Political Corruption and Mergers and Acquisitions

Transcription

1 Political Corruption and Mergers and Acquisitions Nam H. Nguyen, Hieu V. Phan, and Thuy Simpson * * Nam H. Nguyen, nguyen.nam@uqam.ca, Department of Finance, School of Management, Université du Québec à Montréal, 315 St Catherine St E, Montreal, QC H2X 3X2; Hieu V. Phan, hieu_phan@uml.edu, Manning School of Business, University of Massachusetts Lowell, 72 University Avenue, MA Phone: (978) ; Thuy Simpson, simpsoth@gvsu.edu, Seidman College of Business, Grand Valley State University, 50 Front Ave SW, Grand Rapids, MI Phone: (616)

2 Political Corruption and Mergers and Acquisitions Abstract We examine mergers and acquisitions (M&As) as a possible channel through which firms shield their liquid assets from expropriation by local officials. We find that firm acquisitiveness increases but firm targetiveness decreases with the levels of local corruption. Acquiring firms located in highly corrupt areas are more likely to use cash for acquisition payment when they have excess cash and create greater shareholder value from cash-financed acquisitions. Our evidence indicates that M&As help acquiring firms shield their liquid assets from local corruption by converting cash into hard-to-expropriate assets and relocate assets from the high to low corruption areas. JEL classifications: G30, G32, G34 Keywords: Political Corruption; Mergers and Acquisitions; Liquid Assets Shielding I. Introduction Corruption is a cancer: a cancer that eats away at a citizen's faith in democracy, diminishes the instinct for innovation and creativity; already-tight national budgets, crowding out important national investments. It wastes the talent of entire generations. It scares away investments and jobs. Joe Biden Political corruption in the U.S. is prevalent despite the country being among the leading economic powers around the world. Results from the annual Corruption Perception Index surveys from 1995 to 2016 pointed to a decline in ranking over time in perceived corruption in the U.S. 2

3 Further corroborating the survey results is the fact that the total number of public corruption convictions in the U.S. has grown significantly, from 244 cases per year in 1976 to over 1,000 cases per year in the past decade. This emerging trend on the pervasiveness and deteriorating nature of U.S. political corruption proves to be costly for the society and business operations. Previous research documents that rent seeking due to corruption increases transaction costs, uncertainty, inefficient investments as well as misallocation of resources (Shleifer and Vishny (1993), Rose- Ackerman (1978)). Corruption also adversely affects public finances in industrialized nations such as the U.S. via multiple channels, including lowering tax revenues, increasing public expenditures and debt, heightening financial risks, and weakening productivity, competitiveness, and economic growth (Kauffman (2010)). The pervasiveness of corruption also has implications for corporate policies, such as cash holdings and capital structure (Fan, Titman, and Twite (2012), Caprio, Facio, and McConnell (2013), Smith (2016)), and capital expenditures (Campos, Lien, and Pradhan (1999), Fisman and Svensson (2007), Malesky and Samphantharak (2008)). However, no prior research has considered the effect of corruption on mergers and acquisitions (M&As), one of the most important forms of corporate investments. In this research, we ask whether and how political corruption affect M&As. To answer these questions, we investigate the effects of U.S. political corruption on several aspects of M&As including firm acquisitiveness, targetiveness, propensity of M&A deal completion, payment consideration for acquisition targets, and acquirer and target shareholder value. Previous research suggests two different hypotheses for the relation between political corruption and firm behavior. The shielding hypothesis posits that firms consider bribes as a tax they want to avoid paying (Smith (2016)). Thus, faced with local corruption, firms seek to reduce corporate liquidity while increasing debt financing with shorter maturity to shield themselves from 3

4 local officials expropriation (Fan, Titman, and Twite (2012), Caprio, Facio, and McConnell (2013), Smith (2016)). Alternatively, the liquidity hypothesis argues that political connections between firms and government officials can be value-increasing through innocuous means such as corruption. Firms can exploit political connections to obtain more government procurement contracts and financial support (Duchin and Sosyura (2012); Goldman, Rocholl, and So (2013)). To the extent that firms can trade cash for political favors, the liquidity hypothesis suggests that firms operating in a corrupt environment will self-position to pay bribes by maintaining a high level of liquid assets, such as cash. In the M&A context, to the extent that shielding liquid assets from local officials expropriation serves shareholder benefits, a firm located in a more corrupt area will be more likely to acquire another firm located in a less corrupt area for two possible reasons. First, acquisitions can help the acquiring firms to convert their liquid assets into hard assets, which are more difficult to extract. Second, acquisitions can facilitate acquirers assets relocation from a more corrupt to a less corrupt area, which further shields their assets from local officials extraction. Also consistent with the shielding hypothesis, firms located in areas that are more corrupt are unlikely the targets of acquisition due to a greater threat of expropriation to the acquiring firms. Following the liquidity hypothesis, if firms located in highly corrupt areas choose to maintain large cash holdings to trade them for political favors, they will be less likely to engage in acquisitions to convert liquid assets into hard assets. Furthermore, since M&As are business-tobusiness transactions, they are less likely affected by the intervention of local officials. Thus, the liquidity hypothesis suggests a negative relation between local corruption and firm acquisitiveness. However, the liquidity hypothesis does not have a clear prediction about the relation between local corruption and firm targetiveness. 4

5 We begin our analysis by examining the relation between the level of local corruption and firm acquisitiveness and targetiveness. Similar to Butler, Fauver, and Mortal (2009) and Smith (2016), we use the corruption per capita measure as a proxy for corruption. Corruption per capita is measured as the yearly number of convictions of each of the 94 federal judicial districts in the U.S., which is obtained from the Report to Congress on the Activities and Operations of the Public Integrity Section, scaled by the respective district population. We then match the corruption per capita data of judicial districts with the Compustat data and the M&A subsample based on firm headquarters location. We use the headquarters location as the identifier since it is where the majority of plants and operations are based (Bai, Fairhurst, and Serfling (2017)). Using a sample that includes 77,338 firm-year observations of 8,314 unique firms, spanning from 1986 to2014, we find that the levels of local corruption are positively related to firm acquisitiveness but negatively related to firm targetiveness. Since local corruption and corporate investments, including M&As, may be driven by local economic conditions, we control for state GPD growth and GDP per capita as surrogates for local economic conditions in our analysis, but our findings are not sensitive to these controls. To address a concern that both local corruption and M&As may be correlated with other unobservable factors, a cause of endogeneity bias, we use the instrumental variable (IV) model for estimation, but our results hold. Overall, our results are consistent with the shielding hypothesis. Next, we investigate the effect of corruption on payment consideration. Smith (2016) finds that firms located in areas that are more corrupt choose to shield their liquid assets by maintaining lower cash reserves but higher financial leverage. This finding suggests that acquiring firms located in corrupt areas are more likely to use stock for acquisition payment due to their low liquidity and high debt ratios. However, the shielding hypothesis also suggests that acquiring firms 5

6 located in more corrupt areas will be more likely to use cash for payment if they have excess cash. We estimate a stock payment probit model and find a positive relation between the corruption level in the acquirers areas and the likelihood of all-stock payment. Furthermore, we find a negative effect of the interaction between firm excess cash and corruption measure on the likelihood of allstock payment. Our evidence is consistent with the predictions of the shielding hypothesis. Even though acquiring firms from highly corrupt areas may choose to pursue M&As to shield their liquid assets from local expropriation, target shareholders would be concerned about the corrupt practice in the acquirers areas and view their bids less favorably. To explore this proposition, we examine the effects of the level of corruption in an acquiring firm s area on the bid premium and the likelihood of the deal completion. We find a positive effect of corruption on the bid premium and a negative effect of corruption on the likelihood of the deal completion. A possible explanation for these findings is that target shareholders are concerned about the postmerger expropriation by local officials in the acquiring firms areas. Consequently, target shareholders demand higher bid premiums as compensation for the expropriation risk and, to the extent that the premiums are not worth the risk, they may reject the bids, leading to lower probability of deal completion. Moreover, we find a negative relation between cash payment and bid premiums, indicating that acquiring firms can save acquisition costs by paying cash for their acquisition deals. Finally, we examine the relations between corruption and acquirer and target shareholder value and acquisition synergies. Using the three-day M&A deal announcement cumulative abnormal stock returns (CAR) as a proxy for shareholder value, we find that, on average, the effect of corruption level in the acquirer s area on acquirer shareholder value is not statistically different from zero. Intuitively, this result could represent the net effect of two countervailing forces. On 6

7 the one hand, the shielding hypothesis suggests that acquiring firms would benefit from M&As, which help them mitigate the expropriation threat of local officials. Indeed, in a complementary analysis, we find that the interaction between corruption and cash payment has a positive effect on acquirer shareholder value, implying that acquiring firms located in more corrupt areas can create greater shareholder value when they use cash for acquisition payment. On the other hand, the higher bid premiums demanded by target shareholders coupled with a lower probability of deal completion induced by corruption in the acquirers areas may reduce the expected acquisition value to acquirer shareholders. To the extent that the two forces offset each other, the average effect of corruption on acquirer shareholder value could be insignificantly different from zero as we observe. Importantly, we note that our results indicate a positive relation between corruption and cash payment conditional on excess cash, a negative relation between cash payment and bid premiums, and a positive relation between corruption and acquirer shareholder value conditional on cash payment. Taken together, our evidence suggests M&As as an effective channel through which acquiring firms can shield their liquid assets from local expropriation. We investigate and find a negative effect of the corruption levels in the acquisition targets areas on their shareholder value. This finding is consistent with the view that targets located in more corrupt areas suffer from larger corruption discount due to the acquirers concern about the threat of expropriation. To examine the effect of corruption on acquisition synergies, we use the three-day value-weighted average of the acquirer-target CARs (WCARs), where the weight is the market values of equity of the acquirer and target measured one week before the bid announcement, as a proxy for the acquisition synergies. Our WCAR analysis reveals a negative relation between the difference in the corruption levels of the acquirer s and target s areas and acquisition synergies. Our evidence suggests that, on average, an acquisition deal between an 7

8 acquirer located in a more corrupt area and a target located in a relatively less corrupt area can create lower synergy for the merged firm, possibly because corruption in the acquiring firm area undermines its ability to create synergies. Our research makes three important contributions to the corruption and M&A literature. First, to the best of our knowledge, this is the first study that examines the effect of political corruption on M&As, one of the most important investment decisions in corporate America. An advantage of using M&As for empirical research is that they are observable at exactly the points in time when the decisions are made, which allows us to establish a direct link between political corruption and M&As. As M&As enable firms to convert cash into hard-to-extract assets and relocate assets from a more corrupt to a less corrupt area, acquiring firms can shield their liquid assets from local officials expropriation. Thus, our evidence suggests M&As as an effective channel through which firms can shield their liquid assets from local officials expropriation. Second, we add to the M&A literature by suggesting local corruption as a determinant of firm acquisitiveness and targetiveness. To the extent that M&As represent the market discipline to reallocate assets to a better use while improving general economic efficiency, the negative effect of local corruption on firm targetiveness indicates that corruption impedes market discipline and hampers efficient asset reallocation. Finally, our research has important implications for policy makers, corporate managers, and investors. Our evidence indicates that local corruption discourages businesses from investing in their existing areas while encouraging them to relocate their assets to other less corrupt areas. Corruption, thus, distorts corporate investments and raises business costs, which adversely affect local economies and investors benefits. The remainder of the paper is organized as follows. We present a description of the data 8

9 and variables construction in Section II. Section III develops empirical predictions and discusses the research methods and results. Section IV presents robustness checks and Section V concludes the paper. II. Samples, Variables Construction, and Descriptive Statistics We obtain firm accounting data and stock price and return data from Compustat and the Center for Research in Security Prices (CRSP) databases, respectively. We obtain the M&A data from the Securities Data Company s (SDC) Platinum Database. We merge the M&A data and Compustat data to form the full sample to investigate the effect of political corruption on M&A likelihood. Following the literature, we exclude firms from the utility (Standard Industrial Classification (SIC) codes from ) and financial industries (SIC codes ) since these industries are subject to more stringent regulations. We retain firm-year observations with at least one M&A deal to form the M&A subsample for cross-sectional analysis. Moreover, to focus on M&A deals that can have significant effects on the acquiring firms, we exclude small M&A deals with values below one million U.S. dollars from the M&A subsample. The sample period spans from 1986 to Similar to Butler et al. (2009) and Smith (2016), we obtain the number of yearly convictions of each of the 94 U.S. federal judicial districts from the Report to Congress on the Activities and Operations of the Public Integrity Section and use them as a measure of political corruption. The yearly conviction numbers are scaled by the annual population estimates from the U.S. Census for each judicial district to ensure that a high corruption level is not merely due to a bigger population of a district. By construction, a higher level of corruption per capita of a judicial district indicates a more corrupt operating environment in that district. We manually identify the ZIP codes associated with each federal judicial district and merge the corruption data with the full 9

10 sample and the M&A subsample using the ZIP codes of firm headquarters locations. Table 1 presents the annual and 2-digit SIC code industry distribution of the M&A subsample over the sample period in Panels A and B, respectively. The annual number of M&As deals increased over the period , peaking in 1998 before decreasing during the recessions in and Examples of industries that experience high frequency of M&As include business services, electronic and other electrical equipment, chemicals and allied products, instruments and related products, industrial and commercial machinery, and computer equipment. [Insert Table 1 about here] We report the summary statistics of the full sample and the M&A subsample in Panels A and B, respectively, of Table 1. The full sample includes 77,338 firm-year observations while the M&A subsample consists of 7,325 firm-year observations. Acquirer corruption is the corruption per capita of the judicial district in which an acquirer s headquarters is located. Firm size is measured as the natural logarithm of total book value of assets. Market-to-book is calculated as the ratio of the market value of assets to the book value of assets. Book leverage is the ratio of the book value of debt divided by the book value of assets. Past 12-month returns is the acquirer 12- month buy-and-hold stock return in the year prior to the M&A announcement. Average sales growth is the average of 3-year sales growth rate. Noncash working capital is the working capital minus cash, scaled by the book value of assets. Firm age is the number of years a firm has been included in the Compustat database. Other variables are defined in Appendix A. The average corruption per capita of the full sample and M&A subsample are and 0.333, respectively, which are close to that (0.327) reported by Smith (2016). The average firm size, market-to-book 10

11 ratio, past 12-month returns, and firm age of the M&A subsample appear to be larger than those of the full sample. [Insert Table 2 about here] III. Empirical Predictions, Research Methods, Results, and Discussions In the context of macroeconomic and public choices, theory and empirical evidence suggest that corruption impedes transactions, hinders trade, retards financial and economic growth and development (Rose-Ackerman (1978), Shleifer and Vishny (1993), Mauro (1995), and Wei (2000). The impacts of political corruption have also been examined in the context of municipal finance and sovereign credit ratings. To the extent that corruption affects the quality of a country s legal and political environment, Butler and Fauver (2006) document a significant impact of institutional environment on sovereign credit ratings. In taking a closer look at the impact of political corruption on municipal finance, Butler, Fauver and Mortal (2009) find that corruption significantly affects credit risk and municipal bond pricing. Their study documents a corruption premium that is roughly the equivalent of a bond issue being rated more than 2.0 notches lower. A. Political Corruptions, Firm Acquisitiveness, and Firm Targetiveness In this section, we investigate the extent to which U.S. political corruption affects firm acquisitiveness and targetiveness. Shielding hypothesis: As political corruption presents an external force that shapes financial markets and corporate behaviors, a growing body of literature on corruption has begun to examine the impact of corruption on firm behavior and value. In this stream of research, there emerges two competing hypotheses on the effects of political corruption on firm policies. Early theoretical work by Myers and Rajan (1998) and Stulz (2005) lay the foundation for the shielding hypothesis, which posits that firms are more likely to channel its assets to harder-to-extract 11

12 investments to reduce rent-seeking that arises from political extraction. Fan, Titman, and Twite (2012) examine the impact of corruption, among other institutional differences, on the capital structure and debt maturity choices of firms from 39 developed and developing countries. These authors find that firms located in more corrupt countries use more debt financing with shorter maturity to shield their assets from political expropriation. Similarly, Caprio, Facio, and McConnell (2013) study the effect of political corruption on corporate liquidity and find that firms respond to political corruption and threats of rent extraction by holding significantly lower liquid assets. Also consistent with the shielding hypothesis, these firms channel their cash to harder-toextract assets, including property, plant, equipment, inventory, and dividends. These findings suggest that political corruption increases not only the direct costs from rent-seeking but also the indirect costs associated with asset reallocation that deviates from an otherwise optimal structure. Smith (2016) examines the impact of U.S. corruption on firm behavior and finds that firms manage their cash level downward while managing their debt ratio upward to shield themselves from political expropriation. To the extent that shielding liquid assets from local officials expropriation plays a role in M&A decision, we predict that firms located in more corrupt areas are more likely to pursue acquisitions to convert their liquid assets into hard assets, which are more difficult to extract. In addition, acquisitions also allow acquiring firms to relocate assets to less corrupt areas, further shielding their assets from potential expropriation by local officials. In contrast, firms in highly corrupt areas will be less likely to become acquisition targets due to a greater threat of political extraction to the acquiring firms. Liquidity hypothesis: In contrast to the shielding hypothesis, the liquidity hypothesis suggests that corruption could be value-increasing for firms. There are numerous anecdotal 12

13 examples suggesting corruption as a contributing factor to firms winning government contracts and political favors. For example, in 2004, former-governor of Connecticut John Rowland pleaded guilty to corruption that involved $107,000 worth of vacations, [free] work on his cottage and free flights from state contractors and others. 1 In 2013, Kwame Kilpatrick, the former city mayor of Detroit, was sentenced to 28 years in prison for racketeering conspiracy, fraud, extortion, and tax crimes. 2 More recently, former New Mexico state senator Phil Griego was accused of using his position as a lawmaker to profit in a real estate sale. In 2017, he was convicted of fraud, bribery, and felony ethical violations in corruption trial and could face up to 17.5 years in prison. 3 If firms can trade cash for political favors, such as securing governmental contracts, which benefit the firms, the liquidity hypothesis suggests that firms operating in a corrupt environment will selfposition to pay bribes by maintaining a high level of liquid assets such as cash. If firms in highly corrupt areas choose to maintain large cash holdings to trade them for political favors as predicted by the liquidity hypothesis, they will be less likely to acquire other firms because acquisitions result in a conversion of the acquirers liquid assets (i.e., acquirers cash) into target firms hard assets. Furthermore, M&As represent business-to-business transactions, which are more likely subject to competition but less likely affected by the intervention of local government officials, implying a negative relation between local corruption 1 Available at last accessed on January 13, Available at last accessed on January 13, Available at last accessed on January 13,

14 and firm acquisitiveness. The liquidity hypothesis does not have a clear prediction about the relation between corruption and firm targetiveness. We examine the effect of political corruption on firm acquisitiveness using the following probit model: M&A dummyi,t = α + β*district corruptioni,t-1 + λ*ci,t-1 + δyear dummies + γindustry dummies + εi,t (1) where M&A dummy is an indicator variable that takes a value of 1 if firm i engages in at least one acquisition announcement in year t, and 0 otherwise. District corruption measures the level of political corruption of the fiscal year preceding the M&A announcement in the judicial district in which firm i s headquarters is located. 4 Following the M&A literature, we control for various firm characteristics documented to have power in explaining firm acquisitiveness including size, market-to-book ratio, book leverage, past 12-month returns, firm age, non-cash working capital, and average sales growth. The control variables are lagged by one period to alleviate possible endogeneity concern. We additionally control for industry and year fixed effects in our M&A probit model. The definitions of the variables are provided in the Appendix. Panel A of Table 3 reports the M&A probit model results. The coefficients of district corruption are positive (ranging from to 0.043) and statistically significant at the 5% and 1% levels. These results indicate that firms headquartered in more corrupt areas are more likely to pursue M&As. 5 To illustrate the economic effect of corruption, we calculate and find that, holding 4 Our results are qualitatively similar if we use the contemporaneous level of political corruption. 5 In an unreported univariate analysis, we find that acquirers, on average, choose targets located in relatively less corrupt areas. 14

15 other variables unchanged at their sample means, a 1-standard-deviation increase in acquirer corruption measure above its sample mean is associated with a 1.54-percentage-point increase in acquisition probability, which is economically important given the sample s unconditional M&A probability of 25.03%. This finding is consistent with the shielding hypothesis. [Insert Table 3 about here] Both the corruption level and M&A activities could be correlated with unobserved factors, such as the economic conditions of the firms headquarters states, suggesting that our M&A probit model could be prone to the omitted variable problem. This problem raises endogeneity concern that potentially biases the coefficient estimates of the M&A probit model. We address this endogeneity concern by re-estimating the M&A probit model that additionally control for the natural logarithm of the state GDP per capita and state GDP growth rate, and report the results in Column 5 of Panel A of Table 3 (we further address the endogeneity concern using the instrumental variables (IV) regression method as discussed in the robustness check section below). The estimation results indicate that the coefficient of district corruption remains positive (0.058) and statistically significant at the 1% level. This finding suggests that our M&A probit model results are robust. We investigate the effects of political corruption on the firm targetiveness by estimating the following probit model: Target dummyi,t = α + β*district corruptioni,t-1 + λ*ci,t-1 + δyear dummies + γindustry dummies + εi,t (2) where Target dummy takes a value of 1 if a firm is an acquisition target in a given year, and 0 otherwise. Our variable of interest is district corruption, which is measured as corruption per capita of the judicial district where a firm is headquartered. Ci,t-1 is a set of control variables similar to 15

16 that in Equation 1. Panel B of Table 3 reports the estimation results of the targetiveness probit models. The coefficient estimates of district corruption are negative ( and ) and statistically significant at the 1% level, suggesting that firms located in more corrupt areas are less likely to be acquisition targets. In another analysis, we rerun the targetiveness probit model that further controls for the natural logarithm of state GDP per capita and state GDP growth rate and report the results in Column 3 of Panel B, Table 3. The results indicate that the coefficient of district corruption is negative (-0.04) and statistically significant at the 1% level, suggesting that our finding is unlikely subject to the omitted variable bias concern. In summary, our evidence in this section is consistent with the shielding hypothesis according to which the threat of expropriation by local officials increases (decreases) firm acquisitiveness (targetiveness). B. Corruption and Acquisition Payment Consideration In this section, we investigate whether and how political corruption affects payment consideration for acquisition deals. To the extent that firms respond to local officials rent-seeking by either shielding their assets from potential expropriation or by positioning themselves to take advantage by giving bribes, corruption will have different implications for M&As payment consideration. The shielding hypothesis suggests that firms hold lower cash reserves but maintain higher debt ratios to minimize rent-seeking by public officials in highly corrupt operating environment (Smith (2016)). Due to their low level of corporate cash holdings and high level of financial leverage, acquirers in highly corrupt areas are more likely to use stock as a medium of payment for M&A deals. The shielding hypothesis further suggests that if acquiring firms in highly corrupt areas have excess cash, they are more likely to use cash for acquisition payment to reduce 16

17 exposure to local officials expropriation. The liquidity hypothesis predicts that firms operating in more corrupt areas maintain larger cash holdings. As such, we expect that acquiring firms located in more corrupt areas are more likely to use cash to finance M&As. It is noteworthy that the liquidity hypothesis suggests a positive relation between corruption and cash payment on average, whereas the shielding hypothesis suggests a positive relation between the two conditional on excess cash. We use the following probit model to examine the relation between political corruption and payment consideration: Stock_dummyij = α + β* Acquirer corruptioni,t-1 + ϒ*Target corruptioni,t-1 + λ*ci,t-1 + δyear dummies + γindustry dummies + εi,t (3) where stock dummy is an indicator variable that takes a value of 1 if the payment for M&A deal j of firm i is fully in stock, and 0 otherwise. Acquirer corruption (target corruption) is the level of corruption in the acquirer s (target s) judicial district. Following the literature, we control for firm and deal characteristics such as size, market-to-book, past 12-month returns, average sales growth, book leverage, noncash working capital, firm age, excess cash, deal ratio, stock dummy, cash dummy, diversifying dummy, hostile dummy, public dummy, and challenge dummy. Appendix A provides the descriptions of the variables. The probit model estimation results reported in Table 4 indicate a positive and significant relation between the acquirer area s corruption level and stock payment. We further run the stock payment probit model with an interaction between acquirer corruption and corporate excess cash. We estimate excess cash as the residuals from the regression of firm cash holdings on firm size, financial leverage, market-to-book ratio, cash flows, standard deviation of the last 10 years cash flows, net working capital, capital expenditures, research and development expenses, acquisition 17

18 spending, dividend expense, S&P credit ratings, and industry and year fixed effects (Harford (1999)). The estimation results report in Panel B of Table 4 indicate that the coefficients of the interaction between acquirer corruption and excess cash are negative and statistically significant at the 1% level in all models, indicating that acquiring firms with large excess cash and located in more corrupt areas are less (more) likely to use stock (cash) as the medium of payment for M&A deals. The effect of the target district s corruption level on the payment consideration is inconclusive. Taken together, our results of the payment consideration analysis are consistent with the shielding hypothesis. [Insert Table 4 about here] C. Political Corruption, M&As Bid Premiums, and Completion Likelihood In this section, we investigate the effects of corruption on M&A bid premiums and the likelihood of deal completion. To the extent that corruption increases operating costs and impedes business operations, we predict a positive (negative) relation between the level of corruption in an acquirer s area and the bid premium (the probability of deal completion). Moreover, we expect these relations to be more pronounced when an acquirer firm located in an area that is relatively more corrupt than the target s location area. Alternatively, to the extent that corruption facilitates business operation and creates value for firms as predicted by the liquidity hypothesis, we expect a negative (positive) relation between corruption in the acquirer s area and the bid premium (the probability of deal completion). We examine the effect of political corruption on bid premiums by estimating the following regression model: Bid premiumij = α + β* Acquirer corruptioni,t-1 + ϒ*Target corruptioni,t-1 + λ*ci,t-1 + δyear dummies + γindustry dummies + εi,t (4) 18

19 where Bid premium is measured as the percentage difference between the bid price and the target s stock price one week before the deal announcement. C is a vector of control variables that include firm and deal characteristics similar to those in Equation 3 (Officer (2003), Dimopoulos and Sacchetto (2014)). Table 5 reports the results of the bid premium regressions. The coefficients of acquirer corruption are positive, ranging from to 0.058, and statistically significant in all models. However, we observe a negative relation between cash payment and bid premiums. Our results indicate that acquiring firms from highly corrupt areas have to pay higher bid premiums but using cash for acquisition payment, on average, decreases the bid premiums that the acquirers have to pay. On the other hand, the coefficients of target corruption are statistically insignificant. [Insert Table 5 about here] Next, we investigate the relation between political corruptions and the propensity of M&As completion. The dependent variable is completion dummy that takes a value of 1 for a deal completion in a given year, and 0 for a deal withdrawal. Our completion probit model includes either acquirer corruption and target corruption or corruption delta, which is measured as the corruption level of the acquirer s area minus that of the target s area. The Probit model estimation results reported in Panel A of Table 6, which include both acquirer corruption and target corruption, are consistent with the view that corruption impedes business operation. Specifically, the coefficient estimates of acquirer corruption are negative and statistically significant, indicating that acquirers located in more corrupt areas have lower probability of deal completion. Similarly, the negative and highly significant coefficients of corruption delta reported in Panel B of Table 6 suggest that acquirers located in more corrupt areas have lower probability of completing acquisition deals with targets located in less corrupt areas. Intuitively, target shareholders may 19

20 reject the bids of acquirers from highly corrupt areas, perhaps due to their concern about the risk of expropriation in the acquirers areas. [Insert Table 6 about here] C. Political Corruptions and Acquirer Shareholder Value Our empirical results thus far demonstrate that corruption has significant effects on several aspects of M&As including firm acquisitiveness, targetiveness, methods of payment, bid premiums, and the M&A likelihood of completion, which are consistent with the shielding hypothesis. In this section, we are interested in examining the effects of corruption on acquirer and target shareholder values, and the synergies of M&A deals. The shielding hypothesis suggests two possible opposing effects of the level of corruption in the acquirer s area on acquirer shareholder value. On the one hand, acquirer corruption may be negatively related to acquirer shareholder value due to the direct and indirect costs associated with corruption. On the other hand, acquirers from highly corrupt areas may create shareholder value through acquisitions if they help shield firm liquid assets from expropriation by local officials. Therefore, the net value effect of acquirer area s corruption level on acquirer shareholder is an empirical question. On the other hand, the shielding hypothesis suggests a positive relation between target corruption and acquirer shareholder value since acquiring firms can exploit the target s corruption discount. Moreover, acquisitions can also shield the targets liquid assets from expropriation by local officials in the targets areas following the merger if the merged firms operate in a less corrupt environment. Alternatively, the liquidity hypothesis suggests a positive relation between the corruption level in the acquirer s area and acquirer shareholder value. However, the liquidity hypothesis does 20

21 not have a clear prediction about the relation between the corruption level in the target s area and acquirer shareholder value. Panel A of Table 7 reports the results of the cross-sectional regressions of acquirer threeday abnormal stock returns (CAR(-1, 1)) on acquirer corruption, target corruption, and other firm and deal characteristics. We use the market model and value-weighted CRSP index returns to estimate acquirer three-day CARs. Similar to previous M&A research, we control for the following firm and deal characteristics in the CAR regressions: size, market-to-book ratio, financial leverage, excess cash, past 12-month returns, high-tech dummy, cash dummy, stock dummy, deal attitude dummy, target public status, challenge dummy, and industry M&A intensity. All variables are defined in Appendix A. Firms may self-select to pursue M&As and, all else being equal, they are more likely to pursue M&A deals that create larger shareholder value. This observation indicates that our crosssectional regression is subject to the self-selection problem that potentially biases the coefficient estimates. We control for firm self-selection bias by using the Heckman s (1976, 1979) two-step self-selection correction model. Specifically, we use the estimates from the M&A probit model to calculate the inverse Mill s ratio (IMR), then we include IMR in the cross-sectional regressions as an additional control variable. The acquirer CAR regression results reported in Panel A of Table 7 indicate that acquirer corruption is statistically insignificant. It is possible that the two opposing value effects of corruption that we discuss above offset each other, leading to the insignificant coefficient of acquirer corruption. On the other hand, the coefficients of target corruption are positive and highly significant, indicating that acquiring a firm located in a more corrupt area can create more value for acquirer shareholders, which is consistent with the prediction of the liquidity hypothesis. 21

22 Using the coefficient estimates to calculate the economic effect of target corruption on acquirer shareholder value, we find that, holding other variables fixed at their sample means, a one standard deviation increase in target corruption above its sample mean is associated with a 0.21 percentagepoint (i.e. 0.21%) or $10.2 million increase in acquirer shareholder value. [Insert Table 7 about here] In a complementary analysis, we examine the effect of an interaction between payment consideration and corruption on acquirer shareholder value. We rerun the acquirer CAR regression with interactions between acquirer corruption and stock dummy and cash dummy variables (mixed stock and cash payment is left out to avoid perfect collinearity) and report the results in Panel B of Table 7. The results indicate that the coefficients of the interaction between acquirer corruption and cash dummy are positive and statistically significant, indicating that using cash to pay for M&As deals can increase shareholder value for acquiring firms located in highly corrupt areas. This evidence is consistent with the argument of the shielding hypothesis that firms can shield its cash from potential expropriation by using cash for acquisition payment, which is viewed favorably by shareholders. The interaction effect of stock dummy and acquirer corruption on acquirer shareholder value is inconclusive. We examine the effects of political corruption on target CARs and report the results in Panel C of Table 7. The coefficients of acquirer corruption are positive and statistically significant at the 5% level in all specifications, indicating a positive relation between the corruption level in the acquirer s area and target shareholder value. We interpret this result as evidence that acquirers located in more corrupt areas have to pay higher bid premiums that benefit target shareholders. However, the coefficients of target corruption are negative and statistically significant in Columns 1 and 2, suggesting that target firms located in more corrupt areas suffer from a corruption discount 22

23 that decreases shareholder value. The estimation results indicate that, holding other variables unchanged at their sample means, a one standard deviation increase in target corruption above its sample mean is associated with a decrease of 0.48 percentage-point (i.e., 0.48%) or $16.1 million in target shareholder value. It is crucial to examine the effects of political corruptions on acquisition synergies. We use the three-day acquirer and target value-weighted CAR (WCAR) as a proxy for acquisition synergies. Table 8 reports the results of the WCAR regression on acquirer and target corruption levels and other firm and deal characteristics. We control for acquirer and deal characteristics in the WCAR crosssectional regressions, which include size, market-to-book ratio, financial leverage, excess cash, past 12-month returns, high-tech dummy, cash dummy, stock dummy, deal attitude dummy, target public status, challenge dummy, and industry M&A intensity. The results reported in Table 8 indicate that acquisition synergies are negatively (positively) related to the corruption level in the acquirer s (target s) area. The economic effect of political corruptions on the M&A synergies is also important. Our calculation indicates that, holding other variables fixed at their sample means, a one standard deviation increase in acquirer corruption (target corruption) above its sample mean is associated with a decrease (increase) of 0.22 percentage-point or $10.9 million (0.34 percentagepoint or $16.1 million) in WCAR. Intuitively, following the acquisition completion and integration of the target firms into the acquiring firms, the merged firms are more likely affected by the corruption in the acquirer areas. Therefore, the low (high) corruption level in the acquirer areas could reduce (increase) the threat of expropriation by local officials, leading to higher (lower) acquisition synergies. Moreover, an acquisition can shield a target liquid assets from expropriation by local officials in the target s area, leading to greater value for the merged firms. [Insert Table 8 about here] 23

24 To further explore the proposition that the synergistic gains from M&As arise from the improvements in business operating environment due to the change in corporate control, we rerun regression of WCAR on corruption delta and other controls. The results reported in Panel B of Table 8 indicate that the estimated coefficients of corruption delta are negative ( and ) and statistically significant at the 5% and 1% levels, which is consistent with our expectation that an M&As deal creates greater synergy when an acquirer located in a less corrupt area acquires a target located in a more corrupt area. IV. Robustness Checks Although our regressions control for state GDP growth and GDP per capita of the acquirer headquarters states, political corruption and M&As can be jointly correlated with unobservable variables, raising an endogeneity concern due to the omitted variable. Endogeneity could bias the coefficient estimates of the M&As probit models. In the next analysis, we use the instrumental variable (IV) approach to address endogeneity concern. In particular, we use the isolation measure of state capital city suggested by Campante and Do (2014) and Smith (2016) as an instrument for state corruption. 6 This variable measures the concentration of a state s population and its distance relative to a state s capital city. As such, a measure of 0 indicates a state s population concentrates around its capital while a measure of 1 suggests that a majority of the state s residents live as far away from the capital city as possible. Campante and Do argue that politicians are more corrupt in the isolated capital cities due to lower media coverage of politics and less oversight from voters in these cities. The isolation of state capital city is a valid instrument in our analysis since it relates to state corruption while having no direct ties to M&A activities. 6 We thank Campante and Do for making the isolation of capital city data available. 24

25 We report the results of M&A IV probit model in Table 9. The first-stage results of the IV probit model reported in Column 1 indicate that the coefficient of the isolated capital city is positive (0.601) and significant at the 1% level, confirming the positive relation between the instrument and corruption measure. The test statistics of the underidentification and weak identification tests indicate that the selected instrument is strong and relevant. The results of the second-stage of the IV probit model reported in Column 2 indicate that the coefficient of the instrumented acquirer corruption variable is positive (0.709) and statistically significant at the 1% level, suggesting that our findings are robust to endogeneity correction. [Insert Table 9 about here] We estimate the targetiveness IV probit model and report the results in column 3 and 4 of Table 9. The first-stage results and test statistics of additional identification tests reported in column 3 confirm the strength and relevance of the selected instrument. The results of the secondstage IV probit model reported in Column 4 show that the coefficient of the instrumented target corruption variable is negative (-0.203) and highly significant, indicating that the results of the targetiveness probit model are also robust to endogeneity correction. 7 V. Conclusions Previous research argues that faced with political corruption, firms may attempt to shield their assets from local officials expropriation or maintain high liquidity to trade cash for political 7 The coefficient estimates of instrumented acquirer corruption and instrumented target corruption appear to be larger in magnitude than those of acquirer corruption and target corruption of the Probit model. The reason could be that the IV models identify the local average treatment effect of the endogenous variable on the outcome variable (Imbens and Angrist (1994)). 25

26 favors. Using the annual number of convictions in a federal judicial district scaled by the district s population as a proxy for corruption, we find that local corruption is positively related to firm acquisitiveness but negatively related to firm targetiveness. When pursuing acquisition deals, acquiring firms located in highly corrupt areas are more likely to use stock payment, possibly because these firms typically maintain low cash reserves and high debt ratios to avoid local officials extraction. However, when these firms have excess cash, they are more likely to use cash for acquisition payment. This behavior could help acquiring firms shield liquid assets from expropriation by converting liquid assets into harder-to-extract assets and relocate assets from a high to a low corruption environment. We further find that corruption is positively related to the bid premiums but negatively related to the propensity of deal completion. We interpret these results as evidence that target shareholders are concerned about the threat of expropriation associated with corruption in the acquiring firms areas, therefore they demand higher bid premiums as compensation for bearing the expropriation risk or even reject the bids if additional compensation is not worth the risk. However, bid premiums are negatively related to cash payment. We examine the relations between corruption and acquirer and target shareholder value and find that, on average, corruption in the acquirers areas does not have a significant impact on acquirer shareholder value, which could be due to the opposing value effects of corruption offsetting each other. However, we find that acquiring firms from highly corrupt areas that use cash for acquisition payment can create greater value for shareholders, implying the benefits of using M&As as a way to shield the acquirers liquid assets. On the other hand, target shareholders suffer from a corruption discount. Finally, we find that an acquisition tends to create greater synergy when the acquirer from a low corruption area acquires a target in a high corruption area. 26