Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
AN EMPIRICAL INVESTIGATION INTO THE INFORMATION
CONTENT OP THE REQUIRED DISCLOSURE OF
OIL AND GAS RESERVE VALUES
DISSERTATION
Presented to the Graduate Council of the
North Texas State University in Partial
Fulfillment of the Requirements
A/
M
For the Degree of
DOCTOR OF PHILOSOPHY
By
Jiunn-Chang Huang, B.B.A., M.B.A,
Denton, Texas
August, 19 80
© Copyright by
Jiunn-Chang Huang
1980
Huang, J
Information Co
iunn-Chang, An Empirical Investigation into the
ntent of the Required Disclosure of Oil and
Gas Reserve Va lues. Doctor of Philosophy (Accounting),
August, 1980, 1
This emp
and gas reserv
have been util
based on a pre
the developmen
new accounting
(RRA) for oil
new accounting
adopted as a t
versy over val
of income from
this research
assess the use
This dis
addresses two
as if the repo
signals for pr
Second, has th
any response i
employed to pe
04 pp., 9 tables, bibliography, 56 titles.
irical study is concerned with whether the oil
e value data reported by petroleum producers
ized by investors. Reporting reserve value data
sent value approach is the initial step toward
t of the Securities and Exchange Commission's
method called "Reserve Recognition Accounting"
and gas producers. Experimentation with this
concept in the oil and gas industry has been
entative resolution of the long-standing contro-
uation of oil and gas reserves and the measure
oil and gas exploration. Evidence gathered in
will be valuable to the SEC in its efforts to
fulness of RRA.
sertation assumes capital market efficiency and
specific questions. First, do investors behave
rted end-of-year reserve value data are effective
icing securities of oil and gas producers?
5 SEC-mandated reserve value disclosure induced
11 the capital market? Two research designs were
rmit extensive investigation of these two questions
In the f
scaled (by ma
value per sha
Involves form
from randomly
regard to thl
of reserve va
the pricing o
expect the st
equal during
disclosure da
determine if
the returns o
In the s
the end-of-ye
reserve value
long-term Hal
value of addi
of these meas
common shares
size effect
they were ass
represented b
four thirteen
related to th
sure. The as
technique of
irst design, the information of interest is the
rket price of common shares) end-of-year present
re of companies analyzed. The basic procedure
ing equal risk portfolios from sample firms and
selected control firms that differ only with
s attribute of "information." If this disclosure
lue data contained no information pertinent to
f oil and gas producers' securities, one would
ock returns realized by the portfolios to be
a period of twenty trading days surrounding the
2
te. The Hotelling's T statistic is used to
a significant return difference exists between
f the matched portfolios.
econd design, the information of interest includes
ar reserve value, the end-of-year adjusted
(reserve value adjusted for net working capital,
bilities, and preferred stock), and the present
tions to reserves during the current year. Each
ures was scaled, either by the market price of
or by the total revenue, to abstract for the
Five information variables were specified, and
ociated with a set of eight market-based variables
y systematic returns and residual returns from
-week periods, each covering a critical event
e SEC's decision to require reserve value disclo-
sociation test was analyzed by the multivariate
canonical correlation.
This report concludes that
1. The reserve value disclosure under the SEC-
prescribed rules did not have a measurable impact
on the stock prices of oil and gas producers at
the time the data were actually disclosed. It
appears that the information has already been
impounded in security prices prior to the actual
disclosure, possibly around the time the SEC
announced its initial proposal and the time it
issued the final rules;
Investors did not behave as if the static measures
of unadjusted values of proved reserves are effec-
tive signals in pricing securities. The effective
signals are represented by the present value of
additions of reserves during the current year.
However, the adjusted reserve value appears to con-
tain more information than does the static measure
of unadjusted reserve value,
LIST OF TABLE
LIST OP ILLUS
Chapter
I. INTR
0 P R L 0
TRATIONS.
ODUCTION
il and Gas Accounting Explored roblem and Purpose of the Study esearch Procedures imitations and Benefits of the Study rganization of the Study
II
III
IV.
CAPI
S G R B E R S
RESE
R R D S
RESU
R R S
PAL MARKET IMPACT STUDIES
V. CONC
APPENDIX A .
APPENDIX B .
APPENDIX C .
BIBLIOGRAPHY
TABLE OP CONTENTS
Page
ARCH METHODOLOGY,
esearch Design I search Design II ata Collection ummary
LTS OP THE STUDY.
esults of Research Design I suits of Research Design II ummary
LUSIONS AND SUMMARY
Iptrepretation and Conclusion SCtmmary of the Resaerch
IV
v
VI
24
tudies on Oil and Gas Company Stock Prices onedes, Dopuch and Penman Study o Study owman Study skew and Ro Study lationship to the Present Study ummary
42
. 63
78
94
96
98
9 9
Table
II
III
IV.
V.
VI.
VII.
VIII.
IX,
LIST OF TABLES
Daily Differences between Returns on Matched Portfolios ,
Resu Rk P
Its from T Test on Differences between sturns on Matched Minimum Variance Drtfolios with Risk Equal to Unity. . .
Cano I B
Re 1 a s f'
V,
sub-:
Sele A:
Risk P
Risk C
Mean V
ileal Correlations Relating Five iformation Variables to Eight Market-a.sed Variables
tive Importance of Variables Repre-nted by the Structure Coefficients Dr the First Pair of Canonical ariates
-Adjusted Portfolio Returns for Test Drtfolios
-Adjusted Portfolio Returns for ontrol Portfolios
and Standard Deviation for ariables in Research Design II,
Page
64
65
70
hypotheses and Variables Included . .
3ted Results of Canonical Correlation lalysis for Sub-Hypotheses
72
74
75
96
97
98
V
Figure
1. Cumu P
LIST OF ILLUSTRATIONS
lative Return Differences for Matched ortfolios — 20 Trading Days
Page
67
VI
CHAPTER I
INTRODUCTION
This chapter contains an overview of oil and gas account-
ing, a statement of the problem, objective and research
procedures of the study, as well as the limitation, potential
benefits, and organization for this dissertation.
Oil and Gas Accounting Explored
It is commonly agreed that the contemporary accounting
framework as typified by the Paton and Littleton monograph,
An Introduction to Corporate Accounting, which underlies our
present historical cost approach to accounting, is heavily
influenced by neoclassical economic thought. Accordingly,
"historical cost" accounting is infused with the same under-
lying assumptions, and suffers from the same weaknesses as
neoclassical economics. One of these weaknesses is the idea
that "price" measures "value." In today's environment,
characterized by continuous price instability and by great
concern over the shortage of natural resources, particularly
energy resources, the assumptions of price-value equality is
subject to growing criticism because it is argued that economic
theory does not come to grips with depletable resources."'" In
"'"Peter B. Roche, "Energy—Costly Energy Is Wasting Resources, Some Analysts Worry," The Wall Street Journal, May 3, 1979, pp. 1, 30.
th accounting,
long been rec
pleas to sup
measures refl£
for and consu:
With the
subject of oi
The need for
of companies
importance to
interested pa:
mineral resen
e deficiencies in historical cost concepts have
(bgnized. Accounting literature is replete with
plement or even replace historical cost data with
cting the "true" value of resources available
jfied in business operations.
current energy crisis filling the headlines, the
and gas accounting has moved to center stage.
Relevant information about the current economics
n the energy field is identified to be of vital
the public interest. In the minds of many
ies, this relevant information is the value of
ires.
rt
.es Compani
diverse method
basis. Under
small noninte
development c
amortized as
income is thui
changes in the
full-cost meth
unprofitable
operations in
Methods of Accounting for Exploration and Development Activities
in the oil and gas industry have adopted two
s of financial reporting on the historical
"full costing," a method followed especially by
grated oil and gas producers, all exploration and
cpsts are capitalized when incurred and are
il and gas reserves are produced. Reported net
relatively insensitive, in the short-run, to
level of expenditure. A company using the
od, therefore, may be engaged in economically
Operations while capitalizing the costs of these
its balance sheet. On the other hand, under the
o
"successful-e
and gas compa
of mineral re
The costs of
dry holes) ar
discoveries a
duced, an exp
lower reporte
success in fi
portion of th
efforts accou:
tend to raise
of the impact
profitability
Despite
costing nor s
until oil and
actions. The^
greater than
activities.
however, do n
of the econom
exploration c
and relative
economic reso
iy
fforts method," used by most large integrated oil
nies, only costs leading directly to discoveries
serves and to their development are capitalized,
unsuccessful exploration (including exploratory
9 charged to expense. Since revenues from
re not recognized until the reserves are pro-
ansion in exploration activities will tend to
i net income in the short term regardless of the
iding new reserves. Likewise, because a smaller
e total cost is capitalized under successful-
iting, a reduction in exploratory spending will
current net income in the short term regardless
of that decision on the long-run economic
of the company.
heir broad adoption and usage, neither full
uccessful-efforts accounting recognizes revenues
gas are produced and sold in arm's-length trans-
both present as mineral assets an amount no
he costs incurred in exploration and development
osts incurred in exploration and development,
Dt necessarily bear any relationship to the value
ic resources obtained. A small portion of
Dsts may boost economic resources substantially,
significant expenditures frequently yield no
urces. As a result, neither full costing nor
are developed
successful-efforts accounting provides In a company's balance
sheet information about the value of its economic resources
attributable to oil and gas properties. Consequently,
several alternatives to these historical-cost-based methods
have been suggested. Among them, perhaps "discovery-value
accounting" and "current-value accounting" deserve special
attention.
Discovery-value accounting recognizes revenue from
exploration activities at the point of discovery. Mineral
reserves are recorded at their estimated value when the
reserves are discovered or, alternatively, when the reserves
Subsequent to discovery, the carrying amount
of the reserves will not be adjusted for changes in prices;
however, the carrying amount may be adjusted for revisions of
estimated reserve quantities and for the possible capitaliza-
tion of additional development costs as they are incurred.
This discover^ value becomes the recorded value of reserves
that will be amortized against the revenue from the production
and sale of the minerals in the future.
accounting, oi i the other hand, oil and gas reserves are
revalued periodically and changes in their value are reported
as income m
Under current-value
;he period of change. Such a system of accounting
may present separate data for "(a) value increases resulting
'For a c see Joseph E. Untried Method
pmplete discussion of discovery-value accounting, Conners, "Discovery Value—The Oil Industry's ," Journal of Accountancy, 139 (May, 1975), 5^-63.
SCO
ch
lit
from new dis
ment of reseri
resulting froiji
reflect the
Di sco very-vail}
supported on
financial and
investors in £
oil and gas p
However,
questions abo
are to be usee,
the primary fi
accounting and
subjective est
imprecise and
tion of such £
ments could le
over time. Cc
analysts have
at their curre
financial stat
veries, (b) value changes resulting from adjust-
e quantities, and (c) holding gains and losses
revaluing end-of-period reserve quantities to
ange in unit value during the period."
e accounting and current-value accounting are
1j;he theoretical grounds that they provide
operating information that is highly useful to
ssessing and analyzing the operating results of
xfoducers.
from a practical standpoint, there are serious
whether these methods are appropriate if they
for the purpose of reporting reserve values in
nancial statements. Both discovery-value
current-value accounting rely upon highly
imates of reserve values that are inherently
uncertain. It is possible that the incorpora-
ubjective values in the primary financial state-
ad to erratic and manipulated financial results
nsequently, many accountants and financial
suggested that oil and gas reserves be reported
nt (or discovery) values only in supplementary
4 ements.
3 Financia
ing and Report Statement of 5 Connecticut,
4.
1 Accounting Standards Board, "Financial Account-ing by Oil and Gas Producing Companies," inancial Accounting Standards No. 19 (Stamford, 1977), p. 66.
Ibid., p p. 110-112.
The broa
costing metho
numerous vers
over the lack
and gas compa:
of accounting
(ARS 11)
successful-ef
AICPA in 1971
uniform adopt
6
Historical Background
ill acceptance and use of the two historical
ids of full-cost and successful-efforts (and the
.[ions of each method) have created a great concern
of comparability in financial reporting of oil
.nies. For at least fifteen yea.rs the accounting
profession, the regulatory agencies, and the petroleum
industry itsejLf have attempted to establish a single method
(full costing, successful-efforts accounting,
or an alternative method) to be used by all petroleum com-
panies in their financial reports. The first major step
toward accomplishing this goal was taken in 1964 when the
American Institute of Certified Public Accountants (AICPA)
commissioned Robert E. Field to study the various accounting
methods in the extractive industries. The study was published
in 1969 by the AICPA as Accounting Research Study No. 11
In ARS 11 Field recommended the adoption of the
'orts concepts of accounting. Acting upon this
recommendation, the Accounting Principles Board (APB) of the
announced a proposed "opinion" requiring the
on of successful-efforts accounting by the
industry. However, this proposition was opposed by proponents
of full costing and by many others who advocated modified
5t 'Robert Industires," 1969)•
E. Field, "Financial Reporting in the Extractive Accounting Research Study No. 11 (New York,
Po
versions of s
APB withdrew
study the sub J}
The oil
bloom because
the "Energy
act contained
and Exchange
development o
producing comj)
energy data b
the right to
(PASB) (which
Board as the
private sector
Consequently,
with the mandi
Financial Acc
re
and Reporting
No. 19, publi
successful-eft
the option of
19 approach oil
an alternative
In August);
neither of th4
7
uccessful-efforts accounting. As a result, the
its proposal with the announced intention to
ject further.
and gas accounting controversy came into full
of the oil crisis and the resulting passage of
licy and Conservation Act of 1975" (EPCA). This
a directive from the Congress to the Securities
(Commission (SEC). The directive called for the
^ "accounting practices" for oil and gas
anies in order to establish an adequate national
^se. This mandate to the SEC also gave the SEC
ly on the Financial Accounting Standards Board
in 1972, replace the Accounting Principles
Authoritative accounting rule-making body in the
'), to develop those accounting practices.
the SEC took no isolated action in compliance
te until the FASB published its Statement of
(bunting Standards No. 19, "Financial Accounting
by Oil and Gas Producing Companies." Statement
^hed in December 1977, in essence adopted the
orts method of accounting. Thus, the SEC had
either approving and adopting the Statement No.
* of developing a method of its choice as
,1978, the SEC announced that it was endorsing
commonly used historical cost methods as a
permanent req
closure of re
initiate a pr<f)
balance sheet
of a new meth
accounting (R
accounting me
companies to
by Statement
As a first st
ending after
disclosure ou
estimated fut
reserves, alo:
tion, was man
. . the es
and natural g
demonstrate w
future years
8
uirement. Instead, it decided to require dis-
erve value data on oil and gas reserves and to
ject seeking to incorporate this information in
and income statements through the development
<pd of accounting called reserve-recognition
RA).^ Until RRA can be developed as the primary
;hod, the Commission will allow registrant
ase either the successful-efforts method required
fjo. 19 or a full-cost method specified by the SEC.
p in the development of RRA, for fiscal years
December 25, 1978, but before December 25, 1979,
;side financial reports of the present value of
ire net revenues from production of proved
ig with certain financial and operating informa-
7
iated. The SEC defines proved reserves as
timated quantities of crude oil, natural gas,
as liquids which geological and engineering data
ith reasonable certainty to be recoverable in
from known reservoirs under existing economic
Securit ments for Fin and Gas Produ IC-10382; AS-
Ies and Exchange Commission, "Adoption of Require-ancial Accounting and Reporting Practices for Oil ing Activities," Release Nos. 33-5966; 35-20688; 253 (Washington, August 31, 1978)•
7 Securit
and Gas Reser Releases Nos. (Washington,
ies and Exchange Commission, "Disclosure of Oil ves and Operations; Amendments to Regulation S-K," 33-6008; 3^-15^18; 35-20838; IC-10532 December lTl 1978) .
ed
and operating
monetary valua
SEC as follows
1. Estiil
and
pect
condi
2 . A fut
applji
balart
tract
3. The
re due
conce
net
the
The secon
gas producers
g
conditions. . . . " The process of placing a
tion on proved reserves is prescribed by the
ates are made of quantities of proved reserves
tfhe future periods during which they are ex-
to be produced based on year-end economic
tions;
ure revenue amount is to be computed by
ing the prices of oil and gas in effect at the
ce-sheet date, adjusted only for fixed con-
ual escalations;
esulting future gross revenue streams are
ed by estimated future costs to develop and to
produce the proved reserves, based on year-end cost
estimates;
The Resulting future net revenue streams are reduced
to present value amounts by applying a ten percent
discount factor. According to the SEC, information
rning the present value of the estimated future
evenues should be presented as of the end of
iscal year or for the fiscal year, as applicable,
d step was to adopt a requirement for oil and
to include in their 1979 financial reports filed
fi
8 0 . , . Securiti
Financial Accc Producing Act! 35-20836; IC-
es and Exchange Commission, "Requirements for unting and Reporting Practices for Oil and Gas ivities," Releases No. 33-6006; 3^-15^16; 10530; AS-257 (Washington, December 19 ,1978).
with the Comm:
10
ssion a supplemental summary of oil and gas
9
Commission as
producing activities prepared on the basis of RRA.^ The
summary should present the current year's additions and
revisions to proved reserves of oil and gas as well as the
costs associated with the discovery and development of
proved reserves and all nonproductive costs. According to
the SECj the Information generated and the experience gained
from the presentation of this supplemental summary will
provide much <pf the basis for an eventual decision by the
to whether RRA should be required as the
accounting method for financial statements of oil and gas
producers. The ultimate goal of RRA is the preparation of
primary financial statements that reflect
(1) Proved oil and gas reserves as assets in the balance sheet;
(2) Additions to proved reserves and changes in valuations of proved reserves in the income statement; and
(3) All costs associated with finding and developing additions to proved oil and gas reserves, together with all costs determined to be non-productive during the current period, in the income statement.10
Securit ducers Supplei(n Recognition A 35-21222; IC-1979).
ies and Exchange Commission, "Oil and Gas Pro-ental Disclosures on the Basis of Reserve counting," Releases Nos. 33-6126; 3^-16218;
10 875; AS-269 (Washington, September 24,
10o Securi
Requirements for Oil and G 35-20688; IC-p"! W.
ties and Exchange Commission, "Adoption of for Financial Accounting and Reporting Practices as Producing Activities," Releases Nos. 33-5966; .10382; AS-253 (Washington, August 31, 1978) ,
A closer
method for pro
First, by defiifi
Developing the$
trained and exp
estimates are
during developita
estimates. Pr<p
to be ultimate
Rather, these
established by
frequently est
tion to proved
reserves assoc
oftentimes con
The assumption^
tion these prolp
based on data
those on which
The disclosure
reserves in an:
by the SEC sin
Second,
revenues repre
of the timing
11
Problem and Purpose of the Study
examination of the SEC-prescribed valuation
/ed reserves reveals several critical issues,
ition, proved reserves are estimates.
e estimates requires the judgment of persons
erienced in reservoir engineering. These
likely to be imprecise, and knowledge gained
ent and production often necessitates revising
ved reserves are not estimates of oil and gas
jLy recoverable from discovered reservoirs,
stimates must meet engineering criteria
the definition of proved reserves. Companies
mate future recoveries of oil and gas in addi-
reserves. These "probable" and "potential"
ated with proved and unproved properties
;ribute significantly to future production.
prescribed by the SEC exclude from the valua-
able and potential reserves because they are
;hat are significantly more uncertain than
quantities of proved reserves are estimated,
of quantities (but not values) of proved
i|iual filings of all registrants has been required
e 1976.
he present value of estimated future net
ents a current valuation, based on an estimate
<t)f future production of proved reserves, current
:os
th
prices and c
cent. Opponei
generally stre
quantities of
projecting ra
expenditures.
not necessari
rent value of
these valuatioih
estimate of
reserves. An
into account,
recoveries of
anticipated f
raent and prodiji
rates would b
different res4
based on the
market value
relatively ob ;
believes that
outweighs the
nature of the
method.
Various
communities hi
12
ts, and a prescribed discount rate of ten per-
its of the disclosure of present value data
ss the uncertainties inherent in estimating
proved reserves and the need for judgment in
es of production and the timing of development
Many also argue that the valuation method will
y provide a meaningful indication of the cur-
proved reserves. As acknowledged by the SEC,
procedures do not necessarily yield the best
e fair market value of a company's oil and gas
estimate of fair market value should also take
among other factors, the likelihood of future
oil and gas in excess of proved reserves and
ijiture prices of oil and gas and related develop-
ction costs. In addition, different discount
used to reflect the varying risks related to
rves. However, in the SEC's view, a valuation
prescribed rules will be representative of the
f proved oil and gas reserves derived on a
ective and uniform basis. Obviously, the SEC
the usefulness of this supplemental information
limitations associated with the subjective
data and the necessary use of a uniform valuation
Authors and members of the business and financial
ve also presented arguments concerning the
usefulness of
instance, Rictt
natural resoui
Andersen & Co
provide a bas
companies and
company.11 G1
University of
present value
have stated th
The methc reserves arise frc operation and thus than prim gas locat Slope) wo required reserves present v
Albert S. Mart
also given rea
In our op success i of the mi useful in is no rel their val
13
the present value data on proved reserves. For
ard C. Adkerson, an audit partner in the
'ces division of the Houston office of Arthur
, has indicated that this information would
s for comparing proved reserves of different
the changes in proved reserves of a single
enn A. Welsch and Edward B. Deakin of the
Texas at Austin, in their evaluation of the
approach for mineral resource assets disclosure,
at
d would preserve the distinctions between held by different companies. Reserves that m secondary operations would show higher costs per barrel than primary reserves,
would be reported at a lower net present value ary reserves. The wellhead price of oil and ed in distant places (e.g., Alaska's North uld reflect the transportation costs to bring the reserves to market; these thus would be reported at a lower net alue than those in the lower 48 states.12
in, Jr., Controller of Sun Company, Inc., has
sons for his support of reserve value disclosure.
inion, a petroleum company's true economic s largely measured by the current values neral reserves it owns. Cost data are not this repsect, because in our industry there ationship between the cost of reserves and ues. Value data provide an insight into the
11Richard Work?," Journa
12 Glenn A
Reporting The ^Washington,
C. Adkerson, "Can Reserve Recognition Accounting 1 of Accountancy, 148 (September, 1979), 72-81.
. Welsch and Edward Deakin, Measuring and 'Replacement" Cost of Oil and Gas Reserves
19 77), p. 78.
14
qualitative differences of reserves, not otherwise discernible from a purely volumetric disclosure.13
Ma
The counterviews are represented by serveral major oil
ne accounting educators, and a number of account-
ny oil companies are unconvinced that disclosure
ue would result in more meaningful and useful
rmation, in view of the extent to which sub-
d be required in making long-range projections
companies, soi
ing firms.
of reserve va
financial inf<j)
jectivity wou
of ultimate r4
ment expenditu
disclosure fr$:
failure to re
as quality of
technology, ad
location. Pri
opposed the pr
n The plai any indus is fraugh the degre mation ir in planni
serves, production rates, and timing of develop-
14
res. Cooper et al. attacked the SEC
.mework for being arbitrary and because of its
fleet various risks associated with reserves such
reserves, concentration of reserves, changes in
cessibility and recoverability, and reserve
ce Waterhouse & Co., in a recent study, strongly
esentation of RRA data and concluded that fact is that the petroleum industry, like
try or any business or any human endeavor, .t with uncertainty. We can try to reduce e of uncertainty by providing factual infor-the most meaningful, useful format to assist
ng for a more orderly environment. But we
His opi 15 and 16, 197 Oil Corporatic discussions o Turbulence in (August, 1979)
14it n Kerry C ing: A Propose 148 (September
nion was expressed at a symposium held on May 8, in Pittsburgh under sponsorship of the Gulf n. Yuji Ijiri summarized the presentations and that symposium in "Oil and Gas Accounting—
Financial Reporting," Financial Executive, 47 , 18-26.
ooper and others, "Reserve Recognition Account-d Disclosure Framework," Journal of Accountancy, , 1979), 82-91.
must be predict! facts.15
careful not to confuse the fiction of re estimates with the reality of historical
Whether
reserves will
market is an
adequately.
usefulness of
empirically th
Specifically,
enues from pre
Form 10-K repc
information cc
value data con
tors, as the S
believe, one c
disclosure of
reaction exist
content of the
reserve value
On the other h
new informatio
as the counter
observable mar
the
In
15
present value data on proved oil and gas
convey new information to the aggregate capital
empirical issue that has not been addressed
view of the apparent disagreement over the
reserve value data, this study seeks to assess
e usefulness of reserve value data to investors,
the present value of estimated future net rev-
'ducing proved reserves reported in the 1978
>rts provides a unique opportunity to assess the
ntent of the reserve value data. If reserve
.vey any new information important to the inves-
EC and other proponents of additional disclosure
an expect to observe a market reaction to the
the data. To the extent that the market
s, it is possible to evaluate the information
reserve value data and the effect of the SEC
disclosure decision on the pricing of securities,
and, if the reserve value data do not convey any
n about the risk-return prospects of the firms,
view asserts, then there should be no empirically
ket reaction to disclosure of such data. The
15 'Joseph
Pact or Pictio 1979), 99.
E. Connor, "Reserve Recognition Accounting: a?," Journal of Accountancy, 148 (September,
absence of ma:
behave as if
the pricing o
no new informi
.rket reaction implies that investors do not
the reserve value data are effective signals in
securities, and one can conclude that there is
tion content in the reserve value data.
ff
Over the
theoretical a
sion that, as
markets in wh
traded are e
information.
publicly avai
of past prices,
ratios, macroe
policies, and
Much of
usefulness or
the capital m
development ir
ing the effici
foundation for'
the stock mark
prices reflect
unbiasedly. He
value disclosu
nd
16
Research Procedures
last twenty years, a considerable amount of
empirical analysis has pointed to the conclu-
a good working approximation, the securities
ch common stocks of U.S. corporations are
icient with respect to publicly available
That means the current price incorporates all
lable information including the entire sequence
, annual reports, accounting income numbers and
conomic variables such as monetary and fiscal
a variety of other industry-specific information,
tjhe accounting research on topics related to the
the information content of accounting data in
rket context has built upon the extensive
the financial literature studying and document-
ency of capital markets. Similarly, the
the present study is provided by assuming that
ets are efficient to the extent that stock
all relevant public information rapidly and
nee, the capital market impacts of the reserve
re are measured by examining the stock price
behavior of o
ing the initi^
Two sepa
extensive invi
first design,
utilized—the
as of the end
one attribute
the present v
proved reserv^
Additions,
of twenty trad
data are repo
incorporates
disclosure of
earlier market
in the second
is composed of'
reserve value
are contained
Poor's Corpora
and sufficienc
grated compani
examined in th
chosen for the
17
1 and gas producers during the period surround-
1 public disclosure of the data.
ate research designs are formulated to permit
stigation of the issue in question. In the
only one attribute of reserve value data is
present value of estimated future net revenues
of the fiscal year. In the second, more than
of reserve value data are explored including
' .lue of estimated future net revenues from
s added during the current year.
lly, the first design focuses on a test period
ing days surrounding the day the present value
rted by a company, whereas the second design
ijhree more test periods before the initial public
the data to allow for the detection of any
reaction. Weekly security returns are examined
design. The sample of firms used in this study
oil and gas producing companies reporting
data to the SEC whose common stock price data
in the Daily Stock Price Record of Standard &
tion. The sample is limited by the availability
y of the data and by exclusion of large inte-
es from the study. Seventy-two companies were
e first design and fifty-eight firms were
second design.
The proc
steps. First
ratio of pres
hereafter refi
each firm. Ne
according to
measure of in^
based on that
firms in the
in the first c
PV per share t
tional distrib
four portfolio
stock returns,
securities are
are formed. N
the annual rep
date the reser
returns for tw
disclosing dat
after the disc
of the eight p
folio return i
for each of th
are computed f
portfolio retu
ent
18
dures in the first design involve the following
a particular measure of information (i.e., the
value per share to market price per share,
rred to as scaled PV per share) is computed for
xt, the firms are ranked in ascending order
ir respective magnitudes of that particular
ormation and subsequently divided into quartiles
ranking. This grouping procedure suggests that
fourth quartile are more desirable than those
uartile since the former reported higher scaled
han the latter. In order to compare the condi-
ution function of the stock returns of these
s with the unconditional distribution of the
four control portfolios of randomly selected
formed at the same time that the test portfolios
ext, assuming the filing date associated with
orts or its amendment filed with the SEC is the
ve value data are first known to the public,
enty trading days (nine trading days before the
e, the disclosing date, and ten trading days
losing date) for the shares of each firm in each
ortfolios are computed. A risk-adjusted port-
s then computed for each of the eight portfolios
twenty trading days. Next, return differences
or each trading day by subtracting the control
rns from their respective test portfolio returns.
o
se
Finally, the
are statistic
research ques
reported end-
signals for p
In the
values and fid)
Five Informat
Independent o
(full costing
information v
the variable
market value
variables are
1. The
or
tota
2. The
of re
3. The £
value
capi
stoct
share;
4. Th e e
to ptf
19
^eturn differences are analyzed to see if they
ally different from zero. The specific
;ion asked is: Do investors behave as if the
f-year amounts of reserve value are effective
icing securities?
cond design, both static measure of reserve
w measure of reserve values are considered,
on variables are defined and they are all
the historical cost accounting method
successful-efforts) used. In addition, each
friable is adjusted for the size factor, e.g.,
s divided by either total revenue or total
f shares of common stock. The five information
resent value of proved reserves scaled by
market value of common shares;
present value of proved reserves per dollar
venue;
djusted value of proved reserves (present
of proved reserves, plus net working
ijal, minus long-term liabilities and preferred
:) scaled by total market value of common
s;
resent value from the current year's additions
oved reserves scaled by total market value of
common shares; and
5. The
addi
revei
This set of v
variables rep^
returns. The
for each of till
reflect the p
events relate
The four event
disclosing res
the announcem
19 78J and the
period covers
critical event
this design is
related inforri
np
The maj or
set forth as
1. An e
knowledge. Tit
information c
desirab ility
allocation is
tion should b
20
resent value of the current year's
ions to proved reserves per dollar of
lue.
ariables is associated with a set of stock price
esented by systematic returns and residual
e two components of stock returns are obtained
e four test periods, which are designed to
cf)ssible security price effects of four critical
to the SEC reserve value disclosure decision,
s are: an initial proposal in October 1977 for
erve values, a second proposal in August 1978,
^nt of the adoption of the proposal in December
first public disclosure of the data. Each
the six weeks before and the six weeks after the
The specific research question addressed in
: Is there a relationship between reserve-value
ation variables and capital market responses?
Limitations and Benefits of the Study
' limitations and assumptions of the study are
ollows.
irical study per se can only yield descriptive
is research is limited to the investigation of
content. It will not attempt to address
f accounting policy decisions or optimum resource
ues. Consequently, no normative policy implica-
drawn from the findings.
21
Other assets,
2. Lack of data caused some firms to be omitted from
test samples. Extrapolation of the conclusions of this
study must be based on a belief that a sufficiently large
segment of thi universe of oil and gas producers is
included and that the test results are sufficiently strong
to overcome tljie possible effects of omitting firms from
the analysis,
3. The information content tested in this study is
the value of proved reserves. Proved reserves, however,
do not account for all the assets of an oil and gas company,
especially probable and potential reserves
may be significant. To assess the true "value" of the
firm, all assets including probable and proved reserves
need to be considered,
4. The e|xact time that the reserve value data are
made publicly available is unknown. The first design
assumes that the "signature date" in the report filed with
the SEC is the disclosing date, whereas the second design
assumes that t|he week in which the signature date falls is
nformation is released to the market. To the
fferent companies may define their signature
tly, the test of the first design will be
daily return differences are examined. The
is less influenced by this weakness because
e involved.
the week the i
extent that di
dates differer
affected since
second design
weekly data ar
5. This
markets. To
have been bas
York Stock Ex
(ASE). The s
whose stocks
The extent of
exchanges nee
Answers
preceding sec
reached regartjl
the answers t
fits would de]
suggests that
information i
such informat
for making so
analysis base
the disclosure
disclosure of
market impact
market respon
efforts to as
Even if the ai
information
market is sti
re
22
research relies on the efficiency of capital
jiate the studies supporting market efficiency
d on analyses of securities traded on the New
jhange (NYSE) and the American Stock Exchange
amples of this study involve a number of firms
.re traded in the over-the-counter (OTC) market.
efficiency in the OTC market and regional
fis further research and investigation.
:o the two research questions stated in the
ion are needed before any conclusions can be
ing the usefulness of reserve value data. If
both questions are positive, a number of bene-
ive from this study. First, if the evidence
reserve value disclosure does convey new
important to the investors, the argument that
on is not useful and meaningful to the investors
ifmd economic decision is damaged. Secondly, the
on canonical correlation may suggest whether
of static measure of present values or the
additions to reserve values has the larger
, and it may identify the timing of significant
^es. This would be very valuable in the SEC's
ess its experimental feedback in developing RRA.
lswers are negative, the conclusion that the
presented by the reserve value is not new to the
11 useful to policy-making bodies. The SEC and
d
FASB appear s
should, be int
designed, emp
effects of th4
ensitive to the economic consequences issue and
^rested in learning the results of any well-
rically based study related to the economic
ir rules and standards.
le
CI'
Chapter
review of re
methodology e
the relations
Chapter
designs, des
the hypothese
Chapter
of the statis
analysis, supp
are discussed.
Chapter \
23
Organization of the Study
I of this dissertation includes an extensive
vant prior work. Emphasis is placed on the
mployed by each empirical study reviewed and
ip of each to the present study.
II includes an explanation of the research
iption and sources of the empirical data, and
and test procedures that have been formulated,
presents the research results. The results
ical tests and additional evidence from further
lemented with tabular and graphic illustrations,
IV
analyzes the findings and summarizes the study.
w Concern
policy decisic^
disclosures m
evident in bot
to and stateme
Securities and
Accounting St
this area have
empirical rese
have been prov
less extensive
and the capita
Accounting St a:
CHAPTER II
CAPITAL MARKET IMPACT STUDIES
ith the capital market effects of accounting
ns and the capital market impacts of the
ide by firms complying with those decisions is
h the accounting literature and in submissions
nts by policy-making bodies such as the
Exchange Commission (SEC) and the Financial
.ndards Board (PASB). Research activities in
been substantial. Recent reviews of the
arch characterized as efficient market studies
1 2 ided by Gonedes and Dopuch, and Kaplan. A
review with focus on methodological problems
1 market impact of Statement of Financial
ndard No. 19 ("Financial Accounting and
Nicholas Equilibrium, I Techniques: Th Work," Studies Supplement to 48-129.
^Robert S Accounting Num' at the Duke Sy Research on Pr
Gonedes and Nicholas Dopuch, "Capital Market nformation Production, and Selecting Accounting eoretical Framework and Review of Empirical on Financial Accounting Objectives: 1974,
Journal of Accounting Research, 12 (1974),
Kaplan, "The Information Content of Financial bers: A Survey of Empirical Evidence," presented mposium and published in The Impact of Accounting actice and Disclosure (Durham, N.C., 1978) .
24
Reporting by
Poster.J
The obje
vey of capita
and reporting
related to thet
Emphasis is p
empirical stu4;
dissertation.
Oil and Gas Producing Companies") appears in
ctive of this chapter is to present a brief sur-
market studies related to oil and gas accounting
and to review in detail some of the prior work
capital market impact of various disclosures,
laced on the research design employed in each
y reviewed and the relationship of each to this
Studi
Like many
cal work relat
in the context
cient market h
relationship b
available to i
in the literat
the viewpoint
publicly avail
react instanta:
Patz and
oil exploratio
•5 -^George P
Market Researc Stanford Unive
4 Dennis H
mulation in an Accounting Res
25
es of Oil and Gas Company Stock Prices
other capital market effect studies, the empiri-
ed to oil and gas accounting has been performed
of the efficient market hypothesis. The effi-
ypothesis is a theory used to explain the
etween security price changes and information
nvestors in the markets. The evidence presented
ure of accounting and finance generally supports
that stock market prices completely reflect all
able information and that stock market prices
meously and unbiasedly to new information.
4
Boatsman investigated the price performance of
>n companies around the time that the Accounting oster, "Accounting Policy Decisions and Capital h," working paper, Graduate School of Business, rsity, Stanford, California, 1980.
. Patz and James R. Boatsman, "Accounting For-Efficient Markets Environment," Journal of earch, 10 (Autumn, 1972), 392-405.
Principles Boa
requiring comp
to switch to
pronouncement
companies to t
perceive that
accounting cor
forces, one mi.
companies to t
that there was
companies, whi
efficient mark
Similarly
been conducted
of the exposur
of full-cost f
Separate studi
and the SEC6 f|
Statement No
prices of eith
producers. Th
26
.rd issued an exposure draft of an opinion
anies that previously used full-cost accounting
tthe successful-efforts method. If adopted, this
would have caused reported earnings of such
e sharply lowered. If the market could not
the earnings decrease was simply due to an
.vention, and not to any fundamental economic
ght expect the share prices of full-cost
e adversely affected. Patz and Boatsman found
no adverse effect on the share prices of such
ch is consistent with predictions based on the
et hypothesis.
, three major empirical research studies have
to assess the differential effects of issuance
e draft of Statement No. 19 on equity securities
irms versus those of successful-efforts firms,
es by the PASB, conducted by Thomas Dyckman,"'
ound that issuance of the proposed draft of
19 had no significant effects on the stock
er full-cost or successful-efforts oil and gas
e Dyckman study reported that
Thomas R sure Draft on (Stamford, Con
^Securiti vs. SE (succes Change's Compe
. Dyckman, Report on the Effects of the Expo-the Returns of Oil and Gas Company Securities n., October, 1977).
es and Exchange Commission, "PC (full-cost) sful-efforts): A Study of a Proposed Accounting titive Impact" (Washington, February 20, 1978).
The test show a s from the No. 19 o\ Septembeip evidence This eff^
results we were able to conduct do not jatistically significant information effect issuance of the exposure draft of Statement er the nine-week (July 18, 1977 to 16, 1977) test period. . . . there is
of a transitory effect during the issue week, ct did not persist however.7
However, Danie
research proje
Exposure Draft
the returns of
successful-eff'
was sustained
Other cap
gas accounting,
Etebari,1^ and
reached in the
numerous tests
27
8 1 Collins and Warren Dent reported that their
ct indicated that following the release of the
there occurred a significant downward shift in
full-cost firms' stocks relative to those of
'orts companies' stocks, and this downward shift
over a period of six months.
ital market impact studies related to oil and
methods were performed by Robert Eskew,^ Ahmad
11 Baruch Lev, among others. The conclusions
se studies were also conflicting. Although
have been performed to assess the effects of
7 Dyckman,
Daniel W Elimination of leum Industry: quences," Jour 1979), 3-44.
^Robert K Accounting and Industry," The
op. cit., Report 2, p. 2.
Collins and Warren T. Dent, "The Proposed Pull Cost Accounting in the Extractive Petro-An Empirical Assessment of the Market Conse-nal of Accounting and Economics, 2 (Spring,
Eskew, "An Examination of Association between Share Price Data in the Extractive Petroleum Accounting Review, 48 (April, 1974), 317-324.
10 Ahmad E
Effects of Pro Industry," unp Business Adminli Texas, 1979.
tebari, "An Assessment of the Stock Market •posed Accounting Changes in the Oil and Gas ublished doctoral dissertation, College of istration, North Texas State University, Denton,
11 Baruch
Stock Market: Review, 54 (JujL
•ev, "The Impact of Accounting Regulation on the The Case of Oil and Gas Companies," The Accounting h.y, 1979), 485-503. &
the FASB expos
is far from re
studies to exa.:
sure of reserv
information of
required disci
review of prio
market impact
relevant.
G
Gonedes,
the Gonedes st
primarily with
conveys inform
The Gonedes gr
earnings forec
previously by
28
ure draft on stock returns, and the controversy
solution, no results of empirical research
.mine the capital market effect of the disclo-
e value data have been reported. Since the
interest in this study is the effects of
osure of oil and gas reserve values, a detailed
r empirical studies related to the capital
of various types of disclosures appears more
onedes, Dopuch and Penman Study
12
Dopuch and Penman (hereafter referred to as
udy) of the University of Chicago were concerned
the extent to which disclosure of forecast data
ation useful to an analyst in valuing a firm,
oup's approach to assessing the effects of
ast relied on a theoretical framework developed 13 14
Gonedes. Since several subsequent studies
12Nichola "Disclosure Ru Equilibrium: T of Accounting
13Nichola of Special Ite Accounting Res
14
s Gonedes, Nicholas Dopuch, and Stephen Penman, les, Information-Production, and Capital Market he Case of Forecast Disclosure Rules," Journal Research, 1.4 (Spring, 1976), 89-137-
s Gonedes, "Risk, Information, and the Effects ms on Capital Market Equilibrium," Journal of arch, 13 (Autumn, 1975), 220-256.
The sam examine market the effect of price/earnings tent of divide the disclosure
e testing procedure has been used by Harrison to reactions to accounting change, by Basu to test the release of annual earnings information on ratio, by Gonedes to test for information con-nds and extraordinary items, and by Ro to analyze of capitalized lease information.
were based on
theoretical ba.
Let R it,
i from t to t •h
the informatio
thought to be
i at time t.
available afte
time t. Goned
0 . can be ass
of R ^ conditi
distribution f
tion function
If the compari
equal, then on
content, in th
probability re
The Goned
investigation
random variabl
adjusted (seal
share closing
trading day pr
a particular m
each sample fi
Earnings Porec
n
29
the same idea, a brief description of its
sis is presented here.
i=l, n, denote the rate of return on portfolio
1. The n portfolios are formed on the basis of
conveyed by a random variable. 0 ^ is
pertinent in valuing the securities in portfolio
Assume that the realization of 0.. becomes 1X/
r capital market equilibrium is established at
es has shown that the information content of
essed by (a) comparing the distribution function
onal on the value of 0.. to the unconditional 1 Ty
'unction of or (b) comparing the distribu-
of R. , conditional on different values of 0.^. it it
son indicates that the distributions are not
e can infer that 0 does have information 1 U
e sense that some realizations of eL^ produce
assessments.
es group conducted their information content
in the following steps. First, the information
e was defined as the earnings-per-share forecasts
ed) for size effects, as represented by the per
price of the firm's common stock on the tenth
seeding the forecast's publication date. Thus,
sasure of information variable was computed for
rm that was covered by Standard and Poor's
aster. Next, the sample firms were ranked in
ascending ord€'
scaled earning
into quartiles
were formed w
sample with sc
forecasts. Pc
casts in the s
place at the c
ceding each fc
a firm in the
at time t, tha
quartile. Nex
tions of the s
(conditional o
distribution o
were formed by
portfolio retu
portfolios for
returns were o
30
r according to their respective values of
s-per-sahre forecasts and subsequently divided
based on that ranking. Thus, four portfolios
ith portfolio 1 containing all firms in the
aled forecasts in the lowest quartile of scaled
rtfolio 2 contained all firms with scaled fore-
econd quartile, etc. Portfolio formation took
lose of trading on the tenth trading day pre-
recast date. Presumably, a forecast that puts
fourth quartile has more favorable implication,
n a forecast that puts the firm into the first
t, in order to compare the distribution func-
tock returns of these four portfolios
n the value of S ^ ) with the unconditional
f the stock returns, four control portfolios
randomly selected shares. An equal-weighted
15 rn was then computed for each of the eight
each of the 240 days (ten trading days'
btained for each of the twenty-four forecast
15 The ris
within each of mates of syste lower and uppe portfolios wer and a high-ris weighted so as (beta equal to portfolio were weighting fact portfolio retu
k adjustments involve ranking the securities the eight portfolios according to their esti-
matic risk (beta), in ascending order. The r halves of the ranked array for each of those e used to form two sub-portfolios: a low-risk k sub-portfolio. Those sub-portfolios were next to achieve a single portfolio with neutral risk one). The unadjusted returns for each sub-subsequently multiplied by the appropriate
ors and summed to arrive at the risk-adjusted rns .
dates they obs
for each tradi
returns from t
Finally, the r
were statistic
Two concl
study are
1. Earni
pert!
2. Great
the e
extre
compa
Certain s
capital market
utilized a ran
positive resul
to only that s
very important
Second, by all<
portfolios ins
becomes possib
(see Footnote
setting the re
31
erved). Next, return differences were computed
ng day by subtracting the control portfolio
heir respective forecast portfolio returns,
eturn differences were analyzed to see if they
ally different from zero.
usions of interest resulting from the Gonedes
ngs forecasts seemed to reflect information
nent to valuing firms.
er information content could be ascribed to
arnings forecasts of companies with
mely low scaled forecasts as compared to
nies with higher scaled forecasts.
trengths of Gonedes' methodology used to assess
effects can be identified. First, the design
domly selected smaple to validate that any
ts found in the test sample were not specific
ample. Such an internal validity control is
in studies of the market impact of disclosure
:>wing division of the ranked firms into four
tead of only two, a more discriminating test
le. Third, the procedure for risk adjustments
15) is readily performed. The procedure for
lative risks of the respective portfolios
16
16 Foster, op. cit., p. 27 •
32
exactly equal to one another assued that all other things are
held constant, insofar as expected returns are concerned.
Ro Study
Ro tested whether the SEC's 1973 mandated lease dis-
closures under ASR No^ lkj_ had any impact on the pricing of
17 securities. The framework of Ro's research design was
based on the idea developed by Gonedes and on the methodology
1.8
introduced by Harrison to examine market reactions to account-
ing changes. Although the theoretical base for the methodology
used in each of the studies of Gonedes and Ro is similar,
certain differences in application exist.
Ro classified firms into the self-selected categories of
those using lease financing (treatment group) and those not
using lease financing (control group).
To control for confounding factors that might conceal
possible effects of capitalized lease disclosure under
investigation, treatment firms and control firms were indi-
vidually matched according to the estimates of their systematic
risks (betas). Also, in order to examine whether the impacts
of capitalized lease disclosure did vary according to differ-
ent levels of riskiness of firms, firms in each group were
17 B. Y. Ro, "The Disclosure of Capitalized Lease Informa-
tion and Stock Prices," Journal of Accounting Research, 16 (Autumn, 1978), 315-340.
18 T. Harrison, "Different Market Reactions to Discre-
tionary Accounting Changes," Journal of Accounting Research. 15 (Spring, 1977), 84-102.
further class!
and low, using
After the risk
return was corr..
Individual ret
each portfolio
then used as a
Ro Interp
of Information
Impact on the
be derived fro
pacts. First,
claimed to con
portfolios to
allows for a d
impounded by t
firms than It
there Is at le
design. When
estimates of s
are only appro
be needed to a
portfolio risk
33
fied into two different risk portfolios, high
systematic risks (betas) as an attribute,
portfolios were constructed, a portfolio
.puted by taking the arithmetic average of the
urn differences (between matched firms) for
Each monthly average return difference was
2
sample unit for the Hotelling's T test.
rets his results to indicate that the release
in accordance with ASR No. 147 had an adverse
pricing of securities. Certain advantages can
m Ro's design for detecting capital market im-
the individual matching procedure has been
.trol for factors that could cause respective 19
not be strictly comparable. * Second, the design
irect test of the possibility that the news is
he market in a different manner for low-risk
is for high-risk firms. On the other hand,
ast one weakness associated with matching—pair
the firms are matched individually on their
ystematic risk, the resultant portfolio risks
ximately equal. Thus, an additional step may
djust the portfolio returns for differences in
19 When th
select the con control group, financing deci the measuremen beta could be such claims.
e individual matching procedure is applied to trol sample firm instead of randomly selected (1) the effects of production-investment and ions are believed to be held constant and (2)
t error that results from estimation of firm's reduced. See Harrison for justification for
Similarly
his study, put
the impact of
using capital
sidering direc
disclosure lik
lence of lease
Bowman de
for the empiri
the capital as
between market
established.
an accounting
business risk
the analysis t
empirical desi
risk (beta) as
beta,^ debt-t
independent va
reported to th
34
Bowman Study
•, Robert Bowman of the University of Oregon in
lished in April, 1980, investigated empirically
leases on risk in the securities market,
2o
market theory. However, rather than con-
tly the impact of changes in levels of
e the design followed by Ro, the debt equiva-
s was the focus of Bowman's study.
veloped a theoretical model to provide the basis
cal work. First, using the basic framework of
set pricing model, the theoretical relationship
(systematic) risk and accounting variables was
Market risk was found to be directly related to
beta and leverage. The distinction between
and financial risk was then incorporated into
o aid in building an empirical model. Bowman's
gn was a multiple regression model with market
the dependent variable and with an accounting
o-equity ratio, and leases-to-equity ratio as
riables. The capitalized value of leases, as
e SEC under ASR No. lkj_, was used to measure 20d . .
Robert Empirical Inve 1980), 237-253
21The acc of a firm's ac of the market accounting bet
G. Bowman, "The Debt Equivalence of Leases: An stigation," The Accounting Review, 55 (April,
ounting beta was expressed as the covariability counting earnings with the accounting earnings portfolio. In the absence of leverage, the a represents business risk.
the value of 1
betas were coir
of the capital
market value o
equity ratio rc
to capture tha
industries. B
icance of the
risk.
In his in
was not signif
the lease and
gies were then
In each case,
eliminated, le
association te
Eskew anc.
the replacemen.
No. 190 induce
35
ease obligations. The market and accounting
puted from the time periods prior to disclosure
22
ized value of leases. In addition, the
f debt was estimated for use in the debt-to-
easure, and intercept dummy variables were used
t portion of market risk that varies across
owman's test hypothesis concerned the signif-
lease variable in the association test on market
itial test, Bowman found that the lease variable
icant, but there was multicollinearity between
leverage variables. Two different methodolo-
developed to overcome that econometric problem,
when multicollinearity was controlled or
ases made a significant contribution to the
st on market risk.
Eskew and Ro Study
Ro of Purdue University investigated whether
t cost disclosure mandated by the SEC in ASR
d any response in the capital market. 23 Their
22 Both we
for the 60-mon subject to dis
23 Robert
Replacement Cc Volumes," unpu Purdue Univers
re estimated with simple linear regression models th period ending on the fiscal year-end, first closure of lease information under ASR No. 1^7.
K. Eskew and Byung T. Ro, "SEC Mandated st Disclosure, Security Returns and Transaction blished working paper, School of Management, ity, West Lafayette, Indiana, 1978.
research methc
correlation an
relationship b
and a set of p
they attempted
defined from r
variables repr
common stock r
ing variables
1. Unrea
7
histo
Unrea
histo
Reali
cost,
(Real
at hi
Unreal
histoj
Unreal
histor
Reali:
up
According
ratios were s
prices of asse
variable was cd>
36
dology was based on multivariate canonical
alysis, which generally focuses on a linear
etween a set of criterion (dependent) variables
redictor (independent) variables. Essentially,
to associate a set of accounting variables,
eplacement cost data, with a set of market—based
esented by the relative risk and the residual of
eturn and transaction volume. The seven account-
were
lized net holding gain/Total assets at
rical cost,
lized gross holding gain/Total assets at
rical cost,
zed holding gain/Total assets at historical
ized holding gain + Total stockholders' equity
storical cost)/Total assets at historical cost,
ized net holding gain/Net income at
leal cost,
ized gross holding gain/Net income at
leal cost,
zed holding gain/Net income at historical cost,
to Eskew and Ro, the first three accounting
posed to measure the effects of changes in
is on the firm's total assets. The fourth
nsidered as a measure of the impact of inflation
pi
on the firm's
expected to ca
operating leve
period of 100
periods so tha
SEC's action w
separate perio
proposal date,
adoption, and
closure of re
four market-ba:
represented an
to sixteen mark
Eskew and
some extent du
closure. Howe
was not clear,
either to the
to the attenti
response during
long before the
The prese
institutional
behavior is as^
37
financial leverage. The last three ratios were
•Pture the effect of inflation on the firm's
rage in different ways. For the test period, a
weeks was partitioned into four 25-week sub-
t each of the three major events relating to the
ith respect to ASR No. 190 was covered by a
d. Specifically, period I included the initial
period II included the announcement date of
period IV reflected the time of the actual dis-
acement cost data. Since for each sub-period
ed variables were estimated, the research
attempt to associate seven accounting variables
:et-based variables.
Ro found that the capital market responded to
ring the period of the replacement cost dis-
ver, the cause of the capital-market response
They concluded that the response was due
ontent of the replacement cost disclosure or
<|jn effect. Their result also indicated a market
the period of promulgation of ASR No. 190,
^ data were prepared and disclosed.
nt
Relationship to the Present Study
study is primarily empirical within the
etting of capital markets. Capital market
umed to be efficient in the semi-strong
form: securl
including that
the disclosure
All the studie
chapter made s
the link betwe
mation variabl
of implicitly
conclusions in
proves the ass
One of th
impact of a "d
identify and m
variable, in t'.
ature lacks an
measurement ru
Because o
existing empirjl
effect as oppo
effect. For i
variable, capi
measurement of
statistical te
38
ity prices reflect all public information
contained in accounting policy decisions and
s made by firms complying with those decisions,
s, except that by Bowman, reviewed in this
uch an assumption. This assumption provides
en the aggregate market response and the infor-
e of the disclosure and avoids the untenability
assuming market efficiency and then stating
terms of whether the evidence proves or dis-
umption.
critical issues to address in studying the
isclosure" on the capital market is how to
asure the "information," the independent
le research design. Currently, accounting liter-
articulated theory of how specific accounting
Les are linked to the pricing of capital assets,
f the limited theoretical underpinning, most
cal studies test for the existence of an
ed to testing an explanation for an observed
jistance, Ro included only one information
alized lease information, in his study, and the
the information he was willing to make for
t purposes was simply classificatory
24, de For a see Fama, "Eff: Empirical Work
^There a ficatory) , ordili
tailed description of forms of efficient markets, cient Capital Markets: A Review of Theory and " Journal of Finance, 25 (May, 1970, 383-417.
re four levels of measurement: nominal (classi-cal, interval, and ratio. For a detailed
(i.e., disclos
disclosure fir
Similarly, the
as earnings-pe
of such inform
ordinal scale
obviously been
does represent
observed effec
three informat
support althou
last variable-
defined more t'.
measurement of
that Eskew and
ing theoretica
why a particul
present study,
of information
Another e
involves detec
39
ure firms versus non-lease firms, and the PV
26 ms versus the PV-IE disclosure firms ).
Gonedes study defined its information variable
r-share forecasts, and the authors' measurement
ation for testing market impact was in the
although a ratio level of measurement has
achieved. On the other hand, Bowman's study
an attempt to test an explanation for an
t. His multiple regression model included
ion variables, each with adequate theoretical
gh the focus was on the association test of the
-the leases-to-equity ratio. Eskew and Ro also
lan one information variable, and all the
information was in the ratio level. It appears
Ro have more difficulty than Bowman in justify-
Lly which variables to include (and exclude) and
ar variable has been measured that way. In the
both ordinal measurement and ratio measurement
are investigated.
(bually important issue in capital market studies
;ing "confounding events" and choosing between
discussion of parametric Sta 1956), pp. 21-
ASR No
level of measurement, see Sidney Siegel, Non-sistics for the Behavioral Sciences (New York,
value (PV) of interest rate on net income ized.
B O .
147 requires disclosure of (1) the present the minimum future lease commitments, (2) the mplicit in computing the PV, and (3) the impact income effect IE) if such leases were capital-
alternative ap
problem can be
slon studies d
available to t
event time and
effect is diff
studies review
Bowman used a
(the dependent
founding event
neglected due
Gonedes study
effects since
publication pe
in any specifi
reflected in t
Both Ro, and Es;
years must end
behavior of ea
was made to i
control or rem
sample compani
companies, not
de
40
proaches to controlling for them.2^ This
especially severe in accounting policy deci-
ue to the limited time period diversification
he researcher. In most policy decision studies,
real time are the same. Thus, time-period
icult to control or remove. Among the four
ed in the previous sections of this chapter,
60-month period to estimate the market risk
variable in his association test). The con-
s, if any, during that period can usually be
to the nature of this type of study. The
seems to have adequate control over time-period
returns of twenty-four earnings forecast
riods were observed and the confounding event
period might have been averaged out or
he beginning market prices of common shares,
cew and Ro specified that sample firms' fiscal
December 31, and they examined the stock price
ch firm in the same time period. No attempt
ntify the possible confounding events and to
,<bve their effects during the test period. If
<ps were not limited to calendar-year-end
only sample size can be increased but time
2 7„ For an
evidence on th4 alternative app Foster, op. ci
outline of a systematic approach to gaining confounding events problem and suggested
roaches to controlling for these events, see , pp. 24-34.
period diversi
becomes possib
companies' For
internal valid
price behavior
The informatic
disclosure of
1978, when the
accounting, th
had attempted
oil and gas ac
41
fication to average out the confounding effect
le. The present study utilizes different
m 10-K filing dates in an attempt to increase
ity of the study.
Summary
A large rjumber of empirical studies have used stock-
to assess the market impact of new information,
n of interest in this dissertation is the
oil and gas reserve values. Until August 29,
SEC ordered development of reserve recognition
e majority of studies conducted in this area
to measure the capital market effects resulting
from an announced elimination of full costing. Some of the
counting related market impact studies have
been briefly discussed in this chapter, although the review
of four studie
but that are m
the chapter.
Bowman, and by
Additionally,
to this disser
for the unders
in Chapter III
s that are not related to oil and gas accounting,
ethodologically relevant, has been the focus of
The studies by the Gonedes group, by Ro, by
Eskew and Ro have been carefully examined,
the relationship of each of these prior studies
tation has been explained, providing some basis
tanding of the research methodologies (presented
) developed in the present study.
This chap
employed to ac
information co
The methodology
measurements a
hypotheses and
selection and
chapter.
As noted
that stock mar;
prices reflect
The research i
of oil and gas
reserve value
the impacts of
be reflected i:
rather than at
The impor'
is measured in
previous chapt*
relevant to as:
specified, and
CHAPTER III
RESEARCH METHODOLOGY
ter provides a detailed discussion of procedures
hieve the research objective—assessing the
ntent of oil and gas reserve value disclosures,
ical assumptions, the research designs, the
nd the variables employed in the design, the
the appropriate test techniques, and the sample
data collection are all delineated in this
previously, the research design followed assumes
<ets are efficient to the extent that stock
public information rapidly and unbiasedly.
s then designed to examine stock price behavior
producers, primarily at the time when the
data are disclosed. In an efficient market,
such information, if any, would be expected to
a stock prices at or near the disclosure time
a later time.
ance of the manner in which relevant information
the research design has also been noted in the
r. Specifically, if attributes that are not
essing firms' values are inadvertently
if the release does have other attributes that
42
are relevant,
that no inform
opposite infer
against this p
the capital ma
developed and
specifies only
whereas the se
the reserve va
then there is a high probability of inferring
ation content is contained when, in fact, the
ence should be drawn. As an attempt to guard
ossibility, two research designs to examine
rket impacts of reserve value disclosure were
implemented in this study. The first design
one attribute of the reserve value data,
cond design incorporates several measures of
lue data.
In the fi
volves forming
unity. These
particular att
If the data co:
tion of the se
realized by th^
returns of the
measure of dat
framework of
the Gonedes st
forecasts.
th
The proce
measurement of
43
Research Design I
rst research design, the basic procedure in-
eight portfolios whose risks are forced to
portfolios differ only with regard to a
ribute of interest—a measure of "information."
itained no information pertinent to the valua-
urities, one would expect the stock returns
portfolios to be equal. If the realized
portfolio differ, one can infer that the
a. did provide information to investors. The
is design is based on the methodology used by
udy to test for information content of income
Measuring Information
cflure begins with the identification and the
the information variable. The information
variable exami
of estimated f
reserves (PV) .
variable, a pa
for each sampl
the PV of each
shares outstan
per share was
the firm's com!
disclosing dat
confound the t
employed by th
the disclosing
disclosing dat
report, or the
which the PV d.
the prices obs
all informatior
the per share
appropriate si
increase the i
44
ned here pertains to the disclosed present value
uture net revenues from producing proved
To test the information content of this
rticular measure of information was computed
oil and gas producing company. Specifically,
firm was divided by the total number of common
ding at the end of the fiscal year. The PV
then divided by the per-share closing price of
mon stock on the tenth trading day preceding the
to remove the possible size effect that may
st. The scaling process is similar to that
s Gonedes study. The tenth trading day preceding
date"1" is the portfolio formation date and the
js is taken to be the signature date of the 10-K
"amendment date" of the 10-K amendment, in
.ata appeared. Assuming capital market efficiency,
erved on the portfolio formation dates reflect
n publicly available as of those dates. Thus,
closing price of common shares seems to be an
e measure, which should be controlled to
nternal validity of the test.
Companie year, 10-Ks ar<fe fiscal year end pressure to de that permitted the required i before June 30
differ in their disclosure filing date. Every j; usually filed with the SEC in late March for "ed on December 31. However, due to the time yelop valuation data, a grace period was allowed calendar-year oil and gas companies to disclose
nformation for the year ending December 31, 1978, 1979.
Next, the
to their respe
subsequently d
Consequently,
folio 1 being
scaled PV per
firms that rep
folios 2 and 3
ranking. Beca
than a higher
share that put
favorable implji.
puts the firm
portfolios wer
preceding the
ct:
o
ut
In order
stock returns
tional distrib
folios of randcb
same time that
if three oil a
March 30, 1979
1979, and they
45
Forming Test Portfolios
firms were ranked in ascending order according
ive values of scaled PV per share and were
ivided into quartiles based on that ranking.
four test portfolios were formed, with port-
composed of those firms that reported the lowest
hare, and portfolio 4 being composed of those
orted the highest scaled PV per share. Port-
were formed from the middle quartiles of the
use a lower scaled PV per share is less desirable
caled PV per share, a level of scaled PV per
s a firm in the fourth quartile has a more
cation than a level of scaled PV per share that
into the first quartile of the portfolio. The
formed as of the end of the tenth trading day
disclosing date.
Forming Control Portfolios
o compare the distribution function of the
f the four test portfolios with the uncondi-
ion of the stock returns, four control port-
>mly selected securities were formed at the
the test portfolios were formed. For instance,
gas producing companies disclosed PV data on
with a portfolio formation date of March 16,
were grouped into test portfolios 1, 2, 4
nd
respectively,
selected and r
4 on March 16,
46
then three nondisclosing firms were randomly
andomly assigned to control portfolios 1, 2, and
1979.
Returns f
the disclosing
days after the
in each of the
returns are in
to allow for d.
in each portfo
(betas) for ea
market model,
by their betas
form a low-bet
the highest be
The average be
portfolio.
Next, a s
to one) was fo
average of the
low-beta sub-p
strained to sur
portfolio were
weighting fact
Obtaining Measure of Impacts
or twenty trading days (nine trading days before
date, the disclosing date, and ten trading
disclosing date) for the shares of each firm
eight portfolios were then computed. Since
fluenced by the degree of risk, it is necessary
ifferences in risk among the securities included
lio. Accordingly, measures of systematic risk
sh firm were obtained by applying the familiar
The firms in each portfolio were then ranked
The half with the lowest betas were used to
. sub-portfolio and the remaining firms with
as were used to form a high-beta sub-portfolio.
:a was computed for both sub-portfolios in each
.ngle portfolio with neutral risk (beta equal
•med for each portfolio by taking a weighted
high-beta sub-portfolio average beta and the
ortfolio average beta. The weights were con-
:i to one. The unadjusted returns for each sub-
subsequently multiplied by the appropriate
<brs and summed to arrive at the risk-adjusted
portfolio retu
for each tradi
returns from t
return differe
cally differen
St
Testing f
a difficult ta
functions. In
joint distribu
This implies t
tions of retur:
normal. Thus,
inequality of
inequality of
desired joint
test method th
the portfolio
Based on
normality assu
values of retu
^7
rns. Finally, return differences were computed
ng day by subtracting the control portfolio
heir respective test portfolio returns. The
nces were analyzed to see if they are statisti-
t from zero.
atistical Test and Null Hypothesis
or the inequality of distribution functions is
sk without restrictions on the nature of those
this study all conditional and unconditional
2 tions are assumed to be multivariate normal,
lat the conditional and unconditional distribu-
ns on particular portfolios are univariate
for a given pair of portfolio returns, the
distribution can be tested by testing for the
their means. In addition, in order to insure a
level of significance, one needs a multivariate
at simultaneously examines the equality of all
'eturn differences to zero.
•he research design, along with the multivariate
.mption, the main concern was whether the mean
n differences are equal to zero. Thus, the
Fama has returns, the n
concluded that, for monthly security ormal distribution is a good working approxima-oth the univariate t-test and its multivariate selling's T2 test are known to be robust against ;he normality assumption. See Eugen F. Fama, Finance (New York, 1976), and Richard J. Harris,
A Primer of Multivariate Statistics (New York, 1975), p. 87.
tion. Also, b analog, the Ho violations of Foundations of
null hypothesl
N1: Ther
test
ret u
The statl
Hotelling's (1
,2 T provides a
that the two p
group and cont
means on any o
fact the squar
that maximizes
maximized valu
Mathematically
When the null
following dist
with degrees o
pairs) and N-n
above exceeds
(no informatio
The secon
measures of PV
48
s was stated as follows.
e is no significant difference between the
portfolio returns and the control portfolio
rns .
stical technique to test this hypothesis is the
2
931) T statistic. Essentially, the Hotelling's
means of testing the overall null hypothesis
opulations from which the two groups (disclosure
rol group) are sampled do not differ in their
f the p (=4) measures. The T 2 statistic is in
ed standard t-statistic based on a weight vector
the value of the squared t-statistic, and this 2 2
s of t is equivalent to the value of T . 2 2
, T =MAX^t (W), where W is the weight vector,
hypothesis is true, the T 2 statistic has the
ribution: pn N _ n = N ~ n
T2
> n(N - 1)
f freedom n (=4 for the number of portfolio
(where N= 20). If the F value computed as
a critical tabular value, the null hypothesis
ft content) is rejected.
Research Design II
i research design is characterized by various
, no control group, multiple test periods, and
simultaneous t
The design dev
by Eskew and R
^9
est of systematic risk and abnormal return.
loped here is a modification of the one used
0 in their study of replacement cost disclosures
The first
on the reserve
carefully spec
risk and retur
and each was i
versus success
(Ij_) are defin<p
where:
PV
SH
Price
REV
NWC
Measuring Information
step was to define information variables based
value data. Each information variable was
ified so that various aspects of the firm's
1 could be satisfactorily explored and measured,
independent of accounting method (full-costing
ful-efforts) used. The information variables
d below.
= (PV/SH)/Price
= PV/REV
= [(NWC + PV - LTL - PS)/SH]/Price
= (PVC/SH)/Price
= PVC/REV
= present value of estimated future net revenues from production of proved oil and gas reserves, as of the end of the fiscal year;
= total number of common shares outstanding;
= market price of each common share at the beginning of the test period;
= total revenue reported on an historical cost basis;
= net working capital;
LTL
PS
PVC
Since the
nomic resourc
appear straight
proved mineral
at the end of
current year's
increase in pr
share of commo
year. In orde
ful to common
working capital
value and the
stock should b
(NWC + PV - LT
of these per s
price per shar
50
= Long-term liabilities;
= book value of preferred stock; and
= the portion of the PV attributable to proved reserves added during the current year.
PV represents the estimated value of the eco-
esj the meanings of PV/SH and PVC/SH above
forward. PV per share is the present value of
reserves per share of outstanding common stock
the fiscal year, and the present value of the
addition per share is the present value of the
oved mineral reserves during the year for each
n stock outstanding at the end of the fiscal
r to arrive at a reserve value that is meaning-
stockholders of the company, the amount of net
should be added to the end-of-year reserve
amounts of long-term liabilities and preferred
e deducted from the value of the reserves.
L - PS)/SH reflects such an adjustment. Each
hare variables was then divided by the beginning
of common stock to show how large or small the
^Ideally, mineral reserv value. Howeve gas producers other assets u and equipment. and amortizati the amount of identified and
book value of long-term assets other than e assets should also be added to the reserve r, the current reporting framework of oil and does not separate mineral reserve assets from nder the general heading of "property, plant,
Also, typically, depreciation, depletion, on are presented in the aggregate. Therefore, non-mineral reserve long-term assets cannot be added to the reserve value in this study.
51
variable is relative to the original investment (market price).
and Ijj are size-factor abstracted information Thus, 1^, I^
variables. Feb
total revenue
The seco
Unlike the fii
the second de
returns, syst^
taneously. Acf
portion of a
components:
The systemati
of movements c
market factor
reflects that
independently
use of the mark
can be used tc
(1) those ever
reflected in t
nd
r I2 and 1^, the PV and PVC were divided by the
4 m an another attempt to remove size effect.
Measuring Market Response
step was to define dependent variables.
'St design which focused on portfolio returns,
ign examined the two components of security
matic returns and unsystematic returns simul-
cording to the market model, the stochastic
ecurity return can be partitioned into two
. systematic component and a specific component.
component of return reflects the association
f an individual security's return with the
The specific component of return, however,
portion of a security's return that varies
of the market factor. A rationale behind the
•ket model is provided by the fact that the model
> classify events into two major categories:
.ts that have economy-wide effects, which are
he market factor, and (2) those events that have
George F methodological of accounting as one of the profile analys See George Fos Market Researc
oster in his discussion on various topics of problems in measuring the capital market impact policy decisions, has pointed out gross revenues size measures that should be considered in firm is when performing an internal validity check, ter, "Accounting Policy Decisions and Capital h," unpublished paper, Stanford University, 1980.
impact on a pa
specific compo
useful in expl
or unusual ret
For the p
clear a priori
response woul
simultaneously
the market mod
disclosure sam
that its chang
changes.
Unlike th
a twenty-tradi
the second des
might have res
value data wer
The folio-
SEC reserve va
1. Relea
propo
state
inclu
1978.
52
rticular security, which are reflected in the
nent. Therefore, accounting information may be
aining either the systematic risk of a security
urns associated with firm-specific events.
resent design, the assumption that it is not
which component best reflects the market
d allow the researcher to examine both components
Both beta and residuals were estimated via
.el for the individual common stocks of the
pie firms. The residual return was squared so
es can be clearly highlighted as opposed to no
Test Periods
e first design of this study, which focuses on
ng-day period centered on the disclosing date,
ign allowed for the possibility that the market
ponded to the SEC's decision before reserve
actually published.
wing events took place during the course of the
lue disclosure decision.
se No. 33-5878, dated October 26, 1977—initial
al for disclosing reserve values in financial
ments beginning in filings with the SEC that
de fiscal years ending on or after December 25,
2. Relea
proposed that the reserve value data be disclosed
outside financial statements for fiscal years
ending after December 25, 1978, but before
December 25, 1979.
3. Relea|se No. 33-6008, dated December 19, 1978—
final
S-K r
quant
4. The f
value
To design
week of the cr
after the week
gate possible
test periods w
was on the las
initial propos
tured by the t
II and period
No. 33-5967 pr
capital market
the market res
reserve value
by the SEC, is
53
se No^ 33-5967, dated August 31, 1978—
rules that incorporated into Regulation
equirements for the disclosure of reserve
ities and reserve values.
irst public disclosure of the actual reserve
data by individual firms.
the test period, a thirteen-week period (the
itical event, six weeks before and six weeks
of the critical event) was adopted to investi-
market reactions to each event, and these four
ere tested simultaneously, although the focus
t period. Accordingly, any effects of the
al itself on the market is supposed to be cap-
sts for period I. The test results for period
III should reveal the effects of the Release
oposal and the announcement of adoption on the
, respectively. Similarly, it is assumed that
ponse to the (first) actual disclosure of the
data in the firms' Form 10-K reports, as required
reflected in the test results for period IV.
Nu
Given the
formed as foil
N2: No 1
mark
risk
of i
The null
variate canoni
of canonical c
from each of t
correlation be
The analysis i
measure canoni
variables. Th
importance of
the variabilitly
54
11 Hypothesis and Statistical Test
two sets of variables, the null hypothesis was
ows.
inear dependence exists between the set of
t based variables, represented by systematic
and average squared residuals, and the set
The sampl
and gas produc
fiscal years e
nformation variables represented by 1^.
hypothesis was examined by means of the multi-
al correlation analysis. The basic strategy
orrelation is to derive a linear combination
ne sets of variables in such a way that the
5
tween the two linear combinations is maximized,
s performed in two steps. The first step is to
cal correlations between the two sets of
second step is to determine the relative
a particular variable in one set in explaining
of variables in the other set.
Data Collection
Sources
s of firms used in this study is composed of oil
srs reporting reserve value data to the SEC for
iding after December 25, 1978, but before
For an e Cohen and Patr tion Analysis 1975) .
xtensive discussion of the technique, see Jacob icia Cohen, Applied Multiple Regression/Correla-for the Behavioral Sciences (Hillsdale, NJ,
December 25, 1
of oil and gas
list compiled
6
Study. Pour
the list:
1. U.S.
of Co
979. First, a list of the names and addresses
companies was developed based on the company
by the FASB staff for Dyckman's Exposure Draft
sources were used by the PASB staff to compile
Securities and Exchange Commission. Directory
mpanies Required to File Annual Reports with
the S ecurities and Exchange Commission. Washington,
D.C.,
June
2. John
Outlo
3 • The 1
Expos
Accou
1977,
4. N. C.
Succe
Indus
Conse
Michijg
The selec
the selection
55
Government Printing Office, June 30, 1975 and
30, 1977 editions;
S. Herold, Inc. "Report Inventory," Petroleum
ok, February, 1977, p. 39;
ist of firms in L. J. Seidler. "Oil and Gas
ure Draft Outlaws Pull Cost Accounting,"
flting Issues, Bear Stearns & Co., August 15,
pp. 4-13;
O'Connor and D. W. Collins. "Full Cost vs.
ssful Efforts Accounting in the Oil and Gas
tries: A closer Look at the Potential Market
^uences." Unpublished paper, Lansing, Michigan,
an State University, May, 1977.
;ed sources were believed to be adequate. Also,
f specific sources was considered to provide an
6, For the
Thomas Dyckman on the Returns Conn., October
'esulting list of 346 different companies, see Report on the Effects of the Exposure Draft of Oil and Gas Company Securities (Stamford, 19771, Part 3, pp. 12-77.
objective limi
ated. Next, 3
copies of thei
as any amendme
responded posi
price data wer
published by S
56
tation on the number of companies to be evalu-
00 firms from the list were asked to send
r 1976, 1977} and 1978 Form 10—K reports as well
nts to the filings. A total of 150 firms
tively in time for this study. The security
e obtained from the Daily Stock Price Record,
tandard & Poor's Corporation.
fi
The folio-
the 150 respon
Research Desig:
1. The
net r
oil a:
of th<fe
of the
of the
10-K
appea
two f
after
2. The cd>
recent
must
to inct
Sample Selection
Aring sample selection criteria were applied to
dents to arrive at the final test sample of
1 I.
Lrm did report the present value of future
evenues from estimated production of proved
.nd gas reserves. The total number of shares
firm's common stock outstanding at the end
fiscal year as well as the signature date
10-K report or the amendment date to the
Amendment, wherever the present value data
red, must be identifiable. One-hundred-and-
rms out of the 150 respondents were retained
the first screening process.
mpany's total revenue, as reported in a
annual report to stockholders or Form 10-K
e less than five billion dollars. In order
lude in the sample only firms engaged in oil
57
and das exploration and production activities
(i.e., those that could be affected by the
reserve value disclosure), large integrated com-
panies were eliminated. Twelve companies with
annuajl total revenue exceeding five billion
dollajrs were excluded from the sample.
Price data on transactions of the company's stock
must be available from the Daily Stock Price
Recorld. Closing prices on each of fifty-three
Fridays (December 30, 1977, to December 29, 1978),
and closing prices on twenty-one trading days
surrounding the disclosing date were required.
Twelve companies were eliminated because of the
non-availability of stock price data.
Price data on transactions of the firm's stock
must be sufficient during the test period and the
beta estimation period. Six companies were elim-
inated from the study due to insufficient price
data. Even though the Daily Stock Price Record
reports these six companies, their securities were
traded relatively infrequently during the periods."^
7 Lev excl
Draft study in oil and gas com; rience days of are only broker actions during Unfortunately, present study w size. To maint
uded all OTC firms from his Exposure which he examined the daily stock returns of panies. He argues that many OTC stocks expe-:io trading and the quoted bid-ask OTC prices i3' estimates rather than the result of market days of no trading in the specific stocks, the elimination of all the OTC firms from the ould have significantly reduced the sample ain a balance between internal validity and
The above
dents to a tot
have an equal
company with t
companies was
sample compani
thirty-two New
Stock Exchange
As indica
of randomly se
was that price
must be availab
New York Stock
selected from
The sampl
began with the
Two additional
was modified t
1. The fi
fi
long
the
calcut
2. That
to pr<b
58
elimination procedures reduced the 150 respon-
al of seventy-three companies. In order to
number of securities in each portfolio, one
he highest total revenue among the seventy-three
further excluded. Thus, the final number of
;s was seventy-two companies comprised of
York Stock Exchange Firms, nineteen American
Prims, and twenty-one over-the-counter firms,
ted earlier, the control sample was composed
lected securities. The only selection criterion
data on transactions of the company's stock
le from the Daily Stock Price Record for the
Exchange. Seventy-two firms were randomly
he NYSE listing.
e selection procedure for Research Design II
seventy-three firms used for Research Design I.
criteria were employed and criterion 3 above
d determine the final sample of the second design,
Lrm's total revenue, net working capital
erm liabilities, and preferred stock for
Lscal year must be readily identifiable and
able.
portion of the present value attributable
ved reserves added during the current year
external valid, volumes were e Impact of Acco of Oil and Gas 1979), 485-503
ty, only those firms with very thin trading ccluded from the sample. See Baruch Lev, "The unting Regulation on the Stock Market: The Case Companies," The Accounting Review, 54 (July
must
compu
Price
must
Recor
1977,
to Oc
Pebru
round
The result
59
be separately disclosed or can be readily
ted from the present value.
data on transactions of the firm's stock
oe available from the Daily Stock Price
1. Friday closing prices from September 9,
to December 9, 1977; from July 14, 1978,
ober 13, 1978; from November 3, 1978, to
ary 2, 1979; and for thirteen weeks sur-
ng the disclosing week were required.
was the inclusion of fifty-eight firms in the
the second design. The names of firms final sample o
included in each research design, along with the trading
Tirm's common stock, are shown in Appendix A. market of each
The daily
individual firiji
from the Daily
(for Design I)
price on the pr
current trading
relationship o
the week (usua
trading day of
returns, R i t, 1
in the followir
Stock Returns Computation
stock returns and weekly stock returns for
.s that are used in this study were computed
Stock Price Record. The daily stock returns
are based on the relationship of the closing
'evious trading day to the closing price on the
day. The weekly returns are based on the
the closing price on the last trading day of
Friday) to the closing price on the last
the previous week. Specifically, the stock
ere computed from the daily or weekly prices
g manner:
n y
where
Rit
Pit =
Notice that th
include cash di
the exclusion
R., = ln(P., v P. . . ) it it i,t-l'
return on stock i in period t ,
closing price of stock i in period t
(adjusted for stock splits and stock dividends).
s return calculation in the present study did
vidend data. Sharpe and Cooper have shown that
Df dividends from the return data does not
8 alter the statistical results of such empirical studies.
The marke
estimate the r
an individual
is assumed in
expressed as
where R., and it
market portfol
See Will Classes of New Financial Anal}
William Market Equilib Finance, 19 (
10T , T. John Li Selection of R: Budgets," Revi^ 1965), 13-37.
60
Beta Estimation
y 10 model of Sharpe and Lintner was used to
elationship between the return on the shares of
firm and the market portfolio. This relation
he market model to be linear and it may be
Rit ai + 6i(Rmt) + eit
Rmt a r e t h e P e r i o d t returns of firm i and the
.o, respectively; cu and 31 are the regression
.am F. Sharpe and Guy M. Cooper, "Risk-Return York Stock Exchange Common Stock, 1931-1967," sts Journal, 28 (March-April, 1972), 46-54.
Sharpe, "Capital Asset Prices: A Theory of •ium under Conditions of Risk," Journal of
September, 1964), 425-442.
ltitner, "The Valuation of Risk Assets and the sky Investments in Stock Portfolios and Capital w of Economics and Statistics, 47 (February,
coefficients;
sumed to satis
Ordinary
observed value
for the regres
These coeffici
weekly returns
weekly returns
Design II. Th
the Standard &
The a vera
computed as fo]L
1. Obtai
where
t,
stock
and
2. Ob tail
where
U., i it
61
and e l t is the residual return, which is pre-
fy standard regression assumptions.11
least-square regression was performed on the
s of R and R , to obtain estimates a. and b. 11 mt 1 1
sion coefficients cu and respectively.
snts were estimated by using the fifty-two
in 1978 for Research Design I and thirteen
surrounding each cirtical event for Research
e market surrogate used for both estimates was
Poor's 500 Composite Index.
ge squared residual in Research Design II was
lows.
ft residual return:
U., = R., — (a. + b . R ,) it it 1 1 it
R ^ is the actual return on stock i in week
(a^+b^R^) is the forecasted return for
i in week t.
1 average squared residual return:
ASR = (1/13) E (U?.) t=l i r
ASR is the average squared residual return and
the residual return on stock i for week t.
11 The assumptions are
1) E(e ) = 0,
2' a<-Rint>eit' ° °> 3 ) o ( eit' ejt ) = °'
62
Summary
In this chapter, the information of interest—the dis-
closure of reserve value data—has been translated into
independent variables of two research designs. Two major
hypotheses wer^ stated along with the statistical techniques
to be utilized; for testing the data collected from the com-
panies' Form 1D-K reports and the Daily Stock Price Record.
The test results are presented in Chapter IV.
Chapter I
conducted on (
(2) the associ
the market-bas
statistical hy
first section
of the first r
findings of th
Appendices B a
results report
As indica
research desigJi
adjusted returi
portfolios duri
value disclosu
each portfolio
beginning of tl(i
between the ac
ascribed to th£
disclosure. Th
CHAPTER IV
RESULTS OP THE STUDY
V presents results of the statistical tests
1) the risk-adjusted portfolio returns and
ation between the reserve value variables and
d variables, along with evaluations of the
potheses stated in the preceding chapter. The
Df this chapter includes the empirical results
esearch design; the second section details the
e analysis of the second design. In addition,
id C provide further analysis to support the
d in this chapter.
Results of Research Design I
;ed in the preceding chapter, the first
is intended to determine whether the risk-
is of the test portfolios and the control
ng twenty trading days surrounding the reserve
e date are significantly different. Because
s market risk was forced to equal unity at the
e test period, any significant difference
;ual returns of the matched portfolios could be
information conveyed by the reserve value
2 .e Hotelling's T test was used to test the null
63
hypothesis N1.
null hypothesl
each of the tw
control group
The Input
The number ass
This method provides a test of the overall
3 that the means of the four portfolios from
d populations from which the test group and
tfere selected do not differ.
2
data for the T test are presented in Table 1
igned to a portfolio pair is the quartile
TABLE I
DAILY DIFFERENCES BETWEEN RETURNS ON MATCHED PORTFOLIOS
64
Trading
Days
Return Difference for Portfolio Pair
1
- 9 - 8
- 7 - 6
- 5 -4 - 3 - 2
-1 0 1 2 3 4 5 6 7 8 9 10
0 .0035 0 .0037 •0.0014 0 .0041 0 .0047 0 .0077 0 .0265 0 .0055 0 .0201 •0 .0066 0.0001 0 .0013 0 .0084 0.0020 0 .0068 0 .0013 0 .0054 0 .0002 0 .0003 0 .0169
0 .0179 -0 .0135 -0 .0068 0 .0069
- 0 . 0 0 5 1 0 .0140 -0.0122 0 .0014 0 .0039
- 0 . 0 1 7 1 -0 .0117 -0 .0033 0 .0076 0 . 0 0 9 1 0 .0063 -0 .0028 -0 .0119 0 .0004
-0 .0033 •0 .0037
0 .0130 0 .0023
-0 .0047 -0 .0006 0 .0032
-0 .0014 0 .0049 0 .0098 0 .0065 0 .0140 0 .0158 0 .0061 0 .0059
-0 .0070 0 .0104 0 . 0 0 6 1
-0 .0043 -0 .0158 -0 .0097 0 .0061
-0 .0092 - 0 . 0 1 7 1 -0 .0119
0 .0044 0 .0204 0 .0093 -0.0010
0 .0065 0 .0076
- 0 . 0 1 7 1 0 .0404 0 .0125
- 0 . 0 1 6 9 - 0 . 0 1 6 5
0 . 0 0 6 1 0 . 0 0 4 3 0 .0176 0 .0036
- 0 .0065 - 0 .0093
to which the
twenty trading
component test sample portfolio belongs. The
days surrounding the disclosing date (denoted
as zero) are li|sted in the first column of the table. Each
figure in the t
portfolio and t'.
2
The T re
mean is an esti:
on a test portf
RESULTS P ON
able is the return difference between the test
ne control portfolio.
suits are given in Table II. Each estimated
nate of the mean difference between the return
Dlio and the return on a control portfolio.
TABLE II
ROM T2 TEST ON DIFFERENCES BETWEEN RETURNS MATCHED MINIMUM VARIANCE PORTFOLIOS
WITH RISK EQUAL TO UNITY
65
Number of Portfolio Pai
Means of Differences between
Returns on Portfolios
Standard Deviations of Difference
between Returns on Portfolio
1
2
3
4
0 .0043
- 0 . 0 0 1 2
0 .0030
0 .0014
0 .0085
0.0096
0 .0083
0 .0150
value of T = 4
value of F(3,17
probability = 0
*F(n jN-n) portfolio pairs estimation. Sel
2360
= 1 .2634*
3148
2 = (T )(N - n)/n(N - 1), where n is the number of and N is the number of observations used for .ected fractiles of F ( 3 , 1 7 ) are as follows:
F ( 3 , 1 7 ) 1. 51 1 .72 2 .44 3.20 4 . 0 1 5 .18
Fractiles 0 .750 0.800 0 .900 0 .950 0 . 9 7 5 0 .990
Each estimated
deviation of su
portfolio is a
relative risk e
tandard deviation is an estimate of the standard
a differences between portfolio returns. Each
.inimum variance portfolio with estimated
c|ual to unity.
oh
n:
The compu
with the hypot
test portfolio
The correspond)!
below the .10
the test result
returns of com
significantly
panies in any
The findi
content may al
"sign" of the
portfolios in
quartiles of t
return in exce
control sample
mean of differ^
greater than t
value, this me
lower relative
PV per share—
expectation th
than a higher
the daily retu
obtained from
four portfolio
66
ted value of T given in Table II is consistent
lesis that there is no difference between the
returns and the control portfolio returns,
ing value, 1.263^, of F(3,17) is substantially
significance level. Therefore, on the basis of
s, the conclusion is reached that portfolio
panies disclosing reserve values did not behave
differently from randomly selected control com-
Df the trading days involved in the test period,
ng that the disclosures contained no information
o be inferred from a simple observation on the
means of daily differences between returns on
Table II. Both the lowest and the highest
ie sample of oil and gas firms have an average
s of the average return on their matching
portfolios. However, the magnitude of the
nces for the first portfolio is substantially
lat of the fourth portfolio. Taken at face
ans that on the average the market has assigned
values to firms that reported a higher scaled
result not consistent with the pre-test
at a lower scaled PV per share is less desirable
caled PV per share. Additional insights into
n differences on portfolios can also be
Figure I, in which return differences on the
pairs were cumulated and plotted in a time
+
C t i
r r
• - d
C ' j
3 . H c,..< ( 1 ) ^
series. Notic
ences, and not
test, are plot|t
variability pi
differences is
nificant than
is, the variab
return differe
merely the tiro
shows that por
differently in
portfolios app'
folios. Initi
than its respe
has a better p
The performanc'
deviate apprei
general, the f
with unique ma
surrounding th-
The prima
detect if ther
attributes of
of market resp
68
e that in the figure cumulative return differ-
the average return differences used in the
ed. As a consequence, the same amount of
Dtted in a time series of cumulative return
more likely to render it statistically insig-
the related average return difference. That
ility visually represented in a cumulative
nee plots affects the entire cumulation and not
series of the underlying average. Figure 1
fcfolios 1 and 3 did not perform significantly
the test period. However, both of these
ear to outperform their matched control port-
ally, the performance of portfolio 4 was poorer
3tive matched portfolio. However, the portfolio
srformance in the second ten days of the period,
of portfolio 2 during the test period did not
iably from its matched control portfolio. In
our portfolio pairs appear not to be associated
rket behavior during the twenty trading days
reserve value disclosing date.
Results of Research Design II
ry goal of the second research design is to
e is a linear relationship between a set of five
reserve value data and a set of eight measures
onse related to the SEC policy decision.
Canonical corr
maximum amount
variables, was
Sig
slation analysis, which aims at explaining a
of the relationship between two sets of
used to examine the null hypothesis N2.
nificance of Canonical Correlations
The resul
summarized in
1. The c
the first colu:
correlations c
smaller set.
puted. The me
that of a simpfL
coefficient.
simple correla
for a parti
between the tw
first canonica
between the fi
canonical corr
69
ts of the canonical correlation analysis are
Table II, which contains the following.
anonical correlations. These are presented in
.:im of the table. The total number of canonical
Dmputed equals the number of variables in the
Thus, five canonical correlations were com-
aning of canonical correlation is equivalent to
e product-movement (Pearson) correlation
Specifically, a canonical correlation is a
bion between the two canonical variates (X. and 1
2
ular i), and is not a direct correlation
0 sets of original variables. Therefore, the
1 correlation, .7107, is a simple correlation
est pair of canonical variates. Each remaining
elation is interpreted in the same way. Canonica
noted in the p analysis is to sets of variab the two linear linear combina canonical vari intercorrelati involved. A s selected to ac between the tw> first canonica
I variates are pairs of linear combination. As receding chapter, the technique of canonical derive a linear combination from each of the les in such a way that the correlation between combinations is maximized. Many such pairs of tions may be derived. The first pair of ates are selected so as to have the highest on possible, given the particular variables econd set of canonical variates is then 3ount for a maximum amount of the relationship d sets of variables left unaccounted for by the 1 variates, and so forth.
2. The m
in the second
equal in numbe
The eigenvalue
and they can b
shared by the
spond. Thus,
canonical vari
remaining eige
CANONICA VARI
70
agnitudes of the eigenvalues. These are present
column of the table in descending order and
r to the number of canonical correlations.
s are squared values of canonical correlation,
e interpreted as the proportion of variance
pair of canonical variates to which they corre-
the variance shared between the first pair of
ates is approximately 50 .5 percent. Each
nvalue is interpreted in the same way.
TABLE III
L CORRELATIONS RELATING FIVE INFORMATION ABLES TO EIGHT MARKET-BASED VARIABLES
Number Canoni Correla
cal tion Eigenvalue
Wilk's Lamb da
Chi-Square D.F. Signif-icance
1 0 .710 74 o.50515 0 .34563 53 .11993 40 0.080
2 0 .372 27 0.13859 0.69845 17 .94493 28 0.928
3 0 .332 42 0.11050 0.81081 10.48596 18 0 . 9 1 5 4 0 .280 50 0 .07868 0 .91154 4 .63101 10 0 .914
5 0 .103 04 0.01062 0.98938 0 .53375 4 0.970
3. The n
to the signifi
statistic give
null hypothesi
the two variab
variates (if a
significance,
column, follow
ext four columns provide information relative
cance of the canonical correlations. The first
n is Wilk's lambda, which is used to test the
s that no residual linear association between
le sets remains after the preceding canonical
ny) have been extracted. For an actual test of
the chi-square values are given in the next
ed by the degrees of freedom and the actual
significance v
the canonical
hypothesis N2,
of the five co
.10 but not si
these results,
exists some re
the two sets o
earlier, infor:
canonical vari
Accordingly, tji
analysis to de
variables is "
These issues a
variate coeffi
hypotheses fori|ni
variables.
The exact
sponding pairs
first canonica
the first colurti:
for each varia
of the origina
For instance,
71
alues in the last column. An examination of
correlations indicates that for the null
only the first canonical correlation, .7107,
:nputed is significant at a probability level of
gnificant at a probability level of .05. Given
it seems reasonable to conclude that there
lationship between the canonical variates of
f original variables. However, as indicated
mation presented in Table III is drawn from the
a.tes, not the two sets of original variables,
ere is sufficient justification for further
ermine which variable or combination of
•esponsible" for the detected relationship,
•e explored by examining standardized canonical
ients, and by a separate testing of sub-
ied on selected combinations of the original
Relative Importance of Variables
information on the composition of the corre-
of the canonical variates, which produce the
correlation in Table III, is presented in
m of Table IV. The size of the coefficients
•tj>le are indicative of the relative contribution
variables in composing the canonical variates,
the residual return variable (ASR3) of
period III has
the largest of
This implies t
RELATI BY
a structure coefficient of -.8133, which is
all structure coefficiencies in Table IV.
lat its variability is explained better by the
TABLE IV
VE IMPORTANCE OP VARIABLES REPRESENTED THE STRUCTURE COEFFICIENTS FOR THE TRST PAIR OF CANONICAL VARIATES
72
Variable 3 Structure
Coefficients Ranking of Importance
II. -0.1437 4
X2 -0 .0392 5
J3 0 .5564 2
X4 -0.3453 3
X5 -0.7038 1
B1 -0.4032 3 B2 0.0684 7 B3 -0.3654 4 B4 0.3599 5 ASR1 0.4461 2
ASR2 -0.0227 8
ASR3 -0.8133 1 ASR4 -0 .2147 6
five informaticj)
market-based va
separately rani
resulting ranks
A careful
the informati
First, I^ has
on
a
n variables than the variability of any of the
.riables, and the information variables can be
-ordered and interpreted similarly. The
are provided in the second column of Table IV.
examination of the structure coefficients for
variables reveals two interesting points,
substantially larger structure coefficient
than I and 1^
value adjusted
term liabiliti
explain the va
than the unadj
PVC related va
variables than
An examin
market-based v
return variable
III and I appe
and IV. The s.
ranking of str'
variables amon;
importance of •
return variable
the structure •
analysis is ve:
relative impor'
in forming the
analysis does
draw a conclus
test is perfori|n
issue, separat
formed to evalili
combinations o
Thus, it seems that the reported reserve
for the amounts of net working capital, long-
s, and preferred stock appears to better
riability of the market-based variables
usted reserve value. Second, except I^, the
73
riables are, in general, better explanatory
the PV related variables.
ation of the structure coefficients for the
ariables indicates that among the residual
s for the various periods, those for periods
ar more important than the ones for periods II
ame result can be observed by comparing the
acture coefficients for the systematic return
g the four periods. However, the relative
he systematic return variables and the residual
s cannot be determined based on the results of
coefficients. The structure coefficient
y useful in the sense that it explores the
;ance of the original variables in the subset
canonical variates. However, a relative
not provide any basis for the researcher to
ion from the result since no formal significance
ed. To provide additional insight into the
canonical correlation analysis can be per-
ate a set of sub-hypotheses based on particular
variables of interest, as discussed below.
dditional Tests of Sub-Hypotheses
Sub-hypot
determine (1)
(2) the existe
the set of sys
information va
variables and
information co
and the inform
value. All th
That is, there
variables. Ta
variables exam
74
heses one through eight were developed to
the timing of significant market response,
nee or non-existence of relationships between
tematic return variables and the set of the
riables, and between the set of residual return
the set of information variables, and (3) the
atent of the static measure of reserve value
ation content of the flow measure of reserve
e sub-hypotheses take a form similar to N2.
is no linear relationship between two sets of
Die V details each sub-hypothesis along with
ined.
TABLE V
SUk-HYPOTHESES AND VARIABLES INCLUDED
Sub-Hypotheses
51
52
53
54
55
56
57
58
nformation Variables
All
All
All
All
All
All
> 25 " 2
x5
Market-Based Variables
Bl, ASR1
B2, ASR2
B3, ASR3
B4, ASR4
Bl to B4
ASR1 to ASR4
All
All
Representing
Period I
Period II
Period III
Period IV
Systematic Returns
Residual Returns
PV
PVC
The resul
evaluate the e
Notice that on
SELE
75
ts of using canonical correlation anaylsis to
ight sub-hypotheses are presented in Table VI.
ly the information related to the first pair of
TABLE VI
CTED RESULTS OF CANONICAL CORRELATION ANALYSIS FOR SUB-HYPOTHESES
Sub-Hypotheses C
Canonical orrelation
Chi-Square
Degree of Freedom Significance
SI 0 .5173 19 .4102 10 .035** S2 0 .2265 3 .7232 10 .959 S3 0 .5406 20.9605 10 .021** S4 0 .3289 7.5064 10 .677 S5 0 .5618 27.5704 20 .120 S6 0 .5264 23.3905 20 .270 S7 0 .3914 17.2069 24
O
-r 00
S8 0 .6489 32.5076 16 .009*
*p < .01: sig **.01< p < . 05 :
lificant at the .01 probability level significant at the .05 probability level
canonical vari
the table. Fr
the following
Among the
with the timin,
and period III
This finding i
cient analysis
supports the r$
ates of each association test are reported in
3m the statistical results presented in Table VI,
reservation can be made.
first four sub-hypotheses, which are concerned
of significant market response, only period I
are significant at the .05 probability level.
consistent with the previous structure coeffi-
for the major hypothesis N2. The finding also
suit of the first design that no information
content was de
data were actu
component, sys
market respons
disclose any s
rejection of t
level, along w
ship exists be
variables, sug
concerned with
additions than
A descrip
been presented
indicates that
contention tha
groups are equ
disclosed. In
associated wit:
significance (
associated wit
The chi-s
analysis indie
.10 level of s
variables and
76
tected during the period when the reserve value
ally disclosed. With respect to which return
tematic returns or residual returns, reflects
e better, the separate association test did not
ignificant relationship in either case. The
he sub-hypothesis S8 at the .01 significance
ith the finding that no significant relation-
tween PV-related variables and all market-based
gests that market participants are more
the present value of current-year reserve
the end-of-year reserve value.
Summary
tion of the results of statistical tests has
2 in this chapter. The Hotelling's T statistic
the evidence was insufficient to reject the
t the portfolio returns of the test and control
al during the period the reserve value data were
fact, the probability (p=.3l84) actually
a findings, when compared to the .10 level of
y. = .10), infers that little inherent danger is
i failing to reject the null hypothesis Nl.
quare statistic in the canonical correlation
ates that there is a linear relationship at the
Lgnificance between the set of five information
the set of eight market-based variables. The
rejection of t
mation effects
disclosure dec
measured by in
returns. It a
what particula
the detected 1
the variables
identified bas
Separate assoc
also been cond
evidence sugge
sively in peri
value of the c
market impact
and (3) the adj
than the unadj
variability in
77
he null hypothesis N2 suggests that the infor-
of the various events of the SEC reserve value
ision were present when the effects were
dividual firm's systematic returns and residual
lso justifies a further analysis to determine
r combination of variables contirbute most to
inear dependence. The relative importance of
in each set of variables included in N2 has been
ed on the structure coefficient analysis.
iation tests (by canonical correlation) have
ucted on eight sub-hypotheses. The additional
sts that (1) the market responded quite inten-
od I and period III, (2) the reported reserve
urrent year's additions seems to have larger
than the reported end-of-year reserve value,
justed reserve value is relatively more important
usted reserve value in explaining the total
the market-based variables.
Based on
results are in
chapter also s
As noted
usefulness of
the "Descripti
the SEC for th
and on or befo
belief that "m
gas producers)
mental data to
by the Commiss
The empir
CHAPTER V
CONCLUSIONS AND SUMMARY
the findings in Chapter IV, in this chapter
terpreted and conclusions are drawn. The
ummarizes the study.
Interpretation and Conclusion
in Chapter I, many people have doubts about the
reserve value data because of imprecise estimates
of both quantities and value of proved oil and gas reserves.
Nevertheless, the SEC took its first step in the development
of RRA by mandating the disclosure of reserve value data in
Dn of Property" section of reports filed with
e fiscal year ending after December 25, 1978,
re December 25, 1979. The SEC expressed its
eaningful analyses (of operations of oil and
would focus almost exclusively on the supple-
be disclosed persuant to the rules" prescribed
.on.
cal findings of the first design that the
evidence was not sufficient to reject the hypothesis that the
Securiti quirements for for Oil and Ga^ 34-15108; 35-0 1978), p. S.
s and Exchange Commission, "Adoption of Re-Financial Accounting and Reporting Practices Producing Activities," Releases Nos. 33-5966;
^688; IC-10382; AS-253 (Washington, August 31,
78
disclosures co
investors do n
effective sign
limitations an
possible expla
1. Inves
present value
an insight int
reserves and a
basis for comp
However, the 1
value and the
gas producers :
explanation is
the lowest sea
better than fi
2. Pailu
that an effect
The most likelj/
be the initial
dated October
proposed to re<h
assets employed
the present val
information wa$
the second res
79
ntained no new information, suggests that
ot believe that such reserve value data are
als for pricing securities. Subject to the
d assumptions discussed in Chapter I, three
nations of these results can be made.
tors are confused over the implication of the
data. They may believe that value data provide
o the qualitative differences of oil and gas
ttempt to use the reported present value as a
aring proved reserves of different companies,
inkage between the magnitude of the reported
assessment of the financial position of oil and
las not been consistently established. This
supported by the finding that firms reporting
led PVs per share appear to perform relatively
rms generating higher scaled PVs per share,
re to find an information effect does not mean
may not have taken place at an earlier date.
time for an earlier response would appear to
proposal of the SEC in Release No. 33-5878
-6, 1977, which for the first time explicitly
uire that registrants with mineral resource
in oil and gas producing activities disclose
ue data on proved reserves. Whether new
released at that time has been examined in
(fearch design of this study. Release of
information at
the market to
value disclosu
expect a furth^
3. The s
not have inforifi
items in the s
measure of pro
economic resout
prior to the b
have shown som^
related financ
market values.
does not appeaif
risk. Accord!
any new inform,
beyond what had
The test
information imp
disclosure of
reserve value c
related variabl
rejection of th
80
an earlier date could explain the failure of
•eact to information contained in reserve
es. If so, however, there is no reason to
r reaction upon actual disclosure of the data,
atic measure of oil and gas reserve value does
^ation content. Similar to many other asset
tatement of financial position, the static
yed reserve value represents the estimated
ces that have been generated and accumulated
alance sheet date. Prior empirical studies
relationship between certain balance-sheet-
al ratios, such as financial risk and firms'
The reported end-of-year reserve value
to have any direct impact on firms' financial
rjigly, it appears that the data do not convey
tion about the financial risk of the firm
already been known to the market.
•esults of the second design indicate that the
act of the various events (including actual
eserve value data) of the SEC's decision on
isclosure were present when five reserve value
es were examined. Two possible reasons for
e null hypothesis are offered. First,
Hamada sh related to its the Firm's Capi Stock," Journal
ows how the firm's financial risk is market value. See R. S. Hamada, "The Effect of tal Structure on the Systematic Risk of Common of Finance, 27 (May, 1972), 435-452.
although the s
appear to have
value measure
(Ijj and Ij.) ml
the sense that
stockholders'
operating resu
adjusted reser
of proved rese
minus the amou:
stock) may be
position and f.
alone. This 1
In evaluating
In the case of
additions, thl
appear In the
prepared on th^
may provide a
success In find
In Its hlstorl
Second,
be ascribed to
is
th
81
;atic measure of the present value data did not
Information content, the adjusted reserve
and the flow measures of reserve value
ght have Information content to Investors, in
the data conveyed new information about the
potential claims against reserves, and about the
Its of the current period. Specifically, the
/e value (which is the discounted present value
ves, plus the amount of new working capital,
ifits of long-term liabilities and preferred
better indicator of the firms' financial
nancial risk than unadjusted reserve values
^formation could then be utilized by investors
1phe firms' investment and financing decisions.
the present value of current period reserve
3 is one of the two major components that would
^ummary of oil and gas producing activities
basis of reserve recognition accounting. It
etter measure of a firm's success or lack of
ing proved reserves than is currently reflected
(jial cost income statement.
e existence of a linear relationship may also
the inclusion of three additional test periods
The othe See Securities —Supplemental Accounting," Rej; IC-10 875; AS-2q)_9_ (Washington^ September 24, 1979) .
•> component is "revisions to proved reserves." and Exchange Commission, "Oil and Gas Producers Disclosure on the Basis of Reserve Recognition lease Nos. 33-6126; 3^-16218; 35-21222;
reflecting thr
value disclosu
the second expfL
design previou
be maintained
disclosure of :
market. It may
possibility b
effect. Howevi
a designation
prior events s^
impacts.
Of course
necessarily mut[
These two inte
the structure
relating to the
period reserve
response. Evid
period reserve
that a signific
proposal was re
the proposal wa
values, adjuste
and preferred s
reserve value m
82
ee critical events leading to the SEC's reserve
.re decision. This explanation is similar to
anation of results of the first research
ly discussed. In efficient markets, it could
that various actions taken by the SEC to require
eserve value data gradually affected the
be harder to test for confirmation of this
cause of the extended and gradual hypothesized
r, if there are such prior market reactions,
f thirteen weeks surrounding each of the three
.ould provide good opportunities to detect the
the two possible reasons explored are not
ually exclusive. Nor are they exhaustive,
rpretations are further supported, however, by
coefficient analysis and by the additional tests
contribution of the present value of current
additions and the timing of significant market
ence suggests that the present value of current
additions seems to have information content and
ant market response occurred when the initial
leased and during the period when adoption of
s announced. Based on the results, reserve
d for working capital, long-term liabilities,
tock, have been found more important than the
easures without adjustments.
Based on
presented abov
conclusions.
1. The r
not c
data '
the i
price
aroun
propo
2. The s
ad j us
conte
reser
inves
Howev
reser
pert!
Because o
the Energy Pol
accounting has
financial repo
controversy ce
disclosing the
83
:he empirical findings and on the interpretations
it seems reasonable to draw the following
2serve value data required by the SEC did
unvey new information to investors when the
^ere actually disclosed. It appears that
nformation was already impounded in security
s prior to the actual disclosure, possibly
3 the time when the SEC issued the initial
sal and the time when it adopted the proposal,
tatic measure of reserve value, without
fcment, does not appear to have information
at in pricing securities. The adjusted
ve value is a more effective signal to
tors than is unadjusted reserve value.
er, the present value of current period
ve additions appears to possess information
nent to the assessment of security prices.
Summary of the Research
The Problem
f the oil crisis and the resulting passage of
icy and Conservation Act of 1975, oil and gas
been one of the most controversial issues in
rting during the past five years. The basic
nters on the methods of determing the cost and
value of oil and gas reserves.
of exploration
capitalize all
and developmen
efforts method
Statement No.
84
The historical cost of reserves is the money actually
spent by a company to discover and develop its reserves.
There is no question that the cost of drilling successful
wells should be included in the cost of reserves. However,
questions are raised whether the cost of reserves should also
include the co^t of drilling unsuccessful wells and the cost
activities. Supporters of full costing would
(both productive and unproductive) exploration
t costs when incurred because they represent
what business l(iad to pay, in fact, to get the reserves.
Supporters of jshe opposing approach, the successful-efforts
method, would tapitalize only those exploratory costs leading
directly to discoveries of mineral reserves and to their
development, arguing that the unproductive expenditures do
not result in assets and therefore should be charged to expense.
In December, 1977, the FASB issued a ruling on this long-
standing controversy in its Statement of Financial Accounting
Standard No. 19_, "Financial Accounting and Reporting by Oil
and Gas Producing Companies." It adopted the successful-
and prohibited the full-cost method.
9_, however, was met by strong opposition from
ull costing and from various governmental
of Energy, the
supporters of
agencies, including the Department of Justice, the Department
Federal Trade Commission, and the Federal
Power Commission. Consequently, in August, 1978, the SEC
decided to allow companies to use either full costing or
all
successful-eff
reversing the
costing for ex
SEC proposed t
accounting met
under which
as Incurred an
proved oil and
This method of
of all proved
According
to take at lea&
Its eventual
of placing a m
tlon with its
1978 reporting
requires (1) c
year by apply!
ation of price
contractual ar
proved reserve
(based on curr
producing the
value of estim.
of ten percent
attributable
fe
.0
to
85
Drts costing as a transition measure, thus
ASB's pronouncement that proscribed full
ploration costs. As a permanent measure, the
d develop over the next few years a new
lod called "Reserve Recognition Accounting,"
exploration costs are written off immediately
i the estimated value of the additions to
gas reserves is treated as current revenue,
reporting also requires that the present value
<f)il and gas reserves be reported as assets,
to the SEC, the development of RRA is expected
t three to four years, with no assurance about
asibility. At the core of RRA is the process
netary value on proved reserves. In connec-
eserve disclosure requirements adopted for
the SEC prescribed a valuation method that
computing estimated future net revenues for each
jig current prices of oil and gas (with consider-
changes only to the extent provided by
rangements) to estimated future production of
and deducting estimated future expenditures
ent costs) to be incurred in developing and
proved reserves; and (2) computing the present
ated future net revenues using a discount factor
The portion of the resulting present value
proved oil and gas reserves added in years
prior to the c
current year s
In the SE
tional procedu
proved reserve
form basis. T
value approach
the disclosure
and government
of the financi
and (b) to com
results of dif
of many oil co
reserve value
information.
which subjecti
reserves, prod
ditures. Many
resulting from
market value,
will only conf
public at larg
Despite t
reserve value
empirical rese
value disclosu
86
urrent year, and the portion added during the
hould be separately disclosed.
C's view, a valuation based on this computa-
re will be a surrogate for the market value of
s, derived on a relatively objective and uni-
he SEC and many other proponents of the present
to reserve valuation obviously believe that
of reserve value data will enable investors
policy-makers (a) to gain a better understanding
al position and operating results of a company
pare the financial positions and operating
ferent companies. On the other hand, officials
mpanies are unconvinced that disclosure of
would result in more meaningful and useful
Their doubts are based largely on the extent to
vity would be required in estimating proved
uction rates, and timing of development expen-
opponents of RRA also point out that the data
the prescribed method will reflect neither
replacement cost, nor net present value, and
use invesotrs, governmental agencies, and the
6 .
he importance and complexity of the issue of
disclosure, little has been published about
arch undertaken to evaluate the results of
res .
Objective and Foundation of the Study
The objec
1. Do in
matioi
Both ques
tional setting
effects result
costing. Howe
87
ive of this dissertation has been to investi-
gate empirically whether the reserve value data presented by
oil and gas producers have been utilized by market partici-
pants. Two research questions have been formulated in an
attempt to assess the usefulness of the reserve value data.
restors behave as if the reported present
values of proved reserves are effective signals
for pricing securities of oil and gas producers?
Is there a linear dependence between the infor-
i related to reserve values and the capital
marker response?
;ions have been addressed within the institu-
of capital markets (i.e., New York Stock
Exchange, American Stock Exchange, and over-the-counter
market). Capital market behavior is assumed to be efficient
in the semi-strong form.
Prior capital market impact studies in the field of oil
and gas accounting have centered on measuring the information
ng from an announced elimination of full
rer, past research has also reported that the
disclosure or publication of such items as earnings-per-share
forecasts, capitalized lease data, and replacement cost data
appears to have an observable effect on the capital market.
Consequently, the review of the studies by Gonedes' group,
by Ro, by Bowman, and By Eskew and Ro serves to provide a
the research methodology of the present study. foundation for
Two re seai
investigation
firms used in th
fo
companies repoi
cial reports
but before Dec
gas producers v
a total of fift
the second des
single measure
portfolios, co
firms, and rel:
the control gro
design was char
data, no contr
neous test of
The procecfl.
follows. First
per-dollar inv
present value
share—was com'
the firms were
respective magrfi
subsequently d
Thus, four por
88
Research Methodology
'.ch designs were employed to permit extensive
f the two research questions. The sample of
is study is composed of oil and gas producing
'ting reserve value data to the SEC in finan-
r fiscal years ending after December 25, 1978,
<£mber 25, 1979- Reports of seventy-two oil and
lere examined in the first research design, and
y-eight oil and gas producers were included in
gn. The first research design focused on a
of reserve value, involved forming equal risk
retained a control portfolio of randomly selected
ed on a comparison of daily stock returns for
up and the experimental group. The second
°acterized by various measures of reserve value
ol group, multiple test periods, and simulta-
ystematic returns and residual returns.
ures in the first design can be described as
, a particular attribute of reserve value—the
^stment related to the static measure of
er share, or alternatively, the scaled PV per
.jfjuted for each sample oil and gas firm. Next,
ranked in ascending order according to the
itudes of their scaled PVs per share, and were
vided into quartiles based on that ranking.
;folios were formed at the beginning of the
P
o
OS
disclosure (te
order to compa^
stock returns
distribution o
randomly selec^
twenty trading
date were comp
eight portfoli
then adjusted
nique which foi
risk-adjusted i
trading day by
from their res
differences weit5
different from
In the se
by canonical c
Five informati
value data were'
effect of size
were used as a
information vaif
1. Th e s
of pre
commor
o
89
it) period, each representing a condition. In
e the conditional distribution function of the
f these four portfolios with the unconditional
stock returns, four control portfolios of
ed firms were formed. Next, daily returns for
days surrounding the reserve value disclosing
i|ited for the shares of each firm in each of the
The unadjusted daily portfolio returns were
'or the difference in security risk by a tech-
ces portfolio risks to equal unity. Next,
'eturn differences were computed for each
subtracting the control portfolio returns
Elective test portfolio returns. These return
e tested to see if they were statistically
zero.
oond design, an association test was performed
orrelation analysis on two sets of variables,
in variables based on the reported reserve
specified. To remove the possible confounding
both common stock prices and total revenues
denominator to the variables. The five
iables were
tlatic measure of the present value (alone)
ved reserves per share for each dollar of
share market price;
90
2. The static measure of the present value of
proved reserves per dollar of revenues;
3. The adjusted value (present value of proved
reserves plus net working capital, minus long-
term liabilities and preferred stock) per share
per dollar of investment
4. The fjLow measure of reserve value (the present
value of current-year reserve additions) per
share for each dollar of investment;
5. The flow measure of reserve value per dollar
of revenues.
This set of variables was associated with a set of stock
return variables represented by systematic returns and
is. The test periods included not only the
actual disclosure period but also three selected prior
periods related to three critical SEC actions leading to the
actual disclosure. Weekly stock prices were used, and each
thirteen weeks centered on the critical-event
sets of variables were then tested for the
linear dependence between them.
residual retun
period covered
week. The two
existence of a
The stati
that the eviderli
thesis that the
trol groups are:
Findings and Conclusion
tical test conducted in the first design shows
ce was insufficient to reject the null hypo-
portfolio returns of the disclosure and con-
equal during the twenty trading-days period.
ch
a
However, the
analysis of thi
linear relation,
variables and
tional analysi
related to the
and variables
event (the ini
respectively)
the hypothesis
adjusted reser\
reserve value i
variables.
Collective'
1. The re
rules did not h
oil and gas pre
It appears that
security prices
around the time
time it issued
2. Invest
unadjusted valu
pricing securit
the present val
adjusted reserv
does the static
i-square statistic in the canonical correlation
second design indicates that there is a
ship between the set of five information
t he set of eight market-based variables. Addi-
and tests suggest that the variables
present value of current-year reserve additions
ssociated with the first and third critical
ttial proposal and the adoption of final rules,
appear to be responsible for the rejection of
that no relationship exists. Also, the
e value is more important than the unadjusted
n explaining the variability in the market-based
91
ly, the following conclusion is reached.
serve value disclosure under the SEC-prescribed
ave a measurable impact on the stock prices of
ducers when the data were actually disclosed,
the information has already been impounded in
prior to the actual disclosure, probably
the SEC announced the initial proposal and the
the final rules.
ors did not behave as if the static measures of
es of proved reserves are effective signals in
ies. The effective signals are represented by
ue of current-year reserve additions. However,
e value seems to contain more information than
measure of unadjusted value.
es
3. Evidep
current-year r
decision to retjj
to have some m
data is thus s
experimentatio
ce of the usefulness of the present value of
erve additions suggests that the SEC's
uire disclosure of reserve value data appears
erits. The continued presentation of such
upported, and the SEC should persist in its
of developing RRA. n
Recommendation for Future Research
o
The concl
because of the
Chapter I, esp^
firms whose st
limitations op^
disclosure. A^
oil and gas pr
RRA should be
oil and gas pre
determine if thi
to the market
OTC market alsc
sent study. Th
those traded or
whether a diffe
groups. In adc
test of the pot
operating chara.
us
92
ions drawn in this study must be qualified
limitations and assumptions set forth in
cially the lack of data and the inclusion of
cks are traded in the OTC market. These
n areas for future research on reserve value
noted previously, a supplemental summary of
•oducing activities prepared on the basis of
disclosed in the 1979 annual filings of the
ducers. Obviously, it will be interesting to
s RRA-based summary conveys any new information
Limited evidence of market efficiency in the
> suggests a possible modification of the pre-
.at is, by separating OTC-traded firms from
. the national exchanges, one can examine
rential market impact exists between these two
.ition, future research may consider a direct
ential differential impact of size and
cteristics of companies. Also, when the
operating char
alternative to
to exclude fin
from the study
93
acteristics of the firm are considered, an
the direct test of the differential impact is
•rns with substantial activities in refining work
Com pany Name
01 .m
Sy
American Pacif American Petro American Quasa American Resou: Argo Petroleum Argonaut Energy-Beard Oil Co. Belco Petroleu: Burton-Hawks Buttes Gas & C & K Petroleu: Canadian Homes Canadian Merri Canadian Super Cities Service Columbia Gas Crown Central Crystal Oil Diamond Shamro Dome Petroleum Dorchaster Gas Dyco Petroleum El Paso Co. Energy Mineral Energy Reserve Ensearch Corp. Entex Inc. Equitable Gas Equity Oil Co. Pelmont Oil Co Forest Oil Cork Florida Gas General Americ General Explor Getty Oil Co. Gulf Energy & Hamilton Broth
ic International Inc, fina Petroleum Company
rces Management Corp. Corp.
m Corp.
1 Co. Inc.
ead Oils Ltd. 11 Ltd. ior Oil Ltd.
Exp
94
APPENDIX A
LIST OF SAMPLE COMPANIES
Trading Design Design
stem Petroleum
:k Ltd. Corp. Corp. Ltd.
3 Corp. 3 Group
;o.
rp
loration Co. kn Oil-Texas ation Co.
Development Corp. rs Petroleum Corp,
Market I II
OTC DI DII ASE DI DII OTC DI DII OTC DI OTC DI DII OTC DI DII OTC DI DII NYSE DI DII OTC DI DII NYSE DI DII ASE DI DII ASE DI ASE DI DII ASE DI DII NYSE DI DII NYSE DI ASE DI DII ASE DI DII NYSE DI DII ASE DI ASE DI DII OTC DI DII NYSE DI DII OTC DI DII OTC DI NYSE DI DII NYSE DI NYSE DI DII OTC DI ASE DI DII OTC DI DII NYSE DI DII NYSE DI DII ASE DI NYSE DI DII OTC DI OTC DI
Com pan.y Name
Houston Natural Husky Oil Inexco Oil Co. Kerr McGee Louisiana Land Marathon Oil May Petroleum McCulloch Oil McParland Ener McMoRan Oil & Mesa Petroleum MGF Oil Corp. Mitchell Energy Mountain Fuel Murphy Oil Corjs Noble Affiliat North American Natomas Co. Ocean Drilling Patrick Petrol^ Pennzoil Co. Pennzoil Louis Petro-Lewis Co Ranger Oil Ltd Reserve Oil Sabine Corp. Southern Natur Southland Roya Tesoro Petroleii: Texas American Texas Eastern Texas Gas Tran£ Texas Internat Texas Oil & Gai Total Petroleum Trans continent
Gas Corp,
Exploration
orp. gy Corp. 3as Co.
& Development iupply 1 . s Inc. Royalties
95
APPENDIX A—Continued
Trading Design Design
Exploration
Texas Offshore
um
iana P-
(feas
al Resources Inc. Ity Co. m Oil <po. mission ional
North America Ltd, Oil Corp.
Market I II
NYSE DI DII ASE DI DII NYSE DI NYSE DI DII NYSE DI DII NYSE DII OTC DI DII ASE DI DII OTC DI DII NYSE DI DII NYSE DI DII NYSE DI DII ASE DI DII NYSE DI DII NYSE DI DII OTC DI DII ASE DI DII NYSE DI DII OTC DI DII NYSE DI DII NYSE DI DII OTC DI ASE DI DII ASE DI DII NYSE DI NYSE DI NYSE DI DII NYSE DI DII NYSE DI DII OTC DI NYSE DI DII NYSE DI DII NYSE DI DII NYSE DI DII ASE DI DII ASE DI DII
96
APPENDIX B
TABLE VII
RISK-ADJU 3TED PORTFOLIO RETURNS FOR TEST PORTFOLIOS
Trading Risk-Adjusted Portfolio Returns
Days 1 2 3 4
-9 0 .0014 0 . 0185 0.0097 -0.0073
- 8 0.0056 0.0020 -0.0035 -0 .0124
-7 0.0030 -0 .0023 -0 .0043 -0 .0049
-6 0 .0091 0 .0047 -0 .0000 0.0085
-5 0.0022 0.0104 -0.0035 0 .0097
-4 -0 .0040 -0 .0006 0 . 0058 0.0209
-3 o . 0238 -0.0085 0 .0068 0 . 0068
_2 0.0060 0.0034 0 . 0 0 7 3 0 . 0127
-1 0.0210 0 .0016 0 .0030 0.0075
0 -0 .0062 -0.0081 0.0115 -0.0187
1 0 . 0034 -0 . 0091 0.0162 0 . 0 2 5 1
2 0 .0041 0 .0004 0 .0060 0.0131
3 0 .0129 0.0093 0 . 0 0 6 1 -0 .0145
4 0.0088 0 .0089 -0 .0037 -0 .0107
5 -0 .0005 0 .0042 0 . 0119 0.0006
6 0 . 0 0 5 1 -0.0004 0 . 0070 0 . 0 0 5 1
7 -0 .0031 -0 .0039 -0 .0025 0 .0166
8 0 .0049 -0 .0072 -0.0123 -0 .0028
9 0 . 0003 -0.0028 -0 .0095 -0.0045
10 0 .0106 -0 .0079 0 .0059 -0.0046
97
APPENDIX B—Continued
TABLE VIII
RISK-ADJUST ED PORTFOLIO RETURNS FOR CONTROL PORTFOLIOS
Trading Risk-Adjusted Portfolio Returns
Days 1 2 3 4
-9 -0.0021 0 . 0 0 0 6 -0.0033 0 . 0 0 1 9
-8 0.0019 0.0155 -0.0058 0.0047
-7 0.0044 0 .0045 0 .0004 0 . 0 0 7 0
-6 0.0050 -0.0022 0 . 0 0 0 6 0 .0041
-5 -0.0087 0 .0045 0 . 0 0 2 6 0.0005
-4 -0 .0055 -0.0036 -0 .0021 0.0004
-3 -0.0027 0.0037 0 . 0 0 1 9 0 . 0 0 7 8
-2 0 .0005 0.0020 -0.0025 0 .0062
- 1 0 .0009 -0 . 0 0 2 3 -0.0035 -0.0001
0 0 .0004 0 . 0 0 9 0 -0 . 0025 -0 . 0016
1 0.0033 0 .0026 0 .0004 -0.0153
2 0.0028 0 .0037 -0 .0001 0 .0006
3 0 .0045 0.0017 0 .0002 0 .0024
4 0 .0068 -0.0002 0 . 0 0 3 3 0 . 0 0 5 8
5 o .0063 -0.0021 0.0015 -0.0055
6 0 . 0 0 3 8 0.0024 0 . 0 0 0 9 0 . 0 0 0 8
7 0.0023 0 . 0 0 8 0 0 . 0 0 1 8 -0 .0010
8 0 . 0 0 5 1 -0 . 0076 0.0035 -0 .0064
9 -0 .0000 0 . 0 0 0 5 0 .0002 0 .0020
10 -0 . 0 0 6 3 -0 .0042 -0 .0002 0 .0047
MEAN
98
APPENDIX C
TABLE IX
AND STANDARD DEVIATION FOR VARIABLES IN RESEARCH DESIGN II
Variable Mean Standard Deviation
1.3837 0.9392
J2 3.6233 7.9513
X3 0.5949 1.1195
X4 0 .2237 0.2294
0.4634 0.7738
B1 1.2919 0.7952
B2 0.5712 0.6864
B3 1.1886 1.0400
B4 1.7828 1.2229
ASR1 0.0018 0.0021
ASR2 0.0019 0.0017
ASR3 0.0023 0.0019
ASR4 0.0025 0.0023
Cohen, Jacob, Correlati Hillsdale 19 75.
Dyckman, Thoma Capital M Englewood
Fama, Eugene F Books, 19
and Patricia Cohen, Applied Multiple Regression/ on Analysis for the Behavioral Sciences, , NJ, Lawrence Erlbaum Associate, Publishers,
, David Downes, and Robert Magee, Efficient arkets and Accounting: A Critical Analysis, Cliffs, NJ, Prentice Hall, Inc., 1975.
., Foundations of Finance, New York, Basic 76.
Franci Jack ., Investment Analysis and Management, New York, McGraw-Hill Co., 1972.
Harris, Richar York, Aca
Nie, Norman H. brenner, Social Sc 1975.
Siegel, Sidney Sciences,
Adkerson, Rich Work?," J 72-81.
Basu, S., "The Associati Security 599-625.
Beaver, Willia: Implicati ment to T
BIBLIOGRAPHY
Books
3 J., A Primer of Multivariate Statistics, New iemic Press, Inc., 1975.
, C. Hadlai Hull, Jean G. Jenkins, Karin Stein-and Dale H. Bent, Statistical Package for the lences, 2nd ed., New York, McGraw-Hill, Inc.,
, Nonparametric Statistics for the Behavioral New York, McGraw-Hill Book Company" 1956.
Articles
ard C., "Can Reserve Recognition Accounting ournal of Accountancy, 148 (September, 1979),
Effect of Earning Yield on Assessment of the on between Annual Accounting Income Number and Prices," The Accounting Review, 53 (July, 1978),
.:n , "The Behavior of Security Prices and It's on for Accounting Research (Methods)," Supple-le Accounting Review, 47 (1972), 407-432.
99
Beaver, Williaib, "What Should be the PASB's Objectives?," Journal o_f Accountancy, 136 (August, 1973), 49-56.
in Beaver, Wlllla: tion betwi Risk Measii 654-682.
Benston, G., " Evaluatloi American
Required Disclosure and the Stock Market: An of the Securities Exchange Act of 19 34,"
Economic Review, 63 (March, 1973), 132-155.
Bowman, Robert Empirical (April, 1
G., "The Debt Equivalence of Leases: An Investigation," The Accounting Review, 55
980), 237-253.
Collins, Daniel W., "SEC Production-Line Reporting and Market Efficienc 1975), 125-
r," journal of Financial Economics, 2 (June, • 164.
Collins, Danie "Market E . ing in the Journal, ®
Conners, Joseph Untried Me 54-63.
Fiction?, 92-99.
Cooper, Kerry, C. Groth, Disclosure (Septembe
Deakin, Edward major Oil Methods," 722-734.
Eskew, Robert t Extractive 48 (April
Fama, Eugene F and Empir 383-417.
100
_ T. P. Kettler, and M. Scholes, "The Associa-en Market Determined and Accounting Determined re," The Accounting Review, 45 (October, 1970),
W., Warrent T. Dent, and Melvin C. O'Connor, fects of the Elimination of Full Cost Account-Oil and Gas Industry," Financial Analysts
£4 (November-December, 1978), 48-57.
E., "Discovery Value—The Oil Industry's thod," Journal of Accountancy, 139 (May, 1975),
"Reserve Recognition Accounting: Fact or Journal of Accountancy, 148 (September, 1979),
Steven M. Flory, Steven D. Grossman, and John "Reserve Recognition Accounting: A Proposed Framework," Journal of Accountancy, 148
•, 1979), 82-91.
B., "An Analysis of Differences between Non-Firms Using Successful Efforts and Full Cost The Accounting Review, 54 (October, 1979),
, "An Examination of Share Prices Data in the Petroleum Industry," The Accounting Review, 1973), 314-324.
, "Efficient Capital Market: A Review of Theory cal Work," Journal of Finance, 25 (May, 1970),
Gonede; N. J. Accountin of Accoun
101
"Risk, Information and the Effects of Special y Items on Capital Market Equilibrium," Journal ing Research, 13 (Autumn, 1975), 220-256.
Gonedes, N. J. rium, Inf' Technique Empirical tives: 19 12 (197I0
Gonedes, N. J. "Disclosu Market Eq Rules," 89-137.
and Nicholas Dopuch, "Capital Market Equilib-Drmation Production, and Selecting Accounting : Theoretical Framework and Review of Work," Studies on Financial Accounting Objec-74, Supplement to Journal of Accounting Research, 48-129.
Jo
Nicholas Dopuch, and Stephen H. Penman, e Rules, Information-Production, and Capital ailibrium: The Case of Forecast Disclosure urnal of Accounting Research, 14 (Spring, 1976),
Hakansson, Nils H., "Where We Are in Accounting: A Review of •Statement on Accounting Theory and Theory Acceptance,'" The Accounting Review, 53 (July, 1978), 717-725.
Hamada, R. S., the Syste: 27 (May,
Harrison, Tom, and Nondi Accounti ing
Ijiri, Yuji, " Reporting
Kaplan, Robert Accounting paper pre The Impact closure, Keller, Du
Lev, Baruch, " Market: Th Reviewj 5f
Patz, Dennis H in an Eff Research,
Ro, Byung, "The; Stock Pric 1978), 315
"The Effect of the Firm's Capital Structure on i|natic Risk of Common Stock," Journal of Finance, 1972), 435-452.
"Different Market Reactions to Discretionary cretionary Accounting Changes," Journal of Research, 15 (Spring, 1977), 84-107.
(pil and Gas Accounting—Turbulence in Financial " Financial Executive, 47 (August, 1979), 18-26,
., "The Information Content of Financial Numbers: A Survey of Empirical Evidence," ented at the Duke Symposium and published in of Accounting Research on Practice and Dis-
dited by A. Rashad Abdel-khalik and Thomas F. rham, N.C., Duke University Press, 1978.
'he Impact of Accounting Regulation on the Stock e Case of Oil and Gas Companies," The Accounting (July, 1979), 485-503.
and James R. Boatsman, "Accounting Formulation dent Market Environment," Journal of Accounting 10 (Autumn, 1972), 392-405.
Disclosure of Capitalized Lease Information and es," Journal of Accounting Research, 16 (Autumn, -340.
Sharpe, Willia: Equilibri Finance ,
Sharpe, Willia: New York Financial
102
In F., "Capital Asset Prices: A Theory of Market im under Conditions of Risk," Journal of 33 (September, 1964), 425-442.
.m F. and Guy M. Cooper, "Risk-Return Classes of jStock Exchange Common Stocks, 1931-19673" Analysts Journal, 28 (March-April, 1972), 46-54
Dyckman, Thomap on the Re Stamford,
19, Finan Producing Securitie
ISL Daily Stoc dard
ISL Dally Stock dard & Poo
Reports
Report on the Effects of the Exposure Draft ;urns of Oil and Gas Companies1 Securities, Conn., FASB, October, 1977.
Report 3, on the Effects of Statement No 3ial Accounting and Reporting by Oil and Gas Companies, on Returns of Oil and Gas Company Stamford, Conn., FASB, March, 1978.
£ Price Record, New York Stock Exchange, Stan-Poor's Corporation, New York, 1977-1979.
Price Record, American Stock Exchange, Stan-d's Corporation, New York, 1977-1979.
ISL Daily Stock Price Record, over-the-counter market, Stan-dard & Poor's Corporation, New York, 1977-1979.
Welsch, Glenn A., and Edward A. Deakin, Measuring and Report-ing the "Replacement" Cost of Oil and Gas Reserves, Washington n.
Zecher, Richarfl Study of Impact," Washingtoi
J., "Full Cost vs. Successfull Efforts: A a Proposed Accounting Change's Competitive SEC Directorate of Economic and Policy Research, i, D.C., SEC, February, 1978.
Materials of Professional Organizations
Field, Robert Industrie American
Financial Acco and Repor Statement Stamford,
D.C American Petroleum Institute, 1977.
,, "Financial Reporting in the Extractive s," Accounting Research Study No. 11, New York, Institute of CPAs^ 1969.
anting Standards Board, "Financial Accounting ^ing by Oil and Gas Producing Companies," of Financial Accounting Standards No. 19, Conn., FASB, December, 1977.
Paton, W. A., rate Acco American
and A. C. Littleton, An Introduction to Corpo->unting Standards, monograph No. 3, Sarasota, Accounting Association, 1940.
Securities and No. 190,
103
Public Documents
Exchange Commission, Accounting Series Release Washington, Government Printing Office, 1976.
'Adoption of Requirements for Financial Accounting ;ing Practices for Oil and Gas Producing
Releases Nos. 33-5966; 34-15108; 35-20688; AS-253, Washington, Government Printing Office,
and Repor Activitie IC-10383; August, 1
, "Disclosure of Oil and Gas Reserves and Operations; Amendments to Regulation S-K," Releases Nos. 33-6008; 3^-15418; 35-20838; IC-10532, Washington, Government Printing Office, December , 1978,
_, "Disclosure of Oil and Gas Reserves and Operations, Amendments to Regulation S-K," Releases Nos. 34-20691; IC-103o5, Washington, Government
Proposed 33-5967; Printing Dffice, August, 1978,
"Oil and Gas Producers Supplemental Disclosures is of Reserve Recognition Accounting," fos. 33-6126; 34-16218; 35-21222; IC-10875;
Government Printing Office, September,
on the Bap Releases AS-269,Washington 1979.
"P oposed Supplemental Earnings Summary for Oil roducing Activities," Releases Nos. 33-5969; 35-20690; IC-10382, Washington, Government
and Gas P 34-15111; Printing Ijffice, August, 1978.
'Requirements for Financial Accounting and Practices for Oil and Gas Producing Activities,"
Nos. 33-6006; 34-15416; 35-20836; IC-10530;
Reporting Releases AS-257, W December,
Eskew, Robert Cost Disc Volumes," Purdue Un
ashington, Government Printing Office, 1978,
Unpublished Materials
X. and Byung T. Ro, "SEC Mandated Replacement Losure, Security Returns and Transactions unpublished paper, Department of Management,
iversity, West Lafayette, Ind., 1978.
Foster, George Market Re of Busine
Etebari, Ahmad Proposed unpublish Administr Texas, 19
10 4
"Accounting Policy Decisions and Capital learch," unpublished paper, Graduate School is, Stanford University, Stanford, Ca., 19 80.
"An Assessment of the Stock Market Effects of Accounting Changes in the Oil and Gas Industry," ed doctoral dissertation, College of Business ation, North Texas State University, Denton, 79.
Roche, Peter B Some Anal]/ 1979, pp.
Newspapers
"Energy—Costly Energy Is Wasting Resources, •sts Worry," The Wall Street Journal, May 3, 1, 30.