Journal of Futures Markets - 2022 - Dudley

Received: 21 December 2021 | Accepted: 1 April 2022
DOI: 10.1002/fut.22336
RESEARCH ARTICLE
How do firms hedge in financial distress?
Evan Dudley1 | Niclas Andrén2 | Håkan Jankensgård2

1
Smith School of Business, Queen's University, Kingston, Ontario, Canada
2
Department of Business Administration and Knut Wicksell Centre for Financial Studies, Lund University School of Economics and Management,
Lund University, Lund, Sweden
Correspondence
Niclas Andrén, Department of Business Abstract
Administration and Knut Wicksell Centre We examine how firms hedge in financial distress. Using hand‐collected data
for Financial Studies, Lund University
School of Economics and Management,
from oil and gas producers, we find that these firms hedge oil prices during
Lund University, P.O. Box 7080, 220 07 periods of financial distress. Derivative portfolios in these firms are
Lund, Sweden. characterized by short put options. These positions are part of a composite
Email: niclas.andren@fek.lu.se
three‐way (3W) collar strategy that combines buying put options and selling
Funding information put and call options with differing strike prices. Because liquidity demand
Jan Wallanders och Tom Hedelius Stiftelse varies with the degree of financial distress, the 3W collar strategy preserves
samt Tore Browaldhs Stiftelse
incentives for future growth.
KEYWORDS
corporate hedging, economic distress, financial distress, risk management
JEL CLASSIFICATION
G30, G32
1 | INTRODUCTION
The question of whether the use of financial derivatives increases or decreases in response to a deteriorating financial
condition has attracted significant attention in the literature on corporate hedging. In these circumstances, hedging is
valuable, potentially ensuring the survival of the firm. Financially weak firms thus have an apparent motive to mitigate
their risks and protect against further declines in performance. Yet at the same time, they may find risk management
exceedingly costly because hedging positions need to be cash‐financed or supported by collateral (Adam, 2002; Mello &
Parsons, 1999; Rampini et al., 2014). Recent empirical evidence lends some support to the notion that hedging intensity
diminishes as financial health deteriorates (Rampini et al., 2014). What stands out, however, is the overall paucity of
evidence on hedging in financial distress in the literature on corporate risk management. To shed light on this issue,
this paper investigates the types of hedging strategies firms employ in financial distress.
There is considerable variation in terms of how derivative portfolios are composed (Adam, 2009; Croci et al., 2017).
Adam (2002) develops a theory of hedging according to which the optimal composition of derivatives is a function of
the firm's financial health. Financially healthier firms concentrate on protecting against future bad states by purchasing
put options financed with cash‐on‐hand (a convex payoff). Distressed firms, in contrast, hold a derivative portfolio that
focuses on generating liquidity for current investment by selling call options (a concave payoff). Selling calls is costly
This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License, which permits use and distribution in any
medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made.
© 2022 The Authors. The Journal of Futures Markets published by Wiley Periodicals LLC.
1324 | wileyonlinelibrary.com/journal/fut J Futures Markets. 2022;42:1324–1351.

10969934, 2022, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/fut.22336 by Nat Prov Indonesia, Wiley Online Library on [20/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
DUDLEY ET AL. | 1325
because doing so causes underinvestment in good states of the world. In Adam (2002), the heavy discount rate applied
to future payoffs for distressed firms resolves this trade‐off in favor of selling calls.
In this paper, we examine empirically how firms manage the tradeoff between hedging to avoid bankruptcy,
minimizing collateral demands by hedging counterparties, and maintaining incentives for future investment. Historical
conditions in the oil and gas industry provide an interesting laboratory to investigate corporate hedging policies during
financial distress. The oil and gas markets are notorious for their high volatility, and these markets have seen repeated
price crashes. The advantages of the oil and gas industry for studying hedging behavior are also well‐known: firms in
this industry have economically meaningful exposures, extensive disclosures about their hedging policies, and exhibit
significant variation in hedge ratios (Bakke et al., 2016; Jin & Jorion, 2006).
The years 1986, 1998, 2008, 2014, and 2020 all saw dramatic falls in the oil price. In all cases, these episodes were
relatively short‐lived. From the vantage point of shareholders, these shocks generate difficult‐to‐predict episodes of
financial distress of uncertain length. To examine how firms hedge in distress we examine the sudden and unexpected
collapse in the oil price that occurred in the fourth quarter of 2014. Within the space of 10 weeks the oil price
essentially halved after a prolonged period of high and stable oil prices. This price shock abruptly sent large parts of the
oil and gas industry into severe distress. We use hand‐collected data on derivative portfolios from the first quarter of
2013 through the fourth quarter of 2015.
We find, consistent with Adam (2002), that the use of cash‐financed put options used to create insurance‐like
hedging strategies is inversely related to financial distress. Firms instead sell options in addition to buying put options
as distress worsens. We document this finding in both the time‐series and cross‐section dimensions. Following the
fourth quarter of 2014, the frequency of such composite option‐based hedging strategies increased, and this effect was
strongest in financially distressed firms. We confirm in a multivariate setting that strategies involving both long and
short option positions increase in likelihood as the firm's distance‐to‐default declines, that is, as the likelihood of
financial distress increases, controlling for known determinants of corporate hedging policy.
While our empirical results support Adam's prediction that firms closer to distress sell options, a closer look reveals
that the distinguishing characteristic is the selling of put options, not call options as theorized by Adam (2002). Selling
put options is surprisingly common. In our sample, 25.4% of firm‐quarters contain sold puts (conditional on the firm
being a derivative user).
For oil and gas producers, such a derivative position taken in isolation increases risk: the firm agrees to make good
on a financial claim that gets larger the further the oil price falls. But sold put options never appear in isolation.
Instead, they are part of a composite strategy referred to as a “three‐way collar” (3W‐collar), which also contains a
position in bought put options and another in sold call options. The strike price on the bought put options is always
higher than on the sold puts, which means that the strategy offers protection against falling prices, albeit not to the
same extent as cash‐financed put options.
Under hedge accounting rules, oil and gas producers are allowed to report hedging activity on their financial
statements as either cash‐flow hedges or fair‐value hedges.1 All of the 3W‐collar users in our sample report derivatives
hedges as fair value hedges. If the 3W‐collar is an effective hedging strategy, it should be reported as a net derivative
asset rather than a liability by these firms when oil prices are low. We examine quarterly reports of oil and gas firms
filed after the oil‐price collapse. Under fair‐value hedge accounting rules, these firms report the 3W‐collar as a net asset
in just under 90% of firm‐years quarters, suggesting that the 3W‐collar offers protection against a decline in output
prices as intended.
For the 3W‐collar strategy to be effective, the firm's demand for liquidity as a function of output prices (i.e., the
price of oil) should be tiered. To see why, consider firms that fail to achieve profitability at current oil prices. Firms in
such a state no longer require large‐scale funding of investment but may still value the option to continue operations.
Consequently, their demand for external funds decreases as these firms reduce investment programs that are now
unprofitable. Composite strategies that only involve bought puts and sold calls generate excess liquidity when oil prices
are low. For financially constrained firms, this excess liquidity comes at the price of underinvestment when the price of
oil is high.
An additional factor explaining cross‐sectional variation in hedging strategy is the firm's own cash holdings. In
contrast to the 3W‐collar strategy, insurance‐based strategies are cash‐intensive and not self‐financed. The insurance
strategy becomes increasingly costly as the firm nears financial distress, and we expect this strategy to become less
1
FASB 133 describes eligibility rules for hedge accounting. See https://www.fasb.org/st/summary/stsum133.shtml for a summary.
1326 | DUDLEY ET AL.
prevalent as the firm's internal resources fall. Consistent with this explanation, 3W‐collar hedging becomes more
sensitive to the firm's distance to default as the firm's cash balances decline. In contrast, the sensitivity of insurance‐
type hedging to the firm's distance‐to‐default declines with cash balances. These contrasting results suggest that the
choice of corporate hedging strategy is also a function of the firm's own cash holdings.
In further cross‐sectional tests, we examine how corporate hedging strategies vary with the firm's own production
costs. Firms with higher production costs are more likely to cut back on investment following an oil‐price shock, which
introduces a kink in their demand for funds. Consistent with this explanation, we find that 3W‐collar usage is more
prevalent in firms that are economically distressed due to structurally higher costs, which necessitate higher oil prices
to sustain investment programs. In contrast, we do not find a significant link between economic distress and the usage
of other types of hedging strategies that involve either linear contracts such as forwards or insurance strategies that
involve bought put options.
We contribute to the literature on corporate risk management and the use of derivatives. First, we describe
corporate hedging behavior during financial distress. Whether and how firms hedge in distress is currently an open
question in the literature on corporate risk management (Almeida et al., 2020). We empirically document the use of
option‐based hedging strategies for oil and gas firms experiencing financial distress. Second, we describe a novel and
innovative hedging strategy used by large oil and gas producers: the 3W‐collar. This hedging strategy involves aspects of
collateral management, liquidity provision in financial distress, and the management of incentives for future
investment. Third, we shed light on the nature of liquidity demand during distress. Differences in liquidity demand
between bankruptcy and financial distress have been examined theoretically (Purnanandam, 2008), but the empirical
implications about the use of option‐based versus forward‐based hedging strategies have not been tested. We contribute
to this question by finding strong empirical support for option‐based rationales for hedging in financial distress.
2 | HYPOTHESIS DEVELOPMENT
Froot et al. (1993, 1994) argue that risk management primarily creates value by coordinating investment and financing
in financially constrained firms. Their model predicts that firms buy put options when investments or financial
constraints are convex in the hedgable risk factor, for example, when there is a threshold below which firms cannot
further cut investments.
Adam (2002) examines firms' use of option‐based hedging strategies to coordinate investment and financing
activities. In this model, the firm trades off its need to finance current investment against the risk of underinvesting in a
future period. By selling call options a firm can generate a cash inflow that can be used toward financing capital
expenditure. However, the sold calls exacerbate the underinvestment problem in future states in which the market
price exceeds the strike on the options because the negative payoff on the options offsets the payoff on the firm's
investments. This trade‐off is resolved by the cost wedge between internal and external financing. When this cost
wedge is small, the firm buys put options to insure against the risk of underinvestment in future periods in which it
risks a cash shortfall. When this cost wedge is high firms sell call options to alleviate underinvestment in the current
period. Both Froot et al. (1993) and Adam (2002) suggest that a value‐maximizing risk management strategy be tiered
relative to the hedged risk factor (e.g., oil). Table 1 provides a more comprehensive summary of the literature on
theories of corporate hedging.
In practice however, we observe that firms use more complex hedging strategies that involve combinations of sold
calls and bought and sold puts with different strike prices. Anecdotal evidence documents the popularity of 3W collars
in smaller and (possibly) financially distressed oil and gas producers.2 The multiplicity of strike prices, combined with
several different permutations of long/short and calls/puts suggests that firms have a more complex and tiered demand
for liquidity over a broad range of output prices than described in the works cited above.
How could such a tiered demand for liquidity come about? This is illustrated in Figure 1. The first kink in the firm's
liquidity demand curve occurs if there is a set of core investments that are economical above a certain price, denoted
X1. Core investments have a high marginal productivity and are vital to the firm's future viability. These investments
are comprised of maintenance capital expenditures needed to sustain output capacity and offset economic depreciation
2
See https://seekingalpha.com/article/3986405-are-3-way-collars-choking-your-oil-company and https://www.mercatusenergy.com/blog/bid/39927/
hedging-oil-gas-with-three-way-collars for fact‐based anecdotal evidence on the use of three‐way collars.
DUDLEY
ET AL.
TABLE 1 Summary of theoretical predictions.

Hedging strategy
Linear contracts Bought put options Sold call options Sold put options Bought call options
Deadweight costs of distress and Minimize variance to reduce risk of Protect downside while
bankruptcy default (Smith & Stulz, 1985) maintaining upside
(Stulz, 1996)
Coordinate investment and Coordinate investment and Coordinate investment and Finance put options Coordination when
financing policies under financing when hedge ratio is financing state‐by‐state under financial investment outgrows
financial constraints constant (Froot et al., 1993) (Froot et al., 1993) constraints cash flow (Froot
(Adam, 2002) et al., 1993)
Mitigate underinvestment problem Minimize variance to reduce risk of
default (Bessembinder, 1991)
Risk shifting Increase firm's
risk taking
Mitigate free cash flow problem Minimize variance to reduce free Reduce free cash Reduce free cash
cash flow (Morellec & flow in good states flow in bad
Smith, 2007) states
Note: This table summarizes theoretical predictions for different hedging strategies.
| 1327
F I G U R E 1 Tiered demand for liquidity. The demand for

liquidity for a distressed firm is shown. In region A, the
distressed firm cuts back on investment and hoards cash to
survive. In region B, this firm spends on maintenance and
avoids new investments. In region C, this firm expands and
invests in new projects.
of the firm's assets‐in‐place. Above the price X2 the firm in addition pursues expansionary (growth) investments that
require higher levels of capital expenditures.
Tiered hedging demand is driven by the firm's production technology which involves achieving profitability at
current output prices independent of the firm's capital structure. In an industry that produces a homogenous good such
as oil, economic distress can arise in firms with a higher‐than‐average cost base, which reduces operating margins and
cash flows. Economically distressed firms are more likely to hold a derivative portfolio containing sold puts because
they have a shorter distance to the kinks in the liquidity demand curve. Structurally higher cost of operations requires
higher prices to sustain current operations and investment programs. The hedging problem that tiered hedging solves
occurs when there is a material probability that output prices X1 and X2 can be reached.
If the main risk management concern is to protect core investments, the firm should buy put options with strike
price X2. However, as the firm gets nearer to distress (output price below X1), survival becomes its primary objective.
Survival requires less funding of large‐scale investment programs because the firm has few viable investment
opportunities at that point. If the firm has debt in its capital structure, the cash flows required to service this debt
determine demand for funding in this output price range. At this point, the firm's core investments are no longer
economically viable, so the firm only needs to generate enough cash flow to meet its current debt obligations. Under
these conditions, hedging solely with bought puts (irrespective of if the puts are financed with cash or by selling call
options) to protect the core investment program generates excess liquidity for output prices below X1.
Why is excess liquidity a problem? Selling call options with strike price above X2 to finance the put options with
strike price X2 in the traditional 2W‐collar strategy described in many textbooks induces underinvestment when the
output price is greater than X2. By limiting incentives for investment when output prices are high, sold calls imply
below‐optimal investment from the perspective of firm value maximization. This poor coordination of investment and
financing is illustrated in Panel A in Figure 2, where there is excess liquidity in the area A but simultaneously
underinvestment in the area B.
This tension between liquidity management and incentives for investment is resolved by selling puts with a strike
price X1 set below X2 , in a strategy we describe as a 3W collar, or 3W‐collar for short. The sold puts in the 3W‐collar
reduce the problem of excess liquidity below X2 when the strike price on the sold puts is below X2. The firm receives a
premium for the sold puts that it can use to buy put options with strike price set at X2, which in turn reduces the
number call contracts (or alternatively increases the strike price on these contracts) the firm needs to sell to finance its
put‐option purchases. Hence, the associated underinvestment problem is also reduced. The collateral requirements of
the short put positions are covered by the long‐put positions, which are in‐the‐money when output prices are low.
Panel B of Figure 2 illustrates graphically how both excess liquidity (Area A) and underinvestment (Area B) are lower
compared to the 2W‐collar in Panel A.
In theory, if transaction costs are small, one could match more complicated corporate liquidity demand schedules
than the one shown in Figure 1 by using multiple options with different strike prices. Transaction costs will increase as
the number of options used in the hedging strategy increases, and a satisfactory compromise will be a tradeoff between
tracking the firm's liquidity demand schedule and limiting transaction costs.3
3
We thank an anonymous referee for pointing this out.
F I G U R E 2 Hedging strategies under tiered

demand for liquidity. The demand for liquidity
for a distressed firm as well as hedged cash flows
with a two‐way collar (2W‐collar) (Panel A) and
a three‐way collar (3W‐collar) (Panel B) are
shown. The 2W‐collar involves selling call
options with strike price SC and using the
proceeds to purchase puts with strike price LP.
As a result of the hedging strategy, in region A,
there is excess liquidity. In region B, there is
underinvestment. The 3W‐collar involves selling
call options with strike price SC, selling put
options with strike price SP and using the
proceeds to purchase puts with strike price LP.
As a result of the hedging strategy, in region A,
there is some excess liquidity, but less than with
a 2W‐collar. In region B, there is minimal
underinvestment.
There may also be other factors at work determining firms' hedging strategies. Survey evidence indicates that
management's view of the market also shapes hedging decisions.4 For example, firms may signal future prospects by
altering the composition of puts purchased, puts written and calls written for the hedging strategy. If producers have
positive private information on future price recovery, they will sell less calls and sell more puts.
In summary, the 3W‐collar can be an effective risk management strategy when firms are in financial distress. The
short put options reduce excess liquidity when output prices are low while simultaneously limiting the negative “call‐
option overhang” effect on investment when output prices are high. We test this hypothesis by studying the hedging
strategies of US oil and gas producers around the 2014–2015 oil‐price shock.
3 | SAMPLE, VARIABLES, A ND DESCRIPTIVE S TATISTICS
3.1 | Sample
The sample consists of publicly traded oil and gas producers in the United States (SIC code 1311) between Q1:2013 and
Q4:2015. The advantages of using the oil and gas industry for studies of derivative usage are well known. It is one of
very few industries to disclose sufficiently detailed information about derivative positions, and it consists of a larger
number of firms than other commodity‐producing industries. Jin and Jorion (2006) argue that it is a relatively
homogenous industry, yet it exhibits significant variation in hedge ratios and employs a variety of different hedging
4
See the Oil and Gas Journal survey of hedging strategies, April 10, 2015 (https://www.ogj.com/home/article/17294837/hedging-strategies-surveyed).
strategies. Furthermore, according to Bakke et al. (2016) the industry's cash flow volatility is high enough to make risk
management economically important.
Firms are eligible for inclusion in the sample if their headquarters are in the United States, they are publicly listed,
and they have at least $1mn in total assets in all quarters. This yields 2153 firm‐quarters, corresponding to 220 unique
firms. We furthermore require that 10‐Qs (quarterly reports) be available from the online EDGAR database, and that
firms report their derivative positions in sufficient detail to quantify different hedging strategies. The latter criterion
essentially means that firms must report their hedging position in tabular form. Fortunately, most firms use this form
of disclosure. Firms that report a value‐at‐risk or a sensitivity measure, which are also allowed under US accounting
rules, are deleted because the information is insufficient to determine the extent and type of hedging.
Since our object of study is hedging‐instrument choice, we are ultimately constrained to those firms that are
derivative users. We use keyword search to identify sample firms that use derivatives.5 A firm is eligible if it uses
derivatives in at least one quarter of the sample period, thereby allowing firms to begin or stop using derivatives. This
leaves us with 1320 derivative firm‐quarters. The firm must also report production figures to allow us to calculate hedge
ratios, further reducing the sample to 1215 firm‐quarters.
To fully explore the impact of distress on hedging portfolios we balance the sample by requiring firms to have at
least three quarters of data before (Q1:2013–Q3:2014) and after the oil price collapse in the autumn of 2014
(Q4:2014–Q4:2015). This leaves 1116 firm‐quarters, corresponding to 96 unique firms.6
3.2 | The 2014 oil price collapse
The price of oil fell sharply during the fourth quarter of 2014. After fluctuating for a prolonged period around $90–100
per barrel (bbl), the price of West Texas Intermediate oil roughly halved within the space of one quarter (Figure 3).
While a modest decline appeared before Q4, the fall accelerated in early October 2014. On November 27th, OPEC
announced that it would change its production output to recoup market share, fueling the fall.
The accelerated fall in October 2014 was unforeseen by industry analysts and derivative markets, though
indications were present postfacto (Baffes et al., 2015).7 North American nonconventional oil production (i.e., shale and
tight oil) had been booming for the past 5 years but had so far been more than matched by declines in conventional oil
supply, which offset potential downward pressure on oil prices.8 Deloitte added in their Oil and Gas Reality Check 2014,
published in July 2014 that “[t]he expected price stability bodes well for the industry.” The 12‐month forward price of
oil exhibited in Figure 3 is characteristic of forward prices at all maturities: the forward curve provided no indications
of an upcoming crash. Moreover, a poll of 30 oil analysts by Reuters dated October 1st predicted a Brent crude price of
$103 for 2015.9 Even as late as October 26, 2014, Goldman Sachs revised their price forecast for Q1:2015 from $100 to
$85. In the same week, CIBC World Markets maintained their 2015 Brent average price of $100.
A second feature of the shock is the abrupt increase in volatility that accompanied the descent in oil prices. Figure 3
shows that oil‐price volatility correlated inversely with the oil price during the shock. Oil prices were unusually stable
during 2013 and most of 2014, and the CBOE crude oil volatility index (VIX) reached an all‐time low just below 15% in
June 2014. Volatility quadrupled during Q4:2014, to stabilize around 45% during 2015. Baffes et al. (2015) suggest that
supply factors were decisive to the price decline, which can explain the abrupt increase in the volatility of oil prices
over the period of the shock.
The combination of plunging oil prices and skyrocketing volatility provides a unique setting in which to study
corporate hedging policy. Before the oil shock, oil producers were generally profitable and had positive cash flows.
After the price collapse, default risk increased significantly. Figure 4 plots the average distance‐to‐default implied by
5
Hedging positions are identified by carefully reading the 10‐Qs, as well as through keyword search. Examples of search words are: “item 7a,” “hedg,”
“derivative,” “market risk,” “swap,” “collar,” “forward,” “put option,” and “risk management.”
6
Balancing the sample results in a drop of 42 observations; all main results are robust to including these firm‐quarters.
7
Possible reasons behind the sudden price decline include increased supply from North American nonconventional oil producers, declining demand
due to slowing global growth, and a significant shift in OPEC policy (Baffes et al, 2015).
8
The International Energy Agency (IEA) noted in their Medium‐Term Oil Market Report 2014 dated June 2, 2014, that “[o]il markets are in many
ways tighter today than they were at the onset of the US shale and tight oil boom, and considerably tighter than they were a year ago. Not
surprisingly, far from falling back from their highs under the weight of the new nonconventional supply, oil prices have remained stubbornly
elevated.”
9
“Oil price forecasts cut, to stay subdued in 2015: Reuters poll”, published October 1, 2014.
F I G U R E 3 Crude oil price and volatility,

2013–2015. The black curve (left axis) shows the
price of West Texas Intermediate (WTI) crude
oil, while the dotted curve (left axis) shows the
New York Mercantile Exchange (NYMEX) 12‐
month forward price of WTI crude oil. The gray
curve (right axis) shows the Chicago Board
Options Exchange (CBOE) crude oil volatility
index.
F I G U R E 4 Distress risk and hedging, Q1:2013–Q4:2015. The average distance‐to‐default for the balanced subsample comprising firms
that use derivatives in at least one quarter and that have at least three observations in both the preshock and shock periods. For variable
definitions, see Appendix A.
the firms' reported debt levels between Q1:2013 and Q4:2015 (for details on the calculation of the distance‐to‐default
see Appendix A). As shown, the average distance‐to‐default was 6–9 standard deviations away from zero in the
preshock period from Q1:2013 to Q3:2014 but dropped to 0–1.5 in the shock period from Q4:2014 onward. In short,
from Q4:2014 onward the oil and gas industry was in distress. This is not to say that every firm was distressed, but the
fraction of firms with a distance‐to‐default below 1.8 (roughly equivalent to a 3.6% probability of default, which was the
1‐year default frequency for B‐rated US corporates over the period 1981–2018 according to Standard & Poor's 2018
Annual US Corporate Default and Rating Transition Study) grew from 11% in the preshock period to 61% in the second
half of the period. Almost 25% (vs. 73% in the preshock period) of the firms were at least 3.2 standard deviations away
from zero (≈0.07% probability of default, the 1‐year default frequency of A‐rated US corporates over 1981–2018) in the
shock period.
Summarizing, the high incidence of severely distressed firms combined with the wide dispersion in default risk
makes the sample period opportune for studying hedging behavior. While the shock does not lend itself to estimating
causal effects, the impact of the oil shock was heterogeneous and not all firms suffered financial distress. In fact, a
nonnegligible fraction of firms maintained a strong balance sheet throughout this period, meaning that there should be
enough dispersion in default risk to understand determinants of hedging strategy choice.
3.3 | Hedging variables
Based on reported hedging activity we distinguish between buying put options, selling put options, buying call options,
selling call options, and linear contracts. Linear contracts consist of forward, futures, and price swaps, for which the
payoff is a linear function of the underlying commodity. Quantitative data on these basic strategies are hand‐coded
based on quarterly reports (10‐Q). We sum the outstanding positions regardless of maturity for each quarter and create
a variable that takes the value 1 if the sum is positive, zero otherwise. This is done for each generic strategy, giving the
following five indicator variables: Linear, Bought puts, Sold calls, Sold puts, and Bought calls.10 Hedging contracts on
natural gas are converted into barrels of oil equivalents using the standard assumption that 6 Mcf of gas has the same
energy content as 1 barrel (bbl) of oil (Kumar & Rabinovitch, 2013).
Following Adam (2009) we classify individual firms' hedge portfolios into five distinct hedging strategies
based on the character of the provided protection and the cash flow impact (see Appendix A for detailed
definitions). The Linear only strategy comprises hedge portfolios exclusively containing linear contracts, while
the Insurance strategy includes hedge portfolios where the option strategy is based on put options financed by
cash or a combination of cash and sold calls. In the 2W‐collar strategy the firm's option strategy is entirely based
on selling call options to finance the bought put options, which we define as having equal amounts of options on
the two legs. The 3W‐collar strategy instead includes hedge portfolios where the bought put options are financed
by selling put options. Our fifth hedging strategy is the Net sold call strategy where the firm's option strategy is
based on selling call options, either as a standalone strategy or more than any bought put options. 11 The
insurance, 2W‐collar, and net sold call strategies are distinguished by the fraction of sold call options to bought
put options. In the 2W‐collar strategy, the firm sells as many calls as it buys puts, whereas in the insurance (net
sold call) strategy, it buys more (less) puts than it sells calls. As in Adam (2009), the option‐based strategies do
not exclude linear hedging. They are calculated based on the option‐part of the firm's portfolio of derivatives,
whereas the same firm simultaneously may or may not hold linear contracts.
3.4 | Other variables
We measure financial distress with the distance‐to‐default based on Merton's model of the firm. This measure
provides the magnitude (in units of standard deviations of asset returns) an asset must decline before
shareholders rationally default on their debt obligations. We employ the naïve method of estimation based on
realized stock returns and volatility as well as the amount of debt outstanding. For further details, see Bharath
and Shumway (2008).
We also use control variables common in the hedging literature. These variables include measures of firm size,
profitability, leverage, and cash holdings. Definitions are provided in Appendix A. We obtain financial and operating
data from Compustat and Capital IQ. Since our sample is small, all financial variables are winsorized at the 2.5 and 97.5
percentiles.
3.5 | Descriptive statistics
Panel A of Table 2 reports the frequencies for the generic hedging positions. Linear contracts are the most common,
occurring in 81% of firm‐quarters among derivative users. Bought puts and sold calls occur in approximately the same
number of firm‐quarters (63%), indicating that a large fraction of bought puts are financed by selling call options rather
than paying in cash. Sold puts appear in 25% of firm‐quarters, whereas the corresponding number for bought calls is
8%. From Panel B of Table 2, we see that three hedging strategies cover the vast majority of cases. Linear only,
2W‐collar, and 3W‐collar appear in the sample in roughly the same proportions (around 25%). Based on these results,
corporate hedging in the oil and gas industry consists primarily of composite strategies, that is, combinations of various
10
These categories are not mutually exclusive.
11
Our results are robust to alternatively defining the 2W‐collar to include all derivatives portfolios with sold call options = bought put options ±10%,
±20%, or ±30% (and adjusting the definitions of the insurance and net sold call strategies accordingly).
TABLE 2 Hedging strategies, Q1:2013–Q4:2015.

Panel A. Generic strategies
Frequency
Linear 0.806
Bought puts 0.633
Sold calls 0.625
Sold puts 0.254
Bought calls 0.080
Panel B. Composite strategies
Frequency
Linear only 0.250
Insurance 0.100
Two‐way collar 0.247
Three‐way collar 0.254
Net sold call 0.059
No hedging 0.091
Note: The frequencies of common hedging strategies for a balanced sample comprising 1116 firm quarters across 96 firms that use derivatives in at least one
quarter and that have at least three observations in both the preshock and shock periods are shown. For variable definitions, see Appendix A.
TABLE 3 Descriptive characteristics, Q1:2013–Q4:2015.

Variable N Mean SD Min Median Max
WTI ($) 1116 80.5 23.1 44.7 94.3 105.8
WTIVol 1116 30.8 14.5 15.9 21.4 54.8
Assets ($ mn) 1116 5772.0 9899.3 2.505 2067.6 43,438.8
Leverage 1116 0.406 0.238 0.000 0.384 1.196
Cash 1116 0.040 0.056 0.000 0.016 0.243
Tobin's Q 1116 1.374 0.561 0.650 1.206 2.977
Cash flow 1116 0.003 0.075 −0.242 0.026 0.086
Capex 1100 0.061 0.056 0.002 0.049 0.289
Hedge ratio 1116 0.401 0.255 0.000 0.398 0.833
Distance‐to‐default 857 5.117 5.264 −2.398 4.441 18.175
Reserves‐to‐production 1026 57.868 26.575 21.833 52.084 140.982
Production cost ($) 741 14.358 6.778 5.490 12.710 32.160
Note: The descriptive statistics for a balanced sample of 1116 firm quarters comprising 96 firms that use derivatives in at least one quarter and that have at least
three observations in both the preshock and shock periods are shown. For variable definitions, see Appendix A.
generic strategies. Cash‐financed put options (Insurance) and Net sold call are comparatively much rarer strategies (at
10% and 6%, respectively). Generic option positions hardly ever occur in isolation.12
Table 3 reports descriptive statistics for our sample. Previous literature has documented a strong size effect
influencing the decision to hedge. This feature is also true in our sample, where the median book value of assets is
$2067 mn among derivative users. Appendix B compares sample firms to the broader sample of oil & gas firms that
12
Bought puts (sold calls) make up the only options component in 3.2% (2.2%) of the firm‐quarters, while bought calls and sold puts are never used in
isolation.
includes nonhedging firms. Descriptive statistics for the full set of firm‐quarters comprising derivative users and
nonusers are presented in Table B1. Compared to this broader sample, derivative users are larger, more levered, and
hold less cash. They are also more profitable, but have lower Tobin's Q.
Panel A of Table 4 shows descriptive statistics for each of the dominant strategies. As shown, 3W‐collar and net
short call users are larger than firms that predominantly use other strategies. Smaller firms instead seem to prioritize
the insurance or 2W‐collar strategies. The 2W‐collar and 3W‐collar appear to attract very different types of companies:
3W‐collar users have higher leverage, fewer growth opportunities, smaller distance‐to‐default, and are less profitable.
By comparison, firms that exclusively use linear strategies are more profitable while firms that use the insurance
strategy have fewer investment opportunities (as measured by Tobin's Q). Net short call users stand out by having the
TABLE 4 Mean characteristics per hedging strategy, Q1:2013–Q4:2015.

Panel A. Characteristics per hedging strategy
Variable Linear only Insurance Two‐way collar Three‐way collar Net sold call
Assets (mn) 4858 3461 3551 8684 6035
Leverage 0.424 0.413 0.363 0.452 0.566
Cash 0.041 0.040 0.033 0.039 0.034
Tobin's Q 1.405 1.144 1.529 1.229 1.422
Cash flow 0.014 −0.010 0.014 −0.005 −0.021
Capex 0.035 0.013 0.047 0.028 0.019
Hedge ratio 0.372 0.602 0.414 0.488 0.449
Distance‐to‐default 5.145 4.187 6.010 4.754 3.510
Production cost ($) 13.935 15.186 14.344 14.150 12.374
N 283 111 272 288 59
Panel B. Distance‐to‐default per hedging strategy
Distance‐to‐default < 1.8 1.8 ≤ Distance‐to‐default ≤ 3.2 Distance‐to‐default > 3.2
Linear only 0.323 0.070 0.607
Insurance 0.384 0.081 0.535
Two‐way collar 0.259 0.092 0.649
Three‐way collar 0.342 0.082 0.575
Net sold call 0.297 0.216 0.486
All strategies 0.321 0.086 0.593
Panel C. Hedge ratios per distance‐to‐default—hedging strategy group
Distance‐to‐default < 1.8 1.8 ≤ Distance‐to‐default ≤ 3.2 Distance‐to‐default > 3.2
Linear only 0.485 0.310 0.318
Insurance 0.660 0.416 0.593
Two‐way collar 0.433 0.575 0.457
Three‐way collar 0.556 0.482 0.506
Net sold call 0.406 0.370 0.454
All strategies 0.514 0.457 0.447
Note: Panel A reports mean characteristics for each of the dominant hedging strategies for the subsample comprising 1013 firm‐quarters with nonmissing firm
characteristics that use derivatives in at least one quarter and that have at least three observations in both the preshock and shock periods. Panel B reports the
distribution of the dominant hedging strategies across low, intermediate, and high distance‐to‐default (D‐to‐D), where low (high) distance‐to‐default is defined
as a distance‐to‐default less than 1.8 (more than 3.2) standard deviations away from zero. Panel C reports the mean hedge ratio for each of the dominant
hedging strategies across low, intermediate, and high distance‐to‐default (D‐to‐D). Variable definitions are found in Appendix A.
*, **, and *** indicate significance at the 10%, 5%, and 1% levels.
lowest production costs (measured in dollars per barrel produced) in combination with having higher leverage.
Insurance and 3W‐collar users have high hedge ratios (0.602 and 0.488) compared with linear‐only hedgers (0.372).
To appreciate how hedging strategies relate to default risk, Panel B of Table 4 reports the distribution of each hedging
strategy across low, intermediate, and high distance‐to‐default firm‐quarters. High (low) default risk is defined as a distance‐
to‐default lower than 1.8 (higher than 3.2) standard deviations from zero; 32% (59%) of the firm‐quarters have a distance‐to‐
default that is lower than 1.8 (higher than 3.2) standard deviations. The 2W collar is the most prevalent (65%) hedging
strategy among firms with a high distance to default. The net short call strategy is the most frequent (22%) hedging strategy
among firms in the middle tercile of distance to default. Firms employing Insurance and 3W‐collars are over‐represented
(38% and 34%, respectively) in the low default risk category relative to the other strategies.
Panel C of Table 4 reports average hedge ratios for each hedging strategy across default‐risk classes. Among
3W‐collar users, hedge ratios are increasing in default risk (i.e., decreasing in distance to default), which suggests this
type of strategy meets the hedging demands of financially distressed firms.
4 | FINANCIAL DISTRESS A ND HEDGING STRATEGIES
We use the sharp increase in default risk brought about by the oil price shock to explore the importance of distress to
corporate hedging strategy choice. The deteriorating economic and financial performance over the sample period had
important consequences for hedging behavior in the oil and gas industry. The deterioration was accompanied by a shift
from linear hedging strategies toward option‐based ones. Figure 5 shows that as conditions worsened there was a
relative increase in the usage of option strategies and a corresponding relative decline in linear strategies.
To further explore time‐variation in the use of hedging strategies we classify firms in each calendar quarter into two groups
depending on whether they are financially distressed. Financial distress is defined as having a distance‐to‐default less than 1.8.
We compare these firms to nondistressed firms, defined as having a distance‐to‐default greater than 3.2. We then track the
frequency of each hedging strategy quarter‐by‐quarter for the two groups of firms. Results are reported in Table 5.
Comparing the pre Q4:2014 period to the period from Q4:2014 onward reveals that utilization frequencies decline
for linear‐only, 2W‐collars and net sold calls. In contrast, utilization rates for insurance and 3W‐collars increase. The
decline in net sold calls and increase in insurance strategies reflects a greater concern with downside risk among oil
and gas producers during this period.
Comparing the rates of utilization of 3W‐collars between distressed and nondistressed firms reveals that the
increase in utilization rates of this strategy in the fourth quarter of 2014 onward is driven by distressed firms. As shown
in Panel B, which compares frequencies between firms, distressed firms are 8% and 3% more likely to employ this
strategy in Q4:2014 and Q1:2015, respectively. Similarly, the increase in utilization rates of insurance strategies in the
shock period is driven by distressed firms. These firms are 3% and 9% more likely to employ insurance strategies in
Q4:2014 and Q1:2015, respectively.
Overall, Table 5 reveals an important regularity, namely that oil & gas firms increase their usage of certain hedging
strategies as prices in their product market decline: these strategies involve bought puts that are self‐financed in
combination with sold calls and puts (3W‐collar). We next examine the relation between financial distress and hedging
strategy choice in a multivariate context.
4.1 | Multivariate regressions
Table 6 examines the effect of financial distress on hedging policy controlling for known determinants of firms' hedging policy.
We report binomial regression results in which each of the five hedging strategies are regressed on various determinants of
financial status. Panel A uses the variable Post to measure the average impact of the oil‐price shock on the frequency of
different types of hedging strategies. Post equals one on and after the fourth quarter of 2014 and zero otherwise. Panel B instead
uses the distance‐to‐default to measure the probability of financial distress. While the oil‐price shock affects all firms
simultaneously, the effect of the shock will vary with the amount of indebtedness of each firm. This dimension is well‐captured
with the distance‐to‐default, which measures the effect of the oil‐price shock on the likelihood of financial distress. To control
for the effect of size on hedging policy (larger firms are more likely to hedge), the log of assets is included as a control. Panel C
introduces the full model specification where firm‐specific distress is broken down into various components: cash, growth
opportunities (Tobin's Q), and cash flow in addition to the distance‐to‐default. In addition, we include two market‐based
F I G U R E 5 Hedging strategies, Q1:2013–Q4:2015. The fractions of hedging strategy users for the balanced subsample comprising firms
that use derivatives in at least one quarter and that have at least three observations in both the preshock and shock periods. Panel A shows
the fraction of the sample using the linear only strategy. Panel B shows the fraction of the sample using the insurance strategy. Panel C
shows the fraction of the sample using the two‐way collar (2W) strategy. Panel D shows the fraction of the sample using the three‐way collar
(3W) strategy. Panel E shows the fraction of the sample using the net sold call strategy.
TABLE 5 Hedging frequencies for distressed and nondistressed firms.

Panel A: Hedging frequencies by strategy
Linear only Insurance Two‐way collar Three‐way collar Net sold call
Calendar Qtr DD < 1.8 DD > 3.2 DD < 1.8 DD > 3.2 DD < 1.8 DD > 3.2 DD < 1.8 DD > 3.2 DD < 1.8 DD > 3.2
Q1:2013 0.63 0.19 0.00 0.12 0.13 0.31 0.13 0.25 0.00 0.08
Q2:2013 0.43 0.17 0.00 0.09 0.29 0.36 0.14 0.26 0.00 0.06
Q3:2013 0.44 0.22 0.11 0.09 0.22 0.35 0.22 0.27 0.00 0.07
Q4:2013 0.29 0.30 0.00 0.07 0.43 0.33 0.14 0.23 0.14 0.03
Q1:2014 0.25 0.31 0.00 0.05 0.38 0.33 0.25 0.21 0.13 0.03
Q2:2014 0.25 0.28 0.00 0.11 0.50 0.31 0.00 0.25 0.25 0.00
Q3:2014 0.31 0.28 0.08 0.13 0.15 0.28 0.38 0.20 0.08 0.02
Q4:2014 0.27 0.21 0.10 0.07 0.17 0.21 0.29 0.21 0.02 0.04
Q1:2015 0.21 0.19 0.16 0.07 0.19 0.30 0.26 0.22 0.00 0.00
Q2:2015 0.20 0.22 0.15 0.09 0.26 0.26 0.30 0.30 0.00 0.00
Q3:2015 0.15 0.19 0.13 0.13 0.28 0.25 0.34 0.38 0.04 0.00
Q4:2015 0.19 0.18 0.17 0.06 0.17 0.18 0.24 0.35 0.10 0.06
Panel B: Hedging frequency differences between distressed (DD < 1.8) and nondistressed (DD > 3.2) firms
Calendar Qtr Linear only Insurance Two‐way collar Three‐way collar Net sold call
Q1:2013 0.43 −0.12 −0.18 −0.13 −0.08
Q2:2013 0.26 −0.09 −0.07 −0.12 −0.06
Q3:2013 0.23 0.02 −0.12 −0.05 −0.07
Q4:2013 −0.01 −0.07 0.10 −0.09 0.11
Q1:2014 −0.06 −0.05 0.05 0.04 0.09
Q2:2014 −0.03 −0.11 0.19 −0.25 0.25
Q3:2014 0.03 −0.05 −0.12 0.18 0.06
Q4:2014 0.05 0.03 −0.04 0.08 −0.01
Q1:2015 0.02 0.09 −0.11 0.03 0.00
Q2:2015 −0.02 0.07 0.00 0.00 0.00
Q3:2015 −0.04 0.00 0.03 −0.03 0.04
Q4:2015 0.01 0.11 −0.01 −0.11 0.04
Note: The frequencies of the use of each hedging strategy quarter‐by‐quarter are reported. We classify firms in each calendar quarter into two groups depending
on whether they are financially distressed. Financial distress is defined as having a distance‐to‐default less than 1.8, while nondistressed firms are defined as
having a distance‐to‐default greater than 3.2. Panel A reports frequencies for each hedging strategy quarter‐by‐quarter for financially distressed and
nondistressed firms. Panel B reports differences in frequencies between financially distress and nondistressed firms.
variables that influence the relative cost of different hedging strategies: the oil price per barrel in USD (WTI) and the volatility
of the oil price (WTIVol) in annual daily returns.
Panel A in Table 6 confirms the pattern in Figure 5 regarding the drop in the usage of linear hedging strategies: there is a
highly significant drop in this type of strategy post 2014, as measured by the coefficient on the shock period dummy Post. The
usage of the 2W‐collar decreases postshock whereas insurance‐based strategies and 3W‐collar strategies increase with the oil‐
price shock. The sign on Post is negative and statistically insignificant for the Net sold call strategy.
A preference for insurance‐based rather than linear hedging is consistent with linear hedges carrying an opportunity cost
in form of loss of upside potential. Selling call options carries a similar opportunity cost, but at higher strike rates. The relative
increases in cash‐financed insurance and financing with sold puts as opposed to financing buying put options with sold calls
suggest that this loss of upside potential may outweigh the drawback of reducing cash‐on‐hand.
TABLE 6 Determinants of hedging strategies, Q1:2013–Q4:2015—binomial logit evidence.

Panel A. Post‐shock dummy
Constant −0.930*** −2.349*** −0.984*** −1.127*** −2.843***
(−21.4) (−28.0) (−20.2) (−32.0) (−33.1)
Post −0.382*** 0.327*** −0.381*** 0.168*** −0.106
(−4.88) (3.00) (−3.83) (2.62) (−0.49)
N 1116 1116 1116 1116 1116
2
Pseudo R 0.006 0.004 0.006 0.001 0.000
Panel B. Distance‐to‐default
Constant −0.580*** −1.382*** −0.056 −4.128*** −4.684***
(−2.82) (−4.79) (−0.25) (−19.5) (−7.61)
Log(Assets)t−1 −0.089*** −0.091*** −0.169*** 0.441*** 0.273***
(−3.22) (−3.16) (−5.08) (15.1) (3.85)
Distance‐to‐defaultt−1 0.007 −0.023 0.060*** −0.069*** −0.107***
(0.40) (−1.24) (4.13) (−5.69) (−2.60)
N 850 850 850 850 850
2
Pseudo R 0.005 0.008 0.023 0.084 0.037
Panel C. Full specification
Constant −1.225** −0.044 0.526 −3.481*** −5.221***
(−2.03) (−0.07) (0.86) (−6.79) (−4.28)
WTIt−1 0.010* −0.006 −0.012** 0.002 0.002
(1.85) (−1.23) (−2.09) (0.58) (0.17)
WTIVolt−1 0.001 0.008** −0.010* −0.014*** −0.034*
(0.13) (2.199) (−1.91) (−3.52) (−1.88)
Log(Assets)t−1 −0.088** −0.150*** −0.160*** 0.456*** 0.389***
(−2.40) (−5.06) (−4.85) (14.3) (4.03)
Distance‐to‐defaultt−1 −0.007 0.052*** 0.002 −0.059*** −0.194***
(−0.40) (4.07) (0.12) (−3.73) (−3.30)
Casht−1 0.170 4.790** −6.732*** 1.871 0.295
(0.14) (2.42) (−3.60) (1.31) (0.08)
Tobin's Qt−1 −0.146 −0.970*** 0.819*** −0.494*** 0.629*
(−1.15) (−4.37) (5.78) (−4.50) (1.80)
Cash flowt−1 1.917* 0.544 3.833*** −2.133* −1.778
(1.76) (0.52) (2.90) (−1.89) (−0.64)
N 780 780 780 780 780
2
Pseudo R 0.016 0.037 0.060 0.096 0.068
Note: The coefficients and z‐values clustered by quarter (in parenthesis) from bivariate logit regressions in which indicator variables for the different hedging
strategies are dependent variables (each indicator variable takes the value 1 if the firm uses the indicated hedging strategy, 0 otherwise). Data for the estimation
sample consists of firms with nonmissing accounting variables that use derivatives in at least one quarter and that have at least three observations in both the
preshock and shock periods. For variable definitions, see Appendix A.
While Panel A provides clues to the relevance of financial distress to hedging behavior, Post is a blunt distress proxy
in that it assumes that the effect of the shock is homogeneous across firms. Panel B uses the distance to default to
measure distress at the firm‐year quarter level. As shown, this measure is an important determinant of hedging
strategies. Firms closer to default are more likely to use the 3W‐collar, whereas the 2W‐collar is inversely related to
default risk. The Net sold call strategy also decreases in distance‐to‐default, suggesting that selling options in general
(whether puts or calls) becomes more common the closer the firm is to default.
Panel C in Table 6 confirms that these results persist with a full set of control variables. 3W‐collar usage decreases
significantly with the distance to default, as does the net sold call strategy, but 3W‐collar falls when growth options
(measured with Tobin's Q) improve. The likelihood of using the insurance strategy similarly falls when growth
opportunities improve but increases with the distance to default. The 2W‐collar, in contrast, is associated with better
growth opportunities and lower cash balances, but not with the distance to default.
Whereas the likelihood of using the 3W‐collar strategy declines with the distance‐to‐default and falls with
improvements in growth opportunities, the 2W‐collar is associated with higher growth opportunities and liquidity (as
proxied with cash). These results highlight the importance of selling options instead of engaging in linear hedging
strategies as the distance to default shortens. Thus, the 2W‐collar and 3W‐collar seem to attract different users. Linear
strategies are only significantly associated with firm size.
Oil‐price volatility affects firms' choice of hedging strategy. As shown in Panel C, controlling for the distance‐to‐default,
insurance‐based strategies become more prevalent when volatility is high, but oil‐price volatility decreases the frequency of
3W collars. Increases in oil‐price volatility may raise the likelihood of very low oil prices, decreasing the attractiveness of
3W‐collars, which are often put in place in anticipation that oil prices are unlikely to cross the “subfloor” (i.e., X1 in Figure 2)
relative to pure insurance strategies.
In Table B2 in Appendix B, we report robustness tests using Altman Z‐score as a proxy for financial distress. Our
results are qualitatively similar.13
4.2 | Multinomial regressions
Table 7 reports results from multinomial regressions where the no‐hedging choice is the baseline reference point. The
estimates are qualitatively similar to the binomial regression estimates. As shown, distance‐to‐default is significantly
negatively associated with increases in the likelihood of 3W‐collar strategies as well as other types of hedging strategies
(except for the insurance strategy) relative to the baseline no‐hedging outcome. More generally, the prevalence of sold
options is significantly related to the distance‐to‐default, highlighting the need for liquidity and/or self‐financing
hedging strategies as the firm's financial situation deteriorates.
The multinomial regressions show how pure insurance strategies become relatively more attractive than 3W‐collars
as the possibility of very low oil prices increases with WTI volatility. As shown, the coefficient on WTIVol is negative
for both insurance and 3W‐collars, indicating that firms are less likely to employ both hedging strategies compared to a
baseline of no hedging. However, the effect of WTI volatility on insurance is lower in absolute magnitude than on
3W‐collars, indicating that as oil price volatility increases, firms are relatively more likely to employ insurance
strategies, compared with 3W‐collars, consistent with the logit results in Table 6.
4.3 | Cash positions and the effect of distance‐to‐default
Because insurance‐based strategies are cash‐intensive and not self‐financed, utilization rates may vary with firms' cash
positions. Firms may also switch away from a cash‐intensive strategy to strategies that involve a greater degree of self‐
financing such as 3W‐collars as their own cash reserves diminish. To investigate how the propensity to employ
13
Security and Exchange Commission (SEC) reporting requirements for oil and gas producers make the Z‐score an untimely default proxy. Rule 4‐
10(a) of Regulation S‐X states that the price used to value proved oil and gas reserves must be calculated as the unweighted arithmetic average of the
first‐of‐day‐of‐the‐month average price for each month within the 12‐month period before the end of the reporting period. While taking an average
smooths out volatility in prices, the average understates the magnitude of the decline in oil prices experienced by oil and gas firms over a very short
period of time.
TABLE 7 Determinants of hedging strategies, Q1:2013–Q4:2015—multinomial logit evidence.

Constant 2.142* 3.257*** 3.710*** 0.327 −1.558
(1.66) (3.01) (3.85) (0.28) (−0.82)
WTIt−1 0.003 −0.010 −0.014 −0.003 −0.004
(0.24) (−1.30) (−1.12) (−0.24) (−0.19)
WTIVolt−1 −0.045** −0.038** −0.053*** −0.058*** −0.081***
(−2.34) (−2.50) (−3.22) (−3.15) (−2.60)
Log(Assets)t−1 0.165* 0.100 0.109 0.605*** 0.646***
(01.65) (1.28) (1.30) (7.93) (4.42)
Distance‐to‐defaultt−1 −0.111** −0.059 −0.098** −0.160*** −0.298***
(−2.41) (−1.38) (−2.41) (−4.10) (−3.88)
Casht−1 −2.093 1.093 −6.908*** −0.746 −2.043
(−1.31) (0.55) (−3.08) (−0.40) (−0.61)
Tobin's Qt−1 −0.333 −1.064*** 0.211 −0.611*** 0.271
(−1.59) (−2.90) (1.06) (−2.71) (0.99)
Cash flowt−1 5.513** 4.740*** 7.191*** 2.878 2.693
(2.19) (2.56) (3.15) (1.17) (0.68)
N 780
2
Pseudo R 0.074
Note: The coefficients and z‐values clustered by quarter (in parenthesis) from multinomial logit regressions where no‐hedging is the reference point are
reported. Indicator variables for the different hedging strategies are dependent variables. Data for the estimation sample consists of firms with nonmissing
accounting variables that use derivatives in at least one quarter and that have at least three observations in both the pre‐shock and shock periods. For variable
definitions, see Appendix A. Full specification—base outcome = no hedging.
insurance and 3W‐collar hedging strategies varies with cash holdings, we interact the distance‐to‐default with the cash
balance at the end of the prior quarter. Estimation results are reported in Table 8.
Panel A of Table 8 shows that the sensitivity of linear, insurance, and 3W‐collar utilization rates to the distance‐to‐
default varies with cash holdings. Linear only strategies' sensitivity to the distance‐to‐default decrease with cash balances.
Insurance strategies increase with the distance‐to‐default when cash holdings are high, as shown by the positive coefficient
on the interaction term between the distance‐to‐default and cash. In contrast, 3W‐collar usage declines with the distance‐to‐
default and becomes more sensitive in absolute terms to changes in the distance‐to‐default when the firm's cash holdings are
low. These contrasting effects suggest that linear only, insurance, and 3W‐collar users differ significantly in terms of the
likelihood of financial distress and available cash resources.
To quantify the magnitude of the effect of cash holdings on hedging strategy choice, we perform a sensitivity analysis in
which we contrast the sensitivity of the probability of a hedging strategy to variation in the distance to default for the average
firm at different levels of cash.14 We consider three scenarios: cash levels are at the average level in the sample (4% of assets),
cash levels are lower by one standard deviation (5.6%), and cash levels are higher by one standard deviation. We then
compare the sensitivity of the probability of a given strategy to the distance to default across the three levels of cash. As
shown in Panel B, linear strategies decrease with distance‐to‐default when cash levels are high but increase with distance to
default when cash levels are low. Insurance strategies have a positive sensitivity to distance to default regardless of cash
levels. However, this strategy is most sensitive to distance‐to‐default when cash levels are high, with a sensitivity of 0.103
versus 0.007 when cash is low. In contrast, the sensitivity to distance‐to‐default of the 3W‐collar strategy is negative and
14
We measure the sensitivity of hedging policy at average levels of the regressors. See Wooldridge (2010) for marginal effects analysis of binary‐choice
models.
TABLE 8 Cash holdings and hedging strategy choice, Q1:2013–Q4:2015.

Panel A: Logit regression estimates
Constant −1.300** 0.110 0.542 −3.461*** −5.262***
(−2.22) (0.17) (0.90) (−6.31) (−4.15)
WTIt−1 0.010* −0.005 −0.012** 0.003 0.002
(1.80) (−1.08) (−2.05) (0.68) (0.19)
WTIVolt−1 0.002 0.007* −0.010** −0.015*** −0.034*
(0.32) (1.73) (−1.98) (−3.46) (−1.91)
Log(Assets)t−1 −0.092*** −0.146*** −0.158*** 0.462*** 0.394***
(−2.60) (−4.65) (−4.91) (15.6) (3.92)
Distance‐to‐defaultt−1 0.030 0.016 −0.011 −0.085*** −0.209***
(1.24) (0.94) (−0.42) (−5.83) (−3.06)
Casht‐1 4.300*** 0.724 −8.604*** −0.965 −0.546
(3.28) (0.21) (−2.83) (−0.59) (−0.14)
Distance‐to‐defaultt−1 × Casht−1 −0.755*** 0.663** 0.285 0.497** 0.237
(−2.83) (2.39) (0.96) (2.39) (0.73)
Tobin's Qt−1 −0.220** −0.963*** 0.844*** −0.454*** 0.669*
(−2.09) (−4.48) (6.02) (−4.39) (1.77)
Cash flowt−1 2.084* 0.412 3.826*** −2.272** −1.801
(1.93) (0.36) (2.95) (−1.98) (−0.65)
N 780 780 780 780 780
2
Pseudo R 0.023 0.042 0.061 0.099 0.068
Panel B: Sensitivity to the distance‐to‐default across cash levels
Linear only Insurance Three‐way collar
Mean level of cash 0.000 0.055 −0.086
Low cash level (mean −1 SD cash) 0.053 0.007 −0.122
High cash level (mean +1 SD cash) −0.056 0.103 −0.049
Note: Panel A of this table reports the coefficients and z‐values clustered by quarter (in parenthesis) from bivariate logit regressions in which indicator variables
for the different hedging strategies are dependent variables (each indicator variable takes the value 1 if the firm uses the indicated hedging strategy, 0
otherwise). Panel B reports the effect of distance‐to‐default at different levels of cash on the probability of linear only, insurance, and three‐way collar
strategies. Marginal effects are estimated at the mean values of the regressors (except for cash), as described in Chapter 15 of Wooldridge (2010). Data for the
estimation sample consists of firms with nonmissing accounting variables that use derivatives in at least one quarter and that have at least three observations in
both the preshock and shock periods. For variable definitions, see Appendix A.
greatest in absolute terms at −0.122 when cash is low, compared with only −0.049 when cash levels are high. Overall, how
hedging strategy choice varies with distance‐to‐default is contingent on cash levels: firms with low levels of cash are more
likely to engage in 3W‐collars than linear‐only and insurance strategies as the distance‐to‐default falls.
4.4 | 3W‐collars and economic distress
We next examine the relation between hedging and economic distress. Table 9 reports logit estimations in which the
3W‐collar is regressed on indicators of economic distress.
TABLE 9 The three‐way collar and economic distress, Q1:2013–Q4:2015.

Panel A. The 3W‐collar
(1) (2) (3) (4) (5)
Constant −5.295*** −4.479*** −3.590*** −3.494*** −4.872***
(−7.37) (−6.36) (−6.90) (−6.99) (−5.99)
WTIt−1 −0.004 −0.001 0.003 −0.001 0.001
(−0.78) (−0.17) (0.60) (−0.18) (0.22)
WTIVolt−1 −0.023*** −0.023*** −0.014*** −0.014*** −0.015***
(−3.35) (−3.35) (−3.42) (−3.78) (−3.21)
Log(Assets)t−1 0.538*** 0.578*** 0.446*** 0.449*** 0.468***
(11.3) (9.56) (13.8) (14.1) (13.2)
Distance‐to‐defaultt−1 −0.066*** −0.078*** −0.062*** −0.056*** −0.074***
(−3.84) (−4.39) (−3.71) (−3.43) (−4.09)
Casht−1 3.597* 3.556* 1.818 1.865 2.000
(1.75) (1.74) (1.30) (1.28) (1.29)
Tobin's Qt−1 −0.091 −0.046 −0.479*** −0.393*** −0.432***
(−0.80) (−0.36) (−4.66) (−3.33) (−3.89)
Cash flowt−1 −1.037 −0.763 −1.922* −1.071 −1.195
(−1.11) (−0.83) (−1.68) (−0.84) (−0.88)
Log(Production cost)t−1 0.598***
(3.37)
High production costt−1 0.903***
(5.24)
Shalet−1 0.395***
(4.47)
Neg cash flow firm 0.379***
(2.97)
Neg operating profit firm 1.545***
(2.82)
N 568 568 780 780 780
2
Pseudo R 0.120 0.125 0.102 0.101 0.115
Panel B. Economic distress and other hedging strategies
Linear only Insurance Two‐way collar Net sold call
Log(Production cost)t−1 −0.156 −0.726*** −0.372* −1.222**
(−1.32) (−2.64) (−1.91) (−2.48)
N 568 568 568 568
2
Pseudo R 0.032 0.071 0.043 0.010
High production costt−1 −0.712*** 0.214 −0.645*** 0.226
(−2.73) (1.01) (−2.80) (0.41)
N 568 568 568 568
Pseudo R2 0.039 0.058 0.047 0.074
TABLE 9 (Continued)
Panel B. Economic distress and other hedging strategies

Linear only Insurance Two‐way collar Net sold call
Shalet−1 −0.183 0.013 −0.587*** −0.461
(−1.12) (0.06) (−7.47) (−1.34)
N 780 780 780 780
2
Pseudo R 0.018 0.037 0.073 0.073
Neg cash flow firm −0.467*** 0.009 −0.116 −0.928**
(−3.60) (0.05) (−1.20) (−1.98)
N 780 780 780 780
2
Pseudo R 0.024 0.037 0.061 0.086
Neg operating profit firm −0.134 −1.010*** −0.428* −0.849***
(−0.59) (−3.90) (−1.82) (−2.85)
N 780 780 780 780
2
Pseudo R 0.016 0.052 0.063 0.078
Note: The coefficients and z‐values clustered by quarter (in parenthesis) from bivariate logit regressions in which the dependent variable is an indicator variable that takes
the value 1 if the firm uses the three‐way collar strategy, 0 otherwise (Panel A) are reported. In Panel B, indicator variables for the linear only, insurance, 2W‐collar, and
net sold call strategies are dependent variables (each indicator variable takes the value 1 if the firm uses the indicated hedging strategy, 0 otherwise). Each cell in Panel B
represents a separate regression with results on control variables suppressed; model specifications are identical to Panel A. Data for the estimation sample consists of
firms with nonmissing accounting variables and production costs that use derivatives in at least one quarter and that have at least three observations in both the pre‐
shock and shock periods. For variable definitions, see Appendix A
Model 1 uses the (log of the) average production cost, normalized per barrel. This is a direct measure of how the
firm's cost structure compares to that of its peers. Model 2 uses a binary variable that takes the value 1 if the average
production cost lies in the top two deciles of the production‐cost distribution for the quarter; in contrast to Model 1,
Model 2 makes no assumption of monotonicity in the relationship between production costs and usage of the 3W‐collar
strategy. Model 3 uses a binary variable that takes the value 1 if a search on the word “shale” in the firm's quarterly
report yields a positive number of hits, zero otherwise. Extracting shale oil is inherently more expensive than extraction
from conventional wells, and therefore implies a structurally higher cost base.
Model 4 uses a binary variable that takes the value 1 if the firm experienced negative operating cash flows in at least
three out of four quarters during 2015. Negative cash flow has been used in the literature as an empirical proxy for
economic distress (e.g., Allayannis & Mozumdar, 2004). This measure identifies the firms that truly got into severe
economic problems because of the shock. Model 5 instead uses a binary variable that takes the value 1 if the firm
experienced negative operating profits in at least three out of four quarters during 2015.
The effect of distance‐to‐default is robust to the inclusion of proxies for economic distress in all specifications.
Moreover, consistent with our main hypothesis, economically distressed firms are more likely to use the 3W‐collar.
There is a positive and significant association between the 3W‐collar and firms with higher average production costs,
firms more exposed to shale, and firms with predominantly negative EBIT or cash flow in the shock period.
Panel B reports the effect of economic distress on the frequency of other types of hedging strategies. Each cell
represents a separate regression. Results on control variables are suppressed, but model specifications are identical to
Panel A. In stark contrast to the results for the 3W‐collar, the association between economic distress and the other
hedging strategies is either insignificant or negative and shows no clear pattern across economic distress proxies.
4.5 | Risk shifting
Distressed firms appear to be the least apt to assume liabilities that will exacerbate the firm's condition if prices go
down. An alternative possibility is therefore that the preference for sold puts reflects incentives to engage in risk
shifting (i.e., selling insurance).
The risk‐shifting hypothesis cannot explain hedging behavior in our sample, however. The first and most basic observation
is that the short puts never appear in isolation. In the associated composite strategy (3W‐collar) the bought put always has a
higher strike price, so the position will continue to generate a positive cash flow no matter how low the price falls.
The risk management aspect of the strategy is confirmed by an analysis of quarterly reports through the shock. In
Table 10, we show that just over 10% of firms report the 3W‐collar as a net liability. While the short put leg itself is a
liability, it is always dominated by the positive value of the bought puts. Net liabilities, if present, thus arise from the
short call positions. If risk‐shifting were the case, we would have expected the 3W‐collar users to have greater
proportions of net derivative liabilities (net derivative assets, NDA < 0) than linear hedgers (19.4%) and 2W‐collar users
(38.3%) in the post‐2014 period.
4.6 | Collateral constraints
Another theory to consider for explaining the association between distance‐to‐default and the 3W‐collar involves
collateral constraints. Entering linear contracts or short positions often requires collateral in the form of cash or
undrawn credit lines (Mello & Parsons, 1999). If a firm is constrained and therefore cannot set aside any collateral to
support hedging, it may prefer to sell out‐of‐the money puts in combination with bought puts. This is because the sold
puts combined with bought puts do not require collateral if the strike price on the former is set below that of the latter.
The long puts dynamically generate the liquidity needed to deal with collateral requirements that may arise on the sold
puts when the price falls (triggering margin calls from the firm's derivative counterparty). While the positive
association between this strategy and default risk supports this explanation, it is contradicted by the fact that net sold
calls are negatively related to distance‐to‐default (Table 6, Panel C). This indicates that firms can enter such positions
despite being financially weak.
4.7 | 3W‐collars and bankruptcy rates
Another way to determine whether the 3W‐collar represents risk management or risk shifting is to measure default
rates among 3W‐collar users. If the 3W‐collar is employed for hedging by economically and financially distressed firms,
we should not expect a higher fraction of such firms to default compared to other hedging strategies.15 We examine
bankruptcy rates in Appendix C, which shows that bankruptcy rates are no higher for firms that use the 3W‐collar than
other types of hedging strategies.
4.8 | Oil price volatility and break even prices
Our evidence suggests that the goal of preserving investment opportunities dominates the value of the option to default.
Due to high oil‐price volatility going concern values can remain high even when current market prices are insufficient
to break even on current production. High levels of volatility help to preserve the value of real options, as there is a
nontrivial probability of prices rebounding to levels where operations and investment programs would again be
profitable.
The oil and gas markets are in fact notorious for their high volatility and have seen repeated price crashes.
However, in all recent cases, these episodes were relatively short‐lived. Oil‐price volatility also tends to increase in
times of falling oil prices (e.g., Batten et al., 2019; Ewing & Malik, 2017), which was certainly true in our sample period.
Further supporting the view that survival has value is the fact that breakeven prices tend to lag the oil price. This
means that firms that survive the most intense phase of the shock can expect costs to come down as service providers
respond by cutting prices and due to efficiency improvements (Azar, 2017; Kleinberg et al., 2018). We analyze
breakeven prices in Appendix B. Figure B1 reports data on half‐cycle breakeven crude‐oil prices ($/bbl) from
Bloomberg and operating (UOCI) and capital (UCCI) cost indexes for oil and gas production from HIS Markit.16 As
15
Even if the lower strike price is reached, the 3W‐collar throws off enough cash to avoid bankruptcy. The spread between the two put options can in
fact be thought of as “carrying the debt”: it is just enough to survive, but no or little excess cash is accumulating.
16
Bloomberg only started recording half‐cycle breakeven prices on January 1, 2014.
DUDLEY
ET AL.
T A B L E 10 Proportion of net derivative positions reported as net assets (net derivative assets [NDA]).
Linear only Insurance Two‐way‐collar Three‐way‐collar Net sold call
NDA NDA NDA NDA NDA NDA NDA NDA NDA NDA NDA NDA NDA NDA NDA
>0 =0 <0 >0 =0 <0 >0 =0 <0 >0 =0 <0 >0 =0 <0
Q1:2013– 56 60 69 27 13 17 39 70 69 71 74 15 17 10 9
Q3:2014
Fractions 30.3% 32.4% 37.3% 47.4% 22.8% 29.8% 21.9% 39.3% 38.8% 44.4% 46.3% 9.4% 47.2% 27.8% 25.0%
Cum fractions 100% 100% 100% 100% 100%
Q4:2014– 77 2 19 33 2 19 57 1 36 108 7 13 21 1 1
Q4:2015
Fractions 78.6% 2.0% 19.4% 61.1% 3.7% 35.2% 60.6% 1.1% 38.3% 84.4% 5.5% 10.2% 91.3% 4.3% 4.3%
Cum fractions 100% 100% 100% 100% 100%
Note: The proportion of firms for whom each strategy is reported as a net asset (net derivative assets, NDA > 0) is reported. Net derivative assets are defined as Derivative assets–Derivative liabilities. Data on derivative
assets and liabilities is obtained from company 10‐K filings. Data for the estimation sample consists of firms with nonmissing accounting variables and production costs that use derivatives in at least one quarter and
that have at least three observations in both the preshock and shock periods. For variable definitions, see Appendix A.
| 1345
shown, all series exhibit marked reductions in the shock period. The breakeven price fell by 29% between Q3:2014 and
Q1:2016, while operating and capital costs fell by 18% and 28%.
5 | CONCLUSION
We examine the derivative portfolios of publicly traded oil and gas firms around the 2014–2015 shock to oil prices. We
find that distressed firms in the oil and gas industry tend to avoid cash‐financed option strategies. An empirical
regularity is that distressed firms sell call options to finance their risk management in combination with both bought
and sold put options. A strategy involving sold puts in distressed firms is highly suggestive of risk shifting, yet the
evidence does not support this interpretation. We find that sold puts are an important component of oil and gas firms'
derivative portfolios. These puts are typically combined with bought puts and sold calls in a 3W‐collar. Our results
indicate that the combined strategy generates enough liquidity in distressed states for the firm to survive temporary
shocks to output prices, while attenuating the disincentives for investment when oil prices are high. For such firms, a
portfolio containing sold puts more efficiently equalizes the marginal value of internal liquidity over a broad range of
oil prices.
ACKNOWLEDGMENTS
The authors thank Knut Wicksell Financial Centre for Financial Studies for funding the research assistance. Andrén and
Jankensgård gratefully acknowledge the financial support of the Jan Wallander and Tom Hedelius Foundation. The
funding sources had no involvement in any part of the research or preparation of the article. We also wish to thank our
discussant at the FMA 2020 (New York) Annual Meeting for helpful comments as well as seminar participants at the
Knut Wicksell Centre for Financial Studies and the Stockholm School of Business, Stockholm University. Any errors are
the responsibility of the authors.
C O N F LI C T S OF IN T E R E ST
The authors declare no conflicts of interest.
DATA A VAILABILITY S TATEMENT

The data that support the findings of this study are openly available in the Swedish National Data Service research data
catalogue at https://doi.org/10.5878/rmrg-9341, reference number 2022‐47 (Andrén et al., 2022). Restrictions apply to
the availability of financial and operating data from Compustat and Capital IQ, which were used under license for this
study.
ORCID
Niclas Andrén http://orcid.org/0000-0001-5899-814X
REFERENCES
Adam, T. R. (2002). Risk management and the credit risk premium. Journal of Banking & Finance, 26, 24–269.
Adam, T. R. (2009). Capital expenditures, financial constraints, and the use of options. Journal of Financial Economics, 92, 238–251.
Allayannis, G., & Mozumdar, A. (2004). The impact of negative cash flow and influential observations on investment‐cash flow sensitivity
estimates. Journal of Banking & Finance, 28, 901–930.
Almeida, H., Hankins, K. W., & Williams, R. (2020). Do firms hedge during distress? Unpublished manuscript, May 2020.
Altman, E. (1968). Financial ratios, discriminant analysis, and the prediction of corporate bankruptcy. The Journal of Finance, 23, 589–609.
Andrén, N., Dudley, E., & Jankensgård, H. (2022). How do firms hedge in financial distress? Swedish National Data Service research data
catalogue. Data Citation; https://doi.org/10.5878/rmrg-9341.
Azar, A. (2017). Reserve based lending and the outlook for shale oil and gas finance. Columbia Center on Global Energy Policy, May.
Baffes, J., Kose, M. A., Ohnsorge, F., & Stocker, M. (2015). The great plunge in oil prices: Causes, consequences, and policy responses. Policy
research note 15/01. The World Bank.
Bakke, T., Mahmudi, H., Fernando, C., & Salas, J. (2016). The causal effect of option pay on corporate risk management. Journal of Financial
Economics, 120, 623–643.
Batten, J. A., Kinateder, H., Szilagyi, P. G., & Wagner, N. F. (2019). Liquidity, surprise volume and return premia in the oil market. Energy
Economics, 77, 93–104.
Bessembinder, H. (1991). Forward contracts and firm value: Investment incentive and contracting effects. Journal of Financial and
Quantitative Analysis, 26, 519–532.
Bharath, S. T., & Shumway, T. (2008). Forecasting default with the Merton distance to default model. Review of Financial Studies, 21,
1339–1369.
Croci, E., del Guidice, A., & Jankensgård, H. (2017). CEO age, risk incentives, and hedging strategy. Financial Management, 46, 687–716.
Ewing, B. T., & Malik, F. (2017). Modelling asymmetric volatility in oil prices under structural breaks. Energy Economics, 63, 227–233.
Froot, K. A., Scharfstein, D. S., & Stein, J. C. (1993). Risk management: Coordinating investment and financing policies. The Journal of
Finance, 48, 1629–1658.
Froot, K. A., Scharfstein, D. S., & Stein, J. C. (1994). A framework for risk management. Journal of Applied Corporate Finance, 7.3, 22–32.
Jin, Y., & Jorion, P. (2006). Firm value and hedging: Evidence from U.S. oil and gas producers. The Journal of Finance, 612, 893–919.
Kleinberg, R. L., Paltsev, S., Ebinger, C. K. E., Hobbs, D. A., & Boersma, T. (2018). Tight oil market dynamics: Benchmarks, breakeven
points, and inelasticities. Energy Economics, 70, 70–83.
Kumar, P., & Rabinovitch, R. (2013). CEO entrenchment and corporate hedging: Evidence from the oil and gas industry. Journal of
Financial and Quantitative Analysis, 48, 887–917.
Mello, A. S., & Parsons, J. E. (1999). Hedging and liquidity. Review of Financial Studies, 13, 127–154.
Merton, R. C. (1974). On the pricing of corporate debt: The risk structure of interest rates. The Journal of Finance, 29, 449–470.
Morellec, E., & Smith, C. E. Jr. (2007). Agency conflicts and risk management. Review of Finance, 11, 1–23.
Purnanandam, A. (2008). Financial distress and corporate risk management: Theory and evidence. Journal of Financial Economics, 87,
706–739.
Rampini, A. A., Sufi, A., & Viswanathan, S. (2014). Dynamic risk management. Journal of Financial Economics, 111, 271–296.
Smith, C. W., & Stulz, R. M. (1985). The determinants of firms' hedging policies. Journal of Financial & Quantitative Analysis, 204, 391–405.
Stulz, R. M. (1996). Rethinking risk management. Journal of Applied Corporate Finance, 9.3, 8–24.
Wooldridge, J. M. (2010). Econometric analysis of cross‐section and panel data (2nd ed.). MIT Press.
How to cite this article: Dudley, E., Andrén, N., & Jankensgård, H. (2022). How do firms hedge in financial
distress? Journal of Futures Markets, 42, 1324–1351. https://doi.org/10.1002/fut.22336
APPENDIX A: VARIABLE DEFINITIONS

Panel A. Hedging strategies
Linear 1 if the firm uses linear hedging instruments, 0 otherwise
Bought call 1 if the firm buys call options, 0 otherwise
Bought put 1 if the firm buys put options, 0 otherwise

Sold call 1 if the firm sells call options, 0 otherwise
Sold put 1 if the firm sells put options, 0 otherwise
Insurance 1 if there are no put options sold and the number of put options bought exceeds the number of call options, 0
otherwise
Linear only 1 if the firm has linear hedging instruments but no options, 0 otherwise
Net sold calls 1 if there are no put options sold and the number of sold call options exceeds the number of put options
bought, 0 otherwise
Two‐way collar (2 W) 1 if there are no put options sold and the firm has an equal number of put options bought and sold call
options, 0 otherwise
Three‐way collar (3 W) 1 if the firm simultaneously has bought put options, sold put options, and sold call options, 0 otherwise
Panel B. Financial variables
Assets Total assets ($ mn)
Capex (PPEt — PPEt−1 + Depreciation and depletion)/Assets
Cash (Cash + Cash equivalents)/Assets

Cash flow Operating income before depreciation/Assetst−1
(Continues)
Panel B. Financial variables

Distance‐to‐default (DD) Merton's (1974) DD calculated as in Bharath and Shumway (2008). The inputs to the DD model are calculated
as follows: market value of equity ($ mn) is calculated as the product of the end‐of‐month share price and
the number of shares outstanding; the face value of debt is calculated as short‐term debt plus one‐half of
long‐term debt; the firm's annual stock return is calculated by cumulating monthly returns over the
previous year; stock‐return volatility is the annualized percent standard deviation of returns and is
estimated from the prior year stock return data for each month; asset volatility is estimated as a value‐
weighted average of equity volatility and debt volatility, with debt volatility measured as
0.05 + 0.25 × equity volatility. The time horizon is set to 1 year.
Hedge ratio The sum of linear hedging instruments and bought put options (in barrels of oil equivalents) divided by
expected production. Expected production is calculated as a weighted average of the current year's and
next year's production, the weights being 75/25 (current year/next year) for Q1, 50/50 for Q2, 25/75 for Q3,
and 0/100 for Q4.
Leverage (Long‐term debt + Short‐term debt)/Assets

Post 1 for Q4:2014–Q4:2015, 0 otherwise
Tobin's Q (Assets – Book value of equity + Market value of equity)/Assets

High production costs 1 if the firm's production cost per barrel is in the top two deciles, 0 otherwise
Neg operating profit firm 1 if the firm's earnings before interest and taxes (EBIT) is negative in at least three quarters in the shock
period (Post), 0 otherwise
Neg cash flow firm 1 if the firm's operating income before depreciation is negative in at least three quarters in the shock period
(Post), 0 otherwise
Production cost Average production cost per barrel produced
Reserves‐to‐production Total proven reserves (Mmboe)/Total production (Mmboe)
ratio
Shale 1 if the number of hits on the search term “shale” in the firm's quarterly report is positive, 0 otherwise
WTI The end‐of‐quarter price of West Texas Intermediate (WTI) oil ($/barrel)
WTIVol The squared daily returns on the WTI oil price summed over all trading days over the previous year
Z‐score Z‐score is the Altman's Z‐score (Altman, 1968), computed as follows:
(1.2 × X1 + 1.4 × X2 + 3.3 × X3 + 0.6 × X4 + 1.0 × X5), where X1 is Working capital/Assets, X2 is Retained
earnings/Assets, X3 is Earnings before interest and taxes/Assets, X4 is Market value of equity/Total
liabilities, and X5 is Sales/Assets.
A P P E N DI X B : AD D I T I O N A L FIG U R E S A ND T A B L E S
See Figure B1 and Tables B1 and B2
F I G U R E B1 Breakeven prices, Q1:2013–Q4:2016. The black solid curve shows the average half‐cycle breakeven crude‐oil price ($/bbl)
across six US regions (Bakken, DJ Basin, East Eagle Ford, West Eagle Ford, Permian Delaware Basin, and Permian Midland Basin) reported
by Bloomberg (only available from January 2014). The gray solid curve shows the Upstream Capital Costs Index (UCCI), while the black
dotted curve shows the Upstream Operating Costs Index (UOCI), both from IHS Markit and indexed to 100 in Q1:2013. The UCCI tracks the
costs of equipment, facilities, materials, and personnel used in the construction of onshore, offshore, pipeline, and LNG projects, while the
UOCI measures operating costs. Both indexes are calculated for a virtual portfolio of 28 onshore, offshore, pipeline, and LNG projects
around the world.
T A B L E B1 Descriptive characteristics for derivative and nonderivative users, Q1:2013‐Q4:2015.

Variable N Mean SD Min Median Max
Assets ($ mn) 2153 4344 9241 2 671 43,439
Leverage 2123 0.380 0.334 0.000 0.329 1.602
Cash 2153 0.082 0.133 0.000 0.024 0.590
Tobin's Q 2150 1.974 2.188 0.521 1.284 12.135
Cash flow 2105 −0.015 0.090 −0.356 0.014 0.081
Capex 2064 0.023 0.111 −0.267 0.022 0.369
Hedge ratio 2054 0.270 0.321 0.000 0.094 1.450
Distance‐to‐default 1451 5.003 5.255 −2.398 4.180 18.175
Reserves‐to‐production ratio 1614 53.094 116.140 0.000 34.397 3866.98
Production cost 1094 18.050 30.559 0.860 13.775 871.820
Note: The descriptive statistics for all firm‐quarters meeting our inclusion criteria (SIC 1311, headquarters in the US; publicly listed; and total assets ≥ $1mn),
irrespective of if they are derivatives users or nonusers are reported. For variable definitions, see Appendix A.
T A B L E B2 Determinants of hedging strategies, Q1:2013–Q4:2015—binomial logit evidence.

Constant −1.042*** −0.824** 0.012 −4.303*** −4.982***
(−4.91) (−2.35) (0.05) (−17.5) (−7.59)
Log(Assets)t−1 −0.033 −0.204*** −0.150*** 0.467*** 0.310***
(−1.46) (−6.22) (−4.59) (12.8) (4.06)
Distance‐to‐defaultt−1 0.020 −0.035 −0.063** −0.058*** −0.088*
(1.46) (−0.92) (−2.42) (−4.11) (−1.86)
Altman Z‐scoret−1 −0.065* 0.062 0.579*** −0.118*** −0.163**
(−1.95) (0.91) (6.29) (−4.13) (−2.23)
N 780 780 780 780 780
2
Pseudo R 0.002 0.028 0.079 0.091 0.047
Note: The coefficients and z‐values clustered by quarter (in parenthesis) from bivariate logit regressions in which indicator variables for the different hedging
strategies are dependent variables (each indicator variable takes the value 1 if the firm uses the indicated hedging strategy, 0 otherwise) are reported. Data for
the balanced subsample comprising firms that use derivatives in at least one quarter and that have at least three observations in both the preshock and shock
periods. For variable definitions, see Appendix A.
A P P E N DI X C : D E FA U L T ID E N T I F I C A T I O N A ND CA T E G O R I Z AT I O N
This appendix describes how we record out‐of‐sample defaults among our sample firms. To investigate post‐oil shock
bankruptcy rates, we screen Capital IQ for defaults on the firm's debt. We record bankruptcy and 8‐K filings for out‐of‐
court debt amendments and settlements. Amendments and settlements include forbearance or exchange agreements,
out‐of‐court foreclosures, or notifications of being in default on a debt contract.17 A firm in distress may be an attractive
takeover candidate, especially in times of industry‐wide distress, so we also record whether the firm was acquired.
Finally, we identify firms that “go dark,” either voluntarily (by filing Form 15) or involuntarily (by having their
registration revoked by the SEC) delisting or by simply stopping filing with the SEC.
We track each of the 96 firms from the estimation sample (described in Table 3), and record out‐of‐sample
information from Q1:2016 to Q4:2017. For each firm‐quarter during this period, we examine whether the firm filed for
bankruptcy, negotiated an out‐of‐court settlement with its creditors, was delisted, or was acquired. We count the
number of these events by hedging strategy and tabulate this information in Table C1.
Based on this data, we estimate default rates. This analysis allows us to evaluate the performance of 3W‐collar users
relative to other hedging strategies. Since default can occur several quarters following the shock and because we have a
balanced sample through 2015, we estimate default rates out‐of‐sample over the period Q1:2016–Q4:2017.18
17
We do not record covenant reliefs such as waivers of leverage covenants. Such waivers were commonplace (Azar, 2017), but they do not consist of a
failure to meet scheduled debt payments.
18
Both bankruptcies and debt contract amendments peaked in the first half of 2016 (Azar, 2017).
T A B L E C1 Acquisitions and defaults, Q1:2016–Q4:2017.

No hedging Linear only Insurance Two‐way collar Three‐way collar Net sold call Total
Acquisition 0 1 0 1 0 0 2
Bankruptcy 1 4 4 3 6 2 20
Settlement 0 2 2 2 3 0 9
Nb. Events 1 7 6 6 9 2 31
Nb. Firms 13 21 11 20 28 3 96
Default rate 7.7% 28.6% 54.5% 25.0% 32.1% 66.7% 30.2%
T test (p value) (0.529) (0.573) (0.929) (0.742) (0.347)
Note: The company exits for the balanced sample in the form of acquisitions, bankruptcies, out‐of‐court agreements, and settlements, delistings, and firms that
stop filing with the SEC over the period Q1:2016 to Q4:2017 are reported. Events are identified by screening Capital IQ for completed acquisitions, bankruptcy
filings, and SEC. 8‐K filings of out‐of‐court agreements and settlements in form of forbearance agreements, exchange agreements, foreclosure of collateral, or
notifications from lenders of breach‐of‐contract. We also read Form 15 for voluntary termination of registration, voluntary or involuntary termination of
registration filings, SEC revocations of registration of registered securities, and firms that stop filing with the SEC. Each event is verified and classified by
reading the related SEC filings. Hedging strategies are classified based on the firm's dominant hedging strategy over Q4:2014–Q4:2015. Default rates equal the
number of events over the out‐of‐sample period over the number of firms that engage in a given hedging strategy based on the Q4:2014–Q4:2015 period. The
last row reports p values for a T‐test statistic that compares default rates of each group relative to the Linear only category.

Journal of Futures Markets - 2022 - Dudley

Uploaded by

Document Information

Original Description:

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

Journal of Futures Markets - 2022 - Dudley

Uploaded by

Copyright:

Available Formats

Received: 21 December 2021 | Accepted: 1 April 2022

How do firms hedge in financial distress?

Evan Dudley1 | Niclas Andrén2 | Håkan Jankensgård2

1324 | wileyonlinelibrary.com/journal/fut J Futures Markets. 2022;42:1324–1351.

TABLE 1 Summary of theoretical predictions.

F I G U R E 1 Tiered demand for liquidity. The demand for

F I G U R E 2 Hedging strategies under tiered

3 | SAMPLE, VARIABLES, A ND DESCRIPTIVE S TATISTICS

3.2 | The 2014 oil price collapse

F I G U R E 3 Crude oil price and volatility,

3.3 | Hedging variables

3.4 | Other variables

3.5 | Descriptive statistics

TABLE 2 Hedging strategies, Q1:2013–Q4:2015.

TABLE 3 Descriptive characteristics, Q1:2013–Q4:2015.

TABLE 4 Mean characteristics per hedging strategy, Q1:2013–Q4:2015.

4 | FINANCIAL DISTRESS A ND HEDGING STRATEGIES

4.1 | Multivariate regressions

TABLE 5 Hedging frequencies for distressed and nondistressed firms.

TABLE 6 Determinants of hedging strategies, Q1:2013–Q4:2015—binomial logit evidence.

4.2 | Multinomial regressions

4.3 | Cash positions and the effect of distance‐to‐default

TABLE 7 Determinants of hedging strategies, Q1:2013–Q4:2015—multinomial logit evidence.

TABLE 8 Cash holdings and hedging strategy choice, Q1:2013–Q4:2015.

4.4 | 3W‐collars and economic distress

TABLE 9 The three‐way collar and economic distress, Q1:2013–Q4:2015.

Panel B. Economic distress and other hedging strategies

4.5 | Risk shifting

4.6 | Collateral constraints

4.7 | 3W‐collars and bankruptcy rates

4.8 | Oil price volatility and break even prices

DATA A VAILABILITY S TATEMENT

APPENDIX A: VARIABLE DEFINITIONS

Bought put 1 if the firm buys put options, 0 otherwise

Cash (Cash + Cash equivalents)/Assets

Panel B. Financial variables

Leverage (Long‐term debt + Short‐term debt)/Assets

Tobin's Q (Assets – Book value of equity + Market value of equity)/Assets

T A B L E B1 Descriptive characteristics for derivative and nonderivative users, Q1:2013‐Q4:2015.

T A B L E B2 Determinants of hedging strategies, Q1:2013–Q4:2015—binomial logit evidence.

T A B L E C1 Acquisitions and defaults, Q1:2016–Q4:2017.

You might also like