Melanie Freeze and Dr. Jeremy Clayne Pope, Political Science
Voters look to the economy when evaluating incumbent presidential candidates. For example, the record high inflation, interest rates, and unemployment during Carter’s first term in office are said to have contributed to his defeat in the 1980 presidential elections. The significance of the economy was so large in the 1992 presidential election one campaign strategist bluntly stated, “It’s the economy, stupid.” While national economic factors have always been clearly connected to presidential campaigns, relatively little research exists that examines the role state economies play in determining state-level presidential returns. My research primarily sought to understand whether national or state economic conditions have a stronger influence on the state-level outcomes of presidential elections.
Voters use the state of the economy to evaluate the presidential incumbent candidate’s competence and ability to manage the country. My expectation is that when voters use the economy to evaluate candidates, they rely more on state economic conditions than national economic conditions. State economies are closer in proximity to the voter’s personal situation and should reflect individual’s perception of economic conditions better than national indicators.
To understand which level of economy has the strongest influence on state-level vote returns, election and economic data was gathered for the 12 elections spanning 1960 to 2004. The two-party percentage vote given to the incumbent party’s presidential candidate serves as the dependent variable while national and state economic conditions are the primary independent variables. To measure national and state economic conditions, I focused on two main economic indicators: the change unemployment and real per capita disposable income. Due to the long term nature of disposable income (its yearly changes have only a small relative impact on individual while the changes over four years are more noticeable), it was differenced over four years while unemployment was only differenced over a one year period. In addition to these main variables of interest, controls for the intensity of a state’s presidential campaign (intense campaigns defined as having returns where candidates only received 45-55% of the vote), the presence of an incumbent candidate in the elections, and a variable indicating whether a state is in the South. Finally to control for autocorrelation and to capture the contemporaneous effect of the independent variable on the dependent variable, the incumbent party’s vote percent was lagged one year and added to the model as an independent variable. This addition transformed the static model to a dynamic Lagged Dependent Variable (LDV) model. Finally, the significance and magnitudes of the relationships were estimated using an Ordinary Least Squares (OLS) regression.
The results revealed that national economic conditions have a stronger and more significant impact than state economic conditions on the percentage of votes given to the incumbent party’s presidential candidate. When the national economic variables were excluded from the model, both change in unemployment and disposable income had a significant relationship with the incumbent party vote return. As state unemployment decreases by one percent from the prior year, the incumbent party’s candidate receives 2.05 percent more votes. Likewise, as state disposable income increases by one percent since the prior election, the incumbent party candidate’s vote shares increase by 0.29 percent. However, when the national variables were taken into consideration, the significance of these state effects decreased in magnitude and became statistically insignificant (statistical significance being the probability the relationship estimated is the product of reality and not chance). When the national economy improves, voters are more likely to support the incumbent candidate. A one percent decrease in national unemployment over the past year causes a 1.80 percent increase in votes for the incumbent party, and a one percent increase in disposable income since the past election results in a 0.34 percent surge in the vote percent for the incumbent party’s presidential candidate.
As a result of this research project, I was able to conclude that fluctuations in the national United States economy play a distinct role in presidential elections. Although state economies may be closer in proximately to individual voter’s personal situation, changes in state economic conditions fail to predict presidential returns when changes in the national economy are taken into account. Improvement in national disposable income and unemployment rates significantly improves an incumbent candidate’s standing in all U.S. states.
From this ORCA sponsored research, I have developed research and statistical analysis skills that most political science undergraduates never have the chance to explore. I learned of the frustrations of data-collection and panel data statistical modeling, and was able to present my findings at the 2006 Fulton Mentored Research Conference. In addition to the educational strides that I made as a result of this research, I built a valuable academic relationship with faculty. Through weekly interaction with my mentor, I was given advice not only on my research, but also on my academic goals that gave me much needed perspective as I applied to political science graduate programs across the United States.
A few weeks before I graduated, I was notified that both my husband and I were admitted into the Political Science PhD program at Duke University. While a number of factors contributed to my acceptance into Duke, I believe that the research skills that I had the opportunity to gain through this ORCA research project were instrumental in preparing me for graduate school. I will enter Duke with advanced statistical skills and an awareness of the difficulties and rewards of research.