Which States Support Which Ones? Predicting Federal Flow Through From the Feds to the States

While politicians in Washington, who have already authorized trillions of funds for businesses in the past 3 years, have been arguing over allocating funds for state and local governments, there have been a number of suggestions that stimulus packages potentially benefit certain states at the expense to others. The arguments primarily appear to come down to politics – “red,” Republican-led states versus “blue,” Democratically-led states. The insinuation is that that the “blue” states are more poorly run than the red states and that the better run “red” states should not be “bailing out” poorly run “blue” states. A linear regression model was developed that perfectly predicted both federal fund recipients and “red” or “blue” designations. Correlation and principal component analysis was run to determine the factors that were the best predictors. The truth appears to be as Governor Cuomo states, “blue” states are better run, better funded and are more productive than “red” states when measuring gross domestic product, income and net tax revenue per person. Educational funding was a major reason “blue” states were ahead of their “red” neighbors.


Introduction
The Center on Budget and Policy Priorities (2020) estimates that more than 25 percent of state revenues have evaporated because of the covid-19 pandemic. The Federal Reserve Bank in Cleveland estimated major losses at the state and federal levels (Whitaker, 2020). Unemployment neared 15% at one point (Long & Van Dam, 2020) as 40 million workers who lost jobs and filed for unemployment during the pandemic. To date Congress has approved 2 stimulus funding packages totaling over $3 trillion, with discussions or a third (already approved by the House of Representatives) (Foran et al., 2020). www.scholink.org/ojs/index.php/sssr Studies in Social Science Research Vol. 3, No. 1, 2022 Published by SCHOLINK INC. However, this third stimulus is not without controversy. Senate Majority Leader Mitch McConnell called it a "blue state bailout," alluding to the fact that the Covid-19 virus had impacted Democratic states harder than Republican ones at that point (Lahut, 2020). In addition to McConnell, President Trump has often tweeted about the incompetence by local politicians in blue states, whom he also holds responsible for the economic harm wrought on their constituents (Trump tweet, 4/27/2020). He tweeted on April 27th: "Why should the people and taxpayers of America be bailing out poorly run states (like Illinois, as example) and cities, in all cases Democrat run and managed, when most of the other states are not looking for bailout help?" McConnell said he "would certainly be in favor of allowing states to use the bankruptcy route… We're not interested in rescuing them from bad decisions they've made in the past" (Frum, 2020). Former Florida Governor and current Senator Rick Scott weighed in that he was not supportive of helping any states that can't "balance its bloated budget without borrowing and consistently raising taxes to pay for its profligate spending." Scott added, "It's not fair to Florida citizens to send their tax dollars to bail out liberal politicians in states like New York for their unwillingness to make tough and responsible choices." Schultz, 2020). The same question was posed to Governor Kristi L. Noem of South Dakota who asked "why taxpayers in her state should bail out Illinois" (Bump, 2020).
Governor Cuomo of New York pointed out that New York, the hardest hit state early in the coronavirus pandemic, is a net payer to the rest of the country so federal funds are not a bailout but returns for years of payments to benefit others (Lahut, 2020). Senator Chris Murphy (2020) said in a tweet that "If Florida would like to have a conversation about making sure no state gets more money from the federal government than they send to it, Connecticut is ready." Arguments over which states are "better" comes down to an assumption that "red" states are more responsible in their operations than "blue" states, and that there is a net transfer of funds by "red" states to "blue" states. In addition, they argue that "blue" states are less efficient and productive than "red" states. A further argument is that low tax "red" states provide "more cost-effective services to their residents". Bump (2020) suggested that based on graphics developed by the Washington Post, "there's not a strong relationship between how red or blue a state is and how much it contributes or receives from the federal government on net." The question is whether there is merit to any of these regularly pontificated arguments?
The project attempted to use economic data to compare states with the intent of determining the following: 1. Can "red" or "blue," "payor" or "recipient," be predicted? 2. Are "red" states "better run" than "blue" states, meaning are "red" states more "productive," based on per capita incomes, gross domestic product and net federal tax transfers that "blue" states?
If the current dialogue is not correct, the can these same statistics reveal the truth, which will perhaps give some guidance how to correct the situation.

Methodology
The database was developed from a variety of online sources (see the data sources at the end of the paper). These data were used for the analyses with statistical techniques principal component analysis, correlation analysis and linear and logistic regression, with the goal of identifying:  Spearman (1904) and continues to develop (Jolliffe, 2002). PCA techniques are mainly used to reduce the dimensionality of p multi-attributes to two or three dimensions. The mathematics of PCA uses an orthogonal transformation convert observations of possibly correlated variables into a set of values of linearly uncorrelated variables called principal components (Pleitez, 2012;IOS, 2012). Two methods are commonly used for determining the number of factors to be used for interpreting the results: The Scree test (Cattell, 1966) is based on the decreasing curve of eigenvalues. The number of factors to be kept corresponds to the first turning point found on the curve. However, these representations are only reliable if the sum of the variability percentages associated with the axes of the representation space are sufficiently high. If this percentage is high (for example 80%), the representation can be considered as reliable. If the percentage is reliable, it is recommended to produce representations on several axis pairs in order to validate the interpretation made on the first two factor axes.
Ultimately the goal is to determine a consequencei.e. "red" or "blue", incoming funds or net outflow of funds. Linear regression model provides a mechanism to model the data to determine if differences between the active and inactive projects exists. Linear and logistic regression were compared -logistic regression is generally the preferred technique when the dependent variable is binary in nature. Both regression models were run using XLSTAT ® .
With regression models, the goal is to develop a series of weights for the independent variables (all variables to determine which variables meaningfully impact the outcome of success. Regression can be used as a tool to predict the likelihood of success (the dependent variable) using the current dataset. An . For the purposes of the modeling, "blue" states were assigned the value of 0, and "red" states a 1. Likewise, those states that are net outflow were assigned a value of 0, and the recipient states, 1.

Results
To start, a series of histograms were created to compare states. Figure 1 shows the states and their net federal flow of funds per capita. What this graph shows is that Kentucky and Virginia receive the most federal funding per capital than any other state (over $9,000/person   Next a histogram was developed for the gross domestic product (GDP) per capital. The results in Figure 2 shows that the GDP/capita is highest in many of these same states, plus Minnesota, Virginia, and Delaware. These states are productive. The third histogram was developed to shows whether these states are net payors because they have high income? The answer is to a degree, yes, although Alaska, Delaware and Wyoming are high income subsidized states, Alaska particularly so (see Figure 3).
However, Alaska, Delaware and Wyoming are states with falling incomes over the past 5 years, and challenging budgets -Alaska is considering ending payments for oil since their Treasury is now broke.
Note Connecticut also had a falling income as well for 3 rd Quarter 2019.       Based on this data, the initial review suggests that certain states may be operated better than the political rhetoric may indicate, and therefore may provide better services to their populace based on wealth, GDP, education, health and other factors. Running Principal component analysis, the goal was simply to determine if there were consistent factors that made up the states that were payor. After conducting this analysis, the Scree plot indicated on factor that was nearly 40% of variance. It indicated that those that were net payors of funds had:  Vol. 3, No. 1, 2022 The results were the same for both eigenvalues and rotated results.
Running the linear regression models, the linear regression confusion matrix, which identifies whether the predicted value for payor or recipient correctly. The linear regression model predicted the correct result 100 percent of the time. A logistic model was run since the result being predicted was binary. Figure 8 shows the prediction of whether a state was a net payor or received of federal funds. The model correctly predicted the actual situation 100 percent of the time. The regression model was re-run based on being categorized as "Red" or "Blue" based on the overall results of Presidential elections since 1960, the model yielded a correct prediction 100 percent of the time (see Figure 9).  Running the linear regression models, the linear regression confusion matrix, which identifies whether the predicted value for payor or recipient correctly or state color, yielded a correct prediction 90 percent of the time. The logistic regression model was run for comparison and the resulting confusion matrix was also over 90 percent (see Tables 1 and 2). The resulting linear regression predictions are shown in Figure   10 and 11.
Running the linear regression models, the linear regression confusion matrix, which identifies whether the predicted value for payor or recipient correctly. The linear regression model predicted the correct result 90 percent of the time. A logistic model was run since the result being predicted was binary. The model correctly predicted the actual situation 94 percent of the time. The regression model was re-run based on being categorized as "Red" or "Blue" based on the overall results of Presidential elections since 1960, the model yielded a correct prediction 100 percent of the time, just like with the regression model above.

Discussion
Arguments over which states are "better" comes down to an assumption that "red" states are better more responsible in their operations that "blue" states, and that there is a net transfer of funds by "red" states to "blue" states. Aside from the fact that we are all in this together and we are all better off when everyone benefits in the long run, the question isit this red-blue argument really true? The intent of this paper was to answer the following questions: 1. Can "red" or "blue," payor or recipient, be predicted? 2. Are "red" states "better run" than "blue" states, meaning are "red" states more "productive," based on per capita incomes, gross domestic product and net federal tax transfers that "blue" states?
The answer to the first question was clearly "yes," a model can be created using generally available data to predict whether a state was a net payor or recipient, and whether the state was categorized as "red" or "blue." With respect to the question about being "better run," the incomes, education and gross domestic product per capita were higher for the "blue" states which were generally the same states that were net payors. Hence the "blue" states subsidize the "red" states. The grand experiment of reducing government in favor of letting private sector take over is clearly not working in "red" states. The result is that they all rely on net federal transfers of fund, including South Dakota, Florida and Kentucky, the states with elected officials arguing against state and local government bailouts. The truth appears to be as ex-Governor Cuomo stated, blue states are better run, better funded and are more productive than "red" states. In many cases the state taxes are higher in these states, but the taxes are used to benefit the greater whole as opposed to "red" states that appear to, and some of whom brag about, starving government. The states that talk the most about starving government are consistently recipients of federal funds.