RPS 11 Poin 1 Zhang2009

ARTICLE IN PRESS
Journal of Accounting and Economics 47 (2009) 244–264
Contents lists available at ScienceDirect
Journal of Accounting and Economics

journal homepage: www.elsevier.com/locate/jae
Effect of derivative accounting rules on corporate risk-management

behavior$
Haiwen Zhang
Fisher College of Business, The Ohio State University, Fisher Hall 436, 2100 Neil Avenue, Columbus, OH 43210, USA
a r t i c l e i n f o abstract
Article history: I examine the effect of the accounting standard for derivative instruments (SFAS No.
Received 11 September 2006 133) on corporate risk-management behavior. I classify a derivative user as an ‘‘effective
Received in revised form hedger’’ (EH firm) if its risk exposures decreased after the initiation of the derivatives
3 November 2008
program, and as an ‘‘ineffective hedger/speculator’’ (IS firm) otherwise. I find that
Accepted 19 November 2008
volatility of cash flows and risk exposures related to interest rate, foreign exchange rate,
Available online 3 December 2008
and commodity price decrease significantly for IS firms but not for EH firms, suggesting
JEL Classification: that IS firms engaged in more prudent risk-management activities after the adoption of
G1 SFAS No. 133.
G32
& 2008 Elsevier B.V. All rights reserved.
M41
Keywords:
SFAS 133
Derivative financial instruments
Risk-management behavior
1. Introduction
In 1998, FASB issued Statement of Financial Accounting Standards (SFAS) No. 133, Accounting for Derivative Instruments and
Hedging Activities. As the primary directive with respect to accounting treatment for derivatives in the U.S., SFAS 133
requires that entities record all derivatives as either assets or liabilities at fair value and recognize unrealized gains or losses
due to changes in fair value in their income statements. However, while prescribing fair value accounting in general, the
standard allows special accounting treatment for derivatives that qualify as effective hedge instruments. Specifically, under
hedge accounting, a company can record the changes in fair value of an effective hedge instrument and the changes in
fair value of the underlying hedged item in the income statement simultaneously. Thus, only the ineffective portion of
the hedge instrument affects a company’s net income.1 The objective of this study is to examine whether the changes to the
recognition and disclosure requirements for derivative instruments mandated by SFAS 133 has had an effect on corporate
risk-management behavior.
$
I gratefully acknowledge the guidance and comments from my dissertation committee members: Frank Gigler, Chandra Kanodia, Judy Rayburn, Ram
Venkataraman, and especially my adviser Pervin Shroff. I wish to thank Thomas Lys (the editor), an anonymous referee, Jennifer Altamuro, Anne Beatty,
John Dickhaut, Amy Hutton, Rick Johnston, John Kareken, Rashad Abdel-Khalik, Kimberly Smith, Theodore Sougiannis, Ivy Zhang, and workshop
participants at University of Minnesota, The Ohio State University, UIUC, UCLA, Dartmouth, College of William and Mary, UBC, and University of Toronto,
for their helpful comments. All errors are mine.
Tel.: +1 614 292 6547.
E-mail address: zhang.614@osu.edu
1
Please see Section 2.1 for a more detailed discussion on the accounting treatment for derivative instruments.
0165-4101/$ - see front matter & 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jacceco.2008.11.007
ARTICLE IN PRESS
H. Zhang / Journal of Accounting and Economics 47 (2009) 244–264 245
The introduction of SFAS 133 has generated intense debate over how derivatives accounting affects firms’ risk-
management activities. Proponents of the standard maintain that derivatives create new risks that are not appropriately
disclosed or recognized under historical-cost accounting and that fair value-based recognition makes the use of derivatives
more transparent, encouraging prudent risk management. This view is supported by the theoretical finding of Melumad et
al. (1999), who illustrate that hedge accounting with fair value measurement can reveal a firm’s underlying risk exposure
and thereby induce the first-best hedge choice, as in a setting with no asymmetric information. In contrast, opponents of
the standard insist that firms mainly use derivatives to hedge their inherent business risk and that fair value measurement
potentially leads to higher short-term earnings volatility in the financial statements. This may deter the legitimate use of
derivatives for hedging purposes. Investigating this issue from the capital market valuation perspective, Sapra (2002)
reaches a similar conclusion. His model shows that an extreme position in the derivatives market signals favorable private
information about the firm’s future spot price, which, in turn, leads to a higher capital market price. Sapra’s findings
suggest that increased reporting transparency (resulting from SFAS 133) may actually induce a firm to take an excessive
speculative position.
While the controversy relating to SFAS 133 centers on how it affects firms’ risk-management activities, extant empirical
evidence focuses on earnings volatility rather than firms’ real actions. For example, Singh (2004) and Park (2004) find no
significant change in earnings volatility after the adoption of SFAS 133. Both papers conclude that the impact of SFAS 133
may not be as significant as claimed. However, while the change in earnings volatility may be a relevant issue, it is
important to differentiate the effect of SFAS 133 on short-term earnings volatility assuming firms’ risk-management
behavior remains unchanged2 and the effect of SFAS 133 on firms’ risk-management behavior. According to the hedge
accounting treatment specified in SFAS 133, unrealized gains or losses of the hedging instrument can be offset by
unrealized losses or gains of the hedged item in the income statement. Hence, only the ineffective portion of a hedging
instrument is reflected in contemporaneous earnings if hedge accounting is adopted. However, if the increase in earnings
volatility is material and costly and a firm adjusts its derivatives portfolio in anticipation of this potential cost,3 no increase
in earnings volatility will be observed following the adoption of SFAS 133.
In this paper, I separately analyze firms that successfully reduce their inherent risk after holding derivative instruments
(effective hedgers) and firms that fail to reduce their risk exposure after initiating derivatives programs (ineffective
hedgers/speculators). I first provide evidence on whether firms changed their risk-management activities after the
adoption of SFAS 133 and then examine changes in earnings volatility and cash flow volatility given the changes in risk-
management behavior.
Since the inherent business risk for firms with derivative positions is unobservable, I follow the ‘‘new user’’ approach
proposed by Guay (1999) and classify new derivative users as either effective hedgers (EH firms) or ineffective hedgers/
speculators (IS firms).4 First, I identify a sample of 225 non-financial firms that initiated their derivatives programs during
the period of 1996–1999. Next, following previous literature (Wong, 2000; Guay, 1999; Geczy et al., 1997), I estimate new
derivative users’ exposure to (i) interest rate, (ii) foreign currency exchange rate, and (iii) commodity price risk for each of
the following three periods: the period before the initiation of the derivatives program (period 1), the period after the
initiation of the derivatives program but before the adoption of SFAS 133 (period 2), and the period after the adoption of
SFAS 133 (period 3). I classify a firm as an effective hedger (EH firm) if its exposure to a given risk is lower than the expected
level after the initiation of the derivatives program, and as an ineffective hedger/speculator (IS firm) if its exposure is higher
than the expected level after the initiation of the derivatives program, where information about new users prior to the
initiation of the derivative program (period 1) is used to estimate the expected level of new users’ risk exposures. This
classification criterion is based on whether firms successfully use derivative instruments to reduce their risk exposures and
therefore does not differentiate speculators from ineffective hedgers. Since SFAS 133 requires companies to document their
hedge effectiveness both prior to entering derivative contracts and periodically thereafter to qualify for hedge accounting
treatment, the impact of the accounting standard on firms’ financial statements is the same for firms that intend to
speculate and for firms that intend to hedge (establish hedge effectiveness ex-ante) but end up holding ineffective hedging
instruments (fail to establish hedge effectiveness ex-post). Finally, I examine changes in the three types of risk exposure
after the adoption of SFAS 133 for EH firms and IS firms separately.
For IS firms, I find a significant reduction in all three types of risk exposure during the post-SFAS 133 period as compared
with the pre-SFAS 133 period after controlling for other related risk factors and potential changes in firms’ incentives to
hedge/speculate. In contrast, I find no significant change in any of the three types of risk exposure for EH firms. Next, I find
that cash flow volatility for IS firms significantly decreased relative to that for EH firms after the adoption of SFAS 133,
consistent with changes in risk exposure. However, I find no significant change in earnings volatility for either EH firms or
IS firms after controlling for the contemporaneous change in cash flow volatility. Since earnings are affected by both real
2
Hodder et al. (2006) and Barth et al. (1995) find that the volatility of annual income after incorporating changes in the fair value of financial
instruments is significantly higher than the volatility of annual income measured with historical cost accounting.
3
Austin et al. (1995), Subramanyam (1996), and Defond and Park (1997) find that managers make accounting choices (at least in part) to reduce
earnings volatility. The survey evidence in Graham et al. (2005) suggests that managers sometimes manage real activities to reduce earnings volatility.
Barton (2001) documents that managers use hedging instruments and discretionary accruals as partial substitutes to smooth earnings.
4
Using a sample of non-financial firms that initiated their derivatives programs in the early 1990s, Guay (1999) finds that risk exposures of these new
users decrease significantly after the initiation of the derivatives program and concludes that, on average, firms use derivatives to hedge.
ARTICLE IN PRESS
246 H. Zhang / Journal of Accounting and Economics 47 (2009) 244–264
business activities and the accounting treatment, these results suggest that IS firms adjusted their derivatives portfolios
towards more effective hedging activities in the post-SFAS 133 period and therefore the changes in earnings volatility are
insignificant ex-post.
Although it is impossible to identify the exact relationship between firms’ inherent business risk and their derivative
positions based on publicly available data, analysis of changes in notional amount and direction of derivative positions after
the adoption of SFAS 133 for EH and IS firms provides additional insights about how firms adjust their derivative portfolios.
For firms that use derivative instruments to manage their interest rate risk and foreign exchange rate risk, I find that 19%
(18 out of 95) of the IS firms stopped holding derivative positions whereas only 6% (8 out of 121) of the EH firms stopped
holding derivative positions after the adoption of SFAS 133. To the extent that ineffective hedgers are less likely to improve
their hedging effectiveness by holding zero derivative positions, this evidence suggests that SFAS 133 has discouraged firms
from engaging in speculative activities. Analysis of the directions of new users’ derivative positions suggests that IS firms
that manage interest rate risk and foreign exchange rate risk are more likely to hold offsetting derivative positions after the
adoption of SFAS 133. However, I do not find increased offsetting derivative positions for IS firms that use derivative
instruments to manage commodity price risk. Overall, changes in the notional amount and the direction of derivative
positions for new derivative users suggest that IS firms comprise both speculators and ineffective hedgers and that IS firms
generally engage in more prudent risk-management activities after the adoption of SFAS 133.
In this paper, I provide indirect evidence on how changes in the recognition and disclosure requirements for derivatives
mandated by SFAS 133 are associated with firms’ risk-management activities. The results extend the current literature on
the effects of SFAS 133 and shed light on the impact of accounting regulations on firms’ real economic behavior (e.g., Beatty,
2007; Kanodia et al., 2000). In addition, this paper makes the first attempt to empirically identify firms that effectively
hedge their business risk, providing a sharper test of the effects of SFAS 133.
The remainder of the paper proceeds as follows. Section 2 describes the hypotheses. Section 3 discusses the sample
selection and research design. Empirical results are presented in Section 4. Section 5 summarizes various sensitivity tests
and Section 6 concludes.
2. Hypothesis development
2.1. Accounting treatment for derivative instruments
Prior to SFAS 133, the accounting treatment for derivative instruments was incomplete and inconsistent. SFAS 52 (1981),
Foreign Currency Translation, specified the accounting treatment for derivative instruments related to foreign currencies
(e.g., forward exchange contracts and currency swaps). SFAS 80 (1984), Accounting for Futures Contracts, established the
accounting reporting standards for futures contracts except foreign currency futures. Accounting for derivative instruments
not specifically covered by SFAS 52 or SFAS 80 was developed largely by analogy to these two standards. In general, the
accounting treatment for derivative instruments prior to SFAS 133 depended on the claimed purpose of the derivative
investment. If a firm held derivative instruments for trading purposes, it was required to recognize the derivative
instruments at fair value on the balance sheet and to recognize any unrealized gains or losses on the income statement. On
the other hand, if a firm held derivative instruments to hedge the risk of existing assets/liabilities or forecasted
transactions, the accounting treatment for the derivative instruments was determined by the accounting treatment of the
hedged items. If the related hedged items were recorded at fair value (historical cost), the hedging instruments were also
recorded at fair value (historical cost). Since most non-financial firms claimed that they held derivative portfolios for
hedging purposes and the related hedged assets/liabilities or transactions were measured at historical cost, their
derivatives portfolios were also recorded at historical cost, which was often zero or negligible.5
To address the concern that historical-cost accounting may not fully reflect a firm’s underlying risk profile, the FASB and
the SEC issued a series of accounting standards and reporting requirements in the 1990s.6 However, the accounting
guidance for derivative instruments remained incomplete and inconsistent. SFAS 133 (paragraph 235) notes that ‘‘before
the issuance of this statement, accounting standards specifically addressed only a few types of derivatives’’ and that ‘‘many
derivative instruments were carried ‘off-balance-sheet’ regardless of whether they were formally part of a hedging
strategy.’’ In addition, prior to the issuance of SFAS 133, ‘‘the required accounting treatment differed depending on the type
of instrument used in a hedge and the type of risk being hedged’’ and ‘‘the accounting standards were inconsistent on
whether qualification for hedge accounting was based on risk assessment at an entity-wide or an individual-transaction
level’’ (SFAS 133, paragraph 236).
5
SFAS 133 documents in paragraph 219 that ‘‘before the issuance of this statement, many derivatives were ‘off-balance-sheet’ because derivatives
often reflect at their inception only a mutual exchange of promises with little or no transfer of tangible consideration.’’
6
The FASB issued SFAS No. 105 (1990), Disclosure of Information about Financial Instruments with Off-Balance-Sheet Risk and Financial Instruments with
Concentrations of Credit Risk, SFAS No. 107 (1991), Disclosure about Fair Values of Financial Instruments, and SFAS No. 119 (1994), Disclosure about Derivative
Financial Instruments and Fair Value of Financial Instruments. In 1997, the SEC issued Financial Reporting Release 48 (FRR 48), which specifies detailed
disclosure requirement for quantitative and qualitative information about market risk inherent in derivative financial instruments, other financial
instruments, and derivative commodity instruments.
ARTICLE IN PRESS
In contrast to the preceding accounting standards, SFAS 133 requires that an entity recognize all derivatives as either
assets or liabilities on the balance sheet at fair value and recognize changes in the fair value of the derivatives as unrealized
gains/losses on the income statement. If certain conditions are met, a derivative instrument may be specifically designated
as (a) a hedge of the exposure to changes in the fair value of a recognized asset or liability (fair value hedge), or (b) a hedge
of the exposure to variability in the cash flows of a recognized asset or liability, or of a forecasted transaction (cash flow
hedge).7 For a derivative designated as a fair value hedge, the unrealized gains (losses) resulting from the change in the fair
value of the hedging instrument are recognized in current earnings together with the offsetting unrealized losses (gains)
resulting from the change in fair value of the hedged item. For a derivative designated as a cash flow hedge, the effective
portion of the derivative’s gains or losses is initially reported as a component of other comprehensive income and
subsequently reclassified into earnings when the forecasted transaction affects earnings. The ineffective portion of gains or
losses is reported in earnings immediately.
In short, SFAS 133 mandates fair value recognition of all derivative instruments on firms’ financial statements
and provides preferential hedge accounting treatment if certain conditions are met. Any speculative position or
hedge ineffectiveness will directly affect the current-period earnings because there is no offsetting adjustment from the
hedged items.
2.2. The impact of SFAS 133 on firms’ risk-management activities
Before examining whether firms’ risk-management behavior changed after the adoption of SFAS 133, it is important to
understand how a firm chooses its production and derivatives positions. Danthine (1978) shows that a firm bases its
production decisions solely on the futures price of the derivatives market and chooses its derivatives position based on the
expectation of the future spot price. Hence, depending on different private information and levels of basis risk,8 a firm can
hold derivatives to completely offset the inherent business risk (perfect hedge), to partially offset the inherent business risk
(partial hedge), or to increase risk. A firm may increase risk by intentionally speculating in the derivatives market
(speculator) or by failing to hold effective hedge instruments (ineffective hedger). To account for the underlying economics
of the investment in derivatives, SFAS 133 provides preferential treatment for derivatives used as effective hedging
instruments. Thus, I expect the effects of SFAS 133 on firms’ risk-management behavior to be different for firms that
successfully reduce their inherent risk after holding derivative instruments (effective hedgers, i.e., EH firms) than for firms
that fail to reduce their risk exposure after initiating derivatives programs (ineffective hedgers/speculators, i.e., IS firms).
I hypothesize that SFAS 133 has a major impact on IS firms. By definition, ineffective hedging or speculative positions
cannot offset firms’ business risks and thus will be reflected in companies’ contemporaneous earnings. This potentially
induces higher short-term earnings volatility, which is generally less preferable to shareholders or managers. First,
corporate debt covenants and performance pricing metrics usually include earnings as one of the crucial inputs. An
increase in earnings volatility is likely to directly increase the loan spread through performance pricing or increase the
likelihood of debt covenant violation, both of which are costly for the borrowing firm (e.g., Asquith et al., 2005; Beatty and
Weber, 2003). Similarly, the literature on managerial income smoothing suggests that managers prefer less volatile
earnings and may even sacrifice long-term value to smooth earnings because lower earnings volatility helps reduce the
firm’s perceived risk (Graham et al., 2005). Consistent with these arguments, Barton (2001) documents that firms use
hedging instruments and accrual management as partial substitutes to smooth earnings in order to increase managerial
compensation and reduce debt financing costs.9 If the benefit of engaging in ineffective hedging and/or speculative
activities does not change but the potential cost arising from earnings volatility increases after the adoption of SFAS 133, IS
firms are likely to engage in more prudent risk-management activities.
In addition to concerns about earnings volatility, increased reporting transparency is also likely to deter managerial risk-
taking behavior if ineffective hedging/speculative derivative positions do not increase shareholder value. Previous literature
provides limited evidence in support of this conjecture. For example, Geczy et al. (2007) find that frequent speculation is
associated with weaker firm-specific governance mechanisms. Adam and Fernando (2006) and Brown et al. (2006) find
that economic gains from selective hedging are insignificant. Chernenko and Faulkender (2007) suggest that the main
incentives for managers to time the interest rate market with swap contracts are to meet earnings forecasts and to raise
managerial pay. Adam et al. (2007) find that the most likely explanation for selective hedging in the gold mining industry is
that managers erroneously believe they can outperform the market.
Compared with firms that have speculative or ineffective hedge positions, firms that use derivative instruments to
successfully reduce their business risk face different decisions after the adoption of SFAS 133. Accounting practitioners and
7
SFAS 133 specifies type (c): a hedge of the foreign currency exposure of a net investment in a foreign operation, an unrecognized firm commitment,
an available-for-sale security, or a foreign currency-denominated forecasted transaction. In this situation, the accounting treatment still follows either the
fair value hedge or the cash flow hedge depending on the nature of the underlying hedged item.
8
Basis risk refers to the risk arising from the imperfect correlation between the price of the asset being hedged and the price of the asset underlying
the hedging instrument (see Haushalter, 2000). Basis risk exists if the derivatives market is incomplete or imperfect.
9
Barton (2001) documents a negative association between the derivatives’ notional amounts and discretionary accruals. This negative association
suggests that firms use hedging instruments and accrual management as partial substitutes to smooth earnings if derivative instruments are for hedging
purpose and hedging activities effectively reduce cash flow volatility.
ARTICLE IN PRESS
firms that utilize derivative instruments frequently criticize SFAS 133 for its complexity and high implementation cost.
Even if firms with effective hedge instruments are able to document such effectiveness on both an ex-ante and ex-post
basis, they may choose not to do so if the cost of doing so is too high. However, it is unclear whether EH firms will reduce
their hedge position after weighing the cost of the potential increase in earnings volatility due to changes in the fair value
of derivative instruments and the benefit derived from the hedging activity in the post-133 period.
Although risk-management activities are unobservable, previous literature suggests that these activities directly alter
firms’ exposure to risk (Wong, 2000; Guay, 1999). Thus, I estimate the risk exposure to interest rate, foreign exchange rate,
and commodity price for each sample firm for both pre-SFAS 133 and post-SFAS 133 period and infer the change in firms’
risk-management activities based on changes in the three types of risk exposure. Based on the above discussion, I first
hypothesize that IS firms engaged in more effective hedging (less speculative) activities after the adoption of SFAS 133.
H1. Ceteris paribus, (i) interest rate, (ii) foreign exchange rate, and (iii) commodity price risk exposures decrease
significantly for IS firms but remain unchanged for EH firms after the adoption of SFAS 133.
2.3. Impact of SFAS 133 on cash flow volatility and earnings volatility
My second and third hypotheses test the effect of SFAS 133 on firms’ cash flow volatility and earnings volatility, given
the changes in firms’ risk-management behavior.
Changes in cash flow volatility post-SFAS 133 reflect changes in firms’ ‘‘real’’ actions, including risk-management
activities. Changes in earnings volatility, however, will only reflect the pure accounting effect on the income statement after
controlling for the contemporaneous change in cash flow volatility. The analysis of cash flow volatility and earnings
volatility not only makes this research comparable with prior studies but provides additional evidence as to whether firms
changed their risk-management activities after the adoption of the standard.
Overall, I expect changes in cash flow volatility for EH and IS firms to follow the same pattern as changes in risk
exposure.
H2. Ceteris paribus, cash flow volatility decreases significantly for IS firms but remains unchanged for EH firms after the
adoption of SFAS 133.
With respect to earnings volatility, if IS firms anticipate the potential cost of increased volatility and adjust their
derivatives portfolios accordingly, I do not expect to observe any effect of SFAS 133 on a firm’s earnings volatility ex-post.
H3. Ceteris paribus, earnings volatility does not change after the adoption of SFAS 133 for either IS firms or EH firms.
3. Data, sample selection, and research design
In this section, I first describe the process used to identify new derivative users and classify new users as effective
hedgers (EH firms) or ineffective hedgers/speculators (IS firms). I then describe the empirical models used to test
the hypotheses.
3.1. Data and sample selection
To examine the different effects of SFAS 133 on effective hedgers and ineffective hedgers/speculators, I identify a sample
of new derivative users and then classify the new users as either effective hedgers (EH firms) or ineffective hedgers/
speculators (IS firms) based on the change in the given risk exposure after the initiation of the firm’s derivatives program.
Guay (1999) examines whether corporations use derivative instruments to reduce or increase their risks by examining
changes in several risk measures for new derivative users. He finds that firms’ exposures to interest rate risk and foreign
exchange rate risk decrease significantly after the initiation of a derivatives program and concludes that, on average, firms
use derivative instruments to hedge. Instead of focusing on the average change in the risk exposures for all derivative users,
I attempt to classify new users as either EH firms or IS firms based on the direction of the change in their risk exposures
relative to an expected level.
Following Guay (1999), I initially identify the sample of new users by conducting a keyword search of firms’ 10-K filings
in the Compact Disclosure database.10 Since most firms adopted SFAS 133 at the beginning of fiscal year 200111 and the
predecessor standard SFAS 119 became effective in December 1994,12 I conduct the keyword search from fiscal year 1995 to
10
The keyword search is conducted for all years from 1995 to 2001 and flags all companies in the Compact Disclosure database with the following
keywords in their annual reports: forward contract(s), currency exchange contract(s), foreign exchange contract(s), future(s) contract(s), option(s)
contract(s), swap(s), hedging instrument(s), hedge instrument(s), and derivative(s) instrument(s).
11
The original effective date of SFAS 133 was June 15, 1999. Faced with extensive requests from various business entities, FASB delayed the effective
date to June 15, 2000, but encouraged early adoption.
12
SFAS 119: ‘‘Disclosure about derivative financial instruments and fair value of financial instruments’’ required extensive disclosures about derivatives
positions and was superseded by SFAS 133.
ARTICLE IN PRESS
Table 1
Description of sample selection.
Panel A: Sample selection procedure

1995 1996 1997 1998 1999
# of firms in Compact Disclosure database 11,993 12,309 11,425 10,600 11,072

# of non-financial derivative users based on the keyword search 1,289 1,385 1,535 2,293 3,440
Non-financial new derivatives users based on keyword search – 98 151 626 948
Non-financial new derivative users after manual verification – 32 83 117 75
Non-financial new derivative users with available returns and accounting data – 23 60 82 60
Panel B: Distribution of new users by type of risk managed with derivative instruments
1996 1997 1998 1999 Total
Interest rate risk 10 31 52 36 129

Foreign exchange rate risk 10 24 30 23 87
Commodity price risk 6 10 18 11 45
Total 26 65 100 70 261
In panel A, I describe my sample selection procedure. Using the Compact Disclosure database, I conduct keyword search in 10-K filings to identify whether
a firm uses any derivative instruments for a given fiscal year. Keywords include forward contract(s), currency exchange contract(s), foreign exchange
contract(s), futures contract(s), option(s) contract(s), swap(s), hedging instrument(s), hedge instrument(s), and derivative(s) instrument(s). I define ‘‘new
users’’ in year t as non-financial firms that did not report any derivative positions from 1995 to year t1 but reported derivative positions from year t to
the adoption of SFAS 133. I then manually verify the accuracy of the keyword search results. The final sample of new users consists of 225 firms that
initiated their derivatives programs from 1996 to 1999.
In panel B, I describe the distribution of new uses by type of risk managed with derivative instruments. Since some firms report derivative use for more
than one type of risk, the total number of derivatives users by type of risk is more than 225.
fiscal year 2001 and then manually verify the year of the initiation of the derivatives program for each sample firm.
A firm is considered a new derivative user and included in my sample if it meets two conditions: (1) it started holding
derivative instruments between 1996 and 1999; and (2) it continued the derivatives program at least until the SFAS 133
adoption date.
In Table 1 panel A, I describe the sample selection process. From 1995 to 1997, the proportion of non-financial firms
whose 10-K filings contained at least one of the keywords increases steadily, from 11% in 1995 to 13% in 1997. In 1998 when
SFAS 133 was issued, this proportion increases to 22%. Many firms mentioned the issuance of the new accounting standard
in their 10-K filings even though they did not hold any derivative positions. Following the two sample selection criteria, I
identify 1823 non-financial new derivative users based on the keyword search. I then require that each firm have at least 4
years of data in the CRSP monthly files.13 This procedure reduces my sample to 1167 firms. To eliminate possible errors in
the keyword search process, I hand-collect the 10-K filings for the sample of 1167 firms and manually verify the existence
and initiation year of the derivative programs. My final sample consists of 225 new derivatives users that have the
necessary accounting data.14
In Table 1 panel B, I describe the distribution of new derivative users by type of risk associated with the derivative
instruments. Since firms commonly hold derivatives portfolios to manage interest rate risk, foreign currency exchange rate
risk, and/or commodity price risk, I concentrate my analysis on these three types of risk exposure. The number of new
derivative users increases from 23 in 1996 to 82 in 1998 and then decreases to 60 in 1999. The distribution of new users by
type of risk follows the same trend: steadily increasing from 1996 to 1998 and then decreasing in 1999. The breakdown of
new users by type of risk indicates that interest rate derivative instruments are the most commonly used instruments, with
129 out of the 225 (57%) new users holding interest rate-related derivatives. Foreign exchange rate and commodity price
derivatives are held by 87 (39%) and 45 (20%) new users, respectively. Of the 225 sample firms, 36 (16%) firms hold
derivatives in relation to two or more types of risk. The composition of my sample of new users is consistent with Guay
(1999), who identifies 254 new derivative users from 1991 to 1994, with 130 (51%) new users holding interest rate
derivatives, 85 (33%) new users holding foreign exchange rate derivatives, and 53 (21%) new users holding commodity
derivatives. The small decrease in the sample size relative to Guay (1999) is most likely a result of my sample selection
procedure, which requires a longer time series of data.15
13
Monthly stock returns data are used to estimate the risk exposures for each firm-period. I require at least 4 years of data because the latest new
users started their derivative programs in 1999. Thus, the period from at least 1 year prior to initiating the derivatives program to at least 1 year after
adopting SFAS 133 spans at least 4 years.
14
Of the 1167 firms, 225 qualify for my final sample. 494 firms have never held any derivative position; 160 firms have always had some derivative
position; 25 firms started, stopped, and then possibly restarted their derivative programs; 23 firms first used derivatives in 2000; 158 firms do not have
10-K filings available; and 82 firm are dropped due to missing accounting variables necessary for the multivariate regressions.
15
Guay’s (1999) sample selection procedure required 2 years of data (1 year before and 1 year after initiating the derivatives program), whereas I
require a firm to start, as well as to continue, the derivatives program until the adoption date of SFAS 133.
ARTICLE IN PRESS
Period 1 Period 2 Period 3
Initiating derivative programs Adopting SFAS 133
Fig. 1. Time line for each sample firm.
I also use keyword search to identify a sample of non-derivative users. A firm is categorized as a non-derivative user and
enters the control sample if the 10-K filings of that firm did not flag any keyword for at least 4 consecutive years including
at least 1 year after the adoption of SFAS 133. The final sample of non-users consists of 725 firms with data available on
CRSP and Compustat.
3.2. Research design
3.2.1. Classification of effective hedgers (EH firms) and ineffective hedgers/speculators (IS firms)
Fig. 1 describes the time line for each sample firm. I designate period 1 as the period from fiscal year 1995 to the year
prior to the initiation of the derivative program (pre-initiation period), period 2 as the period from the first year of
derivatives use to the year prior to the adoption of SFAS 133 (post-initiation and pre-133), and period 3 as the 4 years
following the adoption of SFAS 133 (post-133). Following the approach in previous research (Wong, 2000; Guay, 1999),
I define a firm’s interest rate, foreign exchange rate, and commodity price risk exposures as
Interest rate risk exposure: The absolute value of the estimated coefficient from a regression of the firm’s monthly stock
returns on the monthly percentage change in LIBOR.
Foreign currency exchange rate risk exposure: The absolute value of the estimated coefficient from a regression of the
firm’s monthly stock returns on the monthly percentage change in the Federal Reserve Board trade-weighted US dollar
index.
Commodity price risk exposure: The absolute value of the estimated coefficient from a regression of the firm’s monthly
stock returns on the monthly percentage change in the given commodity price.16
I use Model (1) to estimate the risk exposure for each firm-period. A minimum of 12 months of return data is required for
estimation purpose:
Ri;t ¼ a0i þ a1i Rmt þ a2i Macrot þ i;t ; (1)
where Rit is the Holding period return for firm i in month t; Rmt the value-weighted market portfolio return for month t;
Macrot the monthly percentage change in the related macro factor ((i) interest rate, (ii) foreign exchange rate, (iii)
commodity price) for month t.
The absolute value of the estimated coefficient a2i captures the risk exposure in relation to the macro factor for firm i in
each period.17
After estimating the three types of risk exposure for each firm-period, I examine the change in each type of exposure
between periods 1 and 2 to classify new derivative users as either EH firms or IS firms. The most challenging task for this
classification procedure to be meaningful is to model the expected change in risk exposures due to changes in firms’
underlying business risk. Relying on previous research findings and using data from period 1 only, I estimate the following
three cross-sectional regressions to quantify how firms’ risk exposures are affected by industry and firm characteristics
before the initiation of the derivatives program:
INT_EXPi ¼ gint 0 þ gint 1 EXP_INDi þ gint 2 RET_VOLi þ gint 3 BMi þ gint 4 SIZEi
þ gint 5 LEVERAGEi þ gint 6 ST_INVi þ i . (2)
FX_EXPi ¼ gfx 0 þ gfx 1 EXP_INDi þ gfx 2 RET_VOLi þ gfx 3 BMi þ gfx 4 SIZEi
(3)
þgfx 5 F_SALESi þ gfx 6 IMPORT_EXPORTi þ i :
COM_EXPi ¼ gcom 0 þ gcom 1 EXP_INDi þ gcom 2 RET_VOLi þ gcom 3 BMi þ gcom 4 SIZEi
þ gcom 5 CASHi þ gcom 6 INVENTORYi þ i . (4)
16
For each commodity derivatives user, I identify the specific commodity (e.g., oil, gas, wheat, gold) tied to the derivative instrument from the firm’s
10-K filings.
17
Following Guay (1999), I use absolute values instead of raw coefficients to measure the risk exposures so that these risk exposures can be
aggregated across firms and differenced across time periods.
ARTICLE IN PRESS
where INT_EXP, FX_EXP, and COM_EXP refer to interest rate risk exposure, foreign exchange rate risk exposure, and
commodity price risk exposure, respectively. I use two sets of explanatory variables to capture the determinants of firms’
risk exposures. The first set of variables includes EXP_IND, the average risk exposure of the non-derivative users within the
same industry (3-digit SIC code) during the same time period,18 RET_VOL, the standard derivation of daily stock returns,
BM, the book to market equity ratio, and SIZE, the log of the sum of market value of equity, total liabilities, and the
carrying value of preferred stock. These four variables directly measure industry and firm-level risk and are common for all
three regressions.
The second set of explanatory variables is specific to the type of risk being managed. Geczy et al. (2006) find
complementary relationships among cash holding, gas storage, and derivative usage for natural gas firms. To control for
other operating risk-management activities for commodity derivatives users, I include in model (4) CASH, defined as the
cash balance deflated by total assets, and INVENTORY, defined as total inventory divided by total assets times a dummy
variable that equals 1 if the firm’s commodity price risk exposure in period 1 (pre-initiation period) is positive and 1 if the
firm’s commodity price risk exposure in period 1 is negative. The dummy variable captures the nature of the underlying
business risk in relation to the commodity price. If the commodity price risk that a firm experiences is mainly on the
revenue side (cost side), then it is an increasing (decreasing) function of the inventory holdings. To estimate the
determinants of foreign exchange rate risk exposure, I follow Geczy et al. (1997) and include F_SALES (the amount of
foreign sales divided by the firm’s total revenue) and IMPORT_EXPORT (the sum of the industry imports and exports
divided by the industry’s GDP) to control for a firm’s dependence on the foreign exchange rate. Finally, I control for
LEVERAGE and ST_INV, short-term investments deflated by total assets, to control for the potential impact on interest rate
risk resulting from leverage and short-term investments.
I report the regression results for models (2)–(4) in Appendix A. Since these three models are estimated using data from
period 1 (pre-initiation period) only, I can reasonably quantify the determinants of firms’ risk exposures before any use of
derivatives.19 I then calculate the expected risk exposures for each sample firm in period 2 using estimated coefficients
from models (2)–(4) and the explanatory variables measured in period 2.20
I designate a firm as an IS firm (EH firm) if its risk exposure in period 2 (post-initiation period) is higher (lower) than the
expected level. More than 80% of the firms in my sample hold derivative positions associated with only one of the three
types of risk exposure. If a firm holds derivative positions associated with two types of risk, I classify it as an IS firm (EH
firm) if both types of risk exposure in period 2 are higher (lower) than the expected level. If a firm holds derivatives
associated with all three types of risks, I classify it as an IS firm (EH firm) if at least two types of risk exposure in period 2
are higher (lower) than the expected levels. Otherwise, a firm is classified as ‘‘neutral’’. This classification procedure results
in 125 effective hedgers (EH firms), 87 ineffective hedgers/speculators (IS firms), and 13 neutral firms.
3.2.2. Hypothesis tests

I use the following regression to test whether the three types of risk exposure change significantly from period 2 to
period 3:
EXPOSUREi;t ¼ b0 þ b1 F133 þ b2 CLASS þ b3 F133 CLASS þ CX þ i;t ; (5)
where EXPOSUREi,t is the risk exposure for firm i in period t; F133 the dummy variable that equals 0 for period 2 and 1 for
period 3; CLASS the dummy variable that equals 0 for EH firms and 1 for IS firms; X the set of control variables; C the vector
of coefficients for control variables.
Model (5) essentially measures changes in risk exposures between periods 2 and 3 with a time dummy. To attribute a
change in risk exposure to the adoption of SFAS 133, I still need to control for the changes in firms’ underlying business risk
and the potential change in firms’ incentives to either hedge or speculate.
I use all explanatory variables in models (2)–(4) to control for firms’ underlying business risk. In addition, I include NOL,
ROA, INT_COV, CFO_DELTA, and CFO_VEGA in model (5) to control for the potential change in firms’ incentives to use
derivative instruments. NOL is a dummy variable that equals 1 for firms that have a net operating loss carried forward and
positive net income, and 0 otherwise. Smith and Stulz (1985) and Graham and Smith (1999) contend that firms with convex
tax functions engage in hedging activities to reduce the potential tax liability. Smith and Stulz (1985) also argue that
financially distressed firms are more likely to engage in hedging activities to reduce bankruptcy cost. I include ROA, defined
as income before extraordinary items divided by the average total assets for a given year, and INT_COV, defined as EBIT
divided by interest expense, as measures of financial distress. Due to data constraints, there is very limited evidence on
18
For example, for a new user with fiscal year-end on March 31 that initiated the derivatives program in fiscal year 1998 and adopted SFAS 133 in
fiscal year 2001, period 1 (period 2) for non-derivative users from the same industry is defined as from April 1, 1995 to March 31, 1998 (April 1, 1998 to
March 31, 2001).
19
One concern about applying the estimated coefficients from one period to another is that there could be a structural shift across periods. Although
my setting partially alleviates this concern because new users started the derivatives programs in different fiscal years, I estimate models (2)–(4) using the
sample of non-users and conduct Chow test (See Green, 2003, p. 130) to test for potential structural changes. The test results suggest that the estimated
coefficients are not significantly different across different time periods and do not reject the null hypothesis that there is no structural changes across
periods.
20
I measure all accounting variables annually. If one period consists of multiple years for a sample firm, I use the mean values of the explanatory
variables for the firm-period in the regressions.
ARTICLE IN PRESS
how extensively firms speculate with derivative instruments. Even fewer studies document why firms speculate with
derivatives. Geczy et al. (2007) and Chernenko and Faulkender (2007) both find that the sensitivity of CFO compensation to
performance and the variability of the firm’s stock price partially explains firms’ market timing behavior. Therefore, I also
include CFO_DELTA and CFO_VEGA as control variables in model (5). The two variables are estimated using the empirical
method specified in Core and Guay (2002).
Since a new derivative user chooses to be a hedger or a speculator, potential sample selection may lead to biased
estimation results. Although sample selection bias in the setting studied here is less of a concern due to the difference-
in-difference research design,21 I specify a selection model and discuss the related regression results in the sensitivity
tests section.
To test the second and third hypothesis, I estimate regression equations (6) and (7) to test whether cash flow volatility
and earnings volatility experience significant changes after the adoption of SFAS 133:
CASH_VOLi;t ¼ b0 þ b1 F133 þ b2 CLASS þ b3 F133 CLASS þ g1 CASHVOL_INDi;t

þ g2 RET_VOLi;t þ g3 BMi;t þ g4 SIZEi;t þ g5 LEVERAGEi;t þ g6 PPEi;t
þ g7 ROAi;t þ g8 CFO_DELTAi;t þ g9 CFO_VEGAi;t þ i;t , (6)
EARN_VOLi;t ¼ b0 þ b1 F133 þ b2 CLASS þ b3 F133 CLASS þ g1 EARNVOL_INDi;t

þ g2 CASH_VOLi;t þ g3 RET_VOLi;t þ g4 BMi;t þ g5 SIZEi;t
þ g6 LEVERAGEi;t þ g7 PPEi;t þ g8 ROAi;t
þ g9 CFO_DELTAi;t þ g10 CFO_VEGAi;t þ i : (7)
where CASH_VOLi,t denotes quarterly operating cash flow volatility for sample firm i in period t. I measure volatility in two
ways. The first measure, CASH_VOL1i,t, is the standard deviation of the quarterly operating cash flows deflated by total
assets. The second measure, CASH_VOL2i,t, is the coefficient of variation of quarterly operating cash flows, calculated as the
standard deviation of quarterly operating cash flows divided by the absolute value of the mean operating cash flows for
firm i in period t. Similar to the measures of cash flow volatility, EARN_VOL1i,t is the standard deviation of quarterly income
before extraordinary items deflated by total assets. EARN_VOL2i,t is the coefficient of variation of quarterly income before
extraordinary items. To reduce the noise in calculating earnings volatility and cash flow volatility, I impose a minimum data
requirement of 8 quarters of non-missing values.
I use EARNVOL_IND and CASHVOL_IND, the average earnings volatility and cash flow volatility of the non-derivative
users, from the same industry (3-digit SIC code) to control for industry effects, and I expect the estimated coefficients on
both variables to be positive. I expect the estimated coefficient on RET_VOL, the annualized daily returns volatility, to be
positively correlated with earnings and cash flow volatility. The book-to-market ratio (BM) and SIZE proxy for risk factors
and are expected to be negatively correlated with earnings and cash flow volatility. I expect the estimated coefficient on
LEVERAGE to be negative because firms with a higher leverage ratio are expected to have relatively stable earnings and cash
flows, reducing the probability of financial distress. Defond and Hung (2003) argue that firms with higher capital intensity
rely on cash flows for routine maintenance. Therefore, I expect the coefficients on PPE, measured as property, plant, and
equipment deflated by total assets, to be negative. I use ROA to control for possible changes in firms’ performance resulting
from changes in economic conditions in the post-SFAS 133 period. Finally, I use CFO_DELTA and CFO_VEGA to control for
managerial risk-taking incentives.
In addition to estimating the above regressions, I collect the notional amount and net direction of the derivatives
positions for each sample firm. The information about the net direction of firms’ derivatives positions in period 2 is helpful
to check the construct validity of the classification procedure. Changes in the notional amount of firms’ derivative positions
in relation to commodity prices, foreign exchange rates, and interest rates between periods 2 and 3 provide additional
evidence about how firms adjust their derivative portfolios after the adoption of SFAS 133.
4. Empirical results
4.1. Descriptive statistics
In Table 2, I report the industry classification of the sample of new users, non-users, and the overall Compustat
database.22 The group of new users represents a similar industry distribution compared with the overall Compustat
database except for mining and services. A greater percentage of firms in the mining industry appears in the new user
sample than in the Compustat database, while firms in services exhibit a smaller percentage of new users relative to non-
21
Nikolaev and Lent (2005) argue that ‘‘The mere fact that some variable represents a decision to the firm is not in itself sufficient for ‘econometric
endogeneity’ to arise. Only if the factors that impact on the decision are also inter-related with the dependent variable will endogeneity exist.’’ This paper
focuses on the changes in risk exposures for EH firms and IS firms separately after the adoption of SFAS 133. Unless the underlying factor that affects a new
user’s incentive changes systematically and differently for EH firms and IS firms between periods 2 and 3, model (5) should be able to capture the impact
of SFAS 133.
22
The overall Compustat database includes all non-financial firms that have at least 4 years of data with at least 1 year after the adoption of SFAS 133.
ARTICLE IN PRESS
Table 2
Industry distribution for new users, non-users, and Compustat firms, respectively.
2-Digit SIC Industry title No. of Percentage No. of non- Percentage No. of Percentage
code new (%) users (%) Compustat (%)
users firms
01-09 Agriculture, forestry, fishing 3 1.3 6 0.7 27 0.45

10-14 Mining 36 15.9 46 6.3 311 5.21
15-17 Construction 2 0.9 3 0.1 78 1.31
20-34, 37-39 Manufacturing except machinery and 51 23.5 253 35.2 1,883 31.54
equipment
35-36 Industrial machinery and equipment 46 20.4 108 14.0 857 14.35
40-49 Transportation, communications, and 22 9.7 30 4.1 708 11.86
utilities
50-51 Wholesale trade 13 5.8 25 3.4 240 4.02
52-59 Retail trading 15 6.6 60 8.3 419 7.02
70-89 Services 36 15.9 184 25.4 1,356 22.71
91-99 Public administration 1 0.4 10 1.4 92 1.54
Total 225 100 725 100 5971 100
I define ‘‘new users’’ in year t as non-financial firms that did not report any derivative positions from 1995 to year t1 but reported derivative positions
from year t to the adoption of SFAS 133. I define ‘‘non-users’’ as non-financial firms that did not report any derivative positions for at least 4 consecutive
years including at least 1 year after the adoption of SFAS 133. To compare the industry distribution of ‘‘new users’’ and ‘‘non-users’’ with the industry
distribution of Compustat population, I report the industry distribution for all non-financial firms covered in Compustat that have at least 4 years of data
including at least 1 year after the adoption of SFAS 133.
Table 3
Median firm characteristics of ‘‘effective hedgers’’ (EH firms), ‘‘ineffective hedgers/speculators’’ (IS firms), and non-users for periods 1, 2, and 3,
respectively.
Firm characteristics (median) EH firms (N ¼ 125) IS firms (N ¼ 87) Non-users (N ¼ 725) EH vs. IS NON vs. EH NON vs. IS
MV of the firm ($MM)

Period 1 285 389 61 ** *** ***
Period 2 461 643 75 * *** ***
Period 3 603 728 80 *** ***
BM
Period 1 0.41 0.53 0.48 ** *
Period 2 0.49 0.65 0.59 ** **
Period 3 0.51 0.69 0.56 ** **
LEVERAGE (%)
Period 1 28.6 37.1 25.9 * ***
Period 2 36.1 49.7 35.1 ** ***
Period 3 44.1 47.9 38.9 * **
INT_COV
Period 1 8.71 7.54 4.11 * *** ***
Period 2 5.82 4.71 3.62 * *** **
Period 3 4.81 5.03 3.44 *** ***
ROA (%)
Period 1 5.08 4.88 1.67 *** ***
Period 2 3.46 3.68 0.98 *** ***
Period 3 2.95 3.07 0.97 *** ***
RET_VOL (%)
Period 1 3.56 2.68 4.18 ** *** ***
Period 2 3.87 3.41 4.60 * *** ***
Period 3 3.64 3.21 4.12 *** ***
ARTICLE IN PRESS
Table 3 (continued )
Firm characteristics (median) EH firms (N ¼ 125) IS firms (N ¼ 87) Non-users (N ¼ 725) EH vs. IS NON vs. EH NON vs. IS
EAR_VOL
Period 1 0.38 0.28 0.50 * *** ***
Period 2 0.40 0.41 0.64 *** ***
Period 3 0.42 0.36 0.60 *** ***
CASH_VOL
Period 1 0.82 0.76 0.99 ** ***
Period 2 0.87 0.95 1.06 *** ***
Period 3 0.94 0.82 0.91 ***
CFO_DELTA ($ thousand)
Period 1 11.72 10.81 2.44 * *** ***
Period 2 13.56 16.62 3.15 *** *** ***
Period 3 14.94 17.01 3.53 *** *** ***
CFO_VEGA ($ thousand)
Period 1 3.72 4.28 0.34 * *** ***
Period 2 4.06 5.66 0.81 *** *** ***
Period 3 4.24 5.85 1.02 *** *** ***
years including at least 1 year after the adoption of SFAS 133. I classify a new user as an EH firm (IS firm) if its risk exposures decreased (increased) relative
to the expected level after the initiation of the derivatives program. For a given ‘‘new user’’, I define period 1 as from fiscal year 1995 to the year prior to
the initiation of the derivatives program (pre-initiation period), period 2 as from the first year of derivatives use to the year prior to the adoption of SFAS
133 (post-initiation and pre-133), and period 3 as the 4 years following the adoption of SFAS 133 (post-133).
MV of the firm is the sum of the market value of equity, total liabilities, and the carrying value of preferred stock, calculated for each firm-year. BM is book
value of equity divided by market value of equity, calculated for each firm-year. LEVERAGE is total liabilities divided by the sum of total liabilities, the
market value of equity, and the carrying value of preferred stock, calculated for each firm-year. INT_COV is interest coverage ratio defined as income
before interest and taxes divided by interest expense, calculated for each firm-year. ROA is income before extraordinary items for a given year divided by
the average total assets for that year, calculated for each firm-year. RET_VOL is the standard deviation of daily stock returns, calculated for each firm-year.
EAR_VOL is the coefficient of variation of quarterly income before extraordinary items (based on a minimum of 8 quarters of data), calculated for each
firm-period. CASH_VOL is the coefficient of variation of quarterly operating cash flows (based on a minimum of 8 quarters of data), calculated for each
firm-period. CFO_DELTA is CFO pay-performance sensitivity, estimated for each firm-year following Core and Guay (2002). CFO_VEGA is CFO pay-risk
sensitivity, estimated for each firm-year following Core and Guay (2002). If one period consists of multiple years for a sample firm, I use the mean value of
the accounting variables across years in the analysis. ***, **, and * Wilcoxon test for difference in distributions at the 1%, 5%, and 10% level, respectively.
derivative users and the Compustat population. The industry distribution of non-users also corresponds to the overall
Compustat database except for utilities. In untabulated analyses, I also compare the industry distribution for EH firms with
IS firms. I find that there is no industry clustering for both groups and that EH firms and IS firms have similar industry
representation.
I present descriptive statistics for the sample of EH firms, IS firms, and non-users for each of the three periods in Table 3.
Cross-sample comparisons in Table 3 indicate that non-users differ significantly from new users in nearly all firm
characteristic, and that these differences are primarily driven by size. For instance, the median value of firms’ market value
in period 1 for EH firms, IS firms, and non-users is $285 million, $389 million, and $61 million, respectively. Consistent with
prior literature, the smaller size of non-users relative to new users suggests that economies of scale are positively
associated with the adoption of derivatives. Compared with new derivative users, non-users also have significantly lower
leverage ratios, lower interest coverage, lower profitability, higher volatility, and lower managerial compensation. However,
compared with previous literature (Singh, 2004; Guay and Kothari, 2003; Hentschel and Kothari, 2001; Guay, 1999), my
sample of new derivatives users consists of much smaller non-financial firms, probably because larger firms started using
derivatives long before 1996.
Although my main objective in this paper is not to document firms’ incentives to hedge or to speculate, further
comparison between EH firms and IS firms provides interesting insights. IS firms, as identified by my classification
procedure, are larger and have higher leverage and book-to-market ratios than EH firms. Further, IS firms experienced
relatively lower volatilities of returns, earnings, and cash flows across all three periods. However, statistical results show
that the increase in the volatility of cash flows between periods 1 and 2 is significantly higher (at the 5% level) for IS firms
than for EH firms, consistent with the notion that IS firms engaged in ineffective hedging or speculative activities in period
2. Changes in CFO pay-performance and pay-risk sensitivities between periods 1 and 2 follow the same pattern. To the
extent that firm size correlates with an information advantage, the overall descriptive statistics are consistent with Stulz
(1996), who argues that managers tend to take more risk when they have an information advantage and/or are encouraged
ARTICLE IN PRESS
Table 4
Simple correlations of key variables.
COM_EXP FX_EXP INT_EXP SIZE INT_COV BM LEV ROA EARN VOL CASH VOL RET_VOL CFO_DELTA CFO_VEGA
COM_EXP 0.51 0.15 0.24 0.09 0.18 0.06 0.22 0.15 0.08 0.47 0.04 0.03
FX_EXP 0.62 0.07 0.18 0.14 0.35 0.08 0.01 0.04 0.06 0.30 0.03 0.01
INT_EXP 0.35 0.29 0.14 0.03 0.01 0.05 0.01 0.09 0.11 0.36 0.05 0.02
SIZE 0.25 0.15 0.18 0.11 0.04 0.12 0.14 0.19 0.12 0.41 0.17 0.12
INT_COV 0.21 0.05 0.06 0.13 0.02 0.34 0.08 0.03 0.08 0.15 0.01 0.01
BM 0.09 0.06 0.07 0.29 0.02 0.14 0.24 0.02 0.03 0.16 0.08 0.10
LEV 0.19 0.08 0.04 0.12 0.55 0.45 0.35 0.14 0.17 0.15 0.06 0.05
ROA 0.20 0.03 0.10 0.16 0.29 0.26 0.45 0.35 0.10 0.06 0.04 0.03
EARN_VOL 0.13 0.05 0.18 0.22 0.01 0.13 0.05 0.29 0.34 0.30 0.01 0.01
CASH_VOL 0.18 0.04 0.12 0.13 0.13 0.14 0.24 0.15 0.35 0.38 0.02 0.01
RET_VOL 0.50 0.34 0.31 0.41 0.10 0.08 0.09 0.31 0.41 0.08 0.03 0.05
CFO_DELTA 0.06 0.05 0.06 0.18 0.01 0.09 0.08 0.07 0.02 0.01 0.06 0.57
CFO_VEGA 0.04 0.01 0.07 0.25 0.01 0.19 0.09 0.05 0.06 0.03 0.12 0.69
I report Pearman correlations above the diagonal and Spearman rank order correlations below the diagonal. Numbers in bold indicate at least 5% level of
significance. COM_EXP denotes commodity price risk exposure, estimated for each firm-period as the absolute value of the coefficient from a regression of
monthly stock returns on the monthly percentage change in the given commodity price. FX_EXP denotes foreign currency exchange rate risk exposure,
estimated for each firm-period as the absolute value of the coefficient from a regression of monthly stock returns on the monthly percentage change in the
Federal Reserve Board trade-weighted US dollar index. INT_EXP denotes interest rate risk exposure, estimated for each firm-period as the absolute value of
the coefficient from a regression of monthly stock returns on the monthly percentage change in LIBOR. SIZE is log of the sum of the market value of equity,
total liabilities, and the carrying value of preferred stock, calculated for each firm-year. INT_COV is interest coverage ratio defined as income before
interest and taxes divided by interest expense, calculated for each firm-year. BM is book value of equity divided by market value of equity, calculated for
each firm-year. LEV is total liabilities divided by the sum of total liabilities, the market value of equity, and the carrying value of preferred stock, calculated
for each firm-year. ROA is income before extraordinary items for a given year divided by the average total assets for that year, calculated for each firm-year.
EARN_VOL is the coefficient of variation of quarterly income before extraordinary items (based on a minimum of 8 quarters of data), calculated for each
firm-period. CASH_VOL is the coefficient of variation of quarterly operating cash flows (based on a minimum of 8 quarters of data), calculated for each
firm-period. RET_VOL is the standard deviation of daily stock returns, calculated for each firm-year. CFO_DELTA is CFO pay-performance sensitivity,
estimated for each firm-year following Core and Guay (2002). CFO_VEGA is CFO pay-risk sensitivity, estimated for each firm-year following Core and Guay
(2002).
to do so by their compensation schemes. Note that since the classification of IS firms and EH firms is by no means perfect,
differences in firm characteristics for the two groups should be interpreted with caution.
In Table 4, I report simple correlations of major variables used in the regression analysis. Pearson correlations are
presented above the diagonal and Spearman rank correlations are presented below the diagonal. Firms with higher risk
exposures are, in general, smaller, less profitable, and have more volatile returns, earnings, and cash flows. Firm size
significantly correlates with most variables, indicating that size is an important driving factor for firms’ risk, performance,
and managerial compensation structure.
4.2. Changes in the three types of risk exposure
In Table 5, I report the univariate analysis of changes in the three types of risk exposure across periods for EH firms and
IS firms. Note that, partially due to construction, all three types of risk exposure decrease (increase) for EH firms (IS firms)
from period 1 to period 2 relative to non-users from the same industry.23 In panel A, I describe the change in interest rate
risk exposure across periods. Between periods 1 and 2, 70 new users who initiated their derivative programs to manage
interest rate risk experienced a decrease in their interest rate exposure relative to the expected level and thus are classified
as EH firms. Similarly, 59 new users are classified as IS firms because their interest rate exposure increased relative to the
expected level. Consistent with my hypothesis, industry median-adjusted interest rate risk exposure declines significantly
for IS firms but not for EH firms after the adoption of SFAS 133. Specifically, the mean and median changes in interest rate
risk exposure for EH firms after SFAS 133 are 0.05 and 0.03, respectively, which is insignificant statistically as well as
economically. In contrast, the mean and median changes in the risk exposure for the IS firms after SFAS 133 are 0.35 and
0.19, respectively. Both are significantly negative at the 1% level.
In panels B and C of Table 5, I report the results of the univariate analysis for foreign exchange rate risk exposure and
commodity price risk exposure. The changes in these two types of risk exposure exhibit, on average, the same pattern as the
changes corresponding to the interest rate risk exposure. The mean change of commodity price risk exposure for IS firms is
insignificant after the adoption of SFAS 133, probably due to the small sample size. Since new users and non-users are quite
different in size, profitability, and firm risk characteristics, industry-adjusted changes in new users’ risk exposures alone do
not provide sufficient evidence on the impact of SFAS 133.
23
The potential concern of mean reversion is addressed in the sensitivity test section.
ARTICLE IN PRESS
Table 5
Univariate analysis of changes in interest rate risk exposure, foreign exchange rate risk exposure, and commodity price risk exposure across periods for
‘‘effective hedgers’’ (EH firms) and ‘‘ineffective hedgers/speculators’’ (IS firms), respectively.
Panel A: Changes in interest rate risk exposure
EH firms (N ¼ 70) IS firms (N ¼ 59)
Mean Median Mean Median
Period 1 1.41 0.93 0.72 0.37

Period 2 0.65 0.52 0.71 0.47
Period 3 0.48 0.47 0.26 0.20
Changes in interest rate risk exposure across periods adjusted for industry median
Period 2 vs. period 1 0.23* 0.16** 0.21** 0.13*
Period 3 vs. period 2 0.05 0.03 0.35*** 0.19***
Panel B: Changes in foreign exchange rate risk exposure

Period 1 3.73 2.42 1.73 1.05

Period 2 2.28 1.53 2.03 1.37
Period 3 2.51 1.25 1.42 0.87
Changes in foreign exchange rate risk exposure across periods adjusted for industry median
Period 2 vs. period 1 0.86*** 0.52** 0.53** 0.41**
Period 3 vs. period 2 0.31* 0.09 0.73*** 0.64***
Panel C: Changes in commodity price risk exposure

Period 1 1.47 1.39 0.50 0.34

Period 2 0.81 0.67 0.70 0.42
Period 3 0.57 0.43 0.51 0.16
Changes in commodity price risk exposure across periods adjusted for industry median
Period 2 vs. period 1 0.33** 0.46*** 0.37* 0.26**
Period 3 vs. period 2 0.09 0.17* 0.09 0.29**
years including at least 1 year after the adoption of SFAS 133. I classify a ‘‘new user’’ as an ‘‘effective hedger’’ (‘‘ineffective hedger/speculator’’) if its risk
exposures decreased (increased) relative to the expected level after the initiation of the derivatives program. For a given ‘‘new user’’, I define period 1 as
from fiscal year 1995 to the year prior to the initiation of the derivatives program (pre-initiation period), period 2 as from the first year of derivatives use
to the year prior to the adoption of SFAS 133 (post-initiation and pre-133), and period 3 as the 4 years following the adoption of SFAS 133 (post-133).
Changes in risk exposures for ‘‘EH firms’’ and ‘‘IS firms’’ across periods are adjusted for changes in median risk exposure of the non-users in the same
industry (3-digit SIC code).
I estimate interest rate risk exposure for each firm-period as the absolute value of the coefficient from a regression of monthly stock returns on the
monthly percentage change in LIBOR. I estimate the foreign exchange rate risk exposure for each firm-period as the absolute value of the coefficient from a
regression of monthly stock returns on the monthly percentage change in the Federal Reserve Board trade-weighted US dollar index. I estimate the
commodity price risk exposure for each firm-period as the absolute value of the coefficient from a regression of monthly stock returns on the monthly
percentage change in the given commodity price. ***, **, and * denote difference in means or medians at the 1%, 5%, and 10% level, respectively.
Table 6
Coefficients and p-values from regressions of interest rate risk exposure, foreign exchange rate risk exposure, and commodity price risk exposure on
dichotomous variable indicating the adoption of SFAS 133 and other industry and firm characteristics.
Predicted sign INT_EXP FX_EXP COM_EXP
Intercept 7 0.48 1.86 0.33

(0.00)*** (0.03)** (0.54)
F133 7 0.04 0.26 0.095
(0.24) (0.32) (0.27)
CLASS + 0.06 0.04 0.05
(0.32) (0.24) (0.26)
F133*CLASS 0.32 0.47 0.27
(0.00)*** (0.02)** (0.09)*
ARTICLE IN PRESS
Table 6 (continued )
EXP_IND + 0.24 0.17 0.32

(0.02)** (0.05)** (0.00)***
RET_VOL + 0.56 1.47 0.78
(0.00)*** (0.00)*** (0.06)*
BM 7 0.01 0.01 0.02
(0.55) (0.66) ( 0.12)
SIZE 0.11 0.13 0.08
(0.06)* (0.07)* (0.09)*
NOL 0.03 0.01 0.02
(0.46) (0.47) (0.48)
ROA 7 0.86 0.99 0.24
(0.13) (0.14) (0.59 )
INT_COV + 0.08 0.06 0.04
(0.22) (0.33 ) (0.69)
CFO_DELTA 7 0.04 0.07 0.12
(0.47) (0.12) (0.07)*
CFO_VEGA + 0.29 0.35 0.47
(0.01)*** (0.03)** (0.01)***
CASH 7 0.45
(0.57)
INVENTORY + 0.97
(0.02)**
F_SALES + 0.38
(0.04)**
IMPORT_EXPORT + 0.07
(0.23)
ST_INV 7 1.25
(0.05)***
LEVERAGE 7 0.22
(0.24)
Adjusted R2 33% 25% 65%

F-test: F133+F133*CLASS ¼ 0 (0.00)*** (0.00)*** (0.00)***
N 258 174 90
I estimate interest rate risk exposure (INT_EXP) for each firm-period as the absolute value of the coefficient from a regression of monthly stock returns on
the monthly percentage change in LIBOR. I estimate foreign exchange rate risk exposure (FX_EXP) for each firm-period as the absolute value of the
coefficient from a regression of monthly stock returns on the monthly percentage change in the Federal Reserve Board trade-weighted US dollar index. I
estimate commodity price risk exposure (COM_EXP) for each firm-period as the absolute value of the coefficient from a regression of monthly stock
returns on the monthly percentage change in the given commodity price.
F133 is dichotomous variable equal to 0 for period 2 (pre-133 period) and 1 for period 3 (post-133 period). CLASS is dichotomous variable equal to 0 for
‘‘effective hedgers’’ and 1 for ‘‘ineffective hedgers/speculators’’. I classify a ‘‘new user’’ as an ‘‘effective hedger’’ (‘‘ineffective hedger/speculator’’) if its risk
exposures decreased (increased) relative to the expected level after the initiation of the derivatives program. EXP_IND is median exposure to the risk at
hand of the non-derivative users in the same industry (3-digit SIC code), estimated for each firm-period. RET_VOL is the standard deviation of daily stock
returns, calculated for each firm-year. BM is book value of equity divided by market value of equity, calculated for each firm-year. SIZE is log of the sum of
the market value of equity, total liabilities, and the carrying value of preferred stock, calculated for each firm-year. NOL is dichotomous variable equal to 1
for firms with positive net income and positive net operating loss carry forward and 0 otherwise, calculated for each firm-year. ROA is income before
extraordinary items for a given year divided by the average total assets for that year, calculated for each firm-year. INT_COV is interest coverage ratio
defined as income before interest and taxes divided by interest expense, calculated for each firm-year. CFO_DELTA is CFO pay-performance sensitivity,
estimated for each firm-year following Core and Guay (2002). CFO_VEGA is CFO pay-risk sensitivity, estimated for each firm-year following Core and Guay
(2002). CASH is cash and short-term investments deflated by total assets, calculated for each firm-year. INVENTORY is total inventory deflated by total
assets times a dichotomous variable that equals 1 if the firm’s commodity price risk exposure in period 1 (pre-initiation period) is positive and 1 if the
firm’s commodity price risk exposure in period 1 is negative, calculated for each firm-year. F_SALES is the amount of foreign sales divided by total revenue,
calculated for each firm-year. IMPORT_EXPORT is sum of the industry’s imports and exports divided by industry GDP, calculated for each industry-year.
ST_INV is short-term investments deflated by total assets, calculated for each firm-year. LEVERAGE is total liabilities divided by the sum of total liabilities,
the market value of equity, and the carrying value of preferred stock, calculated for each firm-year.
The unit of analysis in Table 6 is firm-period. If one period consists of multiple years for a sample firm, I use the mean values of the accounting variables
for the firm-period in the regressions. p-Values in parentheses denote two-tailed tests. I cluster observations at the firm level to account for the potential
correlations of the residuals across the periods for the same sample firm.
In Table 6, I report multivariate regression results for the three types of risk exposures controlling for other relevant
factors that may affect new derivatives users’ business risk and their incentives to hedge. The unit of analysis in Table 6 is
firm-period. I measure all accounting variables annually. If period 2 or period 3 consists of multiple years for a particular
sample firm, I use the mean values of the accounting variables for the firm-period in the regressions. To account for the
potential correlations of the residuals across the periods for each same sample firm, I cluster observations at the firm level
in Table 6.
ARTICLE IN PRESS
The estimated coefficient on F133, representing the change in risk exposures between period 2 (pre-SFAS 133) and
period 3 (post-SFAS 133) for EH firms, is insignificant for all three types of risk exposure. However, the estimated coefficient
on the interaction term F133*CLASS, representing the difference between the change for EH firms and the change for IS
firms, is always negative and significant. The F-test indicates that changes in risk exposures for IS firms are also
significantly negative.
The estimated coefficients on the control variables are generally consistent with my predictions. Both the exposures of
the non-users from the same industry and return volatility are significantly positively associated with the dependent
variables. The estimated coefficient on size is significantly negative, suggesting that larger firms have lower risk exposures.
NOL, ROA, and interest coverage (INT_COV) measures firms’ incentives to hedge resulting from the convex tax function and
the probability of financial distress and therefore indirectly affect firms’ risk exposures. The pay-performance sensitivity
(Delta) of CFOs does not have a significant impact on the exposures to interest rate and foreign exchange rate risk. However,
the CFO Delta is significantly (p-value 0.07) negatively associated with the risk exposure resulting from commodity risk. In
contrast, the pay-risk sensitivity is highly significant and is positively associated with firms’ risk exposures.
For risk-specific control variables, the estimated coefficient on INVENTORY is significantly positive (0.97 with a p-value
of 0.02) in the regression relating to commodity price risk, suggesting that increased inventory increases (decreases) risk if
the risk is on the revenue (cost) side. The estimated coefficient on foreign sales is positively significant, suggesting a higher
magnitude of foreign sales increases firms’ foreign exchange rate risk exposure. Finally, the coefficient estimate on short-
term investments in the regression on interest rate risk exposure is significantly positive (1.25 with a p-value of 0.05),
indicating that stock returns of firms with higher short-term investments in securities are more sensitive to interest
rate changes.
In short, I find that all three types of risk exposure decrease significantly for IS firms after controlling for changes in
firms’ underlying business risk between periods 2 and 3. However, the change in risk exposures for EH firms is insignificant.
This implies that SFAS 133 has had a major impact on ineffective hedgers/speculators (IS firms).
4.3. Changes in cash flow volatility and earnings volatility
In Table 7, I report the regression results for models (6) and (7), examining the changes in cash flow volatility and
earnings volatility after the adoption of SFAS 133. Volatility of cash flows and earnings is measured in two ways, namely,
the coefficient of variation (COV) and the standard deviation (STD) of quarterly operating cash flows/earnings deflated by
total assets.24 The regression results for cash flow volatility indicate no significant change in volatility for EH firms after the
adoption of SFAS 133. However, the estimated coefficients on F133*CLASS are significantly negative with a p-value of 0.08
for the STD measure and a p-value of 0.09 for the COV measure. This finding indicates that cash flow volatility decreases
significantly for IS firms relative to EH firms after the adoption of SFAS 133 and provides additional evidence in support of
the findings with respect to the changes in risk exposures.
If cash flow volatility reflects the ‘‘real actions’’ in firms’ risk-management activities, changes in earnings volatility
capture the pure income statement effect of SFAS 133 after controlling for cash flow volatility. The regression results show
that, from period 2 to period 3, neither EH firms nor IS firms experienced significant changes in earnings volatility after
controlling for cash flow volatility. The estimated coefficients on the control variables are generally consistent with my
predictions. Both earnings volatility and cash flow volatility are positively correlated with returns volatility. Larger firms
and firms with higher leverage ratios tend to have lower earnings and cash flow volatility. I find a significantly negative
association between PPE and the COV measure of cash flow volatility, consistent with the argument in Defond and Hung
(2003) that firms with higher capital intensity may need stable cash flows for routine maintenance.
To summarize, the observed changes in risk exposures and cash flow volatility suggest that after the adoption of SFAS
133, IS firms engaged in more prudent risk-management activities to mitigate the potential cost of higher earnings
volatility imposed by the standard.
4.4. Speculating or ineffective hedging—evidence from changes in notional amount and direction of derivative positions after the
adoption of SFAS 133
A ‘‘hedge’’ position implies that a derivative position is held primarily to reduce risk. If firms can perfectly offset their
business risk by using derivative instruments, their risk exposures should be reduced to zero. However, due to the existence
of basis risk, many firms cannot completely eliminate certain risk even if they intend to do so.25 I classify new derivatives
24
The requirement of a minimum of 8 quarters of non-missing cash flow and earnings data further reduces the number of observations from (124 EH
firms+87 IS firms)*2 periods ¼ 424 firm-period combinations to 378 firm-period combinations when volatility is measured as the coefficient of variation
and 370 firm-period combinations when volatility is measured as the standard deviation of earnings (cash flow) deflated by total assets.
25
For example, Southwest Airlines uses derivative instruments to manage risk resulting from the price change of jet fuel. Since the derivatives market
for jet fuel is not sufficiently liquid, Southwest Airlines frequently uses derivative instruments based on crude oil and heating oil to hedge its risk exposure
in relation to the price volatility of jet fuel. Although the prices of crude oil, heating oil, and jet fuel are highly correlated, product, time, and location basis
risk make a perfect hedge impossible for Southwest Airlines even if the company seeks to eliminate its risk exposure (Carter et al., 2006, Fuel Hedging in
the Airline Industry: the Case of the US Airlines Industry).
ARTICLE IN PRESS
Table 7
Coefficients and p-values from regressions of cash flow volatility and earnings volatility on dichotomous variable indicating the adoption of SFAS 133 and
other industry and firm characteristics.
Predicted sign CASH_VOL1 CASH_VOL2 EARN_VOL1 EARN_VOL2
INTERCEPT 7 0.031 1.36 0.000 1.39

(0.00)*** (0.23) (0.99) (0.29)
F133 7 0.002 0.56 0.002 0.27
(0.25) (0.21) (0.38) (0.36)
CLASS + 0.004 0.38 0.002 0.09
(0.09)* (0.30) (0.19) (0.39)
F133*CLASS 0.008 1.05 0.003 0.41
(0.08)* (0.09)* (0.13) (0.13)
CASHVOL_IND + 0.321 0.21
(0.00)*** (0.00)***
EARNVOL_IND + 0.126 0.66
(0.00)*** (0.00)***
CASH_VOL + 0.111 0.04
(0.00)*** (0.16)
RET_VOL + 0.110 1.04 0.031 3.15
(0.06) (0.23) (0.00)*** (0.02)**
BM 7 0.001 0.03 0.001 0.04
(0.57) (0.30) (0.11) (0.13)
PPE 7 0.002 2.55 0.003 0.49
(0.46) (0.00)*** (0.89) (0.50)
SIZE 0.002 0.14 0.003 0.02
(0.00)*** (0.08)* (0.20) (0.38)
ROA 7 0.021 0.45 0.037 0.65
(0.09)* (0.68) (0.00) (0.70)
LEVERAGE 0.017 1.70 0.009 1.20
(0.01)*** (0.02) (0.00) (0.11)
CFO_DELTA 7 0.003 0.15 0.001 0.01
(0.47) (0.12) (0.26) (0.46)
CFO_VEGA + 0.001 0.58 0.000 0.16
(0.65) (0.10)* (0.48) (0.34)
Adjusted R2 34% 15% 34% 32%

F-test: F133+F133*CLASS ¼ 0 (0.31) (0.37) (0.59) (0.79)
N 370 378 370 378
CASH_VOL1 is the standard deviation of quarterly operating cash flows deflated by total assets (based on a minimum of 8 quarters of data), calculated for
each firm-period. CASH_VOL2 is the coefficient of variation of quarterly operating cash flows (based on a minimum of 8 quarters of data), calculated for
each firm-period. EARN_VOL1 is standard deviation of quarterly income before extraordinary items deflated by total assets (based on a minimum of 8
quarters of data), calculated for each firm-period. EARN_VOL2 is the coefficient of variation of quarterly income before extraordinary items (based on a
minimum of 8 quarters of data), calculated for each firm-period.
F133 is dichotomous variable equal to 0 for period 2 (pre-133 period) and 1 for period 3 (post-133 period). CLASS is dichotomous variable equal to 0 for
‘‘effective hedgers’’ and 1 for ‘‘ineffective hedgers/speculators’’. I classify a ‘‘new user’’ as an ‘‘effective hedger’’ (‘‘ineffective hedger/speculator’’) if its risk
exposures decreased (increased) relative to the expected level after the initiation of the derivatives program. CASHVOL_IND is the median industry-level
(3-digit SIC code) cash flow volatility of the non-derivative users. EARNVOL_IND is the median industry-level (3-digit SIC) earnings volatility of the non-
derivative users. RET_VOL is the standard deviation of daily stock returns, calculated for each firm-year. BM is book value of equity divided by market
value of equity, calculated for each firm-year. SIZE is log of the sum of the market value of equity, total liabilities, and the carrying value of preferred stock,
calculated for each firm-year. PPE is property, plant, and equipment deflated by total assets, calculated for each firm-year. ROA is income before
extraordinary items for a given year divided by the average total assets for that year, calculated for each firm-year. LEVERAGE is total liabilities divided by
the sum of total liabilities, the market value of equity, and the carrying value of preferred stock, calculated for each firm-year. CFO_DELTA is CFO pay-
performance sensitivity, estimated for each firm-year following Core and Guay (2002). CFO_VEGA is CFO pay-risk sensitivity, estimated for each firm-year
following Core and Guay (2002).
The unit of analysis in Table 7 is the firm-period. If one period consists of multiple years for a sample firm, I use the mean values of the accounting
variables for the firm-period in the regressions. My classification procedure identifies 125 EH firms and 87 IS firms. Since the calculation of cash flow and
earnings volatilities requires a minimum of 8 quarters of data, the numbers of observations reported in Table 7 are less than (125+87)*2 ¼ 424. p-Values in
parentheses denote two-tailed tests. I cluster observations at the firm level to account for the potential correlations of the residuals across the periods for
the same sample firm.
users as effective hedgers (EH firms) only if they successfully reduce their risk exposures after using derivative instruments.
For firms whose risk exposures increase after the initiation of the derivatives programs, it is not clear whether these firms
intend to take additional risk, or whether they intend to hedge the business risk but the hedge positions turn out to be
highly ineffective. Since firms’ inherent business risk without derivative positions is unobservable, this paper cannot
provide direct evidence to differentiate speculators from ineffective hedgers. However, examination of the notional amount
and directions of derivative positions for EH firms and IS firms after the adoption of SFAS 133 provides interesting insights.
ARTICLE IN PRESS
Table 8
Changes in notional amount and direction of derivatives positions after the adoption of SFAS 133 for ‘‘effective hedgers’’ (EH firms) and ‘‘ineffective
hedgers/speculators’’ (IS firms), respectively, by type of risk managed with derivative instruments.
Interest rate risk Foreign exchange rate risk Commodity price risk
EH firms IS firms EH firms IS firms EH firms IS firms
Panel A: Number of new users that hold derivative instruments in periods 2 and 3, respectively
Period 2 70 59 51 36 28 17
Period 3 67 50 46 27 27 17
Panel B: Notional amount (scaled by total assets)

Period 2 22.6% 23.5% 30.6% 36.7% 42.1% 43.5%
N ¼ 48 N ¼ 33 N ¼ 30 N ¼ 17 N ¼ 19 N ¼ 11
Period 3 24.8% 25.4% 33.8% 31.5% 45.6% 44.8%
N ¼ 48 N ¼ 33 N ¼ 30 N ¼ 17 N ¼ 19 N ¼ 11
Panel C: Number of firms that have derivative positions offsetting the inherent risk
Period 2 40 (65%) 25 (47%) 24 (60%) 14 (50%) 16 (73%) 9 (70%)
N ¼ 62 N ¼ 53 N ¼ 40 N ¼ 28 N ¼ 22 N ¼ 13
Period 3 37 (69%) 26 (63%) 19 (59%) 15 (71%) 13 (65%) 8 (73%)
N ¼ 54 N ¼ 41 N ¼ 32 N ¼ 21 N ¼ 20 N ¼ 11
from year t to the adoption of SFAS 133. For a given ‘‘new user’’, I define period 2 as from the first year of derivatives use to the year prior to the adoption of
SFAS 133 (post-initiation and pre-133), and period 3 as the 4 years following the adoption of SFAS 133 (post-133). I classify a ‘‘new user’’ as an EH firm (IS
firm) if its risk exposure decreased (increased) relative to the expected level after the initiation of the derivatives program.
In panel A, I report the number of EH firms and IS firms that hold derivative instruments in periods 2 and 3, respectively, by type of risk managed with
derivative instruments.
In panel B, I report the notional amount for EH firms and IS firms in periods 2 and 3, respectively, by type of risk managed with derivative instruments.
Notional amount is the stated amount of the firm’s derivative contracts scaled by total assets. If a period covers multiple fiscal years for a sample firm, I
use the mean value of the notional amount for the firm-period. I only include new users with positive identifiable notional amount in both periods 2 and 3
in panel B. I do not include commodity derivative users that only disclose the quantity of the underlying assets for their derivatives positions due to the
difficulty of obtaining accurate price data for the underlying assets.
In panel C, I report the number of EH firms and IS firms that have derivative positions offsetting the inherent risk for periods 2 and 3, respectively, by type
of risk. I use the sign of firms’ estimated risk exposure from period 1 as the proxy for the direction of firms’ inherent business risk. For example, if the
estimated risk exposure in period 1 for a commodity derivatives user is positive, then the offsetting derivatives position should be ‘‘short’’, i.e., the value of
the derivatives position is negatively correlated with the value of the underlying hedged item.
In panel A of Table 8, I describe the number of new users that still use derivative instruments after the adoption of
SFAS 133. Before the adoption of SFAS 133, 149 EH firms held derivative instruments to manage risk resulting from interest
rate, foreign exchange rate, and commodity price variations. This number decreases slightly to 140 after the adoption of
SFAS 133, representing a 6% reduction. However, the number of IS firms that use derivative instruments decreased from 112
to 94 after the adoption of SFAS 133, representing a 16% reduction. To the extent that ineffective hedgers are less likely to
improve hedging effectiveness by holding zero derivative positions, higher reduction rates in IS firms suggest that at least a
portion of IS firms reduced their speculative activities after the adoption of SFAS 133. In panel B, I report the notional
amount for EH firms and IS firms that hold derivative positions both in period 2 (pre-SFAS 133) and period 3 (post-SFAS
133). The magnitude of the notional amount for new derivative users increases slightly by 2% of total assets on average,
except for IS firms that use foreign exchange rate derivatives. In contrast to other new users, 8 out of 17 IS firms that use
derivatives to manage their foreign exchange rate risk report holding less notional amount after the adoption of SFAS 133
and the average reduction is around 5% of total assets for the 17 firms. The reduction in notional amount for some IS firms
again raises suspicion that some firms in the IS group may have engaged in speculative activities prior to the adoption of
SFAS 133.
In panel C of Table 8, I report the number of firms that have derivative positions offsetting their business risk. I use the
sign of firms’ risk exposure estimated in period 1 as a proxy for the direction of firms’ underlying business risk. Although a
firm’s underlying business risk could change over time, it is less likely to shift to the opposite direction. For example, the
inherent risk exposure in relation to energy prices for Southwest Airlines is usually negative because jet fuel has always
been a major cost item for the airline industry. Sixty-seven percent of IS firms hold offsetting derivative positions on
average after the adoption of SFAS 133, compared with 51% prior to the adoption of SFAS 133 (panel C). However, the change
in percentage for EH firms is less economically significant after the adoption of SFAS 133 (61% versus 65%).
Finally, changes in the notional amount and direction of derivative positions for IS firms occur only for firms that use
derivative instruments to manage interest rate risk and foreign exchange rate risk. I do not observe similar changes for IS
firms that use derivative instruments to manage commodity price risk. A potential explanation for this phenomenon is that
the direction of business risk exposure for firms that face commodity price risk is more transparent. Therefore, firms would
ARTICLE IN PRESS
face higher scrutiny if the direction of the derivative positions fails to offset the inherent business risk. The decrease in the
risk exposure for these firms is more likely due to the adjustment of the maturity, type, and underlying assets of the
derivative instruments.
5. Sensitivity tests
5.1. Test of mean reversion
I interpret the reduction in risk exposures for IS firms as evidence that SFAS 133 encouraged firms to engage in more
prudent risk-management activities. Another potential explanation for the significant change in firms’ risk exposures is
mean reversion. Note that, partly by construction, IS firms have higher risk exposures in period 2 than EH firms. Thus, if
mean reversion exists, one would expect to find the change in risk exposures between periods 2 and 3 to be negatively
correlated with the change between periods 1 and 2 for all firms. My findings reported in Table 5 are inconsistent with the
mean reversion explanation. Specifically, I do not find any significant increase in risk exposures for EH firms after SFAS 133.
To further rule out the mean reversion explanation, I change the cutoff point of periods 2 and 3 and re-do the analysis. I
delete all observations after the adoption of SFAS 133 and designate period 2A as from the year the derivative programs are
initiated to fiscal year 1999 and period 2B as fiscal year 2000. If the results are driven purely by mean reversion, the
reduction in risk exposure should not be sensitive to the period designation. I report the results of the analysis in Table 9.
Between periods 2A and 2B, both EH firms and IS firms experienced the same pattern of change in risk exposures as non-
users and the industry-adjusted changes are insignificant in general. Finally, I conduct the same univariate analysis with
the sample of non-users only and find no consistent pattern.
5.2. Effect of the Sarbanes-Oxley act
Another concern about the inference of my analysis is that the results may be driven by the impact of the Sarbanes-
Oxley Act (2002) rather than by SFAS 133. The Sarbanes-Oxley Act (2002) requires stronger corporate governance and
internal controls and thus may restrict ineffective hedging or speculative activities with derivatives if these activities
conflict with shareholders’ interests. To address this concern, I further divide period 3 into two sub-periods based on the
passage of the Sarbanes-Oxley Act. Period 3A is from the adoption date of SFAS 133 to June 30, 2002 and period 3B is from
Table 9
Robustness test of the effect of mean reversion.
Period 1 (median) Period 2A (median) Period 2B (median) Period 2A vs. 1 (Ind adj.) Period 2B vs. 2A (Ind adj.)
Interest rate risk exposure

EH firms (N ¼ 61) 0.99 0.38 0.79 (0.02)** (0.56)
IS firms (N ¼ 52) 0.43 0.49 0.75 (0.08* (0.28)
Non-users (N ¼ 485) 0.68 0.43 0.73
Foreign exchange rate risk exposure

EH firms (N ¼ 43) 2.52 1.33 2.89 (0.13) (0.42)
IS firms (N ¼ 29) 1.07 1.23 2.20 (0.09)* (0.13)
Non-users (N ¼ 307) 1.69 1.34 2.91
Commodity price risk exposure

EH firms (N ¼ 24) 1.12 0.49 0.69 (0.00)*** (0.25)
IS firms (N ¼ 15) 0.30 0.39 0.59 (0.05)** (0.23)
Non-users (N ¼ 98) 0.69 0.51 0.61
In Table 9, I report the median changes in the three types of risk exposure for EH firms and IS firms across periods 1, 2A, and 2B, respectively. I define ‘‘new
users’’ in year t as non-financial firms that did not report any derivative positions from 1995 to year t1 but reported derivative positions from year t to
the adoption of SFAS 133. I define ‘‘non-users’’ as non-financial firms that did not report any derivative positions for at least 4 consecutive years including
at least 1 year after the adoption of SFAS 133. For a given new users, I define period 1 as from fiscal year 1995 to the year prior to the initiation of the
derivatives program (pre-initiation period), period 2A as from the year the derivative programs are initiated to fiscal year 1999 and period 2B as fiscal year
2000. I classify a ‘‘new user’’ as an EH firm (IS firm) if its risk exposures decreased (increased) relative to the expected level after the initiation of the
derivatives program. I estimate interest rate risk exposure for each firm-period as the absolute value of the coefficient from a regression of monthly stock
returns on the monthly percentage change in LIBOR. I estimate foreign-exchange rate risk exposure for each firm-period as the absolute value of the
coefficient from a regression of monthly stock returns on the monthly percentage change in the Federal Reserve Board trade-weighted US dollar index. I
estimate the commodity-price risk exposure for each firm-period as the absolute value of the coefficient from a regression of monthly stock returns on the
monthly percentage change in the given commodity price. For each new derivative user, I estimate the median risk exposure of non-users from the same
industry (3-digit SIC code) across the same time period and use it as a benchmark. ***, **, and * denote difference in distributions at the 1%, 5%, and 10%
level, respectively.
ARTICLE IN PRESS
Table 10
Robustness test of the effect of the Sarbanes-Oxley Act.
Period 2 Period 3A Period 3B Period 3A vs. 2 Period 3B vs. 3A

(median) (median) (median) (Ind adj.) (Ind adj.)
Interest rate risk exposure

EH firms (N ¼ 64) 0.53 0.43 0.39 (0.69) (0.88)
IS firms (N ¼ 55) 0.41 0.19 0.15 (0.06)* (0.39)
Non-users (N ¼ 508) 0.59 0.42 0.36
Foreign exchange rate Risk exposure

EH firms (N ¼ 46) 1.44 1.26 1.12 (0.65) (0.50)
IS firms (N ¼ 32) 1.47 0.99 0.82 (0.01)*** (0.34)
Non-users (N ¼ 332) 1.69 1.43 1.34
Commodity price risk exposure

EH firms (N ¼ 27) 0.69 0.44 0.35 (0.21) (0.68)
IS firms (N ¼ 16) 0.46 0.20 0.16 (0.25) (0.12)
Non-users (N ¼ 106) 0.77 0.33 0.22
In Table 10, I report the median changes in the three types of risk exposure for EH firms and IS firms across periods 2, 3A, and 3B, respectively. I define
‘‘new users’’ in year t as non-financial firms that did not report any derivative positions from 1995 to year t1 but reported derivative positions from year t
to the adoption of SFAS 133. I define ‘‘non-users’’ as non-financial firms that did not report any derivative positions for at least 4 consecutive years
including at least 1 year after the adoption of SFAS 133. I classify a ‘‘new user’’ as an EH firm (IS firm) if its risk exposures decreased (increased) relative to
the expected level after the initiation of the derivatives program. For a given new users, I define period 2 as from the first year of derivatives use to the year
prior to the adoption of SFAS 133 (post-initiation and pre-133), period 3A as from the adoption date of SFAS 133 to June 30, 2002 and period 3B is from July
1, 2002 to the end of fiscal year 2004. I estimate interest-rate risk exposure for each firm-period as the absolute value of the coefficient from a regression
of monthly stock returns on the monthly percentage change in LIBOR. I estimate foreign-exchange rate risk exposure for each firm-period as the absolute
value of the coefficient from a regression of monthly stock returns on the monthly percentage change in the Federal Reserve Board trade-weighted US
dollar index. I estimate the commodity-price risk exposure for each firm-period as the absolute value of the coefficient from a regression of monthly stock
returns on the monthly percentage change in the given commodity price. For each new derivative user, I estimate the median risk exposure of non-users
from the same industry (3-digit SIC code) across the same time period and use it as a benchmark. ***, **, and * denote difference in distributions at the 1%,
5%, and 10% level, respectively.
July 1, 2002 to the end of fiscal year 2004. In Table 10, I report the univariate analysis of changes in risk exposures for
periods 2, 3A, and 3B. I find that the reduction in interest rate and foreign exchange rate risk exposures for IS firms occurs in
period 3A and the industry-adjusted change in risk exposures between periods 3A and 3B are not significant. However, I
find that the change in commodity price risk exposure is insignificant for both sub-periods. I conjecture that the
insignificance might be due to increased noise in the risk exposure measures when risk exposures are measured over a
shorter period. Overall, the evidence does not support the argument that the observed change in firms’ risk exposures is
due to the Sarbanes-Oxley Act.
5.3. Alternative classification procedures
When classifying new users as either EH firms or IS firms, I use models (2)–(4) to quantify the expected level of risk
exposures. To test for robustness, I also use the matched-firm approach for the classification procedure. For instance,
I match each new user with different non-users based on industry and size, on industry and the level of risk exposure in
period 1, on industry and the level of risk exposure in period 2, or on industry and the change in risk exposures between
periods 1 and 2. I classify a new user as an effective hedger (ineffective hedger/speculator) if its risk exposure decreases
(increases) between periods 1 and 2 relative to the matched firm. The disadvantage of the matched-firm approach is that it
relies heavily on non-derivative users from the same industry as benchmarks and ignores other factors that may affect a
firm’s business risk. Overall, various matched-firm approaches generate qualitatively similar empirical results.
In separate tests, I also use data from periods 1 and 2 and estimate regression equations (6) and (7). Consistent with the
univariate results reported in Table 3, I find that the cash flow volatility for IS firms increases significantly from period 1 to
period 2 relative to EH firms, which further confirms the validity of the classification procedure.
5.4. Selection bias
In this paper, I primarily focus on whether EH firms and IS firms changed their risk-management behavior after the
adoption of SFAS 133. However, firms choose to be effective hedgers or not and this underlying choice may potentially
affect the inference of the regression results of models (5)–(7). To address the potential concern of selection bias, I adopt
two-stage self-selection models. In the first stage, using information from period 2 only, I estimate the following probit
ARTICLE IN PRESS
model to estimate a new user’s propensity of being an IS firm:

CLASSi ¼ a1 þ a2 SIZEi þ a3 LEVERAGEi þ a4 BMi ; þa5 NOLi þ a6 CFO_VEGAi
þ a7 CFO_DELTAi þ ui : (8)
In the second stage, I add the selectivity correction (Lambdai) to the following regressions:
EXPOSUREi;t ¼ b0 þ b1 F133 þ b2 CLASSi þ b3 F133 CLASSi
þ b4 EXP_INDi;t þ b5 RET_VOLi;t þ b6 Sizei;t þ b7 ROAi;t þ b8 INT_COVi;t
þ b9 Lambdai þ ZX þ i;t (9)
^ ^ ^ ^
where Lambdaî ¼ fðCLASSi Þ=½FðCLASSi Þ for IS firms (Class ¼ 1) and Lambdai ¼ fðCLASSÞ=½1 FðCLASSÞ for EH firms
(Class ¼ 0). CLASSi is the fitted value from the first stage. f represents the density function of the normal distribution and
F represents the cumulative normal distribution function.
The independent variables in model (8) include variables that are identified in previous literature as incentives to
hedging/speculative activities. The independent variables in model (9) include variables that directly affect firms’ risk
exposures. For commodity price risk exposure, the set X in model (9) includes CASH and INVENTORY. For foreign exchange
rate risk exposure, the set X in model (9) includes F_SALES and IMPORT_EXPORT. For interest rate risk exposure, the set X in
model (9) includes LEVERAGE and ST_INV. Regression results show that the coefficient b9 in model (9) is insignificant,
suggesting selection bias is not a serious concern.
6. Conclusions
My objective in this paper is to examine whether firms changed their risk-management behavior after the adoption of
SFAS 133. Specifically, I examine changes in firms’ risk exposures (interest rate, foreign exchange rate, and commodity
price), cash flow volatility, and earnings volatility after the adoption of SFAS 133 for effective hedgers (EH firm) and
ineffective hedgers/speculators (IS firms) separately.
I designate new derivative users as either EH firms or IS firms based on the changes in risk exposures after the initiation
of the derivative program. I find that the interest rate risk exposure, foreign exchange rate risk exposure, and commodity
price risk exposure decrease significantly for the group of IS firms following the adoption of SFAS 133, after controlling for
potential changes in the underlying business risk. However, I do not find a significant change in risk exposures for EH firms.
The cash flow volatility for IS firms also declines significantly relative to that for EH firms after the implementation of SFAS
133, consistent with changes in risk exposures, whereas no such pattern is found for earnings volatility.
By sequentially examining changes in risk exposures, cash flow volatility, and earnings volatility, I disentangle the effect
of SFAS 133 on firms’ risk-management activities from the effect of SFAS 133 on firms’ earnings volatility controlling
for corporate risk-management behavior. The empirical evidence suggests that firms engaged in more prudent
risk-management behavior after the adoption of SFAS 133 and the impact of the standard on earnings volatility is limited
ex-post.
Appendix A
Coefficients and p-values from regressions of new users’ interest rate risk exposure, foreign exchange rate risk exposure,
and commodity price risk exposure on industry and firm characteristics, using data from period 1 only.
Intercept 7 0.52 1.98 0.19

(0.05)** (0.45) (0.88)
EXP_IND + 0.69 1.66 0.88
(0.00)*** (0.00)*** (0.00)***
RET_VOL + 1.04 1.57 0.89
(0.00)*** (0.00)*** (0.01)***
BM 0.10 0.89 0.37
(0.56) (0.47) (0.07)*
SIZE 0.13 0.26 0.23
(0.04)** (0.05)** (0.02)**
CASH 7 0.12
(0.32)
INVENTORY + 0.46
(0.03)**
F_SALES + 2.56
(0.06)*
IMPORT_EXPORT + 2.20
(0.11)
ST_INV 7 1.12
(0.09)*
ARTICLE IN PRESS
LEVERAGE + 0.17
(0.25)
2
Adjusted R 26.5% 36.6% 47.9%
N 129 87 45
from year t to the adoption of SFAS 133. For a given new user, I define period 1 as from fiscal year 1995 to the year prior to the initiation of the derivatives
program. I define ‘‘non-users’’ as non-financial firms that did not report any derivative positions for at least 4 consecutive years including at least 1 year
after the adoption of SFAS 133. INT_EXP denotes interest rate risk exposure, estimated for each firm-period, as the absolute value of the coefficient from a
regression of monthly stock returns on the monthly percentage change in LIBOR. FX_EXP denotes foreign exchange rate risk exposure, estimated for each
firm-period, as the absolute value of the coefficient from a regression of monthly stock returns on the monthly percentage change in the Federal Reserve
Board trade-weighted US dollar index. COM_EXP denotes commodity price risk exposure, estimated for each firm-period, as the absolute value of the
coefficient from a regression of monthly stock returns on the monthly percentage change in the given commodity price. EXP_IND is the median risk
exposure to interest rate, foreign exchange rate, and commodity price, respectively, of the non-derivative users in the same industry (3-digit SIC code),
estimated for each firm-period. RET_VOL is standard deviation of daily stock returns, calculated for each firm-year. BM is book value of equity divided by
market value of equity, calculated for each firm-year. SIZE is measured as log of the sum of the market value of equity, total liabilities, and the carrying
value of preferred stock, calculated for each firm-year. CASH is cash and short-term investments deflated by total assets, calculated for each firm-year.
INVENTORY is total inventory deflated by total assets times a dummy variable that equals 1 if the firm’s commodity price risk exposure in period 1 (pre-
initiation period) is positive and 1 if the firm’s commodity price risk exposure in period 1 is negative, calculated for each firm-year. F_SALES is the
amount of foreign sales divided by total revenue, calculated for each firm-year. IMPORT_EXPORT is sum of the industry’s imports and exports divided by
industry GDP, calculated for each industry-year. ST_INV is short-term investments deflated by total assets, calculated for each firm-year. If period 1
consists of multiple years for a sample firm, I use the mean values of the accounting variables for the firm-period in the regressions. p-Values in
parentheses denote two-tailed tests.
References
Adam, T., Fernando, C., 2006. Hedging, speculation, and shareholder value. Journal of Financial Economics 81, 283–309.
Adam, T., Fernando, C., Salas, J., 2007. Why do firms hedge selectively? Evidence from the Gold Mining Industry. Working Paper.
Asquith, P., Beatty, A., Weber, J., 2005. Performance pricing in bank debt contracts. Journal of Accounting and Economics 40, 101–128.
Austin, J., Gaver, J., Gaver, K., 1995. Additional evidence on bonus plans and income management. Journal of Accounting and Economics 19, 3–28.
Barth, M., Landsman, W., Wahlen, J., 1995. Fair value accounting: effects on banks’ earnings volatility, regulatory capital, and value of contractual cash
flows. Journal of Banking and Finance 19, 577–605.
Barton, J., 2001. Does the use of financial derivatives affect earnings management decisions. Accounting Review 76, 1–26.
Beatty, A., 2007. How does changing measurement change management behavior? A review of the evidence. Accounting & Business Research (Special
issue), 63–71.
Beatty, A., Weber, J., 2003. The effects of debt contracting on voluntary accounting method changes. Accounting Review 78, 119–142.
Brown, G., Crabb, P., Haushalter, D., 2006. Are firms successful at selective hedging? Journal of Business 79, 2925–2949.
Carter, D., Rogers, D., Simkins, B., 2006. Does fuel hedging make economic sense? The case of the US airline industry. Financial Management 35, 53–86.
Chernenko, S., Faulkender, M., 2007. Interest rate speculation and managerial compensation? University of Maryland. Working Paper.
Core, J., Guay, W., 2002. Estimating the value of employee stock option portfolios and their sensitivities to price and volatility. Journal of Accounting
Research 40, 613–630.
Danthine, J.P., 1978. Information, futures prices, and stabilizing speculation. Journal of Economic Theory 17, 79–98.
DeFond, M., Hung, M., 2003. An empirical analysis of analysts’ cash flow forecasts’’. Journal of Accounting and Economics 35, 73–100.
DeFond, M., Park, C., 1997. Smoothing income in anticipation of future earnings. Journal of Accounting and Economics 23, 115–139.
Geczy, C., Minton, B., Schrand, C., 1997. Why firms use currency derivatives? Journal of Finance 52, 1323–1354.
Geczy, C., Minton, B., Schrand, C., 2006. The use of multiple risk management strategies: evidence from the natural gas industry. Journal of Risk 8, 19–54.
Geczy, C., Minton, B., Schrand, C., 2007. Taking a view: corporate speculation, governance and compensation. Journal of Finance 62, 2405–2443.
Graham, J., Smith, C., 1999. Tax incentives to Hedge. Journal of Finance 54, 2241–2262.
Graham, J., Harvey, C., Rajgopal, S., 2005. The economic implications of corporate financial reporting. Journal of Accounting and Economics 40, 3–73.
Green, W., 2003. Econometric Analysis, fifth ed., Prentice Hall Publishing.
Guay, W., 1999. The impact of derivatives on firm risk: an empirical examination of new derivative users. Journal of Accounting and Economics 26,
319–351.
Guay, W., Kothari, S.P., 2003. How much do firms Hedge with derivatives? Journal of Financial Economics 70, 423–461.
Haushalter, D., 2000. Financing policy, basis risk, and corporate hedging: evidence from oil and gas producers. Journal of Finance 55, 107–152.
Hentschel, L., Kothari, S.P., 2001. Are corporations reducing or taking risks with derivatives? Journal of Financial and Quantitative Analysis 36, 65–85.
Hodder, L., Hopkins, P., Wahlen, J., 2006. Risk-relevance of fair-value income measures for commercial banks. Accounting Review 81, 337–375.
Kanodia, C., Mukherji, A., Sapra, H., Venugopalan, R., 2000. Hedge disclosures, future prices, and production distortions. Journal of Accounting Research 38,
53–82.
Melumad, N., Weyns, G., Ziv, A., 1999. Comparing alternative Hedge accounting standards: shareholders perspective. Review of Accounting Studies 4,
265–292.
Nikolaev, V., Lent, L.V., 2005. The endogeneity bias in the relation between cost-of-debt capital and corporate disclosure policy. European Accounting
Review 14, 677–724.
Park, J., 2004. Economic consequences and financial statement effects of SFAS No. 133 in bank holding companies. University of Wisconsin at Madison
Dissertation.
Sapra, H., 2002. Do mandatory Hedge disclosures discourage or encourage excessive speculation? Journal of Accounting Research 40, 933–964.
Singh, A., 2004. The effects of SFAS 133 on the corporate use of derivatives, volatility, and earnings management. Pennsylvania State University
Dissertation.
Smith, C., Stulz, R., 1985. The determinants of firms’ hedging policies. Journal of Financial and Quantitative Analysis 20, 391–405.
Stulz, R., 1996. Rethinking risk management. Journal of Applied Corporate Finance, 8–24.
Subramanyam, K.R., 1996. The pricing of discretionary accruals. Journal of Accounting and Economics 22, 249–281.
Wong, F., 2000. The association between SFAS No. 119 derivatives disclosures and the foreign exchange risk exposure of manufacturing firms. Journal of
Accounting Research 38, 387–417.

RPS 11 Poin 1 Zhang2009

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

RPS 11 Poin 1 Zhang2009

Uploaded by

Copyright:

Available Formats

ARTICLE IN PRESS

Journal of Accounting and Economics 47 (2009) 244–264

Contents lists available at ScienceDirect

Journal of Accounting and Economics

Effect of derivative accounting rules on corporate risk-management

2.1. Accounting treatment for derivative instruments

2.2. The impact of SFAS 133 on ﬁrms’ risk-management activities

3. Data, sample selection, and research design

3.1. Data and sample selection

Panel A: Sample selection procedure

# of ﬁrms in Compact Disclosure database 11,993 12,309 11,425 10,600 11,072

Interest rate risk 10 31 52 36 129

Period 1 Period 2 Period 3

Initiating derivative programs Adopting SFAS 133

Fig. 1. Time line for each sample ﬁrm.

3.2. Research design

3.2.2. Hypothesis tests

CASH_VOLi;t ¼ b0 þ b1 F133 þ b2 CLASS þ b3 F133 CLASS þ g1 CASHVOL_INDi;t

EARN_VOLi;t ¼ b0 þ b1 F133 þ b2 CLASS þ b3 F133 CLASS þ g1 EARNVOL_INDi;t

4.1. Descriptive statistics

01-09 Agriculture, forestry, ﬁshing 3 1.3 6 0.7 27 0.45

Total 225 100 725 100 5971 100

MV of the ﬁrm ($MM)

4.2. Changes in the three types of risk exposure

Panel A: Changes in interest rate risk exposure

EH ﬁrms (N ¼ 70) IS ﬁrms (N ¼ 59)

Mean Median Mean Median

Period 1 1.41 0.93 0.72 0.37

Panel B: Changes in foreign exchange rate risk exposure

Mean Median Mean Median

Period 1 3.73 2.42 1.73 1.05

Panel C: Changes in commodity price risk exposure

Mean Median Mean Median

Period 1 1.47 1.39 0.50 0.34

Predicted sign INT_EXP FX_EXP COM_EXP

Intercept 7 0.48 1.86 0.33

Predicted sign INT_EXP FX_EXP COM_EXP

EXP_IND + 0.24 0.17 0.32

Adjusted R2 33% 25% 65%

4.3. Changes in cash ﬂow volatility and earnings volatility

Predicted sign CASH_VOL1 CASH_VOL2 EARN_VOL1 EARN_VOL2

INTERCEPT 7 0.031 1.36 0.000 1.39

Adjusted R2 34% 15% 34% 32%

EH firms IS firms EH firms IS firms EH firms IS firms

Panel B: Notional amount (scaled by total assets)

5.1. Test of mean reversion

5.2. Effect of the Sarbanes-Oxley act

Interest rate risk exposure

Foreign exchange rate risk exposure

Commodity price risk exposure

Period 2 Period 3A Period 3B Period 3A vs. 2 Period 3B vs. 3A

Interest rate risk exposure

Foreign exchange rate Risk exposure

Commodity price risk exposure

5.3. Alternative classiﬁcation procedures

5.4. Selection bias

model to estimate a new user’s propensity of being an IS ﬁrm:

Predicted sign INT_EXP FX_EXP COM_EXP

Intercept 7 0.52 1.98 0.19

You might also like