Executive pensions, risk-shifting, and foreign exchange exposure

Accepted Manuscript Title: Executive Pensions, Risk-Shifting, and Foreign Exchange Exposure Author: Alain A. Krapl Reilly S. White PII: DOI: Reference:

S0275-5319(16)30086-1 http://dx.doi.org/doi:10.1016/j.ribaf.2016.05.001 RIBAF 526

To appear in:

Research in International Business and Finance

Received date: Accepted date:

5-5-2016 13-5-2016

Please cite this article as: Krapl, Alain A., White, Reilly S., Executive Pensions, Risk-Shifting, and Foreign Exchange Exposure.Research in International Business and Finance http://dx.doi.org/10.1016/j.ribaf.2016.05.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Executive Pensions, Risk-Shifting, and Foreign Exchange Exposure

Alain A. Krapl*, Reilly S. White **

This version: May 2016

*

(Corresponding Author), Northern Kentucky University, Haile/US Bank College of Business, Nunn Drive, Highland Heights, KY 41099, [email protected] **

University of New Mexico, Anderson School of Business, 1 University of New Mexico, Albuquerque, NM 87131, [email protected]

Highlights:

   

We analyze the effects of executive pension–based compensation on FX exposures We study pension plans for 272 of the largest U.S. firms over a ten-year period The relation between pension-based compensation and FX exposures is negative Results are strongest when sample firms are close to financial distress

ABSTRACT Using a hand-collected executive compensation database of 272 large U.S. firms from 2000 to 2009, we present the first study of its kind to analyze the effects of executive pension–based compensation on foreign exchange exposure. We find evidence that higher executive pension compensation results in lower foreign exchange exposure among our sample firms, an effect that is strongest when sample firms are closest to bankruptcy. Our results have important implications for the structuring of managerial compensation contracts for multinational firms. JEL Classifications: G32, G38, G15 Keywords: Foreign Exchange Exposure; Executive Compensation; Pensions; Managerial Risk-Taking

1. Introduction Value-maximizing firms are motivated to manage foreign exchange (FX) exposure in order to reduce risk. Several factors determine the level of risk management undertaken by a firm, including impending financial distress, tax policy, and agency issues (e.g. Bessembinder, 1991; Leland, 1998; Smith and Stulz, 1985; Stulz, 1984; Tufano, 1998). Agency effects include the structure of contracts, risk aversion of managers, and wealth incentives. Although several papers have examined the link between compensation structure and hedging, to our knowledge no work has focused on debt-based pension compensation and its effects on corporate FX exposure. Using a hand-collected database and building on existing international and corporate agency literature, we examine the relation between pension-based compensation and firm-level FX exposure. The connection between managerial incentives and risk-taking has been explored in several papers. Notably, Smith and Stulz (1985) demonstrate that the level of managerial risk aversion and the structure of management contracts can be significant drivers of corporate hedging. Further, Rogers (2002) and Adkins et al. (2007), find that managerial incentives are an important determinant in corporate derivatives usage. However, managerial incentives were historically regarded as a matter of equity compensation; only with Sundaram and Yermack (2007) was pension-based compensation considered an important driver of aggregate firm risk. Building on agency literature extending to Jensen and Meckling (1976), Sundaram and Yermack (2007) posit that high proportions of debt-like compensation (such as pensions) reduce

Page 1 of 43

managerial risk-taking behavior. Inversely, greater risk-taking is observed when executive compensation schemes are equity-heavy. Combining the findings from both international finance and corporate finance literature, we posit that managers with high proportions of debt-like compensation are more likely to manage FX exposure aggressively, thereby reducing overall cash flow volatility and assuring the firm’s long-term survival. Although data limitations do not allow us to observe managerial risk-taking directly, we argue that pension-based managerial compensation is ultimately reflected in lower corporate FX exposures. Managerial capital budgeting decisions can also serve as a link between compensation and FX risk exposure. The papers closest to ours are Francis et al. (2013) and Belkhir and Boubaker (2013). Francis et al. (2013) use Delta and Vega measures of CEO compensation to study the relation between CEO compensation and FX exposure. Their study reports that a 4.5% (7.9%) increase in the sensitivity of CEO wealth to stock price (equity return volatility) results in a one standard deviation decrease in FX equity exposure. Using a sample of banks, Belkhir and Boubaker (2013) analyze the effects of pensions and deferred compensation on CEO hedging of interest rate risk. Their study identifies a positive relation between CEOs’ inside debt and the use of interest rate derivatives for hedging purposes. Our paper differs from the Belkhir and Boubaker (2013) and Francis et al. (2013) studies in several ways. First, we focus our analysis on the relation between pensions and FX exposure; FX exposure is consistently identified by survey results as a major source of risk that is commonly managed (Bodnar et al., 1996, 1998; Bodnar et al., 1995). Second, we expand our analysis to include other top executives of the company, not just CEOs. This inclusion is important because often the CFO and/or the Treasurer are more directly involved in risk

Page 2 of 43

management and capital budgeting decisions than the CEO. Third, hedging policies primarily focus on the reduction of the firm’s cash flow volatility. Although equity returns are commonly used as a cash flow proxy when estimating FX exposure, we estimate FX cash flow exposure using an accounting-based cash flow proxy. This alternative provides us with a more direct measure of FX cash flow exposure. In our empirical analysis we study pension plans for 272 of the largest U.S. firms over a ten-year period (from 2000 to 2009). We report two important results. First, we document a negative relation between pension-based executive compensation and firm-level FX exposures. This is consistent with a reduction in managerial risk-taking incentives, which would lower FX-related risk taking and ultimately be reflected in corporate FX exposures. Second, consistent with the notion that the risk-shifting problem is magnified in distressed firms (Eisdorfer, 2008), we document a stronger relation between pension-heavy compensation and FX exposure reduction in a subsample of financially distressed firms. The paper proceeds as follows. Section two provides a discussion of the hypotheses. Section three presents a description of the methodology and sample data used in this study. In section four we discuss the results and its implications. Section five concludes the study.

2. Hypotheses Typically, managers are compensated with a debt-like components (such as pensions and deferred compensation) and equity-like compensation (such as stock awards and stock options). Researchers observe that when the ratio of debt to equity components in the compensation structure of an executive (compensation leverage) is different from the debt to equity ratio of the

Page 3 of 43

firm as a whole (firm leverage), substantial disincentives can develop.1 Executives with high compensation leverage will have interests more closely aligned with the firm’s bondholders. Conversely, executives with low compensation leverage have interests more closely aligned with shareholders. In this paper, we study firm-level FX exposures that reflect managerial risk-taking with respect to FX rates and corporate international investment. We argue that pension-compensated managers, because of their increased incentive to reduce the firm’s cash flow volatility, will more aggressively manage FX exposure and/or make less risky capital budgeting decisions associated

with

corporate

international

diversification.

As

a

result,

firms

with

pension-compensated managers will have lower levels of FX exposure. We test our first hypothesis: H1: Firms with relatively larger executive pension plans will have lower levels of FX exposure. In our second hypothesis, we draw on prior literature covering agency theory, financial distress, and bankruptcy. The risk-shifting incentive increases for managers of financially distressed firms. 2 In such firms, equity holders are more inclined to take large gambles (Eisdorfer, 2008; Jensen and Meckling, 1976) due to relatively larger potential payoffs. From an FX exposure management perspective, equity-compensated managers are more likely to leave 1

Core and Guay (2002) and Coles et al. (2006) describe increased managerial risk-taking when executive compensation packages include high proportions of stock options. Sundaram and Yermack (2007) and Cassell et al. (2012) find that high levels of pension-based compensation have the opposite effect: managers are inclined to engage in less risky projects to preserve the firm’s value in the long run. Edmans and Liu (2011) show that during crises, firm compensation packages that are pension-heavy can preserve long-term firm value. We use the Moody’s KMV distance-to-default measure, which measures how many standard deviations in asset value would need to occur to generate firm bankruptcy (defined as the point at which a firm’s value falls below its short-term debt plus one-half of long-term debt value). Higher values indicate the firm is farther from bankruptcy; lower values indicate the firm is closer to it. 2

Page 4 of 43

FX exposure unhedged or even participate in FX speculation. Conversely, pension-compensated managers have an increased incentive to manage FX exposure when the firm is financially distressed due to their fear of losing their pensions. This leads to our second hypothesis: H2: The FX risk-reducing effect of pension-based compensation is particularly large for firms close to bankruptcy (as measured by distance-to-default).

3. Methodology and sample data 3.1. Compensation leverage Compensation leverage is the ratio of managers’ inside debt divided by the sum of inside equity and inside debt. Inside debt includes the actuarial present value of the managers’ pension entitlement3, whereas inside equity refers to both values of stock awards and the stock options held by the manager:

1 J 1 J

J

 Pension

J

 ( Pension

j

j 1

j

(1)

 Stocks j  Options j )

j 1

where J represents the number of top managers (typically three to five) in each firm in each year. The market value of a manager’s common equity is estimated by the number of shares held by the manager multiplied by the share price. To estimate the value of the unexercised stock-options held by the manager, we follow Core and Guay (2002) and Sundaram and Yermack (2007). Option values are estimated with the Black and Scholes (1973) model.

3

Deferred Compensation, arguably a component of inside debt, is omitted, due to qualifying limitations, variable firm disclosure, and precedent in prior literature (e.g., Sundaram and Yermack, 2007).

Page 5 of 43

Two steps are required to determine the price of unexercised stock options. First, using ExecuComp data, we compute the ratio of realizable in-the-money exercisable options to the number of unexercised exercisable options. Second, we estimate the exercise price by taking this ratio and subtracting it from the firm’s end-of-fiscal-year stock price. Following Sundaram and Yermack (2007), we set the maturity of all outstanding stock options to six years. We measure stock price volatility by the standard deviation of stock returns during the previous 60 months and take the dividend yield over a three-year period. We estimate dividend yield using the approach suggested by Fama and French (1988). Finally, we use the one-year Treasury Bill rate as the risk-free rate of return. For the computation of stock awards, we use end-of-year stock prices.

3.2. Pension actuarial value Under a traditionally defined benefit pension scheme, the IRS establishes the maximum annual disbursement for an employee as the lesser of the highest 3 years of annual salary or $205,0004. This is significantly below the average salary and bonus compensation of most executives. In response, many firms have established Supplemental Executive Retirement Plans (SERPs), to provide higher pension compensation than dictated under IRS guidelines. Roughly three-fourths of our sample firms maintain unfunded plans. Often, SERPs are prefunded by placing the value of a pension aside in an executive-specific trust (rabbi trust). Prior to July 2006, the SEC required that pension disclosure be in the form of a tabled matrix, matching average salaries and years of experience to approximate the annual pension

4

See http://www.irs.gov/Retirement-Plans for current information on pension plan limits.

Page 6 of 43

entitlement (see Table A1 in the Appendix). Firms were not required to disclose the actuarial present value of the benefit, and its calculation was left entirely up to the investors. However, firms with fiscal years ending December 2006 were required to present a table of actuarial present values. To keep our sample consistent throughout this period (2000-2009), we maintain a consistent calculation methodology (a guided example is provided in the Appendix). We use a set of 272 firms drawn from the 700 largest firms (based on equity market capitalization) over a ten-year period (2000-2009). In addition to including CEO-level data, we include all top firm executives in our computation of the firm’s aggregate level of inside debt. We define the present value of a pension annuity as: KA

 n n  max( 0 ,R  A ) ( 1  d ) P( n ) X

(2)

where X is the amount of the annual pension, A is the current age of the executive, R is the minimum retirement age to achieve full retirement benefit, K is the final year of the pension, and P(n) is the probability that the executive will be alive in n years. The Social Security Administration’s Period Life Table provides mortality probabilities based on the general population; using this table, we can determined the probability that executives of age A will be alive in n years. While it is possible an executive can receive a pension benefit indefinitely, the mortality projections of the Social Security Administration end at 119 years, so K is for practical purposes set at 120. The discount rate, d, is defined as the annualized Moody’s Seasoned Aaa bond-rating for a given year, taken from the Federal Reserve Board’s H.15 release.5 Since most of our firms are older and well-established, we find this bond rating is generally appropriate. Furthermore, firms

5

Information is taken from the FRB H.15 Tables

Page 7 of 43

that volunteered present-value data of pensions prior to 2006 used either the ten-year Treasury bond yield or the Aaa bond-rating for that year. The annual pension benefit (𝑋) is computed as follows: P

 k 1

Ct  k M S P

(3)

where Ct refers to the cash salary and bonus compensation to each executive for year t, P refers to the number of years of averaged compensation to determine pension entitlement, and S refers to the executive’s years of service. The average age of an executive and a CEO in our sample are 53 and 56, respectively (see Table 1). M is the amount of pension benefit earned per year of service. For most firms, this figure is between 1.5% and 2.0% of averaged compensation per year of service. The combination of these two equations produces the actuarial present value for the executive pension for that year. We differentiate between ‘CEO-Level’ and ‘Firm-Level’ pension size. ‘Firm-Level’ uses the highest-paid executives (typically five: the CFO, CIO, and other officers) to establish an aggregate pension size for our pensions/assets measure. 3.3. Measuring FX exposures We estimate firm-level FX equity exposures based on the approach introduced by Adler and Dumas (1984). To line up the exposure measures with our annual managerial compensation data from 2000 to 2009, we use 5-year rolling period windows. Thus, the estimation period of Eq. (4) for the end-of-year 2000 exposure measure is January 1996 to December 2000. We use OLS regression to estimate the following model:

Page 8 of 43

𝑅𝑖𝑡 = 𝛼𝑖 + 𝛾𝑖 𝑅𝑋𝑡 + 𝜀𝑖𝑡

(4)

where 𝑅𝑖𝑡 are monthly holding period stock returns of firm 𝑖, 𝑅𝑋𝑡 are the monthly returns of a trade-weighted basket of foreign currencies, and 𝛾𝑖 is the FX equity exposure of firm 𝑖 to changes in FX rates. Dealing with a relatively large cross section of firms without having access to information on which specific currencies each firm is exposed to, we choose the Federal Reserve’s Major Trading Partner Currency Index (MCI) for our study (e.g., Jorion, 1990; Wei and Starks, 2013). Although widely popular, using stock returns as a proxy for cash flows offers indirect estimates of FX cash flow exposure. Theoretically this can be problematic because FX rate changes can affect stock returns through two channels: a cash flow channel and/or a discount rate channel. To gain further insight into the effects of managerial risk-shifting on the cash flow volatility of the firm, we estimate a ‘direct’ FX cash flow exposure measure. Adapting the approach used by Bartram (2008) we estimate the following OLS regression model: 𝐵 𝐵 𝐸𝑃𝑆𝑖𝑡 − 𝐸𝑃𝑆𝑖𝑡−4 = 𝛼𝑖𝐵 + 𝜉𝑖𝐵 𝑅𝑋𝑡 + 𝜙1,𝑖 𝑅𝑆𝑇𝑡 + 𝜙2,𝑖 𝑅𝐷𝑆𝑡 + 𝜀𝑖𝑡𝐵

(5)

where 𝐸𝑃𝑆𝑖𝑡 are quarterly earnings per share for firm 𝑖; we use quarterly frequency data but annual horizons for changes in EPS to address potential seasonality in earnings data. 𝜉𝑖𝐵 is the FX cash flow exposure of firm 𝑖 based on a 5-Year rolling period estimation window. 𝑅𝑆𝑇𝑡 and 𝑅𝐷𝑆𝑡 are short-term interest rate and term-spread control variables that are defined as follows: 𝑅𝑆𝑇 = ∆𝑆𝑅⁄(1 + 𝐿𝑅) and 𝑅𝐷𝑆 = ∆ (𝐿𝑅 − 𝑆𝑅)⁄(1 + 𝐿𝑅) where ∆ denotes a quarterly change, 𝑆𝑅 is the short-rate (we use a 1-Year U.S. Treasury yield), and 𝐿𝑅 is the long-rate (we use a 10-Year U.S. Treasury yield).

Page 9 of 43

3.4. Pension-based compensation and FX exposure – the main model After merging FX exposure measures with firm-level executive compensation numbers, we have panel data with end-of-year observations from 2000 to 2009. To analyze the effects of executive compensation on FX equity and FX cash flow exposures, we estimate the following regression model: 𝐹𝑋𝑖𝑡 = 𝛽0 + 𝛽1 𝐶𝑜𝑚𝑝𝐿𝑒𝑣𝑖𝑡 + 𝛽2 𝑃𝑒𝑛𝑠𝐴𝑠𝑠𝑒𝑡𝑠𝑖𝑡 + 𝛽3 𝑆𝑎𝑙𝑆𝑐𝑎𝑙𝑒𝑑𝑖𝑡 + 𝛽4 𝑂𝑝𝑡𝑆𝑐𝑎𝑙𝑒𝑑𝑖𝑡 + 𝛽5 𝐹𝑖𝑟𝑚𝑠𝑖𝑧𝑒𝑖𝑡 + 𝛽6 𝐵𝑜𝑜𝑘𝑚𝑘𝑡𝑖𝑡 + 𝛽7 𝐷𝑒𝑏𝑡𝑒𝑞𝑖𝑡 + 𝛽8 𝐷𝑡𝑑𝑖𝑡 + 𝛽9 𝐿𝑖𝑞𝑐𝑜𝑛𝑠𝑖𝑡 9

+ 𝛽10 𝑇𝑎𝑥𝑖𝑡 + 𝛽11 𝐸𝑥𝑒𝑐𝑜𝑤𝑛𝑖𝑡 + 𝛽12 𝐹𝑜𝑟𝑠𝑎𝑙𝑒𝑠𝑖𝑡 + 𝛽13 𝑅𝑂𝐸𝑖𝑡 + ∑ 𝛿𝑦 𝐼𝑛𝑑𝑦 𝑦=1 9

(6)

+ ∑ 𝛿𝑗 𝑌𝑒𝑎𝑟𝑗 𝑗=1

where 𝐹𝑋𝑖𝑡 is the magnitude of FX exposure for firm 𝑖 at the end of year 𝑡. Firm-level compensation data includes industry-adjusted compensation leverage (CompLev), total pensions scaled by total assets (Pens/Assets), total salaries scaled by total assets (SalScaled), and total option value scaled by total assets (OptScaled). Several characteristics of a firm can affect its FX exposure. We control for Firmsize, which we define as the natural log of the firm’s market value of equity. Most studies identify a predominantly negative relation between the size of firms and their FX exposures (e.g., Bodnar and Wong, 2003; Dominguez and Tesar, 2006).6 Bookmkt is the market value of firms’ common shares divided by shareholders’ equity. We include Bookmkt to proxy for a firm’s growth 6

A notable exception is presented by He and Ng (1998) who find a positive relation between the size of Japanese multinationals and their FX exposures.

Page 10 of 43

opportunities (Géczy et al., 1997; He and Ng, 1998). According to optimal hedging theory, firms with more growth opportunities are more likely to hedge FX exposure in order to mitigate the underinvestment problem. Thus, we would expect firms with lower Bookmkt ratios to have lower FX exposures. Smith and Stulz (1985) posit that hedging can reduce a firm’s expected financial distress costs. Based on optimal hedging theory, firms with higher expected financial distress costs would have a higher incentive to hedge their FX exposure. Our study uses alternative measures to control for the probability of distress. Following He and Ng (1998), we include Debteq, which is computed by dividing the firms’ long-term debt by total equity. Alternatively we include Dtd, which is the distance-to-default measure based on Moody’s KMV Method. The inclusion of Dtd is also motivated by Wei and Starks (2013) who, in contrast to the arguments put forth by hedging theory, document a positive relation between financial distress risk and FX exposure. Similarly, Liqcons and Tax are dummy variables that capture whether the firm reported negative income or a net operating loss carryover during the current year. We include these measures along with ROE, which is the defined as the return on shareholder equity, to capture increased managerial hedging incentives. Firms that are less profitable or operate at a loss have increased incentives to hedge their FX exposures (He and Ng, 1998). Nevertheless, Wei and Starks (2013) argue that financially distressed firms have a diminished ability to hedge and thus remain more highly exposed to FX rate changes. Based on the work by Francis et al. (2013), we include Execown, which captures equity ownership of the CEO (as a proportion of all firm equity). Based on arguments put forth by Smith and Stulz (1985), CEO stock holdings can influence their approach to risk management. Finally, to capture the effects of a firm’s level of international activity (e.g., Allayannis and Ihrig, 2001; Page 11 of 43

Bodnar and Gentry, 1993; Jorion, 1990), we include Forsales, which is the ratio of firm’s foreign sales to total sales. Based on prior empirical results, we would expect that FX exposure increases with foreign sales ratios. In addition to the aforementioned control variables, we also include sets of dummy variables that control for industry7 (we sort firms by industry groups that are reported in Table 1) and year fixed effects.

3.5. Sample data Our data contains estimated executive pension values from 2000 to 2009. We examine the 700 largest firms by U.S. market capitalization on December 31, 2009. Out of these, 300 offered executive pensions (42%), while 290 (41%) provided values calculable under the Sundaram and Yermack (2007) framework. We omit firms with unclear compensation data, and executive structure. Company financials are obtained from Compustat, stock and market values are retrieved from the CRSP database, interest rate data is retrieved from the Federal Reserve’s H.15 tables, and FX data is obtained from the Federal Reserve’s H.10 tables. Our final sample includes 272 firms and 8,955 executive-year data points, consisting of 2,114 CEOs-years (23.6%) and 6,851 Non-CEO executive-years (76.4%). Non-CEO executives are combined in aggregate with the CEO data to generate firm-level compensation values. In Panel A of Table 1, we provide an overview of our collected sample’s executive and firm-level data. We use 1,581 executive firm-years to compute compensation leverage, and 1,684 executive firm-years for pensions scaled by asset size.

7

The role of industry structure and its effects on FX exposure is documented by Bodnar and Gentry (1993) and Williamson (2001).

Page 12 of 43

In Panel A of Table 1, we observe that CEOs in the sample are on average 56 years old, while non-CEO executives are on average 54 years of age. We also observe that our sample firms are larger than firms of the non-sample general population. This is not surprising and is consistently observed in prior studies. Average absolute value of FX equity exposure |𝛾𝑖 | is 0.554 and the average absolute value of FX cash flow exposure |𝜉𝑖𝐵 | is 2.193 for our sample. Distance-to-default (Dtd) is on average 2.096 standard deviations. In Panel B of Table 1, we divide the sample into 10 industries based on their two-digit SIC codes. Dividing our sample of 272 firms into industries is necessary for the industry-adjustment of our compensation data. Manufacturing firms dominate the overall sample with 130 firms (48%). In addition, 47 firms (17.3%) are in the Financial Sector, 46 (16.9%) in the Utility Sector, and 15 (5.5%) in the Mining Sector. To control for industry effects in the study, we establish the average compensation leverage (or pension/assets) for each sample industry and subtract it from individual firm compensation leverage. 8 In Table 2 we present a correlation matrix of our major variables. Here we observe that FX exposure magnitude (|𝛾|) is positively correlated with relative CEO salaries and CEO stock option values, CEO SalScaled and CEO OptScaled. On the other hand, the magnitude of FX cash flow exposure, |𝜉 𝐵 |, is negatively related to Pens/Assets, SalScaled, and OptScaled. [Insert Tables 1 and 2 approximately here]

4. Empirical results 4.1. Executive pensions and FX exposure (H1)

8

For example, the industry-adjusted compensation leverage for an executive reporting compensation leverage of 0.34 in an industry with an average of 0.30 is 0.34 – 0.30 = 0.04.

Page 13 of 43

We begin our analysis by estimating our main model specified in Eq. (6).

First, we

focus on the compensation schemes at the CEO-level; these results are reported in Table 3. Subsequently we expand our analysis to include the compensation schemes of several top executives. The firm-level results are presented in Table 4. In Table 3, for models [i] and [ii], the dependent variable is the absolute value of FX equity exposure (|𝛾𝑖 |). Models [iii] and [iv] use absolute values of direct FX cash flow exposure estimates (|𝜉𝑖𝐵 |) as the dependent variable. In models [i] and [iii] we measure the relative size of pension-based compensation as compensation leverage (CompLev) whereas models [ii] and [iv] use total pensions scaled by assets (Pens/Assets). We control for year and industry effects in all of our models. Heteroskedasticity (HC) and autorcorrelation (AC) robust standard errors are clustered by firm observations and reported in parentheses. Firms with pension-based compensated CEO’s have lower FX exposures. Both compensation leverage and relative pension size are negatively related to FX equity exposure. We also observe a negative relation between compensation leverage and FX cash flow exposure, which is consistent with the relation documented between CompLev and |𝛾𝑖 |. Moreover, results illustrate a negative relation between the relative size of CEO option holdings in the firm and FX cash flow exposure. This result is somewhat surprising, since it indicates that CEOs who hold a lot of stock options are more likely to hedge their companies’ FX exposures. Consistent with prior literature, results in Table 3 document a positive relation between foreign sales ratios and FX exposure. This indicates that either the firms in our sample are unable to manage their exposures in response to changes in FX rates, or that they choose not to do so. The negative relation between FX cash flow exposure and distance-to-default (in model [iv]) would be consistent with optimal hedging theory; firms with higher probabilities of financial

Page 14 of 43

distress have more incentive to hedge their FX exposures. Also, the negative relation between net operating loss carryover (Tax) and FX equity exposure suggests that firms in our sample are able to hedge FX exposure despite their financial constraints. This result is contrary to the findings reported by Wei and Starks (2013). It is likely that our result differs because, on average, our sample contains bigger firms. We believe that this is also the reason why there is no negative relation between firm size and FX exposure for our sample firms — a finding that is reported by several studies based on larger samples (e.g., Bodnar and Wong, 2003; Dominguez and Tesar, 2006). In Table 4 (at the firm-level), we observe that increasing pension-based managerial compensation reduces the FX equity exposure and FX cash flow exposure of the firm. Additionally, results illustrate a negative relation between the relative size of managerial option holdings in the firm and FX cash flow exposure but a positive relation between managerial option holdings and FX equity exposures. Other interesting results illustrated in Table 4 include the finding that FX exposures are lower for more profitable firms (firms with higher of ROE). Although optimal hedging theory would predict that more profitable firms have less incentive to hedge their FX exposures and thus should have higher FX exposures, more profitable firms are also more likely able to absorb FX shocks. Also, foreign sales ratios are positively correlated with FX cash flow exposures — an observation that is consistent with prior results documented in literature. [Insert Tables 3 and 4 approximately here] Our results could be affected by an endogeneity problem; for example, firms that choose to hedge FX exposure might also have more internal funds available, and therefore be able to allocate more cash to their pension plans. To address this potential problem, we follow Eisdorfer Page 15 of 43

et al. (2014) and identify two instrumental variables that are related to compensation leverage and inside debt, but not to FX exposure. The first variable is executive age, which is positively correlated with pension size. We use the CEO’s age (for the CEO-level analysis) and the average age of all executives (for the firm-level analysis). The second instrumental variable is M, a multiplicative factor that describes the ratio of pension benefits earned per dollar of compensation. Firms with higher M values allocate more money per dollar to pension benefits than those with lower M values. M is typically the same for the CEO and all executive measures, as all executives in a firm are usually under the same executive pension plan. We follow the procedure outlined by Baum et al. (2003) to confirm the appropriateness of our instrumental variables. We establish that our two instrumental variables are neither underidentified (based on the Kleibergen and Paap LM statistic) nor overidentified (based on the Hansen J Statistic), nor are they particularly weak (based on the Cragg and Donald test). Table 5 presents the CEO-level results of the 2-SLS model. We observe that three out of the four models confirm the negative relation between pension-based compensation and FX exposure of the firm. This is similar to the results reported for models [i] through [iii] in Table 3. However, based on the 2-SLS estimates, the relation between pension-based compensation and FX exposure, particularly FX cash flow exposure, is more pronounced compared to OLS estimates. We observe that the estimated slope coefficient of CEO Pens/Assets becomes more negative for model [iii] — it decreases from -0.030 to -0.596. Similarly, the estimated coefficient on CEO Pens/Assets decreases from -0.0761 to -3.272 by switching to the two-stage model. Also, the results based on compensation leverage (CEO CompLev) become more pronounced (the estimated effect decreases from -0.712 to -3.443). In contrast to the strengthening of results at the CEO-level (Table 5), using the 2-SLS models at the firm-level weakens results Page 16 of 43

significantly. Table 6 shows no statistically significant relation between pension-based compensation at the senior management-level and the firm’s FX equity and FX cash flow exposures. At the CEO-level (Table 5) the 2-SLS analysis confirms the positive relation between FX exposures and foreign sales ratios based on models [i] and [iii]. In contrast, estimates of model [ii] illustrate a positive relation between FX equity exposure and distance-to-default, which is consistent with observations reported by Wei and Starks (2013). Moreover, results presented in Table 5 confirm the negative relation between CEO OptScaled and FX cash flow exposures. At the firm-level (Table 6), the 2-SLS estimates confirm that FX exposures are lower for firms with higher of ROE values (based on models [ii] and [iv]). Table 6 also illustrates a negative relation between firm size and FX equity exposure, a result that has been documented by several earlier studies (e.g., Bodnar and Wong, 2003; Dominguez and Tesar, 2006). [Insert Tables 5 and 6 approximately here]

4.2. Pensions and FX exposure in financially distressed firms (H2) In this subsection we test our second hypothesis (H2) by studying the effect financial distress (as measured by distance-to-default, Dtd) on the relation between pension-based executive compensation and FX exposure. First, we divide our sample into four equally sized subsamples based on Dtd. High Dtd values indicate that firms are further from bankruptcy. Consistent with the rest of our paper, we perform our analysis at the CEO-level (Table 7) and

Page 17 of 43

firm-level (Table 8). For the sake of brevity we focus on absolute values of FX equity exposure, 𝛾𝑖 . At the CEO-level (Table 7), we observe that the negative relation between CEO pension-based compensation and FX equity exposure is primarily driven by firms that are closest to bankruptcy according to Dtd. Estimated coefficients on both pension compensation measures, CompLev and Pens/Assets, become increasingly less negative as firms become less likely to default. At the CEO-level, the negative relation between compensation leverage and FX equity exposure is statistically significant for firms in the lowest Dtd quartile. At the Firm-level (Table 8) this finding becomes even more pronounced. Here we observe that that the negative relation between CompLev and FX equity exposure as well as the negative relation between Pens/Assets and FX equity exposure are limited to the low Dtd quartiles. Indeed, for firms that are the least likely to default (highest Dtd quartile), managerial pension-based compensation, as measured by Pens/Assets, is positively related to FX equity exposure. Based on the main results reported in Tables 7 and 8, we conclude that the risk-mitigating effect of pension-based compensation is limited to firms that are subject to increased risk-shifting — firms that are closer to default (Eisdorfer, 2008). The seeming reversal of the relation between pension-based compensation and FX exposure for low-distress firms is puzzling and emphasizes the need to further study the relation between managerial compensation and corporate FX exposures. [Insert Tables 7 and 8 approximately here]

4.3. Robustness tests It is possible that our observed results are partially determined by an endogenous relation Page 18 of 43

between executive compensation and the level of a firm’s multinationalism. It would be problematic if firms that compensate their executives were primarily domestically oriented and thus would have lower FX exposures. To address this concern we collect data on three proxy variables that are commonly used to measure a firm’s level of multinationalism: 1) The number of geographic segments in which the company reports operations (Compustat Geographic Segments database), 2) foreign sales ratios, and 3) foreign asset ratios. To compare firms that offer pension-based executive compensation to those that do not, we expand our sample by 428 firms that were previously omitted from this study because they do not use pensions as compensation. Including the 272 firms contained in our main sample (large firms with pension-based executive compensation), these two data subsets span the 700 largest U.S. traded firms.

Panel A of Table 9 presents summary statistics of multinationality variables for the two subsamples of large U.S. firms. Panel B presents t-test for differences in average values and Wilcoxon Signed Rank (Median) test values for differences in medians with corresponding p-values. In Panel C, we classify firms as either multinational or domestic; A firm is considered multinational if either Geo_segments is higher than 1, the firm reported a positive foreign sales ratio, or the firm reported a positive foreign asset ratio during at least one sample year. Statistical significance of differences is based on Pearson 𝜒 2 tests and corresponding p-values. Summary statistics indicate that firms that offer pensions report operations in more geographic segments, have higher foreign sales ratios, and have higher foreign asset ratios than large firms that do not offer pensions. Of the 272 firms that offer pensions, 71% are multinational (based on the number of geographic segments and foreign sales ratios); 45.30% of firms that offer

Page 19 of 43

pensions have positive foreign asset ratios. In contrast 41% of large firms that do not offer pension-based executive compensation are multinational, while 19% of these firms have positive foreign asset ratios. Based on these results we conclude that the positive relation between pension-based executive compensation and FX exposures is unlikely driven by the aforementioned endogeneity problem. [Insert Table 9 approximately here]

Our results presented in Section 4.2 may be affected by a selection bias. It is possible that cash-strapped firms that are closer to financial distress/bankruptcy (low distance-to-default measures) employ executives that are willing to be paid with deferred compensation. Such managers would then have increased incentive to lower the firm’s risk. For our study this would be particularly problematic if firms that had no pension-based compensation would change their compensation scheme as a result of becoming financially distressed. We explore the magnitude of this potential problem in Table10. In Panel A the 700 largest U.S. firms are sorted into quartiles based on their average Dtd measures, where “Lowest” contains firms with the lowest Dtd measures (riskiest firms). We report the percentages of firms for each distress quartile that offer pensions. Panel B presents the results of a Logistic regression where the dependent variable indicates whether the firm offers pensions or not. Explanatory variables include Dtd, Firmage (the age of the firm in years), Debteq (Debt-to-Equity ratio), and a dummy variable that classifies the firm as multinational or domestic based on one of the three multinationality measures. We observe that the percentage of firms that offer pensions monotonically decreases with default risk. In the high-risk quartile (Lowest Dtd), 6.25% of firms offer pensions, compared to the

Page 20 of 43

low-risk quartile (High Dtd) where 79.17% of firms offer pensions. Results reported in Panel B further confirm the negative relation between default risk and the likelihood of the firm using pension-based executive compensation. We also observe that firms that offer pensions are more likely to be both older and multinational. The results reported in Table 10 indicate that highly distressed firms are less likely to offer pension-based compensation to their executives. [Insert Table 10 approximately here]

Although we cannot rule out that the results reported in Section 4.2 are affected by a selection bias, we believe that this potential bias is likely to have a small effect on our results. In addition to the observations based on Table 10, we offer the following additional points: First, firms have historically not changed compensation schemes for executives when they become financially distressed. In our sample of 272 firms, pensions were given throughout the entire sample period by the vast majority of firms. Only 4 firms initiated new executive pension schemes. Second, cash-poor firms can just as likely compensate managers with equity-based “deferred” compensation. This is quite common in many startup firms and high-growth firms that offer managers stock options and other equity compensation schemes associated with long vesting periods. It is conceivable that financially distressed firms that are cash-strapped could use stock options to compensate their managers.

5. Conclusion In this paper we study a manager-owner agency conflict that has not been previously considered in the literature. We argue that pension compensation (firms with high levels of

Page 21 of 43

inside debt) incentivizes managers to manage FX exposure more aggressively in order to reduce cash flow volatility, ultimately resulting in lower levels of FX exposure. Our argument is based on the idea that pension-compensated managers are primarily interested in preserving the long-term survival of the firm, thus ensuring the payout of their pensions. Empirically, we find evidence of a negative relation between pension-based executive compensation and FX exposures, which is consistent with our main argument. Moreover, we observe that our results are magnified for financially distressed firms. Our results are consistent with agency theory that describes increased risk-shifting in distressed firms (Eisdorfer, 2008; Jensen and Meckling, 1976). Our paper has several important implications: 1) Researchers should consider agency and risk-shifting arguments in future exposure research, 2) Multinational firms need to consider the effects of executive compensation on resulting economic exposures, 3) Principal shareholders and boards of directors can use pension-based compensation to preserve firm value during times of financial distress, and 4) Investors and lenders can gain valuable insight into managerial risk-taking during times of crisis by considering the executive’s compensation structure.

Acknowledgements: The authors would like to thank Andrea Moro and participants at the 2014 EFA, and 2015 FMA annual meetings for helpful comments and suggestions.

Appendix: An example of the pension value estimation procedure

Using TJX Companies, Inc. as an example firm, we can establish how the pension computation is performed for each executive. In this case, Edmond J. “Ted” English, the

Page 22 of 43

President and CEO of TJX Companies in 2005, provides a good representative example. Our example follows the procedure first outlined in Sundaram and Yermack (2007). In Table A1, we have reproduced the same pension table disclosure available to investors of TJX Companies in the fiscal year ending January 2005. While investors may reference annual reports to access these tables, they are presented more conveniently in Definitive 14A statements. The table records years of service in five-year increments on the horizontal axis, and final average earnings in varying $50,000, $100,000, and $200,000 increments on the vertical axis. Final average earnings are defined as the average of the three highest years of salary and bonus awards in the ten years prior to retirement. We assume the most recent five years of Mr. English’s compensation are his five highest years of compensation in the last ten years; we were therefore required to conduct further research into prior SEC statements to determine his compensation in every year beginning in 2001. Our calculation yields a five-year average of $1.654 million in earnings credit towards retirement. For each executive firm-year, a sufficient historical salary and bonus level of each executive was computed. To begin the sample at 2000, firms requiring three years of historical compensation needed SEC data beginning in 1998, and firms requiring five years needed SEC data beginning in 1996. For many executives, especially those requiring five or more years of averaged compensation to compute their earnings, historical data was unavailable for as much time as was needed. To compute average compensation for these executives, salaries and bonuses were ‘downwardly weighted’ to the oldest year. For example, if five years of data were required to average an executive’s compensation and four years were available, the most recent three years were waited equally and the most distant year double-weighted to generate a five-year proxy. Page 23 of 43

Mr. English’s widely-available birth year of 1954 establishes his age at 50 in January 2005; for other executives, age information was obtained from 10-Ks (when available), and using a variety of other sources including old news articles, obituaries, and public records indexing services. Retirement age to achieve full benefit is 65. The multiplicative factor M can be determined algebraically from Table A1: the addition of every $1,000,000 in final average earnings generates $500,000 of additional pension compensation for 20 years of service; this corresponds to 0.50 for 20 years, or 0.025 (2.5%) of final average earnings for each year of service. English, as of 2005, has 22 years of service credit towards retirement We can assume that English will work through his 65th year, at which point he will retire with 37 years of service. 9 Following equation (2), we can calculate his annual pension entitlement credited upon retirement as 0.025 x 37 x $1.654 = $1.530 million. To complete Eq. (1), we require English’s age, A (50); R, the company’s retirement age (65); d, the cost of long-term debt; and P(n), the probability that English will be alive and receiving pension disbursements n years into the future. The cost of long term debt, determined from the Federal Reserve Statistical Release H.15 for Moody’s Aaa rated bonds was d = 0.0523 for 2005. Using the statistical tables provided by the U.S. Social Security Administration, we infer that English has an 85.3% chance of being alive to receive his first payment at the age of 66, an 83.8% chance of surviving until age 67, and so forth until age 120.10 The summation of each year’s actuarial present-value contribution establishes our present value of English’s pension benefit at the beginning of 2005: $6.355 million.

9

Mr. English was 50 with 22 years of service at the very beginning of 2005; he is eligible to achieve full retirement benefits in 2020, at which point he will have 37 years of service (22+(65-50)). 10

The odds of English surviving even to age 111 are so minimal, that no additional present value is added beyond this age. Thus, the age 120 truncation is appropriate based on current longevity estimates.

Page 24 of 43

References Adkins, L.C., Carter, D.A., Simpson, W.G., 2007. Managerial incentives and the use of foreign-exchange derivatives by banks. Journal of Financial Research 30, 399-413. Adler, M., Dumas, B., 1984. Exposure to currency risk: Definition and measurement. Financial Management, 41-50. Allayannis, G., Ihrig, J., 2001. Exposure and markups. Review of Financial Studies 14, 805-835. Bartram, S.M., 2008. What lies beneath: Foreign exchange rate exposure, hedging and cash flows. Journal of Banking & Finance 32, 1508-1521. Baum, C.F., Schaffer, M.E., Stillman, S., 2003. Instrumental variables and GMM: Estimation and testing. Stata Journal 3, 1-31. Belkhir, M., Boubaker, S., 2013. CEO inside debt and hedging decisions: Lessons from the US banking industry. Journal of International Financial Markets, Institutions and Money 24, 223-246. Bessembinder, H., 1991. Forward contracts and firm value: Investment incentive and contracting effects. Journal of Financial and Quantitative Analysis 26, 519-532. Black, F., Scholes, M., 1973. The pricing of options and corporate liabilities. Journal of Political Economy 3, 637-645. Bodnar, G.M., Gentry, W.M., 1993. Exchange rate exposure and industry characteristics: Evidence from Canada, Japan, and the USA. Journal of International Money and Finance 12, 29-45. Bodnar, G.M., Hayt, G.S., Marston, R.C., 1996. 1995 Wharton survey of derivatives usage by US non-financial firms. Financial Management, 113-133. Bodnar, G.M., Hayt, G.S., Marston, R.C., 1998. 1998 Wharton survey of financial risk management by US non-financial firms. Financial Management, 70-91. Bodnar, G.M., Hayt, G.S., Marston, R.C., Smithson, C.W., 1995. Wharton survey of derivatives usage by US non-financial firms. Financial Management, 104-114.

Page 25 of 43

Bodnar, G.M., Wong, M.F., 2003. Estimating exchange rate exposures: issues in model structure. Financial Management, 35-67. Cassell, C.A., Huang, S.X., Manuel Sanchez, J., Stuart, M.D., 2012. Seeking safety: The relation between CEO inside debt holdings and the riskiness of firm investment and financial policies. Journal of Financial Economics 103, 588-610. Coles, J.L., Daniel, N.D., Naveen, L., 2006. Managerial incentives and risk-taking. Journal of Financial Economics 79, 431-468. Core, J., Guay, W., 2002. Estimating the value of employee stock option portfolios and their sensitivities to price and volatility. Journal of Accounting Research 40, 613-630. Dominguez, K.M., Tesar, L.L., 2006. Exchange rate exposure. Journal of International Economics 68, 188-218. Edmans, A., Liu, Q., 2011. Inside debt. Review of Finance 15, 75-102. Eisdorfer, A., 2008. Empirical evidence of risk shifting in financially distressed firms. Journal of Finance 63, 609-637. Eisdorfer, A., Giaccotto, C., White, R., 2014. Do managers skimp on shareholders' dividends to protect their own retirement funds? Forthcoming Publication, Journal of Corporate Finance. Fama, E.F., French, K.R., 1988. Dividend yields and expected stock returns. Journal of Financial Economics 22, 3-25. Francis, B.B., Hasan, I., Hunter, D., Zhu, Y., 2013. Do managerial risk-taking incentives impact exchange rate exposure? Working Paper, Rensselaer Polytechnic Institute/University of South Florida. Géczy, C., Minton, B.A., Schrand, C., 1997. Why firms use currency derivatives. Journal of Finance 52, 1323-1354. He, J., Ng, L.K., 1998. The foreign exchange exposure of Japanese multinational corporations. Journal of Finance 53, 733-753. Jensen, M.C., Meckling, W.H., 1976. Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3, 305-360. Jorion, P., 1990. The exchange-rate exposure of US multinationals. The Journal of Business 63, 331-345. Page 26 of 43

Leland, H.E., 1998. Agency costs, risk management, and capital structure. Journal of Finance 53, 1213-1243. Rogers, D.A., 2002. Does executive portfolio structure affect risk management? CEO risk-taking incentives and corporate derivatives usage. Journal of Banking & Finance 26, 271-295. Smith, C.W., Stulz, R.M., 1985. The determinants of firms' hedging policies. Journal of Financial and Quantitative Analysis 20, 391-405. Stulz, R.M., 1984. Optimal hedging policies. Journal of Financial and Quantitative Analysis 19, 127-140. Sundaram, R.K., Yermack, D.L., 2007. Pay me later: Inside debt and its role in managerial compensation. Journal of Finance 62, 1551-1588. Tufano, P., 1998. Agency costs of corporate risk management. Financial Management, 67-77. Wei, K.D., Starks, L.T., 2013. Foreign exchange exposure elasticity and financial distress. Financial Management 42, 709-735. Williamson, R., 2001. Exchange rate exposure and competition: evidence from the automotive industry. Journal of Financial Economics 59, 441-475.

Page 27 of 43

Table 1 Descriptive statistics Panel A: Firm-level overview Firm-level variables |𝛾𝑖 | |𝜉𝑖𝐵 | Firm CompLev Firm Pens/Assets Firm SalScaled Firm OptScaled Firmsize Bookmkt Debteq Dtd Liqcons Tax Execown Forsales ROE CEO-level variables CEO CompLev CEO Pens/Assets CEO SalScaled CEO OptScaled Instrumental Variables Executive Age (Firm-Level, Avg. of Top 5) Executive Age (CEO-Level) M

N 2,303 2,303 1,581 1,684 1,781 1,741 1,580 1,810 1,761 1,543 1,543 1,543 1,761 2,253 2,244 N 1,741 1,741 1,636 1,642

Mean 0.554 2.193 -0.061 -0.302 0.012 0.057 3.906 0.295 0.869 2.096 0.018 0.722 0.030 0.244 0.191 Mean -0.093 -0.077 0.008 0.032

SD 0.499 3.141 1.140 0.134 0.026 0.149 1.199 0.487 2.730 1.504

P25 0.205 0.432 -0.609 -0.133 0.001 0.005 3.634 0.147 0.322 1.112

P50 0.414 1.094 -0.198 -0.052 0.004 0.018 4.112 0.230 0.600 2.003

P75 0.735 2.543 0.047 0.052 0.012 0.059 4.523 0.376 1.123 3.011

0.125 0.250 1.326 SD 0.119 1.092 0.013 0.062

0.002 0.000 0.093 P25 -0.185 -0.623 0.001 0.002

0.006 0.184 0.149 P50 -0.116 -0.179 0.003 0.009

0.015 0.448 0.223 P75 -0.017 0.010 0.004 0.034

1,464 1,495 1,543

54.14 56.54 0.017

4.069 5.678 0.038

51.50 53.00 0.010

54.20 57.00 0.016

57.00 60.00 0.019

Page 28 of 43

Table 1 Descriptive statistics (continued) Panel B: Industry breakdown Industry Agriculture, Forestry, & Fishing Construction Finance, Insurance, and Real Estate Manufacturing Mining Non-classifiable Establishments Retail Trade Services Transportation & Public Utilities Wholesale Trade Total firms

SIC codes 01-09 15-17 60-67 20-39 10-14 99 52-59 70-89 40-49 50-51

N 1 1 47 130 15 2 13 12 46 5 272

% of total 0.37% 0.37% 17.28% 47.79% 5.51% 0.74% 4.78% 4.41% 16.91% 1.84% 100.00%

Notes: Columns reflect mean, standard deviation, and N, the number of firm-years for each variable. P25, P50, and P75 indicate the 25th, 50th, and 75th percentiles of each variable. We use two measures of FX exposure: the absolute value of FX equity exposure (see Eq. (4) in text), and the absolute value of FX cash flow exposure (see Eq. (5) in text). Instrumental variables include M, a multiplier factor equivalent to the percentage of pension benefit for each dollar of compensation earned, and average ‘top 5’ executive age during the sample firm year. CompLev and Pension/Assets are industry-adjusted measures of relative pension size. CompLev is the ratio of inside debt divided by the sum of inside debt and inside equity; Pension/Assets is the ratio of the actuarial value of the pension(s) to total assets. Additional independent variables include SalScaled, which is salary and bonus compensation for managers scaled by firm asset size and OptScaled, which is option award value scaled by firm asset size. Executive compensation variables are used at the CEO-level as well as firm-level. FX exposure–related control variables include: Firmsize, the natural log of the firm’s market value; Bookmkt, the market value of firm’s common shares divided by shareholders’ equity; Debteq, firms’ long-term debt divided by equity for the year; DtD, distance-to-default defined by the Moody’s KMV Method explained in the text; Liqcons and Tax, dummy variables representing whether the firm reported a negative income or a net operating loss carryover during that firm year; Execown, captures the ownership of the CEO as a proportion of all firm equity; Forsales, the firm’s foreign sales ratio, which is computed by dividing foreign-based sales by total sales; and ROE, return on equity for the sample year. We also test pension effects using two different independent variables: industry-adjusted compensation leverage and scaled actuarial pension value for aggregate executive data at the firm-level. Data is on 272 firms over the period 2000-2009. In Panel B we provide the two-digit SIC industry designations of our sample firms, obtained from Compustat. N refers to the number of firms in that industrial category.

Page 29 of 43

Table 2 Correlation coefficients |𝛾𝑖 | [1] |𝜉𝑖𝐵 | [2] CEO CompLev[3] CEO Pens/Assets [4] CEO SalScaled [5] CEO OptScaled [6] Firmsize [7] Bookmkt [8] Debteq [9] Dtd [10] Liqcons [11] Tax [12] Execown [13] Forsales [14] ROE [15] Bookmkt [8] Debteq [9] Dtd [10] Liqcons [11] Tax [12] Execown [13] Forsales [14] ROE [15]

[1] 1 0.088* 0.037 -0.011 0.074* 0.063* -0.039 -0.005 0.010 -0.031 0.003 -0.031 0.001 0.247* -0.007 [6] 1 -0.002 -0.037 -0.016 -0.067* 0.075* 0.024 -0.042

[2]

[3]

[4]

[5]

[6]

[7]

1 0.009 -0.057* -0.080* -0.069* -0.030 -0.007 0.002 -0.059* 0.002 -0.008 0.049* 0.153* -0.029 [7]

1 0.187* -0.264* -0.239* -0.027 0.023 0.026 -0.003 -0.047 -0.010 -0.016 -0.028 0.004 [8]

1 0.243* 0.218* -0.009 0.009 0.018 0.025 0.005 -0.016 0.001 -0.079* -0.009 [9]

1 0.404* 0.001 -0.008 -0.006 0.035 -0.007 0.019 -0.004 -0.004 0.005 [10]

1 0.026 -0.007 -0.001 0.002 0.007 -0.040 0.032 0.074* 0.015 [11]

1 -0.103* 0.029 0.343* 0.050 0.360* -0.053 -0.012 0.011 [12]

1 -0.023 0.000 0.016 -0.003 -0.039 0.001

1 -0.115* 0.229* -0.036 -0.046 -0.016

1 0.000 0.068* 0.041 0.006

1 -0.062 0.015 -0.016

1 0.024 -0.005

1 0.010

Notes: This table reports Person correlation coefficients for the main variables used in this study. The * denotes that variable correlations are significant at the 0.01 level. We use two measures of FX Exposure: the absolute values of FX equity and FX cash flow exposures (see Eqs. (4) and (5) in text). CompLev and Pension/Assets are industry-adjusted compensation leverage and scaled actuarial pension value at the CEO-level. Additional independent variables include CEO SalScaled, salary and bonus compensation for managers scaled by firm asset size; CEO OptScaled, option award value scaled by firm asset size; Firmsize, the natural log of the firm’s market value; Bookmkt, the market value of firms’ common shares divided by shareholders’ equity; Debteq, the firms’ long-term debt divided by equity for the year; Dtd, distance-to-default defined by the Moody’s KMV Method explained in the text; Liqcons and Tax, dummy variables representing whether the firm reported a negative income or a net operating loss carryover during that firm year; Execown, the ownership of the CEO as a proportion of all firm equity; Forsales, the firm’s foreign sales ratio; and ROE, control for the return on equity for the sample year.

Page 30 of 43

Table 3 FX exposure and executive compensation, CEO-level (OLS model) Model: Dependent Variable: CEO CompLev

[i] -0.5579** (-2.95)

CEO Pens/Assets CEO SalScaled CEO OptScaled Firmsize Bookmkt Debteq Dtd Liqcons Tax Execown Forsales ROE Constant Observations 𝐴𝑑𝑗. 𝑅 2 Year Variables Industry Variables

[ii]

[iii]

|𝛾𝑖 |

-0.0001 (-0.18) 0.0010 (0.36) -0.0091 (-0.72) -0.0001 (-1.28) -0.0004 (-0.12) 0.0115 (0.74) -0.1284 (-1.27) -0.0996** (-2.73) -0.8205 (-0.79) 0.8856*** (8.93) 0.0084 (1.70) 0.8610*** (6.73) 1,232 0.1662 Yes Yes

|𝜉𝑖𝐵 |

[iv]

-0.712** (-3.07) -0.0301** (-3.13) 0.0010* (1.89) 0.0037 (1.12) -0.0101 (-0.81) -0.0001 (-1.30) -0.0017 (-0.38) 0.0120 (0.76) -0.1154 (-1.13) -0.0975** (-2.62) -0.8047 (-0.79) 0.8831*** (8.68) -0.0061 (-3.56) 0.8007*** (5.32) 1,232 0.1545 Yes Yes

-0.0032 (-1.38) -0.0237** (-2.67) 0.0551 (1.00) -0.0001 (-0.17) -0.0083 (-0.60) -0.0149 (-0.65) 0.315 (0.66) -0.130 (-1.07) 3.484 (0.61) 2.310*** (7.57) -0.130 (-1.07) 4.995*** (3.87) 1,265 0.0961 Yes Yes

-0.0761 (-0.96) 0.0199* (1.96) -0.0688** (-2.37) -0.0053 (-0.05) -0.0001 (-1.27) 0.0551 (1.00) -0.173* (-2.05) -0.297 (-0.47) -0.130 (-0.76) 1.451 (0.26) 3.127*** (4.38) -0.130 (-0.76) 5.189*** (3.84) 1,159 0.0831 Yes Yes

Notes: In this table, we present four OLS regression results based on Eq. (6). We use two measures of FX Exposure: the absolute value of FX equity exposure (see Eq. (4) in text) and FX cash flow exposure (see Eq. (5) in text). We also test pension effects using two different independent variables: industry-adjusted compensation leverage and scaled actuarial pension value for aggregate executive data at the CEO-level. Additional independent variables include CEO SalScaled, salary and bonus compensation for managers scaled by firm asset size; CEO OptScaled, option award value scaled by firm asset size; Firmsize, the natural log of the firm’s market value; Bookmkt, the market value of firms’ common shares divided by shareholders’ equity; Debteq, the firms’ long-term debt divided by equity for the year; Dtd, distance-to-default defined by the Moody’s KMV Method explained in the text; Liqcons and Tax, dummy variables representing whether the firm reported a negative income or a net operating loss carryover during that firm year; Execown, the ownership of the CEO as a proportion of all firm equity; Forsales, the firm’s foreign sales ratio;

Page 31 of 43

and ROE, control for the return on equity for the sample year. The sample contains between 1,159 and 1,265 firm-years, representing data on 272 firms over the period 2000-2009. Reported standard errors are adjusted for HC and AC following Rogers (1993) and the corresponding T-statistics are reported in parentheses. Statistical significance is indicated as follows: * p < 0.10, ** p < 0.05, *** p < 0.001

Page 32 of 43

Table 4 FX exposure and executive compensation, firm-level (OLS model) Model Dependent Variable: Firm CompLev

[i] -0.0635*** (-3.81)

Firm Pens/Assets Firm SalScaled Firm OptionsScaled Firmsize Bookmkt Debteq Dtd Liqcons Tax Execown Forsales ROE Constant Observations 𝐴𝑑𝑗. 𝑅 2 Year Variables Industry Variables

[ii]

[iii]

|𝛾𝑖 |

0.0008 (1.03) 0.0057** (2.67) -0.0075 (-0.68) -0.0001 (-0.79) -0.0039 (-0.66) -0.0071 (-0.56) -0.0469 (-0.41) -0.0930** (-2.72) -2.694** (-2.86) -0.0471 (-0.80) -0.0101** (-2.51) 1.214*** (5.15) 1,117 0.0626 Yes Yes

|𝜉𝑖𝐵 |

[iv]

-0.0391 (-0.70) -0.872*** (-4.56) -0.0015 (-1.44) 0.0027 (1.14) -0.0078 (-0.71) -0.0001 (-1.08) -0.0020 (-0.35) -0.0067 (-0.58) -0.106 (-0.96) -0.0741** (-2.30) -2.294** (-2.81) -0.0643 (-0.96) -0.0127** (-3.12) 1.322*** (6.38) 1,196 0.0729 Yes Yes

-0.0017 (-0.66) -0.0151 (-1.74) 0.0192 (0.17) -0.0001 (-1.31) 0.0495 (0.96) -0.139 (-1.76) -0.303 (-0.57) -0.315 (-1.73) -0.263 (-0.06) 3.312*** (5.15) -0.0592* (-1.95) 3.351 (1.10) 1,217 0.0773 Yes Yes

-0.744** (-3.12) -0.0013 (-0.67) -0.0191** (-2.80) 0.0554 (0.98) -0.0001 (-0.19) -0.0097 (-0.70) -0.0137 (-0.59) 0.293 (0.61) -0.136 (-1.08) 3.551 (0.60) 2.335*** (7.78) -0.0455* (-1.99) 5.039*** (3.93) 1,149 0.0994 Yes Yes

Notes: In this table, we present four OLS regression results based on Eq. (6). We use two measures of FX Exposure: the absolute value of FX equity exposure (see Eq. (4) in text) and FX cash flow exposure (see Eq. (5) in text). We use two measures of FX Exposure: the absolute value of the FX equity exposure and FX cash flow exposure (see Eqs. (4) and (5) in text). We also test pension effects using two different independent variables: industry-adjusted compensation leverage and scaled actuarial pension value for aggregate executive data at the firm-level — for most firms this spans the five highest-paid company executives. Additional independent variables include Firm SalScaled, salary and bonus compensation for managers scaled by firm asset size; Firm OptScaled, option award value scaled by firm asset size; Firmsize, the natural log of the firm’s market value; Bookmkt, the market value of firms’ common shares divided by shareholders’ equity; Debteq, the firms’ long-term debt divided by equity for the year; Dtd, distance-to-default defined by the Moody’s KMV Method explained in the text; Liqcons and Tax, dummy variables representing whether the firm reported a negative income or a net operating loss carryover during that firm year; Execown, the ownership of managers as a proportion of all firm equity; Forsales, the firm’s foreign sales ratio; and ROE, control for the return on equity for the sample year. The sample contains between 1,117 and 1,217 firm-years, representing data on 272 firms over the period 2000-2009. Reported standard errors are adjusted for HC and AC

Page 33 of 43

following Rogers (1993) and corresponding T-statistics are reported in parentheses. Statistical significance is indicated as follows: * p < 0.10, ** p < 0.05, *** p < 0.001

Page 34 of 43

Table 5 FX exposure and executive compensation, CEO-level (2-SLS model) Model Dependent Variable: CEO CompLev

[i]

CEO OptScaled Firmsize Bookmkt Debteq Dtd Liqcons Tax Execown Forsales ROE Constant Observations Year Variables Industry Variables

[iii]

-0.0002 (-0.44) 0.0011 (0.42) -0.0200** (-2.35) -0.0001 (-0.62) -0.0016 (-0.68) 0.0130 (1.58) -0.0549 (-0.89) -0.0403 (-1.34) -0.404 (-0.60) 0.540*** (6.22) 0.0025 (0.66) 0.246** (2.57) 1,118 Yes Yes

[iv] |𝜉𝑖𝐵 |

-3.443** (-2.05)

-0.0439 (-0.15)

CEO Pens/Assets CEO SalScaled

[ii] |𝛾𝑖 |

-0.596** (-2.27) 0.0010 (1.53) 0.0211* (1.85) -0.0115 (-0.72) -0.0001 (-0.35) 0.0166 (1.03) 0.0309*** (2.63) 0.0981 (0.71) -0.0441 (-1.09) -0.0920 (-0.08) 0.0561 (0.23) -0.0002 (-0.04) -0.296 (-0.75) 1,098 Yes Yes

-0.00940** (-2.37) -0.0262*** (-2.82) 0.0530 (1.17) 0.0001 (0.35) -0.0144 (-0.72) -0.0150 (-0.84) 0.524 (1.41) -0.147 (-1.23) 4.181 (0.87) 2.110*** (6.45) -0.0459** (-1.99) 0.121 (0.27) 1,083 Yes Yes

-3.272** (-2.43) 0.0686*** (2.77) -0.0463*** (-3.00) 0.136 (1.57) 0.0001 (1.32) 0.0603 (1.42) -0.0991 (-1.13) 0.139 (0.21) -0.0758 (-0.32) 6.032 (0.98) 0.0498 (0.05) -0.0357 (-1.35) 0.746 (0.21) 1,070 Yes Yes

Notes: In this table, we present four 2-SLS regression results with robust standard errors. We use two measures of FX Exposure: the absolute value of FX equity exposure (see Eq. (4) in text) and FX cash flow exposure (see Eq. (5) in text). Instrumental variables include M, a multiplier factor equivalent to the percentage of pension benefit for each dollar of compensation earned, and CEO age during the sample firm year. We also test pension effects using two different independent variables: industry-adjusted compensation leverage and scaled actuarial pension value at the CEO-level. Additional independent variables include CEO SalScaled, salary and bonus compensation for managers scaled by firm asset size; CEO OptScaled, option award value scaled by firm asset size; Firmsize, the natural log of the firm’s market value; Bookmkt, the market value of firms’ common shares divided by shareholders’ equity; Debteq, the firms’ long-term debt divided by equity for the year; Dtd, distance-to-default defined by the Moody’s KMV Method explained in the text; Liqcons and Tax, dummy variables representing whether the firm reported a negative income or a net operating loss carryover during that firm year; Execown, the ownership of the CEO as a proportion of all firm equity; Forsales, the firm’s foreign sales ratio; and ROE, control for the return on equity for the sample year. The sample contains between 1,083 and 1,118 firm-years, representing data on 272 firms over the period 2000-2009. Standard errors are adjusted for HC and AC following Rogers (1993) and corresponding T-statistics are presented in parentheses. Statistical significance is indicated as follows: * p < 0.10, ** p < 0.05, *** p <

Page 35 of 43

0.001.

Page 36 of 43

Table 6 FX exposure and executive compensation, firm-level (2-SLS model) Model Dependent Variable: Firm CompLev

[i]

Firm OptScaled Firmsize Bookmkt Debteq Dtd Liqcons Tax Execown Forsales ROE Constant Observations Year Variables Industry Variables

[iii]

|𝛾𝑖 | -0.545 (-1.45)

Firm Pens/Assets Firm SalScaled

[ii]

0.0027 (1.61) 0.0194 (1.28) -0.0468*** (-3.28) -0.0001 (-0.65) 0.0161 (0.89) 0.0541* (1.94) 0.0591 (0.55) -0.0207 (-0.30) -1.084 (-1.31) 0.137 (0.59) -0.0124 (-1.35) -0.261 (-0.55) 1,037 Yes Yes

|𝜉𝑖𝐵 |

[iv]

-3.084 (-1.22) 0.471 (1.00) 0.0003 (0.34) 0.00306* (1.87) -0.0253*** (-2.62) -0.0001 (-0.30) -0.0011 (-0.22) 0.0105 (1.14) 0.0103 (0.15) -0.0061 (-0.24) -1.361*** (-2.68) 0.0266 (0.45) -0.0068** (-2.24) 0.461** (2.56) 1,029 Yes Yes

0.0118 (1.18) 0.0900 (0.85) 0.0850 (1.02) -0.0001 (-0.39) 0.105 (1.00) 0.0158 (0.11) 0.0441 (0.08) -0.280 (-0.61) -0.259 (-0.05) 2.195 (1.28) -0.0698 (-1.30) -0.268 (-0.05) 1,037 Yes Yes

3.134 (0.80) -0.0018 (-0.29) -0.0051 (-0.30) 0.0191 (0.34) -0.0002 (-1.43) 0.0176 (0.51) 0.0575 (1.14) 0.0097 (0.03) -0.102 (-0.33) -4.451 (-1.41) -0.461 (-1.57) -0.113*** (-2.60) 1.180* (1.90) 1,179 Yes Yes

Notes: In this table, we present four 2-SLS regression results with robust standard errors. We use two measures of FX Exposure: the absolute value of FX equity exposure (see Eq. (4) in text) and FX cash flow exposure (see Eq. (5) in text). Instrumental variables include M, a multiplier factor equivalent to the percentage of pension benefit for each dollar of compensation earned, and average ‘top 5’ executive age during the sample firm year. We also test pension effects using two different independent variables: industry-adjusted compensation leverage and scaled actuarial pension value for aggregate executive data at the firm-level. Additional independent variables include Firm SalScaled, salary and bonus compensation for managers scaled by firm asset size; Firm OptScaled, option award value scaled by firm asset size; Firmsize, the natural log of the firm’s market value; Bookmkt, the market value of firms’ common shares divided by shareholders’ equity; Debteq, the firms’ long-term debt divided by equity for the year; Dtd, distance-to-default defined by the Moody’s KMV Method explained in the text; Liqcons and Tax, dummy variables representing whether the firm reported a negative income or a net operating loss carryover during that firm year; Execown, the ownership of managers as a proportion of all firm equity; Forsales, the firm’s foreign sales ratio; and ROE, control for the return on equity for the sample year. The sample contains between 1,000 and 1,087 firm-years, representing data on 272 firms over the period 2000-2009. Reported standard errors are adjusted for HC and AC following Rogers (1993) and corresponding T-statistics are presented in parentheses. Statistical significance is indicated as follows: * p < 0.10, ** p < 0.05, *** p < 0.001.

Page 37 of 43

Table 7 CEO-level FX exposure and executive compensation by distance-to-default Dep. Var: |𝛾𝑖 |

Distance-to-Default Level Lowest

CEO CompLev

*

-0.164 (-2.18)

Distance-to-Default Level

Low-Mid

High-Mid

High

0.0209 (0.24)

0.147 (1.62)

0.130 (1.20)

CEO Pens/Assets CEO SalScaled CEO OptScaled Firmsize Bookmkt Debteq Dtd Liqcons Tax Execown Forsales ROE Constant Observations 𝐴𝑑𝑗. 𝑅 2 Year Variables Industry Variables

0.0018 (0.97) -0.0005 (-0.08) -0.0224* (-1.91) -0.0001*** (-5.06) 0.0006 (0.14) 0.0132 (0.20) 0.0129 (0.14) -0.0239 (-0.53) -0.599 (-0.47) 0.345*** (4.14) -0.0009 (-0.09) 0.896*** (3.68) 339 0.1483 Yes Yes

-0.0021** (-2.49) 0.0064 (1.05) 0.0005 (0.01) 0.0001 (0.62) -0.0037 (-1.03) 0.0655 (0.93) -0.300*** (-3.80) -0.0615 (-1.68) -2.488** (-3.21) 0.439*** (5.41) -0.0013 (-0.31) 0.247 (1.06) 341 0.1642 Yes Yes

0.0002 (0.48) -0.0021 (-0.60) -0.0387 (-0.74) 0.0001 (0.38) -0.0073 (-1.41) -0.0286 (-0.45) 0.0257 (0.12) -0.0615 (-1.12) -0.293 (-0.38) 0.563*** (5.23) -0.0425 (-0.53) 0.790*** (3.47) 336 0.1773 Yes Yes

0.0011 (1.70) 0.0014 (0.52) 0.0121 (0.23) -0.0001*** (-4.68) -0.0347*** (-3.28) 0.0022 (0.17) 0.0000 (0.00) -0.102* (-2.09) 0.502 (0.36) 0.474*** (4.42) -0.0242 (-0.36) 1.521*** (6.20) 337 0.1736 Yes Yes

Lowest

Low-Mid

High-Mid

Highest

-0.0207 (-1.54) 0.0023 (1.25) 0.0014 (0.23) -0.0219* (-1.90) -0.0001*** (-4.19) 0.00104 (0.23) 0.00615 (0.09) 0.0252 (0.27) -0.0241 (-0.53) -0.475 (-0.40) 0.345*** (4.20) -0.0025 (-0.24) 0.869*** (3.67) 339 0.1470 Yes Yes

-0.0110 (-0.51) -0.0021** (-3.09) 0.0069 (1.08) -0.0008 (-0.02) 0.0002 (0.66) -0.0035 (-0.94) 0.0638 (0.93) -0.293*** (-3.57) -0.0607 (-1.67) -2.465** (-3.06) 0.434*** (5.04) -0.0012 (-0.29) 0.246 (1.07) 341 0.1651 Yes Yes

0.0125 (0.73) -0.0001 (-0.11) -0.0037 (-1.17) -0.0423 (-0.80) 0.0001 (0.28) -0.0075 (-1.42) -0.0354 (-0.57) 0.0223 (0.10) -0.0594 (-1.07) -0.353 (-0.45) 0.559*** (4.85) -0.0510 (-0.60) 0.817*** (3.48) 336 0.1760 Yes Yes

0.0134 (1.44) 0.0006 (1.21) 0.0007 (0.31) 0.0144 (0.27) -0.0001*** (-4.60) -0.0371*** (-3.77) 0.0027 (0.21) 0.0000 (0.00) -0.104* (-2.08) 0.417 (0.29) 0.476*** (4.40) -0.0233 (-0.35) 1.516*** (6.33) 337 0.1728 Yes Yes

Notes: In this table, we present four OLS regression results with robust standard errors based on Eq. (6). Using the absolute value of FX equity exposure (see Eq. (4) in text), we divide the sample into quartiles ranked by distance-to-default. The ‘lowest’ quartile contains firms closest to default. We also test pension effects using two different independent variables: industry-adjusted compensation leverage and scaled actuarial pension value at the CEO-level. Additional independent variables include CEO SalScaled, salary and bonus compensation for managers scaled by firm asset size; CEO OptScaled, option award value scaled by firm asset size; Firmsize, the natural log of the firm’s market value; Bookmkt, the market value of firms’ common shares divided by shareholders’ equity; Debteq, the firms’ long-term debt divided by equity for the year; Dtd, distance-to-default defined by the Moody’s KMV Method explained in the text; Liqcons and Tax, dummy variables representing whether the firm reported a negative income or a net operating loss carryover during that firm year; Execown, the ownership of the CEO as a proportion of all firm equity; Forsales, the firm’s foreign sales ratio; and ROE, control for the return on equity for the sample year. The sample represents data on 272 firms over the period 2000-2009. Reported standard errors are adjusted for HC and AC following Rogers (1993) and corresponding T-statistics are presented in parentheses. Statistical significance is indicated as follows: * p < 0.10, ** p < 0.05, *** p < 0.001

Page 38 of 43

Table 8 Firm-level FX exposure and executive compensation by distance-to-default Dep. Var: |𝛾𝑖 |

Distance-to-Default Level Lowest

Firm CompLev

-0.0264 (-2.13)

*

Distance-to-Default Level

Low-Mid

High-Mid

High

-0.0120 (-0.60)

0.0105 (0.59)

0.0156 (1.78)

Firm Pens/Assets Firm SalScaled Firm OptScaled Firmsize Bookmkt Debteq Dtd Liqcons Tax Execown Forsales ROE Constant Observations 𝐴𝑑𝑗. 𝑅 2 Year Variables Industry Variables

0.0023** (2.28) 0.0023 (0.73) -0.0224* (-1.91) -0.0001*** (-5.06) 0.0006 (0.14) 0.0132 (0.20) 0.0129 (0.14) -0.0239 (-0.53) -0.599 (-0.47) 0.345*** (4.14) -0.0009 (-0.09) 0.896*** (3.68) 339 0.1703 Yes Yes

-0.0008 (-0.37) 0.0029 (0.61) 0.0005 (0.01) 0.0001 (0.62) -0.0037 (-1.03) 0.0655 (0.93) -0.300*** (-3.80) -0.0615 (-1.68) -2.488** (-3.21) 0.439*** (5.41) -0.0013 (-0.31) 0.247 (1.06) 341 0.1451 Yes Yes

0.0001 (0.22) -0.0016 (-0.75) -0.0387 (-0.74) 0.0001 (0.38) -0.0073 (-1.41) -0.0286 (-0.45) 0.0257 (0.12) -0.0615 (-1.12) -0.293 (-0.38) 0.563*** (5.23) -0.0425 (-0.53) 0.790*** (3.47) 336 0.1637 Yes Yes

-0.0001 (-0.04) 0.0006 (0.39) 0.0121 (0.23) -0.0001*** (-4.68) -0.0347*** (-3.28) 0.0022 (0.17) 0.0000 (0.00) -0.102* (-2.09) 0.502 (0.36) 0.474*** (4.42) -0.0242 (-0.36) 1.521*** (6.20) 337 0.1746 Yes Yes

Lowest

Low-Mid

High-Mid

Highest

-0.218** (-3.11) 0.0013 (1.54) -0.0011 (-0.32) -0.0219* (-1.90) -0.0001*** (-4.19) 0.00104 (0.23) 0.00615 (0.09) 0.0252 (0.27) -0.0241 (-0.53) -0.475 (-0.40) 0.345*** (4.20) -0.0025 (-0.24) 0.869*** (3.67) 339 0.1537 Yes Yes

0.0241 (0.26) -0.0018** (-2.56) 0.0033 (0.76) -0.0008 (-0.02) 0.0002 (0.66) -0.0035 (-0.94) 0.0638 (0.93) -0.293*** (-3.57) -0.0607 (-1.67) -2.465** (-3.06) 0.434*** (5.04) -0.0012 (-0.29) 0.246 (1.07) 341 0.1634 Yes Yes

0.186 (1.52) 0.0003 (0.82) -0.0008 (-0.36) -0.0423 (-0.80) 0.0001 (0.28) -0.0075 (-1.42) -0.0354 (-0.57) 0.0223 (0.10) -0.0594 (-1.07) -0.353 (-0.45) 0.559*** (4.85) -0.0510 (-0.60) 0.817*** (3.48) 336 0.1786 Yes Yes

0.240* (1.85) 0.0013 (1.81) 0.0002 (0.13) 0.0144 (0.27) -0.0001*** (-4.60) -0.0371*** (-3.77) 0.0027 (0.21) 0.0000 (0.00) -0.104* (-2.08) 0.417 (0.29) 0.476*** (4.40) -0.0233 (-0.35) 1.516*** (6.33) 337 0.1807 Yes Yes

Notes: In this table, we present four OLS regression results with robust standard errors based on Eq. (6). Using the absolute value of FX equity exposure (see Eq. (4) in text), we divide the sample into fourth ranked by Dtd. The ‘lowest’ quartile contains firms closest to default. We also test pension effects using two different independent variables: industry-adjusted compensation leverage and scaled actuarial pension value for aggregate executive data at the firm-level. Additional independent variables include Firm SalScaled, salary and bonus compensation for managers scaled by firm asset size; Firm OptScaled, option award value scaled by firm asset size; Firmsize, the natural log of the firm’s market value; Bookmkt, the market value of firms’ common shares divided by shareholders’ equity; Debteq, the firms’ long-term debt divided by equity for the year; Dtd, distance-to-default defined by the Moody’s KMV Method explained in the text; Liqcons and Tax, dummy variables representing whether the firm reported a negative income or a net operating loss carryover during that firm year; Execown, the ownership of managers as a proportion of all firm equity; Forsales, the firm’s foreign sales ratio; and ROE, control for the return on equity for the sample year. The sample represents data on 272 firms over the period 2000-2009. Reported standard errors are adjusted for HC and AC following Rogers (1993) and corresponding T-statistics are presented in parentheses.

Page 39 of 43

Statistical significance is indicated as follows: * p < 0.10, ** p < 0.05, *** p < 0.001.

Page 40 of 43

Table 9 Pension plans and corporate multinationality Panel A: Summary statistics of multinationality variables for data subsamples Large firms that offer pensions: 272 firms with a total of 2,303 observations Mean SD P25 P50 |𝛾𝑖 | 0.554 0.499 0.205 0.414 2.193 3.141 0.432 1.094 |𝜉𝑖𝐵 | Geo_segments 3.233 2.789 1.000 3.000 Forsales 0.244 0.250 0.000 0.184 Forassets 0.117 0.205 0.000 0.000 Large firms that offer no pensions or data was unavailable: 428 firms with a total of 4,009 observations |𝛾𝑖 | 0.942 0.819 0.311 0.710 𝐵 1.282 1.847 0.247 0.631 |𝜉𝑖 | Geo_segments 1.921 1.815 1.000 1.000 Forsales 0.123 0.252 0.000 0.000 Forassets 0.042 0.160 0.000 0.000 Panel B: Statistical significance of differentials T-test P-value Wilcoxon stat |𝛾𝑖 | 19.900 <.0001 764 𝐵 -14.270 <.0001 1,299 |𝜉𝑖 | Geo_segments -22.180 <.0001 1,456 Forsales -17.950 <.0001 1,451 Forassets -15.870 <.0001 1,251 Panel C: Percentages of firms being multinational in subsamples Pension No-Pension 𝜒2 Based on Geo_segments 71.79% 41.59% 55.328 Based on Forsales 71.37% 41.12% 55.433 Based on Forassets 45.30% 19.86% 47.695

P75 0.735 2.543 5.000 0.448 0.185 1.361 1.388 2.000 0.109 0.000 P-value <.0001 <.0001 <.0001 <.0001 <.0001 P-value <.0001 <.0001 <.0001

Notes: In this table, we present summary statistics (Panel A) of multinationality variables for two subsamples of large U.S. firms. We define a multinationality variable as a firm characteristic that is commonly used to measure a firm’s level of multinationality. We report the number of geographic segments (Geo_segments) for which firms report operations in the Compustat Geographical Segments data base. Forsales is the foreign sales ratio, similarly Forassets is the foreign asset ratio of a firm. The two subsamples are: 1) Large firms that offer pension-based executive compensation — data on these 272 firms is used for the empirical analysis in this study, and 2) Large (428) firms that are omitted from this study because they either do not offer pensions or data was unavailable. Combined these two data subsets span the 700 largest U.S. traded firms. Panel B presents t-test for differences in average values and Wilcoxon Signed Rank (Median) test values for differences in medians with corresponding p-values. In Panel C, we classify firms as either multinational or domestic. This classification is made based on one of three criteria: 1) The number of geographic segments; 2) the foreign sales ratios, or 3) the foreign asset ratios. A firm is considered multinational if either Geo_segments is more than 1, the firm reported a positive foreign sales ratio, or the firm reported a positive foreign asset ratio during at least one sample year. The density of multinational firms is reported as percentages of firms that offer pensions (Pension) and firms that offer no pensions or data was unavailable (No-Pension). Statistical significance of differences is based on Pearson 𝜒 2 tests and corresponding p-values.

Page 41 of 43

Table 10 Pension plans, financial distress, and corporate multinationality Panel A: Percentage of large firms that offer pensions across financial distress quartiles Distance-to-default level

Average Dtd Percentage of firms with pensions

Lowest

Low-Mid

High-Mid

High

0.137 6.25%

1.799 24.74%

3.705 59.38%

6.485 79.17%

Panel B: Logistic regression results of the following model:

9

9

𝐷𝑢𝑚𝑚𝑦_𝑃𝑒𝑛𝑠𝑖𝑜𝑛𝑖𝑡 = 𝛽0 + 𝛽1 𝐷𝑡𝑑𝑖𝑡 + 𝛽2 𝐹𝑖𝑟𝑚𝑎𝑔𝑒𝑖𝑡 + 𝛽3 𝐷𝑒𝑏𝑡𝑒𝑞𝑖𝑡 + 𝛽4 𝐷𝑢𝑚𝑚𝑦_𝑀𝑁𝐶𝑖𝑡 + ∑ 𝛿𝑦 𝑌𝑒𝑎𝑟𝑦 + ∑ 𝛿𝑗 𝐼𝑛𝑑𝑗 + 𝜀𝑖𝑡 𝑦=1

[i]

[ii] ***

Dtd Firmage Debteq Dummy_ Geo_segments

0.0903 (6.86) 0.0574*** (13.25) -0.0276 (-1.23) 1.0921*** (4.59)

[iii] ***

0.0900 (6.86) 0.0574*** (13.26) -0.0309 (-1.27)

Dummy_Forassets

Observations Pseudo 𝑅 2 Year Variables Industry Variables

0.0849*** (6.59) 0.0577*** (13.16) -0.0351 (-1.42)

1.1864*** (4.96)

Dummy_Forsales

Constant

𝑗=1

-5.1375*** (-11.53) 18,101 0.2923 Yes Yes

-5.1960*** (-11.57) 18,101 0.296 Yes Yes

0.8888*** (4.29) -4.6762*** (-10.70) 18,101 0.2908 Yes Yes

Notes: In this table the sample contains annual data on the 700 largest U.S. firms, out of which 272 firms offer pensions. Data spans the period 2000-2009. In Panel A the 700 firms are sorted into quartiles based on their average Dtd measures, where “Lowest” contains firms with the lowest Dtd measures (riskiest firms). We report the percentages of firms for each distress quartile that offer pensions. In Panel B we present the results of a Logistic regression where the dependent variable indicates whether the firm offers pensions or not. Explanatory variables include Dtd (distance-to-default), Firmage (the age of the firm in years), Debteq (Debt-to-Equity ratio), and a dummy variable that classifies the firm as multinational or domestic based on one of three internationalization measures: 1) The number of geographic segments; 2) foreign sales ratios, or 3) foreign asset ratios. Robust z-statistics are reported in parentheses (standard errors are clustered by firm). The logistic regression models also use industry and year control variables. Statistical significance is indicated as follows: * p < 0.10, ** p < 0.05, *** p < 0.001.

Page 42 of 43

Table A1 Pension plan disclosure for TJX Companies, Inc., FY 2005

Final average earnings 100,000 150,000 200,000 300,000 400,000 500,000 600,000 800,000 1,000,000 1,200,000 1,400,000 1,600,000

Years of credited service at normal retirement 10 15 20+ 25,000 37,500 50,000 37,500 56,250 75,000 50,000 75,000 100,000 75,000 112,500 150,000 100,000 150,000 200,000 125,000 187,500 250,000 150,000 225,000 300,000 200,000 300,000 400,000 250,000 375,000 500,000 300,000 450,000 600,000 350,000 525,000 700,000 400,000 600,000 800,000

This pension benefit table is taken directly from the FY 2005 DEF-14A statement filed by TJX Companies on April 28, 2005, p.19. The Pension Plan Table section of the Definitive 14A provides the following information: “As of January 29, 2005, the years of service for the following executive officers under SERP are as follows: Mr. English, 22 years; Mr. Barron, 25 years; Mr. Campbell, 31 years; Mr. Maich, 20 years; Ms. Meyrowitz, 19 years; and Mr. Smith, 10 years.”

Page 43 of 43