Labor Unionization and Stock Price Crash Risk

Transcription

1 Labor Unionization and Stock Price Crash Risk Abstract This study investigates the impact of labor unionization on a firm s future stock price crash risk. We find that labor unions are negatively associated with the firms crash risk. Such negative relation is more pronounced in situations where the firms can intimate more credible evidence on unfavorable prospects, and for firms that face stronger external shareholder monitoring. Our findings are consistent with the notion that firms take strategic actions to reduce the bargaining advantages enjoyed by labor unions, which leads to lower future stock price crash risk. JEL classification: G14; G30; G32; J52 Keywords: Labor unions; bargaining position; crash risk; external monitoring 1

2 Labor Unionization and Stock Price Crash Risk 1. Introduction Employees play a vital role in a firm s value creation. Unionized employees have a greater ability to influence a firm s operations than their nonunionized peers. For example, unionized firms workers tend to have limited operational flexibility, (Chen, Kacperczyk, and Ortiz-Molina, 2011), retain lower cash holdings (Klasa, Maxwell, and Ortiz-Molina, 2009), and engage in less aggressive tax activities (Chyz et al., 2013). In this paper, we investigate whether labor unionization affects a firm s future stock price crash risk. Stock price crash risk increases with managers incentives to conceal adverse operating outcomes. Managers have various incentives to withhold bad news to inflate current share prices at the expense of long-term firm value (Kothari, Shu, and Wysocki, 2009). When the amount of bad news that managers accumulate during a certain period reaches a tipping point, it will suddenly come out all at once, resulting in a sharp stock price crash (Jin and Myers, 2006; Hutton, Marcus, and Tehranian, 2009). Additionally, managers hiding negative information prevents investors and other monitors from taking timely corrective actions or liquidating unprofitable projects at an early stage. Consequently, poor performance of those bad projects accumulates over a longer period than otherwise, which will eventually materialize and lead to asset price crashes (Bleck and Liu, 2007). Labor unions might reduce incentives and abilities of managers to hoard negative information, and thus lower stock price crash risk. On the one hand, labor unionization is widely associated with raising wages and imposing other sticky costs on employers (Lewis, 1986). Corporate managers often strive to take strategic actions to improve their bargaining position against labor unions. To shelter corporate profits from unionized employees salary increase demand, unionized firms are more inclined to adopt conservative financial reporting policies and to timely intimate a negative outlook to their employees 2

3 (DeAngelo and DeAngelo, 1991; Leung, Li, and Rui, 2012; Bova, 2013), which lowers the likelihood of future stock price crashes. On the other hand, employees, like creditors, primarily hold a fixed claim on a firm s cash flow in the form of wages and salaries. Employees prefer less risk than shareholders or managers since they have little to gain when a firm performs better than expected, but have much to lose when it performs poorly (Faleye, Mehrotra, and Morck, 2006). Employees also require a premium in wages or benefits as compensation for unemployment risk that increases with a firm s preference to risky investments (Li, 1986; Hamermesh and Wolfe, 1990). Unionization enables employees to play a more influential role in refraining managers from undertaking risky investment projects. The greater ability of unionized employees to extract above-market rents and capture future profits reduces returns on risky investments, while compensation premium demand for increased unemployment exposures increases the cost of risky investments. Unionized firms are thus less willing to invest in risky projects on the margin. Additionally, unionized employees play a tighter monitoring role in the firms than individual employees. Their high concerns about downside risks motivate them to urge the implementation of conservative information disclosure polices so that they are able to take protective actions upon receiving bad news in a timely manner. Taken together, labor unions are expected to undermine managers incentives to stockpile bad news and constrain managers excessive risk-taking behaviors. We therefore predict a negative relation between labor unionization and a firm s future stock price crash risk. Following prior literature (Klasa, Maxwell, and Ortiz-Molina, 2009; Chen, Kacperczyk, and Ortiz-Molina, 2011), we use industry unionization rate, defined as the percentage of workers in a firm s industry covered by unions in collecting bargaining agreements, to proxy for labor unions ability to affect the firm s behaviors. Following prior studies (e.g., Chen, Hong, and Stein, 2001; Kim, Li, and Zhang, 2011b; Kim, Li, and Li, 2014), we measure crash risk as the conditional skewness of return distribution, which captures asymmetry in risk, especially downside risk. Based on a sample of 48,020 firm-years over the period 3

4 , our empirical results support that firms with higher unionization rates are less likely to experience firm-specific stock price crashes in the future. To determine whether the negative relation between unionization and future stock crashes is due to managers motivations to weaken labor union s collective bargaining advantage, we further examine how this relation is affected by the credibility of negative outlook intimated by firms. Prior literature argues that firms are more likely to successfully obtain concessions from unions in negotiations if their strategic actions like smaller cash reserves are more credible evidence that they are unable to concede to unions demands. The negative outlook is viewed as more credible for firms that do not pay dividends, are more likely to be financially distressed (DeAngelo and DeAngelo, 1991; Klasa, Maxwell, and Ortiz-Molina, 2009). Consistent with these expectations, we find that the negative relation between unionization and future stock price crash risk is more pronounced for firms without dividend payout, firms with lower Altman-Z scores, and firms with lower profitability. If managers endeavors to counter unions impact on firm operations are in the interests of shareholders, we expect the negative relation between unionization and stock price crashes to be more pronounced for firms with stronger shareholder monitoring. Using the level of institutional ownership and shareholder rights as proxies for the strength of external shareholder monitoring, we find that the mitigating effect of unionization on future stock price crash risk is more pronounced for firms with greater shareholder rights and higher institutional shareholdings. A potential concern is that our empirical analysis may suffer from endogeneity issues. For example, the negative relation between unionization and stock price crashes might be spurious if unions self-select to less risky industries or to firms that choose timely information disclosure practices. We conduct instrumental-variables regression analysis to address endogeneity concerns. Following Chen, Kacperczyk, and Ortiz-Molina (2010, 2011), we use the percentage of female workers in a firm s industry as an instrument. The instrumental-variables results remain qualitatively unchanged. It provides support that labor unionization is negatively associated with future stock price crashes. 4

5 We conduct some additional tests to further understanding the negative relation between unionization and stock price crashes. Matsa (2010) find that the strategic incentives to improve their future bargaining position against unions are greater for unionized firms with higher profit variability. It suggests that the impact of labor unionization on stock price crashes is affected by a firm s profit variability. Consistent with the greater strategic incentives to gain bargaining position, we find that the negative relation between unionization and stock crash risk is more pronounced for firms with higher profit variability. Labor related costs are substantial for labor intensive firms. We predict that such firms are more sensitive to the wages increase demand and thus have stronger incentives to improve their bargaining position against unions. Our empirical results are consistent with expectations that the negative relation between unionization and stock price crashes is more pronounced for more labor intensive firms. This study makes several contributions to the literature on labor unions and stock price crashes. First, this study adds to the recent stream of research on the economic consequences of labor unionization. For instance, as strategic responses to union s powerful rent-seeking motives, unionized firms invest less in long-term and intangible assets (Connolly, Hirsch, and Hirschey, 1986), hold lower cash reserve (Klasa, Maxwell, and Ortiz-Molina, 2009), bear higher debt (Matsa, 2010), and refrain from certain risky but valuable investments (Faleye, Mehrotra, and Morck, 2006; Chyz, et al., 2013). Second, this study is closely related to literature on the firms information reporting strategies. Prior literature shows that firms engage in downward earnings management and adopt conservative financial reporting practices to gain bargaining advantages with unions (DeAngelo and DeAngelo, 1991; D souza, Jacob, and Ramesh, 2001; Leung, Li, and Rui, 2012). Stock price crashes primarily results from firms withholding bad news. Financial reporting strategies are only one of many ways for firms to withhold bad news (Hutton, Marcus, and Tehranian, 2009). Our findings shed additional light on how firms information disclosure practices are affected by labor unionization. Finally, our research extends the emerging literature of forecasting future stock price crash risk. This study provides evidence that the strategic interactions between firms and labor unions are an incremental predictor of future stock crash risk. Moreover, our evidence has implications beyond the labor bargaining environment. The strategic interactions between firms and other 5

6 stakeholders, for example, major suppliers and customers, might affect their information reporting practices and in turn future stock crash risk as well. The remainder of the paper proceeds as follows. Section 2 reviews the related literature and develops the hypotheses. Section 3 describes our sample and research design. Section 4 and 5 discuss the empirical results. Section 6 concludes. 2. Related literature and hypotheses Labor unions substantially affect a firm s decision-making in two different ways. On the one hand, labor unions have higher power to increase employees share of the firm s surplus (Hirsch, 1991, 2008), which induces managers to take strategic actions to mitigate unions collective bargaining advantage and shelter corporate income from union demands. On the other hand, employees are fixed claimants like debtholders and have a higher degree of risk aversion than shareholders or managers (Faleye, Mehrotra, and Morck, 2006). Unionized workers can play a stronger monitoring role in constraining managers excess risk-taking behaviors. Both the firms strategic responses to unions rent-seeking abilities and unions monitoring role over the firms risk-taking incentives will eventually have a nontrivial impact on the firms future stock price crash risk. 2.1 Firms strategic responses to unions rent-extracting advantage Theoretical and empirical evidence suggests that unionized employees are able to obtain higher wages and other better benefits through collective bargaining with their employer firms and the threat of strike (Grout, 1984; Lewis, 1986). Employment contracts established through collective bargaining agreements are not only associated with higher labor costs, but also make wages more sticky and layoffs more costly. Such great ability of unionized labors to extract above-market rents restricts the firms operating flexibility (Chen, Kacperczyk, and Ortiz-Molina, 2011), reduces the firms profitability and shareholder values (Abowd, 1989; Hirsch, 1991). A party s ability to increase its share of the firms economic rents is directly related to its bargaining power. Firms thus take strategic actions to improve their bargaining power vis-à-vis labor unions to reduce the impact of unionization. For instance, unionized firms issue additional debt and thereby 6

7 increase the demands on their cash flow so that they can credibly take a tougher stand against unionized workers (Bronars and Deere, 1991; Matsa, 2010). Klasa, Maxwell, and Ortiz-Molina (2009) find that unionized firms strategically hold less cash to improve bargaining position over unions by credibly signaling that additional concessions to unions would exacerbate the threat of potential liquidity shortages to competitive viability. Most importantly, labor unions need the firms financial and nonfinancial information to formulate and substantiate their bargaining strategies (Brown, 2000). Firms can strategically use the information communication policies to project a negative picture as possible about the likelihood of affording labor s wage demands. Specifically, firms facing unionized labor tend to withhold good news and promote bad news to mitigate unionized employees bargaining advantage. They are more willing to take on conservative accounting practices and manage earnings downward (DeAngelo and DeAngelo, 1991; Bowen, DuCharme, and Shores, 1995; D Souza, Jacob, and Ramesh, 2001; Leung, Li, and Rui, 2012). They are more likely to miss analysts forecasts because they make less effort to guide forecasts downward when expectations are too high, but take more actions to deflate earnings when estimates are too low. Such tendency of unionized firms to miss analysts forecasts is present at all times rather than just prior to labor negotiations (Bova, 2013). Apart from taking various strategic actions to improve bargaining position against unions, unionized firms also adjust their preference for risky investments than otherwise. Labor unions are rent-seekers, and they exercise their bargaining power to extract quasi-rents derived from risky investments (Grout, 1984; Salinger, 1984). To the extent that labor union s rent-seeking behaviors reduce the returns to high risk investments, unionized firms would avoid some risky investments that nonunionized firms would undertake. For example, unionized firms are found to invest less in intangible capital and innovative activity (Connolly, Hirsch, and Hirschey, 1986; Hirsch and Link, 1987), and engage less in even value increasing aggressive tax activities (Chyz, et al., 2013). As a result, both the level and riskiness of firms investments are lower in the presence of stronger labor unions (Hirsch, 1992; Fallick and Hassett, 1999). 2.2 Monitoring role of labor unions 7

8 Employees contractual claims on firm value resemble the payoff to risky debt in that they consist of a fixed claim (current and retired labor s stream of promised wages and benefits) less a put option (whose exercise price is the expected value of labor s claim in bankruptcy). Influential employees would like the firms to choose policies to maximize the value of their fixed contractual claims less that of the put option (Faleye, Mehrotra, and Morck, 2006). All else equal, the value of the put option increases with the variation in cash flows during current labor s horizon. Risky investment projects may shift wealth from fixed claimants to residual claimants such as shareholders and managers (Fama, 1990). Furthermore, employees may lose the value of their firm-specific human capital upon bankruptcy. Better coordination among unionized employees leads to a more effective communication with management and allows employees to exert more powerful pressure on management (Freeman and Medoff, 1984; Gray, Myers, and Myers, 1999). Therefore, labor unions have both stronger incentives and greater abilities to constrain the firms from high risk investments regardless of their present values. Like debt-holders, they also likely exert efforts to constrain managers from overstating earnings and assets for compensations they are not entitled to. In addition, workers demand higher wages, additional benefits, and improved working conditions to compensate for unemployment risk (Abowd and Ashenfelter, 1981; Topel, 1984; Li, 1986; Hamermesh and Wolfe, 1990), which increases with the riskiness of future cash flows. To the extent that all stakeholders suffer from adverse consequences associated with risky investment projects, unionized employees likely suffer more because their locked-in firm specific human capital investment and much longer time horizon in firms than that of mangers or shareholders who can quit the firms more easily. Unionized workers would require additional compensation if they feel cash flows are risker due to the firms investment in risky projects. Such compensation premium increases the marginal cost of risky investments and decreases the returns accrued to shareholders and managers. As a response, unionized firms could rationally avoid risky investments so as to reduce the need to compensate unionized employees for their perception of unemployment risk. 2.3 Hypotheses about the impact of labor union on future stock price crash risk 8

9 Corporate managers are inclined to strategically delay the disclosure of bad news and accelerate the release of good news due to various motivations such as formal compensation contracts, career concerns, and empire building ambitions (Kothari, Shu, and Wysocki, 2009). If managers withhold and accumulate bad news for an extended period, the firms stock price will be overvalued and severely deviated from their fundamental value. Once the accumulated negative information reaches a critical threshold that managers cannot afford to keep hoarding, it will be released to public all at once, resulting in a stock price crash-a large negative outlier in the distribution of returns (Jin and Myers, 2006; Hutton, Marcus, and Tehranian, 2009). Moreover, withholding bad news allows managers to camouflage poor performance of bad projects and prevents investors from timely discerning bad projects and taking corrective actions at an early stage. Consequently, the poor performance of bad projects accumulates over longer time than otherwise and will eventually surface, which leads to asset price crashes (Bleck and Liu, 2007; Benmelech, Kandel, and Veronesi, 2010). Based on the impact of labor union on firms operations as discussed previously, this study predicts that labor unions reduce crash risk for the following reasons. First, employee unionization is associated with raising wages and imposing other costs like constrained operating flexibility on employers. Corporate managers tend to take advantage of information disclosure strategies with the intention to reduce labor union s bargaining position. They strive to intimate an unfavorable prospect to their unionized workers by adopting conservative financial reporting policies (D Souza, Jacob, and Ramesh, 2001; Leung, Li, and Rui, 2012), downward manipulating profitability signals (Bava, 2013), promoting bad news and withholding good news (Chung, et al., 2014). Such asymmetric disclosure strategies would to a great extent offset the managerial tendency to stockpile negative information and make bad news flow into the market more quickly than unverifiable good news, which reduces future stock price crashes (Kim and Zhang, 2014). Second, due to relative fixed contractual claims, employees have a higher degree of risk aversion and prefer less risk than shareholders or managers (Faleye, Mehrotra, and Morck, 2006). Unionized employees are better able to deter managers from undertaking risky investment projects from which they 9

10 likely suffer losses but cannot gain much. They also play a more influential monitoring role in constraining mangers cover-up of poor performance of bad projects and promoting the timely discontinuation of bad projects. This prevents the bad performance of bad projects from accumulating and reduces the likelihood of asset price crashes (Bleck and Liu, 2007). Finally, labor union s superior rent-seeking ability and compensation premium demand for the perception of increased unemployment risk reduce the expected returns on risky investments. This will substantially reduce the attractiveness and viability of certain risky investments. Corporate managers could rationally avoid such risky investments in the first place and don t need to camouflage potential poor performance of those projects in subsequent periods. The ex ant avoidance of risky projects lowers the likelihood of accumulation of both poor performance and associated bad news of bad projects, and in turn reduces crash risk. Taken together, the above discussions lead to our first hypothesis: Hypothesis 1. Firms facing stronger unions exhibit lower future stock price crash risk. Corporate managers promote the release of bad news with the intention to gain concessions in negotiations with labor unions, which results in the negative relation between labor unionization and future stock crashes. Such negative relation is expected to be more pronounced for firms that the disclosed negative information provides more credible signal that the firm cannot concede to unions demands. DeAnglo and DeAngelo (1991) find that firms obtain concessions from unions by cutting dividends that convinces rank-and-file union members that shareholders themselves are forced to make sacrifices to alleviate the firm s financial constraints. Klasa, Maxwell, and Ortiz-Molina (2009) show that smaller cash reserves are more credible evidence about the firm s inability to meet union s demands for firms that are more likely to be financially distressed, strengthening the negative relation between cash holdings and unionization. Along the same lines, we formulate the following hypothesis: 10

11 Hypothesis 2. The negative relation between unionization and future stock crashes is stronger in situations where the unfavorable outlook intimated by the firms is more credible. Labor unionization introduces higher labor costs and reduces a firm s profitability and equity values (Abowd, 1989; Hirsch, 1991). The efforts to dampen the adverse impact of labor unionization will be in the interest of shareholders. Better governance and external monitoring lead to a better alignment of interest between managers and shareholders, and encourage managers to weaken labor union s bargaining position. Thus, the impact of worker unionization on future stock price crash risk should be more pronounced for firms with stronger external investor monitoring. This leads to our last hypothesis: Hypothesis 3. The negative relation between unionization and future stock crashes is stronger for firms with better external shareholder monitoring. 3. Sample and research design 3.1. Sample and data source Our initial sample is drawn from the intersection of data from the Center for Research in Security Prices (CRSP) and COMPUSTAT for the period Following prior literature (e.g. Kim, Li, and Zhang, 2011a, 2011b; Kim, Li, and Li, 2014), we use weekly stock return data to construct crash risk measures. Specifically, we assign weekly stock returns to each firm s fiscal year so as to match the time period of its reported financial data. Consistent with prior literature, we use industry unionization rate as the proxy for union strength and bargaining power (Connolly, Hirsch, and Hirschey, 1986; Bronars and Deere, 1991; Klasa, Maxwell, and Ortiz-Molina, 2009; Chen, Kacperczyk, and Ortiz-Molina, 2011; Chen and Chen, 2013). We obtain annual industry unionization data for the period from the Union Membership and Coverage Database at maintained by Barry Hirsch and David Macpherson. 11

12 We exclude firm-years with non-positive book values and total assets, firm-years with fiscal year-end prices of less than $1, firm-years with fewer than 26 weeks of stock return data, financial services firms and utilities, and firm-years without sufficient data to calculate control variables. Finally, we are left with a sample of 48,020 firm-year observations. 3.2 Measuring firm-specific crash risk This study employs two measures of firm-specific crash risk as constructed in Chen, Hong, and Stein (2001). Both measures are based on firm-specific weekly returns estimated as the residuals from the expanded market model. Specifically, we calculate the firm-specific weekly return, denoted by W, as the natural logarithm of one plus the residual return from the following expanded market model regression r j,τ =α j +β 1j r m,τ-2 + β 2j r m,τ-1 + β 3j r m,τ + β 4j r m,τ+1 + β 5j r m,τ+2 +ε j,τ (1) where r j,τ is the return on stock j in week τ, and r m,τ is the return on the CRSP value-weighted market index in week τ. The lead and lag terms for the market index return are included to allow for nonsynchronous trading (Dimson, 1979). The firm-specific weekly return for firm j in week τ, W jτ, is defined as the natural logarithm of one plus the residual return from Eq. (1). Our first measure of crash risk is the negative conditional return skewness (NCSKEW). NCSKEW for a given firm in a fiscal year is calculated by taking the negative of the third moment of firm-specific weekly returns for each year and normalizing it by the standard deviation of firm-specific weekly returns raised to the third power. Specifically, for each firm j in year t, NCSKEW is calculated as: NCSKEW jt =-[n(n-1) 3/2 W 3 jτ ]/[(n-1)(n-2)( W 2 jτ ) 3/2 ] (2) Our second measure of crash risk is the down-to-up volatility measure (DUVOL) of crash likelihood. For each firm j over a fiscal-year period t, we first separate all the weeks with firm-specific weekly returns into two groups: down weeks if returns are below the annual mean, and up weeks if returns above the annual mean. We then calculate the standard deviation for each of the two groups separately. The DUVOL measure is the natural logarithm of the ratio of the standard deviation in the down weeks to that in the up weeks. 3.3 Research design 12

13 To examine the relation between unionization and future stock price crash risk, we estimate the following cross-sectional regression: NCSKEW jt /DUVOL jt = α 0 +β 1 UNION jt-1 + β 2 DTURN jt-1 + β 3 NCSKEW jt-1 + β 4 SIGMA jt-1 + β 5 RET jt-1 + β 6 SIZE jt-1 + β 7 MB jt-1 + β 8 LEV jt-1 + β 9 ROA jt-1 + β 10 AVGACC jt-1 + IndustryDummies+YearDummies + ε t (3) where NCSKEW jt is the negative skewness of firm-specific weekly returns; DUVOL jt is the down-to-up volatility of firm-specific weekly returns; UNION jt-1 is our proxy for union coverage. Following previous work, we define UNION as the percentage of employed workers in a firm s primary Census Industry Classification (CIC) industry covered by unions in collective bargaining with employers. The set of control variables are drawn from prior literature (e.g., Chen, Hong, and Stein, 2001; Hutton, Marcus, and Tehranian, 2009). The variable DTURN jt-1 is the detrended average monthly stock turnover in year t-1, and included to control for differences of opinion among investors. The variable NCSKEW jt-1 is the negative skewness of prior year firm-specific weekly stock returns. We include it to capture the potential persistence of the third moment of stock returns. We include the variable SIGMA jt-1, the standard deviation of firm-specific weekly stock returns over the prior year, as more volatile stocks are likely to be more crash prone. The variable RET jt-1 is the mean firm-specific weekly returns over the fiscal year t-1 since past returns help to forecast crash risk. The variable SIZE jt-1 is defined as the log value of the market value of equity at the end of year t-1. The variable MB jt-1 is the market value of equity divided by the book value at the end of year t-1. The variable LEV jt-1 is the total long-term debts divided by total assets, and the variable ROA jt-1 is income before extraordinary items divided by lagged total assets. The variable AVGACC jt-1 is the average absolute discretionary accruals over the past three years, where discretionary accruals are estimated from the modified Jones model (Dechow, Sloan, and Sweeney, 1995) by each year and each 2-digit SIC code industry. We include AVGACC jt-1 to control for the impact of earnings management on future crash risk (Hutton, Marcus, and Tehranian, 2009). We finally control for major NAICS industry and year fixed effects. 4. Empirical results 13

14 4.1 Descriptive statistics Panel A of Table 1 reports summary statistics. The mean and median NCSKEW are and , respectively, similar to that reported by Kim, Li, and Zhang (2011a). The average value of DUVOL is , with a standard deviation of The average unionization rate (UNION) of 12.3% is higher than the median value of 8.8%, suggesting a slightly right-skewed distribution similar to that shown in Chen, Kacperczyk, and Ortiz-Molina (2011). The distribution of UNION is also quite dispersed, with a standard deviation of 11.7%. The distributions of other variables are largely similar to those reported in prior studies. [Insert Table 1 around here] Panel B of Table 1 presents the Pearson correlations among crash risk measures and control variables. The two crash risk measures (i.e., NCSKEW and DUVOL) are highly correlated, with a ratio of 0.886, quite similar to that in Chen, Hong, and Stein (2001). They appear to be picking up much the same information even though they are quite different in the construction. Both crash risk measures are negatively correlated with the unionization rate (UNION), which is consistent with our predictions that firms facing stronger unions exhibit lower future crash risk. 4.2 Test of Hypothesis 1 Hypothesis 1 predicts that labor unionization is negatively related to future crash risk because it discourages managers bad news hoarding and risk-taking behaviors in various ways. Table 2 reports results from regression analyses controlling for other potential predictors of crash risk. All reported t- values are based on robust standard errors corrected for firm and year clustering to alleviate concern about potential cross-sectional and time-series dependence in the data (Petersen, 2009). [Insert Table 2 around here] The results indicate that the unionization rate is negatively associated with future stock price crash risk proxied by NCSKEW and DUVOL. Specifically, Column (1) shows that NCSKEW is significantly and negatively associated with UNION. All else equal, an increase of one standard deviation in unionization 14

15 rate in year t-1 leads to a decrease of in NCSKEW, about 40% of sample mean. Column (2) reports a significant and negative association between DUVOL and UNION as well. All else equal, an increase of one standard deviation in unionization rate in year t-1 reduces DUVOL by 0.012, about 31% of sample mean. Thus, the impact of labor unionization on future crash risk is both statistically and economically significant. The coefficients of all control variables except ROA are generally consistent with previous studies (e.g., Chen, Hong, and Stein, 2001; Kim, Li, and Zhang, 2011b). The differences of opinion among investors (DTURN t-1 ), past negative return skewness (NCSKEW t-1 ), past return volatility (SIGMA t-1 ), past returns (RET t-1 ), firm size (SIZE t-1 ), and market-to-book ratio (MB t-1 ) are all positively related to future crash risk, while financial leverage (LEV t-1 ) is negatively associated with crash risk. The coefficient on discretionary accruals (AVGACC t-1 ) is positive but insignificant in both regressions. Overall, the results in Table 2 strongly support our Hypothesis 1 that firms facing stronger unions exhibit lower future stock price crash risk. These findings are consistent with the notion that in the presence of labor unionization, corporate managers are discouraged from undertaking risky investment projects and from hoarding bad news, leading to lower future stock crash risk. 4.3 Test of Hypothesis 2 If the negative impact of labor unionization on future stock crash risk is driven by managers strategic responses to improve their bargaining position against unions, this relation should be more pronounced in situations where the unfavorable outlook derived from managers strategic actions is more credible evidence that firms cannot concede to unions demands, as predicted by Hypothesis 2. We examine this issue using empirical models identical to those in Table 2, but apply them to the separate subsamples divided in terms of the magnitude of factors that affect the credibility of managers strategic responses. DeAngelo and DeAngelo (1991) show that the sacrifices made by shareholders themselves through dividend cuts enhance the credibility of the firm s financial difficulties and enable firms to obtain concessions from unions. We expect that the negative signals conveyed by a dividend-paying firm are less credible evidence that it is unable to comply with union demands. Based on whether a firm disburses cash 15

16 dividends or not, we partition all observations into dividend-paying subsample and no-dividend-paying subsample and re-estimate regressions in Table 2 for these two subsamples respectively. Results in Table 3 show that the unionization rate (UNION) is negative in all regressions but statistically significant only for the no-dividend-paying subsample regardless of which crash risk measure used. Moreover, the magnitude of UNION is significantly higher for no-dividend-paying subsample than for dividend-paying subsample when NCSKEW is used as proxy for crash risk. Thus, the negative relation between unionization rate and future crash risk is more pronounced for firms that do not pay cash dividends. [Insert Table 3 around here] It is also likely that negative outlook intimated by firms closer to financial distress provides more credible evidence about the firms inability to comply with union demands. We divide all observations into two groups based on whether a firm s Altman-Z score is above the year median. A higher value of Altman-Z score indicates that a firm is farther away from financial distress and faces less bankruptcy risk. The results for those subsamples in Table 4 show that the unionization rate (UNION) is significantly negative only for firms with lower Altman-Z scores with either of crash risk measures used. Additionally, the magnitude of UNION is significantly higher for firms with lower Altman-Z scores than for those with higher Altman-Z values no matter which crash risk measure is used. This suggests that the impact of unionization on future crash risk is stronger for firms that are closer to financial distress. [Insert Table 4 around here] Finally, negative information released by firms with higher profitability, defined as operating income divided by total assets, is less likely to convince labor unions that the firms are unable to comply with their demands. Those firms might thus have less incentive to promote the disclosure of bad news, leading to a weaker negative relation between unionization rate and future crash risk. To examine this issue, we classify all observations into two subsamples based on whether a firm s profitability is higher than industry-year median. The regression results for those two subsamples in Table 5 indicate that the unionization rate (UNION) is significantly negative only for firms with lower profitability irrespective of 16

17 the crash risk measure used. Furthermore, the coefficient of UNION is significantly higher in magnitude for firms with lower profitability than for those with higher profitability. It suggests that the impact of unionization on future crash risk is more pronounced for firms that have better profitability. [Insert Table 5 around here] In sum, the results in Table 3-5 support Hypothesis 2 that the negative relation between unionization rate and future crash risk is more pronounced in situations where strategic actions undertaken by corporate managers provides more credible outlook that the firms cannot concede to union demands. These findings corroborate those in Klasa, Maxwell, and Ortiz-Molina (2009) that unionized firms are more likely to hold lower cash reserve if smaller cash holdings are more credible evidence that the firm cannot concede to union demands. 4.4 Test of Hypothesis 3 Labor unions impose higher and stickier labor-related costs on firms, which impairs long-term firm value. To the extent that managers strategic actions such as promoting bad news to reduce union s bargaining power are in the interest of shareholders, the negative relation between labor unionization and future stock crash risk should be more pronounced for firms with effective external monitoring, as predicted by Hypothesis 3. To examine this issue, we focus on two proxies for external monitoring: the institutional ownership and G-index. Institutional investors are more sophisticated than individual investors and play a more effective governance role in ensuring managers acting in the interest of shareholders. We divide all observations into either a low institutional ownership group or a high institutional ownership group based on whether the level of a firm s institutional ownership is above sample median. As predicted, the regression results for those two groups in Table 6 show that the unionization rate (UNION) is significantly negative only for firms with higher level of institutional shareholdings regardless of the crash risk measure used. It suggests that the negative relation between unionization and future crash risk is most evident among firms with the high level of institutional ownership. 17

18 [Insert Table 6 around here] The other measure we use to indicate the external shareholder monitoring is G-index constructed in Gompers, Ishii, and Metrick (2003). G-index is defined as the number of anti-takeover provisions for a firm. Higher G-index suggests lower shareholder rights and weaker external monitoring in that more antitakeover provisions insulate a firm s management from takeover threats from the corporate control market. We group all observations into two categories, one with higher G-index and the other with lower G-index, based on whether a firm s G-index is above the sample median. The regression results for the two separate subsamples are presented in Table 7. When using the lower G-index subsample, the unionization rate (UNION) is negative and significant in the regression with NCSKEW as dependent variable, while marginally significant in the regression with DUVOL as dependent variable. However, the unionization rate (UNION) turns positive and never significant in both regressions using the higher G- index subsample. It suggests that the negative relation between unionization and future crash risk is most evident for firms with stronger shareholder rights. [Insert Table 7 around here] Taken together, the results in Tables 6-7 strongly support our second hypothesis that the effective external monitoring from shareholders strengthen the negative relation between unionization and future crash risk. Our findings are consistent with the notion that corporate managers strategic actions to reduce unions bargaining advantages are value accretive to firms and corporate managers are more likely to take such actions in the presence of stronger external shareholder monitoring. 5. Additional analyses and robustness checks 5.1. Endogeneity issues It is not random that unions choose industry to unionize workers. Unions might be more likely to form in industries that are less risky and more likely to survive in the long run if the fixed costs to unionized workers are large. If this is the case, the previously estimated impact of unionization rate on future crash risk would be biased upward. On the other hand, workers employment and working conditions are more 18

19 likely to be threatened by uncertain economic conditions inherent in riskier industries. They thus have a higher demand for unionization to protect themselves from uncertainties. If this is true, the impact of unionization rate on future crash risk would be biased downward because of unions self-selection to riskier industries. To address such endogeneity concerns, we use a two-stage least squares (2SLS) regression to identify the impact of unionization rate on future crash risk. Following Chen, Kacperczyk, and Ortiz-Molina (2010, 2011), we use the fraction of female workers (FEMALE) in the firm s CIC industry as the instrument for UNION. The labor economics suggests that female workers are less likely to unionize than male workers because they typically have less attachment to the labor market and to specific internal job ladders (Hirsch, 1980). Moreover, the expected benefits from being a union member might be smaller but the costs of unionizing may be higher for female workers than for male workers. Thus, we expect a negative relation between FEMALE and UNION. Meanwhile, neither theory nor evidence suggests that the female workers will directly affect a firm s future crash risk. The data on the fraction of female workers comes from the Census Population Survey conducted by the Bureau of Census for the Bureau of Labor Statistics. The data are available only for the period and the period on the website In the first stage regression, we regress UNION on the instrument FEMALE along with SIGMA, RET, SIZE, MB, LEV, ROA, AVGACC, and industry and year fixed effects. In the second stage regressions, we separately regress our two crash risk variables on the predicted value (HATUNION) of UNION from the first stage regression and all control variables used in Table 2. [Insert Table 8 around here] Table 8 reports the 2SLS results. The first-stage results in Column (1) show that the predicted unionization rate is significantly and negatively related to FEMALE, as predicted. The second-stage results in Columns (2) and (3) show that the instrumented unionization rate remains statistically significant and negative in regressions with each of crash risk measures used. In addition, the coefficient 19

20 of predicted unionization rate is much larger in magnitude than its counterpart in Table 2 estimated from the OLS regressions, suggesting that any potential endogeneity problem may only biases the magnitude of the coefficient downward. Overall, the 2SLS results support that the negative relation between unionization rate and future stock crash risk is not driven by self-selection issues The effect of labor intensity Prior literature suggests that unionized firms that attach more importance to gaining a bargaining advantage over unions are more likely to take strategic actions like lower cash reserve against unions (Klasa, Maxwell, and Ortiz-Molina, 2009). A strong bargaining position would be more valuable for firms with high labor intensity since they are affected more intensely by labor unions. One may expect that the impact of unionization rate on future crash risk should be stronger for firms with a higher level of labor intensity. Following Bowen, DuCharme, and Shores (1995), we measure labor intensity as one minus the ratio of gross property, plant, and equipment to total assets and average it across the current and the prior two years. We divide all observations into higher labor intensity group and lower labor intensity group based on whether a firm s labor intensity is above sample median, and then reestimate the regressions in Table 2 using each subsample, respectively. [Insert Table 9 around here] The results in Table 9 show that the unionization rate (UNION) is negative in all regressions but statistically significant only in regressions using higher labor intensity subsample regardless which proxy for crash risk is used as dependent variable. Additionally, the coefficient of UNION is significantly larger in magnitude from regressions using higher labor intensity subsample than that from regressions using lower labor intensity subsample. Thus, the results in Table 9 confirm that the negative relation between unionization rate and future crash risk is more pronounced for firms with higher labor intensity, consistent with the proposition that such negative relation is to some extent driven by the attempts by unionized firms to gain bargaining advantages over their unionized employees The effect of profit variability 20

21 Matsa (2010) argue that unionized firms with more variable profits have stronger incentives to reduce unions bargaining power because greater profit variability exposes more rents to union capture when future cash flows are high. For instance, firms with greater variability of future cash flows tend to carry more debt in the presence of powerful unions, which reduces their inframarginal wage costs. Following Matsa (2010), we measure profit variability as the standard deviation of the change in earnings before depreciation and amortization, divided by lagged total assets. We then divide all observations into higher profit variability group and lower profit variability group based on whether a firm s profit variability is above sample median, and reestimate the regressions in Table 2 using each subsample, respectively. [Insert Table 10 around here] The results in Table 10 show that the unionization rate (UNION) is negative and statistically significant in all but one regression. More importantly, the coefficient of UNION is significantly larger in magnitude from regressions using higher profit variability subsample than that from regressions using lower profit variability subsample regardless which crash risk measure is used. It suggests that unionized firms with high profit variability are more inclined to take strategic actions to improve their bargaining position over unions, which in turn lowers their future stock price crash risk Alternative proxy for union strength and crash risk Prior studies also use the likelihood of extreme negative returns to measure crash risk (e.g., Hutton, Marcus, and Tehranian, 2009; Kim, Li, and Zhang, 2011a, 2011b; Kim and Zhang, 2014). To check the robustness of our main findings to alternative measure of crash risk, we introduce such measure of crash likelihood in place of the two measures based on the conditional skewness of return distribution. Specifically, we first define crash weeks in a given fiscal year for a given firm as those weeks during which the firm experiences firm-specific weekly returns 3.2 deviations below the mean firm-specific weekly returns over the entire fiscal year. Then we create the measure of crash likelihood, CRASH, as an indicator variable that equals one for a firm-year that experiences one or more crash weeks during the fiscal-year period, and zero otherwise. The results from the logit regression with CRASH as dependent 21

22 variable in Column (1) of Table 11 show that the coefficient of the unionization rate (UNION) is significantly negative as well. [Insert Table 11 around here] The Union Membership and Coverage Database, available at provides labor union membership and coverage estimates not only by detailed industry but also by state. In previous analyses, we use industry union coverage estimates to proxy for union strength. In this subsection, we take state union coverage estimates into account and construct a finer proxy for union strength (UNION_SI) by multiplying union coverage rate in a firm s industry with that in the state where the firm is incorporated. The results using this alternative measure of union strength is reported in Columns (2) and (3) of Table 11. The coefficient of our new union strength measure is still negative and statistically significant in both regressions using different dependent variables. We also conduct other robust checks. Specifically, we re-estimate the regressions in Table 2 by dividing the sample into manufacturing (SIC code from 2000 to 3999) and non-manufacturing firms, since some studies on labor union focus on the manufacturing sectors (e.g., Klasa, Maxwell, and Ortiz- Molina, 2009; Matsa, 2010). The unreported results for brevity show that the negative relation between unionization rate and future crash risk appears in both manufacturing and non-manufacturing industries. Finally, we separate our sample into three different time periods, , , and , to check whether our results are stable over time. The unreported results show that the negative relation between unionization rate and future crash risk remains in all the three time periods. 6. Conclusions Strong labor unions impose higher labor-related costs upon unionized firms and have greater ability to monitor managers actions. On the one hand, firm managers tend to take strategic actions, for example, promote the release of bad news and intimate an unfavorable outlook, so as to reduce unions bargaining 22

23 position. On the other hand, labor unions risk-averse nature and rent-capturing power discourage firm managers from undertaking high risk projects. As a result, both mechanisms through which labor unions affect a firm s operations are expected to have mitigating effect on the firm s future stock price crash risk. This study provides evidence that strategic considerations and monitoring effect that arise in the interactions between a firm and its unionized employees do affect the firm s future crash risk. Specifically, we find a negative relation between unionization and future crash risk. This negative relation is more pronounced in situations that firms strategic responses make a more credible case that a firm is unable to concede to union demands. This negative relation is also more pronounced for firms with stronger external governance that better encourage managers to behave in the interest of shareholders. Our findings suggest that labor unionization might deters managerial bad news hoarding behavior and excess risk-seeking behavior, leading to lower future crash risk. 23