum Maximum Std. Dev.
Yes 4249.68 409 65764 7107.66
No 4169.16 133 122472 12460.4
Total 8418.83 716 188236 19414.9
Perc_Yes 0.63 0.27 0.87 0.11
Perc_No 0.37 0.13 0.73 0.11
Logit 0.55 -1.02 1.86 0.49
Perc_Ind_Agr 0.06 0.003 0.26 0.04
Tot_County_In
come
2.05e+009 6.60e+007 5.45e+010 5.99e+009
Agr_Incooe 3.29e+007 7.14e+006 2.18e+008 2.34e+007
Tot_Subs 4.81e+006 3746 2.22e+007 4.61e+006
Subs_Prop_Ag
r
0.18 0 1.01 0.19
Perc_Corn 0.19 0 0.52 0.11
Perc_Soy 0.16 0 0.41 0.09
Perc_Wheat 0.02 0 0.17 0.03
Perc_Cotton 0.15 0 16.78 1.57
Perc_Rice 0.003 0 0.13 0.02
Net_Dom_Mig
r
-66.19 -3641 2258 573.67
Perc_Net_Dom
_Migr
-0.004 -0.03 0.02 0.01
Pop_Over_25 34553.2 1544 679114 81825.2
7. Perc_HS_Dipl 0.40 0.21 0.52 0.06
Perc_Some_Co
l
0.21 0.06 0.28 0.03
Perc_Associate 0.06 0.03 0.29 0.03
Perc_Bach 0.11 0.04 0.27 0.04
Perc_Grad
Per_Cap_In
0.06
48568
0
30801
0.20
80323
0.03
8626.1
Some of the variables used have a very high standard deviation, while others have a much
lower standard deviation. Subsidies as a proportion of agriculture had a relatively high standard
deviation across counties at nearly 20%. Other variables such as the percent of each county
working in agriculture, domestic migration as a percent of population, and percent of counties at
different levels of education did not vary nearly as much.
Corn received the highest dollar amount of subsidies at just over $2.85 million for
Missouri alone in 2013, closely followed by Soybeans at almost $1.77 million, Wheat at $900
thousand, Cotton at $632 thousand, and Rice at $529 thousand. Subsidies for corn were given to
over 90 thousand recipients in 104 out of 115 counties in Missouri.
8. Figure 1. Counties where Amendment 1 Failed
Figure 1 shows a map of Missouri, with county borders shown. The majority of the state
is shown in white, while there are five distinct areas comprised of 12 counties shown in grey.
The grey shaded counties depict the counties where less than fifty percent of the voting
STL
KC
SCO
COL
SPR
9. population voted yes on Right to Farm, and the bill did not pass. The non-shaded counties are
where the bill received more than fifty percent of yes votes and the bill did pass. The Right to
Farm act passed in 103 counties out of 115.
Besides Scotland County in the northern part of the state, the other four major areas
where counties did not pass the amendment are centered around St. Louis, Kansas City,
Columbia and Springfield. These urban centers were the main areas in which people did not
support the Right to Farm Bill.
My model takes the form
ln(p/1-p) = Bo + B1Perc_Ind_Agr + B2Subs_Prop_Agr + B3Net_Dom_Mig + B4Perc_HS_Dipl
+ B5Perc_Some_Col + B6Perc_Associate +B7Perc_Bach + B8Perc_Grad + B9l_Perc_Cap_Inc
Table 2. Model Estimates – OLS – 115 Observations
Dependent variable: Logit
Coefficient Std. Error t-ratio p-value
const -8.85 3.46 -2.55 0.01 **
Perc_Ind_Agr 3.19 1.04 3.05 0.002 ***
Subs_Prop_Agr -0.37 0.21 -1.76 0.08 *
Net_Dom_Migr -0.0001 7.2e-05 -1.44 0.15
Perc_HS_Dipl 4.07 1.16 3.49 0.0006 ***
Perc_Some_Col 0.87 1.37 0.63 0.52
Perc_Associate -2.91 2.02 -1.44 0.15
Perc_Bach -1.05 1.91 -0.55 0.58
Perc_Grad -0.82 2.75 -0.29 0.76
l_Per_Cap_Inc 0.72 0.33 2.18 0.03 **
Adjusted R-squared 0.41
F(9, 105) 9.85 P-value(F) 8.12e-11
10. The dependent variable is the odds ratio where p is the proportion of population that
voted yes on Amendment 1. I take the log odd ratios so that the estimates of P^ are bounded
between 0 and 1.
Model estimates for the full model are shown in Table 2. Net domestic migration, percent
some college, percent associate’s degree, percent bachelor’s and percent graduate degree all have
insignificant t-values. I perform an F-test to see if the coefficients are all simultaneously equal to
zero. The calculated F-value is 0.9869 so with a critical F-value of 2.29 I can not reject the null
hypothesis. Therefore, I drop the variables from my model.
Table 3. Partial Model Estimates
Dependent variable: Logit
Coefficient Std. Error t-ratio p-value
const -7.56 3.01 -2.51 0.01 **
Perc_Ind_Agr 3.17 1.03 3.09 0.003 ***
Subs_Prop_Agr -0.42 0.21 -2.01 0.047 **
Perc_HS_Dipl 4.9 0.88 5.60 <0.00001 ***
l_Per_Cap_Inc 0.56 0.26 2.14 0.03 **
Adjusted R-squared 0.411644 P-value(F) 7.43e-13 Chi-Square 21.82
Table 3 reports the partial model estimates. The coefficients for the variables percent
industry employed in agriculture, dollar amount of subsidy as a proportion of agriculture, percent
with a high school diploma and the log of per capita income are all significant. As the percent of
a county working in agriculture and percent of population who only have a high school diploma
as their highest level of schooling go up, the county was more likely to vote yes on the
amendment.
11. Oddly though, as the subsidy as a proportion of agriculture went up a county was less
likely to pass the bill, and as the log of per capita income went up so did the likelihood of
passing the bill. These results are not consistent with my original predictions, but are shown to be
significant in the model.
I test for the existence of heteroskedasticity using White’s Test. The critical Chi-square
value is 23.7, and with a calculated Chi-square value of 21.82, therefore I do not reject the null
hypothesis and conclude that heteroskedasticity does not exist in my model.
Table 4. Partial Model Estimates – OLS – 115 Observations, Crop Specific Subsidies
Dependent variable: Logit
Coefficient Std. Error t-ratio p-value
const -8.47316 3.08639 -2.7453 0.00710 ***
Perc_Ind_Agr 3.25073 1.03347 3.1455 0.00215 ***
Corn_Prop_Agr -2.56346 1.92885 -1.3290 0.18670
Soy_Prop_Agr 0.763019 2.75138 0.2773 0.78207
Wheat_Prop_Agr 7.5483 5.04177 1.4972 0.13733
Cotton_Prop_Agr -12.8276 9.5637 -1.3413 0.18270
Rice_Prop_Agr -3.0464 7.84355 -0.3884 0.69850
Perc_HS_Dipl 4.97108 0.89713 5.5411 <0.00001 ***
l_Per_Cap_Inc 0.641768 0.268182 2.3930 0.01847 **
F(8, 106) 10.95 Adjusted R-squared 0.41 P-value(F) 3.73e-11
Table 4 shows what the partial model looks like when the subsidies are broken down into
the five largest types of subsidies given to farmers in Missouri. The results are similar to the
original partial model in Table 3 and the same independent variables are significant except for
subsidies as a proportion of agriculture, which is not included in the model shown in Table 4.
12. IV. CONCLUSION
In this paper I examine various factors that might affect the proportion of a county voting in
favor of Amendment 1, the “Right to Farm” bill. Counties that are in the more urban regions
surrounding St. Louis, Kansas City, Columbia and Springfield along with an outlier, Scotland
County, were the only areas that did not vote to pass this bill. I have also determined that the
percent industry working in agriculture, dollar amount of subsidies as a proportion of agriculture,
the percent of the population with a high school diploma and the log of per capita income all had
significant effects concerning this vote. Although the independent variables net domestic
migration, percent some college, percent associate’s degree, percent bachelor’s degree and
percent graduate degree were not found to be significant, they were included in the original
model shown in Table 2. These results show that migration out of a county had a negative effect
on a county’s likeliness to vote yes on amendment 1. I also conclude that the percent with some
college education increased the vote, but percent with an associate’s bachelor’s or graduate level
degree have an inverse effect on the likelihood of a county voting yes.
13. V. SOURCES
2013 Farm Subsidy Database. Environmental Working Group: Farm Subsidies, 2013.
Web. 30 September, 2014.
American Fact Finder. Census Burea, 2014. Web. 30 September, 2014.
Barrows, Geoffrey, Steven Sexton, and David Zilberman. 2014. “Agricultural
Biotechnology: The Promise and Prospects of Genetically Modified Crops,” Journal of
Economic Perspectives 28(1): 99-120. Print.
Burton, et al. 2001. “Consumer attitudes to genetically modified organisms in food in the
UK,” European Review of Agricultural Economics Vol 28(4): 479-498. Web.
Caswell, Julie. January 1998. “How Labeling of Safety and Process Attributes Affects
Markets for Food.” From the SelectedWorks of Julie Caswell p. 151-158. Web.
NASDA. National Association of State Department of Agriculture, 2014. Web. 30
September, 2014.
Wilson, Janzen, and Dahl. 2003. “Issues in Development and Adoption of Genetically
Modified (GM) Wheats.” AgBioForum Vol 6(3): 101-112. Web.