Collins 1981

Journal of Accounting and Economics 3 (1981) 37-71.
North-Holland Publishing Company
THE ECONOMIC DETERMINANTS OF THE MARKET REACTION

TO PROPOSED MANDATORY ACCOUNTING CHANGES IN THE
OIL AND GAS INDUSTRY
A Cross-Sectional Analysis*
Daniel W. COLLINS and Michael S. ROZEFF

University of Iowa, Iowa City, IA 52242, USA
Dan S. DHALIWAL
University of Arizona, Tucson, AZ 85721, USA
Received April 1979, final version received September 1980
This paper examines the economic reasons for the observed negative abnormal common stock
performance of firms whose reported earnings and stockholders' equity were negatively affected
by the proposed elimination of full cost accounting in the oil and gas industry. Four
explanations of the market effects of this mandatory accounting change are examined: (1) naive
investor theory, (2) modified naive investor theory, (3) contracting cost theory, and (4)
estimation risk theory. These hypotheses are developed in detail and used to generate variables
for a cross-sectional model which explains observed return behavior. The effect of the accounting
change on total stockholders' equity, the existence of financial contracts denominated in terms of
accounting numbers and, to a lesser extent, firm size are shown to be important explanatory
variables. The importance of these variables is consistent with both the contracting cost and
estimation risk theories.
1. Introduction
Both Collins and Dent (1979) and Lev (1979) find that the shares of oil
and gas producing firms employing the full cost method of accounting for
exploration and production activities suffered significant negative abnormal
market returns subsequent to July 15, 1977. On this date, the Financial
Accounting Standards Board (FASB), in an Exposure Draft of Statement of
Financial Accounting Standards No. 19 (SFAS No. 19), proposed the
elimination of full cost (FC) accounting in favor of successful efforts (SE)
*The authors are grateful to Warren Dent, Tim McGuire, Nick Dopuch, Ross Watts, Jerry
Zimmerman, Robert Holthausen and Richard Leftwich for their detailed comments on earlier
versions of this paper. Useful comments were also received frarn participants in accounting
workshops at the University of Iowa, University of Chicago, University of Oregon and
University of Washington. We are indebted to Ed Deakin for providing us with information on
the existence of management compensations plans which was gathered from various SEC filings.
Finally, we wish to acknowledge Bill Salatka for his capable research assistance on this project.
0165--4101/81/0000--0000/$02.50 © North-Holland Publishing Company

38 D.W. Collins et al., Cross-sectional tests of oil and gas returns
accounting. 1 While we interpret this evidence as dearly indicating that the

FASB action was associated with negative equity returns of the FC firms, not
all share this viewpoint. The SEC Directorate of Economic and Policy
Research (1978) and Dyckman and Smith (1979, pp. 69-70) find some
evidence of a negative impact of the FASB proposal on FC shares, but are
not willing to reject the hypothesis of no difference in behavior of the FC
and SE shares.
Foster (1980) presents a general critique of research which attempts to
assess the capital market effects of accounting policy decisions and, in
particular, focuses on the research which has been conducted to date relative
to the oil and gas issue. He notes that a common weakness of these studies
is the failure to develop and test explicit theories as to why mandatory
accounting changes may be expected to affect investors' assessments of future
security return distributions. These shortcomings, in turn, impinge on the
reliability of the inferences which are drawn from such research.
To correct the perceived deficiencies in prior research, in this paper we
advance four theories to explain why the proposed elimination of FC
accounting lowered the value of the equity interest in those firms most
affected by this mandatory accounting change: (1) naive investor theory, (2)
modified naive investory theory, (3) contracting cost theory, and (4)
estimation risk theory. 2 Variables suggested by these theories are
incorporated into a cross-sectional model which is used to explain differential
share price adjustment among firms whose financial statements were most
affected by the FASB's proposal. The effect of the accounting change on total
stockholders' equity, the existence of financial contracts denominated in
terms of accounting numbers and, to a lesser extent, firm size are found to be
important explanatory variables. The importance of these variables is shown
to be most consistent with the contracting cost and estimation risk theories.
Our results do not support the naive investor or modified naive investor
theories.
The paper is organized as follows. Section 2 outlines the effects of SFAS
No. 19 on financial statements and financial contracts. Section 3 presents
alternative theories of the observed share price declines of FC firms. Section
4 develops specific hypotheses and variables to explain the cross-sectional
variation in the market performance of FC firms. A model developed from
these hypotheses is presented in section 5. Section 6 contains the empirical
results and an interpretation and summary of the findings.
~The Exposure Draft was dated Friday, July 15, 1977 but its contents were first discussed in
the financial press in the following week. Our subsequent empirical tests are designed to take
account of this timing issue. For convenience, we refer throughout the paper to the Exposure
Draft of SFAS No. 19 simply as SFAS No. 19.
2For related work, see Holthausen (1979) and Leftwich (1979).
D.W.. Collins et al., Cross-sectional tests of oil and gas returns 39
2. Effects of SFAS N o . 19 on financial statements and contracts

SFAS No. 19 significantly altered key accounting numbers disseminated by
the accounting system of FC firms. Financial statements, loan agreements,
and management compensation agreements, all of which employ accounting
numbers, were affected. As a preliminary toward understanding the possible
responses of investors and managers to the changes, the effects of SFAS No.
19 on these documents are outlined in this section.
2.1. The effect of SFAS No. 19 on financial statements

Prior to SFAS No. 19, oil and gas companies could choose one of two
generally accepted methods of accounting for the costs of discovering oil and
gas. Under the FC method, costs of unsuccessful exploration and drilling
ventures were capitalized (carried as assets) and written off against the
income generated by producing wells. This method reduced the immediate
earnings impact of drilling dry wells and, by averaging costs over several
periods, had the effect of reducing the variability of net income) By contrast,
the SE method expensed the exploration and drilling costs of unsuccessful
ventures in the year they were incurred.
SFAS No. 19 required that all firms in the oil and gas industry thereafter
account for exploration costs under the SE method. Several SE firms also
had their financial statements affected by SFAS No. 19 since they were not
using the particular version of SE accounting stipulated in the
pronouncement .4
The proposed elimination of FC accounting substantially affected both
reported income and stockholders' equity, primarily due to the fact that
current unproductive drilling efforts were to be charged to income while pre-
existing dry wells were to be written off as a prior period adjustment to
retained earnings. During the FASB deliberations, Touche Ross & Co.
(March 1977) surveyed 36 FC companies and estimated that prospective
reductions in reported earnings and stockholders' equity would average 20
percent and 16 percent, respectively. 5 In February 1978, the First Boston
Corporation estimated average reductions in net income for a sample of FC
firms of 55 percent, 22 percent, 38 percent and 41 percent for the years 1974-
1977, respectively. 6 Subsequent estimates of the impact of SFAS No. 19
disclosed by firms on 1977 10-K reports and compiled by the FASB were
consistent with these external estimates. 7
3See Sunder (1976).
4The Collins and Dent (1979) sample and our sample include several such SE firms. For
brevity, we refer to these finns in this sample as FC finns.
5Letter dated March 29, 1977 from Touche Ross & Co. to the Ad Hoc Committee on Full
Cost Accounting.
6First Boston Corporation statement to Department of Energy, February 21, 1978.
7Financial Accounting Standards Board, 'Additional Comments of the Financial Accounting
Standards Board to the Securities and Exchange Commission: Accounting Practices - - Oil and
Gas Producers', SEC File $7-715, May 31, 1978.
Another effect of SFAS No. 19 was a likely increase in the volatility of

reported earnings, because the change to SE accounting forced firms to write
off all dry hole costs when incurred rather than over a period of years.
Practitioners recognized this effect. For example, at the Department of
Energy Hearings (February 21, 1978), John Chalsty, Managing Director of
Donaldson, Lufkin & Jenrette Securities Corporation, stated:
[Successful] efforts accounting will bring about a significant reduction
in the reported earnings of most full cost exploration oriented
companies and a significant increase in the volatility of such earnings.
2.2. The effect of SFAS No. 19 on lending agreements

Since a bond indenture or a lending agreement is usually written in terms
of generally accepted accounting principles [Fogelson (1978), Leftwich
(1979)], a mandated accounting change effectively alters these contracts'
terms. Bond covenants frequently specify conditions which set an upper
bound on the amount a firm is allowed to pay as dividends or use for share
repurchase. Typically, the 'dividend covenant' defines the maximum amount
which can be used in a given period to pay dividends [Smith and Warner
(1979)].
For those oil and gas firms with debt containing a dividend covenant, the
mandatory change from FC to SE accounting had several effects. First, the
substantial reduction in the firm's current and future accounting earnings,
ihad the immediate effect (before potential manager responses could occur) of
reducing the maximum dividend payable in current and future years. Second,
since the time pattern of future earnings was altered, with lower earnings
growth more likely for firms with growing nominal exploration expenditures
[Sunder (1976)], investors and managers could anticipate a lower rate of
growth of the dividend maximum. Together, these reductions in the
maximum dividend increased the probability that the constraint would be
binding, that is, firms would be forced to pay lower dividends, now or in the
future, unless offsetting actions were taken.
Other common covenants, which we call 'debt covenants', prohibit the
issuance of additional debt unless the firm meets certain ratio tests such as
between capitalization and funded debt or between tangible net worth and
debt. By decreasing stockholder equity, SFAS No. 19 decreased such ratios.
The immediate effect (prior to managerial responses) was to decrease a firm's
ability to issue debt. Thus, unless mitigating actions were taken, stockholders
faced the prospects of losing tax advantage of future debt issues, using
alternative financing methods or altering the firm's investment program.
2.3. The effect of SFAS No. 19 on management compensation agreements

According to the Jensen-Meckling (1976) theory of agency costs,
management incentive/compensation plans may be viewed as bonding or
D.W. Collins et al., Cross-sectional tests of oil and gas returns 41
monitoring devices designed to restrict management decisions that would

harm the firm's suppliers of capital. Accounting numbers are frequently an
integral component of such management compensation agreements.
Typically, management compensation is related to positive reported
accounting earnings of the firm (managers do not contribute to make up
losses). 8 A mandated accounting change which alters the standard (earnings)
by which management behavior is motivated and rewarded has the
immediate effect, again prior to managerial responses, of altering the basis
upon which management is motivated and rewarded. Without future action
to alter this situation, stockholders could expect altered management
behavior. The types of actions one might expect management to take under
the more restrictive SE method are spelled out in greater detail below.
3. Theories of stock price decline

Theories of stock price decline are divisible into those assuming investor
irrationality (disregard of real cash flows) and those assuming investor
rationality (use of valuation models which value real cash flows). In the latter
category, price declines are brought about by a decline in expected cash
flows and/or an increase in the rate used to discount future cash flows. In
this section we develop alternative theories of price decline, relative to the
proposed elimination of FC accounting, which fall within each of these
categories.
3.1. Naive or irrational investory theory

Rational valuation models tell us that the price of a financial asset is the
discounted value of the future cash flows provided by the asset, where the
discount rate adjusts for the uncertainty attached to the cash flow stream. By
contrast, the traditional naive or irrational investor hypothesis can be stated
as follows: investors value a firm according to the accounting numbers that
the accounting system generates for that firm. This hypothesis, as usually
stated, implies that if earnings per share decline, even if cash flows do not
decline, the asset's price will decline. Alternatively, if stockholders are
assumed to value the firm according to 'book value', then a decline in
stockholder equity will produce a decline in the market value of the firm,
according to this theory.
Since the mandated change from FC to SE accounting produced declines
in both earnings per share and total stockholder equity, the naive investor
theory implies that the shares of the affected firms would decline in price.
Furthermore, it implies that the larger the relative decline in the accounting
number, the larger the relative decline in share price.
8All the firms in our sample which h a d incentive income as a part of the management
compensation plan tied the incentive income to accounting income.
42 D.W. Collins et al.. Cross-sectional tests O['oil and gas returns
3.2. Modified naive investor theory

Collins and Dent (1979) note that management incentive contracts depend
on accounting earnings and that SFAS No. 19 reduced reported accounting
earnings. Suppose that shareholders in the aggregate value securities
rationally. Nevertheless, some shareholders may value securities irrationally
(according to accounting numbers). These shareholders, upset over lower and
more volatile earnings or lower book value of the stock, may initiate legal
action against the management. To avoid this outcome or lower the
probability of its occurrence, management may respond to reduced current
accounting income by cutting back on current exploration expenditures or
positive net present value projects, thereby lowering the value of the equity.
Alternatively, Benston (1978) and Collins and Dent (1979) suggest that
some potential employers of managers may value management's services
according to its record in producing accounting income rather than real cash
flows. To avoid a downward shift in demand for its services, management
may respond by sacrificing profitable projects in order to maintain current
accounting income. This action also would lower the value of the equity.
This theory does not require that the asset pricing mechanism depend on a
predominance of investors who fixate on accounting numbers, merely that
management takes rational action to avoid the consequences of actions by
some irrational investors. Nevertheless, this theory also implies that the
larger the relative decline in the accounting number, the larger the relative
decline in share price, because it is to be expected that management action
will be in proportion to the change in the accounting numbers.
3.3. Contracting cost theory

A firm supplies information (produced by its particular set of accounting
methods) to capital suppliers, governments, employees, auditors, and others,
in order to inform them of the firm's investment/financing opportunities,
contractual arrangements, etc. Accounting information also appears in
important financial contracts and agreements negotiated by the firm. Because
each accounting method that might be chosen by management to represent
the status of the firm's investment/financing opportunities or to influence its
contractual arrangements has its own set of costs and benefits, 9 it is
reasonable to assume that a wealth-maximizing management will not be
indifferent to its choice of accounting methods.
The contracting cost theory thus assumes that accounting methods are
chosen as part of a wealth-maximizing process. The firm's choice of
9These costs and benefits should not be narrowly construed. For example, the costs of an
accounting method are more than the direct cost of compiling a set of accounting numbers, but
include effects the numbers have on capital costs, legal costs, costs of providing competition with
undue information, etc.
D.W. ('olli1~.~ et al.. ( ' r o ~ - ~ e r t i o m d tests o[ oil aml gas ret,rlt~ 43
accounting method is viewed as being embedded in the overall choice

problem of maximizing share price subject to investment and financing
opportunity loci. Management is assumed to face an opportunity set of
vectors of investment/financing/accounting method possibilities and to select
an investment/financing cum accounting method mix so as to maximize
shareholder wealth.1 o
In this framework, SFAS No. 19 imposes a binding constraint on the
opportunity set which, as a standard price-theoretic result [Hirshleifer
(1958)], produces a decline in welfare, manifested as a decline in stock
price. 11 The decline occurs because the new optimum
investment/financing/accounting method mix has reduced expected cash
flows. In general, expected cash flows may decline because of increased costs
imposed on affected firms, changes in investment/financing decisions or some
combination thereof. Below, we indicate more specifically how these declines
in expected cash flow may come about through increased information costs
and agency costs, which we refer to collectively as contracting costs.
3.3.1. Information costs

As outlined above, firms are assumed to have selected the FC method as
being the optimum accounting procedure conditional upon the existing and
planned investment/financing choices of the firm. 12 One important
consideration in this optimizing process is the cost of searching for and
acquiring capital since the firm expends resources to explain to investors the
potential returns and risks attached to its securities. We hypothesize that
higher information costs were perceived to result from SFAS No. 19 in order
to explain to current and future investors how the lower, more volatile
earnings and the altered balance sheets under SE accounting affected the
1°Much recent research on positive accounting theories, which attempts to identify the
benefits and costs of different accounting methods, suggests that the choice of accounting
method is not independent of the investment/financing choice. Watts (1977) hypothesizes, for
example, that the provision of depreciation numbers is related to the value of a firm's fixed
assets. He also argues that the provision of financial statements is itself related to such financing
decisions as the number of bonding covenants in debt agreements, the fraction of equity held by
management, and the amount of risky debt held by outsiders. Watts and Zimmerman (1978)
suggest and test the hypothesis that lobbying behavior for accounting standards is conditioned
upon firm size. Hagerman and Zmijewski (1979) test the relationship of accounting choice to a
number of economic variables (size, risk, capital intensity, competition, and management
compensation plan).
11In the two period Fisherian model, the constraint alters the shape and/or location of the
investment opportunity locus such that the investor optimum occurs at a lower level of utility.
Z2Evidence to support the assumed interdependency of investment/financing/accounting
method choices in the oil and gas industry is provided in recent work by Deakin (1978) and
Dhaliwal (1980). Specifically, Deakin presents evidence that the more aggressive exploration
oriented oil and gas producing firms tend to sdect FC accounting. Dhaliwal finds a strong
positive relationship between the amount of debt in a firm's capital structure and the use of FC
accounting.
firm's existing and planned investment/financing choices. 13 Hence the costs

of acquiring capital increased.
Additionallly, institutional investors often (for legal reasons) define a
relevant universe of potential investments by such screening devices as
earnings coverage ratios, debt-to-equity ratios and number of years of
positive earnings [Pogue and Soldofsky (1969),1. Given the large adverse
effects which the shift to SE accounting would have on many of these
standard ratios, it is likely that some current and potential future
institutional investors would cease including securities of these firms within
their feasible sets. As a result, new capital suppliers would have to be sought
out, itself a costly process.
A firm forced to switch to SE accounting can prevent a deterioration in its
financial ratios and avoid incurring higher capital acquisition costs by
altering its investment/financing decisions. It can, for example, reduce
earnings volatility by shifting from high risk exploratory drilling in virgin
territory to production drilling in proven areas. This option is not without
costs of its own since, given the assumption that the firm's policies had been
chosen optimally, a reduction in the expected cash flows of the firm will
result from exercise of a new investment/financing policy. We conclude that
SFAS No. 19 imposed increased information costs, reflected in higher costs
of securing capital and/or a reduction in expected cash flows if firms
attempted to avoid higher information costs by revising investment/financing
policies.
3.3.2. Agency costs

Under the agency cost approach to investigating the valuation
consequences of mandatory accounting changes [Leftwich (1979)-1, any
change which restricts the alternatives available to management is
hypothesized to reduce expected future cash flows to stockholders because of
increased bonding and monitoring costs I-Jensen and Meckling (1976),1 and
via a reduction in expected wealth transfers from bondholders to
stockholders. We next consider the process by which rational investors
13Essentially, this point of view was expressed by numerous investment banking firms who
testified at the DOE and SEC hearings that the shift to SE accounting would make it more
difficult for them to market new securities of independent oil and gas producers. The Federal
Trade Commission (1978, p. 25) assessed the implications of these claims in the following terms:
Due to the short-term nature of a new issues underwriting, underwriters have a felt need to
be able to explain quickly the nature of their offering. Successful efforts, however, will
generally increase perceived information costs, thus discouraging underwritings and raising
the cost of capital. Recognizing the time and expense entailed in having to explain the
intricacies of accounting practices and their effect upon the 'paper' as opposed to 'real
economic' performance of the company issuing the securities, underwriters will tend to
demand a higher fee and to set a lower public offering price.
would be led to reduce the value of equity claims in FC oil and gas
producing firms as a consequence of these factors.
Within the agency cost framework, covenants in loan agreements are
viewed as the outcome of an optimizing process designed to minimize agency
costs by controlling the conflict between stockholders and lenders over the
firm's investment/financing program. Similarly, management compensation
agreements are viewed as devices to overcome the conflict between
management and stockholders, and thus minimize agency costs.
A forced change to SE accounting upsets these optimizing processes. For
example, one consequence is that it brings firms closer to violation of
covenants in loan agreements. Thus, some states of the world that would not
produce default under the covenants stated in terms of FC accounting will
now produce default under SE accounting. It follows that expected cash
flows produced by investments are lower under SE accounting by the
amount of expected increased bankruptcy/liquidation costs. These higher
costs are an element included in higher agency costs.
Shareholders have an incentive to avoid the higher bankruptcy costs by
seeking to recontract. This option, of course, is also costly. Thus, the stock
value should decline by the minimum of renegotiation cost, redemption cost,
or expected bankruptcy/liquidation cost.14
A related aspect of agency costs which occurs when the loan covenants are
altered by SE accounting is, as noted previously, a loss in expected wealth
transfers from bondholders to stockholders. Under FC accounting, the
negotiated covenants implicitly granted to management some latitude to
select an investment/financing strategy which would produce wealth transfers
from bondholders to stockholders. As a result, the bonds and stocks of the
firm were priced to reflect the expected extent of wealth transfers. To the
extent that SFAS No. 19 leads management to alter the investment/financing
strategy, a non-optimal degree of wealth transfers results and share prices are
predicted to fall as a consequence.
We have suggested a n u m b e r of costs which result from imposing a
constraint on the management's choice of accounting method. For simplicity
we refer to the pricing implications of these various costs as the 'contracting
cost' theory of price decline.
3.4. Estimation risk theory

Except for the naive investor theory, the explanations of share price
decline given above all rely upon a reduction in the expected cash flows
available to shareholders. We suggest next the hypothesis that SFAS No. 19
14The effect of SFAS No. 19 on management compensation agreements according to the
agency cost approach is to create non-optimal incentive effects. Consequently, stock value should
fall by the lesser of the decline in shareholder wealth produced by the resulting non-optimal
decisions or the cost of renegotiating the compensation agreement. See also section 3.4.2.
JAE--C
increased the 'estimation risk' and thus the assessed variance of future cash
flows of affected firms. In a world of limited diversification, this increase will
be translated via Levy's (1978) general equilibrium pricing model into lower
price.
3.4.1. Explanation of estimation risk

Estimation risk is a term suggested by Klein and Bawa (1976, 1977) to
describe the portfolio selection problem when probability distributions of
security returns have unknown parameters. In the traditional Sharpe--Lintner
(1964, 1965) capital asset pricing model (CAPM), uncertainty of a firm's cash
flows is modeled by a probability distribution over future cash flows. The
traditional CAPM model assumes that investors know the parameters of
these distributions although in reality they must be estimated.
In contrast, Klein and Bawa analyze the case in which investors have prior
distributions over the uncertain means and variances of the return
distributions. They show that if investors possess 'minimal' information on a
subset of available assets, they may rationally neglect these securities and
instead form portfolios only from the subset of securities upon which they
have 'sufficient' information, is Klein and Bawa therefore provide conditions
under which limited diversification would be rational. In other words, they
explain why investors would not fully diversify across all securities in the
market as suggested by the CAPM.
Levy (1978) has developed the general equilibrium pricing implications of
limited diversification. He assumes that 'as a result of transactions costs,
indivisibility of investment, or even the cost of keeping track o f the new
financial development of all securities', investors select a subset of securities
to invest in. He then shows that a stock's individual or residual variance
takes on increased importance relative to beta in the equilibrium pricing
equation. A similar results has been derived by Williams (1977, p. 234) in a
general equilibrium context by assuming heterogeneous beliefs among
investors: 'With heterogeneous beliefs beta fails as a complete measure of
risk. For all finite time, both beta and diversifiable or residual risk affect the
market's assessment of required average returns on securities.'
3.4.2. SFAS No. 19 and estimation risk

Estimation risk will increase if investors become less certain of the firm's
cash flows because of increased uncertainty about, for example, the
~S'Sufficient' information means that the number of sample observations, T, used in forming
estimates of each security's mean and variance is large relative to the number of securities, m, in
the subset, or T>m~ 'Minimal' information means T < m . Suppose investors are allocating
funds between two sets of risky securities, the first of which has sufficient information and the
second of which has minimal information. Klein and Bawa obtain the intuitive result that the
optimal portfolio asymptotically limits diversification among the first set of securities only.
investments available to the firm and/or which investments will be selected

by the management. Estimation risk will also increase if investors become
less certain of the future cash flows through changes in the firm's financing
policy (which impact upon the cash flows available to shareholders).
We hypothesize that, if firms possessed loan covenants and/or management
compensation agreements denominated in terms of accounting numbers,
SFAS No. 19 increased estimation risk. Consider first the effect on bond
covenants of a forced change to SE accounting. When the dividend
constraint is altered, a management response is likely, since the changed
position is unlikely to be optimal. The potential actions include (1) increasing
stock financing, to counteract the decline in the dividend maximum; (2)
investing excess funds that normally would be paid out as dividends; (3)
renegotiating the debt agreement; (4) redeeming or repurchasing the debt; (5)
doing nothing. Presumably, the optimizing firm will select a mix of actions
which minimizes costs since all of the above actions involve costs of one sort
or another.
To investors, the important result is likely to be the increased uncertainty
not only as to the immediate management response b u t also as to future
management actions conditional on any new arrangements. Estimation risk
rises insofar as the investment/financing mix of the firm becomes-less certain.
Estimation risk could also be increased by the existence of debt covenants
which prohibit the issuance of additional debt unless the firm meets certain
ratio tests such as between capitalization and funded debt or between
tangible net worth and debt. By decreasing stockholder equity, SFAS No. 19
decreased such ratios and altered the firm's ability to issue debt. This, of
course, could lead management to substitute equity for debt financing or,
alternatively, narrow the investments it selects. Again, the main point is
simply that investor estimation risk will increase because the
investment/financing mix becomes less certain.
Secondly, estimation risk is likely to increase as a result of the effect of
SFAS No. 19 on management compensation agreements. In the Jensen-
Meckling (1976) agency cost theory, management incentive plans are chosen
as part of a process designed to minimize cost and maximize the value of the
firm. Since incentive plans frequently are written in terms of accounting
earnings, the mandated change to SE accounting temporarily places the firm
in a non-optimal position. To achieve a new optimum, the results may be a
change in management behavior and/or adoption of a new incentive plan.
Either of these actions involve uncertain consequences on management
behavior which increase investor uncertainty of the parameters of the
distribution of future cash flows, i.e., increase estimation risk.
To see this, suppose that management compensation is related to positive
reported earnings of the firm. Assume managers are risk-averse and choose a
preferred combination of expected income and standard deviation of income
for themselves via their adoption of a particular set of investment/financing

opportunities open to the firm, subject to any constraints imposed by the
management compensation agreement. The mandated change to SE
accounting has the initial effect of lowering management utility since
expected income is no higher but the variance of accounting earnings has
increased. Shareholders will rationally expect management to alter the mix of
production activities to achieve a new optimum for itself, but it is unclear a
priori how management will respond without detailed knowledge of the
potential investment opportunity set, the management utility contours, and
the costs of making adjustments, all of which are uncertain. This lack of
knowledge of subsequent management behavior means that, as a result of the
mandated switch to SE accounting, stockholders become less certain about
the set of projects being valued in the capital market and less certain of the
firm's distribution of cash flows.
To sum up, a mandatory accounting change which restricts the extant
investment/financing/accounting method choice set places selected firms in a
new situation about which investors have limited information. In a world of
limited diversification as analyzed by Levy, greater uncertainty about future
investment/financing choices raises the variance of the-stock return
distribution and increases the expected rate of return on the firm's stock,
thus causing a price decline in the firm's shares. We have presented several
explanations of how SFAS No. 19 may have increased the uncertainties with
respect to management actions and thus may have caused a price decline in
the shares of affected firms.
4. Statement of hypotheses
This section states specific testable hypotheses based on the various
explanations of price decline outlined above. Seven variables are suggested to
explain the cross-sectional variation in price:
(i) percentage change in total stockholder equity caused by SFAS No. 19,
(ii) possession of a loan agreement defined in terms of accounting numbers,
or
(iii) management compensation agreement defined in terms of accounting
numbers,
(iv) debt/equity ratio,
(v) proportion of revenues devoted to exploration and production,
(vi) possession of public debt versus private debt,
(vii) size of the firm.
4.1. Percentage change in total stockholder equity

Our first hypothesis is as follows:
Hypothesis 1. The abnormal performance of the common stocks of firms

affected by SFAS No. I9 is positively related to the percentage change in the
.firms' accounting numbers (i.e., the more negative the change in the accounting
number, the more negative the abnormal return).
Several of the explanations of price decline imply a testable implication

such as Hypothesis 1. First, the naive investor theories state that accounting
number changes cause stock price changes. It follows that larger stock price
changes should accompany larger changes in accounting numbers.
Hypothesis 1 expresses this implication. Second, the larger the change in the
accounting numbers, the greater the likelihood that contracts denominated in
terms of accounting numbers will be affected and lead to altered
management behavior. This implies that greater contracting costs and
estimation risk will result. Thus all the price decline theories can be
operationalized by Hypothesis 1.
To measure the financial statement effects of the accounting change, we
use the percentage change in the total stockholder equity account and the
percentage change in earnings per share. It is most consistent with the naive
investor hypothesis if both these numbers, particularly earnings per share
change, are significantly related to price change. If the dividend covenant is
important in creating estimation risk and/or contracting costs, then both the
earnings per share variable and the total stockholder equity should be
significant. If the debt covenant is important in these respects, then the total
stockholder equity variable should be significant.
4.2. Existence of a loan agreement

Both the contracting cost and estimation risk theories suggest that firm
values will decline through increased costs and/or increased uncertainties
about management's investment/financing decisions if firms possess loan
covenants defined in terms of accounting numbers. We therefore test:
Hypothesis 2. Firms with loan covenants defined in terms of reported

accounting numbers experienced greater negative abnormal security returns as
a result of SFAS No. 19 than did thosefirms without such covenants.
4.3. Existence of a management compensation agreement

As with loan agreements, all the explanations of price decline except the
naive investor hypotheses suggest t h a t SFAS No. 19 should negatively
influence the shares of firms possessing management compensation
agreements defined in terms of accounting numbers. This leads to:
50 D.W. Collins et al., Cross-seaional tests of oil and gas returns

affected by SFAS No. 19 is more negative for firms possessing management
compensation plans written in terms of accounting numbers.
In our sample, with three exceptions, every firm with a management

compensation plan tied to reported earnings also had loan covenants defined
in terms of accounting numbers. Thus, in this sample, Hypotheses 2 and 3
cannot be tested separately. Accordingly, in the empirical tests, only one
dummy variable is coded. Firms with loan covenants and/or management
compensation plans defined in terms of reported accounting numbers are
assigned a one, while firms without such contracts written in terms of
accounting numbers are assigned a zero.
4.4. Debt/equity ratio

SFAS No. 19 was shown to have the immediate effect on the dividend
covenant of lowering the maximum dividend distribution. Kalay (1978) finds
that firms with higher debt/equity ratios tend to have lower inventories of
funds available for payment of dividends and be closer to technical violation
of the covenant. A lowering of the maximum dividend payable is thus more
likely to raise the chances of a dividend cut for firms with high debt/equity
ratios. We test the importance of the dividend covenant in creating
contracting costs and increasing estimation risk by testing:

affected by SFAS No. 19 is negatively rela]ed to the debt~equity ratio (i.e., the
higher the debt~equity ratio, the more negative the abnormal performance).
4.5. Proportion of revenues devoted to exploration and production

We have noted that SE accounting raises the variance of reported
earnings. Under the contracting cost hypothesis, recontracting becomes more
likely because violation of loan covenants becomes more likely. Under the
estimation risk hypothesis uncertainty surrounding management action is
increased because the variability in management income rises. To measure
the variance effect, we note that Sunder (1976) has shown that the increase in
variance of accounting earnings resulting from a switch to the SE method is
an increasing function of the proportion of the firm's activities devoted to
exploration and production activities. Therefore, we predict that firms with a
higher proportion of each revenue dollar devoted to exploration and
production activities are affected more adversely by SFAS No. 19 than firms
with a lower proportion of their activities devoted to exploration and
production activities:

affected by SFAS No. 19 is negatively related to the proportion of each
revenue dollar devoted to exploration and production activities (i.e., the higher
the proportion spent on exploration and production activities, the more
negative the abnormal return).
It is clear that firms more heavily involved in exploration and production

should have relatively greater declines in total stockholder equity as a result
of SFAS No. 19. The change in total stockholder equity therefore provides
an alternative measure of the firm's commitment to exploration and
production activities.
4.6. Public debt vs. private debt

Leftwich (1979) and Holthausen (1979) suggest that the costs of
renegotiating or redeeming p u b l i c d e b t are higher than the costs of
renegotiating or redeeming private debt. In the event of a mandatory
accounting change, firms with public debt are more likely to face the
prospect of relatively higher renegotiation costs, leading to a decline in
security prices. The joint hypothesis that contracting costs cause price decline
and the that public/private debt variable measures these costs can be stated
as follows:

affected by SFAS No. 19 is more negative for those firms with public debt in
their capital structures than for those firms without public debt (a negative
relation).
In part, the public/private debt variable measures renegotiation costs

because public debt more frequently contains covenants written in terms of
generally accepted accounting principles. Specifically, of the 20 firms in our
sample with public debt, 18 indicated the presence of debt covenants defined
in terms of reported accounting numbers. However, of the 37 firms with no
public debt, 21 also indicated the presence of such covenants. Thus, in our
sample at least, the public/private debt dichotomy is an imperfect proxy for
the existence of covenants written in terms of generally accepted accounting
principles. Although this no doubt reduces the effectiveness of the
public/private debt variable in measuring the extent to which renegotiation
costs cause share price declines, the public debt variable is included in its
present form due to its major role in other empirical studies. Below, an
alternative test of the renegotiation cost hypothesis is undertaken (see section
6.2.3) which does not rely on the public debt/private debt distinction as a
proxy for the existence of accounting based debt covenants.
4.7. Firm size

One final variable relevant to explaining abnormal returns is firm size.
This section outlines the economic reasons for its inclusion in the model. We
argue that size is a 'comprehensive' variable which proxies for (1) leverage,
(12) public debt, (3) political costs, and (4) factors omitted from the return
generating model. While these proxy effects (discussed below) support the
case for including size in the regression, factors 1, 2 and 4 are predicted to
create a negative association, and factor 3, a positive association, between
abnormal returns and size. We, therefore, cannot predict the sign of the size
variable and we allow the data to reveal the net magnitude of the various
effects.
Hypothesis 4 states that, ceteris paribus, greater leverage is associated with
more negative abnormal returns. There is reason to expect that larger firms
will tend to employ more debt per dollar of equity. Ben-Zion and Shalit
[11975) note that larger firms have lower probabilities of failure, more
diversified investments and economies of scale. These three factors tend to
lower the probability of bankruptcy and make it more likely that large firms
will take on relatively more debt in order to capture the tax advantage of
interest-deductability. Empirically, larger firms tend to have higher
debt/equity ratios. In our sample, the Pearson product moment correlation
between size and leverage is 0.37, which is significant at the 0.05 level.
Therefore, size is a proxy for leverage in our sample. Since size and leverage
are positively correlated, a n d Hypothesis 4 suggests a negative relationship
between leverage and abnormal return, the leverage effect may tend to create
a negative relationship between size and abnormal return, with larger firms
showing more negative abnormal returns.
Secondly, Hypothesis 6 suggests that abnormal returns are more negative
for firms with public debt in their capital structures. We expect firm size to
be related to public debt for two reasons: (1) Smith and Warner (1979) argue
that the riskier the debt, the more likely that it will be privately placed. Since
the debt of smaller firms is likely to be more risky than that of larger firms,
smaller firms are less likely to have public debt in their capital structure than
larger firms. (2) Furthermore, larger firms have lower average costs in issuing
public debt due to the fixed cost component to the flotation cost which falls
per dollar of debt as issue size increases. Thus, we expect smaller firms to
have less public debt. As reported below, the point bi-serial correlation
between firm size (market value of equity plus book value of debt) and the
existence of public debt in the capital structure for our sample was 0.43,
which is significant at well below the 0.05 level. Based on these results, the
public debt effect should tend to produce a negative relation between size
and abnormal performance.
A third reason for including size stems from the finding of Watts and
Zimmerman (1978) that political costs are relatively more important for
larger firms. With SFAS No. 19, the reduction in reported earnings would
result in reduced political costs, and thus be of relatively more benefit to
larger firms. This could result in a positive correlation between size and
abnormal performance.
Fourth, and last, Banz's (1979) recent study of New York Stock Exchange
stocks suggests that even after taking account of beta, smaller firms tend to
earn positive abnormal returns compared to larger firms. Banz argues that
this is consistent with investors requiring a higher rate of return for very
small firms which are relatively unknown and for which information costs
are higher. If Banz's results generalize to our sample, it implies a negative
sign for the size coefficient in the model since relatively lower abnormal
returns will be associated with larger firms, a negative relation. However, we
should point out that there is a much greater size difference in 'large' versus
'small' firms in the Banz study than in the present study. In relative and
absolute terms our firms are clearly small, on the average. 16 The applicability
of Banz's finding to our sample is empirically examined below through tests
outside the 'event' period.
In summary, the effects of size acting as a proxy for leverage, public debt
and factors omitted from the return generating model imply a negative
coefficient on size. The political cost factor implies a positive sign. We
therefore state a non-directional hypothesis:
Hypothesis 7. Firm size is a significant factor in explaining the cross-

sectional variation of securities' abnormal returns associated with the proposed
elimination of FC accounting.
5. Sample selection, data definitions and methodology
5.1. The sample

The initial sample consisted of the 113 firms (72 FC and 41 SE firms)
identified in the Collins and Dent (1979) and Dyckman and Smith (1979)
studies. Three selection criteria reduced the sample size to 57 firms. First, for
reasons discussed in Collins and Dent (1979), the 18 firms domiciled in
Canada were excluded from further consideration. 17 Second, five additional
firms involved in mergers, acquisitions or liquidations during the test period
(defined below) were eliminated. Third, thirty-three additional firms which
did not quantify the financial statement effects of SFAS No. 19 were
excluded. The final sample contained 47 FC and 10 SE firms.
t6The sample firms are also small compared to those studied in Watts and Zimmerman
(1978). Thus, political costs are not likely to be a significant factor in our sample.
17The reader is referred to the Collins and Dent study for further details supporting the
exclusion of Canadian firms.
Table 1 summarizes the estimated effects of SFAS No. 19 on earnings and

stockholders' equity for the 57 firms in the final sample. As expected the
impact of the mandatory change on accounting numbers was much greater
for the FC firms than for the SE firms. The median earnings effect for the FC
firms was - 2 7 ' , 0 while for the SE firms the median earnings effect was
- 1 3 %. The median effect on 1977 total stockholders' equity was - 3 0 % for
the FC firms and approximately - 4 % for the SE firms, is These estimates of
the financial statement effects of SFAS No. 19 are similar to those reported
in the FASB (1978) survey of 294 firms, and suggest that the present sample
of 57 firms is representative of the population of oil and gas producing firms
affected by SFAS No. 19.
The financial statement data necessary to compute exploration
expenditures, total capitalization, debt/equity structure and percentage effects
of SFAS No. 19 were taken from the annual stockholder reports, 10-K's, 10-
Q's or prospectus filings with the SEC for fiscal years 1977 and 1978. The
amount of debt raised publicly vs. privately was obtained from Moody's
Industrial, O T C and Bond Manuals. Information concerning the existence of
a management compensation scheme and/or covenants on debt agreements
tied to reported accounting numbers was obtained from 10-K reports,
registration statements, proxy statements filed with the SEC, and from
questionnaires sent to the sample firms. Standard and Poors Daily Stock
Price Record and CRSP tapes were the primary sources of security price
data.
5.2. Multiple regression model
In order to examine empirically the relationships hypothesized in sections

4.1-4.7 of this paper, a cross-sectional regression of the following form was
estimated:
CARj = a o + a~ TO T C A P i + a 2 TO TSEE i + a 3 E X P L O R i
(1)
+ a4DEMK Tj + asPUBDTj + a 6 C O N T R C S i + ej.
Table 2 defines these variables and table 3 presents summary statistics.
5.3. Dependent variable measures

The dependent variable in eq. (1), C A R i, is the cumulative abnormal
return for the jth firm. In the analysis, CAR's were measured over one and
~aThe relatively large discrepancy between the mean and median earnings effects for the FC
sample is due to three fn'ms (C & K Petroleum, Juniper Petroleum, and Buttes Gas & Oil)
which reported 1977 earnings would decline b y m o r e than 200% if they were forced to switch
from FC to SE accounting. For these three firms, reported income (loss) under FC accounting
was near zero and considerably lower than for prior years.
Table 1
Summary of financial statement effects of SFAS No. 19.
Full cost (47 firms) Successful efforts (10 firms) Total sample (57 firms)
Effect Effect on 1977 Effect Effect on 1977 Effect Effect on 1977

on 1977 total stock- on 1977 total stock- on 1977 total stock-
earnings holders’ equity earnings holders’ equity earnings holders’ equity
No. of firms
providing
point estimates 44 47 9 10 53 57
Mean -47.18% - 34.54 % _ - 11.80% -8.34% -41.17% - 29.95 %
Median - 27.00 % -30.00% - 13.00% - 3.50 % -21.50% - 19.50 %
Standard
deviation 60.69 % 28.62 % 11.07% 9.18% 56.95 % 28.06 %
“Tabled values are changes in earnings (stockholders’ equity), measured as a percentage of the 1977 earnings
(stockholders’ equity) under full cost accounting, that would result from a switch to successful efforts accounting specified in
SFAS No. 19.
Table 2
Definitions of regression model variables.
CAR s = Cumulative (weekly) abnormal return fol: the jth firm.

TOTCAPj =Total capitalization (in $100 millions) of firm j at fiscal year end 1977, measured
by the book value (BV) of total long-term (L-T) debt plus BV of preferred stock
plus the market value (MV) of equity. MV equals the number of common shares
outstanding at fiscal year end times the 200-day moving average price as of that
date as reported in the Standard & Poors Daily Stock Price Record. A number of
firms in the sample were traded in the over-the-counter market and possessed
nominal (low) share price quotations which can change dramatically on a
percentage basis from day to day: Accordingly, the 200-day moving average price
was used as means to obtain a more stable and representative measure of the
market value of the equity of the firm.
BV of L-T debt + BV of preferred stock + MV of equity
rOTC, Pj
$100 million
TOTSEEj =Total stockholder equity effect of SFAS No. 19 on firm j, measured as a
percentage of total stockholders' equity under FC accounting as reported in 1977
annual report.
Total stockholder's equity (SE)-Total stockholder's equity (FC)
TOTSEEj
Total stockholder's equity (FC)
EXPLORj = Exploration expenditures as a percentage of firm f s total revenues. The average
total capital expenditures scaled by total revenues for the years 1974-1977 was
used as a proxy for this variable.
1977 Total capital expenditures~
EXPLORj __1
-~ ~
, = 1974. Total revenues,
OEMKTj =Firm f s leverage, measured by the book value (BV) of long-term debt plus BV of
preferred stock at fiscal year end 1977 divided by the market value (MV) of
equity.
BV of L-T debt + BV of preferred stock
OeMr
MV of equity
PVSDTj = Public debt of firm j, represented by a one/zero dummy variable. Firms with
public debt in their capital structure were assigned a one while those with no
public debt were assigned a zero.
CON TRCSj = Loan covenant and management compensation plan variable of firm j represented
by a one/zero dummy variable. Firms with either debt covenants or management
compensation plans defined in terms of reported accounting numbers were
assigned a one, while firms without such contracts written in terms of accounting
numbers were assigned a zero.
two week periods. The FASB's Exposure Draft was dated Friday, July 15,
1977 although the Wall Street Journal did not report on its contents until
Wednesday, July 20. The one week CAR was measured over the week ending
on July 15, 1977. Lev (1979) reports that price declines for FC shares were
most dramatic on Monday through Wednesday, July 18-20. Therefore, a two
week cumulation period was also tested and was expected to more fully
D. I4(. Collins et al., Cross-sectional tests of oil and gas returns 57
Table 3
Summary statistics - - Variables in regression m o d e l ; 57 firms, a
CAR~ - - one week (%) CAR~ - - two weeks ( ~ )
O n e factor m o d e l T w o factor model O n e factor m o d e l T w o factor model
Median 0.80 0.25 - 3.26 - 0.24

Mean - 0.07 - 0.55 - 3.19 - 0.24
Standard
deviation 7.07 7.08 9.46 9.61
EXPLOR PUBDT C O N T R CS
TOTCAP TOTSEE ( ~ of DEMKT
($100 million) ( % change) revenue) (~,,) Yes No Yes No
Median 1.029 - 19.5 42.8 44.4 20 37 42 15

Mean 2.285 - 29.9 55.9 84.1 -- --
Standard
deviation 3.436 28.1 43.0 140.6
aCAR = c u m u l a t i v e a b n o r m a l residual,
TO T C A P = c o m p a n y s~e,
TOTSEE = percentage effect of S F A S No. 19 o n total stockholders' equity,
EXPLOR = capital expenditures as a percentage of revenues,
DEMK T = B V of d e b t / M V of equity,
PUBDT = whether firm has public debt in capital structure,
CONTRCS =whether firm has debt c o v e n a n t s or m a n a g e m e n t c o m p e n s a t i o n p l a n defined in
terms of reported a c c o u n t i n g numbers.
capture the immediate market adjustments associated with the FASB's

Exposure Draft.
The cumulative abnormal returns (CAR~'s) in eq. (1) were measured in two
ways, by a single factor market model and by a two factor market and
industry model. For the former, the CARj's were estimated using weekly data
and the market model:
(2)
where t is a weekly index for a 70 week estimation period beginning 3/12/76

and ending 7/8/77 (one week prior to the Exposure Draft), Rj, is the jth
firm's return defined as log e (Pricet+Dividendt)/Pricet_l; R,.r is the return on
the market portfolio surrogated by the NYSE equally-weighted retui'n index.
The least squares estimated, ~ and ~j, were combined with actual weekly
market returns to derive risk adjusted abnormal returns (conditional on the
market factor only) in the following manner:
e~, = Rj,-[~j + l~.,Rm~], (3)

where t signifies a one week period ending 7/15/77 or a two week period
ending 7/22/77.
r T=I for 7/15/77,

One factor CAR t = ~ e~t, T= 2 for 7/22/77. (4)
t=l
The second procedure for estimating CARj introduces an industry factor

into eq. (2) to control for industry events and thus remove a major potential
source of cross-correlation in the dependent return measures. Operationally,
the industry index was measured as the equally weighted portfolio return of
the combined FC and SE samples (75 firms) in the Collins and Dent (1979)
study,
1 N
Rn=-Nj ~=xRJt' t=3/12/76 to 7/8/77, N=75. (5)
This index contains relatively more SE firms (40 ~o) than the current sample
(18 ~), and is larger than the current sample. Since the objective in using an
industry index is to remove the influences of unspecified industry factors
which operate irrespective of accounting method, the broader based index
was deemed more appropriate than an index based on the current 57 firm
sample.
To insure that the industry factor was orthogonal to the market factor, the
industry index was first regressed against the market index for the 70 week
estimation period. The residuals, which are uncorrelated with the market
return, were determined according to
(6)
These residuals were then used in the following model to estimate a two
factor version of the market model over the 70 week estimation period:
fl~R,~t+ y~rbt'+ujr (7)
Using procedures analogous to those described in eqs. (3) and (4) above
(with the addition of an industry factor), a two factor CAR~ was calculated
for each firm for the two cumulation periods.
5.4. Interdependence among independent variables

As noted in section 4.7, there is reason to expect firm size (TOTCAP) to
be related to leverage (DEMKT), public debt (PUBDT) and debt
covenants/manageme:~t compensation plans defined in terms of accounting
numbers (CONTRCS). To ascertain the degree of dependence among these

and other regressors included in the cross-sectional model, simple
correlations were computed. These results are reported in table 4.19
Table 4
Simple correlation coefficient matrix between independent variables based on 57 observations, a
TOTCAP TOTSEE EXPLOR DEMKT PUBDT CONTRCS
TO TC A P 1.00
TO T S E E 0.09 1.00
EXPLOR - 0.03 - 0.56 ~ 1.00
DEMK T 0.37 c - 0.27 ~ - 0.22 1.00
PUBDT b 0.43 ~ -0.22 -0.07 0.26 ° 1.00
CONTRCS b 0.42 c -0.09 0.09 0.27 c 0 . 2 7 Ca 1.00
"See footnote a of table 3 for definition of variables.

bpoint biserial correlation coefficients were computed for these dichotomous (1,0) variables
and the continuous independent variables.
cSignificant at the 0.05 probability level or lower.
dSinee both variables are dichotomous a chi-square 1 d.f. (phi) is reported.
The simple correlations in table 4 indicate that TO TCAP is significantly

positively related to DEMKT, PUBDT and CONTRCS. Significant positive
associations are also found among PUBDT, DEMKT and CONTRCS,
indicating that the more highly leveraged firms tend to have public debt in
their capital structure and tend to have restrictive covenants and other
financial contracts ,written in terms of accounting numbers. The strong
negative correlation between EXPLOR and TO TSEE bears out the earlier
assertion that the adverse financial statement effects of SFAS No. 19 would
be most severe for those firms investing the greatest amounts of each revenue
dollar in exploration activities. It also suggests that TO TSEE may proxy for
EXPLOR in measuring a firm's concentration in high risk exploration
activities.
Although interrelationships among firms' financial structures, operating
characteristics, financial contracts, monitoring devices and the financial
statement consequences of mandatory accounting changes are of considerable
interest in their own right, the more immediate concern is how these
dependencies may affect the multiple regression results. In general, the
interdependencies within the set of regressors do not appear severe by
traditional standards, the most serious being between EXPLOR and
TO TSEE. In the subsequent analyses, results for the complete model with all
19Multiple regressions, in which each independent variable was regressed against the
remaining independent variables, were run in an attempt to detect possible dependencies among
linear combinations of independent variables. These results corresponded closely to the simple
correlation results reported in table 4.
variables included will be reported first. Selected variables will then be

omitted to help ascertain which variables are most important in explaining
cross-sectional return performance, and to determine how sensitive the
findings are to the observed interdependencies among selected explanatory
variables.
6. Results
6.1. Multiple regression results

Tables 5 and 6 present the results of estimating eq. (1), using both CAR
measures and both cumulation periods. Two results are that the independent
variables' explanatory power is far greater for the two week period and the
regression results are robust to specification of the process generating
returns. For example, using the two factor CAR's as the dependent variable,
the full model with all variables included explains 12 % of the cross-sectional
variance in abnormal returns in week one [table 5, panel B, model 1] and
33 % of the variance in the two week period [table 6, panel B, model 1]. The
corresponding numbers for the one factor model are 12% and 31%.
Furthermore, if one compares the magnitudes of the coefficients that arise
when using the one and two factor models, they are generally quite similar
[cf. panels A and B of table 6].
The fact that the overall explanatory power of the model is much greater
for the two week period than for the one week period is consistent with Lev
(1979). Thus, although the market reaction was relatively rapid, occurring
within a very few days of the Exposure Draft, the Friday issuance clearly did
not elicit the full impact. Because the two week cumulation period appears to
capture the market reaction to a greater extent than the one week period,
our subsequent discussion will concentrate on the two week period.
Limiting our attention to panel B of table 6, which presents results using
the two factor model over the two week cumulation period, we find that
model 1, which contains the results using all six variables, is significant at the
0.001 level with an adjusted R 2 of 0.25. TOTSEE and CONTRCS have the
hypothesized signs and individually are the most significant explanatory
variables with t-values which are significant at the 0.001 and 0.02 levels,
respectively. Note that firm size (TOTCAP) enters with a negative coefficient
which is significant at the 0.08 level. This result is inconsistent with the
'political cost' hypothesis while tending to support the various reasons which
were posited for expecting a negative relationship. Whether the observed
negative association reflects Banz's finding that investors demand a market
discount for securities of very small firms which are relatively 'unknown' is
addressed later in the paper.
In the full model EXPLOR, DEMKT and PUBDT, contrary to the
hypotheses, enter with positive signs. As noted previously, however,
D . W . C o l l i n s e t al., C r o s s - s e c t i o n a l t e s t s o f oil a n d g a s r e t u r n s 61
,~ f..
o o • o. o. o. o. o o.
I I I
O+
oo t,~ o~ o ¢q o
t,~ ¢q o
e~ ,- o. o. o. o. ,-
o:¢
o e~ o.~. o ¢q o,..~
I I I I I I I I I I I I I I
"~.+
o.~. o.,- o.~.
I I I I I I I I
o
I
v~
o.~. . . o.--.
I f~
.=.
~+
ca,v-,
00"~" ["00
o. cq
o ~
e-. ['~ c5,:

o
I I
e- t"4~ ¢',100 ,..~t"- ¢',1~

c C
rZ
o
"0
c,,t o ca
o+
r~
¢'~ ("4 ~ ("4

o~
~...+
I I I I I I I I II II II II
-~+
0,"~ 00,...~ ¢¢~...-.,
• "'~ P"- ~¢~ ¢',1 ¢'¢'t
• • O t ~
I I I I I I I I
.o
+
0
D~
w~r.. ,~1"0
~" ,...1
I e~
.=.
-=¢J
4.-'
+ ~ ,.:1. ¢¢~
qr't
.,.J
~°'-I I I I I 71 I I ~..~o
P'~ ~4"t P".- ,...~

0oo 0oo
¢ C
r~ 0",-~ 0,"~ 0 0 0("4 o~ oed
I I
"2 "~
0
¢¢'t
EXPLOR has a significant negative correlation with TOTSEE while

D E M K T and P U B D T are positively associated with TOTCAP and with one
another. To determine how sensitive the estimated coefficients are to the
indicated multicollinearity within the set of regression variables, the model
was re-estimated omitting selected variables. Model 2 in tables 5 and 6 reflect
results where firm size (TOTCAP) has been omitted. Note that in these
cases, t h e overall explanatory power of the equation changes only slightly.
PUBDT now enters with, the hypothesized negative sign but is insignificantly
different from zero with a t-value of -0.83. The coefficient on DEMKT
remains positive and is insignificant. The TOTSEE and CONTRCS variables
continue to reveal significant relationships with abnormal returns in the
hypothesized directions. These results suggest that TO TCAP is an
explanatory variable of lesser importance than TOTSEE and CONTRCS,
and that, in part, it surrogates for PUBDT.
Model 3 reflects results where TO TSEE in addition to size has been
omitted from the regression. In this case EXPLOR, DEMKT and PUBDT
enter with the hypothesized negative signs but are insignificantly different
from zero. Of these, E X P L O R gains most in significance and is more than
one standard deviation from zero. This suggests that TOTSEE, at least in
part, is surrogating for EXPLOR. CONTRCS is by far the most significant
explanatory variable in this reduced equation. It should also be noted that
with the exclusion of TO TSEE the overall explanatory power of the model
as revealed in the adjusted R 2 and F-statistics is substantially lower than
when this variable is included.
The preceding analysis suggests that TOTSEE, CONTRCS and to a lesser
extent TO TCAP are the most significant factors in explaining the cross-
sectional variation in abnormal returns associated with the release of the
Exposure Draft. In model 4, only these three variables are retained and the
remaining variables are excluded from the regression equation. For the two
week cumulation period (table 6), the adjusted R 2 of 0.243 and F-statistic of
7.00 suggests that this reduced model performs as well as, or better than, the
full model. Where returns are adjusted for both market and industry
influences (panel B of table 6), the TOTCAP, TOTSEE and CONTRCS
variables are significant at the 0.10, 0.001, and 0.03 levels, respectively.
6.2. A closer look at competing theories

Earlier it was noted that there are competing theories which are consistent
with the observed associations between security returns and the TO TSEE
and TO TCAP variables. This section presents evidence relevant to assessing
the efficacy of these competing theories. The PUBDT variable is also
examined in more detail.
6.2.1. Test of naive investor hypotheses

In sections 3.3 and 3.4 it was argued that the mandatory change from FC
to SE accounting would impose various kinds of costs on affected companies
and/or would tend to increase the estimation risk faced by investors with
respect to these firms. The existence of various types of financial contracts
and monitoring devices defined in terms of accounting numbers led us to
propose that the percentage change in the total stockhoders' equity account
would proxy for the increased costs and/or estimation risk associated with
this accounting change. Therefore, we hypothesized a positive relationship
between TO TSEE and security returns.
The strong positive association observed between TOTSEE and security
returns at or about the time of the Exposure Draft issuance is also consistent
with the naive investor and modified naive investor theories which holds that
accounting changes, per se, can affect stock prices through their effects on
reported numbers, especially earnings numbers. If this theory is, in fact,
driving our results we would expect
(1) a strong positive association between security returns and the current
earnings per share effect (EPSE) of SFAS No. 19, and
(2) that firms with essentially the same EPSE and TOTSEE effects would
experience similar market declines irrespective of any other basis on
which these firms may be partitioned.
Fifty-three of our 57 sample firms provided point estimates of the

percentage change in 1977 EPS that would result from adoption of SFAS
No. 19. We correlated these changes in earnings numbers with the one and
two week abnormal returns adjusted for both market and industry
influences. The Pearson product moment correlation coefficients were 0.10
and 0.08, respectively, neither of which is significant at conventional levels.
We then replaced TO TSEE by EPSE in our full multiple regression model.
This resulted in an adjusted R 2 of 0.129, an F-statistic of 1.14 for the two
week cumulation period, and a t-value for the EPSE coefficient of 0.45. These
results indicate that the current earnings effect of SFAS No. 19 has little or
no value in explaining cross-sectional differences in observed returns. This is
evidence inconsistent with the naive investor theories.
To further distinguish between the increased costs and/or estimation risk
hypotheses vis-a-vis the naive investor theories, we partitioned the sample on
the basis of our CONTRCS variable, i.e., whether or not firms had financial
contracts or monitoring devices defined in terms of reported accounting
numbers. The average TO TSEE, EPSE, and two week abnormal returns
(adjusted for market and industry influences) for these two groups were then
compared. The results of this analysis are reported in table 7.
From table 7 one can observe that for the two groups of firms there is
very little difference between the average total stockholders' equity effect
(TOTSEE) and the average earnings per share effect (EPSE). If investors
were focusing naively on the expected financial statement effects, per se, of
SFAS No. 19, one would expect to observe little or no difference, on average,
between the risk adjusted returns of these two groups of firms. Note,
Table 7
Comparison of selected variables for firms with and without financial contracts/monitoring
deviceS defined in terms of accounting numbers.
Standard H0: )(1 = )f2

Variable Group Mean deviation t-value
Firms with C O N T R C S ~ - 29.4 ~o 25.2 ~o

(42 firms)
TOTSEE 0.21
Firms without C O N T R C S - 31.5 ~ 35.8 ~o
(15 firms)
Firms with C O N T R C S - 43.5 ~o 60.8 ~o

(42 firms)
EPSE - 0.56
Firms without C O N T R C S - 34.6 ~o 45.7
(15 firms)
Firms with C O N T R C S - 2.0 ~ 9.8

Two week (42 firms)
CAR - 2.74 b
Firms without C O N T R C S 4.6 ~o 7.1 ~o
(15 firms)
aFirms are classified according to whether they have financial contracts/monitoring devices
written in terms of reported accounting numbers. Firms without C O N T R C S may in fact have
financial contracts but indicated they are not written in terms of reported accounting numbers.
bSignificant at 0.01 probability level, one-tail test.
however, that there is a dramatic difference in the average abnormal returns

of these two groups. Those firms with financial contracts and/or monitoring
devices defined in terms of reported accounting numbers had an average
cumulative two week return of - 2 ~ o , while those firms without such
contracts had positive abnormal returns averaging 4 . 6 ~ . This difference is
significant at the 0.01 level. Furthermore, if the market model alone is used
to estimate abnormal returns, the corresponding results are - 4 . 7 ~o for firms
with contracts and + 1 . 0 ~ for firms without contracts defined in terms of
accounting numbers.
These results suggest that investors did not fixate merely on the
accounting numbers per se but looked beyond to the potential effects of the
accounting change on management behavior and thus on the probability
distribution of real cash flows. The evidence is consistent with the view that
stock prices were affected negatively if there existed financial contracts

written in terms of accounting numbers to act as a means by which the
mandated accounting change could impact on investor expectations of
management behavior and real cash flows.
6.2.2. Test of Banz hypothesis

As was noted earlier, the observed negative relationship between firm size
(TOTCAP) and abnormal returns is consistent with competing hypotheses.
The evidence (see table 4) suggests that the larger firms in our sample have
greater amounts of public debt, are more highly leveraged and are more
likely to have financial contracts and/or monitoring devices defined in terms
of reported accounting numbers. TO TCAP surrogates for these factors in
combination, and this helps to explain why TOTCAP enters some of the
regression models, albeit with marginal significance.
Alternatively, Banz (1979) presents evidence that, in general, investors
require an incremental return for securities of very small firms which are
relatively unknown (i.e., there is a negative relationship between risk adjusted
returns and firm size). In order to test this possibility, observe that if Banz's
information cost hypothesis is operating to yield differential returns between
'larger' versus 'smaller' firms in our sample, a significant negative relationship
between TOTCAP and risk adjusted returns should be observed in all time
periods, not only during the weeks adjacent to the Exposure Draft issuance.
We arbitrarily selected a two week period beginning one month before the
Exposure Draft issuance and a second two week period beginning one month
after its issuance. Cumulative abnormal returns (adjusted for market and
industry influences) for each of these two week periods were computed in a
fashion described previously. We then applied our full regression model (with
size included) in an attempt to explain the cross-sectional variation in these
returns. The results of this analysis are reported in table 8.
A comparison of the results reported in table 6, panel B (Exposure Draft
period) with the results reported in table 8 (outside of the Exposure Draft
period) yields two interesting findings. First, although the coefficient on size
has a negative sign for both periods, it is considerably smaller and less
significant outside of the Exposure Draft period than it is in the Exposure
Draft period. Second, the overall explanatory power and significance of the
regression model is negligible outside of the Exposure Draft period. From
the former result, we conclude that the significant negative relationship
between size and abnormal returns in the weeks adjacent to the Exposure
Draft issuance is not due to the size/return relationship hypothesized by
Banz. The latter result strengthens the case that the Exposure Draft period
was a significant 'event' period and that meaningful variables have been
found which help explain the abnormal return behavior of individual firms
during that period.
Table 8
Cross-sectional regression estimates - Outside of critical event period; 57 observations.8
Coeffcient
a0 al a2 a3 a4 a5 a6
Model Unadjusted F
Constant TOTCAP TOTSEE EXPLOR DEMKT PUBDT CONTRCS R2 statistic
Pre-exposure - 0.007 -0.001 0.001 0.006 -0.000 0.023 0.012 0.021 0.18
draft period (-0.23) (-0.15) (0.02) (0.16) ( - 0.03) (0.79) (0.40)
@i/12/77-6/23/77)
Post-exposure -0.033 * - 0.003 - 0.022 0.013 - 0.004 0.004 0.029 0.085 0.78
draft period (-1.51) ( - 0.85) (-0.54) (0.46) ( - 0.60) (0.17) (1.31)
(8/8/77-8/19/77)
“See footnote a of table 3. Asymptotic t-statistics in parentheses.
1 /.,, ,/
,, ,,, ,,,
6.2.3. Renegotiation costs of public debt vs. private debt

The evidence presented to this point suggests that the observed negative
market adjustment to the proposed elimination of FC accounting was
primarily a function of the increased contracting costs and/or estimation risk
as surrogated by the TOTSEE and CONTRCS variables. In this section we
consider in further detail one aspect of the contracting cost argument;
namely, that the observed market declines were a function of the higher costs
of renegotiating and/or redeeming public debt vis-a-vis private debt [see
Holthausen (1979) and Leftwich (1979)].
Table 9
Comparison of abnormal returns holding financial contracts or type of debt in capital
structure constant.
Two week CAR (adjusted

for market and industry)
No. of firms Standard

in sample Mean deviation t-value
Public debt with

accounting based 18 - 2.2 % 7.6
covenants
Private debt with -0.245
accounting based 21 - 1.5 ~ 12.1 ~o
covenants
In the present analysis, the public debt variable was found to possess
relatively low explanatory power. Possibly this is because the public
debt/private debt dichotomy is an imperfect surrogate in our sample for the
existence of loan covenants denominated in terms of reported accounting
numbers. Because of this lack of correspondence it is conceivable that the
costs of renegotiation/redemption as measured by the PUBDT variable were
obscured in our analysis. One additional contrast between subsets of firms
within our sample was conducted to investigate this possibility. The results
are reported in table 9.
If the costs of renegotiation or redemption of public debt are a potential
explanation of the valuation consequences of mandatory accounting changes
as Holthausen and Leftwich argue, then one would expect significant
differences in the observed return performance of firms with public debt vs.
no public debt (private debt only) where we hold the existence of accounting
based loan covenants constant. Thirty-nine of the 57 firms in our sample
indicated the presence of accounting based loan covenants in loan
agreements. Eighteen of these firms had public debt in their capital structure
while the remaining 21 firms did not. As can be seen from table 9, the
average cumulative two week abnormal return (adjusted for market and
industry influences) was -2.2Y/o for the firms with public debt and - 1 . 5
for firms with no public debt. The computed t-value for a test of differences
in these means was -0.245 indicating no significant difference. This suggests
that the differences in renegotiation costs of public vs. private debt are
insufficient to cause a significant difference in observed return behavior in
our sample.
6.3. Summary and conclusions

We have considered several explanations for the observed decline in stock
prices associated with the FASB's Exposure Draft issuance which called for
the elimination of FC accounting. These are (1) naive investor theory, (2)
modified naive investor theory, (3) contracting cost theory, and (4) estimation
risk theory. A model was developed using variables based on these theories
to explain differential market reaction to this mandatory accounting change
in the oil and gas industry. The results were consistent with the conclusion
that the FASB's proposal had a measurable negative effect on the equity
values of affected firms.
The set of variables which was hypothesized to measure the increased
contracting costs and/or estimation risk associated with the FASB's
proposed elimination of FC accounting was found to explain a significant
proportion of the cross-sectional variation in abnormal return performance
of our sample firms in the two weeks centered on the Exposure Draft
issuance. The two most significant explanatory variables were TO TSEE (the
percentage change in total stockholders' equity) and CONTRCS (the
existence in the firm's ownership structure of financial contracts denominated
in terms of reported accounting numbers). Firm size (TOTCAP) was found
to be marginally significant. Three variables exhibited relatively low
explanatory power, namely, EXPLOR (exploration costs as a percent of
revenue), DEMKT (leverage), and P U B D T (the existence of public debt in
the firm's capital structure).
Due to the fact that several of the independent variables proxy for both a
reduction in expected cash flows and an increase in estimation risk, we
cannot tell which theory, contracting costs or estimation risk, provides a
better explanation for the downward revaluations in the equity securities of
affected companies. Thus while we have identified relevant variables to
explain the relative price adjustments and can make reasonably unambiguous
statements about their relative explanatory power, further research is
required to isolate the relative influences of the alternative hypothesized
causes for the decline in share prices (i.e., reduced cash flows and/or
increased estimation risk). One aspect of the contracting cost hypothesis (i.e.,
public debt is more costly to renegotiate than private debt) and the naive
70 D.W.. Collins et al., Cross-sectional tests of oil and gas returns
investor hypothesis were considered as alternative explanations for our

findings. Evidence from additional empirical analyses was inconsistent with
these competing hypotheses.
Along with the Leftwich (1979) study this paper represents one of the first
attempts to consider competing explanations for the economic effects of
mandatory accounting changes, and the first to suggest estimation risk as a
possible explanation. Two difficulties encountered are that surrogate
variables are used for the various economic variables and that the sample
size is relatively small (57 firms). Therefore, we are aware that a more
complete understanding of the capital market behavior awaits further
research in this area. Nevertheless, the effects of mandated accounting
changes and our understanding of how they occur appear now to be well
enough documented that a suggestion to policy-making bodies is in order. In
particular, the results suggest that accounting policy-makers who are
considering the elimination or alteration of existing accounting alternatives
should recognize that accounting choice decisions and investment/financing
decisions of the firm are interdependent, and that changes in the former may
affect the latter, thereby creating wealth losses and/or wealth transfers among
the firm's capital suppliers.
References
Banz, Rolf W., 1979, The relationship between market value and return of common stocks,
Unpublished manuscript (Center for Research in Security Prices, University of Chicago,
Chicago, IL).
Benston, George J., 1978, An evaluation of studies on the effect of the FASB's Exposure Draft
and Statement 19 on the share prices of oil and gas producing companies, Statement before
the Securities and Exchange Commission, May 4, SEC File $7-715.
Ben-Zion, Uri and Sol S. Shalit, 1975, Size, leverage, and dividend record as determinants of
equity risk, Journal of Finance XXX, 1015-1026.
Collins, Daniel W. and Warren T. Dent, 1979, The proposed elimination of full cost accounting
in the extractive petroleum industry, Journal of Accounting and Economics l, 3--44.
Deakin, Edward B., 1978, An analysis of differences between nonmajor oil firms using successful
efforts and full cost methods, Unpublished manuscript (University of Texas, Austin, TX).
Dhaliwal, Dan S., 1980, The effect of the firm's capital structure on the choice of accounting
methods, Accounting Review LX, no. 1, 78-84.
Dyckman, Thomas R. and Abbie J. Smith, 1979, Financial accounting and reporting by oil and
gas producing companies, Journal of Accounting and Economics 1, 45-75.
Federal Trade Commission, 1978, Response of the Federal Trade Commission to the request by the
Securities and Exchange Commission for comments on accounting practices for oil and gas
producers, Feb. 27, SEC File $7-715.
Financial Accounting Standards Board, 1978, Additional comments of the Financial Accounting
Standards Board to the Securities and Exchange Commission: Accounting practices - - Oil
and gas producers, SEC File $7-715.
Fogelson, Joseph S., Jr., 1978, The impact of changes in accounting principles on restrictive
covenants in credit agreements and indentures, Business Lawyer XXXIII, 769-787.
Foster, George, 1980, Accounting policy decisions and capital market research, Journal of
Accounting and Economics 2, 29-62.
Hagerman, Robert L. and Mark E. Zmijewski, 1979, Some economic determinants of accounting
policy choice, Journal of Accounting and Economics 1, 141-161.
Hirshleifer, Jack, 1958, On the theory of optimal investment decision, Journal of Political
Economy LXVI, 329-352.
Holthausen, R.W., 1979, Bond covenants and the choice of accounting techniques: The case of
alternative depreciation methods, Unpublished manuscript (University of Rochester,
Rochester, NY).
Jensen, M.C. and W.H. Meckling, 1976, Theory of the firm: Managerial behavior, agency costs,
and capital structure, Journal of Financial Economics 3, 305-360.
Kalay, A., 1978, Towards a theory of corporate dividend policy, Unpublished manuscript
(University of Rochester, Rochester, NY).
Klein, Roger W. and Vijay Bawa, 1976, The effect of estimation risk on optimal portfolio choice,
Journal of Financial Economics 3, 215-231.
Klein, Roger W. and Vijay Bawa, 1977, The effect of limited information and estimation risk on
optimal portfolio diversification, Journal of Financial Economics 5, 89-111.
Leftwich, Richard, 1979, The impact of mandatory changes in accounting principles on
corporate loan agreements, Unpublished manuscript (University of Rochester, Rochester,
NY).
Lev, Baruch, 1979, The impact of the FASB's oil and gas exposure draft on the stock market,
Accounting Review LIV, 485-503.
Levy, Haim, 1978, Equilibrium in an imperfect market: A constraint on the number of securities
in the portfolio, American Economic Review 68, 643-658.
Lintner, John, 1965, The valuation of risk assets and the selection of risky investments in stock
portfolios and capital budgets, Review of Economics and Statistics, 13-37.
Pogue, Thomas F. and Robert M. Soldofsky, 1969, What's in a bond rating?, Journal of
Financial and Quantitative Analysis 4, 201-228.
SEC Directorate of Economic and Policy Research, 1978, Full cost vs. successful efforts: A study
of a proposed accounting change's competitive impact, Unpublished manuscript of the
Securities and Exchange Commission.
Sharpe, William F., 1964, Capital asset prices: A theory of market equilibrium under conditions
of risk, Journal of Finance 19, 425--442.
Smith, C.W. and J.B. Warner, 1979, On financial contracts and optimal capital structure: An
analysis of bond covenants, Journal of Financial Economics 7, 117-161.
Sunder, Shyam, 1976, Properties of accounting numbers under full costing and successful efforts
costing in the petroleum industry, Accounting Review LI, no. 1, 1-18.
Watts, Ross L., 1977, Corporate financial statements, A product of the market and political
processes, Australian Journal of Management 2, 53-75.
Watts, Ross L. and Jerold L. Zimmerman, 1978, Towards a positive theory of the determination
of accounting standards, Accounting Review 53, 112-134.
Williams, Joseph T., 1977, Capital asset prices with heterogeneous beliefs, Journal of Financial
Economics, 219-239.

Collins 1981

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Collins 1981

Uploaded by

Copyright:

Available Formats

Journal of Accounting and Economics 3 (1981) 37-71.

North-Holland Publishing Company

THE ECONOMIC DETERMINANTS OF THE MARKET REACTION

Daniel W. COLLINS and Michael S. ROZEFF

Received April 1979, final version received September 1980

0165--4101/81/0000--0000/$02.50 © North-Holland Publishing Company

accounting. 1 While we interpret this evidence as dearly indicating that the

2. Effects of SFAS N o . 19 on financial statements and contracts

2.1. The effect of SFAS No. 19 on financial statements

Another effect of SFAS No. 19 was a likely increase in the volatility of

2.2. The effect of SFAS No. 19 on lending agreements

2.3. The effect of SFAS No. 19 on management compensation agreements

monitoring devices designed to restrict management decisions that would

3. Theories of stock price decline

3.1. Naive or irrational investory theory

3.2. Modified naive investor theory

3.3. Contracting cost theory

accounting method is viewed as being embedded in the overall choice

3.3.1. Information costs

firm's existing and planned investment/financing choices. 13 Hence the costs

3.3.2. Agency costs

3.4. Estimation risk theory

3.4.1. Explanation of estimation risk

3.4.2. SFAS No. 19 and estimation risk

investments available to the firm and/or which investments will be selected

for themselves via their adoption of a particular set of investment/financing

4.1. Percentage change in total stockholder equity

Hypothesis 1. The abnormal performance of the common stocks of firms

Several of the explanations of price decline imply a testable implication

4.2. Existence of a loan agreement

Hypothesis 2. Firms with loan covenants defined in terms of reported

4.3. Existence of a management compensation agreement

Hypothesis 3. The abnormal performance of the common stocks of firms

In our sample, with three exceptions, every firm with a management

4.4. Debt/equity ratio

Hypothesis 4. The abnormal performance of the common stocks of firms

4.5. Proportion of revenues devoted to exploration and production

Hypothesis 5. The abnormal performance of the common stocks of firms

It is clear that firms more heavily involved in exploration and production

4.6. Public debt vs. private debt

Hypothesis 6. The abnormal performance of the common stocks of firms

In part, the public/private debt variable measures renegotiation costs

4.7. Firm size

Hypothesis 7. Firm size is a significant factor in explaining the cross-

5. Sample selection, data definitions and methodology

5.1. The sample

Table 1 summarizes the estimated effects of SFAS No. 19 on earnings and

5.2. Multiple regression model

In order to examine empirically the relationships hypothesized in sections

Table 2 defines these variables and table 3 presents summary statistics.

5.3. Dependent variable measures

Effect Effect on 1977 Effect Effect on 1977 Effect Effect on 1977

CAR s = Cumulative (weekly) abnormal return fol: the jth firm.

CAR~ - - one week (%) CAR~ - - two weeks ( ~ )

O n e factor m o d e l T w o factor model O n e factor m o d e l T w o factor model

Median 0.80 0.25 - 3.26 - 0.24

Median 1.029 - 19.5 42.8 44.4 20 37 42 15

capture the immediate market adjustments associated with the FASB's

where t is a weekly index for a 70 week estimation period beginning 3/12/76

e~, = Rj,-[~j + l~.,Rm~], (3)

r T=I for 7/15/77,

The second procedure for estimating CARj introduces an industry factor

fl~R,~t+ y~rbt'+ujr (7)