Negation of Sanctions: The Personal Effect of Political Contributions

Abstract We show that political contributions are associated with reduced civil and criminal sanctions for fraudulent executives. These managers benefit more from contributions if their firm also gained from the fraud, if they occupy top positions in firms with weak boards, or if they contribute to powerful politicians. Political contributions reduce budgetary resources for government enforcers and lengthen the Securities and Exchange Commission’s case time-to-resolution. They also facilitate penalty transfer from fraudulent managers to the firm, resulting in their entrenchment and long-term destruction of shareholder value. Our findings highlight an agency cost of political contributions and a mechanism undermining the disciplining effect of regulations.


I. Introduction
Do executives use political contributions to avoid government sanctions?The existing literature examining the efficacy of regulatory discipline has highlighted the role of political contributions.Most of these studies focus on the governance effect of political contributions at the firm level.For instance, politically connected firms are less likely to be investigated for fraud by government agencies (Yu and Yu (2011)).Despite political contributions being more likely to be used to "seek favors for managers rather than firms" (Aggarwal, Meschke, and Wang (2012)), there is little direct evidence of how these contributions benefit individual executives.
In this article, we examine whether and how political contributions affect the severity of government sanctions imposed on executives charged with fraud.Exploring this impact on sanctions at the individual level allows us to evaluate to what extent political spending distorts the enforcement process and shifts penalties from fraudulent managers to shareholders.
We collect information on political contributions and manually assemble a data set comprised of penalties imposed by government agencies on individual executives.This allows us to explore government sanctions in multiple dimensions: civil penalties imposed by the U. S. Securities and Exchange Commission (SEC) in terms of monetary fines and officer bans, and criminal penalties imposed by the Department of Justice (DOJ) in terms of probation and prison sentences.
We document a negative relationship between political contributions and the severity of government enforcement on executives accused of engaging in fraudulent activities.A $10,000 increase in annual political contributions is associated with a reduction in monetary penalty equivalent to 1.88% of annual compensation and 53 fewer days banned as an officer or director.Given that an average fraudulent executive in our sample earns $3.2 million per year, a 53-day reduction in officer ban translates into a saving of just over $458,000 in compensation. 1Each additional $10,000 in contributions is also associated with 2.67% lower probability that a fraudulent executive faces criminal investigation, and for those who are criminally prosecuted, 97 fewer days of probation and 89 fewer days of prison.
There are significant distributional effects as fraudulent executives benefit from political contributions to a larger extent if their firms also gained from the fraud, or if they hold top executive positions rather than lower-ranked ones, especially when their boards are in a weaker position to discipline these managers.Contributions to powerful politicians (such as those who remain in office at the time of penalty assessment or those who are ranked highly within their parties) matter more.Arguably, these politicians are more capable of directly and indirectly influencing government agencies' ability to penalize fraudulent executives.
In most of our analyses, we saturate the empirical models with fraud type-, executive role-, settlement year-, and industry-fixed effects, which allow us to control nonparametrically for fraud characteristics, the capacity of the executive to engage in fraud, and time-and industry-specific shocks.In addition, we consider two alternative hypotheses: "benefit exceeding harm" and "earned leniency."The former postulates that when assessing penalties, the SEC and DOJ may take the net effect of the work of an accused executive into consideration if he has otherwise managed the firm well and increased shareholder wealth.The latter suggests that if a firm has generally exercised good governance, which allows for putting its best efforts forth to comply with SEC regulations, this compliance may earn that firm (as well as its executives) leniency for any wrongdoing.We provide evidence that the net benefits accrued to shareholders and earned leniency do not fully explain the disparities in government penalties between fraudulent executives who make generous political contributions and those who do not.
To further consider endogeneity in political contributions, we conduct a difference-in-differences analysis, exploiting the enactment of the Bipartisan Campaign Reform Act (BCRA)a regulatory event that affects the intensity and effect of contributions.Alternatively, we construct a Hainmueller (2012) entropy-balanced matching sample.These tests confirm the negative effect of political contributions on the severity of government penalties.
Next, we explore two possible channels through which political contributions may allow fraudulent managers to dodge harsh sanctions.First, researchers have generally acknowledged that resource limitations constrain the activities of government agencies and affect the overall effectiveness of their enforcement (e.g., Jackson and Roe (2009), Kedia and Rajgopal (2011)).We collect budgetary and staffing information from the SEC and DOJ websites and construct proxies to capture case load per employee and per budget.Fraudulent managers appear to receive lenient penalties for their misconduct when their political contributions lead to tighter resource allocation within these agencies.
Second, we postulate that a fraudulent executive could potentially reduce the severity of his penalty if he can delay resolution of the case long enough.This is because, within the SEC, enforcement actions that take longer to resolve are often closed with reduced or no penalties, particularly when a new attorney is assigned to the case, or when a new SEC Commissioner is appointed (Katz (2010)).We find that political contributions increase the case time-to-resolution, allowing for less severe penalties from the SEC.
In the last part of the article, we provide evidence consistent with political contributions facilitating a transfer of penalty from fraudulent managers to the firm and its shareholders, a finding that previous analyses performed at the firm level have belied.As individual executives enjoy more lenient government sanctions as a result of political contributions, penalties imposed on the firm increase.Further, the penalty-shifting associated with political contributions reduces the likelihood that the fraudulent executive is terminated after the fraud.The entrenchment of these managers further amplifies the negative impact on the firm's shareholders.In particular, firms where executives contribute politically and are able to entrench themselves see a significant reduction in their valuation and performance relative to firms without these politically connected, entrenched managers.The effect can last up to 5 years after the case resolution, or more than 9 years after the end of the fraud.Overall, the penalty transfer and managerial entrenchment exacerbate the damages inflicted on shareholders, who have already suffered adverse consequences from the fraud.
Our article contributes to the large literature studying the real impacts of political connections and contributions.On one hand, political contributions and connections benefit shareholders in terms of higher market values (Fisman (2001), Faccio (2006), Claessens, Feijen, and Laeven (2008), and Cooper, Gulen, and Ovtchinnikov (2010)), more government contracts (Goldman, Rocholl, and So (2013)), and a higher likelihood of government bailouts (Faccio, Masulis, and McConnell (2006)).On the other hand, political connections can pose challenges for corporate governance.In particular, Yu and Yu (2011) find that political spending allows firms to deter fraud detection.Correia (2014) shows that politically connected firms are less likely to be involved in SEC enforcement and for those firms that are, the monetary penalties assessed from the SEC to the firm are less.
This strand of literature mostly focuses on how political contributions benefit or harm the firm.To the best of our knowledge, our article is among the first to employ a comprehensive data set comprising civil and criminal sanctions on individual executives to examine how political connections can generate personal benefit.Examining penalties in multiple dimensions and at the individual level helps us to gauge to what extent political contributions impair different regulatory disciplining mechanisms.
By offering new evidence that entrenched fraudulent managers impose a lasting negative impact upon shareholders, we also add to prior studies documenting that political contributions lead to a reduction in firm penalties (e.g., Correia (2014)), which potentially benefits shareholders.Our findings suggest that any benefits accrued to shareholders from reduced firm penalties may be depleted (or even overwhelmed) if political spending facilitates penalty transfer from fraudulent managers to their firm, resulting in the postfraud entrenchment of these managers and long-term destruction of shareholder value.As such, our findings imply that political contributions can skew the government enforcement process in ways that benefit managers, allowing executives to shift the consequences of fraud they committed away from themselves, even if that means shifting the consequences to shareholders who have already been hurt by the fraud.
Our article also contributes to the literature on corporate fraud (see Yu (2013) for a survey).A strand of this literature investigates the effectiveness of mechanisms including government agenciesthat are designed to detect fraud (e.g., Dyck, Morse, and Zingales (2010), Kedia andRajgopal (2011), andYu andYu (2011)).Instead of the efficiency of government agencies in detecting fraud, we focus on how political contributions affect their ability to regulate and discipline after the fraud is discovered.By documenting that political spending influences the government enforcement process above and beyond fraud detection, our article suggests that penalties are not optimal and that the negative effect of political contributions on corporate governance may have been previously underestimated.In light of Karpoff, Lee, and Martin (2008), who show that managers responsible for corporate financial fraud suffer negative career and monetary consequences, we identify a mechanism that undermines this disciplining effect for managers and potentially exacerbates an executive's ex ante incentive to commit fraud.
The rest of the article is organized as follows: Section II introduces the methodology and Section III describes the data.Sections IV-VI present the empirical results.Section VII concludes.Variable definitions are in the Appendix.In the Supplementary Material, we describe, respectively, test results considering endogeneity in political contributions, robustness analyses on penalty transfer and entrenchment of politically connected fraudulent executives, differential effects of PAC versus individual contributions, and another potential channel through which political contributions affect the severity of penalty.

II. Methodology
A. Measuring the Severity of Government Enforcement Following a "trigger event" and a formal investigation, the SEC files a complaint against named defendants.After an investigation, the SEC either drops the case or continues with administrative or civil litigation.The SEC may or may not refer the matter to the DOJ for criminal litigation proceedings.Once an executive is found liable for fraud, the government can impose civil penalties, criminal penalties, or both.
Figure 1 displays the timeline from the initiation of the fraud to the postpenalty period for our sample.An average fraud in our sample lasts for 3.31 years.The SEC enforcement period on average lasts about 3.44 years, followed by another 1.39 years to reach the case resolution (and resulting penalties).In later analysis, we investigate the impact on the firm during the postresolution period, which lasts up to 5 years following the regulatory period.
When an action is resolved, the SEC may impose a monetary penalty on the fraudulent executive, which consists of civil penalties and fines, disgorgement of

Timeline of SEC Enforcement Action
Figure 1, modified from Karpoff et al. (2008), illustrates the timeline of the SEC enforcement actions for our sample of fraudulent executives.BEGINNING_FRAUD is the fiscal year in which the executive first became engaged in fraudulent behavior.END_FRAUD is the last fiscal year of the alleged fraud.TIME_TO_ENFORCEMENT is the difference between the last fiscal year of the fraud and the formal filing of the SEC complaint (for civil actions).REGULATORY_PERIOD represents the time from the formal filing of the action by the SEC until the final resolution by the SEC.The resolution date typically refers to the resolution of the civil action; criminal actions may not be resolved until a later point in time.POST_RESOLUTION_PERIOD represents the time after the final resolution by the SEC.In our sample, the fraud period is 1990-2012, the regulatory period is 1996-2014, and the postresolution period is 1999-2019.

Penalty Shift
(CARs) illicit profits, and sanctions.Our first proxy for civil penalties is thus MONETARY_PENALTY, calculated as the natural log of the total monetary penalty imposed on the executive scaled by the executive's total annual compensation, which consists of salary, bonus, restricted stock, and options package (TDC1 in ExecuComp).By construction, this variable takes into account that executive compensation affects the magnitude of fraud (Peng and Röell (2014)), and consequently, the severity of sanctions.It also captures the degree of monetary damage relative to a fraudulent executive's personal wealth.For instance, a $100,000 penalty (our sample median) is likely felt less by a CEO making $6 million per year than a Vice President making $250,000 per year.In cases where an individual executive's compensation information is missing, we scale monetary penalty by the industry-year-executive role average compensation, as it is a widespread practice to benchmark executive pay by their industry peers (Faulkender and Yang (2013)).
The SEC can also ban a fraudulent executive from serving as an officer or director at a public company.Being banned as an officer prevents an individual from serving in a top managerial role, representing a significant economic penalty in the form of reduced lifetime earnings.The longer the ban period, the greater the potential compensation loss accrued to an individual.
We use OFFICER_BAN to capture the severity of such a ban, calculated as the natural logarithm of 1 plus the number of years the fraudulent executive is banned as an officer or director of a publicly listed company.This variable is set to 0 if no officer ban is imposed.In cases where an executive receives a permanent officer ban, we determine the effective length of the ban by estimating the number of years the executive could have served as an officer, had the ban not been imposed.Specifically, the length of the effective ban is computed as the difference between the retirement age and the executive's age at the time the permanent ban was imposed.In our main analysis, we use 75 as the cut-off for retirement age.Some, but not all, firms may have mandatory retirement ages ranging from 72 to 75 for directors.Alternative retirement ages of 72 and 78 do not alter our findings.
In instances where the fraud is determined to be criminal, the action may be referred to and prosecuted by the DOJ.We construct a dummy variable, DUM(CRIMINAL_INVESTIGATION), to capture the likelihood that the DOJ launches a criminal investigation.When such an investigation indeed occurs, the possible penalties that could be assessed are probation or a prison sentence.We thus compute PROBATION (PRISON) as the natural logarithm of 1 plus the number of years of probation (prison time) the executive received.These variables are set to 0 when the DOJ does not assess a penalty.2Note that an individual may receive probation instead of prison or receive a prison sentence only.In some cases, an accused executive receives both probation and prison terms.For this reason, we examine probation and prison separately.
Lastly, we rank the overall extent of severity across all types of sanctions, which also allows us to consider probation as a potential substitute penalty for prison.Specifically, SEVERITY is a categorical variable equal to 5 if an accused executive receives both officer ban and prison term, 4 if he receives prison term, 3 if there is an officer ban, 2 if there is probation, 1 if there is monetary penalty, and 0 if no penalties are imposed.

B. Measuring Political Contributions
We aim to assess the overall impact of an executive's political connection on government enforcement.In reality, an executive can channel contributions to a politician via his firm's PAC(s) as well as contribute directly as an individual.In light of Levitt and Dwyer (2003) and Babenko, Fedaseyeu, and Zhang (2020), our variable of interest, PC, is calculated as the natural logarithm of 1 plus the average annual contribution (in $10,000 s) made by an accused executive and his firm (via PAC) during the fraudulent period.This variable is set to 0 for those that do not contribute.
By aggregating both contribution channels, this variable captures the cumulative effect of corporate executives' spending on building and maintaining political connections.This is because, while personal contributions may suggest a more aggressive and clear preference of the contributor, it is often limited by a lower cap compared to a firm's PACs, thus may understate the contribution effort. 3By contrast, PACs not only allow an executive to shift the cost of spending on political contributions to shareholders, but also, more importantly for our research question, offer anonymity for the contributors.Consistent with this notion, Richter and Werner (2017) find that when political candidates refuse to take PAC contributions, CEOs contribute individually, implying that both channels are used and that executives prefer to contribute through their PAC. 4 Babenko et al. (2020) provide causal evidence that employees contribute significantly more money to political candidates supported by the CEO, suggesting that PAC contributions are likely an extension of contributions from top management of the firm.In the Supplementary Material, we reestimate our regressions, separating the two sources of contributions for fraudulent executives.We show that our baseline findings are mostly driven by PAC contributions.

C. Measuring Post-Fraud Firm Performance
We assess the consequence of the entrenchment of fraudulent managers using several firm performance proxies measured over various horizons during the postresolution period as illustrated in Figure 1.To capture the postresolution stock price 3 Corporations are prohibited from making direct political contributions in federal elections but may contribute indirectly either through contributions to PACs or contributions of its officers.PAC contributions are limited in two ways: individuals cannot contribute more than $5,000 to a single PAC in any given year, and PACs may not contribute more than $5,000 to a given candidate in a given election cycle.The Bipartisan Campaign Reform Act of 2002 (BCRA) limited "soft money" contributions to federal parties and indexed contribution limits for inflation, increasing it from $1,000 to $2,000.The limit for individual contributions per candidate was further increased to $2,500 for the 2011-2012 election cycle, while the PAC limit has remained at $5,000.For the purpose of campaign contribution limits, the primary election and general election are considered separate election cycles. 4Specifically, the increase in executive's contributions to a candidate approximately equals the average contribution from the average PAC to the average candidate, "suggesting that executives are in fact using their personal giving to substitute for their firm's linked PACs' inability to contribute to certain candidates" (Richter and Werner (2017)).
Alternatively, we consider a firm's profitability (measured by its ROA), Tobin's Q, and likelihood of financial distress (measured by ALTMAN_Z_ SCORE).We calculate the 1-year, 3-year average, and 5-year average of these proxies following the case resolution.

A. Government Enforcement
To identify government sanctions imposed on fraudulent executives, we start with the SEC's Accounting and Auditing Enforcement Releases (AAERs). 5To ensure that our data set is as accurate and complete as possible, we supplement this initial source of data with hand-collected data from various sources including SEC Litigation Releases, 10-k filings, the websites of the Offices of the US Attorneys, the Corporate Fraud Task Force Report to the President, and Lexis-Nexis news searches in order to avoid any missing fraud categories, duplicates, or omissions.
We manually screen over 2,400 AAERs for civil actions filed against executives of public companies during the period of Jan. 1999 to Dec. 2013.Executives in our sample comprise the following roles: the Chief Executive Officer (CEO), Chief Financial Officer (CFO), Chief Operations Officer (COO), Chief Information Officer (CIO), Chief Accounting Officer (CAO), Director, Controller, Treasurer, President, Vice President, General Counsel, or Chairman of the Board.Our initial manual screening yields 588 accused executives from 302 firms.We exclude 66 executives where either the SEC enforcement actions had not been resolved by Dec. 31, 2014, 6 or they had missing firm-level information in Compustat.Our final sample includes 522 executives from 225 firms.Of these executives, CFOs are the most prevalent at 173 observations.CEOs and Vice Presidents are second at 96 observations, and controllers are third, at 69 observations.For these 522 executives, we manually collect information from the AAERs on fraud duration, the amount of damages assessed by the government agencies, the dates the action was filed and finalized, how the case was resolved, and fraud type.Note that our sample includes executives who were investigated by the SEC (and possibly also the DOJ) for whom there were no resulting penalties imposed.The AAER period from 1996 to 2014 corresponds to the date in which the first 5 AAERs are the copies of court documents or summaries of court rulings on actions instituted against a firm and/or individuals by the SEC, or notices of settlement or court rulings.According to Karpoff, Koester, Lee, and Martin (2017), the AAER database performs the best of four commonly used fraud databases in terms of scope as well as the seriousness of fraud and performs second best in terms of fraud type omissions within a fraud event.The only feature for which AAERs do not perform well is initial revelation dates, which we do not use in our analysis. 6Criminal charges (which sometimes occur after civil charges) and penalties in our sample are resolved as of Dec. 31, 2015.investigation commenced and the year in which the last (complete) case was resolved.Since it often takes years to detect fraud and additional time to reach a final resolution once the fraud has been detected, the period in which executives commit fraud ranges from 1990 to 2012.
Information on civil penalties is from the AAERs and cross-checked by the SEC Litigation Releases.When information on outcomes and penalties is missing or incomplete from these two sources, we supplement with an exhaustive search on10-k filings and Lexis-Nexis news database.
Obtaining data on criminal penalties is less straightforward because the DOJ is comprised of 94 individual districts ("Offices of the United States Attorneys"), each with its own caseload.Unlike the AAER database, there is no central database that lists all actions taken by the DOJ.Also, unlike the AAERs, for which press releases are issued whenever the SEC files an action against a firm or an individual, only the most significant financial fraud cases are reported on the individual district websites.
To avoid the selection bias in which only the most publicized cases are reported, we proceed as follows: We start with our sample of executives accused by the SEC and determine whether the DOJ also pursued criminal charges against these executives.Since the SEC only has the authority to bring civil actions, the AAERs focus on civil penalties.Nevertheless, the AAERs often contain information relating to pending criminal litigation or criminal sanctions against the executive.We then cross-reference our list of executives with the individual websites of the 94 Offices of the United States Attorneys to determine criminal outcomes and penalties.We supplement information on criminal investigations and penalties using a variety of sources, including 10-k filings and the Corporate Fraud Task Force Report to the President.We also perform a Lexis-Nexis news search of all executives in our sample to minimize the possibility of missing information on fraud scope, type, coverage, and outcome.Doing so ensures that we assemble a comprehensive database of both civil and criminal outcomes for these executives.Overall, we are able to identify 181 executives in our sample that were also criminally charged by the DOJ.

B. Political Contributions and Other Data Sources
PAC contribution data from 1990 to 2004 comes from Cooper et al. (2010).We manually collect data on PAC contributions between 2005 and 2012 from the detailed committee and candidate summary contribution files of the Federal Election Commission (FEC).We manually collect individual executive contributions for the entire sample term from the Center for Responsive Politics.
The total compensation of the executive is TDC1 from ExecuComp (i.e., the sum of the executive's salary, bonus, other annual, total value of restricted stock granted, total value of stock options granted, long-term incentive payouts, and all other total), measured at the beginning of the fraud period.For executives from non-S&P 1500 firms, whose compensation is not available in ExecuComp, we manually search the proxy statement (DEF 14A) in SEC EDGAR using ExecuComp's definition of TDC1.Many of the executives in our sample are not in the 5 highest compensated executives; thus, their compensation is not required to be disclosed.

Fulmer, Knill, and Yu 2791
As a result, we are able to collect total compensation for 194 of our 522 executives.Existing literature has established that executive pays are often benchmarked by industry peers (e.g., Faulkender and Yang (2013)).For this reason, for those executives with missing information on compensation, we replace with industryyear-executive role average compensation.Doing so yields additional 217 observations, or 411 observations total.
Firm financial information is from Compustat.Information on independent directors is from BoardEx.Cumulative abnormal returns over various duration windows in the 1999-2016 postpenalty period are calculated using Eventus and CRSP.Data used to construct 1-, 3-, and 5-year ROA, Tobin's Q and Altman Z Score during the postresolution period of 2000-2019 comes from Compustat.Other data sources are described as we introduce them in the analysis.

C. Descriptive Statistics
Panel A of Table 1 reports the descriptive statistics for fraudulent executives.The annual contribution across candidates and election averages $15,200 per year during the fraudulent period.Among those who contribute, which accounts for 36% of our sample executives, the average annual political contribution is approximately $42,786.This is comparable to that reported in the existing studies (Ansolabehere, Snyder, and Tripathi (2002), Cooper et al. (2010)).
In terms of fraud characteristics, a fraud on average lasts for 3.31 years.There is also evidence about the collusion inside the executive suites: on average, 3.53 executives within the same firm as the fraudulent executive were also accused of fraud.The assessed total harm to the shareholders brought by the fraud event averages around $456.4 million per firm.Sixty-nine percent of the executives are ultimately terminated after fraud.
Panel A of Table 1 also reports the types of government enforcement outcomes that fraudulent executives receive during the sample period.The average accused executive is fined with $5.71 millionequivalent to 2.86 times his annual compensationand receives 7.83 years of officer ban.Roughly 33% of executives accused by the SEC are also subject to a criminal investigation by the DOJ.An average executive receives 0.35 year (127 days) of probation and 0.61 year (224 days) in prison.
Panel B of Table 1 shows a breakdown of the number of executives by fraud type.Note that an individual may be accused of more than one type of fraud.Earnings fraud is the most common type of fraud, with 88.5% of the 522 executives accused of manipulating earnings or revenue.The next most frequent type of fraud is securities fraud, accounting for 18% of the accused executives.It is also evident that a great majority of frauds occur within the executive suite.
Panel C of Table 1 presents the descriptive statistics of firm performance during the postpenalty period, up to 5 years after the case resolution.The valueweighted standardized CARs for our sample firms range from À4% over the [À1, þ30] day window to À3% over the [À1, þ360] day window following the resolution of the case.The ROA per firm on average is 1.04% in the first year after the resolution, À0.03% over the 3 years after fraud resolution, and 1.63% over the 5-year postresolution period.Tobin's Q averages 1.68, 1.78, and 1.91 over the 1-year, 3-year, and 5-year periods after case resolution, respectively.

IV. Political Contributions and the Severity of Penalties A. Baseline Results
To explore the link between political contributions and the severity of government sanctions imposed on fraudulent executives, we estimate the following empirical model: The dependent variable is one of the penalty measures described in Section II.A.The key explanatory variable, PC, is the natural logarithm of 1 plus the annual average amount of political contributions during the fraud period.We control for a vector of time-varying fraud and firm characteristics, Z, which may affect the extent of enforcement.For instance, the size and severity of the fraud are positively linked to legal and monetary penalties (Karpoff, Lee, and Martin (2007)).We measure the severity of fraud by DAMAGES, computed as the natural logarithm of the amount of damages assessed by the SEC, which is the total loss to the firm (or shareholders) as a result of the fraud, independent of whether the fraud was determined to be criminal or civil. 7An executive can be accused of more than one type of fraud.In this case, the amount is the cumulative effect of the fraud(s).Hence, it is not necessarily a summation of the damages from each fraud type, but rather the net effect of the multiple fraud types (allowing for overlapping damages).
Damages can be more significant when a fraud lasts for a longer period of time (Yu and Yu (2011)), resulting in more severe penalties.We thus control for the duration of the fraud (FRAUD_DURATION), calculated as the natural log of the number of years (or portion thereof) from the initiation of the fraud to its conclusion.
The severity of the penalties that an accused executive receives may be affected by the nature of the fraud itself.A fraud taking more time to investigate tends to be more complex and severe.The SEC might have more authority to impose sanctions for some types of fraud than others, or certain natures of fraud are more destructive than others.For these reasons, in all regressions, we include TIME_TO_FILING, calculated as the number of years between the end of the fraud and the filing by the SEC, as well as fraud-type fixed effects (δ F ).
Executives may work for companies that are well poised for using their market power or having "deep pockets" to fend off charges, pay for civil penalties, hire superior legal counsel, and provide directors and officers (D&O) insurance.We, therefore, include proxies for firms' deep pockets, financial positions, and market power, such as firm size, leverage, and market share.Firm size is calculated as the natural logarithm of the firm's average market cap during the fraudulent period.Following Kedia and Rajgopal (2011), we also include a dummy for small firm to capture the potential nonlinear effect of firm size in affecting the extent of government investigation and enforcement; a firm is considered small if its market value at the beginning of the fraudulent period is less than $200 million.Leverage is calculated as the natural logarithm of 1 plus the sum of current liabilities and total liabilities scaled by total assets.A firm's market share is its sales scaled by the average sales of firms operating in the same 1-digit SIC industry, thus capturing the relative market power over its industry peers.
The age and role of the executive may also influence penalties.Penalties tend to be more lenient for older fraudsters.CEOs or CFOs could be held more responsible than those in other roles.Misconduct is more likely and more complex in groups than in individuals (Kocher, Schudy, and Spantig ( 2018)); therefore, the collusion among executives may affect to what extent the SEC and DOJ assess and impose penalties on individual officers.We thus control for the age of the fraudulent executive, the number of accused managers in the fraud, and executive role fixed effects (δ E ).Lastly, we include industry fixed effects (δ I ) and settlement year-fixed effects (δ Y ) to control for, respectively, unobserved industry-specific characteristics and unobserved time trend at the time the fraud investigation was finalized and penalties imposed.
Table 2 reports the baseline regression results.The unit of analysis is at the executive level. 8Since the dependent variables contain a nontrivial fraction of zero values, we estimate Tobit regressions (Wooldridge (2012)) and tabulate the marginal effects.
Columns 1 and 2 reveal that political contributions are negatively related to the magnitude of civil penalties for fraudulent executives.The effects are both statistically and economically significant.A $10,000 increase in political contributions is associated with a reduction in monetary penalty that amounts to 1.88% of the executive's annual compensation (column 1) and a 0.144-year reduction (or 53 fewer days) in officer ban (column 2).Given that fraudulent managers in our sample earn on average $3,181,539 annually, these numbers translate into roughly, $60,000 and $458,000 reductions in lost wages, respectively.
In columns 3-5, we explore the effect of political contributions on criminal sanctions imposed by the DOJ.Criminal outcomes tend to be particularly impactful to executives, that is, loss of freedom through prison or probation sentences and impacts on future job prospects.As such, an executive may be eager to avoid a criminal investigation, especially given that he has already been subjected to a civil investigation for fraud by the SEC.Table 2 reveals that political contributions are negatively and significantly related to the propensity of the executive being investigated by the DOJ (column 3) and among those who are criminally prosecuted, the number of years of probation/prison they receive (columns 4-5).In terms of economic magnitude, a $10,000 increase in political contributions is associated with a 2.67% lower probability that a fraudulent manager faces a criminal investigation, a reduction of 0.265 years in probation and 0.243 years in prison, accounting for 75.6% and 39.8% of the sample means, respectively.
The results in columns 1-5 suggest that political contributions are associated with a reduction in various types of government penalties.To further probe this intuition, we consider the overall severity of these government sanctions.In column 6, we continue to observe a negative relation between political contributions and the extent of severity of penalties.
We obtain these estimates with controls for firm-specific, fraud-specific, and executive-specific characteristics, as well as industry-, fraud type-, executive role-, and settlement year-fixed effects.This allows us to control for nonparametric industry-and time-specific shocks, as well as a variety of alternative explanations related to fraud-specific and executive-specific characteristics.Overall, the results in Table 2 support the view that fraudulent executives benefit from political contributions.Not only do they pay reduced fines and bear shorter officer bans compared with fraudulent executives that do not contribute, but they are also less likely to be investigated by the DOJ and spend less time on probation and prison.
To further understand the nature of the penalties imposed on fraudulent executives, in what follows, we investigate how the effect of political contributions  varies along with the nature of misconduct, the role of fraudulent executives, and the power of politicians.

B. For What Type of Fraud Can Political Contributions Help the Most?
By including fraud-type fixed effects, our baseline analysis absorbs unobserved heterogeneity in the nature of the fraud on influencing the severity of penalties imposed on fraudulent executives.Nevertheless, the extent to which the firm (via a PAC) and executives engage in political contributions may vary depending on the nature of the misconduct, and subsequently, the severity of the imposed penalty.For instance, the political contribution literature has shown that maintaining political connections helps firms to secure government contracts and build sales (e.g., Goldman et al. (2013), Akey (2015)).It is possible that contributions are more common among revenue-dependent firms, which are more incentivized to inflate sales artificially or boost revenue via bribery.As such, misconduct such as earnings fraud may get penalized less than other types of misconduct that are beneficial solely to the individuals committing the fraud, such as insider trading and option backdating.
We distinguish between misconduct that is more likely motivated by the firm (i.e., earnings fraud, securities fraud, or bribery) and misconduct where the direct gains almost exclusively benefit individual executives (i.e., insider trading, options backdating, or embezzlement).Accordingly, we construct PERSONAL_ BENEFIT, a dummy variable set to 1 if the fraud type is insider trading, options backdating, or embezzlement, and 0 otherwise.
Panel A of Table 3 reports the distribution of contributions across fraud types.The unit of analysis is the executive-fraud type observations.Column 1 reports the number of observations.Column 2 reports the number of executives accused of a given misconduct type that contribute politically.Columns 3-5 report, respectively, total dollar amount of political contributions, accused executive's contributions, and PAC contributions within each misconduct type.There appears to be significant heterogeneity in contributions among various fraud types.Not only does misconduct that benefits the firm beyond the accused executives account for a great majority of the cases, corroborating with evidence in Panel B of Table 1, but also, it attracts more political contributions.For instance, 167 cases of earnings fraud and 24 cases of securities fraud involve political contributions, accounting for 36% and 26% of such cases, respectively.These two types of fraud also attract the largest political contributions, totaling $7.56 million and $0.994 million, respectively.Separating contributions into individual and PAC contributions, we observe that both accused executives and their firms direct more of their political spending to these types of misconduct.On the other hand, the total dollar amount of political spending on misconduct with benefits limited to the fraudulent executives, such as insider trading, option backdating, and embezzlement, tends to be much smaller.
Since fraud type drives the intensity of political contributions, we conjecture that the extent to which political contributions affect penalties on executives also hinges on the nature of the misconduct.As such, we augment our baseline regressions in Table 2 by including the interaction between our variable for political contributions and PERSONAL_BENEFIT.Corroborating the results in Panel A of Fulmer, Knill, and Yu 2797 Table 3, Panel B reveals a positive and significant coefficient associated with the interaction term, suggesting that political contributions are less effective at reducing penalties for types of misconduct from which managers are most likely to benefit personally.
Overall, the results in Table 3 suggest that there exists heterogeneity in contributions across different types of misconduct, and that fraud type affects the extent to which political contributions influence government sanctions.

C. Who Benefits More from Political Contributions?
By including the executive role fixed effect, our baseline regression analysis also absorbs the heterogeneity in the roles of fraudulent executives.Individuals serving different roles assume different power and can harvest the benefits of political contributions to different extents.For instance, compared to a general counsel, a CFO is likely more responsible for earnings fraud.At the same time, he has deeper pockets and more authority to direct both his firm's political spending and resource allocation, thus benefiting more from the impact of political contributions in terms of reduced penalties.

TABLE 3
The Nature of Fraud and Political Contributions Panel A of Table 3 reports the distribution of political contributions along fraud types.The unit of analysis is the executive-fraud type observations.Panel B examines the effect of political contributions on the severity of government penalties.The dependent variables are indicated at the top of each column.PERSONAL_BENEFIT is a dummy variable set to 1 if the fraud belongs to insider trading, option backdating, or embezzlement, and 0 otherwise.Control variables (identical to those in We postulate that executives who are in more powerful positions, such as those in the executive suite, can appropriate the benefits of political contributions to a greater extent than those that are in lower ranks.In Table 4, we augment the baseline regressions in Table 2 by including the interaction between our variable for political contributions and TOP_5, a dummy variable set to 1 if the fraudulent executive is a CEO, CFO, Chairman of the Board, President, or Vice President, and 0 otherwise. Column 1 shows that the interaction is negative and significantly associated with monetary penalty.Occupying a top executive position increases the extent to which political contributions lessen penalties for misconduct.Interestingly, compared to those that do not hold a top executive position, top 5 executives not only have greater discretion in determining campaign contributions, but are also more likely responsible for the misconduct.To validate this intuition, in column 2, we further include an interaction with WB, a dummy variable for a weak corporate board, captured by (ex ante) lower board independence (Adams and Ferreira (2009)).A weak board consisting of fewer independent directors is especially ineffective in monitoring managers, curbing executive greed, and reining in their power (Haynes, Campbell, and Hitt (2017)).WB is thus set to 1 if the firm's board independence, measured at the beginning of the fraudulent period, falls below the bottom sample decile.
Column 2 reveals that top executives enjoy the benefit of political contributions on reduced monetary penalty to a greater extent than lower ranked executives if their firms' boards are inherently dysfunctional.Column 4 shows similar findings for officer ban: while political contributions are associated with fewer years in officer ban, connected top 5 executives in firms with weak boards can receive a less severe officer ban than lower ranked executives.
In terms of criminal penalties, there appears to be no statistically significant difference in the propensity of being investigated by the DOJ or in the number of years of probation between the top 5 and nontop-5 executives (columns 5-8).However, when considering the most severe penalty that the government agencies can impose upon an individual for misconductprison termsthe estimate associated with the interaction term PC Â TOP_5 Â WB is negative and statistically significant at the 1% level (column 10).With the inclusion of fraud type fixed effect, this suggests that increasing political spending allows top 5 executives to serve fewer years in prison than lower rank officers (column 9) for the same type of misconduct, especially if their firms have weak internal governance (column 10).
Lastly, when considering the overall extent of severity across all types of penalties, column 12 provides further support that government sanctions imposed upon the most senior executives in firms with ineffective boards are more lenient than those on lower-ranked ones.Compared to their lower-ranked colleagues, top-ranked executives are generally paid more.They are also more likely to be responsible for the misconduct and are more capable of influencing campaign contributions.The results in Table 4 indicate that these executives are able to appropriate reduced sanctions from political contributions to a greater extent, especially when their boards are in a weaker position to discipline these managers.These findings help to explain how the role of executives amplifies or mitigates the effect of political contributions on government sanctions.

D. With Which Politicians Can Contributions Be More Effective?
Our proxy for political contributions seeks to measure the overall political spending of the manager.As such, we count contributions to all candidates, including those that remain in office, those that ultimately lose their election, or may otherwise not be in office at the time when the penalties are being assessed.One may argue that the effect on penalties should primarily come from backing the politicians who are actually in a position to influence the government agencies imposing the penalties.This implies that contributions made during the fraudulent period should matter more if they are made to individuals who are ultimately in office at the time the penalties are meted out.Put differently, our main proxy captures the lower bounds of the impact from political contributions.
In this subsection, we explore heterogeneity among politicians' ability to exert influence.We first consider a politician to be more powerful if he or she stays in the office at the time when the penalties are determined.To identify contributions made to politicians that remained in office versus those who left office, we manually check every personal contribution for each executive and every contribution from his firm's PAC(s).We then calculate PC_(IN_OFFICE) and PC_(OUT_OF_ OFFICE), which are, respectively, the natural logarithm of 1 plus the dollar amount of contributions made during the fraud period to those politicians who stay in political office, and to those who have lost their race or otherwise left the office at the time when penalties are decided.

Executives and Political Contributions
Table 4 relates political contributions to the severity of government penalties.The dependent variables are monetary penalty, officer ban, the probability of criminal investigation, probation, prison, and severity, respectively in columns 1-2, 3-4, 5-6, 7-8, 9-10, and 11-12, respectively.TOP_5 is a dummy variable set to 1 if the executive is a CEO, CFO, Chairman, President, or Vice President, and 0 otherwise.WB is an indicator variable for weak corporate board, set to 1 if the percent of outsiders on the board is less than or equal to the bottom 10th percentile.Control variables (identical to those in Columns 1-6 in Panel A of Table 5 show that contributions to candidates who remain in political office at the time of penalty assessment are negatively and significantly related to monetary penalty, officer ban, propensity of criminal investigation, probation, prison, and overall severity of sanctions.In columns 7-12, we control for, additionally, contributions made to candidates who lost their election or have left the office by that time.We continue to observe a negative and significant coefficient for PC_(IN_OFFICE) across all six penalty outcomes.This suggests that our baseline results are largely driven by backing candidates who later on remain in office and thus can potentially exert influence on government agencies. 9 The results in Panel A of Table 5 indicate that the effect of political contributions on the severity of government penalties varies with salient traits of politicians; namely, their capacity to exert influence largely depends on their ability to remain in political office.By the same token, those long-time veteran politicians, who oftentimes occupy higher ranks within the party and have built and maintained extensive networks, are more effective in facilitating direct or indirect impacts on polices and government agencies on behalf of their backers.As such, they are in a better position to influence government sanctions.
Following Stewart (2017), we also classify a politician to be more powerful if he or she ranks as the top 10% of politicians in each chamber (5% within each party).Specifically, this ranking process yields 10 Senators (i.e., 5 Democrats and 5 Republicans) and 44 Representatives (i.e., 22 Democrats and 22 Republicans).To identify contributions made to powerful and nonpowerful politicians, we manually check every personal contribution for each of the contributing executives and each contribution from their firm's PAC(s).Accordingly, PC_(TOP_RANKED) and PC_(LOWER_RANKED) are, respectively, the natural logarithms of 1 plus the political contributions made during the fraud period to the powerful and nonpowerful politicians.
Panel B of Table 5 reveals that contributions to powerful politicians are negatively and significantly related to monetary penalty, officer ban, propensity of criminal investigation, probation, prison, and overall severity (columns 1-6).The magnitude of these coefficients is mostly larger than that of the baseline results, which do not distinguish between contributions to powerful and nonpowerful ones.The largest increase happens for criminal sanctions, where an additional $10,000 of contributions is associated with a reduction of 206 days of probation and 106 days of prison time (vs.97 and 89 days, respectively, for the baseline).Inclusion of contributions to less powerful politicians does not alter the main findings (columns 9 By contrast, in most cases, contributions to candidates who are no longer in political office do not lead to lower penalties.Nevertheless, compared to those for PC_(IN_OFFICE), the estimators associated with PC_(OUT_OF_OFFICE) can be noisier and should be interpreted with caution for two reasons.First, contributions to politicians who lose the elections average $532 per year, considerably smaller than the $13,132 that politicians who win the elections and stay in the offices receive annually.Second, unlike those who remain in office, it is difficult to determine the exact timing of the departure relative to case resolution and the termination of the influence for those who lose the elections.For instance, politicians who are defeated in an election may remain in the office until the opening of the next congress, during the period in which they can still exert influence.They may rerun and win during the next election cycle because of their existing networks in the political arena, which potentially still allow them to attract contributions and exert some influence.Finally, it is possible that a case is actually resolved before the politician officially leaves office.

Powerful Politicians
Table 5 relates political contributions to the severity of government penalties.The dependent variables are indicated at the top of each column.In Panel A, PC_(IN_OFFICE) and PC_(OUT_OF_OFFICE) are, respectively, natural logarithm of 1 plus the political contributions to those who stay in political office, excluding contributions to those who lost the race or left the office.In Panel B, PC_(TOP_RANKED) and PC_ (LOWER_RANKED) are, respectively, natural logarithm of 1 plus the political contributions to the powerful and nonpowerful politicians.A politician is considered powerful if he or she belongs to the top 5% of each party/10% for each chamber.Control variables (identical to those in 7-12).10While we do not rule out the possibility that lower-ranked politicians may be well connected to the powerful ones, or that the number of connected politicians is a driving force, Panel B provides evidence suggesting that our baseline results are largely driven by contributions made to highly ranked politicians, who are more capable of directly and indirectly influencing government agencies' ability to penalize fraudulent executives.

E. Alternative Economic Causes and Additional Robustness Tests
The decision to make a political contribution is likely endogenous.As such, reduced government penalties may be driven by omitted variables or serve as a manifest of other effects caused by political contributions.In this subsection, we first explore two possible alternative hypotheses that may otherwise explain the disparate penalties between fraudulent executives that make generous contributions and those that do not.We then summarize additional robustness tests for our baseline results.
A large literature has documented the impact of political connections on firm value (e.g., Fisman (2001)).Thus, the SEC and DOJ may impose a less severe sanction on an executive from a firm whose net benefit to shareholders is believed to be positive.Put differently, if an executive being accused of fraud has otherwise managed the firm well and has, on balance, increased shareholder wealth through his contribution to firm activities, the SEC/DOJ may take the net effect of his work into consideration and propose a less severe penalty since shareholders are better off with the firm in place.Another possibility is that if a firm (or by extension, its top management) has put its best efforts forth to comply with SEC regulations, this compliance may earn that firm, as well as its executives, leniency for any wrongdoing.In fact, Files (2012) finds evidence of earned leniency in the form of reduced monetary penalties.
To consider the potential net benefit to shareholders, we include in our baseline regressions a proxy for the benefit accrued to the shareholders by accused executives: the firm's profitability (ROA), calculated as income before extraordinary items scaled by total assets measured at the beginning of the fraudulent period. 11o take into account the potential leniency that the firm may have earned by being a "good corporate citizen," we also include a proxy for internal corporate governance (BOARD_INDEPENDENCE), calculated as the fraction of independent directors measured at the beginning of the fraudulent period.If political contributions do not directly affect the severity of government enforcement, but instead, simply proxy for the potential benefit to shareholders and/or for earned leniency, we would expect that adding these two controls will cause our variable of interest to lose statistical significance.
Table 6 shows that the baseline results are robust when we control for the potential effect from these two alternative causes.While the government agencies appear to consider the gains accrued to shareholders and internal corporate governance environment when assessing certain sanctions, political contributions continue to be significantly and negatively related to the severity of penalty.This suggests that the "benefit exceeds the harm" and "earned leniency" hypotheses do not fully explain the disparate penalties between contributing executives and noncontributing executives.
To further alleviate the concern for causality, in the Supplementary Material, we perform a difference-in-differences analysis exploiting the exogeneity of the impact on political contributions brought about by the passage of the Bipartisan Campaign Reform Act (BCRA) (Bradley, Pantzalis, andYuan (2016), andAyyagari, Knill, andSyvrud (2019)).Specifically, the enactment of BCRA enhances both the power and intensity of PAC and personal contributions, but it is not designed to cater to the severity of government penalties imposed upon fraudulent executives.Alternatively, we apply our estimation to an entropy-balanced matching sample (Hainmueller (2012)).These test results offer further evidence that our baseline findings are unlikely tainted by endogeneity.In the Supplementary Material, we also show that our baseline results are invariant to alternative ways of computing political contributions and classifying industries, as well as to the inclusion of industry Â year-fixed effects.

V. Economic Channels
How do political contributions generate reduced penalties for fraudulent executives?Identifying such a channel is empirically challenging because interactions between politicians and executives are nearly unobservable, and could take the form of phone calls, e-mails, or even conversations at social/professional events.

Alternative Hypotheses
Table 6 reports the results examining the effect of political contributions on the severity of government penalties with additional controls for alternative causes: the operating performance benefit to the firm's shareholders (measured by ROA) and corporate governance (measured by board independence) in comparison to the damage caused by the fraud.The dependent variables are indicated at the top of each column.Control variables (identical to those in Table 2) and fixed effects (described in the table) are included in estimations, but coefficients are not tabulated.Industry is a firm's 1-digit SIC code.Variable definitions are in the Appendix.Robust standard errors clustered at the firm level are reported in square brackets.***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.Indeed, finding a reliable, concrete link between political contributions and favors promised by members of Congress is a difficult task.In support of this logic, a Wall Street Journal article notes that when executives and politicians do not want a record of communication, they often use the code LDL, which stands for "let us discuss live."12Nevertheless, in this section, we propose two potential mechanisms through which political contributions may allow fraudulent executives to receive more lenient sanctions.

A. Resource Constraints
Recent studies highlight that budgetary and staffing resources are vital to the oversight intensity and enforcement potency of government agencies (e.g., Jackson and Roe (2009), Blackhurne (2014), Christensen, Hai, and Leuz (2016), and Del Guercio, Odders-White, and Ready ( 2017)).Not only can politicians directly interfere with SEC investigations, but also, perhaps even more preferably, they can do so indirectly by using budget allocation or affecting SEC officials' careers (Weingast (1984)).In the context of our analysis, contributions to politicians may result in tighter budgetary and staffing resources of these agencies, distorting the efficiency of their investigations and increasing the cost for them to assess and impose harsh penalties on fraudulent executives.
To probe this intuition, we manually collect budgetary and staffing information from the SEC and DOJ websites.The budget and staffing data are available for most measures from 2002 onward for the SEC and 1999 onward for the DOJ.To measure the caseload with respect to staffing constraints, we construct CASES/ EMPLOYEE, calculated as, respectively, the natural logarithm of annual number of SEC (DOJ) cases scaled by the numbers of SEC (DOJ) employees.The SEC website also provides a breakdown of its labor force, allowing us to extract information on the staff size in its enforcement department.Therefore, we also construct CASES/ENFORCEMENT_EMPLOYEE, which is the natural logarithm of annual number of SEC cases divided by the number of employees in its enforcement division.To measure the caseload with respect to budgetary constraints, we compute CASES/BUDGET, which is, respectively, the natural logarithm of the number of SEC (DOJ) cases divided by the dollar amount (in $1,000 s) of SEC (DOJ) annual budget.A higher value of these variables indicates a bigger caseload, thus a scarcer resource, of the SEC and DOJ.
Table 7 first validates that a more intensive caseload, thus a greater staffing and budgetary resource constraint at the SEC (columns 1-3) and the DOJ (columns 7-8), is associated with more lenient government sanctions.Next, we explore whether the resource constraint at these agencies is the underlying factor affecting the difference in the severity of penalties between fraudulent executives that make generous contributions and those that do not.Following Core, Holthausen, andLarcker (1999), Francis, Olsson, andSchipper (2005), and Yu and Yu (2011), we decompose each of our caseload variables into the predicted and residual components by regressing these variables individually on PC.By construction, the predicted components of these variables (i.e., PREDICTED_CASES/EMPLOYEE, PREDICTED_CASES/ENFORCEMENT_EMPLOYEE, and PREDICTED_ CASES/BUDGET) capture the extent of increase in resource constraint at government agencies brought about by political contributions from accused executives.
Columns 4-6 and 9-10 of Table 7 show that the predicted caseload is negatively and significantly related to the severity of penalty.The results are consistent with the argument that fraudulent managers receive more lenient penalties for their misconduct when their political contributions lead to an increase in resource constraints at the government agencies.

B. Case's Time to Resolution
Katz (2010) points out that enforcement actions inside the SEC that take longer to resolve are often closed with no resulting penalties, particularly when a new attorney is assigned to the case or when a new commissioner is appointed.One potential reason for this could be budgetary.The longer a case takes, the more resources it consumes.This implies that if an executive can delay resolution of the case long enough, he could potentially also reduce the severity of the penalty imposed by the SEC.

Resource Constraints
Table 7 presents the ordered probit regression results examining how the effect of political contributions on the severity of government enforcement varies with a government agency's resource constraints.Columns 1-6 refer to SEC cases, and columns 7-10 refer to DOJ cases.The dependent variable is SEVERITY, a variable equal to 5 if both prison and officer ban are imposed, 4 for prison term, 3 for an officer ban, 2 for probation, 1 if there is monetary penalty, and 0 if no penalty is imposed.CASES/EMPLOYEE is the natural logarithm of SEC (DOJ) cases per employee.CASES/ENFORCEMENT_EMPLOYEE is the natural logarithm of SEC cases per enforcement department employee.CASES/BUDGET is the natural logarithm of SEC (DOJ) cases per thousand dollars of SEC (DOJ) budget.PREDICTED_CASES/EMPLOYEE, PREDICTED_CASES/ENFORCEMENT_EMPLOYEE, and PREDICTED_CASES/BUDGET are the predicted value from regressing, respectively, CASES/EMPLOYEE, CASES/ENFORCEMENT_EMPLOYEE, and CASES/BUDGET on PC.Control variables (identical to those in Table 2) and fixed effects (described in the table) are included in estimations, but coefficients are not tabulated.Industry is a firm's 1-digit SIC code.To mitigate the concern that the nature of the misconduct may drive the time to resolve the casefor instance, less important matters are pushed off and take more time to settle, we measure CASE_AGE as the number of years from initiation of the formal charges for fraud until a new SEC commissioner assumes power.The value of this variable reflects how "old" the case is when the new commissioner begins his/her assignment.
We first provide graphical evidence validating Katz (2010).Figure 2 presents a stacked bar graph plotting the distribution of civil penalties of different extents of severity for quintile bins of case age, with 1 representing the shortest case age quintile and 5 the longest quintile. 13We observe a downward trend overall for the likelihood of a monetary penalty as case age rises.The likelihood of an officer ban (the most severe civil penalty) initially increases with case age, then declines for quintiles 4 and 5. Overall, Figure 2 provides suggestive evidence that cases taking distinctly longer to resolve to receive more lenient treatment.
We further corroborate the patterns in Figure 2 in a regression framework, regressing the severity of civil penalty on CASE_AGE and its squared term.The inclusion of the squared term of CASE_AGE takes into account that it affects the severity of penalty in a nonlinear way.We defined the severity of civil penalty as a categorical variable set to 2 if a fraudulent executive receives the harshest civil penaltyofficer ban, 1 if there is only monetary penalty, and 0 if no civil penalties are involved. 14he negative coefficient on the squared term in columns 1-2 of Table 8 suggests that fraud cases taking sufficiently longer time to resolve result in less severe sanctions for the executive when a new commissioner is appointed.This is consistent with the No Penalty Officer Ban Monetary Penalty notion described in Katz (2010), which is that the SEC cleans house when new commissioners arrive, leading to old cases resolved with lower penalties than they would perhaps otherwise receive if the new commissioner had not arrived.
Next, in a similar vein as the tests of resource constraints, we examine whether political contributions help lengthen case time-to-resolution, and by doing so allow fraudulent executives to harvest more lenient sanctions.We decompose CASE_ AGE into the predicted and residual components by regressing CASE_AGE on PC.By construction, the predicted component captures the extent of the delay in case resolution until the arrival of a new SEC commissioner that is driven by political contributions from accused executives.
Columns 3-4 of Table 8 reveal that the predicted case age is negatively and significantly related to the severity of civil penalty.We interpret this as evidence consistent with the argument that by lengthening the case's time to resolution, political contributions allow for less severe civil penalties imposed on fraudulent managers.

VI. Penalty Transfer and Executive Entrenchment
A. Penalty Transfer from Executives to Shareholders So far, we provide novel evidence that political contributions limit the adverse effect of government penalties imposed upon fraudulent executives.In light of prior evidence that political spending reduces monetary penalties Case Age at a New SEC Commissioner Table 8 presents the ordered probit regression results examining how political contributions affect the civil penalty imposed upon fraudulent executives via increasing time-to-resolution with the SEC.The dependent variable is CIVIL_SEVERITY, a categorical variable set to 2 for officer ban, 1 for monetary penalty, and 0 for no civil penalty.CASE_AGE is the number of years between the date of the case filed and the time when a new SEC commissioner assumes power.If the case is resolved prior to the appointment of a new commissioner, this variable is the difference between the date of case resolution and the time when the case is filed.PREDICTED_CASE_AGE is the predicted value from regressing CASE_AGE on PC.Control variables (identical to those in imposed on a fraudulent firm by the SEC (Correia (2014)), our findings imply that political spending enables regulatory capture not only at the firm level, but also at the individual level, and that regulatory capture can occur in more than one government agency.Importantly, our data allows us to assess whether political contributions affect the distribution of penalty between fraudulent executives and their firms.On one hand, SEC officials frequently view a large amount of (total) penalties assessed to be indicative of the great diligence, rigor, and success of the agency's enforcement efforts.As such, the civil penalty amounts are regularly trumpeted in press releases and highlighted in its end-of-the-year statistics (Rosenfeld (2019)).On the other hand, while the SEC has a strong incentive to obtain large penalties, the extent to which the agency can impose monetary fines on a natural person is more restrictive than that on corporations. 15This implies that in order to secure sufficiently large civil monetary penalties, the SEC favors assessing a large penalty on the firm, potentially leaving a smaller amount on the manager and by doing so, shifting penalties from an accused executive to the firm's shareholders.
Several regulators have already voiced their concerns on this distortion in the enforcement process.For instance, Katz (2010) writes: "Even when a public corporation pays a large penalty, its punitive value is diminished by the knowledge that the penalty is not paid by a culpable individual but by the shareholders of the company, who typically were the victims of the fraud."Chester Spatt, Chief Economist at the SEC in 2007, stated that "…to the extent that the firm's shareholders were themselves victimized by the fraud, it is not clear that additional sanctions should be imposed upon the firm to be borne by the current shareholders."Steinway (2014) likewise suggests that deterrents of corporate fraud are undermined if the employer of fraudulent executives (or by extension, its shareholders) pays a financial penalty.
In this subsection, we aim to understand whether political contributions facilitate the distortion in the enforcement process that benefits managers, even if that means shifting the consequences to the shareholders who have already lost value from the fraud.Specifically, we examine how the likelihood and amount of fines that the SEC imposes on firms vary with the severity of sanctions it applies to their executives, and to what extent political contributions affect this process.
We first consider the likelihood that a fraudulent executive receives a penalty if his or her firm is fined by the SEC.DUM(FIRM_FINE) is an indicator variable set to 1 if a firm is fined by the SEC separately due to the misconduct by its accused executives.We also consider the dollar amount of the fine imposed on the firm by the SEC, scaled by all fines, which consists of both fines paid by the firm and the amount of monetary penalty imposed on its fraudulent executives.We then construct the variable FIRM_FINE/ALL_FINES by taking the natural logarithm of 1 plus this number.This variable is set to 0 if the firm is not fined by the SEC.
In comparison to the dummy for firm fine, this variable is particularly relevant in our penalty transfer analysis, as it explicitly allows us to capture the extent of penalty borne by the firm due to its fraudulent managers' misconduct relative to penalties received by the fraudulent executive himself or herself.Alternatively, we construct FIRM_FINE/TOTAL_ASSETS by taking the natural logarithm of 1 plus firm fine-scaled by total assets.Similar to our mechanisms analyses, we first regress SEVERITY on PC to generate the predicted component, PREDICTED_SEVERITY, which captures the extent that political contributions reduce the severity of government penalty imposed upon an accused executive.We then relate it to the likelihood of a firm being fined by the SEC as well as the amount of firm fine.
Panel A of Table 9 shows that PREDICTED_SEVERITY is negatively and significantly associated with DUM(FIRM_FINE) (columns 1-2), FIRM_FINE/ ALL_FINES (columns 3-4), FIRM_FINE/TOTAL_ASSETS (columns 5-6).As the severity of the penalty imposed upon the fraudulent executives goes down due to their larger political contributions, both the likelihood that a firm is fined by the SEC and the amount of a firm's fine increases.The results prevail when we also include whether or not alternative market-based disciplining mechanisms have occurred, such as class action lawsuit and delisting, which may affect the likelihood and amount of the firm fine (columns 2, 4, and 6).Overall, our results suggest that political contributions facilitate the "transfer" of penalty between fraudulent managers and shareholders. 16 Karpoff et al. (2008) find that executives are at a higher risk for termination following a restatement due to fraud.In Panel B of Table 9, we explore whether fraudulent executives are able to avoid termination following the shift of penalties from them to the shareholders.In a similar vein, we use the predicted component from regressing FIRM_FINE/ALL_FINES on PC to capture the extent that political contributions influence the amount of government punishment a firm receives.In light of the findings in Panel A, where political contributions allow for a stronger substitution between penalties imposed on fraudulent executives and on their firm, a larger fitted value suggests a greater degree of penalty transfer.
We then regress the likelihood of executive termination on PREDICTED_ FIRM_FINE and report the results in Panel B of Table 9.In this set of tests, we control for, additionally, executive ownership, calculated as the percent of the firm's equity owned by the executive, which is highlighted in the extant literature as an important determinant of managerial turnover (e.g., Denis, Denis, and Sarin (1997)).In columns 1 and 2, we focus on the impact of the predicted firm fine on the likelihood of executive termination.In columns 3 and 4, we also include alternative market-based disciplining mechanisms such as class action lawsuit and delisting, which may contribute to the likelihood of an executive turnover.Panel B shows that when the firm receives larger fines as predicted using political contributions, its executives are less likely to face job termination.Put differently, as the fraudulent 16 In untabulated regressions, we find that the predicted severity is unrelated to the combined sum of firm fines and monetary penalties on accused executives.This suggests that our findings likely capture a shift in the penalty, rather than an increase in the total monetary penalties.
executive shifts the government penalty from himself onto the firm's shareholders, he decreases the likelihood of his own termination.
Overall, Table 9 provides evidence that less severe penalties imposed on fraudulent managers are associated with larger firm fines from the SEC that cost shareholders, and that political contributions exacerbate such a relationship.By distorting the enforcement process, political contributions appear to allow fraudulent executives

Penalty Transfer and Executive Entrenchment
Panel A of Table 9 reports the linear probability regression estimates (columns 1-2) and Tobit regression estimates (columns 3-6) relating the severity of penalty imposed on accused executives on the fine their firm receives from the SEC.The dependent variable in columns 1-2 is a dummy variable equal to 1 if the firm receiving a fine from the SEC, and 0 otherwise, in columns 3-4 is FIRM_FINE/ALL_FINES, and in columns 5-6 is FIRM_FINE/TOTAL_ASSETS.SEVERITY is a categorical variable equal to 5 if an executive receives both prison term and officer ban, 4 for prison term, 3 for an officer ban, 2 for probation, 1 if there is monetary penalty, and 0 if no penalty is imposed.Panel B reports the linear probability regression estimates relating predicted firm fine on the likelihood of job termination for a fraudulent executive.The dependent variable is EXECUTIVE_TERMINATION, a dummy set to 1 for termination of the fraudulent executive, and 0 otherwise.We generate PREDICTED_SEVERITY (PREDICTED_FIRM_FINE) by regressing SEVERITY (FIRM_FINE/ALL_FINES) on PC.Control variables (identical to those in to circumvent the traditional disciplining mechanisms and may even allow for them to entrench themselves further.17

B. Consequences of Executive Entrenchment
Existing literature has established that shareholders suffer substantial value loss upon the revelation of fraud (e.g., Dechow, Sloan, and Sweeney (1996), Palmrose, Richardson, and Scholz (2004)).Our evidence that political contributions mitigate the likelihood of job termination for fraudulent managers, a disciplinary force against executive misconduct, suggests that value destruction by these fraudsters may persist beyond the aftermath of their misconduct.In this subsection, we examine how the entrenchment of fraudulent managers, aided by their political contributions, affects shareholders after the case resolution.
Through shifting sanctions to shareholders, who already bear the losses from their misconduct, contributing executives are more likely to retain their job.This suggests that firms with entrenched executives may suffer a greater decline in shareholder value and take longer to recover from the fraud than those whose fraudulent executives are not able to entrench themselves through political contributions.To test this conjecture, we compare the standardized cumulative abnormal returns (CAR) following the resolution of the fraud between firms with and without entrenched, politically connected managers.As described in Section II.C, we benchmark CARs using the equally weighted and value-weighted CRSP market returns, respectively, and vary the length of event windows (in days) from [À1, þ30] to [À1, þ360], with day 0 being the time when the case reaches its final resolution.We also consider a firm's profitability (measured by its ROA), Tobin's Q, and likelihood of financial distress (measured by ALTMAN_Z_SCORE) in the 1-, 3-and 5-year windows following the case resolution.
To isolate the effect of entrenchment arising from political contributions rather than fraud-, executive-, firm-, and industry-specific characteristics, we conduct this set of analyses on an Abadie-Imbens (2006) matched sample of fraudulent executives.Matches are created exactly on industry Â year, executive roles, as well as indicator variables for class action and firm fine.We use nearest neighbor matching for firm size, executive age, damages, fraud duration, and fraud types.We then construct EXECUTIVE_ENTRENCHMENT, a dummy variable set to 1 if the fraudulent executive made political contributions during the fraudulent period and retained his job following the penalty phase, and 0 otherwise.
Panel A of Table 10 reports the regression results from comparing the postresolution stock returns of firms with executives that both contribute politically and manage to keep their job (the treated sample) to returns of firms without politically connected, entrenched executives (the control sample).We observe a negative and significant relationship between executive entrenchment and stock returns across all event windows, suggesting that the losses experienced by shareholders of firms whose fraudulent executives were able to entrench themselves through political contributions are significantly larger than the losses to those without entrenchment of politically connected executives.The effect appears to last up to at least 360 trading days beyond the resolution of the case.These results indicate that shareholders of firms whose executives received lighter penalties through political contributions and avoided job termination continue to suffer value losses more than 1 year following the resolution of the fraud case, or, on average, over 6 years after the end of the fraud (Figure 1).
Because cumulative abnormal return can be noisy in longer-term windows (see, e.g., Fama (1970)), In Panel B of Table 10, we replace standardized CARs with firm performance proxies averaged over the 1-, 3-, and 5-year windows after the case resolution, respectively.It is evident that firms with executives that The Consequences of Fraudulent Executive Entrenchment Table 10 relates the entrenchment of fraudulent executives to firm performance after the resolution of the fraud using an Abadie-Imbens (2006) matched sample.The dependent variables in Panel A are the standardized CAR over various windows and in Panel B are ROA, TOBINS_Q, and ALTMAN_Z_SCORE measured over 1-, 3-, and 5-year terms postcase resolution.We multiple the annual average ROA by 100.EXECUTIVE_ENTRENCHMENT is a dummy variable set to 1 if the accused executive contributed politically during the fraud period and is not terminated, and 0 otherwise.Matches are created exactly on industry Â year, executive roles, class action, firm fine, and using nearest neighbor matching for firm size, executive age, damages, fraud duration, fraud types, and whether or not the firm was delisted.The full set of matching variables reflects the control variables in Table 2 contribute politically and are able to avoid termination experience significantly lower profitability and valuation.They also have an increased likelihood of bankruptcy relative to firms damaged by fraud with executives that were unable to subvert the government enforcement process.This impact persists for at least 5 years for Altman Z score and up to 3 years for ROA and Tobin's Q.The result is again consistent with the notion that contributing executives who were able to avoid suitable consequences for their actions continue to harm shareholders, even after the case resolves. 18 Taken together, the results in Table 10 suggest that not only are the fraudulent executives able to use political contributions as a means of shifting penalties away from themselves in the short term, but also those who survive market discipline (i.e., termination) cause further negative consequences to the firm and its shareholders. 19

VII. Conclusions
Using a manually assembled data set on government sanctions, we find that an increase in political contributions is associated with more lenient civil and criminal penalties imposed on fraudulent executives.Contributing executives pay smaller monetary fines and are banned for fewer years as an officer for their wrong doings.They are less likely to be investigated by the DOJ, and they serve less time on probation or prison.The effect varies depending on the type of the fraud and the role of executives, with the top 5 executives in firms with weak boards being able to appropriate more benefit of reduced sanctions from political contributions than less senior ones.Intuitively, contributions to powerful politicians also matter to a greater extent.When identifying potential mechanisms, we find that political contributions allow fraudulent executives to evade harsh sanctions by imposing resource constraints on these agencies and by lengthening the case's time-to-settle with the SEC.
These findings shed light on the factors determining the effectiveness of existing disciplinary mechanisms to discourage fraud.In particular, prior literature indicates that managers responsible for corporate fraud suffer negative career and monetary consequences.By documenting that political contributions lead to less 18 It is possible that the differential effect on valuation and share returns between firms with and without entrenched, politically connected fraudulent managers comes from the fact that share prices were more informative for the entrenched group, as there were fewer surprises in the after-fraud period.To consider this possibility, we reestimate Table 10, replacing dependent variables with several proxies for firm information environment (untabulated).We find no evidence that firms in the entrenched group have more informative share prices.Our findings add to the debate surrounding the SEC's ability to apply civil penalty to both public companies and individuals, which potentially generates substitution between managers' penalties and the firm's penalties.Proponents pointed out that such a substitution can be an efficient way to enforce securities laws, acting as a powerful deterrent to corporate misconduct.Critics argued that the cost of fines was ultimately borne by the shareholders, who are already hurt by the fraud.In the context of our analysis, the two competing views have different implications.The former would imply that, by bearing larger fines, shareholders are better incentivized in sharpening corporate governance and discipling managers.The latter would imply that penalty substitution generates further damages to shareholders.The results from Tables 9 and 10 do not support the "efficient enforcement" view; instead, there is evidence that penalty transfer induces executive entrenchment, resulting further destruction of shareholder value.
severe sanctions, we highlight a mechanism that can undermine this disciplining effect for fraudulent managers and potentially exacerbate an executive's ex ante incentive to commit fraud.
We also provide the first evidence consistent with political contributions facilitating a shift of penalties from fraudulent executives to already damaged shareholders, thereby enabling the entrenchment of these executives in the aftermath of the fraud and resulting in longer-term damage to shareholders.These findings identify an agency cost associated with political contributions that prior studies at the firm level are unable to explore.

Figure 2
FIGURE 2Case Age and PenaltiesFigure2illustrates the distribution of civil penalties for 5 bins representing quintiles of case age, with 1 representing the shortest quintile and 5 representing the longest quintile.Case age is measured as the number of years between the date of the case filed and the time when a new SEC commissioner assumes power. 19

TABLE 1
Descriptive StatisticsPanel A of Table1reports the summary statistics of fraud, penalties incurred by executives accused of fraud, and political contributions.Panel B reports the distribution of the roles of the accused executives and the types of fraud of which they are accused.CEO is the Chief Executive Officer.CFO is the Chief Financial Officer.COO is the Chief Operating Officer.CIO is the Chief Information Officer.CAO is the Chief Accounting Officer.Panel C reports the descriptive statistics of firm performance during the postresolution period.The unit of analysis is the executive observations in Panels A and B and the executive-firm observations in Panel C. The regulatory period is 1996-2015.The postresolution period is 1999-2016.

TABLE 2
Political Contributions and Government Penalties Table 2 relates political contributions to the severity of government penalties.The dependent variables are indicated at the top of each column.Fixed effects are described in the table, but coefficients are not tabulated.Industry is a firm's 1-digit SIC code.Variable definitions are in the Appendix.Robust standard errors are clustered at the firm level and reported in square brackets.***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

Table 2 )
and fixed effects (described in the table) are included in estimations, but coefficients are not tabulated.Industry is a firm's 1-digit SIC code.Variable definitions are in the Appendix.Robust standard errors are clustered at the firm level and reported in square brackets.***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

TABLE 4
Table 2) and fixed effects (described in the table) are included in estimations, but coefficients are not tabulated.Industry is a firm's 1-digit SIC code.Variable definitions are in the Appendix.Robust standard errors are clustered at the firm level and reported in square brackets.***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

TABLE 5
Table 2) and fixed effects (described in the table) are included in estimations, but coefficients are not tabulated.Industry is a firm's 1-digit SIC code.Variable definitions are in the Appendix.Robust standard errors are clustered at the firm level and reported in square brackets.***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

TABLE 7
Variable definitions are in the Appendix.Robust standard errors are clustered at the firm level and reported in square brackets.***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

TABLE 8
Table 2) and fixed effects (described in the table) are included in estimations, but coefficients are not tabulated.Industry is a firm's 1-digit SIC code.Variable definitions are in the Appendix.Robust standard errors are clustered at the firm level and reported in square brackets.***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

Table 2 )
and fixed effects (described in the table) are included in estimations, but coefficients are not tabulated.Industry is a firm's 1-digit SIC code.Variable definitions are in the Appendix.Robust standard errors are clustered at the firm level and reported in square brackets.***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

TABLE 10
. Industry is a firm's 1-digit SIC code.Variable definitions are in the Appendix.Robust standard errors are clustered at the firm level and reported in square brackets.***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.