Effects of Election Results on Stock Price Performance: Evidence from 1976 to 2008 Andreas Oehler, Bamberg University Thomas J. Walker, Concordia University Stefan Wendt, Bamberg University 2012 FMA Annual Meeting Atlanta, GA, October 17-20, 2012
Agenda (1) Introduction (2) Related literature (3) Dataset & methodology (4) Results (5) Conclusions 2
(1) Introduction Effects of election results may influence a country s overall economy single companies or sectors. U.S. political system: Democratic Party vs. Republican Party (GOP). We examine effects of the 1976 to 2008 presidential election results on the stock price performance of U.S. companies and sectors, to identify/quantify the perceived favoritism/biases. We find that election results prompt abnormal company and sector returns, effects are driven by individual presidents rather than by parties. 3
(2) Related literature Effects of U.S. presidential election on the performance of the overall U.S. stock market (Niederhoffer, Gibbs, and Bullock 1970; Huang 1985; Gärtner and Wellershoff 1995; Leblang and Mukherjee 2005; Snowberg, Wolfers, and Zitzewitz 2007a), single sectors in a single election (Roberts 1990; Herron et al. 1999). International evidence for influence on the overall market: GB (Herron 2000; Leblang and Mukherjee 2005), GR (Siokis and Kapopoulos 2007), NL (Brunner 2009), ES (Furió and Pardo 2010). sectors: D (Bechtel and Füss 2010) Causality between the economic and the political sphere flows both ways (Gerber, Huber and Washington 2009; Knight 2007; Mattozzi 2008). 4
(3) Dataset & methodology Dataset Daily stock price data (CRSP/WRDS) Company characteristics (Compustat/WRDS) Four-digit SIC-Codes Stock market capitalization Book value of debt and total assets, net income One month T-bill rate, Fama-French factors, momentum factor (WRDS) Sorted into 8 industry groups (mining; construction; manufacturing; transportation; wholesale trade; retail trade; financials; services) 5
(3) Dataset & methodology Descriptive statistics President Election Date Mining Construction Manufacturing Transportation Wholesale Retail Financials Services J. Carter 02.11.1976 18 5 212 12 6 3 34 17 R. Reagan 04.11.1980 21 3 311 23 13 7 57 24 R. Reagan 06.11.1984 49 6 554 29 36 19 110 77 G. Bush 08.11.1988 75 9 781 55 58 41 186 130 W. Clinton 03.11.1992 88 13 899 76 73 41 222 131 W. Clinton 05.11.1996 95 16 1012 91 82 42 346 148 G. W. Bush 07.11.2000 101 26 1107 107 73 48 352 227 G. W. Bush 02.11.2004 82 21 848 90 62 20 347 119 B. Obama 04.11.2008 92 25 673 54 40 19 220 97 This table provides summary statistics for each election covered by our analysis. For each president elected during our 1976-2008 sample period we document the respective political party (Republican or Democratic) and the number of companies in each of the outlined industry divisions included in the analysis. 6
(3) Dataset & methodology Methodology Event study Stock market participants will price their expectations about political change into stock prices prior to an election. However, expectations about election results are not always clear-cut. We assume that pre-election polls are not able to fully forecast election results, and that the election itself will reveal new information which, in turn, will be incorporated into stock prices. 7
(3) Dataset & methodology Methodology Event study Abnormal stock price returns for company i following the election day: [ R ] AR i, t = R i, t E i, t Expected returns calculated using four-factor model (Carhart 1997): [ ] ( ) i, t = Rf, t + β i,1 * Rm, t Rf, t + βi,2 * SMBt+ βi,3 * HMLt + βi, 4 * MOMt E R 8
(3) Dataset & methodology Methodology Event study Cumulative abnormal returns (2-, 4-, 10-, 18-, and 26-week horizon):, t t = 2 1, t2 i, CAR i AR t = t 1 t Industry cumulative abnormal returns (equally weighted): CAR = 1 N industry CAR industry t1, t2 i, t1, t2 Nindustry i= 1 where N industry is the number of companies in the respective sector. 9
(3) Dataset & methodology Methodology Regression analysis We regress the CAR on industry dummies and control variables: CAR i, t t = D i i i 1, 2 α + β * + γ *ln( mcap ) + λ * lev + θ * income + ε D: vector of dummy variables for seven industries (except retail trade) ln(mcap i ): logarithm of stock market capitalization lev i : leverage income i : net income / total assets 10
Period (4) Results Event study Election 1980 Ronald Reagan (first term) Mining Services 0-0.5-0.3 0.6 *** 0.6 1.2-0.1 0.0-1.1 +2w -2.3 1.0 0.8 * 4.8 ** 1.1 3.4-0.1-0.6 +4w -3.2-2.9 0.1 3.7 1.9 3.0-0.3 0.9 +10w -6.2 *** -4.2 0.9 1.7 1.0-2.5-2.4 2.9 +18w -6.9 * -8.1 2.7 *** 0.8 3.1-1.2-4.4 ** 4.5 +26w -13.2 *** 2.0 3.7 *** -2.4 7.9 * 1.7-5.5 *** 7.4 We report mean cumulative abnormal stock price returns for eight industries following the presidential election in 1980. The event windows are denoted as 0 (the day following the election), +2w (two weeks following the election), etc. All values represent percentage returns. The symbols ***, **, and * denote statistical significance at the one, five, and ten percent level, respectively. Financials Construction Manufacturing Transportation Wholesale Retail 11
Period (4) Results Event study Election 1992 William Clinton (first term) Mining Services 0-0.3 3.0 ** 0.1 1.0 * -0.3-0.1 0.2-0.4 +2w -0.8 5.8 2.2 *** 4.7 *** 0.4 0.8 1.7 *** 2.1 +4w 1.3 10.9 ** 3.0 *** 9.2 *** 1.3 2.4 2.2 *** 5.5 * +10w 3.1 15.7 ** 5.2 *** 13.9 *** 3.7 ** 7.9 ** 5.9 *** 14.0 * +18w 11.9 *** 22.7 ** 5.0 *** 16.6 *** 5.1 10.9 ** 10.2 *** 17.6 * +26w 25.7 *** 17.9 6.8 *** 22.0 *** 6.7 * 9.3 ** 9.5 *** 25.4 ** We report mean cumulative abnormal stock price returns for eight industries following the presidential election in 1992. The event windows are denoted as 0 (the day following the election), +2w (two weeks following the election), etc. All values represent percentage returns. The symbols ***, **, and * denote statistical significance at the one, five, and ten percent level, respectively. Financials Construction Manufacturing Transportation Wholesale Retail 12
(4) Results Event study Election of Ronald Reagan (Republican Party) in 1980 Longer-run significant negative effect on mining and financial sector. Slight positive effect on manufacturing sector. All other sectors largely insignificant. Election of William Clinton (Democratic Party) in 1992 Mixed short-run effects: only slight effect of the election itself. Significant positive abnormal longer-run returns in all sectors. The market appears to positively reflect the first political decisions of the incoming president. 13
Period (4) Results Event study Election 2000 George W. Bush (first term) Mining Financials Construction Manufacturing Transportation Wholesale Retail Services 0 0.4 0.4-0.4 ** 1.0 ** 1.0-1.2 ** 0.4 ** 0.8 +2w -5.5 *** -1.5-1.7 *** 0.8-2.1-8.0 *** 0.0-1.4 +4w -8.3 *** -3.8-2.4 *** -0.5-3.2-10.5 *** 2.1 *** -4.0 *** +10w 4.9 1.4 0.1 4.6 * 4.2-5.4 4.6 *** 5.3 *** +18w 1.4 6.7 5.3 *** 6.0 * 11.1 *** 3.4 12.0 *** 9.0 *** +26w 0.2 12.4 7.4 *** 11.2 *** 15.2 *** -1.3 12.4 *** 10.5 *** We report mean cumulative abnormal stock price returns for eight industries following the presidential election in 2000. The event windows are denoted as 0 (the day following the election), +2w (two weeks following the election), etc. All values represent percentage returns. The symbols ***, **, and * denote statistical significance at the one, five, and ten percent level, respectively. 14
Period (4) Results Event study Election 2008 Barack Obama Mining Services 0 1.3 * 1.5 0.1-0.3 0.5-0.5-0.8 0.1 +2w -20.3 *** -8.4 ** -5.4 *** -2.1-7.9 ** -0.6 2.6 * -4.9 ** +4w -17.2 *** 4.2-4.3 *** -3.7-9.7 ** 0.8 4.4 *** -3.6 +10w 11.1 ** 16.1 *** 3.1 ** 2.5 1.5-2.8 4.3 ** 5.4 * +18w 15.7 ** 11.7 * -5.0 *** -10.0-7.7-14.0 1.5 3.8 +26w 66.0 *** 26.9 *** 16.8 *** 27.3 14.8 ** -5.8-1.8 31.9 *** We report mean cumulative abnormal stock price returns for eight industries following the presidential election in 2008. The event windows are denoted as 0 (the day following the election), +2w (two weeks following the election), etc. All values represent percentage returns. The symbols ***, **, and * denote statistical significance at the one, five, and ten percent level, respectively. Financials Construction Manufacturing Transportation Wholesale Retail 15
(4) Results Event study Election of George W. Bush (Republican Party) in 2000 Short-term negative abnormal returns in most sectors, except financials and transportation. Followed by a reversal and even positive abnormal returns in the longer run. A negative first effect appears to be dampened and even reversed as consequence of actual decision making. Election of Barack Obama (Democratic Party) in 2008 Overall picture is unclear: Very mixed results across the industries and across time. The immense extent of the longer-run effect is somewhat puzzling. 16
Period Andreas Oehler, Thomas J. Walker, Stefan Wendt (4) Results Regression analysis Panel A: Elections won by a Democrat candidate 0 +2w +4w +10w +26w Constant 0.005 0.033 ** 0.037 ** 0.061 ** 0.043 Mining 0.003-0.076 *** -0.066 *** 0.015 0.248 *** Contruction 0.013-0.041 0.017 0.073 * 0.098 Manufacturing -0.001-0.022-0.019-0.023 0.011 Transportation 0.003-0.005 0.003 0.005 0.082 Wholesale -0.004-0.029-0.037 * -0.037-0.010 Financials -0.002-0.001 0.003-0.019-0.022 Services -0.006-0.025-0.012 0.009 0.107 ** ln(mcap) -0.001 ** -0.003 *** -0.003 ** 0.000 0.005 leverage 0.000 0.000 0.000 ** 0.000 0.000 ** net inc./tot. ass. 0.001 0.059 0.080 * 0.076 0.043 Adj. R-squared 0.001 0.013 0.008 0.002 0.016 Prob(F-stat) 0.195 <0.001 <0.001 0.038 <0.001 We report regression results with mean CAR following the presidential election days in 1976, 1992, 1996, and 2008 as dependent variables. The event windows are denoted as 0 (the day following the election), +2w (two weeks following the election), etc. The symbols ***, **, and * denote statistical significance at the one, five, and ten percent level, respectively. 17
Period Andreas Oehler, Thomas J. Walker, Stefan Wendt (4) Results Regression analysis Panel B: Elections won by a Republican candidate 0 +2w +4w +10w +26w Constant -0.004-0.036 *** -0.043 *** -0.023 0.026 Mining 0.005 0.011 0.003 0.014-0.030 Contruction 0.009 * 0.019 0.014 0.046 0.048 Manufacturing 0.003 0.028 ** 0.030 ** 0.028 0.019 Transportation 0.007 ** 0.035 *** 0.038 ** 0.039 * 0.030 Wholesale 0.006 0.022 * 0.034 ** 0.044 ** 0.059 * Financials 0.005 * 0.028 ** 0.042 *** 0.027 0.022 Services 0.006 0.022 * 0.018 0.035 * 0.026 ln(mcap) 0.000 0.001 ** 0.002 ** 0.002 * -0.002 leverage 0.000 0.000 0.000 0.000 0.000 net inc./tot. ass. 0.012 0.083 *** 0.040 0.091 * 0.154 * Adj. R-squared 0.000 0.005 0.005 0.001 0.001 Prob(F-stat) 0.355 <0.001 <0.001 0.072 0.044 We report regression results with mean CAR following the presidential election days in 1980, 1984, 1988, 2000 and 2004 as dependent variables. The event windows are denoted as 0 (the day following the election), +2w (two weeks following the election), etc. The symbols ***, **, and * denote statistical significance at the one, five, and ten percent level, respectively. 18
(4) Results Regression analysis Elections won by a Democratic candidate: Hardly any influence of the industry the companies belong to. Some influence of the mining sector. Market participants do not appear to expect industry-specific bias in decision making after the election of a Democratic candidate. Elections won by a Republican candidate: Significant positive influence for companies within the manufacturing, transportation, wholesale trade, and financial sectors. This effect is largely reversed in the longer run. Overall explanatory power is weak. 19
(5) Conclusions Elections of all recent U.S. presidents have prompted abnormal company and sector returns, especially in the longer run. Two potential interpretations: the market remains uncertain; the market struggles to reconcile the effects of political changes. Our results support the hypothesis that, following a presidential election, the market corrects, and thus reflects changes in the underlying governing philosophy. No consistent party-specific influence, effects are primarily driven by individual presidents. 20
Thank you very much for your attention! 21