Printer Friendly

Derivatives use, information asymmetry, and MNC post-acquisition performance.

We utilize a sample of US acquiring firms that engaged in international M&As to document the effects of corporate derivatives use on post-M&A long-term performance. We find that derivatives users outperform nonusers. Furthermore, we find that acquirers with derivative policies that are more comprehensive and sophisticated outperform those with less comprehensive and sophisticated policies. They, in turn, outperform acquirers with no existing policies in place. Our results are consistent with the notion that the use of derivatives lowers information asymmetry related agency problems. Furthermore, our evidence indicates that derivatives use is an important corporate activity that has a profound effect on post-M&A performance.

**********

There is substantial evidence to support the premise that multinational corporations (MNCs) underperform relative to purely domestic firms (Denis, Denis, and Yost, 2002). This can be partly attributed to the fact that their business operations are more complex than those of purely domestic firms (Doukas and Pantzalis, 2003; Kim and Pantzalis, 2003). In addition, since they have large amounts of foreign currency denominated revenues, geographically diversified firms are subject to significant exposure to currency fluctuations. Consequently, it is increasingly difficult for financial markets to correctly assess their values (Lin, Pantzalis, and Park, 2007a; DaDalt et al. 2002). This relatively elevated level of informational asymmetry can also increase the likelihood and magnitude of agency problems for MNCs, leading to reduced valuation.

Merger and acquisition (M&A), especially international M&A, activity is often related to agency conflicts. (1) Acquiring firms subject to more severe information asymmetry problems are more likely to overinvest, and are more likely to exhibit inferior risk-adjusted long-term returns than those without such problems. (2)

In this paper, we hypothesize that corporate hedging policy (proxied here by the use of derivatives) can play an important role in improving MNCs' long-term performance after the acquisition of international targets. We suggest that this is because the use of derivatives can reduce information asymmetry between managers and the financial markets. (3) Guay, Haushalter, and Minton (2003) present evidence indicating that the errors in analysts' forecasts and the dispersion in those forecasts are significantly correlated with unexpected shocks that are not transparent to investors or analysts. Additionally, firms' hedging strategies are related to this earnings uncertainty. Their findings provide empirical support for DeMarzo and Duffie's (1995) assumption that outsiders encounter difficulty interpreting the impact of risks, which corporations can hedge away. Dolde and Mishra (2007) demonstrate that geographically diversified firms use substantially greater amounts of foreign exchange derivatives than purely domestic firms. DaDalt et al. (2002) present evidence to support the premise that the use of derivatives is associated with lower information asymmetry. Therefore, acquirers using derivatives are expected to have lower information asymmetry related agency problems and better long-term performance.

This hypothesis is also explained by a catering theory, which argues that stock market mispricing might influence individual firms' investment decisions. In a recent paper, Polk and Sapienza (2009) find that overpriced firms tend to overinvest and that their subsequent stock returns are low. (4) gin, Pantzalis, and Park (2007a) demonstrate that corporate hedging policy alleviates the level of equity mispricing by improving the firm's transparency. Thus, it follows that the corporate use of derivatives may alleviate the overinvestment problem.

We analyze the long-term performance of acquirers of foreign targets by classifying them on the basis of their use of derivatives. We find that acquirers using derivatives exhibit significantly less information asymmetry (i.e., they are more transparent) and better long-term performance than acquirers that do not use any derivatives. We interpret these findings as consistent with the notion that managers of acquiring firms that use derivatives are less susceptible to agency conflicts and more likely to pick foreign targets according to the value-maximization principle than acquirers that do not use derivatives. Furthermore, we find that acquirers that employ more comprehensive and sophisticated policies of derivatives use outperform those with less comprehensive and sophisticated policies. They, in turn, outperform acquirers that do not use any derivatives.

The rest of the paper is organized as follows. In the next section, we describe data sources and sample selection. Section II describes empirical methodologies and reports univariate and multivariate test results. Section III includes a summary and concluding remarks.

I. Data and Sample Selection

We collect derivative usage data for all nonfinancial corporations listed in the Database of Users of Derivatives published by Swaps Monitor Publications, Inc. over the 1992-1996 period. (5) The database compiles information that firms are required to report according to SFAS 105. SFAS 105 requires firms to report information about financial instruments, such as forwards, futures, options, swaps, etc., that have off balance sheet risk. In the majority of our tests, we use an indicator variable that takes the value of one if the firm used derivatives and zero otherwise. In our tests of users' policy characteristics, we also use the number of different derivative contracts the firms used and the notional amount of derivative contracts. Swaps Monitor lists the notional dollar amounts for seven different contracts spanning two general types of derivatives: 1) interest rate (IR) derivatives and 2) foreign exchange (FX) derivatives. The seven different contracts are: 1) IR-options, 2) IR-swaps, 3) IR-forwards/futures, 4) FX-options, 5) FX-swaps, 6) FX-futures, and 7) FX-forwards. The database contains information for 1,698 firms that list notional amounts of over-the-counter and exchange-traded currency derivatives outstanding at period end.

Our M&A sample consists of M&As of non-US target firms by US acquirer firms reported by the Securities Data Corporation (SDC). We require that: 1) the announcement date lie between January 1, 1992 and December 31, 1996; 2) acquirer firms be public; and 3) self-tender offers, repurchases, and rumored deals be excluded. The SDC sample compiled based on these requirements consists of 2,852 US firms' international M&As. Our sample drops to 1,020 observations after merging with Swaps Monitor's 1,698 firms.

We extract analyst coverage data from the Institutional Brokers' Estimate System (I/B/E/S) database. We select the number of fiscal year-end analyst forecasts issued in June of each year for all stocks covered by security analysts. All valuation and analyst coverage measures used in the study are aligned on the month of June (as in Fama and French, 1992, 1993). Also, we include transactions only if the acquirer's financial and accounting information is available. Financial data are collected from Compustat, and price and return data from the Center for Research in Security Prices (CRSP). To construct the estimation period, we require that all firms have price and return information in the CRSP database over the 60 months prior to the M&A announcement.

The final sample, after combining I/B/E/S, Compustat, and CRSP information, contains 450 firm-year observations. This is a sample of firms with substantial international operations providing the basis for a good natural experiment in measuring the impact of their derivative usage policies, both financial and currency, on post-merger performance. Table I reports the distribution of users and nonusers in each category by year and region. Clearly, in this sample of US MNCs engaging in international M&As, a majority of firms (over two-thirds) use derivatives. This pattern is observed in all years, as well as in all regions.

II. Empirical Results

A. MNCs' Derivative Usage Policy

Belote we investigate the effect of corporate use of derivatives on long-term post-M&A performance, we document, in Table Il, the summary statistics based on the indicator variable of derivative usage (in Panel A) and notional dollar amount of derivatives (in Panel B) used by our sample firms. Sixty-eight percent of the firms in our sample use derivatives. In particular, 59% of the firms use FX derivatives, while 35% use IR derivatives. This confirms our expectation that these are firms that are heavily dependent on their foreign operations, thus the need for currency risk management. The average dollar amount of FX derivatives is also larger than that of IR derivatives. Within the IR derivatives, swaps account for the majority, indicating the long-term nature of IR risk management. In contrast, FX derivative usage is spread out across the different instruments, possibly indicating a more complex/sophisticated approach in FX risk management.

B. Measures of Information Asymmetry

To measure the level of information asymmetry between insider managers and outside investors, we utilize the following five measures to construct a composite index of information asymmetry:

1. Institutional ownership (INSTP): Because institutional investors often have close communication links with management, they can be viewed as informed investors with monitoring capabilities (Brickley, Lease, and Smith, 1988) that can reduce information asymmetry.

2. Number of analysts' forecasts (NAF): This measure of the extent of security analyst coverage has been used by numerous studies (e.g., Hong, Lim, and Stein, 2000) as a proxy of informational asymmetry on the premise that diffusion of information increases with the number of analysts covering a firm.

3. Absolute forecast error (FE): This variable is used here as a proxy for the predictability of future earnings and also serves as a proxy of information asymmetry as was done in several prior studies (Christie, 1987; Atiase and Bamber, 1994). FE is defined as follows:

FE = [absolute value of FMD - A]/[absolute value of FMD], (1)

where [absolute value of FMD - A] is the absolute value of the difference between the median forecast (FMD) and the actual earnings per share (A).

4. Dispersion of analysts' earnings forecasts (DISP): This measure is computed as follows:

DISP = FSD/[absolute value of FMD], (2)

where [absolute value of FMD] is the absolute value of the median forecast, and FSD is the standard deviation of one-year-ahead forecasts. Even though DISP reflects both diversity of analyst beliefs and the uncertainty (lack of precision) of analyst forecasts (Barron, Kim, Lim, and Stevens, 1998), it has been used as a measure of information asymmetry in several previous studies (D'Mello and Ferris, 2000; Krshnaswami and Subramaniam, 1999). We constructed NAF and DISP from security analysts' one fiscal year ahead forecasts issued every June and extracted from the I/B/E/S Summary Database.

5. Standard deviation of residual returns of the market model (SDRES): This idiosyncratic risk variable serves as a proxy for the amount of firm-specific information not shared by the market, but in the hands of insiders. We compute it by first regressing the firms' daily stock returns on market returns over the estimation period (i.e., from days -252 to -46 in calendar year y - 1) and obtaining residual returns computed as the difference between actual returns and estimated returns. Then, we calculate the standard deviation of daily residual returns over each calendar year y. This measure has been used in Bhagat, Marr, and Thompson (1985) and Blackwell, Marr, and Spivey (1990), among others.

To fully utilize all the information available from these five measures and at the same rime alleviate the impact of outliers, we create an information asymmetry index (INFO_ASYMM) by combining all the information measures used above. First, we rank each firm in the sample by the magnitude of each of the information variables above. Then, we compute the average of all the ranks of the five different information measures as our information asymmetry index. This composite measure has the advantage of balancing all of the information measures while incorporating each of them and is defined as follows: (6)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], (3)

where [Rank.sub.k]([INFO.sub.ik]) is the rank function that assigns rank for each observation from the least asymmetric to the most asymmetric information, [INFO.sub.ik] is the kth measure of information symmetry (FE, DISP, SDRES, and the inverse of INSTP and NAF) for firm i in our sample, and K is the dimensions of information measures. For each information variable, the firm with the least information asymmetry (highest value of FE, DISP or SDRES, or highest inverse value of INSTP or NAF) is ranked as [N.sub.k] while the firm with the most information asymmetry is ranked as one. The denominator, [K.sub.i], averages the ranks regardless of the number of information values of the firm in the sample. For instance, the firm that has only three information measures in records is divided by [K.sub.i] = 3. A firm with all five measures is divided by [K.sub.i] = 5. This construction scales the variable INFO_ASYMM to a value between zero (least information asymmetry) and 1 (most information asymmetry). (7)

C. Does Derivatives Use Reduce the Level of Information Asymmetry?

As shown in Dolde and Mishra (2007) and DaDalt et al. (2002), corporate hedging alleviates information asymmetry. Due to the higher degree of uncertainty and increased information asymmetry in international M&As, these acquirers are firms with the greatest potential need and potential gain from derivatives use. In this subsection, we directly test whether the use of derivatives is negatively associated with information asymmetry.

Table III reports univariate tests for all measures of information asymmetry based on whether the firms are users or nonusers. The information asymmetry index is significantly lower for derivatives users than for nonusers. Individual measures of information asymmetry also show consistent results. On average, users have a higher percentage institutional ownership and analyst following and lower idiosyncratic risk than nonusers. These differences are statistically significant, while the differences in forecast error and dispersion are not statistically significant.

In Table III, we also compare users and nonusers of derivatives in terms of other variables describing M&A transactions and acquirer firms. As documented in other studies, the users of derivatives are larger than nonusers. However, the M&A transaction characteristics are generally not significantly different between the two subsamples, indicating that both users and nonusers pursue similar targets. The only variable with significant difference is Focus, an indicator variable that signifies focus-increasing M&As. Acquirers using derivatives are more likely to pursue diversifying M&As than those that are nonusers. This finding is consistent with Lin, Pantzalis, and Park (2007b) who report evidence that hedging reduces the negative valuation effect of corporate diversification.

D. Long-Term Performance of US Acquirer Firms

As mentioned earlier, corporate policy on derivative usage can have an impact on firm value. We expect that for firms engaging in international M&As this valuation impact will be realized in long-term post-merger performance. To test this hypothesis, we examine the long-term risk adjusted stock performance of US acquirer firms after foreign takeover announcements. We examine abnormal long-term performance by employing two different methods in computing expected returns: 1) the market model and 2) Fama and French's (1993) three-factor model (the FF model). Both use the same estimation (-60 to -1) and test (0 to +60) windows in months. We employ conventional event-study methodology in computing abnormal returns (ARs) and cumulative abnormal returns (CARs). Based on the prior literature (Kothari and Warner, 1997; Loughran and Ritter, 2000), we use several methodologies and examine whether the long-run performance results remain consistent when we measure performance in different ways. (8)

Table IV reports monthly cumulative average abnormal returns (CAARs) for various windows after international M&A announcements. When we do not differentiate between derivatives users and nonusers, we observe a statistically significant negative performance of about 10% up to 48 months post-merger (-6.7% after 36 months using the FF model) confirming earlier evidence that acquirer firms' performance tends to lag for a substantial period after merger. The average magnitude of underperformance in our sample is similar to that reported in earlier studies (Agrawal, Jaffe, and Mandelker, 1992; Rau and Vermaelen, 1998). When we separate derivatives users from nonusers, we observe two distinct results. First, in Panel A, although the CAARs in the first 24 months after the announcements are less negative for users than for nonusers, the differences in the negative CAARs are statistically insignificant, indicating that both groups of firms experience about the same level of sub-par performance. Second, beyond 24 months, users substantially outperform nonusers. The 36-, 48-, and 60-month CAARs are all significantly higher for users than for nonusers. For example, the 60-month CAAR differences are, 23.32% and 22.01% based on the market model and the Fama and French three-factor model, respectively. These differences are not only statistically significant, but also economically sizeable. Furthermore, when we examine CAARs over time, the negative performance observed over the first 24 months after the announcement becomes insignificant for the users over the 36- and 48-month windows, and finally turns significantly positive over the 60-month window. Conversely, the post-merger performance for nonusers stays negative and is significant for all windows up to 60 months after the announcement. This indicates that users' performance rebounds after 24 months, while nonusers' performance does not.

Next, we examine whether and how derivative use policy characteristics affect long-term performance. The derivatives use policy characteristics variables are: 1) NTYPE, which captures comprehensiveness of derivative usage by taking the value of zero if the firm does not use derivatives, the value of one when the firm uses only FX derivatives or only IR derivatives, and the value of two if the firm uses both types of derivatives; 2) NCONTR, which captures sophistication of derivative usage by taking the number of different derivative contracts (ranging from 0 to 6) the firm uses; and 3) AMT, total notional dollar amount of derivatives used.

The results in Panel B through Panel D demonstrate that a more expansive and sophisticated policy has a more pronounced effect on long-term performance. In Panel B, acquirer firms are classified based on NTYPE. Group I represents firms using both FX and IR derivatives, while Group 2 represents those using either FX or IR. Group 3 includes only nonusers. After 60 months, comprehensive users (firms using both FX and IR derivatives) outperform one type users by over 22% (18% using the FF model) and nonusers by over 37% (33% using the FF model), all statistically significant. Furthermore, the differences between comprehensive users (Group 1) and nonusers (Group 3) are larger than the differences between users and nonusers documented in Panel A. This clearly indicates that more comprehensive derivatives use policies are associated with better firm performance.

In Panel C, we group acquirer firms based on the number of different derivatives contracts (NCONTR) they use. In our sample, the maximum number of contracts is six. We classify firms using four to six different contracts (i.e., more sophisticated firms) in Group 1, firms using one to three different contracts (i.e., less sophisticated) in Group 2, and nonusers in Group 3. We observe a stark difference in the performance of sophisticated users (firms using four to six contract types). Their CAARs are continuously and increasingly positive after the merger and, on average, up to +37% (27% using the FF model) after 60 months. In contrast, less sophisticated users, as well as nonusers, show negative CAARs throughout the five-year period after merger. Confirming earlier results, sophisticated users outperform less sophisticated users, who, in turn, outperform nonusers. The mean differences in Panel C are the largest in all four panels. This suggests that a sophisticated derivatives using policy entailing the utilization of as many derivatives' contracts as possible is the best risk managing strategy in terms of enhancing long-term performance. In addition, as reported in Table II, only a small percentage of firms use IR-options (5.3%) and IR-forward/futures (0.7%). This fact, coupled with the evidence in Table IV, implies that firms using these instruments are likely to be characterized by greater sophistication in derivative usage. The 60-month performance difference between sophisticated users and nonusers is almost 55% (almost 39% using the FF model), clearly substantial in economic terms, as well.

Lastly, we classify firms into three subgroups based on the notional dollar amount of derivatives. Group 1 consists of firms with notional dollar amount of derivatives that is higher than the sample median. Group 2 consists of firms with notional amount of derivatives that are positive, but lower than the sample median, and Group 3 consists of nonusers. As seen in Panel D, while the CAAR differences between Group 1 and Group 2 are mostly positive, they are not statistically significant. The differences between firms with high notional dollar amounts of derivatives and nonusers are significant for the longer windows, but the magnitudes are smaller than in the other panels. Nonetheless, the evidence in Panel D is consistent with the findings in the other panels.

In Figure 1, we plot and compare CAARs up to 60 months after the international M&A announcements of different groups of firms. There are clear performance differences associated with the degree of comprehensiveness and sophistication in derivative usage (Panels A to D). Table IV and Figure 1 both demonstrate that: 1) derivatives use policy characteristics are important factors in explaining long-term post-M&A performance, and 2) the more comprehensive and sophisticated the derivatives use policy, the more positive is the firm's long-term performance. (9)

[FIGURE 1 OMITTED]

E. Multivariate Analyses

To further examine the effect of derivative usage policy characteristics on US MNCs engaging in international M&As, we employ a cross-sectional regression model. We regress long-term performance, CAR(O, +60M), on derivatives use variables, transaction variables, and acquirer firm characteristics as outlined in the following model

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], (4)

where i indexes firms, and y is a yearly time index. CAR(O,+6OM) is the cumulative abnormal return over the 60-month window from event (announcement) day. The abnormal return (AR) of each month is computed as actual return minus estimated return based on Fama and French's (1993) three-factor model estimated in the estimation period. Our key independent variable DERIVATIVES represents the different characteristics of derivative usage, which include: 1) use of derivatives indicator (USER), 2) number of types of derivatives (NTYPE), 3) number of different derivative contract types (NCONTR), and 4) the notional dollar amount of derivatives (AMT). For AMT, we also consider a potential curvi-linear effect of the notional dollar amount of derivatives on valuation by including both the first and the second order terms (AMT and [AMT.sup.2]) in the regression model. The rationale for including both first- and second-order terms of AMT in the model is based on the possibility that high AMT values are indicative of speculation in derivatives or managerial hubris in their ability to use derivatives. This possibility implies that it is likely that beyond a certain range in notional amount, the use of derivatives may have a detrimental (negative) effect on firm valuation. Therefore, we expect a positive coefficient for AMT and a negative coefficient for [AMT.sup.2]. In addition to policy characteristics, the model also controls for other factors that may impact the magnitude of acquirers' long-term performance after the announcements. Examining the method of payment choice, Travlos (1987) and Franks and Harris (1989) find that cash-financed, as opposed to stock-financed, acquisitions are associated with higher bidder gains. We control for the method of payment by including a dummy variable, CASH, which takes the value of one if the acquisition is financed entirely with cash and the value of zero otherwise. In addition, we include a relative size (RELSIZE) control variable which is equal to the transaction size, measured as the value of the transaction divided by the acquirer's size, measured by the book value of total assets. If the US bidder has not acquired 100% of the target's shares, it is likely that the overall effect on abnormal returns would be weakened. In our sample, the average percent of shares acquired by US bidders is 85%. Seventy-four percent of bidders acquire the full 100% of shares. The dummy variable SHAREIO0 is set to one for these bidders, and zero for bidders who acquire a majority, but less than 100% of shares. M&A literature has noted that related acquisitions (i.e., increasing focus by acquiring firms in same industry) are associated with higher bidder shareholder gains (Doukas and Travlos, 1988; Markides and Ittner, 1994; Comment and Jarrell, 1995; Eun, Kolodny, and Scheraga, 1996). We include FOCUS, a dummy variable that takes the value of one if the acquirer and target firms are in the same four-digit SIC code industry, and the value of zero if they are not in the same industry. We also control for acquirer firms' characteristics such as size, book-to-market ratio, and leverage. SIZE is the NYSE decide ranking of the acquirer's total market value of equity, and BM is the decile ranking of the acquirer's book-to-market ratio (computed as in Fama and French, 1992). LEV is total long-term debt normalized by total assets. These variables are important as past studies have shown that manager's hubris is associated with size and book-to-market ratio. The hubris hypothesis argues that the manager may convince himself/herself that market valuation is wrong and the manager's value estimate is correct. This is more likely to be seen in glamour (larger) firms rather than in value firms (smaller). Rau and Vermaelen (1998), for example, find that the poor post-acquisition performance for bidder firms is driven by low book-to-market (i.e., glamour) firms.

Table V indicates that use of derivatives has an economically and statistically significant, positive effect on post-M&A performance. The USER, NTYPE, and NCONTR variables have significant positive coefficients after controlling for transaction variables and acquirer characteristics. The dollar amount of derivatives (AMT) is also positively related to the long-run performance, although statistically insignificant. There is also indication of a concave relation between AMT and CAR, but again the relationship is not statistically significant.

Similar to prior studies, the cash payment variable's coefficient is positive in all five models, but not statistically significant. Relative size is positively related to bidder wealth gains at the 1% level. This result is consistent with previous studies that found target companies that are large in relation to their acquirers are able to provide greater synergies in mergers than smaller targets can offer (Asquith, Bruner, and Mullins, 1983; Bruner, 1988; Song and Walkling, 1993). Furthermore, since the transaction value includes the premium paid for the target firm, this finding suggests that bidder shareholders do not perceive that the bidder has overpaid for the target on average. High book-to-market acquirers (i.e., managers are less likely to have hubris) demonstrate better performance in the long-run period, confirming the results in Rau and Vermaelen (1998).

F. Robustness Tests

In this subsection, we conduct several robustness checks, which endeavor to determine whether the findings in Table V are due to the particular model used to estimate long-run performance or to the estimation methodology used.

First, we consider the issues of sample selection bias, endogeneity, and causality. It is possible that our sample is subject to a significant large firm selection bias since the information of derivatives use is more likely to be available for larger firms covered in the Swaps Monitor database. Additionally, the relationship between derivatives use, firm size, and information asymmetry may be endogenous. That is, larger firms are more likely to use derivatives because of a more favorable cost-benefit ratio and firms with high information asymmetry may have stronger incentives to use derivatives.

We utilize two-stage and three-stage models in order to control for problems associated with sample selection bias and endogeneity. The literature considers both models as appropriate methods to resolve such problems. In particular, the three-stage model that addresses both selection bias and endogeneity has been widely used in econometrics literature (Mroz, 1987; Leung and Yu, 1996; Wooldridge, 2002). l0 In the first stage of the three-stage model, known as the selection equation, we use a probit model to estimate the possibility of an observation being included in the sample. In addition to firm characteristics such as size, book-to-market, and leverage, we consider additional factors such as exchange listing indicators and an S&P 500 indicator. The rationale for using these variables in the selection equation is that firms belonging to a major index and/or listed in a particular exchange may be more likely to be covered in the Swaps Monitor database, and thus included in our final sample. Il Using all available information from Compustat, we create a pooled data set for the sample period 1992-1996, that contains 20,737 firm-year observations with all six explanatory variables in Equation (5). Then, we combine this sample with the derivative data set from the Swaps Monitor. The dependent variable, SAMPLE, in the first-stage model is a dummy that indicates whether an observation is included in the final sample. (12)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], (5)

where SP500 dummy takes the value of one if a firm is included in the S&P 500 index, and zero otherwise. AMEX and NASDAQ dummies indicate the firm's primary exchange. In the second stage, the reduced-form regression, we estimate the derivatives use variable (i.e. derivatives use policy characteristics USER, NTYPE, NCONTR, or AMT) that will be used as a key independent variable in the structural equation.

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. (6)

Based on prior studies' evidence regarding the determinants of derivatives use, we use seven explanatory variables along with the inverse of Mills ratio (MILLS, the probability that the firm is in our sample as estimated in stage one) to estimate the use of derivatives. First, we control for SIZE as large firms are likely to use derivatives given the fact that they face a more favorable cost-benefit ratio. Book-to-market (BM) is included based on Froot, Scharfstein, and Stein (1993) and DeMarzo and Duffie (1995) who argue that growth firms use more derivatives. (13) Hedging can reduce the probability of financial distress by reducing the volatility of cash flows. Therefore, it is expected that derivatives use can allow the firm to use more debt (LEV). We consider the firm's interest coverage ratio (INTCOV) that is computed as EBIT divided by interest expense. If the firm has relatively large interest expense (i.e., low INTCOV), it has an incentive to use derivatives. The nature of a firm's operations can also influence the level of derivatives used. Dolde and Mishra (2007) report that both the decision regarding whether to use derivatives and the extent of their use are significantly related to firm complexity. We use the number of segments (NSEG) to control for the firm's operational complexity. Smith and Stulz (1985) theoretically prove that if a firm's tax curve is convex, it is more likely to hedge and, thus, to increase firm value (i.e., after-tax cash flows). Accordingly, we also include the tax rate (TAX) as an additional independent variable. The information asymmetry between managers and investors can affect the firm's decision to use derivatives. We expect that a firm that suffers from high levels of information asymmetry is more likely to use derivatives. (14) Given this, we include INFO ASYMM and expect a positive relation with the use of derivatives. Finally, the inverse Mills ratio from the selection equation is included in Equation (6) to control for possible selection bias.

In the third stage, the structural equation includes controls for the estimated derivatives use variable from the second stage (reduced-form regression), the inverse of the Mills ratio from the first-stage (selection regression) equation, and the other existing independent variables as outlined in Equation (7) below.

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. (7)

In addition to the three-stage model, we also estimate a two-stage model that controls only for endogeneity. In this model, we omit the first stage (Equation (5)) that estimates the probability that a firm is selected in the sample, and estimate only the regressions in the second stage (Equation (6)) and the third stage (Equation (7)). Table VI indicates the results of the two-stage and the three-stage models. We find that our previous results are not altered, even after we control for sample selection bias and/or endogeneity. The different characteristics of derivatives use policy are still found to be associated with superior MNCs' post-performance. Overall, the results are consistent with the ones obtained from the OLS models in Table V. Moreover, the magnitude and significance of the derivatives use variables' coefficients is generally enhanced relative to those in the OLS models.

We also conduct several additional robustness tests that we include in the Appendix. First, we examine the issue of causality since it is possible that good performance motivates derivatives use rather than the use of derivatives really improving performance. If this is correct, the use of derivatives can be merely an indicator of better performance. Our additional test results imply that although derivatives use is related to operational performance, the derivatives use effect on stock performance exists above and beyond an operational performance effect. We also test whether our results are sensitive to the method used to estimate long-run performance. After reestimating our regression models using buy-and-hold returns based on the Daniel, Grinblatt, Titman, and Wermers (1997) methodology to identify a matched firm for each firm in our sample, we find similar results to the ones presented here. Finally when we reestimate our models separately for high and low information asymmetry subsamples, we find significant results only for the high information asymmetry group. These results further highlight the larger performance benefit from derivatives use for firms subject to high levels of information asymmetry.

III. Summary and Conclusions

This study investigates the effect of derivative usage policy on the long-term post-acquisition performance of US MNCs engaging in international M&As. Focusing on a sample of firms for whom there is a clear need for currency and financial risk management, we are able to document pervasive and consistent evidence that the comprehensiveness and sophistication in the policy of derivatives use are associated with different post-merger performance for these firms. We find strong evidence that comprehensive and sophisticated practice of derivatives use has a positive impact on acquiring firm performance. More comprehensive and sophisticated users outperform their less comprehensive and sophisticated counterparts, while nonusers perform the worst. In particular, firms using more than three derivative contracts actually outperform the market throughout the 60-month window after the merger announcement. This evidence is in stark contrast to that of earlier studies reporting post-acquisition underperformance for acquirer firms. We attribute this positive value impact of corporate use of derivatives to the associated reduction in information asymmetry and related agency problems, as well as other well documented positive effects of risk management on firm valuation. We also provide direct evidence of a stronger valuation effect from derivatives use when firms have higher information asymmetry, evidence of the information related benefit from derivatives use.

Our findings provide new evidence regarding the effect of financial risk management on firm performance. We also contribute to the M&A literature by asserting that the use of derivatives is an important corporate activity that has a profound effect on post-merger performance. Our findings also complement the recent literature concerning the effect of market valuation on corporate real activities.

Appendix

In this appendix we outline several additional robustness tests that are not included in the paper. First, we examine the issue of causality since it is possible that good performance motivates derivative use rather than the use of derivatives really improves performance. If this is correct, the use of derivatives can be just an indicator of better performance. We employ several tests to determine the cause-and-effect relation. For example, we test the relation between operating performance and stock returns in the post-merger period. In the cross-sectional model, we control for two performance variables instead ofderivative variables. First, GOODPERFORM is a dummy variable that takes the value one ifa firm's average ROA over the post-merger period (y toy + 5) is above the median value of sample firms, and otherwise zero. Second, PERFORM is a continuous variable and is simply a firm's average ROA over the saine period. We find that the coefficient of GOODPERFORM is 0.170 with a t-statistic of 1.17, which is consistent with the positive impact of derivative use but is weaker than the magnitude of effect of the derivatives user dummy variable (USER), which was 0.228 and statistically significant in Table V. We also find that the pattems of CARs are similar to the ones in Figure 1. However, the differences in CAARs between acquirers with better operating performance and the ones with poor operating performance are smaller than the differences in CAARs between users and nonusers in Figure 1. Overall, these results imply that derivative use is related to operational performance but the characteristics of derivatives use policy are more strongly related to stock performance, that is, we find that the derivative-usage effect exists above and beyond a performance effect.

Next, as discussed above, we test whether our results are sensitive to the measurement methodology or not. Many authors have argued that Fama and French three-factor model is not able to completely explain cross-sectional stock returns. However, we expect that if there are significant abnormal returns, most estimating models should provide similar results. In order to test whether this is the case or not, we measure long-term performance in a different way by using buy-and-hold returns and re-estimate the regression models. Following the triple-sorting method of Daniel et al. (1997), we find a matched firm for each sample firm in terms of size, book-to-market ratio, and momentum. In each year, we rank all firms listed on the NYSE, Amex, and Nasdaq based on three characteristics. We classify every firm into decile rankings and then match each sample firm with similar firms based on size and book-to-market ratio. After choosing similar firms, we compare momentum at the time of M&A announcements, based on the cumulative returns over the six-month pre-acquisition period. We choose the firm whose stock shows the smallest absolute difference in cumulative returns in that period relative to our firm's returns, as the matched firm. We compute buy-and-hold abnormal returns, BHARs

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (A1)

where i indexes firms, and d (y) is a daily (yearly) time index. [R.sub.i,d] is firm i's daily return on day d, and [R.sub.match.d] is daily return of firm i's matched firm. [tau] denotes the number of days since the first day after the announcement. Then, we use BHAR as dependent variable in the regressions.

The results of these robustness checks are reported in Table AI. We find that all regressions show a consistent pattern in the derivative-usage variables. Corporate policy of derivative use is positively and significantly associated with long-term buy-and-hold abnormal returns. Therefore, our several robustness tests confirm the previous results and suggest that our results are not sensitive to the particular model and to the measurement of long-term performance.

Earlier univariate evidence indicates a role of information asymmetry, as associated with the use of derivatives, in post-merger long-term performance. This section provides additional multivariate evidence to test for the role of information asymmetry. We run regressions for two different groups of firms: high- versus low-information-asymmetry firms. Acquirer firms are classified into high (low) information asymmetry if the firms are ranked higher (lower) than the median of INFO ASYMM. Recall that INFO ASYMM is a comprehensive variable computed by averaging the ranks in five measures of informational asymmetry: inverse value of institutional ownership (INSTP), inverse value of number of analysts' forecasts (NAF), forecast error (FE), dispersion of forecasts (DISP), and standard deviation of residual returns (SDRES).

The results are reported in Table AII and confirm that derivative use has generally a positive effect on post-merger performance. It is instructive to note that the coefficients are uniformly larger in the regressions for the sample of firms that have high levels of information asymmetry. The differences in coefficients of DERIVATIVES variables are larger and statistically significant when we use the derivative policy characteristics that denote comprehensiveness (NTYPE) and sophistication (NCONTR) in the policy. These results further highlight the negative association between information asymmetry and the derivatives use policy, and the larger performance benefit from derivative use for firms suffering from high information asymmetry to start with. The difference in stock performance between high- and low-information-asymmetry groups is also well evident in Panels (e) and (f) of Figure 1. For the acquirers whose asymmetric information is high, the effect of derivative use on stock performance can add up to an increase of up to 49% over 60 months relative to the case where no derivatives are used. However, we fail to find any distinct difference in stock returns for acquirers whose asymmetric information is low. For these firms, all of the DERIVATIVES variables are uniformly insignificant. That is to say, derivative use as a corporate activity loses its function in generating positive performance (via reducing information asymmetry and associated agency costs) when there is low information asymmetry already!
Table AI. The Effect of Derivative Use on Long-Term Stock Performance:
Using Buy-and-Hold Returns

This table reports estimates of coefficients of OLS regressions

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII],

where BHAR(0, + 60M) is buy-and-hold abnormal return over sixty months
after the announcement.

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII],

where i indexes firms, and d (y) is a daily (yearly) time index.
[R.sub.i,d] is firm i's daily return on day d, and [R.sub.match,d] is
daily return of firm i's matched firm. [tau] denotes the number of
days since the first day after the announcement. DERIVATIVES is
alternatively one of the following: USER is a dummy variable that
takes value one if a firm uses any kind of derivatives against foreign
exchange risk or interest rate risk, and otherwise zero; NTYPE is the
number of types of derivatives used, that is, FX and/or IR, or none
and thus takes values from zero to two; NCONTR is the number of
different derivative contracts used by the firm and based on Swaps
Monitor's database it can take values from zero to seven; and AMT is
the total dollar amount of derivatives used. CASH is a dummy variable
that takes the value of one if the acquisition is financed entirely
with cash, and zero otherwise. RELSIZE is equal to the ratio of
transaction value to acquirer's total assets. SHARE100 is a dummy
variable that takes one when 100% of shares are acquired through M&As.
FOCUS is a dummy variable that indicates whether the US acquirer and
non-US target firms are in the same four-digit SIC code industry. SIZE
(BM) is the decile ranking of acquirer's total market value of equity
(book-to-market ratio), following Fama and French (1992). LEV is the
ratio of long-term debt to total assets.

 [1] [2] [3]

USER 1.066 **
 (2.41)
NTYPE 0.643 *
 (1.97)
NCONTR 0.468 **
 (1.99)
AMT

[AMT.sup.2]

CASH 0.474 0.454 0.405
 (0.69) (0.68) (0.58)
RELSIZE -0.248 -0.152 -0.098
 (-0.96) (-0.64) (-0.43)
SHARE100 -0.056 -0.122 -0.104
 (-0.13) (-0.28) (-0.24)
FOCUS -0.370 -0.310 -0.25
 (-0.84) (-0.70) (-0.58)
SIZE -0.173 -0.195 -0.236 *
 (-1.48) (-1.57) (-1.97)
BM -0.002 -0.008 0.012
 (-0.01) (-0.05) (0.08)
LEV 0.960 0.928 0.986
 (0.78) (0.71) (0.80)
Intercept 0.131 0.492 0.688
 (0.11) (0.38) (0.54)
[R.sup.2] 9.02% 8.44% 10.34%
F-stat. 1.73 1.80 * 2.03 *
[Prob. >F] [0.102] [0.086] [0.051]

 [4] [5]

USER

NTYPE

NCONTR

AMT 2.28 x [10.sup.-4] * 0.001 *
 (1.72) (1.81)
[AMT.sup.2] 3.90 x [10.sup.-8]
 (-1.41)
CASH -0.412 -0.444
 (-0.84) (-0.90)
RELSIZE -0.250 * -0.228
 (-1.76) (-1.56)
SHARE100 -0.096 -0.100
 (-0.20) (-0.21)
FOCUS -0.323 -0.333
 (-0.67) (-0.69)
SIZE -0.205 * -0.249 **
 (-1.69) (-2.03)
BM -0.107 -0.105
 (-0.73) (-0.69)
LEV 1.450 1.224
 (1.05) (0.83)
Intercept 1.827 2.111
 (1.44) (1.65)
[R.sup.2] 13.17% 14.75%
F-stat. 1.89 * 2.15 **
[Prob. >F] [0.075] [0.036]

 ** Significant at the 0.05 level.

 * Significant at the 0.10 level.

Table AII. Information Asymmetry and the Effect of Derivative Use on
Long-Term Stock Performance

This table reports estimates of coefficients of OLS regressions

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

The model is separately tested for high-and low-information-asymmetry
firms. INFO_ASYMM is computed as averaging the ranks in five measures
of informational asymmetry: inverse value of institutional ownership
(INSTP), inverse value of number of analysts' forecasts (NAF),
forecast error (FE), dispersion of forecasts (DISP), and standard
deviation of residual returns (SDRES). INSTP is percentage of shares
owned by institutional investors. NAF is number of forecasts issued in
June of McYR (I-B-E-S Summary Data). FE is the absolute value of the
median forecast error, computed as the difference between the median
one-year-ahead EPS forecast and the actual EPS. DISP, the dispersion
of forecasts, is computed as the standard deviation of the one-year-
ahead forecasts divided by the absolute value of the median forecast.
FE and DISP are measured in June of each year. SDRES is the standard
deviation of residual returns of the market model. Acquirers are
classified into high (low) information asymmetry if the firms are
ranked higher (lower) than the median of INFO_ASYMM. The cumulative
abnormal return of firm i from the announcement to the 60th month,
[CAR.sub.i,y] (0, +60 M) = [[summation].sup.60.sub.m=0] [AR.sub.i,m]
where [AR.sub.i,m] is the abnormal return on firm i in the month m,
and is computed as the difference between actual monthly return and
the monthly return that is estimated by Fama and French's (1993)
three-factor model. DERIVATIVES is alternatively one of the following:
USER is a dummy variable that takes value one if a firm uses any kind
of derivatives against foreign exchange risk or interest rate risk,
and otherwise zero; NTYPE is the number of types of derivatives used
i.e., FX and-or IR, or none and thus takes values from zero to two;
NCONTR is the number of different derivative contracts used by the
firm and based on Swaps Monitor's database it can take values from
zero to seven; and AMT is the total dollar amount of derivatives used.
CASH is a dummy variable that takes the value of one if the
acquisition is financed entirely with cash, and zero otherwise.
SHARE100 is a dummy variable that takes one when 100% of shares are
acquired through M&As. FOCUS is a dummy variable that indicates
whether the US acquirer and non-US target firms are in the same four-
digit SIC code industry. SIZE (BM) is decile ranking of acquirer's
total market value of equity (book-to-market ratio), following Fama
and French (1992). LEV is the ratio of long-term debt to total assets.
Estimated coefficients are divided by 100.

 Use of Derivatives
 (USER)

 High Info. Low Info. Coeff.
 Asymm. Ssymm. Diff.

USER 0.367 * 0.105 0.262
 (1.78) (0.88) (1.10)
NTYPE

NCONTR

AMT

[AMT.sup.2]

CASH 0.128 -0.008 *** 0.136
 (0.53) (-0.06) (0.48)
RELSIZE 0.807 *** 1.596 *** -0.789
 (2.93) (3.10) (-1.35)
SHAREI00 -0.103 -0.179 0.076
 (-0.59) (-1.38) (0.35)
FOCUS -0.444 ** -0.075 -0.370
 (-2.02) (-0.63) (-1.49)
SIZE 0.020 0.083 -0.064
 (0.40) (1.56) (-0.88)
BM 0.113 ** 0.023 0.090 *
 (2.61) (0.96) (1.83)
LEV -0.428 0.515 -0.943
 (-0.42) (0.73) (-0.76)
Intercept -0.843 * -1.023 ** 0.180
 (-1.91) (-2.18) (0.28)
RZ 34.63% 16.45%
F-stat. 6.68 *** 3.00 ***
[Prob. >F] [0.000] [0.008]

 # of Types of
 Contracts Used (NTYPE)

 High Info. Low Info. Coeff.
 Asymm. Asymm. Diff.

USER

NTYPE 0.341 *** 0.109 0.232
 (2.72) (1.28) (1.53)
NCONTR

AMT

[AMT.sup.2]

CASH 0.146 0.009 0.136
 (0.62) (0.06) (0.48)
RELSIZE 0.866 *** 1.691 *** -0.825
 (3.32) (3.27) (-1.42)
SHAREI00 -0.164 -0.175 0.011
 (-0.92) (-1.36) (0.05)
FOCUS -0.472 ** -0.069 -0.403 *
 (-2.24) (-0.58) (-1.67)
SIZE -3.98 x 0.069 -0.070
 [10.sup.-4]
 (-0.01) (1.32) (-0.98)
BM 0.100 ** 0.015 0.084 *
 (2.42) (0.57) (1.73)
LEV -0.407 0.400 -0.807
 (-0.42) (0.56) (-0.67)
Intercept -0.681 -0.919 * 0.238
 (-1.56) (-1.93) (0.37)
RZ 38.81% 17.81%
F-stat. 6.85 *** 2.84 **
[Prob. >F] [0.000] [0.011]

 # of Different
 Contracts (NCONTR)

 High Info. Low Info. Coeff.
 Asymm. Asymm. Diff.

USER

NTYPE

NCONTR 0.247 *** -0.006 0.253 ***
 (3.48) (-0.12)
 (2.95)
AMT

[AMT.sup.2]

CASH 0.104 -0.037 0.141
 (0.45) (-0.00) (0.51)
RELSIZE 0.844 *** 1.604 *** -0.761
 (3.29) (3.15) (-1.33)
SHAREI00 -0.157 -0.181 0.024
 (-0.93) (-1.37) (0.11)
FOCUS -0.395 * -0.088 -0.306
 (-1.98) (-0.72) (-1.31)
SIZE -0.018 0.092 -0.110
 (-0.38) (1.67) (-1.52)
BM 0.094 *** 0.026 0.068
 (2.33) (1.14) (1.48)
LEV -0.330 0.529 -0.859
 (-0.34) (0.75) (-0.72)
Intercept -0.545 -0.991 ** 0.446
 (-1.29) (-2.06) (0.70)
RZ 42.28% 15.80%
F-stat. 8.09 *** 2.28 **
[Prob. >F] [0.000] [0.036]

 Total Dollar Amount of
 Contracts (AMT)

 High Info. Low Info. Coeff.
 Asymm. Asymm. Diff.

USER

NTYPE

NCONTR

AMT 0.002 * -7.25 x 0.002 *
 [10.sup.-6]
 (1.91) (-0.12) (1.89)
[AMT.sup.2] -9.02 x 5.96 x -9.08 x
 [10.sup.-7] * [10.sup.-9] [10.sup.-7] *
 (-1.94) (-1.24) (-1.93)
CASH 0.144 0.027 0.116
 (0.55) (0.12) -0.34
RELSIZE 0.836 *** 1.515 *** -0.679
 (3.19) (3.71) (-1.41)
SHAREI00 -0.074 -0.129 0.055
 (-0.37) (-0.83) (0.22)
FOCUS -0.394 * -0.149 -0.245
 (-1.87) (-1.25) (-1.01)
SIZE -0.044 0.0526 -0.096
 (-0.60) (0.66) (-0.89)
BM 0.015 0.020 -0.005
 (0.28) (0.80) (-0.08)
LEV -0.720 1.200 -1.920
 (-0.73) (1.37) (-1.46)
Intercept -0.032 -0.792 0.760
 (-0.06) (-1.19) -0.90
RZ 30.99% 26.75%
F-stat. 7.59 *** 11.36 ***
[Prob. >F] [0.000] [0.000]

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.


References

Agrawal, A., J.F. Jaffe, and G.N. Mandelker, 1992, "The Post-Merger Performance of Acquiring Firms: A Re-examination of an Anomaly," Journal of Finance 67, 1605-1621.

Ang, J.S. and Y. Cheng, 2006, "Direct Evidence on the Market-Driven Acquisition Theory," Journal of Financial Research 29, 199-216.

Asquith, P., R.E Bruner, and D.W. Mullins, 1983, "The Gains to Bidding Firms from Merger," Journal of Financial Economics 11,121-139.

Atiase, R.K. and L.S. Bamber, 1994, "Trading Volume Reactions to Annual Accounting Earnings Announcements," Journal of Accounting and Economics 17,309-329.

Barron, O., O. Kim, S. Lim, and D. Stevens, 1998, "Using Analysts' Forecasts to Measure Properties of Analyst' Information Environment," The Accounting Review 73, 421-433.

Benou, G., K.C. Gleason, and J. Madura, 2007, "Impact of Visibility and Investment Advisor Creditability on the Valuation Effects of High-Tech Cross-Border Acquisitions," Financial Management 36, 69-89.

Bhagat, S., M.W. Marr, and R. Thompson, 1985, "The Rule 415 Experiment: Equity Markets," Journal of Finance 40, 1385-1401.

Blackwell, D., M.W. Marr, and M. Spivey, 1990, "Self Registration and the Reduced Due Diligence Argument:

Implications of the Underwriter Certification and the Implicit Insurance Hypotheses," Journal of Financial and Quantitative Analysis 25,245-259.

Breeden, D. and S. Viswanathan, 1996, "Why Do Firms Hedge? An Asymmetric Information Model," Duke University Working paper.

Brickley, J.A., R.C. Lease, and C.W. Smith, 1988, "Ownership Structure and Voting on Antitakeover Amendments," Journal of Financial Economics 20, 267-291.

Bruner, R. E, 1988, "The Use of Excess Cash and Debt Capacity as a Motive for Merger," Journal of Financial and Quantitative Analysis 23, 199-217.

Butler, A.W., G. Grullon, and J.P. Weston, 2005, "Stock Market Liquidity and the Cost of Issuing Equity," Journal of Financial and Quantitative Analysis 40, 331-348.

Christie, A., 1987, "On Cross-Sectional Analysis in Accounting Research," Journal of Accounting and Economics 9, 231-258.

Comment, R. and G.A. Jarrell, 1995, "Corporate Focus and Stock Returns," Journal of Financial Eeonomics 37, 67-87.

DaDalt, P., G.D. Gay, and .1. Nam, 2002, "Asymmetric Information and Corporate Derivatives Use," Journal of Futures Markets 22, 241-267.

Daniel, K., M. Grinblatt, S. Titman, and R. Wermers, 1997, "Measuring Mutual Fund Performance with Characteristic-Based Benchmarks," Journal of Finance 52, 1035-1058.

DeMarzo, P. and D. Duffie, 1995, "Corporate Incentives for Hedging and Hedge Accounting," Review of Financial Studies 8, 743-771.

Denis, D.J., D.K. Denis, and K. Yost, 2002, "Global Diversification, Industrial Diversification, and Firm Value," Journal of Finance 57, 1951 - 1979.

D'Mello, R. and S.P. Ferris, 2000, "Information Effects of Analyst Activity at the Announcement of New Equity Issues," Financial Management 29, 78-95.

Dolde, W. and D.R. Mishra, 2007, "Firm Complexity and FX Derivatives Use," Quarterly Journal of Business and Economics 46, 3-22.

Dong, M., D. Hirshleifer, S. Richardson, and S.H. Teoh, 2006, "Does Investor Misvaluation Drive the Takeover Market?" Journal of Finance 61,725-762.

Doukas, J. and C. Pantzalis, 2003, "Geographic Diversification and Agency Costs of Debt of Multinational Firms," Journal of Corporate Finance 9, 59-92.

Doukas, J. and N.G. Travlos, 1988, "The Effect of Corporate Multinationalism on Shareholders' Wealth: Evidence from International Acquisitions," Journal of Finance 43, 1161-1175.

Eun, C.S., R. Kolodny, and C. Scheraga, 1996, "Cross-Border Acquisitions and Shareholder Wealth: Tests of the Synergy and Internalization Hypotheses," Journal of Banking and Finance 20, 1559-1582.

Fama, E.E and K.R. French, 1992, "The Cross-Section of Expected Stock Returns," Journal of Finance 47, 283-465.

Fama, E.E and K.R. French, 1993, "Common Risk Factors in the Returns on Stocks and Bonds," Journal of Financial Economics 33, 3-56.

Franks, J. and R. Harris, 1989, "Shareholder Wealth Effects of Corporate Takeovers: The U.K. Experience 1955-1985," Journal of Financial Economics 23, 225-249.

Froot, K., D. Scharfstein, and J. Stein, 1993, "Risk Management: Coordinating Corporate Investment and Financing Policies," Journal of Finance 48, 1624-1658.

Geczy, C., B.A. Minton, and C. Schrand, 1997, "Why Firms Use Currency Derivatives," Journal of Finance 52, 1323-1354.

Guay, W., D. Haushalter, and B. Minton, 2003, "The Influence of Corporate Risk Exposures on the Accuracy of Earnings Forecasts," University of Pennsylvania Working Paper.

Hong, H., T. Lira, and J.C. Stein, 2000, "Bad News Travels Slowly: Size, Analyst Coverage, and the Profitability of Momentum Strategies," Journal of Finance 55, 265-296.

Jensen, M.C., 1986, "Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers," American Economic Review 76, 323-329.

Kim, C.E and C. Pantzalis, 2003, "Global/Industrial Diversification and Analyst Herding," Financial Analysts Journal 59, 69-79.

Kothari, S.R and J.B. Warner, 1997, "Measuring Long-Horizon Security Price Performance," Journal of Financial Economics 43, 301-339.

Krishnaswami, S. and V Subramaniam, 1999, "Information Asymmetry, Valuation, and Corporate Spin-Off Decision," Journal of Financial Economics 53, 73-112.

Lang, L.H.P., R.M. Stulz, and R.A. Walkling, 1991, "A Test of the Free Cash Flow Hypothesis: The Case of Bidder Returns," Journal of Financial Economics 29, 315-336.

Leung, S. and S. Yu, 1996, "On the Choice Between Sample Selection and Two-Part Models," Journal of Econometrics 72, 197-229.

Lin, J.B., C. Pantzalis, and J.C. Park, 2007a, "Corporate Hedging Policy and Equity Mispricing," University of South Florida, Working paper.

Lin, J.B., C. Pantzalis, and J.C. Park, 2007b, "Corporate Use of Derivatives and Excess Value of Diversification," Journal of Banking and Finance 31, 889-913.

Loughran, T. and J.R. Ritter, 2000, "'Uniformly Least Powerful Tests of Market Efficiency," Journal of Financial Economics 55, 361-389.

Markides, C.C. and C.D. Ittner, 1994, "Shareholder Benefits from Corporate International Diversification: Evidence from US International Acquisitions," Journal of International Business Studies 25, 343-366.

Masulis, R.W., C. Wang, and F. Xie, 2007, "Corporate Governance and Acquirer Returns," Journal of Finance 62, 1851-1889.

Mayers, D. and C.W. Smith, 1987, "Corporate Insurance and the Underinvestment Problem," Journal of Risk and Insurance 54, 45-54.

Mroz, T., 1987, "The Sensitivity of an Empirical Model of Married Women's Hours of Work to Economic and Statistical Assumptions," Econometrica 55, 765-799.

Polk, C. and P. Sapienza, 2009, "The Stock Market and Corporate Investment: A Test of Catering Theory," Review of Financial Studies 22, 187-217.

Rau, P.R. and T. Vermaelen, 1998, "Glamour, Value and the Post-Acquisition Performance of Acquiring Firms," Journal of Financial Economics 49, 223-253.

Renders, A. and A. Gaeremynck, 2006, "Corporate Governance and Performance: Controlling for Sample Selection Bias and Endogeneity," Katholieke Universiteit Leuven, Working paper.

Rhodes-Kropf, M., D.T. Robinson, and S. Viswanathan, 2005, "Valuation Waves and Merger Activity: The Empirical Evidence," Journal of Financial Economics 77, 561-603.

Roll, R., 1986, "The Hubris Hypothesis of Corporate Takeovers," Journal of Business 59, 197-216.

Shleifer, A. and R. Vishny, 2003, "Stock Market Driven Acquisitions," Journal of Financial Economics 70, 295-311.

Smith, C. and R.M. Stulz, 1985, "The Determinants of Firms' Hedging Policies," Journal of Financial and Quantitative Analysis 20, 391-405.

Song, M.H. and R.A. Walkling, 1993, "The Impact of Managerial Ownership on Acquisition Attempts and Target Shareholder Wealth," Journal of Financial and Quantitative Analysis 28, 439-457.

Travlos, N.G., 1987, "Corporate Takeover Bids, Methods of Payment, and Bidding Firm's Stock Returns," Journal of Finance 42, 64-71.

Wooldridge, J., 2002, Econometric Analysis of Cross Section and Panel Data, Cambridge, MA, MIT Press.

(1) Overinvestment by acquiring firms can be partially explained by the free cash flow hypothesis (Jensen, 1986; Lang, Stulz, and Walkling, 1991) and the hubris hypothesis (Roll, 1986). Recently, Masulis, Wang, and Xie (2007) demonstrate that managers protected by more antitakeover provisions (i.e., subject to weak market-imposed discipline) tend to indulge in empire building acquisitions that cause negative shareholder returns.

(2) Benou, Gleason, and Madura (2007) consider and test the importance of target firms' asymmetry of information. They find that visibility (proxied by a high level of media attention) and credibility (proxied by the endorsement of the acquisition by a top investment bank) enhance the perceived benefits of acquiring foreign targets that have substantial intangible assets and a high level of informational asymmetry. Our focus is on acquirer firms' information asymmetry as it has a more direct effect on the acquirer's abnormal returns around the takeover announcement date. We would like to control for the effects of target information asymmetry but lack the data to do so in a rigorous fashion. Instead, we attempt to partially account for target information asymmetry effects by using the RELSIZE (relative size) variable in our multivariate analysis.

(3) In addition, several theories have been proposed to support the view that firm value is improved by the use of derivatives: 1) Mayers and Smith (1987) argue that hedging decreases agency costs emanating from the underinvestment problem; 2) Breeden and Viswanathan (1996) and DeMarzo and Duffie (1995) suggest that hedging allows managers to reduce the noise associated with signaling firm quality; 3) Smith and Stulz (1985) suggest that hedging reduces expected tax liability and, hence, increases firm value because the tax function is convex due to the nature of tax rules and regulations; and 4) Smith and Stulz (1985) maintain that hedging reduces costs of financial distress.

(4) Similar evidence can also be found in other studies. For example, Rhodes-Kropf, Robinson, and Viswanathan (2005) and Dong, Hirshleifer, Richardson, and Teoh (2006) provide evidence that irrational misvaluation affects firms' takeover behavior, supporting the model proposed by Shleifer and Vishny (2003). Ang and Cheng (2006) find that share overvaluation is an important motive for firms to make stock acquisitions.

(5) Swaps Monitor ceased compiling this database in the third quarter of 1997. Thus, our sample is restricted to the five-year period where complete annual derivatives usage data are available. The Database of Users of Derivatives was compiled from annual reports and filings with regulatory agencies. It, therefore, does not contain information on firms that used derivatives, but made no disclosure of that fact.

J. Barry Lin, Christos Pantzalis, and Jung Chul Park *

We appreciate helpful comments and suggestions from Bill Christie (the editor), an anonymous referee, and participants at the 2006 Financial Management Association Annual Meeting and 2006 Multinational Finance Society Annual Meeting.

* J. Barry Lin is an Associate Professor of Finance at Simmons College in Boston, MA. Christos Pantzalis is an Associate Professor of Finance at University of South Florida in Tampa, FL. Jung Chul Park is an Assistant Professor of Finance at Louisiana Tech University in Ruston, LA.

(6) As expected, most component measures of 1NFO_ASYMM are highly correlated with expected signs. Also, by construction, all individual information asymmetry measures are significantly correlated with the information asymmetry index, INFO_ASYMM. This suggests that INFO_ASYMM balances out the effects and shortcomings of the individual information asymmetry measures, while aggregating their informativeness.

(7) In constructing INFO_ASYMM, we employ the methodology outlined in Butler, Grullon, and Weston (2005). In their paper, they create a liquidity index that comprises the effects of ranking on six different liquidity measures.

(8) In addition to the two models presented in this paper, we have used a four-factor model including a momentum factor and also used different estimation periods without obtaining significantly different results from the ones reported here. Also, in the Appendix, we present results based on long-term buy-and-hold abnormal returns (BHRs). These are qualitatively similar to the ones presented here.

(9) In this study, a user is defined as the firm that uses derivatives in the same year it makes an M&A announcement. Out argument that derivatives use bas a positive impact on long-run performance is based on the assumption that firms using derivatives do not change their derivative policies frequently. This assumption is fairly reasonable, based on the evidence of past studies. For example, Lin, Pantzalis, and Park (2007b) used the same database as this paper (Swaps Monitor Publications) and demonstrated that derivative policies remained "sticky" over the 1992-1996 period & which is the same period as the one examined in our study. We also find that 76% of sample firms do not change their derivatives use policy over the five-year period and 90% of sample firms keep the same policy at least two years. When we retest all models with those sample firms that exhibit a consistent derivatives use policy, we find that most results are qualitatively similar to ones we report in the paper.

(10) In a recent paper, Renders and Gaeremynck (2006) use the three-stage model to test the relation between corporate governance and performance by controlling for sample selection bias and endogeneity.

(11) However, these variables do not seem to be seriously related to either derivatives use variables or Iong-run performance, so they can be good instruments in estimating the probability of being in the sample.

(12) We also tested the model in Equation (5) using the SDC database sample and found results similar to the ones in Table VI.

(13) Following Geczy, Minton, and Schrand (1997), we substitute R&D expenditure (scaled by sales) for book-to-market and find similar results.

(14) As we illustrate in the Appendix, the effect of derivatives use on long-run performance is much stronger for firms with high levels of asymmetric information than for firms with low levels of asymmetric information.
Table I. Annual and Geographical Numbers of US Acquirers in the
Sample, 1992-1996

Reported are numbers and corresponding percentages of US acquirers
during the period 1992-1996. Panel A and Panel B show the numbers
based on years of M&A announcements and areas where target firms
locate, respectively. In both panels, we report numbers of all US
acquirers and subsamples classified on the usage of derivatives. If
a firm used any type of derivatives, then the firm is classified
into user. The first row in Panel B demonstrates, for example, that
271 US firms bought European firms from 1992 to 1996. Among them,
182 acquirers used derivatives, while 89 acquirers did not.

Panel A. Numbers and Percentages of US Acquirers Based on Year of
M&A Announcements

Year of M&A All US Users of Nonusers of
Announcements Acquirers Derivatives Derivatives

1992 64 51 13
 (14.22%) (11.33) (2.89)
1993 90 69 21
 (20.00) (15.33) (4.67)
1994 75 53 22
 (16.67) (11.78) (4.89)
1995 122 74 48
 (27.11) (16.44) (10.67)
1996 99 58 41
 (22.00) (12.89) (9.11)

Panel B. Numbers and Percentages of US Acquirers Based on
Areas Where Target Firms Locate

Area Where All US Users of Nonusers of
Targets Locate Acquirers Derivatives Derivatives

European 271 182 89
 (0.60) (35.78) (19.78)
South America 74 52 22
 (16.44) (10.44) (4.89)
North America 51 31 20
 (11.33) (5.33) (4.44)
Asian and Pacific 52 38 14
 (11.56) (6.67) (3.11)
Other areas 2 2 0
 (0.44) (0.44) (0.00)
Total 450 305 145
 (100.00) (67.78) (32.22)

Table II. Acquirer Firms' Policy of Derivatives Use

Reported are descriptive statistics of US acquirers' policy of
derivatives use. The sample contains 450 firms covered in the Swaps
Monitor Publications' Database of Users of Derivatives over the
period 1992-1996. Swaps Monitor indicates the usage of derivatives
as well as the notional amounts for seven different contracts
spanning two general types of derivatives: 1) interest rate (IR)
derivatives and 2) foreign exchange (FX) derivatives. The seven
different contracts are: 1) IR-options, 2) IR-swaps, 3)
IR-forwards/futures, 4) FX-options, 5) FX-swaps, 6) FX-futures,
and 7) FX-forwards. If a firm used any type of derivatives, then
the firm is classified as a user. Variables in Panel A are dummy
variables indicating the use of certain type of derivatives.
Variables in Panel B represent the notional dollar amounts of
corresponding derivatives.

 Mean Median Std. Dev.

Panel A. Use of Derivatives

USER 0.678 1.000 0.468
FX 0.587 1.000 0.493
FX-swaps 0.244 0.000 0.430
FX-options 0.204 0.000 0.404
FX forwards 0.564 1.000 0.496
FX futures 0.000 0.000 n/a
IR 0.351 0.000 0.478
IR-swaps 0.324 0.000 0.469
JR-options 0.053 0.000 0.225
IR forwards/
 futures 0.007 0.000 0.081

Panel B. Notional Dollar Amount of Derivative
Contracts (Million Dollars)

AMT_H 580.554 24.600 2,629.9
AMT_FX 415.522 0.000 1,518.1
AMT_FX-swaps 88.067 0.000 293.416
AMT_FX-options 187.979 0.000 1,045.0
AMT_FX-forwards 292.163 0.000 1,133.7
AMT_FX-futures 0.000 0.000 n/a
AMT_IR 221.364 0.000 1,372.4
AMT_IR-swaps 152.483 0.000 699.467
AMT_IR-forwards
 /futures 19.595 0.000 293.265
AMT_IR-options 46.490 0.000 459.443

 Minimum Maximum

Panel A. Use of Derivatives

USER 0.000 1.000
FX 0.000 1.000
FX-swaps 0.000 1.000
FX-options 0.000 1.000
FX forwards 0.000 1.000
FX futures 0.000 0.000
IR 0.000 1.000
IR-swaps 0.000 1.000
JR-options 0.000 1.000
IR forwards/
 futures 0.000 1.000

Panel B. Notional Dollar Amount of Derivative
Contracts (Million Dollars)

AMT_H 0.000 31,031.5
AMT_FX 0.000 14,921
AMT_FX-swaps 0.000 2,600
AMT_FX-options 0.000 10,100
AMT_FX-forwards 0.000 14,818
AMT_FX-futures 0.000 0.000
AMT_IR 0.000 19,364.5
AMT_IR-swaps 0.000 8,310
AMT_IR-forwards
 /futures 0.000 4,399
AMT_IR-options 0.000 6,655.5

Table III. Mean Differences of Information Variables, Transactions'
Characteristics, and US Acquirer Firms' Characteristics for
Subgroups Classified on Derivatives Use

This table reports mean differences of various variables between
acquirers that use derivatives and acquirers that do not use
derivatives. If a firm used any type of derivatives, then the firm
is classified into user. The variables examined are defined as
follows. INFO_ASYMM is computed as averaging the ranks in five
measures of informational asymmetry: the inverse value of
institutional ownership (INSTP), the inverse value of number of
analysts' forecasts (NAF), forecast error (FE), dispersion of
forecasts (DISP), and the standard deviation of residual returns
(SDRES). INSTP is the percentage of shares owned by institutional
investors. NAF is the number of forecasts issued in June of MeYR
(I/B/E/S Summary Data). FE is the absolute value of the median
forecast error computed as the difference between the median
one-year-ahead EPS forecast and the actual EPS. DISP, the
dispersion of forecasts, is computed as the standard deviation of
the one-year-ahead forecasts divided by the absolute value of the
median forecast. FE and DISP are measured in June of each year.
SDRES is the standard deviation of residual returns of the market
model. CASH is a dummy variable that takes the value of one if the
acquisition is financed entirely with cash, and zero otherwise.
RELSIZE is equal to the ratio of transaction value to acquirer's
total assets. SHARE100 is a dummy variable that takes one when 100%
of the shares are acquired through M&As. FOCUS is a dummy variable
that indicates whether the US acquirer and non-US target firms are
in the same four-digit SIC code industry. SIZE (BM) is the decile
ranking of acquirer's total market value of equity (book-to-market
ratio), following Fama and French (1992). LEV is the ratio of
long-term debt to total assets.

 Users Nonusers
 (N = 305) (N = 145)

Panel A. Information Variables

Information asymmetry 0.476 0.530
Institutional ownership (INSTP) 58.426 55.028
 Low INSTP = high info. asymmetry
Number of analysts' forecasts (NAF) 13.089 9.791
 Low NAF = high info. asymmetry
Forecast error (FE) 0.098 0.099
 High FE = high info. asymmetry
Dispersion of analysts' forecasts
(DISP) 0.002 0.002
 High DISP = high info. asymmetry
Std. Dev. of Market Model Residuals
(SDRES) 0.035 0.042
 High SDRES = high info. asymmetry

Panel B. Transaction Characteristics

Cash finance (CASH) 0.819 0.822
Relative target size (RELSIZE) 0.126 0.141
Acquire 100% of target shares
 (SHARE100) 0.649 0.715
Focus-increasing M&As (FOCUS) 0.218 0.303

Panel C. Acquirer Firms' Characteristics

Rank of size (SIZE) 8.157 6.634
Rank of book-to-market ratio (BM) 3.639 3.434
Leverage (LEV) 0.166 0.147

 Mean Difference Tests

 t-stat:
 Users- Mean
 Nonusers Diff. = 0

Panel A. Information Variables

Information asymmetry -0.054 *** -3.03
Institutional ownership (INSTP) 3.398 * 1.86
 Low INSTP = high info. asymmetry
Number of analysts' forecasts (NAF) 3.298 *** 3.77
 Low NAF = high info. asymmetry
Forecast error (FE) -0.001 -0.05
 High FE = high info. asymmetry
Dispersion of analysts' forecasts
(DISP) 7.08 x [10.sup.-6] 0.01
 High DISP = high info. asymmetry
Std. Dev. of Market Model Residuals
(SDRES) -0.006 *** -4.01
 High SDRES = high info. asymmetry

Panel B. Transaction Characteristics

Cash finance (CASH) -0.003 -0.05
Relative target size (RELSIZE) -0.015 -0.20
Acquire 100% of target shares
 (SHARE100) -0.066 -1.35
Focus-increasing M&As (FOCUS) -0.086 ** -1.97

Panel C. Acquirer Firms' Characteristics

Rank of size (SIZE) 1.523 *** 6.71
Rank of book-to-market ratio (BM) 0.205 0.97
Leverage (LEV) 0.019 1.25

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table IV. Long-Term Stock Performances of Subsamples Based on Acquirer
Firms' Derivative Policies

This table reports the cumulative average abnormal returns (CAARs) of
subsamples after M&A announcements. Sample firms are classified by the
policy of derivatives use: the use of derivatives (USER), the number
of types of contracts used (NTYPE), the number of different contracts
(NCONTR), and the total dollar amount of derivatives (AMT). In Panel
A, Group 1 = all firms, Group 2 = users, and Group 3 = nonusers. In
Panel B, Group 1 = firms using both types (FX and IR), Group 2 = firms
using one of the two types, and Group 3 = nousers. In Panel C, Group 1
= firms using four to six contracts, Group 2 = firms using one to
three contract(s), and Group 3 = nonusers. In Panel D, Group 1 = firms
using a higher dollar amount that is greater than the median value,
Group 2 = firms using a lower dollar amount, and Group 3 = nonusers.
Sixty month estimation interval (-60 to -1M) and various windows up to
60 months are used. Two models are used to estimate abnormal returns.
The abnormal returns are based on either the one-factor market model
or Fama and French's (1993) three-factors model.

One-factor (market) model: [R.sub.i,m] - [RF.sub.m] = [[gamma].sub.0]
+ [[gamma].sub.1] ([RM.sub.m] - [RF.sub.m]) + [e.sub.i,m]

Fama and French's three-factor model: [R.sub.i,m] - [RF.sub.m] =
[[psi].sub.0] + [[psi].sub.1] ([RM.sub.m] - [RF.sub.m]) +
[[psi].sub.2][SMB.sub.m] + [[psi].sub.3] [HML.sub.m] + [e.sub.i,m],

where i indexes firms, and m is a monthly time index. [R.sub.i,m] is
the rate of return on firm i in month m, and [RM.sub.m] is the rate of
return on the market portfolio. We use the one-month Treasury bill
rate for the risk-free rate ([RF.sub.m]). [gamma]'s are the estimated
intercept and coefficient, and [e.sub.i,m] is a random error term.
[SMB.sub.m] captures the performance of small stocks relative to big
stocks by computing [SMB.sub.m] = 1-3 (small value + small neutral +
small growth) -1-3 (big value + big neutral + big growth), while
[HML.sub.m] captures the performance of value stocks relative to
growth stocks by computing [HML.sub.m], = 1-2
(small value + big value)-1-2
(small growth + big growth), where the median NYSE market equity is
used as the size breakpoint, and the 30th and 70th NYSE percentiles of
book-to-market equity are used as the book-to-market breakpoints. We
apply conventional event-study methodology in computing CAARs. The
cumulative abnormal return of firm i from the announcement to the
[omega]th month, C A [R.sub.i,y] (0, [omega]))
[[summation].sup.[omega].sub.m=0] A [R.sub.i,m], where [AR.sub.i,m] is
the abnormal return on the firm i in the month m, and is computed as
the difference between actual monthly return and estimated monthly
return. In order to see the statistical significance of the CAAR, we
compute Z-score, Z(0, [omega]). First, the average standardized
abnormal return of event firms of the subsample in the mth month,
[ASAR.sub.m] is computed as [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE
IN ASCII.], where [N.sub.s] is the number of firms in the subsample,
and standardized abnormal return, S A [R.sub.i,m] is the firm is
abnormal return in the mth month divided by the standard deviation of
firm i's residuals from the model within the estimation period. Then,
cumulative average standardized abnormal return over a given period,
CASAR(0, [omega]), is measured as [[summation].sup.[omega].sub.m=0] A
S A [R.sub.m]-[square root of [omega] + 1]. Finally, Z-score for the
CAAR is obtained by the following equation: Z(0, [omega])) = CSAAR(0,
[omega]) [square root of [N.sub.s].

Panel A. Subsamples Classified by the Use of Derivatives

Windows [1] [2] [3] Mean
 All Firms Users Non-users Difference
 (N = 450) (N = 305) (N = 145) Tests
 [2]-[3]

 CAARs Based on Market Model

(0,+12M) -4.06% ** -3.103 * -6.327 3.224
 (-2.35) (-1.83) (-1.03) (1.11)
(0,+24M) -7.28 ** -6.341 * -9.525 3.184
 (-2.51) (-1.94) (-1.27) (0.82)
(0,+36M) -10.12 *** -5.750 -20.516 *** 14.767 ***
 (-3.18) (-1.22) (-3.61) (3.02)
(0,+48M) -9.99 *** -4.033 -23.409 *** 19.376 ***
 (-2.84) (-0.22) (-4.56) (3.44)
(0,+60M) -1.52 5.648 * -17.676 *** 23.324 ***
 (-0.11) (1.84) (-2.75) (3.76)

 CAARs Based on Fama and French Three-Factor Model

(0,+12M) -3.06% * -2.364 -4.718 2.354
 (-1.80) (-1.36) (-0.55) (0.83)
(0,+24M) -4.93 * -4.826 -5.173 0.347
 (-1.72) (-1.57) (0.27) (0.09)
(0,+36M) -6.70 ** -2.709 -16.205 *** 13.496 ***
 (-2.54) (-0.73) (-3.05) (2.95)
(0,+48M) -4.30 0.990 -16.237 *** 17.226 ***
 (-1.21) (1.09) (-3.47) -3.24
(0,+60M) 3.07 9.833 *** -12.179 * 22.013 ***
 (1.61) (3.37) (-1.82) (3.68)

Panel B. Subsamples Classified by NTYPE

Windows [1] [2] [3]
 NTYPE = 2, NTYPE = 1, NTYPE = 0,
 Firms Use Firms Use Firms
 Both FX Either FX Use No
 and IR or IR Derivatives
 (N = 117) (N = 188) (N = 145)

 CAARs Based on Market Model

(0,+12M) -5.580% ** -1.547 -6.327
 (-2.57) (-0.29) (-1.03)
(0,+24M) -6.099 ** -6.493 -9.525
 (-2.17) (-1.27) (-1.27)
(0,+36M) -0.585 -8.992 -20.516 ***
 (-0.42) (-1.36) (-3.61)
(0,+48M) 1.574 -7.659 -23.409 ***
 (0.84) (-0.61) (-4.56)
(0,+60M) 18.879 *** -3.255 -17.676 ***
 (6.30) (0.12) (-2.75)

 CAARs Based on Fama and French s Three-Factor Model

(0,+12M) -4.526% * -1.006 -4.718
 (-1.94) (-0.19) (-0.55)
(0,+24M) -4.930 ** -4.761 -5.173
 (-2.04) (-0.85) (-0.27)
(0,+36M) -0.104 -4.344 -16.205 ***
 (-0.87) (-0.54) (-3.05)
(0,+48M) 4.645 ** -1.374 -16.237 ***
 (2.22) (0.52) (-3.47)
(0,+60M) 20.360 *** 2.749 -12.179 *
 (8.12) (1.42) (-1.82)

Windows Mean Mean
 Difference Difference
 Tests Tests
 [1]-[2] [1]-[3]

 CAARs Based on Market Model

(0,+12M) -4.033 0.747
 (-1.28) (0.21)
(0,+24M) 0.394 3.426
 (0.09) -0.73
(0,+36M) 8.407 19.931 ***
 (1.54) -3.63
(0,+48M) 9.233 24.983 ***
 (1.50) -3.64
(0,+60M) 22.134 *** 36.555 ***
 (3.32) (4.83)

 CAARs Based on Fama and French's
 Three-Factor Model

(0,+12M) -3.520 0.192
 (-1.16) (0.06)
(0,+24M) -0.169 0.243
 (-0.04) (0.05)
(0,+36M) 4.240 16.101 ***
 (0.83) (3.07)
(0,+48M) 6.019 20.882 ***
 (1.03) (3.20)
(0,+60M) 17.611 *** 32.539 ***
 (2.76) (4.41)

Panel C. Subsamples Classified by NCONTR

Windows [1] [2] [3]
 NCONTR = 4-6, NCONTR = 1-3, NCONTR = 0,
 Firms Have Firms Have Firms Have
 4-6 Different 1-3 Different No Contracts
 Contracts Contract(s) (N = 145)
 (N = 33) (N = 272)

 CAARs Based on Market Model

(0,+12M) 2.032% -3.754 ** -6.327
 (0.65) (-2.17) (-1.03)
(0,+24M) 8.985 *** -8.286 *** -9.525
 (7.21) (2.62) (-1.27)
(0,+36M) 20.308 *** -9.057 ** -20.516 ***
 (4.69) (-2.26) (-3.61)
(0,+48M) 28.465 *** -8.012 -23.409 ***
 (6.10) (-1.34) (-4.56)
(0,+60M) 37.229 *** 1.784 -17.676 ***
 (7.58) (0.74) (-2.75)

 CAARs Based on Fama and French s Three-Factor Model

(0,+12M) 3.064% -3.053 * -4.718
 (0.97) (-1.79) (0.55)
(0,+24M) 9.340 ** -6.624 ** -5.173
 (2.13) (-2.20) (-0.27)
(0,+36M) 15.520 *** -5.023 -16.205 ***
 (3.29) (-1.46) (-3.05)
(0,+48M) 24.327 *** -1.868 -16.237 ***
 (5.66) (0.13) (-3.47)
(0,+60M) 27.197 *** 7.709 *** -12.179 *
 (5.71) (2.65) (-1.82)

Windows Mean Mean
 Difference Difference
 Tests Tests
 [1]-[2] [1]-[3]

 CAARs Based on Market Model

(0,+12M) 5.786 8.358
 (1.19) (1.52)
(0,+24M) 17.271 *** 18.510 ***
 (2.68) (2.63)
(0,+36M) 29.365 *** 40.825 ***
 (3.56) (4.72)
(0,+48M) 36.477 *** 51.874 ***
 (3.86) (4.72)
(0,+60M) 35.445 *** 54.904 ***
 (3.38) (4.47)

 CAARs Based on Fama and French's
 Three-Factor Model

(0,+12M) 6.117 7.782
 (1.31) (1.47)
(0,+24M) 15.964 ** 14.513 **
 (2.59) (2.14)
(0,+36M) 20.542 *** 31.725 ***
 (2.64) (3.89)
(0,+48M) 26.195 *** 40.564 ***
 (2.91) (3.83)
(0,+60M) 19.489 * 39.377 ***
 (1.92) (3.24)

Panel D. Subsamples Classifed by AMT

Windows [1] [2] [3]
 Firms Using a Firms Using a Firms Using
 Higher Dollar Lower Dollar No Dollar
 Amount of Amount of Amount of
 Derivatives Derivatives Derivatives

 CAARs Based on Market Model

(0,+12M) -5.506% * -2.841 -6.327
 (-1.65) (-0.26) (-1.03)
(0,+24M) -8.798 *** -10.688 -9.525
 (-2.63) (-1.40) (-1.27)
(0,+36M) -6.133 ** -16.414 ** -20.516 ***
 (-2.19) (-2.10) (-3.61)
(0,+48M) -5.982 -16.658 ** -23.409 ***
 (-1.19) (-1.96) (-4.56)
(0,+60M) 1.536 -4.492 -17.676 ***
 (1.10) (-0.47) (2.75)

 CAARs Based on Fama and French's Three-Factor Model

(0,+12M) -4.417% -3.694 -4.718
 (-1.20) (-0.78) (-0.55)
(0,+24M) 9.572 *** -8.106 -5.173
 (-3.13) (-1.14) (-0.27)
(0,+36M) -7.523 *** -8.002 -16.205 ***
 (-3.24) (-0.99) (-3.05)
(0,+48M) -3.549 -5.515 -16.237 ***
 (-0.42) (-0.70) (-3.47)
(0,+60M) 7.077 *** 3.940 -12.179 *
 (4.38) (0.53) (1.82)

Windows Mean Mean
 Difference Difference
 [1]-[2] [1]-[3]

 CAARs Based on Market Model

(0,+12M) -2.664 0.821
 (-0.70) (0.22)
(0,+24M) 1.889 0.727
 (0.39) (0.16)
(0,+36M) 10.281 14.384 **
 (1.60) (2.52)
(0,+48M) 10.676 17.427 **
 (1.59) (2.48)
(0,+60M) 6.028 19.212 **
 (0.89) (2.50)

 CAARs Based on Fama and French's
 Three-Factor Model

(0,+12M) -0.723 0.301
 (-0.20) (0.09)
(0,+24M) -1.466 -4.398
 (-0.31) (-0.97)
(0,+36M) 0.478 8.682
 (0.08) (1.60)
(0,+48M) 1.967 12.688 *
 (0.30) (1.89)
(0,+60M) 3.137 19.257 **
 (0.45) (2.54) ***

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

** Significant at the 0.10 level.

Table V. The Effect of Derivatives Use on Long-Term Stock Performance

This table reports estimates of coefficients of OLS regressions

CAR[(0, +60M).sub.i,y] + [[beta].sub.0] + [[beta].sub.1]
[DERIVATIVES.sub.i,y] + [[beta].sub.2][CASH.sub.i,y] +
[[beta].sub.3][RELSIZE.sub.i,y] + [[beta].sub.4][SHARE100.sub.i,y] +
[[beta].sub.5][FOCUS.sub.i,y] + [[beta].sub.6][SIZE.sub.i,y] +
[[beta].sub.7][BM.sub.i,y] + [[beta].sus.8][LEV.sub.i,y] +
[e.sub.i,y].

The cumulative abnormal return of firm i from the announcement to the
60th month, [CAR.sub.i,y] (0, +60M) = [[summation].sup.60.sub.m=0]
[AR.sub.i,m] where [AR.sub.i,m] is the abnormal return on firm i in
the month m, and is computed oas the difference between actual monthly
return and the monthly return that is estimated by Fama and French's
(1993) three-factor model. DERIVATIVES is alternatively one of the
following: USER is a dummy variable that takes the value of one if a
firm uses any kind of derivatives against foreign exchange risk or
interest rate risk, and zero otherwise; NTYPE is the number of types
of derivatives used (i.e., FX and/or IR), or none and takes values
from zero to two; NCONTR is the number of different derivative
contracts used by the firm and based on Swaps Monitor's database, it
can take values from zero to seven; and AMT is the total dollar amount
of derivatives used. CASH is a dummy variable that takes the value of
one if the acquisition is financed entirely with cash, and zero
otherwise. RELSIZE is equal to the ratio of transaction value to
acquirer's total assets. SHARE100 is a dummy variable that takes the
value of one when 100% of shares are acquired through M&As. FOCUS is a
dummy variable that indicates whether the US acquirer and non-US
target firms are in the same four-digit SIC code industry. SIZE (BM)
is the decile ranking of acquirer's total market value of equity
(book-to-market ratio) following Fama and French (1992). LEV is the
ratio of long-term debt to total assets. Estimated coefficients are
divided by 100.

 [1] [2] [3]

USER 0.228 *
 (-1.81)
NTYPE 0.207 ***
 (-2.67)
NCONTR 0.113 **
 (-2.53)
AMT

AMT 2

CASH 0.062 0.082 0.067
 (-0.37) (-0.49) (-0.39)
RELSIZE 0.719 *** 0.762 *** 0.787 ***
 (-3.44) (-3.72) (-3.86)
SHARE100 -0.114 -0.121 -0.113
 (-1.13) (-1.22) (-1.13)
FOCUS -0.192 * -0.194 * -0.183 *
 (-1.76) (-1.82) (-1.71)
SIZE 0.044 0.028 0.028
 (-1.45) (-0.89) (-0.85)
BM 0.085 *** 0.077 *** 0.086 ***
 (-3.58) (-3.24) (-3.83)
LEV 0.131 0.007 0.125
 (-0.24) (-0.01) (-0.22)
Intercept -0.902 *** -0.782 *** -0.792 ***
 (-3.09) (-2.69) (-2.73)
[R.sup.2] 21.40% 23.76% 23.23%
F-stat. 7.53 *** 6.76 *** 6.52 ***
[Prob. >F] [0.000] [0.000] [0.000]

 [4] [5]

USER

NTYPE

NCONTR

AMT 5.63 x [10.sup.-5] 7.11 x [10.sup.-5]
 (-2.68) (-1.01)
AMT 2 -1.52 x [10.sup.-9]
 (-0.28)
CASH 0.110 0.111
 (-0.47) (-0.47)
RELSIZE 0.870 *** 0.869 ***
 (-3.96) (-3.94)
SHARE100 -0.101 0.099
 (-0.91) (-0.89)
FOCUS -0.276 ** -0.277 **
 (-2.42) (-2.44)
SIZE 0.029 0.027
 (-0.80) (-0.68)
BM 0.067 *** 0.067 ***
 (-3.01) (-3.00)
LEV 0.031 0.015
 (-0.05) (-0.02)
Intercept -0.671 ** -0.659 **
 (-2.10) (-2.01)
[R.sup.2] 20.21% 20.25%
F-stat. 7.97 *** 19.55 ***
[Prob. >F] [0.000] [0.000]

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table VI. The Effect of Derivatives Use on Long-Term Stock
Performance: Two-Stage and Three-Stage Models to Control for Selection
Bias and Endogeneity

This table conducts two-stage and a three-stage models to control for
selection bias and endogeneity, and reports estimates of coefficients
of the third-stage model. In the three-stage model, the first equation
is

[SAMPLE.sub.i,y] = [[beta].sub.0] + [[beta].sub.1] [SIZE.sub.i,y] +
[[beta].sub.2] [BM.sub.i,y] + [[beta].sub.3] [LEV.sub.i,y] +
[[beta].sub.4] [SP500.sub.i,y] + [[beta].sub.5] [AMEX.sub.i,y] +
[[beta].sub.6] [NASDAQ.sub.i,y] + [e.sub.i,y]

where SAMPLE is a dummy variable that takes the value of one if data
are included in the final sample, and zero otherwise. SIZE (BM) is the
decile ranking of acquirer's total market value of equity
(book-to-market ratio) following Fama and French (1992). LEV is the
ratio of long-term debt to total assets. SP500 dummy takes the value
of one if a firm is included in the S&P 500 index, and zero otherwise.
AMEX and NASDAQ dummies indicate the firm's primary exchange. In the
second stage, the reduced-form regression estimates the derivative
variable (USER, NTYPE, NCONTR, or AMT):

[DERIVATIVES..sub.i,y] [[beta].sub.0] + [[beta].sub.1] [SIZE.sub.i,y]
+ [[beta].sub.2] [BM.sub.i,y] + [[beta].sub.3] [LEV.sub.i,y] +
[[beta].sub.4] [INTCOV.sub.i,y] + [[beta].sub.5] [NSEG.sub.i,y] +
[[beta].sub.6] [TAX.sub.i,y] + [[beta].sub.7] [INFO_ASYMM.sub.i,y] +
[[beta].sub.8] MILLS + [e.sub.i,y].

DERIVATIVES is alternatively one of the following: USER is a dummy
variable that takes the value of one if a firm uses any kind of
derivatives against foreign exchange risk or interest rate risk, and
zero otherwise; NTYPE is the number of types of derivatives used
(i.e., FX and/or IR), or none and takes values from zero to two;
NCONTR is the number of different derivative contracts used by the
firm and, based on Swaps Monitor's database, it can take values from
zero to seven; and AMT is the total dollar amount of derivatives used.
We consider the firm's interest coverage ratio (INTCOV), computed as
EBIT divided by interest expense. NSEG is the number of reported
segments. R&D is research and development expenditure scaled by sales.
Tax rate (TAX) is calculated as total income taxes divided by pre-tax
income. INFO_ASYMM is the average of ranks in five measures of
informational asymmetry: inverse value of institutional ownership
(INSTP), inverse value of the number of analysts' forecasts (NAF),
forecast error (FE), dispersion of forecasts (DISP), and standard
deviation of residual returns (SDRES). INSTP is percentage of shares
owned by institutional investors. NAF is number of forecasts issued in
June of MeYR (I/B/E/S Summary Data). FE is the absolute value of the
median forecast error, computed as the difference between the median
one-year-ahead EPS forecast and the actual EPS. DISP, the dispersion
of forecasts, computed as the standard deviation of the one-year-ahead
forecasts divided by the absolute value of the median forecast. FE and
DISP are measured in June of each year. SDRES is the standard
deviation of residual returns of the market model. MILLS is the
inverse Mills ratio that is estimated in the first equation. On the
third stage, the structural equation model's independent variables
consist of the estimated derivative variable from the second stage,
the inverse Mills ratio from the first stage, and several control
variables.

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where the cumulative abnormal return of firm i from the announcement
to the 60th month, [CAR.sub.i,j](0, + 60M) =
[[summation].sup.60.sub.m=0] [AR.sub.i,m] where [AR.sub.i,m] is the
abnormal return on the firm i in month m and is computed as the
difference between actual monthly return and the monthly return that
is estimated by Fama and French's (1993) three-factor model. CASH is a
dummy variable that takes the value of one if the acquisition is
financed entirely with cash, and zero otherwise. RELSIZE is equal to
the ratio of transaction value to acquirer's total assets. SHARE100 is
a dummy variable that takes the value of one when 100% of shares are
acquired through M&As. FOCUS is a dummy variable that indicates
whether the US acquirer and non-US target firms are in the same
four-digit SIC code industry. In the two-stage model, we conduct the
second and the third equations to control for endogeneity only.
Estimated coefficients are divided by 100.

 [1] [2] [3]
 2-Stage 3-Stage 2-Stage
 Model Model Model

USER 1.330 *** 1.206
 (3.19) (2.84)
NTYPE 0.770 ***
 (3.00)
NCONTR

AMT

[AMT.sup.2]

CASH 0.027 -0.109 0.027
 (0.15) (-0.92) (0.15)
RELSIZE 0.740 *** 0.617 *** 0.732 ***
 (4.87) (3.60) (4.81)
SHARE100 -0.126 -0.192 * -0.124
 (-1.23) (-1.89) (-1.21)
FOCUS -0.254 ** -0.206 * -0.254 **
 (-2.33) (-1.79) (-2.35)
SIZE -0.226 -0.122 -0.028
 (-2.29) (-1.14) (-0.65)
BM -0.042 -0.027 0.058 **
 (-1.00) (-0.65) (2.32)
LEV -2.513 ** -2.185 ** -0.063
 (-2.44) (-2.00) (-0.12)
Inverse mills -0.279
 ratio (-1.27)
Intercept 1.424 * 0.399 -0.800 ***
 (1.85) (0.41) (-2.68)
[R.sup.2] 24.43% 30.45% 23.66%
F-stat. 5.84 *** 5.60 *** 5.88 ***
[Prob. >F] [0.000] [0.000] [0.000]

 [4] [5] [6]
 3-Stage 2-Stage 3-Stage
 Model Model Model

USER

NTYPE 0.901
 (2.96)
NCONTR 0.349 *** 0.376 ***
 (2.75) (2.86)
AMT

[AMT.sup.2]

CASH -0.099 0.023 -0.104
 (-0.82) (0.12) (-0.86)
RELSIZE 0.612 *** 0.735 *** 0.609 ***
 (3.74) (4.76) (3.68)
SHARE100 -0.194 * -0.119 -0.193 *
 (-1.91) (-1.15) (-1.89)
FOCUS -0.230 ** -0.241 ** -0.220 *
 (-2.03) (-2.22) (-1.93)
SIZE -0.019 -0.022 -0.006
 (-0.25) (-0.50) (-0.08)
BM 0.045 * 0.069 *** 0.047 *
 (1.73) (2.86) (1.82)
LEV -0.001 0.15) 0.213
 (-0.00) (0.30) (0.43)
Inverse mills -0.051 -0.117
 ratio (-0.22) (-0.52)
Intercept -0.718 -0.660 ** -0.725
 (-0.99) (-2.24) (-1.00)
[R.sup.2] 30.81% 22.73% 30.46%
F-stat. 5.68 *** 5.68 *** 5.68 ***

[Prob. >F] [0.000] [0.000] [0.000]

 [7] [8]
 2-Stage 3-Stage
 Model Model

USER

NTYPE

NCONTR

AMT 1.45 x [10.sup.-4] ** 1.34 x [10.sup.-4 **
 (2.10) (2.27)
[AMT.sup.2]

CASH 0.010 -0.127
 (0.05) (-1.05)
RELSIZE 0.720 *** 0.586 ***
 (4.56) (3.38)
SHARE100 -0.113 -0.188 *
 (-1.09) (-1.82)
FOCUS -0.215 * -0.192
 (-1.95) (-1.62)
SIZE 0.022 0.138 ***
 (0.61) (2.73)
BM 0.043 -0.002
 (1.60) (-0.04)
LEV 0.311 0.350
 (0.61) (0.69)
Inverse mills -0.534 **
 ratio (-2.12)
Intercept -0.535 -1.837 ***
 (-1.59) (-2.81)
[R.sup.2] 20.62% 28.33%
F-stat. 5.14 *** 5.27 ***
[Prob. >F] [0.000] [0.000]

 [9] [10]
 2-Stage 3-Stage
 Model Model

USER

NTYPE

NCONTR

AMT 1.36 x [10.sup.-4] 9.51 x [10.sup.-5]
 (0.78) (0.62)
[AMT.sup.2] 4.54 x [10.sup.-10] 1.61 x [10.sup.-9]
 (0.06) (0.28)
CASH 0.009 -0.131
 (0.05) (-1.09)
RELSIZE 0.718 *** 0.576
 (4.42) (3.30)
SHARE100 -0.113 -0190 *
 (-1.08) (-1.82)
FOCUS -0.215 * -0.189
 (-1.95) (-1.61)
SIZE 0.022 0.149 **
 (0.57) (2.19)
BM 0.041 -0.011
 (1.03) (-0.22)
LEV 0.316 0.373
 (0.59) (0.71)
Inverse mills -0.569 *
 ratio (-1.98)
Intercept -0.531 -1.929 **
 (-1.55) (-2.53)
[R.sup.2] 20.63% 28.37%
F-stat. 4.85 *** 4.82 ***
[Prob. >F] [0.000] [0.000]

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.
COPYRIGHT 2009 Financial Management Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2009 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:multinational corporations
Author:Lin, J. Barry; Pantzalis, Christos; Park, Jung Chul
Publication:Financial Management
Article Type:Report
Geographic Code:1USA
Date:Sep 22, 2009
Words:15878
Previous Article:Short selling and the weekend effect for NYSE securities.
Next Article:Trade receivables policy of distressed firms and its effect on the costs of financial distress.
Topics:


Related Articles
Corporate expatriate HRM policies, internationalization, and performance in the world's largest MNCs.
From "Starworks" to networks and heterarchies? Theoretical rationale and empirical evidence of HRM organization in large multinational corporations.
National influences on multinational corporation control system selection.
Towards a Capital Structure Theory for the Multinational Company.
The use of entry methods in identifying multinational companies' AdaptStand behavior in foreign markets.
Competence development through business relationships or competitive environment?--Subsidiary impact on MNC competitive advantage.
Subsidiary role and skilled labour effects in small developed countries.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters