Journal of Applied Corporate Finance

c/o Wiley-Blackwell
350 Main Street
Malden, MA 02148-5018
VOLUME 25
|
NUMBER 4
|
FALL 2013
J
O
U
R
N
A
L

O
F

A
P
P
L
I
E
D

C
O
R
P
O
R
A
T
E

F
I
N
A
N
C
E
R
i
s
k

M
a
n
a
g
e
m
e
n
t
V
O
L
U
M
E

2
5


|


N
U
M
B
E
R

4


|


F
A
L
L

2
0
1
3
In This Issue: Risk Management
Navigating the Changing Landscape of Finance
8 James Gorman, Chairman and CEO, Morgan Stanley
Reforming Banks Without Destroying Their Productivity and Value
14 Charles W. Calomiris, Columbia University
How Companies Can Use Hedging to Create Shareholder Value
21 Ren Stulz, Ohio State University
Do Trading and Power Operations Mix?
30 John E. Parsons, MIT Sloan School of Management
Aligning Incentives at Systemically Important Financial Institutions:
A Proposal by the Squam Lake Group
37 The Squam Lake Group
Managing Pension Risks: A Corporate Finance Perspective
41 Gabriel Kimyagarov, Citigroup Global Markets, and
Anil Shivdasani, University of North Carolina at Chapel Hill
Synthetic Floating-Rate Debt: An Example of an Asset-Driven Liability Structure
50 James Adams, J.P. Morgan Securities, and Donald J. Smith,
Boston University
Hedge Fund Involvement in Convertible Securities
60 Stephen J. Brown, New York University; Bruce D. Grundy Uni-
versity of Melbourne; Craig M. Lewis, Vanderbilt University;
and Patrick Verwijmeren, Erasmus University Rotterdam
Fine-Tuning a Corporate Hedging Portfolio: The Case of an Airline
74 Mathias Gerner, European Business School and
Ehud I. Ronn, University of Texas at Austin
A Primer on the Economics of Shale Gas Production
Just How Cheap is Shale Gas?
87 Larry W. Lake, University of Texas; John Martin, Baylor
University; J. Douglas Ramsey, EXCO Resources, Inc.; and
Sheridan Titman, University of Texas
Evidence from German Companies of Effects of Corporate
Risk Management on Capital Structure Decisions
97 Julita M. Bock, Otto von Guericke University
The Simon Business School is ranked #7 in the world
for fnance and #8 for accounting by the Financial Times.
In addition to the fagship two-year MBA program and several
one-year masters degree programs at its main campus
in Rochester, New York, Simon offers one-year, part-time
masters degree programs in fnance and management at its
New York City location. Designed for working professionals,
with alternate weekend classes in midtown Manhattan
and residency weeks to further complement the classroom
instruction, these programs feature senior Simon faculty, small
class sizes, and personalized program support. For additional
details, please visit www.simon.rochester.edu.
APPLI ED CORPORATE FI NANCE
Journal of
Journal of Applied Corporate Finance Volume 25 Number 4 Fall 2013 87
A Primer on the Economics of Shale Gas Production
Just How Cheap is Shale Gas?
1. We want to thank Patrick Gonzalez, John McCormack, Jeffrey Nelson, and L. J. R.
Bert Scholtens for their thoughtful comments on an earlier draft of the paper. Respon-
sibility for any remaining errors, however, is ours.
2. The American Energy Outlook (published by the U.S. Energy Information Adminis-
tration or EIA) estimated total US recoverable US shale gas resources in 2009 to be
269.3 trillion cubic feet, increased its estimate to 317 trillion cubic feet in 2010 and
then doubled that again for 2011 (http://www.eia.doe.gov/forecasts/aeo/). The EIA also
increased its estimates of worldwide recoverable shale gas volumes from 6,622 TCF in
2011 to 7,299 in 2013. Over the same time frame the worldwide recoverable tight oil
reserves increased from 32 million barrels to 345 million. (http://www.eia.gov/analysis/
studies/worldshalegas/).
3. Guy Chazan, Big Oil Heads Back Home, The Wall Street Journal, December 5,
2011, R1.
4. Based upon his analysis of the production history of the Barnett Shale, Arthur Ber-
man, staff member of The Oil Drum (http://www.theoildrum.com), questions whether
shale gas can ever meet the hype surrounding its current popularity. He argues that
rather than 100 year supply of natural gas, shale gas offers as little as seven years.
5. Natural gas has been produced from shale formations for over one hundred years
but recent technological advances have made the economics of its production much
more attractive.
6. Although fracking has been around since the 1940s its use has increased signif-
cantly since 2000 and become a focus of intense media scrutiny. Fracking fuid is 99.5%
sand and water and most of the injected water is later extracted during production. There
is always some risk, however, of ground water contamination, and objections to more
traffc and noise related to drilling. See John Walton and Arturo Woocay, Environmental
Issues Related to Enhanced Production of Natural Gas by Hydraulic Fracturing, Oil, Gas,
& Mining, Volume 1, Issue 1, (August 2013).
7. The Haynesville shale spans a 9,000 square mile area in North Louisiana. Al-
though not as large geographically as the Marcellus shale which covers 95,000 square
miles, it has almost as much gas: an estimated 251 trillion cubic feet (Tcf) compared to
just 262 Tcf for the much larger Marcellus shale.
B
by Larry W. Lake, University of Texas; John Martin, Baylor University; J. Douglas
Ramsey, EXCO Resources, Inc.; and Sheridan Titman, University of Texas
1
T
he U.S. government doubled its estimate of the
countrys economically recoverable natural gas
reserves between 2009 and 2011.
2
Tis increase
is directly related to unconventional gas, specif-
ically gas held in shale formations that are now recoverable
using horizontal drilling and hydraulic fracturing (i.e.,
fracking) techniques developed in the last decade. More-
over, the promise of unconventional gas and oil resources
has led some to predict that the U.S. will be the top global
oil and gas producer, surpassing Russia and Saudi Arabia.
3

Others caution that overly optimistic estimates of shale gas
reserves combined with low gas prices may keep shale gas
from becoming the game changer many believe it will be.
4

Specifcally, there are three big issues that arise with respect
to the production of natural gas from shale:
5

First, traditional methods used to estimate gas
reserves may overstate recoverable shale gas reserves. Since
the horizontal drilling and fracturing methods used to extract
gas from shale formations is relatively new (often less than
5 years old), analysts have limited experience to draw upon
when trying to estimate the volume of technically recoverable
reserves. Moreover, production experience suggests that the
production rate from shale wells declines much more rapidly
than production from vertical gas wells.
Second, the economic viability of producing shale
gas has been questioned. Extracting natural gas from shale
formations is both difcult and expensive. It requires horizon-
tal drilling and fracturing of formations using large quantities
of water. It is difcult to forecast the long-term production
of shale gas wells. Te productive life of shale gas may be
very short with roughly 70% of available reserves extracted
in the frst year of production. Tis means that a depend-
able supply of natural gas from shale formations requires a
sustained program of drilling, completion, and exploration.
Finally, there are serious environmental concerns
about the production of shale gas. Te drilling techniques
used to extract shale gas require large volumes of water and
some chemicals that could result in ground water contamina-
tion. Potential environmental costs could seriously erode the
economic viability of producing gas from shale formations.
6

Ultimately all of these concerns impact the economics
of shale gas production. If shale gas reserves have been over-
estimated then this will result in reduced production and
possibly higher gas prices. Environmental concerns will likely
lead to regulatory changes that result in increased costs of
extracting shale gas.
In this paper we frst develop a base case model to evalu-
ate the economic viability of producing natural gas from shale
formations. To construct our base model we use data from a
shale gas well in the Haynesville shale region of North Louisi-
ana.
7
Our analysis explores the valuation of the predicted gas
volumes from that well using estimated production costs for
the region in conjunction with estimated natural gas prices.
To generalize the model we use both sensitivity analyses of the
key value drivers, as well as simulation analysis. We fnd most
shale gas wells are proftable under the assumed conditions of
our model data. However, the key driver of NPV is the price
of natural gas, which at the time of this writing is very nearly
equal to breakeven levels (assuming the current method used
to estimate production volume is correct).
Te conventional view of a hydrocarbon accumulation
is that three things are required: a source rock, a reservoir,
88 Journal of Applied Corporate Finance Volume 25 Number 4 Fall 2013
8. This is a bit of an oversimplifcation for even with conventional wells the operator
might choose to frack the well after drilling should the well not fow. Thus, conventional
wells offer the operator the option to complete the well which may include fracking. In
contrast, shale wells must be fracked before their economic viability can be determined.
9. The severance tax is a tax imposed by states for the extraction of natural resources
such as natural gas which in the State of Louisiana is $0.0331 per Mcf. The ad valorem
tax is a tax levied on the difference between the price of natural gas and the cost of
production. In Louisiana this tax is levied for each well as follows: In the frst year the tax
is $955 per month. In years two through twenty the tax declines by 4% per year and for
all years thereafter it is equal to the year 20 total.
much higher than in conventional wells that seek out smaller
reservoirs. Tis means that the opportunity to make add-on
investments is greater for shale gas wells than conventional wells.
VII. Exploration costs. The vastness of shale forma-
tions means that there is little discovery risk and few wells are
absolutely unproductive. Tese diferences indicate that the
economics of extracting shale gas can be quite diferent than
that of extracting conventional gas deposits. Te frst fve difer-
ences tend to decrease the value of shale gas in comparison with
conventional formations but the last two favor shale very much.
We built a model for a specifc shale gas well found in
the Haynesville shale region of North Central Louisiana
(Haynesville #1). Te key elements are the production decline
curve forecast for the well, the cost of drilling and completing
the well, the annual operating costs for producing the gas,
and the price of natural gas.
We wanted to build a general model of shale gas produc-
tion to replicate a wide range of conditions encountered across
diferent regions of the country. We frst calibrated our base
model to the actual Haynesville shale conditions and then
vary the key value drivers.
Panel A of Table 1 contains the original data estimates
for the Haynesville #1 well. Tis data is from a single well
and not the entire resource play. In year 0 the well calls for an
investment of $1,912,500 which equals the owners 15.75%
working interest in the well multiplied by the total drilling
and completion costs.
Beginning in year 1 the annual cash fows from the
example gas well are calculated as follows:
(1)
Net Gas
Production
(mcf)
Price of
Natural Gas
per mcf

Total Net
Operating
Expenses
Severance &
Ad Valorem
Expenses
Depletion
Allowance
1-
Corporate
Income Tax Rate
+
Depletion
Allowance

Net gas production is equal to the owners proportionate
working interest in the well revenues (15.75%) multiplied by
the estimated gross annual gas production. Total net operat-
ing expenses include the owners share of the cash expenses
required to extract, transport and sell the gas production
from the well. Severance and Ad Valorem expenses are taxes
imposed on gas producers and the depletion allowance is a
non-cash expense available to gas producers to refect the
depletion of the asset represented by the well.
9

Energy frms have sometimes computed return on invest-
ments in gas wells on a before corporate tax basis. Tis practice
arose, in part, because their investments were often unprof-
itable. Tis has been much less true recently. For example,
Chesapeake Energy paid average tax rates of 38% and 39%
and a seal. Hydrocarbons are generated in the source rock,
usually organic-rich shale, by slow cooking over hundreds
of thousands of years. Once converted, hydrocarbons are
expelled from the source and migrate upwards until encoun-
tering and being trapped by the seal. Te goal of conventional
exploration was to fnd the seal. Te actual location of the
source rock was unimportant; sometimes it was hundreds of
kilometers away from the reservoir, and for many conven-
tional reservoirs the location of the source was unknown.
Source rocks are very widely distributed across the world.
But they were assumed to be so impermeable that many eons
would be required for hydrocarbons to ooze out of them. And
the hydrocarbons were not hydrocarbons in the usual sense;
they were only precursors to hydrocarbons.
Tat was the conventional view. We now know that
source rocks contain a good deal of hydrocarbon itself,
especially natural gas, not just its precursor. Source rocks are
essentially impermeable however, but we can deal with this by
brutalizing the rock through horizontal drilling and fractur-
ing (often referred to as fracking). Interestingly it remains
true that source rocks are abundant.
Te economics of conventional natural gas production
difers from unconventional shale gas in several ways:
I. Total production volumesa conventional gas well
might produce 30 to 40 billion cubic feet of gas over its life
whereas a shale well would produce a fraction of this amount.
II. Rate of decline in production volumesshale gas
wells have a very steep rate of decline compared to conven-
tional wells, especially in the initial production period.
III. Production methodsshale gas is trapped in rock
formations that must be fractured before the gas can fow.
Fracturing, which involves injecting water and sand at high
pressure into the formation, is expensive and may entail
environmental risks.
IV. Horizontal wellsshale gas production typically uses
horizontal drilling whereas conventional gas wells are drilled
using vertical wells.
V. Completion after drillingthese wells must be
fractured before the viability of producing the well can be
determined. Tis means that the decision to drill the well is
tantamount to a commitment to complete the well. Tis is
diferent than a conventional gas well in which a drilling log and
pressure measurements can be used to estimate the volume of
recoverable reserves before the well is completed.
8
If the well is
deemed uneconomic it can be plugged and abandoned thereby
avoiding completion expenses.
VI. Follow on investmentssince shale formations are
typically very large, the probability of success of follow on
wells (in what is commonly referred to as the resource play) is
89 Journal of Applied Corporate Finance Volume 25 Number 4 Fall 2013
Table 1 Base Case Model for Haynesville #1

Panel A Cash fow estimates
Date Period Estimated
Gross Gas
(Mcf)
Estimated Net
Gas
(15.75% of
Gross Gas)
(Mcf)
Price of
Natural Gas/
Mcf
Estimated
Total Net
Revenue
Estimated
Total Net
Operating
Expense
Estimated
Severance &
Ad Valorem
Expense
Estimated
Depletion
Allowance
Estimated
Equity
Investment
Estimated
After-tax Net
Cash Flow
Dec-09 0 $(1,912,500) $(1,912,500)
Dec-10 1 2,405,443.0 378,857.3 $4.50 $1,704,858 $(13,568) $(1,908) $(483,695) 1,327,676
Dec-11 2 773,147.5 121,770.7 4.50 547,968 (11,833) (36,625) $(155,467) 396,298
Dec-12 3 474,995.6 74,811.8 4.50 336,653 (9,747) (25,794) $(95,514) 239,432
Dec-13 4 343,838.1 54,154.5 4.50 243,695 (9,747) (19,644) $(69,140) 170,755
Dec-14 5 269,659.5 42,471.4 4.50 191,121 (9,747) (15,791) $(54,224) 132,176
Dec-15 6 221,876.7 34,945.6 4.50 157,255 (9,747) (13,165) $(44,616) 107,425
Dec-16 7 188,506.2 29,689.7 4.50 133,604 (9,747) (11,265) $(37,906) 90,186
Dec-17 8 163,873.9 25,810.1 4.50 116,146 (9,747) (9,827) $(32,952) 77,486
Dec-18 9 144,941.4 22,828.3 4.50 102,727 (9,747) (8,701) $(29,145) 67,739
Dec-19 10 129,934.0 20,464.6 4.50 92,091 (9,747) (7,797) $(26,128) 60,021
Dec-20 11 117,744.9 18,544.8 4.50 83,452 (9,747) (7,054) $(23,677) 53,759
Dec-21 12 107,647.8 16,954.5 4.50 76,295 (9,747) (6,433) $(21,646) 48,575
Dec-22 13 99,146.5 15,615.6 4.50 70,270 (9,747) (5,906) $(19,937) 44,213
Dec-23 14 91,890.3 14,472.7 4.50 65,127 (9,747) (5,454) $(18,478) 40,492
Dec-24 15 85,870.8 13,524.7 4.50 60,861 (9,747) (5,076) $(17,267) 37,406
Dec-25 16 80,858.5 12,735.2 4.50 57,308 (9,747) (4,761) $(16,259) 34,838
Dec-26 17 76,149.7 11,993.6 4.50 53,971 (9,747) (4,463) $(15,312) 32,426
Dec-27 18 71,715.1 11,295.1 4.50 50,828 (9,747) (4,182) $(14,421) 30,155
Dec-28 19 67,538.7 10,637.3 4.50 47,868 (9,747) (3,916) $(13,581) 28,017
Dec-29 20 63,605.6 10,017.9 4.50 45,080 (9,747) (3,665) $(12,790) 26,005
Dec-30 21 59,901.5 9,434.5 4.50 42,455 (9,747) (3,428) $(12,045) 24,109
Dec-31 22 56,413.1 8,885.1 4.50 39,983 (9,747) (3,204) $(11,344) 22,325
Dec-32 23 53,127.8 8,367.6 4.50 37,654 (9,747) (2,993) $(10,683) 20,645
Dec-33 24 50,033.9 7,880.3 4.50 35,462 (9,747) (2,793) $(10,061) 19,063
Dec-34 25 47,120.2 7,421.4 4.50 33,396 (9,747) (2,605) $(9,475) 17,574
Dec-35 26 44,376.1 6,989.2 4.50 31,452 (9,747) (2,427) $(8,923) 16,171
Dec-36 27 41,791.8 6,582.2 4.50 29,620 (9,747) (2,259) $(8,404) 14,851
Dec-37 28 39,358.1 6,198.9 4.50 27,895 (9,747) (2,101) $(7,914) 13,607
Dec-38 29 37,066.0 5,837.9 4.50 26,271 (9,747) (1,952) $(7,453) 12,436
Dec-39 30 34,907.5 5,497.9 4.50 24,741 (9,747) (1,811) $(7,019) 11,334
Dec-40 31 32,874.6 5,177.8 4.50 23,300 (9,747) (1,678) $(6,611) 10,296
Dec-41 32 30,960.1 4,876.2 4.50 21,943 (9,747) (1,553) $(6,226) 9,318
Dec-42 33 29,157.2 4,592.3 4.50 20,665 (9,747) (1,434) $(5,863) 8,398
Dec-43 34 27,459.2 4,324.8 4.50 19,462 (9,747) (1,323) $(5,522) 7,531
Dec-44 35 25,860.1 4,073.0 4.50 18,328 (9,747) (1,218) $(5,200) 6,714
Dec-45 36 24,354.1 3,835.8 4.50 17,261 (9,747) (1,119) $(4,897) 5,946
Dec-46 37 22,935.8 3,612.4 4.50 16,256 (9,747) (1,025) $(4,612) 5,222
Dec-47 38 21,600.2 3,402.0 4.50 15,309 (9,747) (937) $(4,343) 4,540
Panel B Project Valuation (Cost of capital = 10%)
Corporate Tax Rate, % 30 0
NPV, $ 309,201 893,995
IRR, % 16.47 30.86
90 Journal of Applied Corporate Finance Volume 25 Number 4 Fall 2013
10. Based on a study published in April 2009 and prepared for the U.S. Department
of Energy Offce of Fossil Energy and National Energy Technology Laboratory titled Mod-
ern Shale Gas Development in the United States: A Primer, p. 17.
11 Larry Benedetto, Unconventional Gas and its Impact on Domestic Supply, a re-
port issued by Howard Weil (http://www.ocsbbs.com/hottopics/2008/PLANO_Larry_
May2008.pdf)
12. This is the cost estimate of our example well. However, the average cost of shale
wells in the Haynesville shale has been reported to be somewhat higher. In 2010 both
Petrohawk (HK) (Enercom Oil & Gas Conference: August 24, 2010) and EXCO (XCO)
(Fourth Quarter and Full Year 2009 Review: February 2010) reported an average cost of
their shale gas wells in the Haynesville Shale of $9.5 million.
13. R.E. Allen is credited with mentioning four types of decline curves in 1931 and
that J.J. Arps later expounded upon. See R. E. Allen, Control of California Oil Curtail-
ment, Trans. A.I.M.E., 92, 47, 1931 and J. J. Arps, Estimation of Primary Oil Re-
serves in Petroleum Transactions, AIME, Vol. 207, 1956.
14. The method we use to estimate production volumes follows industry practice;
however, new methods have been proposed that claim to improve the accuracy of vol-
ume projections. These include the one mentioned in the M.S. thesis of A.J. Clark. See
A. J. Clark, Decline Curve Analysis in Unconventional Resource Plays using Logistic
Growth Models, M.S. thesis, University of Texas at Austin, 2011; A. J. Clark, Larry W.
Lake, and Tad Patzel, Production Forecasting with Logistic Growth Models, SPE
144790, presented at the 2011 Annual Technical Conference and Exhibition of the So-
ciety of Petroleum Engineers, Denver, Colorado; and Peter Valko and W. John Lee, A
Better Way to Forecast Production from Unconventional Gas Wells, SPE134231, pre-
sented at the 2010 E Annual Technical Conference and Exhibition of the Society of Pe-
troleum Engineers, Florence, Italy.
completing the well is not as valuable for a shale gas well as
it is for a conventional gas well.
In a typical shale gas well in the US, the investor will
have a complete working interest. A working interest less than
100% can represent an attempt to diversify risk exposure
or simply to accommodate capital constraints. For example,
in the model we assume that the investor owns 15.75% of
well revenues but its share of expenses is 22.5% since royalty
owners do not share the expenses.
Gas well revenues are a function of two key variables:
Te price per thousand cubic feet (Mcf) of natural gas sold,
and the volume of gas produced. We will assume that the
price of natural gas is fxed at $4.50/Mcf over the life of the
investment just as was done by the investor in the example
well described in Table 1. Although gas prices can fuctuate
widely, it is common practice to use rolling, forward contracts
dated out to fve years to hedge price risk. Te fve-year gas
price assumption should refect forward market prices at the
time of evaluation since that is what can be hedged.
The second key variable is the volume of gas it will
produce. Te common methods used to estimate oil and gas
reserves rely on an empirical extrapolation based on physical
characteristics of the reservoir. Over sixty years ago, J.J. Arps
defned a set of empirical production decline curves based on
the following hyperbolic function:
13
q
t
= q
i
(1+nD
i
t)

1
n
(2)
where q
t
is the production rate at time t (i.e., Mcf/year); q
i

is the initial production rate at time t= 0; D
i
and n are two
constants (the former is the initial rate of decline in produc-
tion and n is the rate of change in D
i
over time); and t is the
time period for which production is being estimated. Te
Appendix discusses the particulars of applying the hyperbolic
decline curve to production estimates found in Panel A of
Table 1.
14

Te production estimates for Haynesville #1 are based
on a combination of two types of production decline curves:
Initially production declines based on a hyperbolic decline
curve with the following parameters:
q
i
(initial production rate at time 0) = 19,500 Mcf/day;
D
i
(initial rate of decline in production) = 85% for the
frst year;
in 2010 and 2011 respectively. We believe our assumption of
a 30% corporate tax rate is a conservative one.
Panel B of Table 1 contains the estimated net present value
(NPV) and internal rate of return (IRR) for the well. Te NPV
is $309,201 and the IRR 16.47%, which exceeds the 10% cost
of capital used in the analysis. If one were to assume a zero
tax rate, the NPV would be $893,995 and the IRR 30.86%.
Obviously, the tax assumption will have a very large impact
on the reported results.
Table 2 presents the results of the models forecast of the
cash fow found in the Haynesville #1 well described in Table
1. Tere are three basic components that must be modeled
to evaluate an investment in a shale gas well: Drilling and
completion costs, gas revenues, and operating costs. Lets
consider each in turn.
Drilling costs are directly related to the depth of the shale
formation and the number of fractures or fracks used to
provide a permeable path for the gas to fow through the
tight shale formation. Te number of fracks in the pay zone
tends to be 8-10 for most wells. However, the depth of the
shale formation varies from one shale area to another. For
example, the depth of the shale formation in the Haynesville
region is 10,000-13,500 feet whereas in the Fayetteville shale
the formations are 1,000-7,000 feet and in the New Albany
shale, the wells can be as shallow as 500 feet.
10
To illustrate
the range of drilling costs, the cost of drilling and completing
a well in the Woodford shale (Oklahoma), where the shale
is 6,000-11,000 feet deep is estimated to be $6.7 million,
11

whereas the cost of drilling and completing a new well in the
Haynesville shale is estimated to be $9.5 million.
12
Te cost of
Haynesville #1, however, was slightly lower than the average
at $8.5 million. Note that drilling costs change from month
to month and also vary by location. Te cost of drilling the
well is approximately 60% of this total and completion costs
make up the remaining 40%. For conventional gas wells the
operator has a valuable option to shut-in the well if measure-
ments taken during drilling indicate it will be uneconomic.
Consequently, the division between drilling and complet-
ing costs is very important. With shale wells, however, the
productivity of a shale gas deposit cannot be known until
after the well has been fractured and this entails incurring
the bulk of the costs of completing the well. Consequently,
the option to shut-in the well prior to incurring the costs of
91 Journal of Applied Corporate Finance Volume 25 Number 4 Fall 2013
15. The slope term, n, being equal to one indicates that the hyperbolic decline func-
tion reduces to the special case of a harmonic decline function, which is discussed in the
appendix.
16. Given the relatively brief time that horizontal, hydraulically fractured gas wells
have been producing, the productive life of these wells is still unknown. Although con-
ventional wells have produced for 30 years and longer, the initial evidence from the
Barnett Shale is that the average productive life of a well is 7.5 years and the early evi-
dence from the Marcellus suggests its wells also might have similar (short) lifespans (see
www.marcellus-shale.us/Marcellus-production.htm). A short life span does not mean
these wells are not economic, but to be proftable the volumes of initial production have
to carry the investment.
17. This switch to the exponential decline function is important because a hyperbolic
curve with value of n greater than one implies an infnite recovery amount, (i.e., com-
monly referred to as the wells estimated ultimate recovery or EUR).
18. Although we report actual production volumes on the vertical axis, engineers
typically report log production. This practice probably arose out of the convenience where
an exponential production decline function was used and the log transformation resulted
in a linear or near-linear function that was easy to extrapolate future production by hand.
19. There are two methods for computing depletion: the cost method and percent of
revenue method. The cost method, which we use, bases the allowance on the original
cost of the income-generating property and any subsequent capitalized costs incurred
(e.g., work over expense) can be used by all energy companies. The percent of revenues
method is not tied to the original cost of the well such that the total depletion for a well
is not limited to the original investment in the well using this method. The percent of
revenue method can be used by an independent producer or royalty owner. However,
certain refners and certain retailers and transferees of proven oil and gas properties can-
not claim percentage depletion. We use cost depletion to be conservative.
model total net operating expenses as a percent of revenues
for the frst three years followed by a fxed dollar amount for
years four and beyond that is equal to the year three expense.
Te percentages of revenues used in the analysis are consis-
tent with actual estimates for the example well whose data
provides the basis for our model.
We combine severance and ad valorem taxes and model
them as a quadratic function of net gas revenues as shown
in Figure 2. Te ft of the relationship to the data provided
for Haynesville #1 was nearly perfect. Tis is unsurprising
as these taxes are a function of revenues that are estimated
using the production data provided to us for Haynesville #1.
Te production estimates were made using the production
technique described in the Appendix. In 2010 the sum of
these taxes is much smaller than in subsequent years because
in Louisiana, where Haynesville #1 is located, there is no
severance tax for the frst years production.
Te depletion allowance is like depreciation expense in
that it represents the allocation of the cost of drilling and
completing the gas well against revenues over the life of the
well. For modeling purposes we use cost depletion expense
per unit of production for the year multiplied by the total
amount of gas produced.
19

and n (the rate of change in D
i
) = 1.0.5
Te production decline curve switches to an exponential
decline when the annual rate of decline in production falls to
7% at which time the decline function switches to the follow-
ing exponential function with a rate of decline of 6%, i.e.,
16
q
t
=q
t-1
e
.06
Te switch from the hyperbolic to the exponential decline
function occurs between years 13 and 14 for Haynesville #1.
17
Figure 1 contains the estimated production volumes for
2010 through 2047.
18
Note the initial steep production rate
decline as noted above. Using estimates based on this hyper-
bolic decline curve in the model we can later consider the
efects of deviations from the estimated parameters on the
value of the shale gas well.
Multiplying the annual production volumes by the $4.50/
Mcf price for natural gas provides an estimate of the gross
gas revenues from the well. Adjusting these estimates for the
working interest of the investor produces the net gas revenue.
Tere are three categories of annual operating expenses
(see Equation 1). Tese are Total Net Operating Expense;
Severance and Ad Valorem Expense; and Depletion Allow-
ance. Based on the estimates provided for Haynesville #1 we
Figure 1 Estimated Gross Gas Production (Mcf/year) for Haynesville #1.


y = 2,015,779.16x
-1.19

R = 0.99
0
500,000
1,000,000
1,500,000
2,000,000
2,500,000
3,000,000
0 5 10 15 20 25 30 35 40
Annual Production Estimates
G
r
o
s
s

G
a
s

P
r
o
d
u
c
t
i
o
n

(
M
c
f
/
Y
e
a
r
)
Production function switches from
hyperbolic to exponential between
years 13 and 14
92 Journal of Applied Corporate Finance Volume 25 Number 4 Fall 2013
20. Recall that the production decline function used to model Haynesville #1 is a
hybrid model containing a mixture of a hyperbolic and exponential function. In the sen-
sitivity analysis used here we do not alter the exponential portion of the decline curve,
which describes the tail of the curve. In the initial estimates we found that the exponen-
tial function is used between 13 and 14 years in to the productive life of the well.
21. Recall from equation (2) that there is a third variable in the hyperbolic decline
curve equation. This variable determines the rate of change in the annual decline rate in
production and was designated by n. When we evaluated the sensitivity of NPV to this
variable we found that there was no positive value for n for which NPV was equal to zero.
NPV declined with smaller values of n, but it reached a minimum near $100,000 as n
approached zero. Remember that we were only allowing n to vary while holding all other
variables constant so that estimates of this variable are important even though in this
example this variable was not able to drive the NPV to zero.
22. We used a switching model that is initially a hyperbolic function and then con-
verts to an exponential function where the rate of decline in production falls to 6%.
23. John Dizard of the Financial Times described the dissention within the petroleum
engineering profession in 2010 regarding the amount of natural gas that will ultimately
be recovered from shale formations and at what cost. See Debate over Shale Gas De-
cline Fires Up, FT.com (October 10, 2010).
Table 3 Sensitivity Analysis of the Production Estimate
Parameter Original
Parameter
Estimate
Break-even
Parameter
Value
Percent
Change in
Parameter
Value
Percent
Change in
Estimated
Ultimate
Recovery
(EUR)
Initial
Production
Rate (q
i
)
19,500 Mcf/
day
16,551 Mcf/
day
-15.12% -15.12%
Initial Decline
in Production
in Year 1 (D
i
)
85% 87.7% 3.15% -13.2%
Clearly, the NPV of the project is very sensitive to produc-
tion volumes. An increase in the initial rate of decline in
production for year one from 85% to 87.7% (an increase of
just 3.15%) leads to a zero NPV estimate. Furthermore, a
drop of 15.12% in the initial rate of production from 19,500
Mcf/day to 16,551 Mcf/day can also result in a zero NPV.
21

Although we have used the same hyperbolic decline
function (as in equation 2) widely used with conventional
gas wells,
22
to estimate the output of our shale gas well, we
do not yet know whether this or some other function is more
appropriate for unconventional shale gas wells. Some geosci-
entists suggest that a more rapid exponential function would
be more appropriate. Only time will tell as we gain more
production experience.
23

Table 2 shows the NPV and IRR generated by the model.
Te model estimates are very close to the calculated values for
NPV and IRR found when using the original data supplied
for the project. As we estimate drilling plus completion costs
total $1,912,500, our share of the 16.17% increase in these
costs would reduce the NPV to zero.
Table 2 Summary Measures for the Original Data and
Model of Haynesville #1
Original Data Estimates Model Estimates
NPV, $
308,222

309,201
IRR, %
16.45

16.47
Te wells NPV is sensitive to two key parameters of the
hyperbolic production decline curve found in equation (2).
Tese are the initial production rate and the initial decline
in production for year 1.
20
Table 3 shows how much each
of the two parameters would have to change to reduce the
NPV to zero.
Figure 2 Actual and Estimated Severance and Ad Valorem Expenses.

Equation in the fgure is for the ftted line.
y = 0.00x
2
- 0.13x - 1,654.88
R = 0.99
100,000 200,000 300,000 400,000 500,000 600,000
S
e
v
e
r
a
n
c
e

&

A
d

V
a
l
o
r
e
m

T
a
x
e
s


Net Gas Revenues
(40,000)
(35,000)
(30,000)
(25,000)
(20,000)
(15,000)
(10,000)
(5,000)
0
2
93 Journal of Applied Corporate Finance Volume 25 Number 4 Fall 2013
24. See the appendix for a description of the method used to calculate q
i
25. Technically we are modeling forward prices or future spot prices for different dates
in the future. Following common practice we assume that the mean drift or rate of
change in prices over the forecast period is zero. See C. Blanco, S. Choi, and D. Soronow,
Energy Price Processes Used for Derivatives Pricing & Risk Management, Commodities
Now, (March 2001), 74-80.
26. The simulated value of the mean NPV and the original estimate diverged because
of skewness in the probability distributions used in the simulation.
27. For an excellent overview of the use of option pricing analysis to evaluating oil and
gas investments, see J. McCormack, Raoul LeBlanc, and Craig Heiser, Turning Risk into
Shareholder Wealth in the Petroleum Industry, Journal of Applied Corporate Finance,
15, 2 (Winter 2002).
28. This discussion follows the dichotomy set forth by S. Titman and J. Martin, pages
437-439 of Valuation: The Art and Science of Corporate Investment Decisions 2
nd

edition, (Upper Saddle River, NJ: Prentice Hall, 2011).
Producers often use a gas price assumption that matches
the forward prices at which the frms could actually hedge
future price risk (i.e., enter into forward contracts to sell gas).
However, production for Haynesville #1 is estimated out 38
years, which is far beyond any frms ability to lock in gas
prices such that the frm is subjected to price risk.
To model natural gas prices, we use a geometric Brown-
ian motion model
25
that has two parameters: the mean rate
of change and standard deviation in the rate of change in gas
prices (volatility). We assume that the mean rate of change
in future gas prices is zero and the annual volatility in future
gas price changes is modeled using a triangular distribution
using an assumed minimum value of .05, most likely value
equal to .10, and maximum value of .30.
Figure 3 shows frequency distributions for the NPV of the
shale gas well. Te expected NPV is $267,776 but the proba-
bility of a NPV of zero or greater is 60.13%, meaning there
is a 39.87% probability that the well will produce a negative
NPV. Tis estimate of project NPV is about $40,000 lower
than the earlier deterministic estimate. Te earlier estimate,
however, corresponds to the results from one particular gas
well in the Haynesville shale, whereas we have, through the
engineers experience, modeled the experience of several wells
throughout the entire area.
26

It has long been recognized that oil and gas production
typically ofers valuable real options to the developer.
27
Tese
options can be thought of in terms of options that exist before
a shale gas drilling project has been launched and options that
arise after the investment launch.
28

We assumed a constant natural gas price for all future
periods equal to $4.50 per Mcf. Should gas prices drop
15.66% to an average of $3.80, the NPV of the well falls
below zero. Historically, gas prices have been volatile. Te
average annual per Mcf well-head price of natural gas was
$4.00 in 2001 but rose to more than $10.00 in 2008 and
then collapsed to $3.60 in October 2013.
Although the sensitivity analysis provides some indication
of the range of outcomes, it says little about the probabilities
of extreme outcomes. To gain insight into possible extreme
outcomes we used Monte Carlo simulation. We chose the
triangular distribution to model uncertainty in the three
key production function variables (the initial production
rate (q
i
), the initial decline in production experienced in the
frst year of production (D
i
), and the slope coefcient that
determines the change in the rate of decline in production
(n)). Te triangular distribution ofers fexibility in model-
ing a wide variety of outcomes. Knowledgeable engineers
and management personnel easily understand its parameters
(minimum, most-likely, and maximum) and they can readily
provide estimates. Table 4 contains the values used to defne
the distributions for each of these parameters.
Table 4 Parameter Estimates for the Production
Function Variables
Parameter Minimum Value Most-Likely Value Maximum Value
Initial Production
Rate (Mcf/day) (q
i
)
4,053 19,500 31,616
Initial Decline in
Production in Year 1
(D
i
), %
76.5 85 93.5
Slope Coeffcient (n) 0.50 1.0 1.5
Te parameter estimates for the initial production rate
(q
i
) are based upon a conversation with the chief reservoir
engineer for the Haynesville gas well. He indicated that the
initial year of production for wells such as ours could be as
small as 500,000 Mcf (0.5 Bcf ) or as large as 3,900,000
Mcf (3.9 Bcf ) although the most likely level of production
is roughly 2,500,000 Mcf (2.5 Bcf ). Tese annual produc-
tion estimates correspond to the three daily production
estimates found in Table 4 for the minimum, most likely, and
maximum values.
24
For the distribution around the initial
decline in production in year one, we assumed maximums
and minimums that were 10% larger and smaller than the
most likely value.
Figure 3 Frequency Distribution of NPV
94 Journal of Applied Corporate Finance Volume 25 Number 4 Fall 2013
29. See J. McCormack and Gordon Sick, Valuing PUD Reserves: A Practical Applica-
tion of Real Option Techniques, Journal of Applied Corporate Finance, 13, 4 (Winter
2001), 110-115.
30. Some shale plays, such as the Haynesville shale, produce only dry gas, whereas
the northern most region of the Eagle Ford Shale in South Texas produces oil, the middle
region produces wet gas (some liquids) and the southernmost region is dry gas only.
of the timing option. However, researchers have not found
evidence of a strong association between drilling costs and
the price of oil and gas.
29

Finally, exercising a timing option in a shale gas play
is made difcult by virtue of the fact that it involves the
coordination of multiple outside contractors including drill-
ing contractors and companies that provide fracking services.
Since it can be very difcult to move drilling and fracking
rigs around the country in a timely way, this friction (and its
associated cost) diminishes the value of the timing option.
Operating options refer to the potential to design a drill-
ing and production program in such a way as to provide the
frm with the fexibility to respond to changing economic
circumstances. For example, the development of drill-
ing pads for drilling multiple shale gas wells from a single
location can dramatically reduce the cost of drilling out a
lease when compared to conventional drilling methods.
Once a gas well has been drilled there are a host of difer-
ent sources of operating fexibility or options available to
the operator of the well. Some of the more important types
of options include growth options, shutdown options, and
abandonment options.
Because the areas covered by shale formations are so
large, shale gas investments ofer a potentially valuable option
to extend production once a viable well has been developed.
Tis is due to the high probability of successful wells in the
immediate vicinity of a successful well. Tis attribute of shale
wells has important implications for the option value of shale
versus conventional gas wells that favors shale.
In addition to the timing option, the operator has an
option to select the site that is most advantageous at the
time of the drilling. For the most part, the selection is deter-
mined by the mix of hydrocarbons at the potential sites and
the price of those hydrocarbon resources at the time of the
drilling.
Shale formations not only yield dry gas but often produce
hydrocarbon liquids, so called rich gas or gas condensates, as
well as oil. Depending on where you drill, you will produce a
mix that may be almost exclusively dry gas, or may be almost
exclusively liquids. Hence, once the play has been explored
and liquid rich areas identifed, the developer has the option
to choose specifc sites depending on the relative prices of the
various hydrocarbon liquids and gas that will be produced.
30

Te source of value to this option arises from the volatil-
ity of the market price diference between hydrocarbon
liquids and gas. In recent history this diference has been
quite volatile, and currently, the market price of the liquids
is quite high relative to the gas. For example, the energy
equivalent of 1 barrel of hydrocarbon liquids is about 7 Mcf
gas. Using this fgure and $5.3/Mcf, the price of 1 barrel of
Prior to drilling, both conventional and shale gas wells
possess three fundamental types of options: staged investment
options, timing options, and operating options. However,
there are diferences in value between the options attached
to conventional and shale gas wells.
Oil and gas drilling programs provide classic examples
of staged investment options. Initial exploratory wells deter-
mine the potential of an investment play before full drill out
and production commences. Te sheer size of shale forma-
tions make the likelihood of successful follow-on wells quite
valuable. For example, the Marcellus shale formation covers
an area of more than 90,000 square miles. Once a productive
site has been identifed, multiple horizontal wells are often
drilled from the same site going out in diferent directions
into the shale formation.
Timing options arise out of the legal right to postpone an
investment in order to maximize value. Both conventional
and shale gas well investment options involve fnite term drill-
ing leases that typically extend for three years. If the energy
frm fails to drill within this three-year window, its lease
with the land-owner lapses. So there is substantial pressure
on the lessee to do at least minimal exploration in order to
hold the lease to enjoy the opportunity to fully develop a
property at some date. But the full production potential of
the property does not have to be exploited in the initial lease
period. Once some minimal level of success has been estab-
lished then the option to defer or delay investing in drilling
out the property is fully available to both the conventional
and shale gas producer.
Because shale gas wells produce high volumes in the frst
year and because they usually also present extensive follow-
on drilling opportunities, they may possess an important
timing option not possessed by conventional wells. Tat is,
shale gas producers may have the ability to time the drilling
of shale gas wells in response to short-term price movements
for natural gas. Te key determinants of the timing option
are the cost of drilling and completing the shale gas well,
the likelihood of success, and the volatility of gas prices. In
a world of highly volatile natural gas prices the option to
develop gas producing properties quickly and with a high
degree of certainty as to the productive capacity of the new
wells in response to price spikes can be extremely valuable.
Tis analogy is very similar to the use of electric generation
peaker plants that can be switched on and of in response
to price spikes for power.
Note, however, that the value of the timing option is
diminished if drilling costs are positively correlated with
the price of natural gas. If a high gas price results in a high
demand to drill gas wells, and this leads to an increase in the
rates charged for drilling rigs, this would reduce the value
95 Journal of Applied Corporate Finance Volume 25 Number 4 Fall 2013
31. We are grateful to John McCormack for pointing out this second type of shut-in
option to us. See J. McCormack and Gordon Sick, Valuing PUD Reserves: A Practical
Application of Real Option Techniques, Journal of Applied Corporate Finance, 13, 4
(Winter 2001), 110-115.
its relative infancy, and it remains to be seen whether the
production curves of these wells will follow this path. Recog-
nizing this, we used diferent parameter estimates for the
production decline curve. Finally, the price path of natural
gas is modeled using a simple difusion process that does not
attempt to incorporate the possibility of shifts in price paths
because of major changes in either the supply of, or demand
for, natural gas.
Conclusion
Breakthroughs in shale gas extraction techniques such as
horizontal drilling and fracturing (fracking) have dramat-
ically increased estimates of recoverable gas reserves in the
US. Although some observers have concluded from this that
the US will enjoy great supplies of low-priced gas for a long-
time to come, others are not so sure.
Te model of US shale gas drilling economics and NPV
simulations presented here indicate that shale gas exploitation
is probably sustainable (with a 60% likelihood) but major
questions remain. NPVs are highly sensitive to gas price
assumptions and projected production volumes. Te base case
assumes the current gas futures price curve, but a fall in price of
just 17% along that curve would reduce the well NPV to zero.
Te model uses standard engineering assumptions about
conventional wells for the relationship between frst year
production declines and subsequent production. Although
it is clear that production from shale wells declines faster
than from conventional wells, engineers have too little history
to forecast ultimate production from shale wells with great
confdence.
Nevertheless, shale gas drilling opportunities also present
energy companies with valuable follow-on real options that
are not captured in NPV analysis. Tis additional source
of value is inherent in vast shale gas formations where one
successful well leads to additional development opportunities
on attractive terms.
Larry W. Lake holds the Sharon and Shahid Ullah Chair in Petroleum
Engineering at the University of Texas at Austin.
John Martin is Collins Professor of Finance at Baylor University.
J. Douglas Ramsey is Director of Strategic Planning and Special Proj-
ects at EXCO Resources, Inc.
Sheridan Titman holds the McAllister Chair in Finance at the Univer-
sity of Texas at Austin.
liquid should be about $35/barrel, a fgure that is about 1/3
of the world oil price. It is no surprise that operators in shale
gas plays are moving to liquids production when it is possible
and deemphasizing dry gas production. However, the price
ratio of gas and liquids can easily change over time, and if
the gas prices change, drillers have the option to move their
rigs to areas with dryer gas.
Te option to shut down operations or shut in a gas well
comes in two basic forms: temporary and permanent. Te
ability to temporarily shut in a gas well is a timing option.
Te owner can produce gas when prices are high and shut in
when prices are very low. To a limited degree it has long been
recognized that conventional gas wells ofer some degree of
fexibility in timing production. But the timing option for
shale gas wells is not yet fully understood because horizon-
tal drilling and fracking technology is relatively new. We
do not know what will happen to production volumes if
wells are shut down for a period of time and then re-opened.
Conceivably, intermediate ways of shutting down wells such
as choking may actually increase the long-term production
of the shale gas well.
Both shale and conventional gas wells that also produce
liquids through pumping present opportunities to tempo-
rarily suspend production. Tese shut in wells can later be
brought back on line should gas prices rise to economic levels.
Wells that require pumping are more costly to operate than
dry gas wells that do not, and sufciently high costs may
make shutting the well in economic. It is not clear though,
whether the well will produce at the same level just prior
to the shut in after it is re-opened.
31
It is clear that options
attached to both conventional and shale gas wells are poten-
tial sources of value but we cannot generalize. Each drilling
play must be evaluated individually.
Our most important fnding is that a large number of
drilling opportunities such as those in Haynesville shale
region of North Louisiana have positive net present values
even in a historically low gas price environment. Neverthe-
less, we acknowledge that our analysis is limited in some
important ways.
First, our model used production and cost estimates
from a single well. Tat said, our results are more generally
applicable because of our thorough sensitivity and simula-
tion analysis. Second, we did not consider the possibility
of producing very valuable liquids along with gas. The
economics of liquids are critical in certain shale regions
such as the Eagle Ford of south Texas. Tird, we based our
production decline curve on the combination of a hyper-
bolic and exponential curve that extended out over three
decades. However, the Haynesville shale production is in
96 Journal of Applied Corporate Finance Volume 25 Number 4 Fall 2013
32. R.E. Allen described four types of production decline curves in 1931. J.J. Arps
later expounded upon three of these (the exception being the frst, the constant decline
curve).
33. W.W. Cutler, Jr., Estimation of Underground Oil Reserves by Well Production
Curves, USBM Bull, 228 (1924).
34. Petrohawk Energy Corporation, Enercom Oil & Gas Conference presentation (April
24, 2010).
35. A. J. Clark, Decline Curve Analysis in Unconventional Resource Plays using Lo-
gistic Growth Models, M.S. thesis, University of Texas at Austin, 2011.
Equation (1A) is the basis often used for reporting of
decline curves to the media. For example, the Petrohawk
Energy Corporation (HK) in a report dated April 24, 2010,
said that its initial recovery wells in the Haynesville shale had
an initial production rate of 8.6 Mmcf/d, an initial decline
rate of 50%, and a hyperbolic exponent (n) of 0.9.
34
Figure 1A
contains the production decline curve for the above inputs:
Te estimated ultimate recovery (EUR) for the Petro-
hawk example reported in Figure 1A was 9.9 Bcf.
When the hyperbolic exponent n is greater than one Eq.
(1A) extrapolates to an infnite estimated ultimate recovery
(EUR). Te EUR is the integral of Eq. (1A) over all time.
When this occursand it appears to for more than one-half
of wells examined
35
the production decline curve must be
converted to another type of function at some point so as to
restrict the EUR to be fnite. Te switching point is arbitrarily
set. For alternates that do not have this arbitrariness see the
papers already mentioned by Clark, Lake, and Patzel and by
Valko and Lee.
AppendixProduction Decline Curves
Te methods used by reservoir engineers to estimate oil and
gas reserves often rely on an empirical extrapolation based on
physical characteristics of the reservoir. Over sixty years ago,
J.J. Arps defned a set of empirical production decline curves
based on the following three-parameter hyperbolic function:
32
q
t
= q
i
(1+nD
i
t)

1
n

(1A)
where q
t
is the instantaneous production rate at time t (i.e.,
Mmcf/year); q
i
is the initial production rate at time 0; D
i
and
n are two constants (the former is the initial rate of decline in
production and n is the rate of change in D
i
over time); and
t is the time period for which production is being estimated.
Equation (1A) can be reduced in two special cases where n
= 0 and n = 1. When n = 0 equation (1A) becomes an expo-
nential decline, i.e.,
q
t
=q
i
e
-D
i
t
(2A)
When n = 1 equation (1A) reduces to the harmonic
decline function, i.e.,
q
t
= q
i
(1+D
i
t)
-1
(3A)
In 1924, W.W. Cutler documented empirical support for
the hyperbolic function with exponent values for n between
0 and 0.7 with most wells falling being between 0 and 0.4.
33

Using Equation (3A) to estimate production decline rates
can be confusing because of how companies choose to report
the initial rate of decline, D
i
. Specifcally, the initial decline
rate reported is typically the efective rate of decline based on
the secant method. Tat is, if the initial rate of production
per year is 10 at year 0 and the production rate is 5 at the end
of one year, then the initial decline rate reported would be
50%. However, this is not the nominal rate of decline that
should be used in Equation (3A). We calculate the nominal
rate of decline from the reported rate using equation (4A):
D
i
=
1-D
esi
-n
-1
n
for n 0 (4A)
where D
esi
is the efective initial decline rate calculated using
the secant method described above.
Figure 1A Production Decline Curve (Initial Production =
8.6 Mmcf/day; Initial Decline rate = 50%;
and the Hyperbolic Exponent = .90)
8.60
4.30
2.82
1.13
0.36
.0
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
9.00
10.00
0 5 10 15 20 25 30 35 40
E
s
t
i
m
a
t
e
d

M
m
c
f
/
D
a
y

Year
ADVISORY BOARD EDITORIAL
Journal of Applied Corporate Finance (ISSN 1078-1196 [print], ISSN 1745-6622
[online]) is published quarterly, on behalf of Cantillon and Mann by Wiley Subscrip-
tion Services, Inc., a Wiley Company, 111 River St., Hoboken, NJ 07030-5774.
Postmaster: Send all address changes to JOURNAL OF APPLIED CORPORATE
FINANCE Journal Customer Services, John Wiley & Sons Inc., 350 Main St., Mal-
den, MA 02148-5020.
Information for Subscribers Journal of Applied Corporate Finance is published in
four issues per year. Institutional subscription prices for 2013 are:

Print & Online: US$514 (US), US$615 (Rest of World), 398 (Europe), 317
(UK). Commercial subscription prices for 2013 are: Print & Online: US$647 (US),
US$772 (Rest of World), 500 (Europe), 394 (UK).
Individual subscription prices for 2013 are: Print & Online: US$113 (US), 63
(Rest of World), 94 (Europe), 63 (UK). Student subscription prices for 2013
are: Print & Online: US$39 (US), 22 (Rest of World), 33
(Europe), 22 (UK).
Prices are exclusive of tax. Asia-Pacifc GST, Canadian GST and European VAT
will be applied at the appropriate rates. For more information on current tax rates,
please go to www.wileyonlinelibrary.com/tax-vat. The price includes online ac-
cess to the current and all online back fles to January 1st 2009, where available.
For other pricing options, including access information and terms and conditions,
please visit www.wileyonlinelibrary.com/access.
Journal Customer Services: For ordering information, claims and any inquiry con-
cerning your journal subscription please go to www.wileycustomerhelp.com/ask or
contact your nearest offce.
Americas: Email: [email protected]; Tel: +1 781 388 8598 or
+1 800 835 6770 (toll free in the USA & Canada).
Europe, Middle East and Africa: Email: [email protected];
Tel: +44 (0) 1865 778315.
Asia Pacifc: Email: [email protected]; Tel: +65 6511 8000.
Japan: For Japanese speaking support, Email: [email protected];
Tel: +65 6511 8010 or Tel (toll-free): 005 316 50 480.
Visit our Online Customer Get-Help available in 6 languages at
www.wileycustomerhelp.com
Production Editor: Joshua Gannon (email:[email protected]).
Delivery Terms and Legal Title Where the subscription price includes print issues
and delivery is to the recipients address, delivery terms are Delivered at Place
(DAP); the recipient is responsible for paying any import duty or taxes. Title to all
issues transfers FOB our shipping point, freight prepaid. We will endeavour to fulfl
claims for missing or damaged copies within six months of publication, within our
reasonable discretion and subject to availability.
Back Issues Single issues from current and recent volumes are available at the cur-
rent single issue price from [email protected]. Earlier issues may be obtained
from Periodicals Service Company, 11 Main Street, Germantown, NY 12526, USA.
Tel: +1 518 537 4700, Fax: +1 518 537 5899, Email: [email protected]
This journal is available online at Wiley Online Library. Visit www.wileyonlineli-
brary.com to search the articles and register for table of contents e-mail alerts.
Access to this journal is available free online within institutions in the developing
world through the AGORA initiative with the FAO, the HINARI initiative with the
WHO and the OARE initiative with UNEP. For information, visit www.aginternet-
work.org, www.healthinternetwork.org, www.healthinternetwork.org, www.oare-
science.org, www.oarescience.org
Wileys Corporate Citizenship initiative seeks to address the environmental, social,
economic, and ethical challenges faced in our business and which are important to
our diverse stakeholder groups. We have made a long-term commitment to stan-
dardize and improve our efforts around the world to reduce our carbon footprint.
Follow our progress at www.wiley.com/go/citizenship
Abstracting and Indexing Services
The Journal is indexed by Accounting and Tax Index, Emerald Management Re-
views (Online Edition), Environmental Science and Pollution Management, Risk
Abstracts (Online Edition), and Banking Information Index.
Disclaimer The Publisher, Cantillon and Mann, its affiliates, and the Editor
cannot be held responsible for errors or any consequences arising from
the use of information contained in this journal. The views and opinions
expressed in this journal do not necessarily represent those of the
Publisher, Cantillon and Mann, its affliates, and Editor, neither does the pub-
lication of advertisements constitute any endorsement by the Publisher,
Cantillon and Mann, its affliates, and Editor of the products advertised. No per-
son should purchase or sell any security or asset in reliance on any information in
this journal.
Cantillon and Mann is a full-service financial services company active in
the securities, investment management, and credit services businesses.
Cantillon and Mann may have and may seek to have business relationships with any
person or company named in this journal.
Copyright 2013 Cantillon and Mann. All rights reserved. No part of this pub-
lication may be reproduced, stored or transmitted in any form or by any means
without the prior permission in writing from the copyright holder. Authorization to
photocopy items for internal and personal use is granted by the copyright holder
for libraries and other users registered with their local Reproduction Rights Organi-
zation (RRO), e.g. Copyright Clearance Center (CCC), 222 Rosewood Drive, Dan-
vers, MA 01923, USA (www.copyright.com), provided the appropriate fee is paid
directly to the RRO. This consent does not extend to other kinds of copying such
as copying for general distribution, for advertising or promotional purposes, for
creating new collective works or for resale. Special requests should be addressed
to: [email protected].
Yakov Amihud
New York University
Mary Barth
Stanford University
Amar Bhid
Tufts University
Michael Bradley
Duke University
Richard Brealey
London Business School
Michael Brennan
University of California,
Los Angeles
Robert Bruner
University of Virginia
Christopher Culp
University of Chicago
Howard Davies
Institut dtudes Politiques
de Paris
Robert Eccles
Harvard Business School
Carl Ferenbach
Berkshire Partners
Kenneth French
Dartmouth College
Stuart L. Gillan
University of Georgia
Richard Greco
Filangieri Capital Partners
Trevor Harris
Columbia University
Glenn Hubbard
Columbia University
Michael Jensen
Harvard University
Steven Kaplan
University of Chicago
David Larcker
Stanford University
Martin Leibowitz
Morgan Stanley
Donald Lessard
Massachusetts Institute of
Technology
Robert Merton
Massachusetts Institute of
Technology
Stewart Myers
Massachusetts Institute of
Technology
Richard Ruback
Harvard Business School
G. William Schwert
University of Rochester
Alan Shapiro
University of Southern
California
Clifford Smith, Jr.
University of Rochester
Charles Smithson
Rutter Associates
Joel M. Stern
Stern Stewart & Co.
G. Bennett Stewart
EVA Dimensions
Ren Stulz
The Ohio State University
Alex Triantis
University of Maryland
Laura DAndrea Tyson
University of California,
Berkeley
Ross Watts
Massachusetts Institute
of Technology
Jerold Zimmerman
University of Rochester
Editor-in-Chief
Donald H. Chew, Jr.
Associate Editor
John L. McCormack
Design and Production
Mary McBride