*3.2. Measure of Active Management of Currency Risk*

Based on the above sample, we examine the firms' foreign currency risk management by examining their hedging strategy against the currency risk. To retrieve hedging data, we perform a keyword search for financial derivatives uses in 10-K filings compiled in the SEC EDGAR database. In this study, we focus on foreign currency derivatives. The foreign currency derivatives are most commonly used for hedging strategy by non-financial U.S. firms. Géczy et al. (1997) investigate the use of derivatives for a sample of Fortune 500 non-financial firms in 1993 and find that currency derivatives are used most frequently by corporation (52.1%). For each type of derivative, we use a set of relevant key words as specified above for the corresponding currency derivative instruments. When a keyword is found, we review the context in which the keyword appears in the report to confirm the use of derivatives for hedging and to collect hedging information. We use two proxies from literature to describe corporate

<sup>1</sup> We also follow Doukas and Pantzalis (2003) to consider the case that a firm might be subject to foreign currency risk due to the competitive environment. To do so, we include the keywords related to foreign currency risk and market risk in our textual searching program to read if firms explicitly state their foreign currency exposures in 10-K filings. We also perform the additional check by identifying MNCs by setting the ratios of foreign assets, foreign sales or foreign income greater than 10%. This classification is based on the requirements of the Statement of Financial Accounting Standard No. 14 (Financial Accounting Standards Board 1976), which defines a firm as a multinational company if it reports foreign assets and foreign sales ratios of 10% or more. Both ways show the qualitatively consistent results.

hedging activities. Following Nance et al. (1993), Géczy et al. (1997), and Chen and King (2014) we first use a dummy variable of hedging to represent if a firm implements hedging in a given year. The hedging dummy variable takes the value of one if a firm holds a hedging position using any types of foreign currency derivatives at the end of the fiscal year or has transactions involving one or more foreign currency derivatives for the purpose of hedging during that year, and zero otherwise. In addition, we use the notional amount of foreign currency derivatives in a given year to quantitatively capture the intensity of hedging activity.

#### *3.3. Measures of Pension Incentive*

As one of the main variables of interest for our study, we follow the literature to construct the measures of executive pension incentive. Sundaram and Yermack (2007) point out that retirement compensation and deferred compensation are two primary benefits generating CEO pension incentives. In the Execucomp database, the value of retirement compensation is defined to be the aggregate present (actuarial) value of the executive's accumulated benefits under the firm's pension plans, and deferred compensation is computed as the aggregate balance under the non-tax-qualified deferred compensation plan. Pension incentive motivates the executives to stand with the creditors to claim residual value during firm liquidations. Deferred compensation refers to the part of compensation which is deferred under the voluntary act of the executives to pay at pre-specified dates in the future. Such two types of compensation arrangements may work as the liability-like security to align CEO incentives with creditors and to induce less risky firm policies. Consequently, we expect a positive relation between hedging activity and executive pension incentive.

Although the equity-based components of CEO compensation have been documented in extant literature to cause risk-taking behavior, liability-based compensation is expected to lead to managerial risk aversion. Particularly, Jensen and Meckling (1976) conjecture that a compensation package with equal-weighted equity and debt instruments is superior to 100% equity compensation. In contrast, Edmans and Liu (2011) suggest granting executives equally weighted liability and equity compensation is typically inefficient. With the component of pension, the executives' wealth is aligned with both the incidence of bankruptcy and firm liquidation value, which makes them less incentive to transfer wealth from debtholders to stockholders and attenuates the stockholder-bondholder conflicts. However, on the other hand, executives paid with excessive pension compensation might engage in unnecessarily conservative policies and reallocate wealth from stockholders to debtholders. As a result, to quantitatively capture the incentive of liability-based compensation relative to equity-based compensation and also to consider the external influence of firm capital structure exposed on the executives, we form the two main proxies to gauge the relative magnitude of pension incentive. The first is relative pension leverage, which is defined as CEO pension leverage divided by firm leverage. We also adopt a dummy variable, which takes the value of one if the relative pension leverage is greater than one and zero otherwise, as a way to capture the possible non-linear relation between dominant liability-based incentives and managerial risk attitude. These measures are also suggested by Edmans and Liu (2011) and Wei and Yermack (2011). We speculate that firms with a higher relative pension leverage have a higher likelihood of hedging and a larger notional amount of hedging derivatives.
