Drilling and Debt

• Author: ERIK P. GILJE,DANIEL MURPHY,ELENA LOUTSKINA
• DOI: http://doi.org/10.1111/jofi.12884
• Published date: 01 June 2020
• Date: 01 June 2020

THE JOURNAL OF FINANCE •VOL. LXXV, NO. 3 •JUNE 2020

Drilling and Debt

ERIK P. GILJE, ELENA LOUTSKINA, and DANIEL MURPHY∗

ABSTRACT

This paper documents a previously unrecognized debt-related investment distortion.

Using detailed project-level data for 69 firms in the oil and gas industry, wefind that

highly levered firms pull forward investment, completing projects early at the expense

of long-run project returns and project value. This behavior is particularly pronounced

prior to debt renegotiations. We test several channels that could explain this behavior

and find evidence consistent with equity holders sacrificing long-run project returns

to enhance collateral values and, by extension, mitigate lending frictions at debt

renegotiations.

UNDERSTANDING HOW DEBT AFFECTS a firm’s investment decisions is one of

the central questions in finance. Incentive conflicts between debt and equity

claimants have the potential to result in inefficient and value-destroying de-

cisions (Jensen and Meckling (1976)). Existing theoretical and empirical work

has focused largely on the size, prevalence, and mitigation of investment distor-

tions linked with the traditional agency costs of debt such as underinvestment

and risk-shifting.1

In this paper, we use new detailed project-level data to document a previously

unrecognized debt-related investment distortion. We show that high-leverage

∗Erik P. Gilje is with the Wharton School, University of Pennsylvania and NBER. Elena Lout-

skina and Daniel Murphy are with the University of Virginia Darden School of Business. Wewould

like to thank Steve Baker; David De Angelis; Michael Faulkender; Vincent Glode; Rustom Irani;

Don Keim; Doron Levit; Song Ma; Greg Nini; Krishna Ramaswamy; Julio Riutort; Michael Roberts;

Mike Schwert; Phil Strahan; Mathieu Taschereau-Dumouchel; Luke Taylor; Hongda Zhong; and

seminar participants at Wharton, Dartmouth (Tuck), University of Virginia (Darden), IU Bloom-

ington, GWU, the 2016 MIT Junior Faculty Conference, 9th Annual FSU SunTrust Beach Confer-

ence, 2017 PNC University of Kentucky Finance Conference, 14th Annual Conference in Financial

Economic Research IDC, 2017 Frontiers of Finance Conference, 2017 Financial Intermediation Re-

search Society Conference, and Finance—UC 12th International Conference for helpful comments.

We would also like to thank Julia Tulloh for her valuable research assistance, and the Rodney L.

White center for financial support. We have read The Journal of Finance disclosure policy and

have no conflicts of interest to disclose.

Correspondence: Erik P. Gilje, The Wharton School, University of Pennsylvania, 3620 Locust

Walk - SHDH 2456, Philadelphia, PA,19104; e-mail: gilje@wharton.upenn.edu.

1Theoretical work that focuses on these issues includes, for example, Aghion and Bolton (1992),

Bolton and Scharfstein (1990), Hart and Moore (1995, 1998), Jensen and Meckling (1976), and

Myers (1977). Empirical work includes Andrade and Kaplan (1998), Parrino and Weisbach (1999),

Eisdorfer (2008), Almeida, Campello, and Weisbach (2011), and Gilje (2016).

DOI: 10.1111/jofi.12884

C2020 the American Finance Association

1287

1288 The Journal of Finance R

firms facing debt renegotiations pull investment forward and complete projects

early at the expense of project net present value (NPV). These negative project-

level investment distortions likely aim to enhance collateral value, thereby

mitigating lending frictions and increasing firms’ access to finance. Overall,

our results suggest that by increasing collateral value, high-leverage firms

(dependent on asset-based loans) mitigate financing frictions that arise at debt

renegotiations. This finding also highlights a previously unexplored hidden

cost of collateral-based financing.

Identifying how debt affects firms’ actions is empirically challenging. First,

it is difficult to observe actions at the project level and identify how these

actions affect firms’ cash flows. Second, even if one can observe managers’

detailed actions, assessing whether a decision is value maximizing requires

that a clear, unambiguous counterfactual decision exists and that its value

be observable. Third, leverage and the composition of credit agreements are

not randomly assigned, and omitted endogenous variables could be related to

both firm-level investment decisions and leverage, making it difficult to infer a

causal relationship. Finally, the distortive effects of debt tend to arise only in

the presence of market frictions that intensify as leverage increases (see, e.g.,

Stein (2003)).2

We exploit an empirical setting that allows us to make significant progress

on the above challenges. Specifically, we use detailed project-level completion

decisions from North American shale oil drilling projects to examine how oil

and gas companies with different levels of leverage react to the severe contango

episode that began in December 2014. This setting has several advantages.

First, we observe high-frequency project-level company decisions and can

quantify the effects of completing an individual oil well versus delaying com-

pletion of the well. Our data set contains detailed project-level data on 3,557

North American shale oil wells operated by 69 publicly traded oil and gas

firms. We know the date of well spudding (project start) and well completion

(first production and project cash flow), as well as the precise location of the

well.

Second, the 2014 to 2015 contango episode offers a clear instance in which

completion should be delayed. During this period, oil spot prices were signifi-

cantly lower than oil futures prices. In February 2015, for example, six-month

oil futures prices were 11% higher than spot prices (Figure 1). The futures curve

provides us with a clear counterfactual as it allows us to estimate the value

of delaying completion. We show that a decision to accelerate well completion

and to start oil production early is not value maximizing at the project level.3

Third, the detailed project-level data allow us to make progress in control-

ling for the endogeneity of potential differences in investment opportunities

2In a frictionless world, investment is independent of firm leverage and depends only on invest-

ment opportunities (Modigliani and Miller (1958)).

3Production from shale oil projects is highest during the first month and declines in each

subsequent month. Consequently, pricing at the time of initial production is a key determinant of

project-level returns.

Drilling and Debt 1289

Panel

0

1

1

1

futures price rel ative t o current s pot pric e

Pane

B. February

0

.95

1

.05

1.1

.15

1.2

.25

1.3

el A. Crude

y 2015 and

5

Matur

i

Futures curve

s

Futures curve

Futures curve

Oil Futures

September

10

i

ty of futures cont

r

s

before contang

o

as of September

as of February 2

0

s Prices over

2014 Oil Pr

15

r

act (Months )

o

and during cont

a

2014

0

15

r Time

rice Futures

20

a

ngo

s Curves

Figure 1. Crude oil futures price dynamics. Panel A plots the difference between the six-

month (two-year) futures and the spot price normalized by the spot price between 2012 and 2016,

crude oil futures prices came from Bloomberg. The shaded area represents the contango period

that we focus on in our study.Panel B plots the crude oil futures curve normalized by the spot price

at two different points in time: September 2014, prior to the contango, and February 2015, during

the contango episode. (Color figure can be viewed at wileyonlinelibrary.com)