I exploit the sudden and dramatic jolt that Osama Bin Laden’s capture gave to Barack Obama’s 2012 re-election prospects to gauge the relationship between presidential prospects and stock market valuation changes. Using campaign contributions as an indicator of political support, I find that following Bin Laden’s death, firms that had previously supported Democrats registered significant positive returns, whereas firms that had supported Republicans registered significant negative returns. Across the S&P 500, the president’s transformed re-election prospects shifted market capital worth $101 billion over one day and $245 billion over one week. My findings indicate that the relationship between the presidency and firm valuations is associated with patterns of past political support, substantively and significantly important, and more pronounced for the presidency than for Congress.
Osama Bin Laden’s capture on May 1, 2011 unexpectedly and dramatically augmented President Barack Obama’s re-election prospects. Because the operation to locate Bin Laden was shrouded in secrecy, market participants had no prior knowledge of it. I use this exogenous jolt to Obama’s re-election prospects to study whether firms that had previously donated to Democrats (or Republicans) gained (or lost) stock market value from the suddenly increased probability that the president would win a second term. My approach neither adjudicates the causal effect of contributions on firm valuations, nor assumes that investors believe that contributions influence political outcomes. It accurately ascertains, however, the value that the presidency bestows on firms, and provides evidence that this relationship is associated with patterns of past support.
The small but growing literature investigating how political events impact politically oriented firms has produced contradictory results. Although some studies argue that political alignments do not matter for firm valuations (e.g., Ansolabehere et al., 2004; Fisman et al., 2012; Werner, 2011), others demonstrate that they do matter (e.g., Jayachandran, 2006; Knight, 2006; Mattozzi, 2008; Monroe, 2010). These inconsistent results likely stem from the nature of the event analyzed; it may be that only entirely unanticipated and highly salient political events generate discernible market reactions.1 The event I study satisfies both requirements. In addition to being a complete surprise, Bin Laden’s capture resulted in the largest increase and the highest level reached in the president’s re-election probability during the entire election cycle, thereby making it an ideal case for adjudicating between competing claims in the literature.
Apart from leveraging a rare and potent exogenous event, my study takes advantage of methodological improvements and new sources of data to build upon and extend the findings of its predecessors. My approach follows that of Jayachandran (2006), whose study shows that partisan realignments in the Senate impact the valuations of politically oriented firms. I focus on the presidency, however, and I improve on the study’s method by using prediction markets data to independently identify the extent and duration of the president’s electoral boost.2
Studies concentrating on presidential electoral probabilities (e.g., Knight, 2006; Mattozzi, 2008) explore how investors use stock markets to hedge against political uncertainty or how policy platforms are capitalized into equity prices, not how contributions correlate with firm returns, which is my focus.3 Moreover, they investigate presidential races over extended election cycles during which confounding developments such as concurrent political races (Shon, 2010) or economic events can affect both electoral probabilities and stock prices (Snowberg et al. 2011). The intra-day and short-term market reactions I study, by contrast, alleviate concerns about reverse causality and omitted variable bias that arise from using protracted event windows. Furthermore, I disentangle firm contributions from industry contributions in order to study how the presidency shapes both industry and firm-level prospects.
My study documents the precise magnitude of the effect of the president’s re-election probability on firm valuations. There is considerable debate about the impact of the presidency on broader economic outcomes, especially in regard to inequality, taxation, and redistribution (Hacker and Pierson, 2012), and the impact of the presidency on firm-level outcomes is similarly unclear.4 Prior studies have documented a relationship between the presidency and market-wide outcomes, and have shown that the presidency typically has a bigger impact than Congress on markets (Snowberg et al., 2007). My results indicate that the presidency is more consequential than Congress for the valuation of politically oriented firms.
1 Measuring Stock Price Responses to Presidential Prospects
1.1 Event
As the mastermind behind the September 11th terrorist attacks and the head of al-Qaeda, Bin Laden’s capture had immediate and dramatic significance for both foreign and domestic politics. Given the absolute secrecy of Operation Neptune Spear, it is implausible that market participants could have anticipated either the timing of Bin Laden’s capture or its political repercussions (e.g., Bergen, 2012). The event was made public on Sunday, May 1, 2011. I thus analyze changes in equity valuations after Friday, April 29th, the previous trading day.
1.2 Effect
Receiving credit for Bin Laden’s capture unambiguously enhanced Obama’s political standing, at least temporarily. Following the event, media pundits and equity research analysts all opined that Obama’s re-election chances had substantially improved; additionally, a Pew Research Center poll conducted on the heels of Bin Laden’s capture showed that Obama’s approval rating had increased by 19%.5 Prediction markets — websites that allow traders to make real-time bets on key events — quantify the extent to which Bin Laden’s capture affected Obama’s re-election prospects. On May 2nd, there was a 19% increase in Obama’s re-election probability, the sharpest change on any day during the entire election cycle, as well as one of the highest probabilities reached.6 In short, Bin Laden’s capture was widely viewed as a significant political victory for Obama and the Democrats, and investors knew it.
1.3 Event Window
The prediction markets data suggests that the spike in Obama’s re-election probability was dramatic but did not last very long. After a sharp increase on the first trading day after Bin Laden’s capture, the predicted probability subsided but remained above pre-May 1st levels for one week.7 To the extent that prediction markets accurately reflected stock market expectations about Obama’s re-election probability, a one-day period would be the appropriate event window. Yet, investors potentially responded not only to the capture itself but also to polling data disclosures and media commentary about the impact of the event on Obama’s re-election probability, which dominated wires for several days (see Online Appendix). Because Obama’s re-election probability remained elevated for a week, and because it is plausible that prediction markets did not precisely aggregate stock market expectations during this early stage of the election cycle, I also examine returns for the entire week. Furthermore, as a sensitivity check, I test whether my effect persisted or dissipated once re-election probabilities fell below pre-event levels. That both prediction and stock markets registered strong initial reactions to Bin Laden’s death indicates that the event was materially significant; the possibility that the event was not material would only bias my analysis toward null results.
1.4 Exclusion Restriction
Could Bin Laden’s capture have affected firm valuations independently of its impact on re-election prospects? It is plausible that the perceived geopolitical consequences of Bin Laden’s death might have moved market-wide valuation trends.8 It is possible, additionally, that Bin Laden’s death might have had different geopolitical consequences for industries that also happened to have partisan orientations. A search of news articles and analyst reports in the week of May 2nd reveals no evidence for these hypotheses.9 Bin Laden’s capture was viewed both as a harbinger of global stability and as a potential instigator of retaliatory terrorist attacks, and it appears that its net geopolitical impact was minimal.
1.5 Methodology
Using standard event study methodology (e.g., Jayachandran, 2006), I first estimate the equation: , where represents the stock return for firm i during the event window e, and Democrati and Republicani are continuous variables representing a firm’s prior donations to the Democratic and Republican Party, respectively. Returns are calculated as the change in stock price over the window, divided by the stock price prior to the event. I test whether βD > 0 and βR< 0.
I next control for the relationship between an individual stock and the broader market by using abnormal returns. First, I estimate the linear relationship between an individual firm’s returns and market returns for each day t in a pre-event period, using the equation: Returnit = αi + βi Market Returnt + εit.10 Based on this relationship, I predict the expected return of the security during each event window. Second, I subtract the expected return from the actual return, in order to the isolate the firm-specific component of the valuation changes, by using the formula: .
1.6 Data
My sample of firms is the Standard & Poor’s list of the 500 largest companies that are actively traded in the United States, which I obtained, as of April 29th, from Bloomberg. I collected stock price data and market data from the Center for Research on Security Prices (CRSP), firm-specific variables from Compustat, and intraday trading data from the Trade and Quote Database.11 To measure business PAC contributions, I matched committee contributions data from the 2007–2008 and 2009–2010 election cycles compiled by the Center for Responsive Politics (CRP) to firms on the S&P 500 list. The Online Appendix discusses the matching procedures and provides descriptive statistics.
2 Empirical Findings
Table 1 demonstrates that firms that had contributed to Democrats during the prior two election cycles received a significant market boost after Bin Laden’s death, while firms that had contributed to Republicans performed significantly worse. These results are qualitatively the same whether I utilize contributions data from the 2009–2010 election cycle (Panel A) or the 2007–2008 election cycle (Panel B). Additionally, whether I consider abnormal returns (Column 1) or actual returns (Column 2), or control for firm-specific indicators (Column 3), these results persist.12 Column 3 shows that every $100,000 donated to the Democratic Party in a past election cycle was associated with a 0.194% higher abnormal return using a one-day event window, while every $100,000 donated to the Republican Party was associated with a 0.159% lower abnormal return.
Relationship between corporate contributions and stock returns.
| One-day window | ||||||||
|---|---|---|---|---|---|---|---|---|
| Abnormal returns | ||||||||
| Variables | Abnormal returns (1) | Raw returns (2) | Firm controls (3) | Intensive margins (4) | Regulated industry (5) | Industry donations (6) | Abnormal dummy (7) | One-week window (8) |
| Panel A: 2009–2010 data | ||||||||
| Donations to Democrats/$100,000 | 0.00195 | 0.00197 | 0.00194 | 0.00180 | 0.00193 | 0.00127 | 0.09766 | 0.00689 |
| (0.00054) | (0.00053) | (0.00055) | (0.00054) | (0.00055) | (0.00058) | (0.02576) | (0.00182) | |
| Donations to Republicans/$100,000 | −0.00148 | −0.00140 | −0.00159 | −0.00150 | −0.00157 | −0.00106 | −0.07738 | −0.00702 |
| (0.00054) | (0.00052) | (0.00056) | (0.00058) | (0.00056) | (0.00052) | (0.02533) | (0.00190) | |
| Regulated industry dummy | 0.00043 (0.00181) | |||||||
| Industry donations to Democrats/$100,000 | 0.00030 (0.00016) | |||||||
| Industry donations to Republicans/$100,000 | −0.00028 (0.00013) | |||||||
| Observations | 500 | 500 | 496 | 292 | 496 | 496 | 496 | 496 |
| Panel B: 2007–2008 data | ||||||||
| Donations to Democrats/$100,000 | 0.00207 | 0.00200 | 0.00190 | 0.00191 | 0.00188 | 0.00150 | 0.10470 | 0.00797 |
| (0.00083) | (0.00082) | (0.00087) | (0.00083) | (0.00085) | (0.00082) | (0.03785) | (0.00273) | |
| Donations to Republicans/$100,000 | −0.00149 | −0.00133 | −0.00144 | −0.00149 | −0.00142 | −0.00106 | −0.07891 | −0.00758 |
| (0.00080) | (0.00078) | (0.00082) | (0.00079) | (0.00081) | (0.00077) | (0.03624) | (0.00250) | |
| Regulated industry dummy | 0.00051 (0.00176) | |||||||
| Industry donations to Democrats/$100,000 | 0.00023 (0.00020) | |||||||
| Industry donations to Republicans/$100,000 | −0.00021 (0.00017) | |||||||
| Observations | 500 | 500 | 496 | 284 | 496 | 496 | 496 | 496 |
| Firm-specific controls | No | No | Yes | Yes | Yes | Yes | Yes | Yes |
Note: Robust standard errors clustered by industry in parentheses. One-day return calculated as of May 2, 2011, the first trading day after Bin Laden’s capture. One-week return calculated as of May 6, 2011. Firm-specific controls are log of total assets (proxy for firm size), fixed-to-total assets ratio (proxy for asset tangibility), total assets-to-total liabilities (proxy for leverage), earnings before interest and taxes-to-total assets (proxy for operating profitability), and sales growth (proxy for growth opportunities). Regulated industries refer to railroad, public utilities, banking, finance, and insurance industries (SIC codes 40, 48, 49, 60, 61, and 63). “Abnormal dummy” is a dummy variable that takes a value of one if abnormal returns were positive.
I next conduct several analyses to explore political factors related to the event study. First, I restrict my sample to only those firms that donated money (Column 4) in order to ascertain the extent to which contributor/non- contributor margins explain the results, and find similar coefficients to those in the extensive margins analysis. I also investigate whether firms in regulated industries are driving my results (Column 5), as it is possible that these firms are more sensitive to the executive; the coefficient on the regulated industry dummy is statistically indistinguishable from zero, suggesting that Democrat and Republican donor firms had larger stock price changes than regulated firms. Additionally, I explore whether industry-wide contributions reflect political alignments that benefit individual firms. For each firm in my sample, I subtract the firm’s own contributions from the total contributions of all other S&P 500 firms in the industry, and include this variable in the regression (Column 6).13 The coefficients on individual firms’ donations are larger and have more predictive power than the coefficients on industry donations. Moreover, to study the role of outliers, I convert my dependent variable into a dummy that takes a value of one if abnormal returns are positive (Column 7). Every $100,000 donated to Democrats was associated with a 9.77% higher probability that the firm gained market value; donations to Republicans have the opposite effect. Last, I extend my analysis to a one-week period (Column 8), and find larger effects in this window.
Figure 1 charts the estimated coefficients and 95% confidence intervals on business campaign contributions to Democrats and Republicans, respectively, over the one-week period following Bin Laden’s capture using the specification presented in Column 3. The figure shows that effects were registered on the first trading and persisted throughout the week. When I extend the analysis to the subsequent week, during which re-election probabilities fell below pre-event levels, the effect dissipates and the coefficients become indistinguishable from zero (see Online Appendix). Figure 2 plots the intraday tick-by-tick stock trade prices of politically oriented firms on May 2nd.14 The timing and consistency of these valuation differences provide additional evidence that stock prices responded specifically to Bin Laden’s capture, as opposed to unrelated events that might potentially be associated with political contributions.
An alternate way to interpret these findings is to study the total change in firm market valuations following Bin Laden’s capture. I multiply the predicted residuals for each firm from the model presented in Column 3 by the firm’s market capitalization on the trading day prior to the event. The sum of the absolute values of these amounts represents the magnitude of the market value change that can be attributed to Obama’s perceived electoral boost. Across my sample of firms, the total change in market capitalization was $101 billion over one day and $245 billion over one week, between one and two percent of the firms’ total market value.15
3 Discussion
My findings show that the fortunes of firms parallel the fortunes of the politicians they support. The data neither captures the causal effect of contributions on valuations nor identifies the reason for this correlation. It is possible, for example, that businesses donate to supportive politicians, such that policy outcomes remain unaffected by donations. Yet, that firms gain or lose from political events in ways that are correlated with past support suggests that donations might serve ends beyond simple consumption, a view found in previous work (Ansolabehere et al., 2003). It is equally plausible that contributions proxy the favorability of political parties toward certain industries. By disaggregating firm- and industry-level contributions, however, I provide evidence that political outlooks might influence firm-level prospects in addition to industry-level prospects. Whether investors disentangle the relationship between contributions, political prospects, and firm outlooks is an open question — one that I leave to future studies.
Comparing the magnitude of these effects with the results of prior event studies suggests that the presidency is more important than Congress for the valuation of politically oriented firms.16 This is perhaps surprising, because Congress — not the president — determines appropriations and other types of policies that donating firms might conceivably find interesting. One potential explanation for this finding is that the president shapes business outlooks, and that business contributions mirror the president’s policy preferences. Another possibility is that investors do not differentiate between the various political influences on business policies, but instead simply look to the president as the bellwether for the overall direction of public policy in the nation.
This paper leverages a short-term spike in electoral probabilities to study temporary valuation differentials in equity prices. My preferred interpretation of the finding is that the temporary jolt to both electoral probabilities and firm values reflected fundamental changes in the expectations of investors, and that this relationship sheds light on how a permanent change in electoral probabilities (and, ultimately, the president’s election itself) would lead to sustained changes in the valuation of politically oriented firms. The short-lived nature of the event can reasonably be viewed to reflect updated market expectations in light of other events that overtook Obama’s electoral prospects, as well as new disclosures about the passing impact of Bin Laden’s death on the presidential election. This interpretation, however, is speculative for a number of reasons. First, the temporary change in re-election probabilities and firm economic outlooks in this case may have reflected investor speculation rather than fundamental changes. Second, the effect of a permanent change might be different than a temporary one, as firms and parties adapt their behavior over time. Scholars are interested in long-term relationships, yet measuring incremental valuation differentials over protracted time periods is fraught with methodological difficulties; the benefit of my approach is that short-term price reactions afford precise measurements. Although this methodology does not sort through questions related to the causal interpretation of the findings, it nonetheless accurately captures how exogenous jolts to electoral expectations trigger instantaneous valuation differentials for politically oriented firms. That political outcomes matter for market performance suggests that firms might face incentives to influence these outcomes via campaign contributions.
Even if the judicial verdicts on campaign finance analyzed by Ansolabehere et al. or Werner contained information entirely new to shareholders, to the extent that these verdicts were material, investors might have developed prior expectations about them based on media coverage, expert commentary, or courtroom disclosures. Similarly, in studies that look at politicians’ health shocks (e.g., Fisman et al.), investors might previously have priced health concerns into stock prices. Likewise, even in events with considerable uncertainty, such as tossup elections, stock prices should already have incorporated the probability that the election would swing in either direction.
Snowberg et al. (2011) argue that prediction markets data can be used to ascertain event windows. The start of the event window is difficult to identify in Jayachandran’s study, because rumors about the Senate shift were aired for days before the switch was announced. Additionally, the study considers event windows ranging from four days to six months, but does not offer an ex ante rationale to favor any particular time frame of analysis.
Knight selects a sample of 70 firms that equity analysts identify as partisan-leaning firms, and only uses campaign contributions as a reliability check. Mattozzi applies contribution thresholds to identify a small subset of firms that potentially fare differently under Democrat or Republican control. Both studies investigate how portfolio movements correlate with electoral polls, not how contributions are associated with stock returns.
On the one hand, the presidency might offer few direct benefits to firms. Presidents depend on the support of broader constituencies, receive heightened media scrutiny, and endure a sustained vetting process that arguably makes them less amenable to capture by special interests. On the other hand, media coverage and round-the-clock attention ensures that market participants are acutely aware of factors influencing presidential prospects, and the downstream implications of these prospects for politically oriented firms.
See, e.g., Business Insider, May 2, 2011; KeyBanc Capital Markets, May 2, 2011; J. P. Morgan, May 2, 2011; Deutsche Bank, May 2, 2011; Pew Research Center Publications, May 3, 2011; and Online Appendix.
The only other period of a higher re-election probability was 13 days in September and October 2012.
My methodology relies on the assumption of market efficiency, which holds that stock prices instantaneously incorporate changes in expectations of materially relevant future events. This methodology only requires that investors at the time thought that Obama’s re-election prospects had materially changed. Relying on prediction markets data to ascertain the stock market event window presupposes an additional assumption of efficiency in the prediction markets. Scholars have demonstrated that prediction markets accurately gauge stock market sentiments, especially close to election periods (e.g., Snowberg et al., 2011). The average daily trading volume, one indicator of market efficiency, in the prediction markets in early 2011 was a small fraction of the levels that were registered closer to the election in late 2012. It is possible that during this early stage of the election cycle, prediction markets traders were not as representative of stock market investors as in later stages. One reasonable strategy in this case, therefore, would be to consider the prediction markets data as a suggestive, rather than a definitive, guide for ascertaining the stock market event window.
The market-wide reaction to Bin Laden’s capture was muted. Although there was an initial uptick in equities during early trading on May 2nd, by the close of trading the market had reverted to earlier levels and actually lost some value. (See, e.g., The New York Times, May 2, 2011.) Additionally, my methodology controls for market-wide movements and ascertains the differential effects of contributions to Democrats and to Republicans.
I read through every news article (6,758) and equity research report (271) that discussed Bin Laden’s capture during the period of analysis in order to identify industries that the event might have directly influenced. Excluding these industries from the analysis produces no qualitative difference in the results.
I use a one-year period beginning one week prior to the event (April 25, 2010–April 25, 2011).
For Brown-Forman Corp., Berkshire Hathaway Inc., CBS Corp., Constellation Brands Inc., Lennar Corp., McCormick & Co., and Molson Coors Brewing Co., which have two classes of common shares, I analyze the classes that were listed on the S&P 500 or had more common stock outstanding; my results are insensitive to the class of shares considered for these stocks. My market index is CRSP’s value-weighted market portfolio.
I report robust standard errors clustered at the industry level using 4-digit SIC codes; results are insensitive to 4-digit or 3-digit codes. Federal Election Commission data also include negative values for refunds or reattributions of contributions; I include these negative values in my contribution totals, but my results remain the same if these are excluded. Control variables are calculated as four-year averages lagged to the prior presidential election cycle (Claessens et al., 2008).
I use CRP categorizations of industries that have shared interests and alignments in the political arena. I also analyze the extent to which variation in contributions within industries are associated with stock returns after the variation in contributions across industries is taken into account (Online Appendix).
After indexing the 11 million trades on May 2nd to pre-event closing prices, for every second of the trading day, I plotted the median of all trades for firms that had donated to Democrats and Republicans, respectively. The Online Appendix contains a black and white version of this figure.
My results are robust to several sensitivity tests. To explore whether alternate developments were influencing my results, I exclude firms with earnings announcements and industries with unrelated developments from my analysis, and find qualitatively similar results. To explore the exclusion restriction, I exclude industries that equity analysts and news commentators speculated might directly be influenced by Bin Laden’s death, and also find similar results. Additionally, I conduct different types of placebo tests and use alternate model specifications. See Online Appendix.
Jayachandran (2006) estimates, for example, that the shift in the Senate from Republican to Democratic majority decreased the valuations of firms that had donated to Republicans by 0.3%, and had a statistically insignificant impact on firms that had donated to Democrats. Using a comparable event window and regression specification, I find that the increased likelihood of the Democrats retaining the presidency decreased Republican-oriented firm returns by 0.7%, and increased Democratic-oriented firm returns by 0.7%. When Jayachandran considers a shorter event window, her coefficients become statistically indistinguishable from zero; by contrast, even with a one-day event window, the magnitudes of my effects are large and significant. Although these comparisons focus only on the Senate, there are few reasons to doubt that the findings do not generalize to Congress.
