Strategic Risk Management
and Product Market Competition
Tim R. Adam
National University of Singapore
Risk Management Institute (RMI)
Amrita Nain
McGill University
Abstract
The objective of this paper is to test whether corporate risk management strategies are
interdependent across firms as Adam, Dasgupta and Titman (2007) and Mello and Ruckes (2006)
have recently suggested. A necessary condition for such interdependence would be that
aggregate hedging decisions affect product prices. Indeed, we find that output prices are less
sensitive to FX shocks if more firms hedge their FX risks. Furthermore, we find that the FX
exposure of derivatives users is negatively correlated with the level of hedging in an industry,
while the FX exposure of derivatives non-users is positively correlated with the level of hedging.
The correlations between exposures and levels of hedging further support the hypothesis of
interdependence of derivatives strategies. Finally, we find that industry structure matters for the
level of hedging in an industry: the fraction of derivatives users is negatively correlated with the
degree of competition. When competition is strong firms may refrain from hedging their FX
risks in order to gain a strategic advantage when prices move in their favour. These results
indicate that firms consider both internal (firm-specific) factors as well as external (industryspecific) factors when deciding upon their risk management strategies.
JEL Classification: G32
Keywords: Corporate risk management, competition, foreign exchange risk, exposure, passthrough
Correspondence address for Adam: Department of Finance, NUS Business School, National University of Singapore, 1 Business
Link, Singapore 117592. E-mail: biztra@nus.edu.sg
Correspondence address for Nain: Faculty of Management, McGill University, 1001 Sherbrooke Street, Montreal, QC H3A 1G5,
Canada. Tel: (514) 398 8440. E-mail: amrita.nain@mcgill.ca.
The authors thank Yiorgos Allayannis, Sugato Bhattacharyya, Greg Brown, Sudipto Dasgutpa, Serdar Dinc, Martin Dierker,
Amy Dittmar, Art Durnev, Philippe Jorion, Lutz Kilian, E. Han Kim, Francine Lafontaine, Adair Morse, Vikram Nanda, Linda
Tesar, Anjan Thakor; participants of the 2004 European Finance Association Meetings, 2005 American Finance Association
Meetings, 15th Annual Derivatives and Risk Management Conference (2005), 16th Annual Conference on Financial Economics
and Accounting (2005); and participants of finance seminar series at – Bank of Canada, College of William and Mary, H.E.C.
Montreal, McGill University, University of California (Irvine), University of Georgia, University of Miami, University of
Michigan, University of Oklahoma, University of Toronto and University of Virginia (Darden) for helpful comments. Any errors
are our own.
According to the Modigliani-Miller paradigm, corporate risk management is irrelevant for firm
value since individual investors can adjust their portfolios to obtain the desired risk exposure. However,
corporate use of derivatives has risen steadily. The International Swaps and Derivatives Association
reports that the notional amounts of interest rate and currency derivatives held by its members, many of
whom are end-user corporations, increased from $865 billion in 1987 to $213 trillion in 2005.1 Existing
theories appeal to managerial incentives or market imperfections such as taxes, financial distress costs
and underinvestment costs to explain firms’ incentive to lower volatility through hedging. A number of
empirical studies examine whether firms’ hedging decisions are consistent with existing theory.2 Crosssectional evidence regarding these firm-specific reasons for hedging is mixed, with different empirical
studies finding conflicting evidence for most theoretical predictions.
Although our understanding of who hedges and why is still poor, a number of studies show that
hedging foreign currency and interest rate risk increases firm value. The documented ‘hedging premium’
appears to be economically large and statistically significant ranging from 4% in Allayannis and Weston
(2001) to 14% in Carter, Rogers and Simkins (2006). The size of the hedging premium has lead to
concerns that the results may be spurious or that the improvement in value may be caused by some other
unobserved factor that also determines the decision to hedge (see Guay and Kothari (2003)). More
recently, Jin and Jorion (2006) show that in the oil and natural gas industry, hedging does not affect a
firm’s market value and thus, cast further doubt on the link between hedging and firm value.
To obtain an understanding of a firm’s decision to engage in derivatives hedging and the value
implications of hedging, we must have a good grasp of how derivatives affect exposure to the underlying
risk. Empirical tests of existing risk management theory consider derivatives usage to be risk-reduction
tool. This assumption is supported by Guay (1999) and Allayannis, Ihrig and Weston (2001) who
document lower foreign exchange exposures for firms that use foreign currency derivatives. In contrast,
Hentschel and Kothari (2001) find no differences in the over-all risk measures of derivatives users and
non-users. The mixed evidence in all these areas of research suggests that although corporate risk
management literature has made progress on many fronts, we have a lot to learn about how derivatives
affect a firm’s risk profile, a firm’s incentive to engage in risk management and whether hedging
increases value.
This paper shows that an important consideration in a firm’s incentive to hedge has remained
largely unexplored in the finance literature. We provide evidence that the prevalence of derivatives
1
Source: International Swaps and Derivatives Association, Inc. http://www.isda.org/.
For risk-management theories, see Stulz (1984), Smith and Stulz (1985), Froot, Scharfstein and Stein (1993),
DeMarzo and Duffie (1991, 1995), Breeden and Viswanathan (1998), Leland (1998) and Morellec and Smith
(2003). For empirical findings, see Nance, Smith and Smithson (1993), Dolde (1995), Mian (1996), Tufano (1996),
Geczy, Minton and Schrand (1997), Schrand and Unal (1998), Haushalter (2000), Graham and Rogers (2002), and
Knopf, Nam and Thornton (2002). For evidence from a case study, see Brown (2001).
2
1
hedging in an industry affects how product prices respond to common cost shocks faced by all firms in an
industry. Once this effect of hedging on product prices is allowed for, two interesting results emerge.
First, an individual firm’s exposure to the shock depends not only on its own hedging decision, but also
on the hedging decisions of its competitors. Second, firms face lower exposure when they conform to the
majority’s decision. These are important considerations for a firm’s incentive to hedge and the value
implications of hedging. Insofar as the risk-management literature ignores these issues, the mixed
empirical evidence is not surprising.
Using comprehensive hand-collected data on foreign currency derivatives (FCD) usage, we show
that industry output prices become less sensitive to exchange rate shocks when more firms in the industry
use FCD. The response of output prices to exchange rate shocks, often referred to as exchange rate passthrough, has been extensively studied but not previously been linked to FCD usage. Our results indicate
that when FCD usage in an industry is non-existent, the pass-through elasticity is -0.08. A negative passthrough elasticity implies that as the dollar depreciates and import costs rise, U.S. firms charge higher
prices in the domestic market. However, as FCD usage in an industry rises, pass-through elasticity
approaches zero. For an industry with an average level of FCD usage, depreciation of the U.S. dollar is
not associated with an increase in industry selling prices. This result holds after controlling for industry
concentration, reliance on imported inputs, import competition faced by an industry, export sales and
industry capital intensity. The results are robust to alternative estimation methods and alternative
definitions of hedging in the industry. Our empirical methodology allows FCD usage to be an
endogenously determined variable. To our knowledge, this is the first paper to show that the use of FCD
reduces the pass-through of foreign exchange shocks to product prices.
In an imperfectly competitive industry, the dampening effect of industry-wide FCD usage on the
correlation between industry prices and exchange rate shocks has noteworthy implications for the
currency exposure of individual firms. The intuition is similar to the arguments put forth by De Meza
(1986) in the context of firms choosing between fixed and variable cost technologies and later applied by
Maksimovic and Zechner (1991) to a firm’s choice of optimal capital structure. Consider a firm that
leaves its foreign exchange exposure unhedged (an FCD non-user) when most of the other firms in the
industry are engaged in derivatives hedging. This firm is exposed to any cost shocks that might arise due
to changes in the exchange rate. However, since pass-through is low and industry prices insensitive to
exchange rate movements, the FCD non-user does not obtain changes in selling prices to offset the cost
shocks. An FCD-user, on the other hand, experiences neither the currency cost shock nor a price change.
Therefore, in this industry, FCD users (non-users) have low (high) exchange rate exposure. In contrast,
consider an industry where very few firms use FCD. In this industry, exchange rate shocks are passed
through to prices. FCD non-users have low exposure because changes in selling prices offset exchange
2
rate induced cost shocks. FCD users on the other hand, experience variation in selling prices but have no
exchange rate induced cost shock. Thus, if only a few firms in the industry use FCD, the profitability of
FCD users (non-users) is expected to fluctuate more (less) with exchange rate shocks.
Our empirical tests provide strong evidence of the above-mentioned patterns in the exposure of
FCD users and non-users. We estimate the foreign exchange exposure of each firm and then examine how
the exposure varies cross-sectionally with the level of hedging in a firm’s industry. We find that as the
extent of hedging in an industry increases, the foreign exchange exposure of an FCD user declines while
that of an FCD non-user increases. Thus, a firm faces lower exposure to the underlying exchange rate risk
when its hedging decision conforms to that of the majority. This result is robust to alternative estimation
methods and alternative definitions of hedging in an industry. In our argument, the observed association
between an individual firm’s exposure and the hedging decision of its competitors arises because
exchange rate pass-through in an industry is lower when hedging is more prevalent. To test this argument,
we directly examine the effect of exchange rate pass-through on the exposure of FCD users and nonusers. The results are strongly supportive of the pass-through explanation. As the extent of exchange rate
pass-through to domestic prices increases, the foreign exchange exposure of FCD users increases while
that of FCD non-users decreases. As expected, these results are driven by imperfectly competitive
industries where firms have the ability to affect output prices. When the sample is restricted to
competitive industries (that is, industries with very low values of the Herfindahl index, the four-firm
concentration ratio or price-cost margin), the industry level hedging measure is not related to the exposure
of individual FCD users or non-users. That the link between an individual firm’s exposure and the
hedging choices of its competitors holds only in industries where some pricing power exists is further
evidence that financial hedging influences product market decisions.
Our findings raise questions about the previously documented relation between currency hedging
and firm value. Derivatives users are valued higher presumably because they reduce volatility and the
associated financial distress costs, underinvestment costs etc. Our results show that FCD non-users
already enjoy low exposure when FCD usage in their industry is low and exchange rate pass-through
high. Thus, their need to engage in derivatives-hedging is low. There is little reason for these firms to
trade at a discount relative to FCD users. Therefore, we revisit the relation between FCD usage and firm
value conditional on the extent of hedging in a firm’s industry. We find that the previously documented
hedging premium is driven only by industries where FCD usage is very widespread. The decision to not
use FCD appears to hurt value only when most competitors are engaged in FCD hedging. While this is
consistent with our finding that FCD non-users face high currency risk only when many of their
competitors use FCD, it leaves a fundamental question unanswered. Why would a handful of firms refrain
3
from using FCD even though they could improve value by using derivatives like most of their
competitors do?
Further analysis shows that these lower-valued FCD non-users in highly hedged industries had
lower profit margins, lower return on assets and lower stock returns than the rest of the sample in the five
years preceding our value analysis. This suggests that the decision to remain unhedged when many
competitors consider it prudent to hedge may be a product of the same unobserved firm characteristics
that lead to underperformance and lower firm value. In fact, once we allow these measures of past
performance to determine a firm’s decision to hedge, the positive association between FCD usage and
firm value disappears even in industries where FCD usage is widespread. Over all, our analysis shows
there is no observable difference in the value of FCD users and non-users once the hedging choices of
competitors are taken into account. When considered in the context of their industry, firms appear to
make optimal choices about foreign currency risk management. This finding is consistent with the
industry-equilibrium argument of Adam, Dasgupta and Titman (2006) where the level of hedging in an
industry adjusts till no difference remains between the value of a hedged and unhedged firm.
The results in this paper are relevant for understanding past evidence as well as for guiding future
work on corporate risk management. Empirical studies that consider FCD usage tantamount to riskreduction and studies that compare the exposure or value of FCD users and non-users without regard to
competitors’ hedging choices are bound to arrive at mixed or conflicting evidence. The use of derivatives
as an empirical proxy for hedging is sometimes criticized on the grounds that some firms admittedly
speculate with derivatives and may consequently face higher risk.3 Our results suggest that even if a
firm’s derivatives program is completely non-speculative in nature, the firm can face greater exposure to
the underlying risk if its competitors are not hedging the same risks. Therefore, it is imperative that
researchers consider the interaction between financial hedging and output markets when using derivatives
to test corporate risk management theory. Finally, we note that since a firm’s exposure is affected by the
hedging decisions of competitors, it is no surprise that models that examine risk-management decisions in
a single-firm framework enjoy little empirical support. We provide persuasive evidence that a firm’s
decision to hedge must arise out of a model that allows for strategic interactions between firms’ hedging
and output choices. Adam, Dasgupta and Titman (2006) and Mello and Ruckes (2006) take a step in this
direction by modeling the hedging decision in an industry setting. Our results complement these theories
3
The Wharton/CIBC World Markets 1998 Survey of Financial Risk Management by U.S. Non Financial Firms
asked firms what best describes the motivation behind their risk management activities: 40% of firms chose higher
profits rather than lower volatility. Moreover, 32% of firms that use derivatives reported that their market view of
exchange rates leads them to “actively take positions” at least occasionally (see Bodnar, Hayt and Marston (1998)).
Finally, the findings of Faulkender (2004) and Adam and Fernando (2006) suggest that firms time the market with
derivatives and try to generate positive cash flows with their derivatives positions.
4
and suggest that future empirical research on the decision to hedge draw on the empirical predictions of
such industry based models.
The rest of this paper is organized as follows. Section 1 provides a motivation for the empirical
questions addressed in this paper along with references to related literature. Section 2 describes the data.
Section 3 describes the methodologies used and also presents empirical results. Section 4 contains
robustness checks. Section 5 concludes.
1. Motivation for Empirical Tests
In existing risk management literature, a firm’s exposure to a risk factor is generally considered to
be a function of the firm’s own characteristics and its own decision to hedge. For example, a firm’s
foreign exchange exposure is determined by its involvement in foreign trade, its decision to engage in
financial or operational hedges, and the financial constraints faced by the firm (see for example Bartov
and Bodnar (1994), He and Ng (1998)).4 However, in imperfectly competitive industries, a firm’s
exposure to the underlying risk should depend not only on its own hedging decision but also on the
hedging decisions of competitors. To see why, consider an imperfectly competitive industry in which all
firms are subject to a common shock to their marginal cost of production. When the shock occurs, firms
that are exposed to the shock adjust their profit maximizing output accordingly. If all firms face the cost
shock and adjust output, industry prices co-vary with the cost shock (see De Meza (1986)). Suppose that,
prior to the realization of the cost shock, firms are able to enter into derivative contracts that enable them
to lock in their cost of production. Ceteris paribus, firms that have completely protected themselves from
the shock make no change to output when the shock occurs. As more firms choose to hedge the shock,
industry price becomes less sensitive to the cost shock because fewer firms adjust output.5
This dampening effect of hedging on the correlation between industry prices and costs implies
that the volatility of an individual unhedged (hedged) firm increases (decreases) with the fraction of
hedged firms in the industry. In an industry where all firms hedge the shock, prices do not fluctuate with
the cost shock. If a firm in this industry remains unhedged, it faces the cost shock but does not obtain an
offsetting change in prices. Hedged firms, on the other hand, face constant costs as well as constant
Exceptions are Bodnar and Gentry (1993) who model a firm’s exchange rate exposure as a function of its
industry’s involvement in foreign trade and Allayannis and Ihrig (2001) who suggest that firms in more competitive
industries face greater foreign exchange exposure.
5
The hedge could be a forward contract by the firm to buy Y units of an input at t=1 at $X per unit. There are two
situations in which this type of contract may not affect a firm’s marginal cost of production. First, if the market price
M at t=1 is higher than X, it may be optimal for the firm to forego production and sell the input on the market after
purchasing it at X, pocketing a profit of $(M-X) per unit. Second, if the contract amount Y falls short of the firm’s
input requirement forcing the firm to obtain the remaining on the market, the marginal cost will be determined by
the market price prevailing at t=1 and not by the contract entered into at t=0. While these scenarios bias us against
finding support for the hypothesis that derivatives-hedging reduces price-cost correlation, they also increase the onus
of eliminating alternative explanations if support is found.
4
5
prices. In this industry, profit volatility of unhedged (hedged) firms is high (low). In contrast, when most
firms in an industry are unhedged, prices co-vary with costs, causing the profit volatility of unhedged
firms to be low. If a firm in this industry chooses to hedge, it has certain costs but faces uncertain prices
because of the output choices of its unhedged competitors. Thus, in largely unhedged industries,
unhedged (hedged) firms have low (high) profit volatility.
The fundamental intuition is that as the number of firms adopting a given production technology
increases, output price corresponds more closely with the cost of production, providing firms with a
natural shelter against changes in production costs (see De Meza (1986), Maksimovic and Zechner
(1991)) and Adam, Dasgupta and Titman (2006)). In this paper, we investigate empirically whether
derivatives hedging affects the sensitivity of product prices to the shock being hedged, thereby making the
exposure of firms in an industry interdependent. If these relationships are found to be robust, then
empirical and theoretical research on who hedges with derivatives and why will need considerable
rethinking.
Since, in practice, foreign currency exposure is the most commonly hedged risk, the tests are
conducted using comprehensive, hand-collected data on the use of FCD by publicly listed firms in the
United States. We address the two key questions that arise from our discussion above. First, does the use
of FCD reduce the pass-through of exchange rate shocks to domestic producer prices? There is an
extensive literature on the effect of exchange rate shocks on import and export prices and a relatively
sparse literature on the impact of exchange rates shocks on domestic prices.6 However, none of these
pass-through studies examine how the use of FCD affects the pass-through of exchange rate shocks to
product prices. Thus, our first question contributes to both the risk-management and pass-through
literatures. The second question we address is the following. Does the exposure of an FCD user (FCD
non-user) to foreign exchange fluctuations decrease (increase) as the level of FCD usage in an industry
rises and pass-through declines? Numerous papers study firms’ exchange rate exposure measured as the
sensitivity of stock returns to changes in the exchange rate.7 The link between exchange rate fluctuations
and stock returns is known to be weak. It has been suggested that the weak relation between exchange
rate shocks and stock returns exists because of the failure to account for firm’s risk-management
practices. Our second question addresses this concern by explicitly accounting for firms’ use of FCD.
Arguably, the tests in this paper would be the most directly applicable to commodity hedging data
where the commodity being hedged is an input used by the industry in question. However, this paper uses
6
For exchange rate pass-through to import or export prices see for example, Mann (1986), Feenstra (1987),
Krugman (1987), Froot and Klemperer (1989), Ohno (1989), Knetter (1989, 1993), Marston (1990), Yang (1997)
and Bodnar, Dumas and Marston (2002). For exchange rate pass-through to domestic prices, see Feinberg (1986,
1989).
7
See for example, Jorion (1990), Bodnar and Gentry (1993), Bartov and Bodnar (1994), He and Ng (1998) and
Griffin and Stulz (2001 ).
6
foreign exchange hedging data instead of commodity hedging data for the following reasons. First,
commodity hedging is not as widespread as foreign currency hedging and is limited to a handful of
industries which makes cross-sectional tests infeasible. An additional data advantage of working with
foreign currency hedging is that we can use the same macroeconomic shock (exchange rate fluctuations)
for all industries, taking into account each industry’s involvement with foreign trade. In contrast, with
commodity hedging, one would require a different data series to capture the shock for each industry
depending on the commodity being hedged. This places additional constraints on already sparse data.
2. Derivatives Data
Data on currency derivative holdings of U.S. firms as of fiscal year 1999 are obtained by
searching the financial footnotes and Management Discussion and Analysis of SEC 10-K filings for text
strings such as “hedg,” “swap,” “cap,” “forward” etc. SFAS 105 requires all firms to report information
about financial instruments with off balance sheet risk for fiscal years ending after June 15, 1990. If a
reference is made to any of the search terms and the firm is not a financial firm, we read the surrounding
text to confirm that it refers to foreign currency derivatives holdings and classify the firm as foreign
currency derivatives (FCD) user in that year.8 Information on the gross notional amounts of foreign
exchange forwards, swaps and options outstanding is collected as of fiscal year ending in 1999. In cases
where there were no contracts outstanding as of fiscal year end, but the firm did engage in foreign
exchange risk-management during the year, we take the notional amounts that expired during the year
1999. If there are no references to the keywords, the firm is classified as an FCD non-user in that year.
These data are matched with COMPUSTAT and only non-financial firms that have positive values for net
sales, total assets and market value of equity are retained in the sample. We also collect less detailed
hedging data for 1997. A firm is classified as a foreign currency derivatives user in 1997 if it discloses the
use of foreign exchange forwards, swaps and options as of fiscal year ending in 1997.
An advantage of this comprehensive sample is that it enables us to determine, for each firm, how
many competitors use foreign currency derivatives. Most previous studies on foreign exchange hedging
either focus on a single industry or use sample-selection criteria that do not give a complete picture of
hedging activity in any given industry. For example, Allayannis and Weston (2001) use a sample of nonfinancial firms that have total assets of more than 500 million in each year between 1990 and 1995.
Geczy, Minton and Schrand (1997) study Fortune 500 non-financial firms. Graham and Rogers (2002)
8
The arguments presented in this paper are applicable to linear derivatives contracts (like forward contracts) and do
not hold for non-linear contracts like foreign currency options. Our data show that 91% of derivatives users employ
forward contracts. Moreover, less than 10% of the total notional amounts of FCD outstanding are accounted for by
options contracts. Given the predominance of forward contracts in the data, excluding options makes no difference
to our overall conclusions.
7
use a randomly selected sample of non-financial firms. Our sample, on the other hand, is more
representative of the universe of firms.
For certain tests, we focus on firms that face ex-ante exchange rate exposure. This allows us to
interpret the absence of derivatives as a choice not to use derivatives, rather than an indication of lack of
exposure to foreign exchange risk. Following Graham and Rogers (2002), firms are defined as having exante currency exposure if they disclose foreign assets, sales or income in the COMPUSTAT Geographic
segment file, or disclose non-zero values of foreign currency adjustment, exchange rate effect, foreign
income, or deferred foreign taxes in the annual COMPUSTAT files. From the initial sample of 10,400
Compustat firms, 4,300 firms (forty-one percent) face ex-ante exchange rate exposure. 549 firms facing
ex-ante exchange rate exposure engage in currency derivatives hedging. The gross notional amounts of
foreign exchange swaps, forwards and options outstanding are summarized in Table I, Panel A. The
descriptive statistics are comparable to previous studies. Graham and Rogers (2002) report a mean foreign
currency derivatives notional amount of $558 million for the year 1994-1995 which is on average 8.06
percent of total assets. The mean in our sample is $745 million, 8.89 percent of total assets. The mean
notional amount of swaps, forwards and options scaled by total assets are 4.90 percent, 7.84 percent and
6.35 percent respectively. These numbers are comparable to those reported by Purnanandam (2007) who
also uses a more comprehensive sample to study derivatives usage.
Panel B of Table I reports the distribution of two measures of industry level hedging at the 3-digit
SIC level. Both measures are a market value-weighted estimate of the fraction of hedgers in an industry.
The first measure is based on the sub-sample of firms facing ex-ante exchange rate exposure as defined
above. It is calculated as the sum of market values of FCD users in a 3-digit SIC industry who face exante currency exposure divided by sum of market values of all firms in the industry who face ex-ante
currency exposure. 9 Market value of a firm is the market value of equity plus the book value of debt. We
use a market value-weighted measure of the fraction of hedgers to account for the possibility that firms
differ in their ability to affect prices. Larger firms account for a bigger fraction of industry output and thus
their hedging choices are more important for the industry. The first column of Panel B shows that in half
the industries, the extent of hedging is less than 2.5%. In the top quartile of industries, the fraction of
hedgers is 48% or higher.
9
It is not mandatory for firms to disclose whether they are long or short on a currency and whether they are hedging
foreign revenues or expenses. This restricts us from selecting only those firms that hedge foreign expenses.
However, since the total number of firms in an industry that use currency derivatives is likely to be positively
correlated with the number of firms that use currency derivatives to hedge expenses, our measure of industry
hedging serves as an acceptable proxy. Moreover, changes in revenue translate into marginal cost shocks if a firm
has positive and convex costs of raising external finance (see Froot, Scharfstein and Stein (1993) and Mello and
Ruckes (2006)). Therefore, insofar as firms exposed to foreign exchange revenue shocks are financially constrained,
including them in our sample is not unreasonable, and at worst, biases us against finding support for the arguments
made above.
8
A firm’s decision to use FCD depends on whether it is exposed to exchange rate risk. Exposure to
foreign exchange shocks is difficult to identify accurately. For example, our measure of ex-ante exposure
ignores firms that have no involvement in foreign trade and yet face exchange rate exposure due to the
export or import activities of competing firms. To avoid biasing our results due to the definition of exante exposure used, we also calculate a measure of industry hedging based on the entire sample regardless
of ex-ante foreign exchange exposure. It is calculated as the sum of market values of FCD users in the
industry divided by sum of market values of all firms in the industry. Panel B of Table I shows that, as
expected, the fraction of hedgers based on the entire sample is lower than the fraction of hedgers based on
exposed firms only.
In Table II, a detailed break up of hedging by industry is provided for 43 out of the 48 FamaFrench industries.10 The extent of hedging is the highest in the following industries - Recreation,
Automobiles and Trucks, Shipbuilding and Railroad Equipment, and Electronic Equipment. It is the
lowest in Printing and Publishing, Tobacco, and Entertainment. It is worth noting here that weighting the
fraction of hedgers by market value can make a big difference. For example, Fama-French industry
number 35 (Computers) has 1500 firms out of which 752 face ex-ante exchange rate exposure. In this
industry, 55 firms disclosed the use of FCD in 1999. Of these 55 FCD users, 53 are classified as have exante exposure. An unweighted measure of the fraction of hedgers would imply that the extent of hedging
is 3.6% using the entire sample and 7% using the sample of exposed firms. However, since the largest
players in the market (e.g. Microsoft, IBM, and Oracle) use FCD, the market value-weighted measures
are 41% and 49% respectively.
For all the tests in Sections 3, we use the market value-weighted measure based on firms with exante exposure. In the robustness section (Section 4), we demonstrate that our main results hold when we
use the market value-weighted measure of industry hedging based on the entire sample regardless of exante exposure. In Section 4 we also discuss the outcome of using an unweighted measure of the fraction
of hedgers in an industry.
3. Methodology and Results
We begin our empirical investigation with an overview of the foreign exchange exposures of
FCD users and FCD non-users in Section 3.1 below. In Section 3.2, we study the pass-through of
exchange rate shocks to domestic prices conditional on the extent of hedging in the industry. In Section
3.3, we examine how the foreign exchange exposures of FCD users and FCD non-users are affected by
the hedging decisions of the industry as a whole and by the extent of exchange rate pass-through in the
10
We exclude Financial Services, Real Estate, Trading, and Other due to lack of hedging data.
9
industry. Section 3.4 examines the relation between derivatives usage and firm value in light of the
findings of Sections 3.2 and 3.3.
3.1. Foreign Exchange Exposure of FCD Users and FCD Non-Users
In this sub-section, we calculate the foreign exchange exposure of FCD users and FCD non-users
and conduct simple univariate comparison of the exchange rate exposures of the two groups.
3.1.1 Methodology
Existing studies examine exchange rates exposure by regressing stock returns on changes in
exchange rates, controlling for the overall market return (see for example Jorion (1990), He and Ng
(1998) and Dominguez and Tesar (2001)). Following this literature, we estimate the following time-series
regression for each hedged (FCD user) and unhedged (FCD non-user) firm in the sample using monthly
data.
rit  i0  ix EXCH t  imrmt   it
(1)
In equation (1), rit is the monthly rate of return on the firm’s stock for the years 1996 till 2000; rmt
is the corresponding monthly rate of return on the value-weighted market index. The variable EXCH t is
the monthly change in value of the U.S. dollar orthogonal to the market return.11 The exchange rate is the
trade-weighted value of the U.S. dollar in terms of its major trading partners as calculated by the Federal
Reserve Board. The coefficient  ix measures a firm’s exposure to exchange rate movements after taking
into account the overall market’s exposure to currency fluctuations. Firms are classified as FCD users if
they disclose the use of foreign currency derivatives in 1999. All other firms are classified as FCD nonusers.12 In selecting the estimation period for stock return exposure, we face a trade-off between having a
sufficiently long time series to estimate equation (1) for each firm, while remaining reasonably close to
the two sample years for which hedging data are available (1997 and 1999). We choose an estimation
period from 1996 till 2000. This ensures 60 data points for each time-series estimation of  ix .
3.1.2 Results
Table III presents summary statistics of  ix for all firms, and separately for FCD users and FCD
non-users. Note that we have not made any ex-ante judgments about which firms should display high
Using changes in exchange rate that are orthogonal to the market return,  EXCH , may lead to econometric
difficulties (see Jorion (1991)). We repeat the analyses using changes in the exchange rate, X, itself instead of the
residuals from a regression of exchange rates on the stock market return. The results of the paper remain unchanged.
12
In alternative specifications we classify FCD users as firms that disclosed the use FCDs in both 1997 and 1999
and FCD non-users as firms that did not disclose the use of FCDs in either 1997 or 1999. This alternative
classification makes no difference to the results in the paper.
11
10
stock return exposure to exchange rates. The coefficients are estimated for all firms that have
uninterrupted stock return data from 1996-2000 and also meet the data requirements of the multivariate
analysis outlined later in Section 3.3 below. As in previous studies, we find that the median foreign
exchange exposure coefficients are small. Approximately 11% of the firms have significantly positive or
significantly negative foreign exchange exposure. We are interested in comparing the foreign exchange
exposure of FCD users with that of FCD non-users regardless of whether the exposure is positive or
negative. Therefore, we take the absolute value of all exposure coefficients and calculate the mean
‘absolute’ exposure for the two groups of firms. Panel A of Table III shows that, on average, firms using
FCD have significantly lower exposure to foreign exchange fluctuations than firms that do not use FCD.
In Panels B and C we present the same analysis separately for firms with positive and negative  ix . In
both sub-samples, the exposure of FCD users is significantly smaller (closer to zero) than that of FCD
non-users. Thus, univariate tests suggest that, on average, FCD users are less exposed to currency
fluctuations than FCD non-users.13
The histograms presented in Figure I provide a pictorial comparison of the distribution of
exposure coefficients of FCD users and FCD non-users. Note that to make the scale of the two histograms
visually comparable, we have dropped 1% of observations on each side of the distribution for FCD nonusers. Despite this, the distribution plots make it evident that the exposure coefficients of FCD users are
much more tightly distributed around 0. FCD non-users display a greater incidence of high positive or
high negative exposure to exchange rate fluctuations. It is also apparent that both groups of firms display
high cross-sectional variation in the extent of foreign exchange exposure. We have hypothesized that
competitors’ hedging choices affect an individual firm’s exposure. Specifically, we expect FCD users
(non-users) to have lower (higher) foreign exchange exposure when more firms in the industry use FCD.
Next, we conduct a univariate test of this hypothesis.
In this test, we first focus on the sample of FCD users and split this sample into two groups. The
first group contains 5% of FCD users with the most positive foreign exchange exposure coefficients and
5% of FCD users with the most negative foreign exchange exposure coefficients. The second group
contains 10% of FCD users with exposure coefficients closest to zero. We compare the average level of
industry hedging for the two groups of FCD users. As shown in Table IV, FCD users with extreme
exposures (first group), belong to industries where, on average, 37% of the industry is hedged. In contrast,
FCD users with exposures close to zero (second group) belong to industries where, on average, 48% of
the industry is hedged. The difference in the level of industry hedging for the two groups of FCD users is
13
This finding is consistent with the results of Allayannis and Ofek (2001) who also find that FCD users have lower
foreign exchange exposures.
11
statistically significant. Thus, exposure coefficients of FCD users are lower when they belong to
industries where FCD usage is more prevalent.
We repeat the same test for our sample of FCD non-users. FCD non-users are divided into two
groups. The first group contains 5% of non-users with the most positive exposure coefficients and 5% of
non-users with the most negative exposure coefficients. The second group contains 10% of FCD nonusers with exposure coefficients closest to zero. In Table IV, we see that FCD non-users with extreme
exposures belong to industries where, on average, 39% of the industry is hedged. In contrast, FCD nonusers with exposures closer to zero belong to industries where, on average, 34% of the industry is hedged.
Again, the difference in the level of industry hedging for the two groups of FCD non-users is statistically
significant. Exposure coefficients of FCD non-users are lower when they belong to industries where FCD
usage is less prevalent.
This univariate comparison supports the hypothesis that the foreign exchange exposure of a firm
is smaller when the hedging strategy of the industry is similar to its own.14 We argued that the link
between an individual firm’s exposure and the hedging choices of competitors arises because hedging
interacts with output markets. In the next section, we investigate this idea further.
3.2. Industry Hedging and Pass-through
This section tests the hypothesis that industry output prices are less sensitive to exchange rate
shocks in industries where FCD usage is more prevalent.
3.2.1 Methodology
Most of the literature on the pass-through of exchange rate shocks to product prices examines
how import prices (the price at which foreign firms sell in the domestic market) respond to exchange rate
shocks. Since we do not have hedging data on foreign firms, we cannot study the effect of derivatives
usage on import prices. Instead we follow the methodology of a smaller stream of literature which
examines the pass-through of exchange rate shocks to U.S domestic producer prices (see, for example,
Feinberg (1989)). The dependent variable is the relative producer price index, RPPI, at the three-digit SIC
level, calculated as the producer price index divided by the overall GDP price deflator.15 The producer
price data are obtained from the Bureau of Labor Statistics. The overall GDP price deflator is obtained
from the Bureau of Economic Analysis. Foreign exchange movements, X, are measured as the trade14
We present a more detailed, multivariate version of this test in Section 3.3.
We use the producer price index (PPI) instead of the consumer price index (CPI) because the PPI more accurately
captures the price of commodities produced in the United States. The CPI is the price paid by U.S. customers for a
basket of goods and this basket of goods includes imported products. Since we do not have data on derivatives usage
by foreign firms, we exclude import prices from our analysis and focus only on the pricing behavior of domestic
firms.
15
12
weighted value of the U.S. dollar in terms of its major trading partners as calculated by the Federal
Reserve Board. An increase in X indicates an appreciation of the U.S. dollar. We obtain monthly data on
producer prices and exchange rates from 1996 till 2000, the same period used to estimate stock return
exposure in Section 3.1.
The effect of various industry characteristics on the extent of pass-through is obtained by
interacting the industry characteristics with the exchange rates in a pooled regression. The main industry
variable we are interested in is the fraction of hedgers, FRACTION, calculated at the three-digit SIC level
as discussed in Section 2 above. We use the fraction of hedgers based on the sub-sample of firms that face
ex-ante exchange rate exposure. In industries that are dependent on imported inputs, output prices are
expected to increase as the U.S. dollar depreciates (because the cost of importing increases).16 Therefore,
the pass-through of a depreciating dollar into higher domestic prices should be more pronounced in
industries that rely on imported inputs. An industry’s reliance on foreign inputs, FORINP, is calculated as
in Allayannis and Ihrig (2001) using monthly industry import data provided by United States
International Trade Commission (USITC) and the 1997 benchmark input-output tables provided by the
Bureau of Economic Analysis of the U.S. Department of Commerce.
The extent of import competition faced by an industry is also likely to affect pass-through. When
the U.S. dollar appreciates, the cost of production of a foreign firm falls relative to that of a U.S. based
competitor. If foreign firms reduce prices in the U.S. market when the dollar appreciates, domestic firms
face more pressure to remain competitive by lowering prices. 17 Thus, when examining the pass-through
of exchange rate shocks to selling prices of domestic firms, it is important to control for the extent of
foreign competition in the U.S. market – which we term ‘import penetration’ as in Feinberg (1989).
Import penetration, IMPORTS, for an industry is calculated as the monthly general customs value of final
goods imports scaled by domestic shipments for that industry in that year.18 Feinberg (1989) finds that
pass-through to domestic prices is lower in more capital intensive industries. Therefore, we control for
capital intensity, KS, calculated as the average value of total assets as a percentage of sales per three-digit
SIC industry group. Since greater competition in an industry is associated with a loss in pricing power,
exchange rate pass-through is less likely in more competitive industries (see Allayannis and Ihrig (2001)).
16
Firms that have foreign revenues but do not import may also experience cost shocks from exchange rate changes.
According to Froot, Scharfstein and Stein (1993), in the presence of convex costs of external financing, a firm’s
marginal cost of production will be affected by exchange rate shocks.
17
Past literature suggests that an appreciating dollar can make domestic markets more competitive. Froot and
Klemperer (1989) show that foreign firms compete aggressively by lowering prices in response to permanent dollar
appreciations, but may compete less aggressively in response to temporary fluctuations in the dollar. Knetter (1994)
provides weak evidence that the U.S. domestic market became more competitive during the large dollar appreciation
in the 1980s.
18
Value of domestic shipments is obtained from the Annual Survey of Manufacturers conducted by the Census
Bureau.
13
We include the four-firm concentration ratio, CONC, obtained from the 1997 U.S. Census of
Manufacturers as a proxy for competition.19 If an industry is engaged in exports in addition to imports, it
is somewhat naturally hedged against exchange rate fluctuations. In this case, domestic prices may be less
response to exchange rate shocks. Therefore, we also include industry exports, EXPORTS, as a control
variable. Industry exports are obtained from USITC and scaled by domestic shipments. Finally, the U.S.
dollar LIBOR, r, is included to control for the overall macroeconomic environment. Table V presents
correlations of the industry level variables. The three foreign trade variables FORINP, IMPORTS and
EXPORTS are highly correlated with correlation coefficients ranging from 0.28 to 0.71. We address this
concern in the empirical methodology below.
Equation (2) below presents our regression model. Since product prices, exchange rates and
interest rates may be non-stationary, these series are included in log differences. The panel data may
involve partially correlated errors across time and across industries. We use Newey-West standard errors
to account for correlated errors across time and account for industry fixed effects parametrically by
including industry dummies.
 ln RPPI jt   0  1 ln X t 1   2  ln X t 1 * FRACTION
jt
  3  ln X t 1 * FORINP jt   4  ln X t 1 * IMPORTS
  5  ln X t 1 * KS jt   6  ln X t 1 * CONC j   7  ln X t 1 * EXPORTS jt   8  ln rit   jt
jt
(2)
Although not shown, all variables that appear in the interaction terms are also included separately
as control variables. To reduce the problem of multicollinearity due to the foreign trade variables, we
scale the foreign trade variables by subtracting their time series mean from each observation. This
significantly reduces the correlation of EXPORTS with IMPORTS and FORINP. However, the correlation
between FORINP and IMPORTS remains high. Therefore, we include FORINP and IMPORTS in separate
regressions.
We need to be wary of a potential endogeneity problem in this regression. If lower pass-through
increases the exposure of unhedged firms to exchange rates, thereby, increasing their incentive to hedge,
then the level of hedging in the industry will be an endogenously determined variable. To address this
concern, we need an instrument for the level of FCD usage in an industry. The instrument must be
correlated with the level of foreign currency hedging in an industry but unrelated to foreign exchange pass
through in that industry. The instrument we use is the fraction of firms in the industry that use interest rate
derivatives. Previous studies show that firms that use derivatives to hedge foreign currency risk are also
more likely to use other hedging instruments like interest rate derivatives, possibly due to the high initial
cost of setting up hedging programs (see for example, Geczy, Minton and Schrand (1997)). Consistent
with this, our data show that the fraction of firms using interest rate derivatives in an industry is
significantly positively correlated with the fraction of firms using foreign currency hedging instruments.
19
In alternative specifications, the Herfindahl index is used instead of the four firm concentration ratio.
14
Since interest rate derivatives usage is unlikely to affect or be affected by foreign exchange pass-through
to product prices, it meets the requirements of a good instrument.20
It is possible that in the presence of industry fixed effects, the Newey-West estimation delivers
biased standard errors (see Petersen (2005)). In such cases, standard errors clustered along a crosssectional dimension will be unbiased. Therefore, in Section 4, we estimate equation (2) using clustered
standard errors and time dummies. Since cluster standard errors are robust to arbitrary intra-group
correlations, the methodology in Section 4 also addresses the concern that there is almost no variation in
FRACTION over time.
3.2.2 Results
Table VI presents results for equation (2). In Column I, the measure of import competition,
IMPORTS is excluded. In Column II, the measure of imported inputs, FORINP is excluded. Recall that
lower values of the exchange rate measure represent depreciation of the U.S. dollar and, consequently,
higher cost of imported inputs. The coefficient on the exchange rate, Δ lnX, is negative and significant in
both specifications indicating that a depreciating U.S. dollar is associated with a rise in the domestic
producer price index. The coefficient on the interaction of the exchange rate with FRACTION, is positive
and significant. Thus, in industries where derivatives usage is widespread, industry prices rise (drop) less
in response to a depreciating (appreciating) dollar. The pass-through elasticity is -0.08 if hedging in the
industry is non-existent. As the extent of hedging increases, the pass-through elasticity approaches zero.
For an industry with the mean level of hedging, depreciation of the U.S. dollar is not associated with a
rise in domestic prices. This result provides support for the hypothesis that currency hedging mitigates the
correlation between product prices and exchange rate related cost shocks.
Other coefficients in the pass-through regression are largely in agreement with theory. Industries
that use more imported inputs experience larger cost shocks when the dollar fluctuates. Consistent with
this, we find that the coefficient on the interaction of FORINP and the exchange rate is negative,
indicating that prices rise more when the dollar depreciates in industries that are more dependent on
imported inputs. A negative coefficient on the interaction of CONC with the exchange rate implies that
20
It is worthwhile to discuss one channel through which interest rate derivatives usage could be linked with foreign
exchange pass-through. Previous research shows that large firms are more likely to use interest rate derivatives. It is
possible that industries that have a high incidence of interest rate derivatives usage are oligopolies with a handful of
large firms. Firms in oligopolist industries have pricing power and are more likely to pass exchange rate shocks
through to product prices relative to more competitive industries (see, for example, Allayannis and Ihrig (2001)).
However, our hypothesis is that FCD usage reduces pass-through. Therefore, if high interest rate derivatives usage
proxies for concentrated industries, then using it as an instrument biases us against finding support for the
hypothesis.
15
pass-through to domestic prices is greater in more concentrated industries.21 The positive coefficient on
the interaction of EXPORTS with the exchange rate implies that pass-through of currency cost shocks to
domestic prices is lower in industries that export more. Since export revenues improve when the dollar
depreciates and offset the increase in import costs, this result suggests that the operational hedge provided
by export revenues reduces the need to pass-through exchange rate related cost shocks to prices. A
negative coefficient on the interaction of IMPORTS with the exchange rate indicates that, as expected,
pass-through to domestic prices is greater in industries that face more import competition. Finally, we
find that an industry’s capital intensity appears not to affect pass-through.
These results show that pass-through is lower in industries in which FCD usage is more prevalent.
An individual firms’ exposure to exchange rate shocks and its need to hedge the shock will depend on
how much of the shock is passed through to selling prices. Therefore, the extent of FCD usage in an
industry will itself be determined by various industry characteristics that affect pass-through. We do not
assert unidirectional causality and nor do we assume that industry hedging is exogenous. Rather, we seek
to establish that derivatives usage and exchange rate pass-through are intuitively linked such that an
individual firm’s exposure and, therefore, its hedging decision, will depend on the hedging choices of its
competitors.
3.3. The Effect of Industry Hedging and Pass-through on Individual Firms’ Exposures
In the previous section we found that industry prices are less sensitive to foreign exchange related
cost shocks in industries where FCD usage is more common. Since prices are less likely to offset foreign
exchange shocks in these industries, FCD non-users (users) are expected to be more (less) exposed to
exchange rate shocks. This section examines whether the exchange rate exposure of FCD non-users
(users) estimated in Section 3.1 is higher (lower) in industries where FCD usage is widespread and passthrough lower.
3.3.1. Methodology
We test the relation between ˆix and fraction of hedgers in a multivariate framework. Since a
more negative ˆix or a more positive ˆix can both reflect higher exposure to exchange rates we take the
absolute value of the exposure coefficient as a dependent variable. The following regression is estimated
using ordinary least squares:
21
When the Herfindahl index is used in place of the four firm concentration ratio, the coefficient is still negative but
not statistically significant.
16
abs ( ˆix )   0  1 Di   2 Di * F j   3 F j   4 Size i   5 LTDratio i   6 QuickRatio i 
 7 ForeignSal esi   8 PayoutRatio  ui
(3)
The dummy variable Di takes a value equal to one if the firm disclosed the use of FCD in 1999 and zero,
otherwise. Fj is our measure of the prevalence of hedging in firm i’s industry and is the same as the
variable FRACTION used in Section 3.2 with one minor change. When calculating the fraction of hedgers
in a firm’s industry, we exclude the firm’s own hedging decision because the dummy variable, D, already
captures the firm’s hedging decision. Since there is almost no variation in F within an industry, we
cannot include industry dummies. However, observations on individual firms in a given industry may not
be independent and equation (3) could suffer from correlated errors. Therefore, we estimate regression (3)
using industry clustered standard errors.22 Section 4 presents an alternative methodology in which,
instead of taking the absolute value of the exposure coefficient, we examine the differential impact of
industry hedging on firms with positive and negative ˆix .
The univariate analysis in Section 3.1 indicated that on average, FCD users have lower absolute
exposure than FCD non-users. Therefore, we expect that coefficient  1 < 0 in equation (3). Since the
exposure of an FCD user (non-user) is expected to decrease (increase) with the fraction of hedged firms in
the industry, we include an interaction of D with F. A wider gap between the exposure of hedged and
unhedged firms implies  2 < 0. That is, in industries where hedging is widespread, unhedged firms are
increasingly more exposed to the foreign exchange shock relative to hedged firms.
To present the results from the point of view of FCD non-users we repeat this estimation with
only an illustrative modification shown in equation (4). We include a dummy that equals 1 if the firm
does not use FCD and interact this non-user dummy variable with the fraction of FCD non-users in the
industry.
abs ( ˆix )   0  1 (1  Di )   2 (1  Di ) * (1  F j )   3 (1  F j )   4 Sizei   5 LTDratio i 
 6QuickRatio i   7 ForeignSalesi   8 PayoutRatio  ui
(4)
Although this equation is econometrically identical to equation (3), it helps us see how the exposure of an
FCD non-user is affected as more competitors also choose to remain unhedged. We expect that, FCD nonusers are, on average, more exposed to currency fluctuations than FCD users (  1 > 0). However, in
industries where FCD usage is less common, unhedged firms have lower exposure and therefore, the gap
22
It is well recognized that in regression models where the dependent variable is an estimate, variation in the
sampling variance causes heteroskedasticity. A common approach to this problem is to use weighted least squares
technique. However, Lewis and Linzer (2005) show that weighted least squares usually leads to inefficient estimates
and underestimated standard errors. They find that in many cases, OLS with heteroskedasticity consistent standard
errors yields better results. The clustered standard errors used here are heteroskedasticity consistent.
17
between the exposure of hedged and unhedged firms is lower (  2 < 0). In equations (3) and (4), the
following control variables are included: Size of the firm measured as log of total assets, LTDratio which
is calculated as long term debt divided by total assets, QuickRatio calculated as current assets minus
inventories divided by current liabilities, ForeignSales calculated as foreign sales divided by total sales,
and PayoutRatio calculated as dividend per share divided by earnings per share. Table VII presents
estimates of equations (3) and (4). Before discussing the results in Table VII, we address an important
concern.
These tests are motivated by the argument that in industries where FCD usage is widespread,
pass-through of currency shocks to product prices is lower, thus causing the exposure of FCD users (nonusers) to be lower (higher). However, we have to be wary of alternative explanations. If the prevalence of
hedging in an industry is correlated with an unobserved risk characteristic of that industry, then the
predicted coefficients may simply indicate that FCD non-users have greater exposure in industries that are
inherently riskier. In this case, it could be argued that our results do not speak to the dampening effect of
currency hedging on exchange rate pass-through
To address this concern, we directly examine the relation between the foreign exchange exposure
of FCD users (non-users) and the degree of exchange rate pass-through in the industry. We first estimate
aggregate foreign currency pass-through to domestic prices over the same estimation period (1996-2000)
for each industry using a time series regression. The pass-through coefficient is  j in the regression
 ln RPPI t   j 0   j  ln EXCH t 1   j1  ln rt  u jt . It captures the aggregate sensitivity of domestic producer
prices in industry j to exchange rate shocks. Since lower values of the exchange rate measure indicate a
weaker dollar, a negative value of the pass-through coefficient means that domestic prices rise (fall) when
cost of imported inputs increases (decreases). Thus, the pass-through coefficient, πj, is more negative in
industries with greater pass-through of foreign exchange cost shocks to prices.
To examine whether the exposure of FCD users (non-users) is higher (lower) when pass-through
is higher, we use the pass-through coefficient, πj, as an explanatory variable in equations (3) and (4)
instead of the level of hedging Fj. The prediction that α1 < 0 in equation (3) and α1 > 0 in equation (4)
remains unchanged. Since the estimated pass-through coefficient is more negative in industries with
greater pass-through, the coefficient α2 in equation (3) should be less than 0 when πj is used instead of F.
That is, when pass-through is lower (i.e. pass-through coefficient πj higher), the exposure of FCD users
declines relative to that of FCD non-users. In equation (4), if (1- πj) is used as an explanatory variable
instead of (1-F), the coefficient α2 < 0. In other words, when pass-through is high (higher values of 1- πj),
the exposure of FCD non-users declines because currency shocks are offset by product price changes.
Thus, the gap between the exposure of FCD users and non-users declines.
18
Table VII presents estimates of equations (3) and (4) with the fraction of hedged firms, F, as the
right-hand side industry variable. Table VIII presents estimates of the same equations with the passthrough coefficient πj serving as the right-hand side industry variable instead of F.
3.3.2. Results
Estimates of equation (3) and (4) are presented in Columns 1 and 2 respectively of Table VII. The
coefficients are as predicted. FCD users, on average, have lower absolute exposure coefficients than FCD
non-users. However, the gap in the level of exposure of FCD users and FCD non-users is significantly
higher in industries where FCD usage is more widespread. The negative coefficient on α2 confirms that
when more firms in an industry use FCDs, the exposure of FCD users drops further relative to that of
FCD non-users. Table VIII shows estimates of equation (3) and (4) if the pass-through coefficient πj is
used instead of F. These coefficients are also consistent with our hypothesis. FCD users have lower
exposure than FCD non-users (α1 < 0). We see that in industries where the pass-through is lower (that is,
pass-through coefficient higher) the gap between the exposure of FCD users and FCD non-users is higher.
That is, the negative sign for α2 confirms that when pass-through is lower, the exposure of FCD users
declines relative to that of FCD non-users.23
Regarding the control variables used in Tables VII and VIII, we see that exposure is lower for
firms that have high foreign sales as a fraction of total sales. In additional analysis not shown here, we
find that the negative coefficient on foreign sales is significant only for firms that rely on imported inputs
(firms belonging to industries with above median values of FORINP). Thus, our results suggest that
foreign sales serve as an operational hedge for firms that use imported inputs. Finally, the role of financial
constraints on foreign exchange exposure is mixed. Firms with high payout ratios have low foreign
exchange exposure coefficients which is consistent with the notion that less financially constrained firms
have lower foreign exchange exposure. However, we find that firms with higher long-term debt ratios
also have lower foreign exchange exposures.
The link between an individual firm’s exposure and competitors’ hedging decisions should be a
feature of imperfectly competitive industries where some pricing power exits. In highly competitive
industries, a firm’s output choice does not affect price, and consequently, the relation between industry
hedging and exchange rate pass-through to output prices is weak. Thus, in competitive industries, a firm’s
23
Since this regression framework examines relative exposure, the higher gap between the exposure of FCD users
and non-users when industry hedging is higher (or pass-through lower) could be driven by lower exposure of FCD
users, higher exposure of FCD non-users or both. In additional tests (shown in Section 4) we examine the relation
between the exposure coefficient and industry hedging for separate samples of FCD users and FCD non-users.
Results indicate that as the level of hedging in an industry increases, the average exposure of the FCD-user sample
decreases and that of FCD-non-users sample increases.
19
exposure is not expected to depend on the hedging choices of competitors. We re-estimate equation (3)
separately for highly competitive and less competitive industries. We classify highly competitive
industries as those belonging to the bottom quintile of the four-firm concentration ratio. Results for highly
competitive (less competitive) industries are reported in Column 1 (Column 2) of Table IX. We find that a
statistically significant relation between an individual firm’s exposure and the fraction of hedgers in the
industry holds only in less competitive industries. In highly competitive industries, an individual firm’s
exposure is not affected by the hedging decisions of competitors. For robustness, we also use low
Herfindahl index and low price-cost margin instead of the four-firm concentration index to identify highly
competitive industries and find similar results.
Our results are consistent with Bodnar, Dumas and Marston (2002) who show that lower passthrough is associated with higher foreign exchange exposure. We make two contributions to this line of
research. First, we show that pass-through in an industry is closely linked to FCD usage in that industry
and second, the exchange rate exposure of FCD users responds very differently to pass-through than the
exposure of FCD non-users.
3.4 FCD Usage and Firm Value
A number of previous studies demonstrate a positive relation between FCD usage and firm
value.24 It is argued that derivatives users are valued higher because they reduce volatility and the
associated financial distress costs, underinvestment costs etc. Jin and Jorion (2006), on the other hand,
find no relation between hedging and firm value in the oil and natural gas industry. They suggest a
possible reason for why foreign currency hedging is valuable but oil and gas hedging is not. It is possible
that investors buy oil and natural gas stock precisely to gain exposure to these commodity prices and thus,
hedging oil and gas exposures is not beneficial for shareholders. In contrast, foreign exchange risk is often
incidental to the firm’s core business, and usually difficult for an investor to assess properly. Therefore, a
corporation can benefit from hedging away foreign exchange risk on behalf of its shareholders.
Even if shareholders want corporations to reduce foreign exchange exposures, our results indicate
that the commonly documented link between FCD usage and firm value needs further investigation. We
have shown that in industries where FCD usage is low, firms that don’t use FCD have naturally low
foreign exchange risk because exchange rate shocks are passed through to prices. In fact, in these
industries, hedged firms may be more exposed to foreign exchange risk. Thus, the usual explanation for
the existence of a currency hedging premium does not hold for industries where FCD usage is rare. In
24
See, for example, Allayannis and Weston (2001), Graham and Rogers (2002), Allayannis, Lel and Miller (2003),
Carter, Rogers and Simkins (2006), Bartram, Brown and Fehle (2004) and Lookman (2004).
20
order to shed more light on this issue, this section reexamines the link between FCD usage and firm value
conditional on level of hedging in the industry.
3.4.1. Methodology
As in previous studies, Tobin’s Q is used as a measure of firm value (see, for example, Allayannis
and Weston (2001)). It is equal to the market value of equity (price times shares outstanding from CRSP)
plus assets minus the book value of equity, all divided by assets. Book value of equity is equal to common
equity plus deferred taxes. As in previous literature, the sample is restricted to firms that face ex-ante
exposure to exchange rates and, thus, the absence of foreign currency derivatives usage can be interpreted
as a decision not to hedge foreign exchange risk rather than a lack of foreign exchange exposure. A firm’s
hedging decision is captured by a derivatives non-user dummy that equals one if the firm does not
disclose the use of foreign exchange swaps, forwards or options and zero otherwise. In this test, all data
are as of 1999.
When estimating the effect of the hedging decision on firm value, we control for factors that are
known to affect firm value, namely, growth opportunities, size, leverage, profitability and industrial
diversification. Research and development expense over sales and capital expenditures over sales are used
as proxies for growth opportunities. Log of total assets serves as the measure of firm size. Leverage is
calculated as total long-term debt divided by total assets. Return on assets serves as a proxy for
profitability and is calculated as net income over total assets. Industrial diversification is captured with a
dummy that equals one if a firm operates in more than one segment and zero otherwise. Following the
findings of Lookman (2004), we use managerial stock-ownership and institutional ownership as controls
for potential agency conflicts between managers and shareholders. To reduce the influence of outliers, Q,
long-term debt ratio, research and development expense, and return on assets are winsorized at the 1
percent level.
We examine the effect of currency derivative hedging on firm value by modeling firm value as
Vi   0  1 X i   2 (1  Di )  ei
(5)
where Vi is value, X i is a set of exogenous observable characteristics of the firm, (1-Di) is a dummy
variable that takes the value of 1 if the firm is a currency derivatives non-user and 0 otherwise, and ei is
the error term. A firm’s decision to engage in risk management may be correlated with some unobserved
variables that also affect firm value. Thus, (1-Di) may be correlated with the error term in equation (4.5),
rendering OLS estimates of δ2 biased. To control for potential self-selection of firms that hedge, we use
Heckman’s (1979) two-stage procedure in which the hedging decision is modeled as a function of firmspecific variables that have been shown to affect a firm’s incentives to hedge exchange rate risk,
21
specifically, foreign sales, size, leverage, research and development expense, and institutional ownership.
We also use a two-stage least squares approach which we will refer to as the instrumental variable (IV)
regression. In the first stage of the IV approach, a firm’s hedging decision is predicted using the same
instruments as in the Heckman method. For both methods, we require an instrument that is correlated
with the hedging dummy but uncorrelated with firm value. Unlike the method in Section 3.2, we cannot
use interest rate derivatives usage as an instrument because previous research has shown a link between
firm value and interest rate derivatives usage. One possible instrument is the lagged value of the currency
derivatives non-user dummy. Recall that the derivatives non-user dummy equals one if a firm did not
engage in currency hedging during the year 1999 and zero otherwise. We create another dummy variable,
called lag_hedge that equals 1 if a firm did not engage in foreign currency hedging in the year 1997 and
zero if it did. The derivatives non-user dummy for 1999 is significantly positively correlated with
lag_hedge. Firms that engaged in foreign currency risk-management in the past are much more likely to
do so again in 1999 than firms that did not hedge previously. It is unlikely that the firm’s hedging
decision in 1997 affects Tobin’s Q in 1999 other than through its association with the current hedging
decision. Thus, lag_hedge satisfies the requirements of a good instrument.
Equation (5) is estimated using the methods described above for three sub-groups – firms with the
highest (top 1/3rd) values for industry hedging, firms with middle 1/3rd of values for industry hedging and
firms with bottom 1/3rd of values for industry hedging. For convenience, we call these samples ‘highly
hedged’ industries, ‘moderately hedged’ industries and ‘unhedged’ industries respectively. A firm’s
market value may be positively correlated with the market value of other firms in its industry. Therefore,
splitting the sample using the market value-weighted fraction of hedgers in an industry can lead to a
spurious relationship between a firm’s value and its hedging decision. To avoid this problem, we split the
sample using an unweighted measure of the fraction of hedged firms in an industry. Results are discussed
in the next section.
3.4.2. Results
Table X presents estimates of equation (5) for all firms that meet the data requirements as well as
for the three sub-samples based on industry hedging. Column I replicates the results of previous studies
by estimating equation (5) for the entire sample. The negative and significant coefficient on the FCD nonuser dummy confirms the common finding that firms that do not use derivatives suffer a value discount
relative to derivatives-users. In Columns, II-IV, we repeat the regression for firms belonging to the three
groups of industries using the Heckman procedure. We see that the derivatives non-user dummy is
significant only in the sub-sample where industry hedging is high. FCD non-users suffer a value discount
only if they belong to industries where FCD usage is widespread. The IV estimation presented in columns
22
V-VII confirms this finding. In industries where FCD usage is low and unhedged firms enjoy a natural
hedge against exchange rate shocks, unhedged firms do not suffer a value discount relative to hedged
firms. This outcome is quite consistent with the pass-through and exposure results of Sections 3.2 and 3.3.
Unhedged firms in highly hedged industries face greater foreign exchange exposure and are also valued
lower. Unhedged firms in relatively unhedged industries face low foreign exchange exposure and do not
experience a value discount.
While this finding is consistent with the overall theme of the paper, it does not resolve the
fundamental question about why a value difference persists at all. It appears irrational for some firms to
remain unhedged even though the choice to not hedge when most competitors are hedging results in lower
firm value. If a simple act of hedging can improve firm value significantly, then why do these firms not
use derivatives? One possible explanation is that they are unable to use derivatives even if they want to.
These could potentially be firms that cannot find counterparties to derivatives transactions due to low
credit quality, poor performance etc. Firms that don’t hedge when most competitors do hedge may have
fundamentally different, but unobservable, characteristics that are associated with lower firm value. Such
an explanation challenges the causality between derivatives usage and firm value and suggests instead
that the decision not to use derivatives is related to other factors that result in low firm value.
To investigate this idea further, we compare the past five-year (1994-1998) operating and stock
performance of FCD non-users in ‘highly hedged’ industries (which appear to suffer a value discount)
with the past performance of (i) FCD non-users in ‘moderately hedged’ and ‘unhedged’ industries and (ii)
FCD users. Table XI presents comparisons of return on assets, gross profit margins and stock returns of
these groups of firms. In Panel A of Table XI, we find significant differences in the performance of the
two groups of FCD non-users. FCD non-users in ‘highly hedged’ industries have significantly lower
industry adjusted profit margins, lower return on assets and lower stock returns during the past five years
relative to FCD non-users belonging to the remaining industries. In Panel B of the same table, we see that
FCD non-users belonging to ‘highly hedged’ industries also significantly underperform FCD users in the
previous five years. They have significantly lower return on assets and profit margins than hedged firms.
Thus, the group of firms that experiences a value discount in 1999 happens to have underperformed the
rest of the sample in the previous five years. This persistent underperformance may be an explanation for
the lower value of FCD non-users in ‘highly hedged’ industries as well as for their decision to not use
derivatives. To test this, we repeat the Heckman estimation of equation (5) for firms belonging to ‘highly
hedged’ industries, this time including the past averages of profit margins, return on assets and stock
returns in the self-selection equation. The two-stage least squares method is also repeated with the past
performance averages included in the first stage. Table XII presents estimates of these regressions. We
23
see that once these factors are allowed for in the decision to hedge, there is no observable relation
between hedging and firm value, even in industries where hedging is widespread.
These results suggest that foreign currency derivatives usage does not cause an observable
improvement in firm value. Rather, the failure to hedge risks that most competitors consider prudent to
hedge is symptomatic of the same firm characteristics that caused poorer profits and stock returns in the
recent past. Although foreign currency risk is often an incidental and not a core business risk, our results
demonstrate that once competitors hedging choices are taken into account, firms appear to make optimal
decisions about foreign currency risk-management.
4. Robustness
This section demonstrates that our results are robust to alternative measures of industry level
hedging. We also revisit the pass-through relation of Section 3.2 and the exposure results of Section 3.3 in
order to demonstrate robustness to alternative methodologies.
4.1 Alternative measures for industry level hedging
In the results presented above, hedging at the industry level is measured as the market value of
FCD users that face ex-ante foreign exchange exposure divided by market value of all firms that face exante foreign exchange exposure. Firms are defined as having ex-ante currency exposure if they disclose
foreign assets, sales or income in the COMPUSTAT Geographic segment file, or disclose non-zero values
of foreign currency adjustment, exchange rate effect, foreign income, or deferred foreign taxes in the
annual COMPUSTAT files. Since ex-ante foreign exposure is hard to measure, this method may lead us
to ignore firms that do face foreign exchange exposure but choose not hedge. To prevent our results from
being biased by the measure of ex-ante exposure, we calculate industry-level hedging without restricting
the sample to exposed firms only. Our alternative measure of industry hedging is the market value of all
FCD users divided by market value of all firms in the industry (see Panel B of Table I for summary
statistics of this measure). Table XIII shows that the pass-through results of Section 3.2 are robust to this
alternative measure of the level of hedging in an industry. The negative coefficient on the exchange rate
indicates that the domestic retail producer price index rises, albeit insignificantly, when the U.S. dollar
depreciates. However, pass-through is stronger in industries that use more imported inputs, in more
concentrated industries and also in industries that face greater import competition. The pass-through of
exchange rate shocks to prices is weaker in industries that use more FCD and in industries that export
more.
Table XIV revisits the relation between an individual firm’s exposure and the fraction of hedged
firms in the industry, as shown in equation (3) of Section 3.3. However, this time the fraction of hedgers
is calculated using all firms regardless of ex-ante exposure. Table XIV shows that the foreign exchange
24
exposure of FCD users is lower than that of FCD non-users. Moreover, as the level of hedging in the
industry rises, the exposure of FCD users drops further relative to that of FCD non-users. Thus, the results
presented in Section 3.3 are robust to a value-weighted measure of industry hedging which is based on all
firms and not only on firms with ex-ante exchange rate exposure
In tests not presented here, we also use an unweighted measure of the fraction of hedgers in an
industry. The pass-through results are robust to the unweighted measure – a higher fraction of hedgers in
an industry is still associated with lower pass-through. However, the association between industry
hedging and individual firms’ currency exposure becomes weaker (statistically insignificant).
4.2 Exchange rate exposure of FCD-users and non-users
The estimated exposure coefficient, ˆix , can be either positive or negative. A more negative or a more
positive ˆix both reflect higher exchange rate exposure. Since values closer to zero indicate lower
exchange rate exposure, in Section 3.3, we used the absolute value of the exposure coefficient as the
dependent variable. We now examine the differential effect of industry hedging on firms with positive
and negative exposure coefficients using a methodology based on He and Ng (1998). The following
regression is estimated using ordinary least squares with industry clustered standard errors:
ˆix   0  1 (1  P)   2 P * F j   3 (1  P) * F j   4 Sizei   5 LTDratio i   6QuickRatioi
  7 ForeignSalesi  8 PayoutRatio  ui
(6)
where P is a dummy variable equal to one if ˆix is positive and zero otherwise. All other variables are as
described in Section 3.3. The regression is estimated separately for FCD users and FCD non-users.
Results are presented in Table XV.
Since the dummy P is equal to 1 for firms with positive ˆix , the constant is positive by design
and the coefficient on the dummy (1-P) negative. In Panel A, which presents the estimates of equation (6)
for FCD non-users, we observe that the coefficient on the interaction of P with the level of hedging in the
industry, F, is positive and significant. This indicates that FCD non-users with positive exposure to
exchange rates have even higher exposure when more competitors are hedged. The coefficient on the
interaction of (1-P) with the level of hedging in the industry is negative but insignificant. The negative
sign of this interaction term suggests that hedging by competitors also exacerbates the exposure for firms
with negative ˆix but this effect is not statistically significant. Thus, a rise in the fraction of hedged
competitors increases foreign exchange exposure of unhedged firms on average, and this result appears to
be driven by firms whose returns improve when the dollar appreciates (i.e., firms with positive ˆix ).
Firms with positive ˆix are more likely to be importing firms whose costs rise as the dollar depreciates.
Since the theoretical arguments behind these tests apply to firms that face cost shocks, it is not altogether
25
surprising that the results are driven by firms whose cost of importing changes as the dollar value
changes.
Panel B provides estimates of equation (6) for FCD users. We see that the coefficient on the
interaction of (1-P) with the level of hedging in the industry, F, is positive and significant. This means
that as the level of hedging in the industry increases, negative exposure coefficients of FCD users move
significantly closer to zero. Similarly, the negative sign on the interaction of P and F indicates that as the
level of hedging in the industry increases, positive exposure coefficients of FCD users move
insignificantly closer to zero. These results confirm the finding in Section 5 that the foreign exchange
exposure of an FCD user is lower in industries where many other firms also use FCD.
4.3 Pass-through analysis with clustered standard errors
In Section 3.2, we examined the relationship between industry hedging and pass-through, by
estimating a panel model in which errors are possibly correlated over time and across observations. We
dealt with the industry-fixed effect parametrically by including industry dummies and with the time effect
by using Newey-West standard errors. The regression was estimated using two-stage least squares.
Petersen (2005) suggests that in the presence of industry-fixed effects, Newey-West standard errors may
be biased. Another concern of the methodology in Section 3.2 is that, given the low variation in the
hedging measure over time, the presence of industry-fixed effects makes it difficult to capture the relation
between hedging and industry prices (see Zhou (2001)). Both these issues can be addressed by using
standard errors clustered across a cross-sectional dimension. We re-estimate the pass-through equation (2)
with standard errors clustered at the 2-digit SIC level using generalized method of moments (GMM).
Since producer price data are limited to the manufacturing industries, the number of 2-digit SIC clusters is
insufficient to include all time dummies. Therefore, we estimate the equation using year dummies only.
Table XVI presents results of equation (2) estimated using GMM with clustered standard errors. As
before, we use the extent of interest rate derivatives usage as an instrument for FCD usage in an industry.
We see that domestic producer prices rise when the dollar depreciates in industries that use more imported
inputs and face greater import competition. However, this pass-through is lower in industries where FCD
usage is more prevalent and in industries that export more. Thus, the relation between hedging and
industry pass-through is robust to this alternative methodology.
5. Conclusion
The evidence in this paper sheds new light on the corporate risk management literature. We show
that when more firms in an industry use FCD, industry output prices become less sensitive to exchange
rate shocks. Specifically, as the dollar depreciates and import costs rise, U.S. firms charge higher prices in
the domestic market. However, as FCD usage in an industry rises, the correlation between domestic prices
26
and exchange rate shocks drops. For an industry with an average level of FCD usage, depreciation of the
U.S. dollar is not associated with an increase in industry selling prices. To our knowledge, this is the first
paper to provide evidence that financial hedging influences output markets.
This finding is important for the corporate risk management literature because the link between
FCD usage and output markets makes the foreign exchange exposures of firms in an industry
interdependent. We find that as the level of hedging in an industry increases (and pass-through decreases),
the foreign exchange exposure of an FCD user declines while that of an FCD non-user increases. A firm
faces lower exposure to the underlying exchange rate risk when its hedging decision is similar to
competitors’ hedging decisions. Thus, an FCD user can face more exposure to the underlying risk factor
than a non-user, even if the FCD user is not trading speculatively. Not surprisingly, empirical tests of risk
management theory that consider FCD usage tantamount to risk-reduction arrive at mixed or inconclusive
evidence.
We also find that once the hedging choices of other firms in the industry are taken into account,
there is no observable difference between the value of an FCD user and FCD non-user. This suggests that,
when studied in the context of their industries, firms make optimal choices about foreign currency risk
management. Our finding that a firm’s exposure to a risk factor is affected by the hedging choices of
competing firms implies that risk-management decisions should be determined within an industry
equilibrium. The theoretical work of Adam, Dasgupta and Titman (2006) takes a step in this direction by
examining how a firm’s hedging choice depends on the hedging choices of competitors.
Existing theories that examine a firm’s decision to hedge in isolation from its industry enjoy little
empirical support. Guay and Kothari (2003) contend that empirical support for risk-management theory is
weak because the usual empirical proxy for risk-management - derivatives usage - constitutes too small a
part of a firm’s hedging program. However, we demonstrate that the extent of derivatives hedging in an
industry significantly affects product prices as well as an individual firm’s exposure to exchange rates.
Our results suggest that previous empirical research is inconclusive not because derivatives hedging is
unimportant for a firm’s risk-profile, but because the role derivatives usage plays in the firms’ product
markets is ignored.
27
References
Adam, Tim, and Chitru Fernando, 2006, Hedging speculation and shareholder value, Journal of Financial
Economics 81, 283-309.
Adam, Tim, Sudipto Dasgupta, and Sheridan Titman, 2006, Financial constraints, competition, and
hedging in industry equilibrium, Journal of Finance, forthcoming.
Allayannis, George, and Jane Ihrig, 2001, Exposure and Markups, Review of Financial Studies 14, 805835.
Allayannis, George, Jane Ihrig, and James Weston, 2001, Exchange-rate hedging: Financial versus
operational strategies, American Economic Review 91, 391-395.
Allayannis, George, and Eli Ofek, 2001, Exchange rate exposure, hedging, and the use of foreign
currency derivatives, Journal of International Money and Finance 20, 273-296.
Allayannis, George, and James P. Weston, 2001, The use of foreign currency derivatives and firm market
value, Review of Financial Studies 14, 243-276.
Allayannis, George, Ugur Lel, and Darius P. Miller, 2003, Corporate governance and the hedging
premium around the world, unpublished manuscript.
Bartram, Söhnke, Gregory Brown, and Frank Fehle, 2004, International evidence on financial derivatives
usage, unpublished manuscript.
Bartov, Eli, and Gordon Bodnar, 1994, Firm valuation, earnings expectations, and the exchange rate
exposure effect, Journal of Finance 49, 1755-1786.
Breeden, D. and S. Viswanathan, 1998, Why do firms hedge? An asymmetric information model,
unpublished manuscript.
Brown, Gregory, 2001, Managing foreign exchange risk with derivatives, Journal of Financial
Economics 60, 401-448.
Bodnar, Gordon, Bernard Dumas, and Richard Marston, 2002, Pass-through and exposure, Journal of
Finance 57, 199-231.
Bodnar Gordon, and William Gentry, 1993, Exchange rate exposure and industry characteristics:
Evidence from Canada, Japan, and the U.S.A, Journal of International Money and Finance 12, 2945.
Bodnar Gordon, Gregory S. Hayt, and Richard C. Marston, 1998, Wharton 1998 survey of risk
management by U.S. non-financial firms, Financial Management 27, 70-91.
Carter, David, Daniel Rogers, and Betty Simkins, 2006, Does hedging affect firm value? Evidence from
the US airline industry, Financial Management 35, 53-86.
DeMarzo, Peter, and Darrell Duffie, 1991, Corporate financial hedging with proprietary information,
Journal of Economics Theory 53, 261-286.
28
DeMarzo, Peter, and Darrell Duffie, 1995, Corporate incentives for hedging and hedge accounting,
Review of Financial Studies 8, 743-771.
De Meza, David, 1986, Safety in conformity but profits in deviance, Canadian Journal of Economics 19,
261-269.
Dolde, Walter, 1995, Hedging, leverage, and primitive risk, Journal of Financial Engineering 4, 187-216.
Dominguez, Kathryn, and Linda Tesar, 2001, A reexamination of exchange rate exposure, American
Economic Review 91, Papers and Proceedings, 396-399.
Faulkender, Michael, 2004, Hedging or market timing? Selecting the interest rate exposure of corporate
debt, Journal of Finance 60, 931-962.
Feenstra, Robert, 1987, Symmetric pass-through of tariffs and exchange rates under imperfect
competition: An empirical test, Journal of International Economics 27, 25-45.
Feinberg, Robert, 1986, The interaction of market power and exchange rate effects on German domestic
prices, Journal of Industrial Economics 35, 61-70.
Feinberg, Robert, 1989, The effects of foreign exchange movements on U.S. domestic prices, Review of
Economics and Statistics 71, 505-511.
Froot, Kenneth, David Scharfstein, and Jeremy Stein, 1993, Risk management: Coordinating corporate
investment and financing policies, Journal of Finance 48, 1629-1657.
Froot, Kenneth, and Paul Klemperer, 1989, Exchange rate pass-through when market share matters,
American Economic Review 79, 637-654.
Geczy, Christopher, Bernadette Minton, and Catherine Schrand, 1997, Why firms use currency
derivatives, Journal of Finance 52, 1323-1354.
Graham, John, and Daniel Rogers, 2002, Do firms hedge in response to tax incentives, Journal of Finance
57, 815-838.
Griffin, John, and Rene Stulz, 2001, International competition and exchange rate shocks: A cross-industry
analysis of stock returns, Review of Financial Studies 14, 215-241.
Guay, Wayne, and S. P. Kothari, 2003, How much do firms hedge with derivatives, Journal of Financial
Economics 70, 423-461.
Guay, Wayne, 1999, The impact of derivatives on firm risk: An empirical examination of new derivatives
users, Journal of Accounting and Economics 26, 319-351.
Haushalter, David, 2000, Financing policy, basis risk, and corporate hedging, evidence from oil and gas
producers, Journal of Finance 55, 107-152.
He, Jia, and Lilian Ng, 1998, The foreign exchange exposure of Japanese multinational corporations,
Journal of Finance 53, 733-753.
29
Heckman, James, 1979, Sample selection bias as specification error, Econometrica 47, 153-161.
Hentschel Ludger, and S. P. Kothari, 2001, Are corporations reducing or taking risks with derivatives,
Journal of Financial and Quantitative Analysis 36, 93-118.
Jin, Yanbo, and P. Jorion, 2006, Firm value and hedging: Evidence from U.S. oil and gas producers,
Journal of Finance 61, 893-919.
Jorion, Phillippe, 1990, The exchange-rate exposure of U.S. multinationals, Journal of Business 63, 331345.
Jorion, Phillippe, 1991, The pricing of exchange rate risk in the stock market, Journal of Financial and
Quantitative Analysis 26, 363-376.
Knetter, Michael, 1989, Price discrimination by U.S. and German exporters, American Economic Review
79, 198-210.
Knetter, Michael, 1993, International comparisons of pricing-to-market behavior, American Economic
Review 83, 473-486.
Knetter, Michael, 1994, Did the strong dollar increase competition in the U.S. product markets, Review of
Economics and Statistics 76, 192-195.
Knopf, John, Jouahn Nam, and John Thornton, Jr., 2002, The volatility and price sensitivities of
managerial stock option portfolios and corporate hedging, Journal of Finance 57, 801-813.
Krugman, Paul, 1987, Pricing to market when the exchange rate changes, in Sven W. Ardnt and J. David
Richardson eds., Real Financial Linkages Among Open Economies (Cambridge, MA: MIT Press
1987)
Leland, Hayne, 1998, Agency costs, risk management, and capital structure, Journal of Finance 53, 12131243.
Lewis Jeffery, and Drew Linzer, 2005, Estimating regression models in which the dependent variable is
based on estimates, Political Analysis, 345-364.
Lookman, Aziz, 2004, Does hedging increase firm value? Comparing premia for hedging ‘big’ versus
‘small’ risks, unpublished manuscript.
Maksimovic, Vojislav, and Josef Zechner, 1991, Debt, agency costs and industry equilibrium, Journal of
Finance 46, 1619-1645.
Mann, Catherine, 1986, Prices, profits, margins, and exchange rates, Federal Reserve Bulletin 72, 366379.
Marston, Richard, 1990, Pricing to market in Japanese manufacturing, Journal of International
Economics 29, 217-236.
Mello, Antonio, and Martin Ruckes, 2006, Financial hedging and product market rivalry, unpublished
manuscript.
30
Mian, Shehzad, 1996, Evidence on corporate hedging policy, 1996, Journal of Financial and Quantitative
Analysis 31, 419-439.
Morellec, Erwan., and Clifford. W. Smith, 2003, Investment policy, financial policies, and the control of
agency conflicts, unpublished manuscript.
Nance, Deana, Clifford Smith, Jr., and Charles Smithson, 1993, On the determinants of corporate
hedging, Journal of Finance 48, 267-284.
Ohno, Kenichi, 1989, Export pricing behavior of manufacturing: A U.S. – Japan comparison,
International Monetary Fund Staff Papers 36, 550-579.
Petersen, Mitchell, 2005, Estimating standard errors in finance panel data sets: Comparing approaches,
unpublished manuscript.
Purnanandam, Amiyatosh, 2007, Financial distress and corporate risk management: theory and evidence,
unpublished manuscript.
Schrand, Catherine, and Haluk Unal, 1998, Hedging and coordinated risk management: evidence from
thrift conversions, Journal of Finance 53, 979-1015.
Smith, Clifford, and René Stulz, 1985, The determinants of firm’s hedging policies, Journal of Financial
and Quantitative Analysis 48, 391-405.
Stulz, René, 1984, Optimal hedging policies, Journal of Financial and Quantitative Analysis 19, 127-140.
Tufano, Peter, 1996, Who managers risk? An empirical examination of risk management practices in the
gold mining industry, Journal of Finance 51, 1097-1137.
Yang, Jiawen, 1997, Exchange rate pass-through in U.S. manufacturing industries, Review of Economics
and Statistics 79, 95-104.
Zhou, X., 2001, Understanding the determination of managerial ownership and its relationship to firm
performance: comment, Journal of Financial Economics 62, 559-571.
31
TABLE I
Descriptive Statistics of Foreign Currency Derivatives Use
Panel A summarizes foreign currency derivatives (FCD) usage as of fiscal year 1999 by 549 U.S. firms that face exante exchange rate exposure. A firm is defined as having exchange rate exposure if it discloses foreign assets, sales
or income in the COMPUSTAT Geographic segment file, or discloses positive values of foreign currency
adjustment, exchange rate effect, foreign income, or deferred foreign taxes in the annual COMPUSTAT files. A firm
is classified as an FCD user if it discloses the use of foreign currency forwards, swaps or options in its 10-K
disclosures. Panel A gives mean, 25th percentile, median and 75th percentile of total FCD usage as well as a break up
by type of derivative (swaps, forwards and options). The table provides total notional amounts as well as notional
amounts scaled by book value of total assets. All values are in dollar millions.
Panel B provides the distribution of FCD usage at the 3-digit SIC industry level. Two measures of FCD usage are
presented. The first column gives the distribution of the market value-weighted fraction of hedgers based on the subsample of firms facing ex-ante exchange rate exposure. It is calculated as the sum of market values of FCD users in
the industry who face ex-ante currency exposure divided by sum of market values of all firms in the industry who
face ex-ante currency exposure. Market value of a firm is the market value of equity plus the book value of debt. N
is the number of observations. The second column of Panel B gives the distribution of the market value-weighted
fraction of hedgers based on the entire sample regardless of ex-ante foreign exchange exposure. It is calculated as
the sum of market values of FCD users in the industry divided by sum of market values of all firms in the industry.
PANEL A : Foreign Currency Derivative Usage (FCD) in $ millions
N
549
Mean
745.69
8.89%
25th
percentile
7.60
1.30%
Median
42.90
3.76%
75th
percentile
265.84
9.74%
Std. Dev
3427.99
23.86%
Foreign Currency Swaps
Scaled by Total Assets
74
850.48
4.90%
35.00
1.39%
158.12
3.08%
603.00
6.18%
2917.45
5.26%
Foreign Currency Forwards
Scaled by Total Assets
502
606.11
7.84%
7.00
1.12%
35.78
3.14%
210.00
7.68%
2811.58
24.13%
Foreign Currency Options
Scaled by Total Assets
91
463.52
6.35%
17.70
0.74%
79.4
2.26%
403.00
7.06%
908.77
11.75%
Total FCD
Scaled by Total Assets
PANEL B : Distribution of FCD Usage by 3-digit SIC Industry Group
Percentiles
Fraction of Hedgers
Fraction of Hedgers
(Based on Exposed Firms
(Based on All Firms)
Only)
10%
0.00
0.00
25%
0.00
0.00
50%
2.50%
0.30%
60%
19.60%
9.70%
75%
48.20%
37.30%
90%
75.80%
65.40%
95%
94.30%
81.85%
Industries
Mean
Std. Dev
256
0.25
0.32
275
0.20
0.28
32
TABLE II
Fraction of Firms Using Foreign Currency Derivatives (FCD) by Fama-French Industry Groups
This table reports two measures of the extent of hedging for 43 out of 48 Fama-French industry groups as of fiscal year 1999. The first measure is based only on
firms that face ex-ante foreign currency exposure. It is calculated as the sum of market values of FCD users in the industry who face ex-ante currency exposure
divided by sum of market values of all firms in the industry who face ex-ante currency exposure. Market value of a firm is the market value of equity plus the book
value of debt. Firms are defined as having ex-ante exposure if they disclose foreign assets, sales or income in the COMPUSTAT Geographic segment file, or
disclose non-zero values of foreign currency adjustment, exchange rate effect, foreign income, or deferred foreign taxes in the annual COMPUSTAT files. The
second measure is based on the entire sample regardless of foreign exchange exposure. It is calculated as the sum of market values of FCD users in the industry
divided by sum of market values of all firms in the industry.
Fama
French
Industry #
1
Agriculture
N
28
Fraction of Hedgers
(Based on Exposed Firms Only)
0.34
Fraction of Hedgers
(Based on All Firms)
0.19
2
Food Products
136
0.58
0.45
3
Candy & Soda
27
0.45
0.43
4
Beer & Liquor
33
0.80
0.78
5
Tobacco Products
13
0.00
0.00
6
Recreation
88
0.94
0.94
7
Entertainment
201
0.00
0.00
8
Printing and Publishing
76
0.03
0.02
9
Consumer Goods
152
0.27
0.27
10
Apparel
108
0.22
0.20
11
Healthcare
143
0.20
0.09
12
Medical Equipment
275
0.66
0.62
13
Pharmaceutical Products
469
0.57
0.54
14
Chemicals
149
0.34
0.34
15
Rubber and Plastic Products
82
0.47
0.44
16
Textiles
42
0.28
0.22
17
Construction Materials
142
0.30
0.28
18
Construction
110
0.52
0.35
19
Steel Works
122
0.40
0.36
20
Fabricated Products
36
0.24
0.19
21
Machinery
266
0.55
0.55
22
Electrical Equipment
84
0.33
0.33
23
Automobiles and Trucks
163
0.86
0.86
24
Aircraft
118
0.49
0.49
25
Shipbuilding, Railroad Equipment
30
0.86
0.85
26
Defense
15
0.05
0.04
27
Precious Metals
13
0.00
0.00
28
Non-Metallic and Industrial Metal Mining
49
0.55
0.48
29
Coal
42
0.29
0.26
30
Petroleum and Natural Gas
9
0.00
0.00
31
Utilities
340
0.35
0.35
32
Communication
396
0.27
0.16
33
Personal Services
446
0.31
0.24
34
Business Services
94
0.23
0.17
35
Computers
1500
0.49
0.41
36
Electronic Equipment
445
0.75
0.71
37
Measuring and Control Equipment
522
0.31
0.31
38
Business Supplies
195
0.41
0.39
39
Shipping Containers
107
0.56
0.53
40
Transportation
27
0.31
0.29
41
Wholesale
212
0.26
0.20
42
Retail
347
0.41
0.32
43
Restaurants, Hotels, Motels
428
0.58
0.35
Industry Name
33
Table III
Summary of Foreign Exchange Exposure Coefficient
This table provides the distribution of foreign exchange exposure coefficient, ˆ ix , estimated in the regression rit   i 0   ix EXCH t   im rmt   it , where rt is
the monthly rate of return on a firm’s stock for the years 1996 till 2000, rmt is the corresponding monthly rate of return on the value-weighted market index,
EXCH t is the monthly change in value of the U.S. dollar orthogonal to the market return. Panels A shows summary statistics of ̂ ix for the entire sample, as well
as for sub-samples of FCD Users and FCD Non-Users. FCD users are firms that disclosed the use of foreign currency forwards, swaps or options at the end of
fiscal year 1999. All remaining firms are classified as FCD non-users. Panel B (Panel C) shows summary statistics for firms with positive (negative) exposure
coefficients only. t-statistics are in parenthesis. For samples with an even number of firms, the median is an average of two coefficients and, therefore, no t-statistic
is reported. Significance is indicated by bold font with superscripts a, b, and c denoting significance at the 1%, 5% and 10% levels respectively.
Panel A : All Firms
Minimum
All Firms
FCD Non-Users
FCD Users
-1.03a
-1.03a
-0.06a
(-10.80)
(-10.80)
(-3.36)
Median
0.003
0.004
0.001
(0.16)
(0.10)
Maximum
1.22c
1.22c
0.06a
(1.77)
(1.77)
(2.75)
Mean “Absolute” Exposure Coefficient
0.02
0.01
Difference in Mean “Absolute” Exposure Coefficients
a
0.01
(3.59)
Number of firms with significant exposure at least at the 10% level
344
292
52
Total Number of Firms
3036
2635
401
Panel B : Firms with positive exposure only
Minimum
Median
All Firms
FCD Non-Users
FCD Users
0.00
0.00
0.00
(0.00)
(0.00)
(0.00)
0.012
0.013
0.008
(0.60)
Maximum
(0.80)
1.22c
1.22c
0.06a
(1.78)
(1.78)
(2.75)
Mean Exposure
0.020
0.011
0.018a
Difference in Mean Exposures
(2.94)
Number of firms with significant exposure at least at the 10% level
216
190
26
Total Number of Firms
1777
1558
219
FCD Non-Users
FCD Users
Panel C: Firms with negative exposure only
Minimum
Median
Maximum
Mean Exposure
All Firms
a
a
-1.03
-1.03
-0.06a
(-10.80)
(-10.80)
(-3.36)
-0.009
-0.009
-0.007
(-0.60)
(0.52)
-0.00
-0.00
-0.00
(0.00)
(0.00)
(0.00)
-0.014
-0.016
-0.010
-0.005b
Difference in Mean Exposures
(1.96)
Number of firms with significant exposure at least at the 10% level
128
102
26
Total Number of Firms
1259
1077
182
34
Figure I
Exposure Foreign Exchange Exposure Coefficients of FCD Users and FCD Non-Users
This figure plots the foreign exchange exposure coefficients, ˆ ix , estimated in the regression rit  i0  ixEXCH t  imrmt  it , where rt is the monthly rate of
return on a firm’s stock for the years 1996 till 2000, rmt is the corresponding monthly rate of return on the value-weighted market index, EXCH t is the monthly
change in value of the U.S. dollar orthogonal to the market return. Figure I-A plots ̂ ix for FCD Users and Figure I-B plots ̂ ix for FCD Non-Users. FCD users are
firms that disclosed the use of foreign currency swaps, options or forwards at the end of fiscal year 1999. All remaining firms are classified as FCD non-users.
0
10
20
Density
30
40
50
Figure I-A
-.1
0
Exposure Coefficients of FCD Users
.1
0
10
20
Density
30
40
50
Figure I-B
-.1
0
Exposure Coefficients of FCD Non-Users
.1
35
Table IV
Firms’ Exposure Coefficients and Industry Hedging: A Univariate Comparison
This table compares the prevalence of hedging in the industries of firms with very high or very low exchange rate exposures. A foreign currency derivatives
(FCD) user is defined as having extreme exposure if the exposure coefficient is among 5% of the most positive coefficients or 5% of the most negative
coefficients within the sub-sample of FCD users. An FCD user is defined as having very low exchange rate exposure if the coefficient is among 10% of exposure
coefficients closest to zero within the sub-sample of FCD users. A similar classification of very high and very low exposure is applied for FCD non-users. The
prevalence of hedging in an industry is captured by the market value-weighted fraction of hedgers calculated as the sum of market values of FCD users in the
industry who face ex-ante currency exposure divided by sum of market values of all firms in the industry who face ex-ante currency exposure. Market value of a
firm is the market value of equity plus the book value of debt. Firms are defined as having ex-ante exposure if they disclose foreign assets, sales or income in the
COMPUSTAT Geographic segment file, or disclose non-zero values of foreign currency adjustment, exchange rate effect, foreign income, or deferred foreign
taxes in the annual COMPUSTAT files. Significance is indicated by bold font with superscripts a, b, and c denoting significance at the 1%, 5% and 10% levels
respectively.
FCD users with extreme positive or extreme negative exposure coefficients
FCD users with exposure coefficients close to zero
40
39
Mean Fraction
of Hedgers in
Industry
0.37
0.48
FCD non-users with extreme positive or extreme negative exposure coefficients
FCD non-users with exposure coefficients close to zero
261
264
0.39
0.34
N
Difference
t-stat
-0.11a
(2.64)
0.05a
(2.51)
36
Table V
Correlation Coefficients of Industry Level Variables
This table presents correlation coefficients for seven industry level variables used in the paper. FRACTION is calculated as the sum of market values of FCD
users in the industry who face ex-ante currency exposure divided by sum of market values of all firms in the industry who face ex-ante currency exposure. Δln
RPPI is the log change in relative producer price index, RPPI, at the three-digit SIC level, calculated as the producer price index divided by the overall GDP
price deflator. The producer price data are obtained from the Bureau of Labor Statistics. The overall GDP price deflator is obtained from the Bureau of Economic
Analysis. An industry’s reliance on foreign inputs FORINP, is calculated as in Allayannis and Ihrig (2001) using monthly industry import data provided by
United States International Trade Commission (USITC) and the 1997 benchmark input-output tables provided by the Bureau of Economic Analysis of the U.S.
Department of Commerce. Import penetration, IMPORTS, for an industry is calculated as the monthly general customs value of final goods imports scaled by
domestic shipments for that industry in that year. Industry exports, EXPORTS, are obtained from USITC and scaled by domestic shipments. Capital intensity, KS,
calculated as industry total assets as a fraction of industry sales. The four-firm concentration ratio, CONC, is obtained from the 1997 U.S. Census of
Manufacturers. t-statistics are provided in parenthesis. Significance is indicated by bold font with superscripts a, b, and c denoting significance at the 1%, 5% and
10% levels respectively.
FRACTION
FRACTION
Δln RPPI
FORINP
IMPORTS
EXPORTS
KS
CONC
1
Δln RPPI
-0.021
(0.13)
1
FORINP
0.160a
(0.00)
0.030
(0.02)
1
IMPORTS
0.192a
(0.00)
-0.017
(0.17)
0.713a
(0.00)
1
EXPORTS
0.117a
(0.00)
-0.018
(0.17)
0.278a
(0.00)
0.524a
(0.00)
1
KS
0.038b
(0.01)
0.008
(0.58)
-0.003
(0.84)
-0.001
(0.94)
0.023
(0.10)
1
CONC
(0.06)a
(0.00)
0.003
(0.80)
-0.037a
(0.00)
-0.012
(0.35)
0.042a
(0.00)
0.01
(0.47)
1
37
Table VI
Foreign Exchange Pass-through to Domestic Prices
This table shows the relation between industry prices and the external value of the U.S. dollar conditional on the extent of currency hedging in an
industry. Estimates of the following regression are presented.
 ln RPPI jt   0  1 ln X t 1   2  ln X t 1 * FRACTION
jt
  3  ln X t 1 * FORINP jt   4  ln X t 1 * IMPORTS
  5  ln X t 1 * KS jt   6  ln X t 1 * CONC j   7  ln X t 1 * EXPORTS
jt
jt
  8  ln rit   jt
Although not shown, all variables that appear in the interaction terms are also included separately as control variables. The dependent variable Δln
RPPI is the log change in the domestic relative producer price index, RPPI. RPPI is calculated as the producer price index divided by the overall
GDP price deflator. Foreign exchange movements, X, are measured as the trade-weighted value of the U.S. dollar in terms of its major trading
partners as calculated by the Federal Reserve Board. The fraction of hedgers in an industry, FRACTION, is calculated as the sum of market values
of FCD users in the industry who face ex-ante currency exposure divided by sum of market values of all firms in the industry who face ex-ante
currency exposure. The fraction of firms in an industry using interest rate derivatives is used as an instrument for FRACTION. An industry’s
reliance on foreign inputs FORINP, is calculated as in Allayannis and Ihrig (2001) using monthly industry import data provided by United States
International Trade Commission (USITC) and the 1997 benchmark input-output tables provided by the Bureau of Economic Analysis of the U.S.
Department of Commerce. Import penetration, IMPORTS, for an industry is calculated as the monthly general customs value of final goods
imports scaled by domestic shipments for that industry in that year. Industry exports, EXPORTS, are obtained from USITC and scaled by domestic
shipments. Capital intensity, KS, is calculated as industry total assets divided by industry sales. The four-firm concentration ratio, CONC, is
obtained from the 1997 U.S. Census of Manufacturers. The U.S. LIBOR, rit is included in log differences. The numbers in parenthesis are tstatistics based on Newey-West standard errors. Significance is indicated by bold font with superscripts a, b, and c denoting significance at the 1%,
5% and 10% levels respectively. t-statistics are in parenthesis.
Dependent Variable : Δln RPPI
Δ ln X
I
II
-0.076c
-0.077c
(1.71)
(1.75)
Δ ln X * Fraction of FCD Users in Industry
0.433b
0.436b
(2.20)
(2.23)
Δ ln X * Imported Inputs
-0.964c
(1.77)
Δ ln X * Imports
-1.457b
(2.35)
Δ ln X * Capital Intensity
Δ ln X * 4-Firm Concentration Ratio
1.154
0.994
(0.43)
(0.37)
-2.253c
-2.18c
(1.86)
(1.82)
b
5.949b
Δ ln X * Exports
4.565
(1.99)
(2.22)
Δ ln r
0.013a
0.013a
(3.24)
(3.23)
0.003
0.003
(0.34)
(0.34)
Fraction of FCD Users in Industry
Imported Inputs
-0.009b
(2.05)
-0.008b
Imports
(2.15)
Capital Intensity
4-Firm Concentration Ratio
Exports
0.009
0.008
(0.21)
(0.19)
-0.003
0.016
(0.05)
(0.27)
0.055
0.062
(1.46)
(1.60)
Industry Dummies
Yes
Yes
Observations
5211
5211
F statistic
3.46a
3.55a
38
Table VII
The Relation between Firms’ Foreign Exchange Exposure and Industry FCD Usage
This table reports estimates of the relation between a firm’s foreign exchange exposure, ˆ ix , and the prevalence of
hedging in a firm’s industry. ˆ ix is estimated in the regression rit   i 0   ix EXCH t   im rmt   it , where rt is the
monthly rate of return on a firm’s stock for the years 1996 till 2000, rmt is the corresponding monthly rate of return on
the value-weighted market index, EXCH t is the monthly change in value of the U.S. dollar orthogonal to the market
return. The first column in Panel A below presents estimates of the equation
abs( ˆix )   0  1 Di   2 Di F j   3 F j   4 Sizei   5 LTDratio i   6QuickRatioi   7 ForeignSalesi   8 PayoutRatio  ui
The dummy variable Di takes a value equal to one if the firm disclosed the use of FCD in 1999 and zero otherwise. Fj is
our measure of the fraction of hedgers in firm i’s industry calculated as the market value of firms in the industry that
face ex-ante exchange rate exposure and use FCD divided by the market value of all firms in the industry who face exante exchange rate exposure. When calculating F, we exclude the firm’s own hedging decision. Size of the firm is
measured as log of total assets, LTDratio is calculated as long term debt divided by total assets, QuickRatio is
calculated as current assets minus inventory divided by current liabilities, ForeignSales is calculated as foreign sales
divided by total sales, and PayoutRatio is calculated as dividend per share divided by earnings per share. The second
column in Panel A presents estimates of the equation
abs( ˆix )   0  1 (1  Di )   2 (1  Di )(1  F j )   3 (1  F j )   4 Sizei   5 LTDratio i   6QuickRatioi   7 ForeignSalesi
  8 PayoutRatio  ui
t-statistics based on clustered (by industry) standard errors are provided in parenthesis. Significance is indicated by
bold font with superscripts a, b, and c denoting significance at the 1%, 5% and 10% levels respectively.
Dependent Variable = abs( ˆ ix )
FCD User Dummy
FCD User Dummy * Fraction of FCD Users in Industry
Fraction of FCD Users in Industry
-0.122
(0.67)
-0.990b
(2.20)
0.816a
(2.72)
-0.001a
(3.68)
-1.117a
(3.01)
0.063
(1.45)
-0.685a
(3.20)
-0.053a
(3.40)
1.111a
(3.57)
-0.990b
(2.20)
0.174
(0.59)
-0.001a
(3.68)
-1.117a
(3.01)
0.063
(1.45)
-0.685a
(3.20)
-0.053a
(3.40)
2826
10.83a
2826
10.83a
FCD Non-User Dummy
FCD Non-User Dummy * Fraction of FCD Non-Users in Industry
Fraction of FCD Non-Users in Industry
Size
LTD Ratio
Quick Ratio
Foreign Sales/Net Sales
Payout Ratio
Observations
F-statistic of overall significance
39
Table VIII
The Relation between Firms’ Foreign Exchange Exposure and Pass-through
This table reports estimates of the relation between a firm’s foreign exchange exposure, ˆ ix , and foreign
exchange pass-through to domestic industry prices. ˆ ix is estimated in the regression
rit   i 0   ix EXCH t   im rmt   it , where rt is the monthly rate of return on a firm’s stock for the years
1996 till 2000, rmt is the corresponding monthly rate of return on the value-weighted market index,
EXCH t is the monthly change in value of the U.S. dollar orthogonal to the market return. The first column
below presents estimates of the equation
abs(ˆix )   0  1Di   2 Di j   3 j   4 Sizei  5 LTDratioi   6QuickRatioi   7 ForeignSalesi  8 PayoutRatio  ui
The dummy variable Di takes a value equal to one if the firm disclosed the use of FCD in 1999 and zero
otherwise. πj is the pass-through coefficient in firm i’s industry. Size of the firm is measured as log of total
assets, LTDratio is calculated as long term debt divided by total assets, QuickRatio is calculated as current assets minus
inventory divided by current liabilities, ForeignSales is calculated as foreign sales divided by total sales, and
PayoutRatio is calculated as dividend per share divided by earnings per share. The second column presents
estimates of the equation
abs(ˆix )   0  1 (1  Di )   2 (1  Di )(1   j )   3 (1   j )   4 Sizei   5 LTDratioi   6 QuickRatio i
  7 ForeignSal es i   8 PayoutRati o  ui
t-statistics based on clustered (by industry) standard errors are provided in parenthesis. Significance is indicated by
bold font with superscripts a, b, and c denoting significance at the 1%, 5% and 10% levels respectively.
Dependent Variable = abs( ˆ ix )
FCD User Dummy
FCD User Dummy * Pass-through Coefficient
Pass-through Coefficient
-0.575a
(5.21)
-3.795b
(2.09)
1.009
(0.63)
FCD Non-User Dummy
FCD Non-User Dummy * (1-Pass-through Coefficient)
(1- Pass-through Coefficient)
Size
LTD Ratio
Quick Ratio
Foreign Sales/Net Sales
Payout Ratio
Observations
F-statistic of overall significance
-0.002a
(3.80)
-1.296a
(2.93)
0.066
(1.48)
-0.590a
(3.10)
-0.055a
(3.33)
2826
11.55a
4.370b
(2.32)
-3.795b
(2.09)
2.786
(1.38)
-0.002a
(3.80)
-1.296a
(2.93)
0.066
(1.48)
-0.590a
(3.10)
-0.055a
(3.33)
2826
11.55a
40
Table IX
The Relation between Firms’ Foreign Exchange Exposure and Industry FCD Usage:
Sample Split by Industry Concentration
This table reports estimates of the relation between a firm’s foreign exchange exposure, ˆ ix , and the prevalence
of derivatives hedging in the industry conditional on industry competitiveness. The sample is divided into two
groups. The first group contains firms belonging to industries in the lowest quintile of the four-firm
concentration ratio (Column I). The second group contains all remaining firms (Column II). For reach firm, ˆ ix is
estimated using the regression
rit   i 0   ix EXCH t   im rmt   it
where rt is the monthly rate of return on a firm’s stock for the years 1996 till 2000, rmt is the corresponding
monthly rate of return on the value-weighted market index, EXCH t is the monthly change in value of the U.S.
dollar orthogonal to the market return. The following regression is estimated:
abs(ˆix )   0  1Di   2 Di F j  3 F j   4 Sizei   5 LTDratioi   6QuickRatioi   7 ForeignSalesi  8 PayoutRatio  ui
where the dummy variable Di takes a value equal to one if the firm disclosed the use of FCD in 1999 and zero
otherwise. Fj is our measure of the fraction of hedgers in firm i’s industry calculated as the market value of firms
in the industry that face ex-ante exchange rate exposure and use FCD divided by market value of all firms in the
industry who face ex-ante exchange rate exposure. When calculating F, we exclude the firm’s own hedging
decision. Size of the firm is measured as log of total assets, LTDratio is calculated as long term debt divided by
total assets, QuickRatio is calculated as current assets minus inventory divided by current liabilities,
ForeignSales is calculated as foreign sales divided by total sales, and PayoutRatio is calculated as dividend per
share divided by earnings per share. t-statistics based on clustered (by industry) standard errors are provided in
parenthesis. Significance is indicated by bold font with superscripts a, b, and c denoting significance at the 1%,
5% and 10% levels respectively.
Dependent Variable = abs( ˆ ix )
FCD User Dummy
FCD User Dummy * Fraction of FCD Users in Industry
Fraction of FCD Users in Industry
Size
LTD Ratio
Quick Ratio
Foreign Sales/Net Sales
Payout Ratio
Observations
F-statistic of overall significance
Bottom 1/5th by
Four Firm
Concentration
Ratio
-0.124
(0.52)
-0.522
(0.81)
0.247
(0.53)
-0.001
(1.44)
-1.162
(1.36)
0.199
(1.13)
-1.398b
(2.21)
-0.038a
(4.54)
416
8.79a
Remaining
Firms
-0.126
(0.56)
-1.061b
(2.03)
0.901a
(2.78)
-0.002a
(3.18)
-1.037a
(2.63)
0.060
(1.40)
-0.623a
(2.94)
-0.069b
(2.46)
2410
9.31a
41
Table X
FCD Usage and Firm Value
This table displays the effect of FCD use on firm value conditional on the extent of hedging in a firm’s industry. The sample is restricted to firms that face ex-ante foreign exchange exposure. A firm is
defined as having ex-ante exchange rate exposure if it discloses foreign assets, sales or income in the COMPUSTAT Geographic segment file, or discloses positive values of foreign currency adjustment,
exchange rate effect, foreign income, or deferred foreign taxes in the annual COMPUSTAT files. The dependent variable is the natural log of Tobin’s Q, which is calculated as market value of equity
(calculated as shares outstanding times share price) plus total assets less common equity and deferred taxes, all scaled by book value of assets. FCD non-user dummy equals 1 if the firm does not disclose the
use of foreign currency swaps, options or forwards in its 1999 10K reports and 0 otherwise. The regression is estimated using Heckman’s (1979) two stage procedure (Panel A) as well as with two-stage least
squares (Panel B). In the Heckman two-step estimation, the decision not to use FCD is modeled as a function of firm size, leverage, research and development expense, foreign sales, institutional ownership
and the lagged hedging dummy. Lambda is the self-selection parameter. The coefficient on lambda indicates the prevalence of self-selection in the model. The same instruments are used in the first stage of
the two-stage least squares regression presented in Panel B. In Column I, the regression is estimated for the entire sample provided data requirements are met. Columns II-VII present estimates for three subsamples: firms with the lowest 1/3rd, middle 1/3rd and top 1/3rd of values for the industry hedging measure. Industry hedging is the fraction of hedgers in an industry calculated as the number of firms with exante exposure who use FCD divided by the total number of firms with ex-ante exposure. The following control variables are used; Size calculated as the log of total assets, RND /Sales is Research and
Development Expense divided by Net Sales. CAPEX/Sales is capital expenditures divided by Net Sales. Long-term Debt Ratio is long-term debt divided by total assets. Multiple Segment Dummy equals one
if the firm operates in more than one segment and zero otherwise. Return on Assets is calculated as net income over total assets. Managerial Stockownership/ Total Assets is the market value of shares owned
by executives of the company divided by total assets. Significance at least at the ten percent level is indicated by bold font. The superscripts a, b, and c denote significance at the 1%, 5% and 10% levels
respectively.
Dependent Variable: Tobin’s Q
PANEL A: HECKMAN
I
FCD non-user dummy
Size
RND/Sales
CAPEX/Sales
Long-Term Debt Ratio
Multiple Segment Dummy
Return on Assets
Managerial Stockownership/Total Assets
Self Selection Parameter (Lambda)
Observations
R-Squared
PANEL B: 2SLS
II
III
IV
Hedging in the Firm’s Industry lies in the
V
VI
VII
Hedging in the Firm’s Industry lies in the
ALL
Bottom 1/3
Middle 1/3
Top 1/3
Bottom 1/3
Middle 1/3
Top 1/3
-0.296a
(2.97)
-0.031c
(1.94)
0.186a
(4.86)
-0.182a
(4.67)
-1.637a
(12.58)
-0.211a
(5.63)
1.003a
(6.41)
0.049a
(15.77)
-0.190
(1.13)
-0.023
(0.82)
0.029
(0.30)
-0.025
(0.26)
-1.956a
(8.25)
-0.219a
(3.21)
0.180
(0.80)
0.037a
(9.60)
-0.197
(1.17)
0.021
(0.83)
2.607a
(8.37)
0.315c
(1.86)
-1.227a
(5.35)
-0.163b
(2.51)
2.692a
(7.48)
0.055a
(6.64)
-0.297b
(2.28)
-0.023
(1.02)
0.329a
(3.90)
-0.526
(1.49)
-0.974a
(5.67)
-0.110b
(2.29)
1.723a
(7.00)
0.082a
(11.01)
0.055
(0.28)
-0.003
(0.10)
0.050
(0.50)
-0.046
(0.46)
-2.023a
(8.56)
-0.216a
(3.13)
0.196
(0.86)
0.037a
(9.43)
-0.107
(0.68)
0.028
(1.13)
2.630a
(8.43)
0.330c
(1.91)
-1.212a
(5.19)
-0.156b
(2.38)
2.683a
(7.26)
0.055a
(6.57)
-0.345a
(3.18)
-0.014
(0.81)
0.300a
(3.49)
-0.412
(1.15)
-0.996a
(5.95)
-0.117b
(2.43)
1.684a
(6.80)
0.080a
(10.78)
0.173b
(2.42)
1095
0.046
(0.44)
376
0.143
(1.33)
323
0.131
(1.59)
396
376
0.43
323
0.52
396
0.48
42
Table XI
Past Operating Performance and Stock Performance
This table compares the past operating and stock performance of FCD non-users in ‘highly-hedged’ industries with that of (i) FCD non-users in the remaining
industries and (ii) all FCD users. Firms that do not disclose the use of foreign currency derivatives in 1999 are classified as FCD non-users. Firms that disclose
the use of foreign currency derivatives in 1999 are classified as FCD users. Industries are classified as ‘highly hedged’ if the fraction of hedged firms in that
industry lies in the top third. For each firm, industry adjusted return on assets, gross profit margin, and stock returns are calculated from 1994-1998. Industry
adjustment involves subtracting the industry median from the firm’s performance. These industry adjusted firm characteristics are averaged over the 5 years for
each firm. Return on assets is calculated as net income divided by total assets. Gross profit margin is calculated as net sales less cost of goods sold and selling,
general and admin. expenses, all divided by net sales. Bold font indicates that the difference is significant at least at the ten percent confidence level. Superscripts
a, b and c indicate significance at the 1%, 5% and 10% levels respectively.
PANEL A
Past five year average of:
Industry Adjusted Return on Assets
Industry Adjusted Gross Profit Margin
Industry Adjusted Stock Return
N
212
212
208
FCD Non-Users in
‘Highly Hedged’
Industries
0.022
-0.012
0.060
N
501
500
483
N
FCD Non-Users in
‘Highly Hedged’
Industries
N
Hedged Firms
212
212
208
0.022
-0.012
0.060
371
370
364
0.055
0.121
0.088
FCD Non-Users in
Remaining Industries
0.047
0.069
0.200
Difference
-0.025b
0.081c
-0.139a
PANEL B
Past five year average of:
Industry Adjusted Return on Assets
Industry Adjusted Gross Profit Margin
Industry Adjusted Stock Return
Difference
-0.032a
-0.133a
-0.028
43
Table XII
FCD Usage and Firm Value after Adjusting for Past Performance
(‘Highly-Hedged’ Industries)
This table displays the effect of FCD use on firm value for firms that belong to industries where hedging is widespread. The sample is restricted to
firms with the highest 1/3rdof values for the industry hedging measure. Industry hedging is the fraction of hedgers in an industry calculated as the
number of firms with ex-ante exposure who use FCD divided by the total number of firms with ex-ante exposure. The dependent variable is the
natural log of Tobin’s Q, which is calculated as market value of equity (calculated as shares outstanding times share price) plus total assets less
common equity and deferred taxes, all scaled by book value of assets. FCD non-user dummy equals 1 if the firm does not disclose the use of foreign
currency swaps, options or forwards in its 1999 10K reports and 0 otherwise. The regression is estimated using Heckman’s (1979) two stage
procedure (Column I) as well as with two-stage least squares (Column II). In the Heckman two-step estimation, the decision not to use FCD is
modeled as a function of firm size, leverage, research and development expense, foreign sales, institutional ownership, the lagged hedging dummy,
and past averages of return on assets, stock returns and gross margins. Lambda is the self-selection parameter. The coefficient on lambda indicates the
prevalence of self-selection in the model. The same instruments are used in the first stage of the two-stage least squares regression presented in
Column II. The following control variables are used; Size calculated as the log of total assets. RND /Sales is Research and Development Expense
divided by Net Sales. CAPEX/Sales is capital expenditures divided by Net Sales. Long-term Debt Ratio is long-term debt divided by total assets.
Multiple Segment Dummy equals one if the firm operates in more than one segment and zero otherwise. Return on Assets is calculated as net income
over total assets. Managerial Stockownership/ Total Assets is the market value of shares owned by executives of the company divided by total assets.
Significance at least at 10% is indicated by bold font. The superscripts a, b, and c denote significance at the 1%, 5% and 10% levels respectively.
Dependent Variable: Tobin’s Q
Heckman
2SLS
FCD non-user dummy
0.056
(0.45)
0.028
(1.31)
4.351a
(8.99)
-0.093
(0.28)
-0.622a
(3.74)
-0.093b
(2.09)
2.266a
(9.52)
0.080a
(8.45)
-0.148
(1.25)
0.007
(0.39)
4.155a
(8.58)
-0.058
(0.17)
-0.607a
(3.67)
-0.094b
(2.08)
2.232a
(9.24)
0.080a
(8.33)
Size
RND/Sales
CAPEX/Sales
Long-Term Debt Ratio
Multiple Segment Dummy
Return on Assets
Managerial Stockownership/Total Assets
Self Selection Parameter (Lambda)
Observations
R-squared
-0.091
(1.15)
379
379
0.48
44
Table XIII
Foreign Exchange Pass-through to Domestic Prices
Alternative Measure of Hedging
This table shows the relation between domestic industry prices and the external value of the U.S. dollar conditional on the extent of currency hedging in an
industry. Estimates of the following regression are presented.
 ln RPPI jt   0  1 ln X t 1   2  ln X t 1 * FRACTION
jt
  3  ln X t 1 * FORINP jt   4  ln X t 1 * IMPORTS jt
  5  ln X t 1 * KS jt   6  ln X t 1 * CONC j   7  ln X t 1 * EXPORTS jt   8  ln rit   jt
Although not shown, all variables that appear in the interaction terms are also included separately as control variables. The dependent variable Δln RPPI is the
log change in the domestic relative producer price index, RPPI. RPPI is calculated as the producer price index divided by the overall GDP price deflator.
Foreign exchange movements, X, are measured as the trade-weighted value of the U.S. dollar in terms of its major trading partners as calculated by the Federal
Reserve Board. The fraction of hedgers in an industry, FRACTION, is calculated as the sum of market values of FCD users in the industry divided by sum of
market values of all firms in the industry. The fraction of firms in an industry using interest rate derivatives is used as an instrument for FRACTION. An
industry’s reliance on foreign inputs FORINP, is calculated as in Allayannis and Ihrig (2001) using monthly industry import data provided by United States
International Trade Commission (USITC) and the 1997 benchmark input-output tables provided by the Bureau of Economic Analysis of the U.S. Department
of Commerce. Import penetration, IMPORTS, for an industry is calculated as the monthly general customs value of final goods imports scaled by domestic
shipments for that industry in that year. Industry exports, EXPORTS, are obtained from USITC and scaled by domestic shipments. Capital intensity, KS,
calculated as industry total assets divided by industry sales. The four-firm concentration ratio, CONC, is obtained from the 1997 U.S. Census of Manufacturers.
The U.S. LIBOR, rit is included in log differences. The numbers in parenthesis are t-statistics based on Newey-West standard errors. Significance at least at the
10% level is indicated by bold font. The superscripts a, b, and c denote significance at the 1%, 5% and 10% levels respectively.
Dependent Variable: Δln RPPI
Δ ln X
Δ ln X * Fraction of FCD Users in Industry
Δ ln X * Imported Inputs
-0.077
(1.59)
0.054b
(2.06)
-1.248b
(2.08)
Δ ln X *Imports
Δ ln X * Exports
Δ ln X * Capital Intensity
Δ ln X *4-Firm Concentration Ratio
Δ ln r
Fraction of FCD Users in Industry
Imported Inputs
5.120b
(2.07)
-0.088
(0.03)
-2.767c
(1.96)
0.013a
(3.26)
-0.003
(0.33)
-0.007
(1.59)
Capital Intensity
4-Firm Concentration Ratio
Industry Dummies
Observations
F statistic
-1.675b
(2.31)
5.471b
(2.25)
-0.124
(0.04)
-2.761c
(1.96)
0.013a
(3.24)
-0.003
(0.33)
0.053
(1.29)
0.005
(0.10)
0.016
(0.68)
-0.007b
(1.58)
0.059
(1.58)
0.006
(0.18)
0.007
(0.12)
Yes
5295
3.25a
Yes
5295
3.31a
Imports
Exports
-0.077
(1.60)
0.541b
(2.08)
45
Table XIV
Foreign Exchange Exposure and Industry FCD Usage
Alternative Measure of Hedging
This table reports estimates of the relation between a firm’s foreign exchange exposure, ˆ ix , and the prevalence of hedging in a
firm’s industry. ˆ ix is estimated in the regression rit   i 0   ix EXCH t   imrmt   it , where rt is the monthly rate of return on a
firm’s stock for the years 1996 till 2000, rmt is the corresponding monthly rate of return on the value-weighted market index,
EXCH t is the monthly change in value of the U.S. dollar orthogonal to the market return. The first column in Panel A below
presents estimates of the equation
abs(ˆix )   0  1 Di   2 Di F j   3 F j   4 Sizei   5 LTDratioi   6 QuickRatioi   7 ForeignSalesi   8 PayoutRatio  ui .
The dummy variable Di takes a value equal to one if the firm disclosed the use of FCD in 1999 and zero otherwise. Fj is our
measure of the fraction of hedgers in firm i’s industry calculated as the market value of firms in the industry that use FCD
divided by market value of all firms in the industry. When calculating F, we exclude the firm’s own hedging decision. Size of
the firm is measured as log of total assets, LTDratio is calculated as long term debt divided by total assets, QuickRatio is
calculated as current assets minus inventory divided by current liabilities, ForeignSales is calculated as foreign sales divided by
total sales, and PayoutRatio is calculated as dividend per share divided by earnings per share.
The second column in Panel A presents estimates of the equation
abs( ˆix )   0  1 (1  Di )   2 (1  Di )(1  F j )   3 (1  F j )   4 Sizei   5 LTDratio i   6 QuickRatioi   7 ForeignSalesi
  8 PayoutRatio  ui
t-statistics based on clustered (by industry) standard errors are provided in parenthesis. Significance at least at the 10% level is
indicated by bold font. The superscripts a, b, and c denote significance at the 1%, 5% and 10% levels respectively.
Dependent Variable = abs( ˆ ix )
FCD User Dummy
FCD User Dummy * Fraction of FCD Users in Industry
Fraction of FCD Users in Industry
-0.165
(0.93)
-0.943b
(1.98)
0.764b
(2.44)
FCD Non-User Dummy
FCD Non-User Dummy * Fraction of FCD Non-Users in Industry
Fraction of FCD Non-Users in Industry
Size
LTD Ratio
Quick Ratio
Foreign Sales/Net Sales
Payout Ratio
Observations
F-statistic of overall significance
-0.002a
(3.75)
-1.115a
(3.06)
0.063
(1.44)
-0.685a
(3.24)
-0.052a
(3.36)
2826
9.97a
1.108a
(3.25)
-0.943b
(1.98)
0.179
(0.56)
-0.002a
(3.75)
-1.115a
(3.06)
0.063
(1.44)
-0.685a
(3.24)
-0.052a
(3.36)
2826
9.97a
46
TABLE XV
Exposure of FCD Users and FCD Non-Users to Foreign Exchange Fluctuations
This table reports estimates of the relation between a firm’s foreign exchange exposure coefficient, ˆ ix , and the extent of
hedging in the firm’s industry for separate samples of FCD users and non-users. ˆ ix is estimated in the regression
rit   i 0   ix EXCH t   im rmt   it , where rt is the monthly rate of return on a firm’s stock for the years 1996 till 2000, rmt is the
corresponding monthly rate of return on the value-weighted market index, EXCH t is the monthly change in value of the U.S.
dollar orthogonal to the market return. The table shows results of the following regression:
ˆix   0  1 (1  P)   2 P * F j   3 (1  P) * F j   4 Sizei   5 LTDratioi   6 QuickRatioi   7 ForeignSalesi   8 PayoutRatio  ui
The regression is estimated separately for FCD non-users (Panel A) and FCD users (Panel B). FCD users are firms that
disclosed the use of foreign currency derivatives at the end of fiscal year 1999. All remaining firms are classified as FCD nonusers. In this equation, P is a dummy variable equal to one if ˆix is positive and zero otherwise. Fj is our measure of the
prevalence of hedging in firm i’s industry and is calculated as the market value of FCD users who face ex-ante exchange rate
exposure divided by the market value of all firms facing ex-ante exchange rate exposure. When calculating the fraction of
hedgers in a firm’s industry, we exclude the firm’s own hedging decision. The control variables included are as follows - Size
of the firm is measured as log of total assets, LTDratio is calculated as long term debt divided by total assets, QuickRatio is
calculated as current assets minus inventory divided by current liabilities, ForeignSales is calculated as foreign sales divided by
total sales, and PayoutRatio is calculated as dividend per share divided by earnings per share.
t-statistics based on clustered (by industry) standard errors are provided in parenthesis. Significance at least at the 10% level is
indicated by bold font. The superscripts a, b, and c denote significance at the 1%, 5% and 10% levels respectively.
Dependent Variable: ˆ ix
Panel A: FCD Non-Users
Industry Hedging (F) * P
Industry Hedging (F) *(1-P)
Size
LTD Ratio
Quick Ratio
Foreign Sales/Net Sales
Payout Ratio
(1-P)
Constant
Observations
F-statistic
R-squared
Panel B: FCD Users
a
1.051
(3.82)
-0.442
(1.20)
-0.001a
(2.99)
-0.158
(0.62)
0.023
(1.38)
0.274
(0.93)
-0.039b
(2.49)
-1.689a
(9.99)
1.194a
(8.88)
2296
60.94a
0.29
Industry Hedging (F) * P
Industry Hedging (F) *(1-P)
Size
LTD Ratio
Quick Ratio
Foreign Sales/Net Sales
Payout Ratio
(1-P)
Constant
Observations
F-statistic
R-squared
-0.231
(0.36)
0.757b
(2.06)
-0.001
(1.42)
-0.234
(0.47)
-0.073
(1.01)
0.732
(1.60)
-0.033
(1.19)
-1.876a
(5.37)
1.133a
(3.02)
389
26.20a
0.27
47
TABLE XVI
Foreign Exchange Pass-through to Domestic Prices
(Using Clustered Standard Errors)
This table shows the relation between domestic industry prices and the external value of the U.S. dollar conditional on the extent of currency hedging in an
industry. The following regression is estimated using generalized method of moments (GMM) and clustered standard errors.
 ln RPPI jt   0  1 ln X t 1   2  ln X t * FRACTION
jt
  3  ln X t 1 * FORINP jt   4  ln X t 1 * IMPORTS jt
  5  ln X t 1 * KS jt   6  ln X t 1 * CONC j   7  ln X t 1 * EXPORTS jt   8  ln rit   jt
Although not shown, all variables that appear in the interaction terms are also included separately as control variables. The dependent variable Δln RPPI is the
log change in the domestic relative producer price index, RPPI. RPPI is calculated as the producer price index for each 3-digit SIC code divided by the overall
GDP price deflator. Foreign exchange movements, X, are measured as the trade-weighted value of the U.S. dollar in terms of its major trading partners as
calculated by the Federal Reserve Board. The fraction of hedgers in an industry, FRACTION, is calculated as the sum of market values of FCD users in the
industry who face ex-ante foreign exchange exposure divided by sum of market values of all firms in the industry who face ex-ante foreign exchange exposure.
An industry’s reliance on foreign inputs FORINP, is calculated as in Allayannis and Ihrig (2001) using monthly industry import data provided by United States
International Trade Commission (USITC) and the 1997 benchmark input-output tables provided by the Bureau of Economic Analysis of the U.S. Department
of Commerce. Import penetration, IMPORTS, for an industry is calculated as the monthly general customs value of final goods imports scaled by domestic
shipments for that industry in that year. Industry exports, EXPORTS, are obtained from USITC and scaled by domestic shipments. Capital intensity, KS,
calculated as industry total assets divided by industry sales. The four-firm concentration ratio, CONC, is obtained from the 1997 U.S. Census of Manufacturers.
The U.S. LIBOR, rit is included in log differences. The numbers in parenthesis are t-statistics based on clustered (at the 2-digit SIC level) standard errors.
Standard errors are clustered at the 2-digit SIC level. Significance at least at the 10% level is indicated by bold font. The superscripts a, b, and c denote
significance at the 1%, 5% and 10% levels respectively.
Dependent Variable: Δln RPPI
Δ ln X
Δ ln X * Fraction of FCD Users in Industry
Δ ln X * Imported Inputs
-0.058
(0.95)
0.372c
(1.86)
-0.662a
(2.50)
Δ ln X * Imports
Δ ln X * Capital Intensity
Δ ln X * 4-Firm Concentration Ratio
Δ ln X * Exports
Δ ln r
Fraction of FCD Users in Industry
Imported Inputs
-0.160
(-0.07)
-2.33
(1.35)
3.947a
(2.81)
0.013a
(3.49)
-0.001
(0.35)
-0.011b
(2.15)
4-Firm Concentration Ratio
Exports
Year Dummies
Observations
F statistic
-1.06b
(2.12)
-0.015
(0.07)
-2.32
(1.34)
4.26a
(2.78)
0.013a
(3.52)
-0.001
(0.31)
0.028
(0.55)
0.019
(0.55)
0.059
(1.23)
-0.010b
(2.25)
0.027
(0.53)
0.019
(0.55)
0.067
(1.35)
Yes
4731
18.37a
Yes
4731
9.77a
Imports
Capital Intensity
-0.059
(0.97)
0.375c
(1.89)
48
I. Introduction
How does competition affect corporate hedging strategies? Allayannis and Ihrig (2001)
predict that firms, which operate in more competitive industries, are more likely to hedge.
In contrast, Mello and Ruckes (2005) predict that firms hedge less if competition is more
intense.25 Adam, Dasgupta, and Titman (2007) show that competition can have a positive
or negative impact on the number of firms that hedge in equilibrium, depending on
whether hedging or not hedging is optimal in the absence of any competitive interaction
between firms.
The objective of this paper is to empirically investigate how competition affects corporate
hedging strategies. We examine the impact of competition on the incentive of an individual
firm to hedge, on the fraction of firms that hedge in equilibrium, and on the extent of
hedging. We also examine how other industry factors, such as market size, industry
concentration, the maturity of an industry, financial heterogeneity, etc. affect firms’
incentives to hedge and their extent of hedging, and whether firm-specific and industry
factors are more important in explaining variations in corporate hedging strategies.
Several studies find significant variation in hedging strategies across industries. For
example, Nain (2004) finds that within an industry the fraction of firms that use FX
derivatives ranges from 0-100%. Geczy, Minton and Schrand (1997) also find significant
variation in hedging practices across industries. Most of this variation cannot be explained
by firm-specific factors alone, highlighting the potential importance of industry effects in
explaining corporate hedging strategies.
There is a large empirical literature that examines the correlations between firm-specific
characteristics, such as firm size, financial constraints, taxes, agency problems, etc. and
hedging strategies. We add to this literature by focusing on the relations between industry
characteristics and hedging strategies.
Our results suggest that competition reduces a firm's incentive to hedge, but has no
measurable impact on the extent of hedging. In particular, we find that in concentrated
industries with low price-cost margins fewer firms hedge than in less concentrated
industries with high price-cost margins.
We find no consistent correlations between the fraction of firms that use derivatives and
other variables, such as the average firm size, sensitivity of the PPI to FX, production
convexity, the average market share per firm, and the fraction of firms with better than A
(BBB) rating.
…
The impact of competition on firms’ risk exposures, and by deduction risk management
strategies has been observed by several researchers. For example, Lewent and Kearney
Firms also hedge less if firm’s financial conditions are similar, products are more homogeneous, and if
firms operate under higher fixed costs.
25
49
(1990) write on p. 19 “… the impact of exchange rate volatility on a company depends
mainly on the companies’ business structure, its industry profile, and the nature of its
competitive environment.” Indeed, He and Ng (1998), and Williamson (2001) find that the
degree of competition affects exposures: more competition implies larger exposures.
Brown (2001) reports that firms view their risk management strategies as a competitive
tool and carefully track the exposures of competitors in order to identify their own optimal
policies and product market strategies. Nain (2004) studies how a firm’s incentive to hedge
depends on the extent of hedging in the same industry. She finds that firms are more likely
to hedge FX risk if many of its competitors are doing the same. This effect is stronger in
less competitive industries. Firms that do not follow the hedging strategies of the majority
face a higher total risk exposure, and suffer a value discount. Finally, Froot, Scharfstein
and Stein (1993) suggest that the structure of competition can affect a firm’s optimal risk
management strategy and may lead to hedging decisions that are different from their rivals.
Allayannis and Weston (1999) find that firms which operate in lower mark-up (more
competitive) industries are more likely to use FX derivatives. Our results seem to
contradict their findings. However, Allayannis and Weston (1999) examine the decision to
use derivatives using a probit estimation, we examine the fraction of derivatives users. The
extent to which industry factors affect hedging strategies has not been examined
empirically in a broad cross-sectional study.
The importance of industry factors on firms’ real and financial decisions has been noted by
several other authors. For example, Maksimovic and Zechner (1991) show in a partial
equilibrium framework that firms whose production technologies are similar to the median
technology choose less leverage than firms whose technology departs from the industry
norm. MacKay and Phillips (2005) find supporting empirical evidence for this prediction
in competitive but not in concentrated industries.
The remaining paper is structured as follows. Section II reviews the theory and derives
testable predictions. The data sample and variables used in the econometric analysis are
described in Section III. Section IV contains the econometric analysis, and Section V
concludes.
50
Table 1
Descriptive statistics (firm level data)
This table lists descriptive statistics of firms that face ex-ante FX exposures.
Mean
Median
Std. dev.
Min
Max
Obs.
Use of derivatives
(dummy variable)
0.153
0
0.360
0
1
2,806
Extent of using derivatives
0.012
0
0.087
0
2.96
2,794
Extent of using derivatives
(hedgers only)
0.079
0.028
0.213
0.000
2.96
417
Size of FX exposure
0.357
0.293
0.273
0.000
1
2,398
Market value of assets
(in millions of US$)
4,302
347.7
18,736
0.076
408,030
2,398
Tobin’s q
2.129
1.475
1.906
0.525
19.51
2,387
R&D expense / sales
0.185
0.017
1.118
0
29.37
2,804
Debt-equity ratio
0.565
0.146
1.398
0
22.09
2,393
Quick ratio
1.820
1.283
1.688
0.053
16.54
2,713
Dividend payout ratio
0.130
0
0.618
0
15
2,719
0.132
0
0.663
0
12.39
2,392
0.240
0
0.427
0
1
2,806
Tax-loss carry-forwards /
book value of assets
Existence of credit rating
(dummy variable)
51
Table 2
Descriptive statistics (industry level data)
Median variables are unweighted medians based on all firms in a 6-digit NAICS industry.
Mean
Median
Std. dev
Min
Max
Obs.
0.195
0
0.324
0
1
802
0.010
0
0.041
0
0.481
787
0.576
0.797
0.433
0
1
766
Median exposure (exposed firms)
0.318
0.242
0.258
0.000
1
526
Median market value of assets
(in millions of US$)
834
163.4
3209
0.439
62,000
770
Median Tobin’s q
1.565
1.324
0.948
0.581
14.12
766
Median debt-equity ratio
0.601
0.293
0.983
0
8.66
770
Median quick ratio
1.300
1.143
0.888
0.000
8.512
783
Median dividend payout
0.054
0
0.157
0
1.489
776
Fraction of firms with investment
grade rating
0.040
0
0.116
0
1
802
Market share per firm
682.2
199.4
2557
0.016
60,072
797
Sales growth
1.004
0.906
0.458
0.194
3.639
606
Price sensitivity
0.003
0.003
0.088
-0.382
0.561
314
1.361
1.246
0.746
0.013
6.062
560
0.716
0.619
0.501
0.004
4.641
582
Operating leverage
0.036
0.007
0.133
-0.329
0.969
664
Cost convexity
0.002
0.000
0.179
-1.824
1.863
458
Market value weighted fraction of
derivatives users (exposed firms)
Industry average hedge ratio
(exposed firms)
Weighted fraction
of exposed firms
Coefficient of variation (debtequity ratio)
Coefficient of variation (quick
ratio)
52
Table 3
Measuring the degree of competition
Panel C shows the number of industries as a function of the PCM and the Herfindahl index. Only
industries that are covered by the 1999 Annual Survey of Manufacturers are considered. Below
median values of both the PCM and the Herfindahl index indicate that an industry is more
competitive than average. Above median values indicate that an industry is less competitive than
average.
Panel A: Descriptive statistics
Mean
Median
Std. dev.
Min
Max
Obs.
PCM
0.324
0.305
0.163
0
1
701
PCMCensus
0.337
0.329
0.099
0.094
0.818
350
Herfindahl indexCensus
0.423
0.394
0.265
0.009
0.999
237
0.423
0.406
0.209
0.036
1
349
0.553
0.561
0.223
0.066
1
346
9.5
3
28.7
1
582
787
Concentration ratio
(top 4 firms)
Concentration ratio
(top 8 firms)
Number of firms per
6-digit NAICS industry
Panel B: Correlation matrix of competition measures
PCM
PCMCensus
Herfindahl
indexCensus
PCMCensus
0.457
Herfindahl indexCensus
0.141
0.142
0.030
0.246
0.970
0.028
0.226
0.941
Concentration ratio
(top 4 firms)
Concentration ratio
(top 8 firms)
Concentrat. ratio
(top 4 firms)
0.974
Correlation coefficients that are significant at the 1% level are in bold font.
Panel C: Number of industries (observations) in sample
Herfindahl indexCensus
PCMCensus
Below median
Above median
Total
Below median
74
54
128
Above median
45
64
109
Total
119
118
237
53
Table 4
The determinants of hedging policy choice (univariate comparisons)
Panels A and B show the average fractions of firms that use derivatives in an industry as a function
of the price cost margin (PCM) and the Herfindahl index. Only industries that are covered by the
1999 Annual Survey of Manufacturers are considered.
Panel A: Fraction of firms that use derivatives (firms with an ex-ante FX exposure only)
Herfindahl indexCensus
PCMCensus
Below median
Above median
Total
Below median
0.209
0.247
0.225
Above median
0.285
0.340
0.318
Total
0.238
0.298
0.268
Panel B: Industry average hedge ratio (firms with an ex-ante FX exposure only)
Herfindahl indexCensus
PCMCensus
Below median
Above median
Total
Below median
0.007
0.013
0.010
Above median
0.013
0.013
0.013
Total
0.009
0.013
0.011
54
Table 5
The determinants of hedging policy choice (firm-level regressions)
This table reports the estimation results of a Heckman two-step selection model. The first stage
evaluates the determinants of the decision to use derivatives. The second stage evaluates the
determinants of the extent of using derivatives. The extent of using derivatives is measured by the
notional principal of outstanding FX derivatives scaled by a firm’s book/market value of assets.
Figures in parentheses denote t-statistics.
All firms
Firms with FX exposure only
Decision to use
derivatives
Extent of using
derivatives
Decision to use
derivatives
Extent of using
derivatives
Intercept
-2.796***
(-15.62)
-0.683
(-0.77)
-2.129***
(-9.25)
-1.558
(-0.66)
Size of FX exposure
0.725***
(4.83)
0.183
(1.08)
0.099
(0.51)
0.073
(0.45)
ln(Market value of assets)
0.288***
(11.24)
0.048
(0.72)
0.233***
(7.53)
0.109
(0.63)
Tobin’s q
-0.062**
(-2.42)
-0.020
(-1.17)
-0.046
(-1.52)
-0.031
(-0.75)
-0.230
(-1.50)
-0.092
(-0.92)
-0.272
(-1.04)
-0.292
(-0.79)
-0.149***
(-2.65)
-0.038
(-0.97)
-0.161**
(-2.41)
-0.089
(-0.71)
Quick ratio
0.006
(0.28)
-0.001
(-0.10)
0.021
(0.62)
0.011
(0.34)
Dividend payout ratio
-0.054
(-0.94)
-0.002
(-0.08)
-0.058
(-0.96)
-0.028
(-0.46)
Tax-loss carry-forwards
-0.123
(-0.62)
0.231
(1.71)
-0.080
(-0.39)
0.209
(0.77)
Existence of credit rating
(dummy variable)
0.212**
(-2.11)
0.072
(1.17)
0.304***
(2.61)
0.185
(0.77)
PCMCensus 
Concentration ratio
-0.107
(-0.57)
0.074
(1.06)
-0.216
(-0.95)
0.011
(0.05)
Number of obs.
2,525
300
1,133
249
R&D expense / sales
Debt-equity ratio
Wald test
Pseudo R2
354.53***
0.249
159.43***
0.149
55
Table 6
Competition and the fraction of firms that use derivatives
This table reports industry-level tobit regressions of the market-value-weighted fraction of
derivatives users (based on firms that have an ex-ante FX exposure) on measures of the degree of
competition and control variables. Industries are classified by the 6-digit NAICS. Figures in
parentheses denote t-statistics.
I
II
III
IV
V
VI
Intercept
-0.703***
(-5.73)
-0.671***
(-4.07)
-0.385*
(-1.89)
-0.480***
(-3.28)
-0.545**
(-2.60)
-0.565***
(-3.81)
PCM
0.665***
(3.55)
1.296***
(3.54)
PCMCensus
Herfindahl
indexCensus
0.362**
(2.01)
Concentration ratio
(top 4 firms)
0.483***
(2.67)
PCMCensus 
Herfindahl index
0.814***
(3.66)
PCMCensus 
Concentration ratio
0.661***
(4.07)
Fraction of
exposed firms
0.495***
(6.50)
0.260**
(2.55)
0.367***
(2.75)
0.306***
(2.96)
0.324**
(2.46)
0.279***
(2.74)
ln(Median firm size)
0.050***
(2.80)
0.049**
(2.39)
0.019
(0.60)
0.029
(1.31)
0.019
(0.63)
0.034
(1.62)
Median Tobin’s q
-0.128***
(-2.95)
-0.086
(-1.23)
-0.077
(-0.87)
-0.005
(-0.07)
-0.127
(-1.41)
-0.051
(-0.76)
Number of obs.
659
338
231
337
231
337
Log likelihood
-465.4
-266.7
-183.5
-269.1
-178.5
-264.2
Pseudo R2
0.086
0.057
0.041
0.047
0.067
0.065
***, ** and * denote significance at the 1%, 5%, and 10% level respectively.
56
Table 7
Competition and the fraction of firms that use derivatives
This table reports industry-level tobit regressions of the market-value-weighted fraction of
derivatives users (based on firms that have an ex-ante FX exposure) in an 6-digit NAICS industry
on measures of the degree of competition and additional control variables. Figures in parentheses
denote t-statistics.
I
II
III
IV
V
VI
Intercept
-0.506**
(-2.44)
-0.650***
(-2.89)
-0.335
(-1.36)
-0.360*
(-1.80)
-0.525**
(-2.01)
-0.476**
(-2.31)
PCM
0.679**
(2.45)
1.083***
(2.84)
PCMCensus
Herfindahl
indexCensus
0.102
(0.50)
Concentration ratio
(top 4 firms)
0.066
(0.27)
PCMCensus 
Herfindahl index
0.571**
(2.32)
PCMCensus 
Concentration ratio
Weighted fraction
of exposed firms
0.433**
(2.20)
0.415***
(3.24)
0.374***
(2.67)
0.574***
(3.56)
0.441***
(3.12)
0.508***
(3.18)
0.391***
(2.77)
ln(Median firm size)
0.001
(0.04)
0.010
(0.26)
0.024
(0.50)
0.004
(0.11)
0.021
(0.45)
0.003
(0.08)
Price sensitivity
0.502
(1.29)
0.962*
(1.91)
1.273*
(1.78)
0.809
(1.59)
1.099
(1.57)
0.837*
(1.66)
Cost convexity
0.459*
(1.76)
0.272
(0.89)
-0.102
(-0.23)
0.252
(0.80)
-0.117
(-0.26)
0.213
(0.69)
ln(market share)
0.025
(0.59)
0.028
(0.59)
-0.011
(-0.17)
0.034
(0.68)
-0.011
(-0.18)
0.027
(0.55)
-0.192
(-0.59)
-0.373
(-1.08)
-0.308
(-0.52)
-0.217
(-0.63)
-0.328
(-0.57)
-0.272
(-0.79)
212
183
132
183
132
183
Log likelihood
-155.0
-137.2
-100.3
-141.2
-97.7
-138.8
Pseudo R2
0.090
0.093
0.092
0.067
0.115
0.083
Fraction of firms
with investment
grade rating
Number of obs.
***, ** and * denote significance at the 1%, 5%, and 10% level respectively.
57