Examining the Effect of Election Outcome on the Short-Term Price Movement of the S&P 500 Index

Examining the Effect of Election Outcome on the Short-Term
Price Movement of the S&P 500 Index

See the source image

Section I: Introduction

The legislation passed by the US Congress can often have a significant impact on large corporations as it influences taxing and business regulations.

One example of this is the Dodd-Frank Act of 2010 which made critical changes to regulations in the financial industry. As 2020 was an election year for both houses of congress in addition to the presidential election, the impact of elections on the stock market found new significance.

Data from the past, shows volatility in the stock market increases leading up to an election day since investors and markets dislike uncertainty.

Political election effects on the market are mainly because of investor’s expectations regarding the potential changes in policy.

Many websites and publications speculated about the potential impact that the 2020 election could have on the market.

One such article was from USBank.com that went through 8 possible scenarios that were analyzed by the U.S. Bank Wealth Management and what each could mean for the market.

As seen in the table from the slides in the presentation, the article predicts that unless there’s an election sweep in favor of one party or another, significant change in the volatility is unlikely.

Moreover, the article predicts that there will be no preference for either party.

We gathered data to test these predictions and see if our data and analysis could support the article’s claims. Also, we analyzed some statistics about partisan control of the White House and Congress. The following sections present some groundwork research that we did before gathering information. That is in turn followed by our data gathering process and finally, our results and conclusions. 

Section II: Motivation

There is plenty of research regarding the correlation between the changes in leadership of the American government and the stock market. We began our research questioning which political party saw the best returns when in office as president. Since 1900, the Dow Jones Industrial Average has gained an average of 3.5% per year under Republican presidents versus an average of 6.7% annual gain. Even more recently, since 1993, there has been a 3% annual gain in U.S. equities under Republican presidents compared to a 14.5% annual gain under Democratic presidents (Smith & Woodley 2020). Given these findings on the difference in average performance between Democratic and Republican presidents we then shifted our focus to how U.S. equities respond to pre-election expectations. We aimed to discover how the market forecasts future growth according to which party wins the presidential election.

We found that leading up to elections there has been an average daily gain of .041% when a Republican candidate leads the pre-election polls in comparison to an average daily gain of .004% when a Democratic candidate leads the pre-election polls. (Smith & Woodley 2020)This suggests that there is a public notion that the U.S. market performs better under a Republican president, however the previously mentioned evidence contradicts this notion. Lastly, we considered the correlation between different combinations of political party majorities in the houses of congress and the president. We discovered that when both houses of congress and the president are under the control of one political party, the Dow Jones Industrial Average gains 10.7% in average annual returns. When there’s a split congress, there’s an average gain of 9.1% in annual returns. However, the worst returns are when the president is of the opposite party of both houses of congress, averaging a 7% annual return. (Smith & Woodley 2020)

Section III: Data Construction
To construct our dataset, we retrieved the adjusted closing price of the S&P 500 index for the 10 days before and after every election time from 1990 to 2020. A 10-Day volatility for the 10 days leading up to the election and a 10-Day volatility for the 10 days after the results were announced was calculated. We calculated the percentage change in the index between the day before the voting happened and the day after results were announced. Same was done for the percentage change in 10-day volatility of return leading up to and after results were announced.

A total of 8 binary variables were generated to signal the possible scenarios of the election outcome. The exhaustive list of variables included the five possible midterm election outcomes (Status-quo, R_flip, D_flip, R_sweep, D_sweep), two presidential election outcomes (Republican, Democratic) as well as the variable No_Pres, which indicates that presidential election did not occur during that cycle. Below is one example of the process of gathering adjusted close prices, calculating before and after 10-day volatilities, and measuring percentage change in volatility. The sample from excel is for the 2012 election. 

Examining the Effect of Election Outcome on the Short-Term
Price Movement of the S&P 500 Index

Section IV: Methodology
In order to analyze the correlation between election outcomes and the fluctuation of the index price, the difference between the pre- and post-election volatility measures and the percentage change in index price were linearly fitted to several combinations of the binary variables. A total of six models were considered for our statistical analysis: 1. volatility ~ senate election, 2. volatility ~ presidential election, 3. volatility ~ overall, and such models identically applied to the Index price change.

Furthermore, these regressions were performed on different time frames to cross validate the significance of the output. Various statistics, including but not limited to the F-statistics and p-value were retrieved through this procedure, which provide insight on not only the effect of election on the index, but also the significance of the eight election scenarios.

Section V: Analysis

Table 1. Coefficients of the Regression Models

In Table 1, the D_sweep and No_pres act as the base, resulting in the same coefficients as the intercept. (.), (*), (**), (***) shown in model #4 each stand for significant levels 0.05, 0.01, 0.01, and 0. 

The result shows that the Price Change Models, models #1,  #2, and #3, did not bring any compelling results. We can see this by looking at the R squared of the models where no model exceeded 10%. However, the Volatility Change Models, models #4, #5, and #6, presented noticeable differences. R squared for these models were shown much higher than those of the Price Change Models, which indicates the significance of the result. Out of the three models, model #6, the Volatility Change Model upon both Presidential and Senate elections, showed the most significance, the R squared data being nearly 75%. The Volatility Change Model upon Senate Elections also resulted in noteworthy findings, the R squared data being nearly 72%. 

On the other hand, model #5, the Volatility Change Model upon Presidential Elections did not show any greater significance than those of Price Change Models, its R squared being only 5%. 

From these findings, we can verify that the advent of the election period and the outcome certainly affects the volatility of the market. It is notable that the presidential election itself does not have a statistically significant impact on the volatility, but when the variables are added to the evaluation with congress elections, the adequacy of the model fit is enhanced. On the other hand, by checking the forecasting power of the model with R squared values, the elections do not seem to have a significant impact on the S&P 500 index price in the very short term. 

Section VI: Conclusion

Overall, as mentioned above, the changes in the congress majority and minority are significant in affecting volatility in the weeks leading up to the election and the weeks after it. This effect is not very significant in the very short run as shown in the insignificance of the percentage change in index.

However, 10 days before and after the election, the effects are significant as demonstrated by the regression. Moreover, the research confirms the conjecture by the USBank article that sweeps have a bigger impact than flips and that there isn’t much preference for one party over the others. As very few elections exist, a limitation of this research is its limited sample size. In order to expand the scope of the work, the same method was applied to data for every election from 1970 to 2020. However, the results were not as promising. Several factors can explain this. One is that from 1970-1990 elections were far less competitive than those from 1990-2020. There were almost no elections in which houses of congress changed hands. As most elections resulted in SQ, different expectations existed in the market. Different expectations and sentiment meant the investors were less likely to look at congress elections as a major factor in changes in the S&P 500 index.

This is a critical challenge in measuring the effect of elections on the market. While data exists going as far back as the 1920s, the nature of elections is less like the contemporary ones, the further we go back. Before the 1970s, elections were even less competitive than 1970-1990. An extra layer of difficulty with older elections is that the parties were simply different from what they are today in terms of political ideology. Their different approaches to economics could shape different expectations in the investors leading to different results in the market. This means the range of data available to work with is limited by the nature of the project. However, every two years, as more data is added, the estimation is improved.

Another limitation is that this project uses an OLS estimation to simplify the statistical model. More accurate results could come from using methods from time-series analysis. Another way in which the project can be expanded is repeating the same procedure but with other indicators of the market like Nasdaq or Dow Jones industrial average. 

Examining the Effect of Election Outcome on the Short-Term
Price Movement of the S&P 500 Index
Written by: Won Ha Chang, Levi Holmes, Claire Kang & Junsup Shin


Smith, Anne Kates, and Kyle Woodley. “How Presidential Elections Affect the Stock Market.” Kiplinger, Kiplinger, 26 Oct. 2020, www.kiplinger.com/investing/stocks/601629/how-presidential-elections-affect-the-stock-market 

Link to the USBank.com article: https://www.usbank.com/investing/financial-perspectives/market-news/how-presidential-elections-affect-the-stock-market.html

Examining the Effect of Election Outcome on the Short-Term
Price Movement of the S&P 500 Index

Examining the Effect of Election Outcome on the Short-Term
Price Movement of the S&P 500 Index

Ai Investing : The Future of Finance