Accelerat ing t he world's research.

A Market Based Analysis of Income

Smoothing
Stuart Michelson

Journal of Business Finance & Accounting

Cite this paper Downloaded from Academia.edu 

Get the citation in MLA, APA, or Chicago styles

Related papers Download a PDF Pack of t he best relat ed papers 

Income Smoot hing and Risk-Adjust ed Performance

St uart Michelson

Firm-Specific Det erminant s of Income Smoot hing in Bangladesh: An Empirical Evaluat ion
ahsan habib

Environment al uncert aint y and t he market pricing of earnings smoot hness

ahsan habib
See discussions, stats, and author profiles for this publication at:
https://www.researchgate.net/publication/229703974

A Market Based Analysis of

Income Smoothing

Article in Journal of Business Finance & Accounting · December

2006
DOI: 10.1111/j.1468-5957.1995.tb00900.x

CITATIONS READS

74 181

3 authors, including:

Stuart Michelson Charles W. Wootton

Stetson University Eastern Illinois Univer…
31 PUBLICATIONS 211 12 PUBLICATIONS 274
CITATIONS CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Stuart Michelson on 07 April 2014.

The user has requested enhancement of the downloaded file.

 Joumat of Business Finance & Accounting, 22(8), December 1995, 0306-686X

A MARKET BASED ANALYSIS OF INCOME

SMOOTHING
STUART E, MICHELSON, JAMES JORDAN-WAGNER AND
CHARLES W , WOOTTON*

INTRODUGTION

Over the last three decades income smoothing has been analyzed in various
ways. Several studies have focused on three issues: (a) do firms actually smooth
income; (b) the smoothing ability of various accounting techniques; and (c)
conditions under which smoothing is effective (Lev and Kunitzky, 1974, p,
268), Smoothing studies have also focused on (a) objectives of smoothing
(management motivation), (b) objects of smoothing (operating income, net
income), (c) dimensions of smoothing (real or artificial), and (d) smoothing
variables (i,e,, extraordinary items, tax credits) (Ronen and Sadan, t98t, p, 6),
However, little attention has been given to the reaction in the marketplace
to income smoothing. Yet for forty years, many assumptions and conclusions
pertaining to relationships between smoothing and the marketplace have been
set forth, Hepworth (1953) advanced the idea that stable earnings give owners
and creditors a more confident feeling toward management, Gordon (t964,
p, 262) proposed that stockholder satisfaction increases with the rate of growth
in a company's income and the stability of its income, Beidleman (1973, p,
655) suggested that as analysts become more enthusiastic about self smoothers,
smoothing may indirectly widen the market for a firm's shares and there should
be a favorable effect on share value and cost of capital. Lev and Kunitzky (1974,
p, 268) stated that their results indicate that income variability is significantly
correlated with both overall and systematic risk measures, Moses (1987, p,
366) stated that smoothing implies a direct, cause-and-effect relationship
between earnings fluctuations and market risk.
The purpose of this paper is to test for an association between income
smoothing and performance in the marketplace. In doing so, the paper examines
(a) the tendency of major corporations to become income smoothers, (b) the
difference in the mean returns on the common stock of smoothing and non

• The authors are respectively, Associate Professor of Finance, Associate Professor of Finance,
and Professor of Accountancy, at Eastern Illinois University, They would like to thank the
participants at the SWFAD Decision Sciences Annual Meeting and an anonymous referee for helpful
comments and suggestions. Any errors are the authors responsibility, (Paper received September
1993, revised and accepted May 1994)
Address for Correspondence: Stuart E, Michelson, Associate Professor of Finance, Department
of Accountancy & Finance, Lumpkin GoUege of Business, Eastem Illinois University, Charleston,
IL 61920, USA,

© Blackwell Publishers Ltd, 1995, 108 Cowley Road, Oxford OX4 IJF, UK 1179
and 238 Main Street, Cambridge, MA 02142, USA,
1180 MICHELSON, JORDAN-WAGNER AND WOOTTON

smoothing companies, and (c) the relationship between perceived market risk
and income smoothing.

BACKGROUND

Income smoothing can result from either natural smoothing or intentional

smoothing. Natural smoothing implies that the income process inherently
produces a smooth income stream (Eckel, 1981, p. 28). The classic example
of this type of process would be public utilities. In contrast, intentional
smoothing of income is usually attributed to the actions of management and
can be classified as either real or artificial. Real smoothing is interpreted as
the endeavors taken by management in response to changes in economic
conditions. In contrast, artificial smoothing implies a deliberate effort to
'artificially' reduce the variability of the income stream (Imhoff, 1981, p. 24).
Changes in a company's earnings can ensue from either real or artificial
smoothing. Horwitz (1977) asserts that normally only real smoothing can affect
the cash flow of a company. If increased demand for a product results in
increased sales and a corresponding increase in earnings, this is an example
of real smoothing. Changes in income can eilso result from acts that do not
directly affect cash flows. For example, a company can change its income by
(1) accounting dimensions, (2) allocation methods, or (3) classification of
expenses (Barnea et al., 1976, p. 11).
Many of the early smoothing studies were directed toward identifying the
smoothing devices that companies used to smooth income. In these analyses,
individual variables such as extraordinary items or investment tax credits often
were put forth as logical smoothing devices. However, Zmijewski and
Hagerman (1981) proposed that companies do not select accounting procedures
independently, but rather they consider the overall effect of all accounting
procedures on income. Similarly, Eckel (1981) suggested that the selection of
one accounting procedure may smooth income, but the selection of another
may work in the opposite direction. Therefore, it is the overall outcome of these
attempts that must be examined.
Income smoothing has been defined as a deliberate attempt by management
to signal information to financial users (Ronen and Sadan, 1981, p. 2).
Therefore, the smoothing efforts of management must be successful for a
company to be considered a smoother (Imhoff, 1981). It is this overall,
successful, effect of smoothing that recent researchers (Eckel, 1981; Belkaoui
and Picur, 1984; and Albrecht and Richardson, 1990) have used to classify
companies as being smoothers or non smoothers, and it is this method of
classification that is utilized in the present study.

RELATED STUDIES
In one of the first articles on income smoothing, Hepworth (1953) offered
various motivations for income smoothing and presented accounting techniques

© Blackwell Publishers Lid. 1995

 ANALYSIS OF INCOME SMOOTHING 1181

that could achieve smoothing. Gordon (1964) predicted that management

smooths income and that stockholder satisfaction increases with the stability
ofa firm's income. Copeland (1968) defined what a good smoothing device
must possess and found that increasing the time series reduced misclassifications
of smoothers. Dascher and Malcom (1970) examined fifty-two firms in the
chemicEil and chemical preparations industries and found that deliberate
smoothing practices had been employed. Beidleman's (1973) results strongly
suggested that many firms employ certain devices to normalize reported
earnings. In an exjimination of 260 companies. Lev and Kunitzky (1974) found
that the smoothness of several variables (e.g., sales, production, capital
expenditures) was significantly correlated with both the overall and systematic
common stock risk measures. However, they did not test the relationship
between smoothness of these variables with the market performance of the
common stock. Imhoff (1977) suggested that income smoothing can be studied
by comparing the variance of sales to the variance of income. In a study of
ninety-four industricd companies, he found no evidence of income smoothing.
Later, Imhoff (1981) confirmed that the smoothing model employed had a
significant impact on the identification of the smoothing sample. In a study
of sixty-two industrial companies over a twenty year time frame, Eckel (1981)
found only two firms that appeared as if they smoothed income. In an extensive
study, Ronen and Sadan (1981) established motives for income smoothing and
the engagement of management in smoothing practices. In an examination
of 117 core companies and 57 periphery companies, Belkaoui and Picur (1984)
concluded that income smoothing is stronger in the periphery sector than in
the core area. Moses (1987) demonstrated that smoothing was associated with
firm size, bonus plan incentives, and deviations from expected earnings.
Albrecht and Richardson (1990), in a study of 128 core companies and 128
periphery companies, concluded that income smoothing existed and was fairly
evenly distributed in various sectors of the economy.

METHODOLOGY
This study examines the 500 stocks contained in the Standard and Poor's 500
Index (S&P 500) on December 31, 1991. The data for computing beta, sales,
and the four classification variables (operating income after depreciation, pretax
income, income before extraordinary items, and net income) were collected
from Standard and Poor's COMPUSTAT for 1980 through 1991. As
COMPUSTAT reclassifies firms' financial data in order to make intercompany
comparisons possible, COMPUSTAT's definition for each component (e.g.,
net income) was used throughout the study. The daily return for each stock,
the stock price, the totad number of outstanding shares, and the SIC codes were
gathered from the Center for Research in Security Price at the University of
Chicago (CRSP) tapes for 1982 through 1991. Companies that became public
during the period or for which complete data were not available for any of

© Blackwell Publishers Ltd. 1995

1182 MICHELSON, JORDAN-WAGNER AND WOOTTON
the required variables were eliminated from the sample. The final sample of
firms consisted of 358 firms, before eliminating firms based on the specifications
of each of the individual models.
The procedure used to determine the presence of income smoothing is based
upon the method presented by Albrecht and Richardson (1990) in their study,
'Income Smoothing By Economy Sector.' Their study, in turn, applied the
coefficient of variation method developed by Eckel (1981) to measure income
and sales variability, A sufficient measure of income smoothing attempts to
separate the causes of income smoothing. If it is assumed, as many studies
do, that changes in sales are the results of real smoothing, then the coefficient
of variation can be used as a measure of the variability of sales and other
measures of income (e,g., operating income) (Eckel, 1981; Imhoff, 1981; Moses,
1987; and Albrecht and Richardson, 1990). The coefficient of variation is useful
because it is a dimensionless measure of the variability of a sample and it allows
for a comparison of variances between groups (Albrecht and Richardson, 1990,
p. 728), Also, since the coefficient of variation is a relative measure, it is useful
in comparing data sets that have different means and standard deviations
(Mason and Lind, 1993, p. 136). Using Albrecht and Richardson's (1990, p.
717) model, 'a firm is not classified as an income smoother if:'

where AI = one period change in income,

AS = one period change in sales,
CV = coefficient of variation
_.. Standard Deviation
Expected Value
The actual objectives of income smoothing with respect to the income statement
have been interpreted in various ways. Some studies (e.g,, Copeland, 1968)
have suggested net income as the ultimate aim of smoothing, while others have
proposed ordinary income (e.g., Ronen and Sadan, 1975). Imhoff s (1981)
possible measures of income smoothing included fully diluted EPS, net income,
net income before extraordinary items, operating income, and gross margin.
In this study, the four objectives of income smoothing are (1) operating income
after depreciation (OIADP), (2) pretax income (PI), (3) income before
extraordinary items (IB), and (4) net income (NI). We use the ratio of the
coefficient of variation of these four measures with respect to the coefficient
of variation in s£iles to identify our sample of firms as smoothers or non
smoothers for each income measure. We calculate these ratios based on twelve
years of data, from 1980 through 1991, to provide an adequate time series of
data to realistically identify firms that have been smoothing over a number
of years. This procedure is consistent with Moses' (1987, p. 362) suggestion
that multiperiod studies capture achievements of smoothing, whereas one period
studies reflect attempts to smooth.

© Blackwdl Publishers Ltd. 1995

 ANALYSIS OF INGOME SMOOTHING 1183
Daily returns, including distributions, are utilized to calculate ten-year
annualized returns for each stock. We use the daily returns over the time period
1982 through 1991 to capture any market effect of the smoothing measures.'
The daily return is the change in the total value of an investment in a common
stock over a one day trading period per dollar of initial investment. The daily
returns, provided by CRSP, are calculated as follows:

pj.t-i
where r^, = the daily stock market return, including distributions, earned by
company J during day t,
pj, = last sale price for company _;' during day t
pjt-i = last sale price for company^ during day t—l
dj, = distributions (cash dividends, capitcd adjustments, and other
distributions to shareholders) for company j during day t.
Logarithmic stock market returns are used to calculate geometric returns
from the daily return data over the ten-year sample period, as follows:
R^ = X;
(-I

ARj
where Rj = the logarithmic stock market return earned by company^' during
time period k, the ten-year sample period,
rj, = the daily stock market return, including distributions, earned by
company 7 during day t,
ARj = the annualized return for company j over the ten-year sample
period.
The methodology employed to test the hypothesis of a difference in mean
returns between smoothing and non smoothing firms involves a comparison
of the cross-sectional mean returns using the difference of means test to compute
the ^-statistics, as follows:

N^s + Ns- 2

where AR,^ = annualized return for the non smoothing firms,

ARs = annualized return for the smoothing firms,
A'^^ = number of non smoothing firms,
A/5 = number of smoothing firms,
SNS = standard deviation of annualized returns for non smoothing
firms,

© Blackwell Publishers Ltd. 1995

1184 MICHELSON, JORDAN-WAGNER AND WOOTTON
5y = standard deviation of annualized returns for smoothing
firms.
We use four models to determine if a firm is a smoother or a non smoother
utilizing our four smoothing variables. First, we take the absolute value of the
ratio of CV^ariabi/CV^ai^. Then, to eliminate bias in our sample due to outliers
(i.e. firms that have had exceptionally large increases or decreases in one of
the variables due to information events), we eliminate all firms that have a
ratio of greater than ten. Finally, we classify our firms into four models, as
follows:
Model 1 — Firms are classified as smoothers if any of the ratios of the four
variables (OIADP, PI, IB, or NI) are between zero and one.
Model 2 — This model eliminates firms that may be marginally classified as
smoothers by requiring more than one classification variable to
fit the smoothing definition. Firms are classified as smoothers if
at least three out of the four ratios of the variables (OIADP, PI,
IB, or NI) are between zero and one.
Model 3 — This model applies a stricter definition of smoothing than Model
2. Firms are classified as smoothers if at least three out of the
four ratios of the variables (OIADP, PI, IB, or NI) are between
zero and one, and additionally, firms with classification variables
between 0.9 and 1.1 are eliminated.
Model 4 — This model equEilizes the sample sizes of smoothers and non
smoothers. Firms are classified as smoothers if at least three out
of the four ratios of the variables (OIADP, PI, IB, or NI) are
between zero and one, additionally firms with classification
Vciriables between 0.9 and 1.1 are eliminated, and the sample sizes
are equalized by deleting the lower range of non smoothers. The
firms with the highest ratio of CV^/CV^, but still with a ratio
of less than ten, are selected as non smoothers to match the sample
of smoothers.*
Based on the above definitions of the four models, the sample sizes for each
model are as follows:
Model Total Firms Non Smoothers Smoothers
1 256 154 102
2 256 195 61
3 212 162 50
4 100 50 50

RESULTS
Table 1 presents the mean smoothing variables for the full sample and for each
of the first three models. The variables are fairly consistent between smoothers

© Blackwell Publishers Ltd. 1995

 ANALYSIS OF INCOME SMOOTHING 1185
Table 1
Mean Smoothing Variables for Models 1 Through 3 and Full Sample

Classification Mkt. Equity OIADP PI IB NI

(000)

Full Sample
Mean 6,169,969 -0,9138 2,1671 1.2402 -0,4123
Std, Dev, 10,999,735 75,9966 25.7043 39,8440 40,2325
Model 1
Non Smoother 6,083,419 1.4607 1,7246 1,5637 1,6864
Smoother 8,572,141 0,3738 0.5211 0,5404 0,7724
Model 2
Non Smoother 6,188,153 1,2672 1.5574 1,4368 1,6255
Smoother 9,910,084 0,2590 0,2313 0,2412 0,3235
Model 3
Non Smoother 6,375,301 1,3858 1,6524 1.4896 1,5542
Smoother 10,439,315 0,3194 0,2225 0,2444 0,3624
Note:
The variables presented in these tables utilize the coefficient of variation of the classification variable
divided by the coefficient of variation of sales. The classification variables are as follows: (1) operating
income after depreciation (OIADP), (2) pretax income (PI), (3) income before extraordinary items
(IB), and (4) net income (NI),

and non smoothers for each of the three models. There is some variation in
the smoothing variables for the full sample, which would be expected when
we combine smoothers and non smoothers. Additionally the full sample still
includes the outliers, i,e. firms with CV^CV^ ratios of greater than ten. This
table also presents data on the market value of equity for the full sample and
each of the individual models. The market value of equity is calculated by
multiplying the number of shares outstanding by the market price of the stock
at the end of our sample period December 1991, The market value of equity
gives a relative comparison of mean firm size between each of the individual
models and the full sample. It appears that there is little variation in firm size,
although smoothers are consistently larger than non smoothers and this tendency
increases from Model 1 to Model 3, This result will be discussed further in
a later section.
Table 2 presents the number and percentage of firms in each two digit
Standard Industrial Classification Code (SIC) for smoother and non smoother
firms based on Model 2,^ The breakdown between smoother and non
smoother firms appears to be fairly uniform and there does not appear to be
any major inconsistencies between the two groups. As expected, the largest
segment for both groups is manufacturing firms. Note that the mining industry
constitutes 18% ofthe smoother firms but only 6% ofthe non smoother firms.
Specifically, eight ofthe eleven smoothing mining firms are in the oil and gas
industry. These results are consistent with the findings of Jordan-Wagner and

© Blackwell Publishers Ltd. 1995

1186 MICHELSON, JORDAN-WAGNER AND WOOTTON

Table 2
Standard Industrial Classification Codes
Number of firms (percent) in eacb two digit SIC Code. Based on Model 2,
firms are classified as smootbers if tbree out of tbe four variables are between
0 and 1. Classification variables are OIADP, PI, IB, and NI.

Description Two Digit SIC Non Smoother Smoother

Code Number of Firms Number of Firms
(Percent of Non (Percent of
Smoothers) Smoothers)

Mining 10-14 11(6) 11 (18)

Construction 15-17 1(1) 1 (2)
Manufacturing 20-39 125 (64) 33 (54)
Transportation, 40-49 7(4) 7(12)
Communication,
etc.
Wholesale Trade 50-51 7(4) 1 (2)
Retail 52-59 17(9) 4(7)
Finance, 60-67 13(7) 3 (5)
Insurance, and
Real Estate
Services 70-89 14(7) 1 (2)
Total 195 (100)' 61 (100)'

NoU:
* Totals are not exactly 100% due to rounding.

Wootton (1993) tbat indicate evidence of income smootbing in tbe oil and gas
industry.
Cordon (1964, p. 262) suggested tbat stockbolders' satisfaction increased
witb tbe stability of a company's income. Beidleman (1973, p. 655) contended
tbat smootbing sbould widen tbe market for a company's sbares and bave a
favorable effect on sbares' value. In contrast. Lev and Kunitzky (1974, p. 268)
stated tbat it was not self evident tbat stockbolders prefer smootber income
streams. To test tbese bypotbeses, we compared tbe ten year annualized returns
for smootbing and non smootbing companies. Tables 3, 4, 5, and 6 present
tbe results of tbe difference of means test on tbe ten year annualized returns
for Models 1 tbrougb 4. Tbe mean annual return for tbe non smootbers varies
from 17.75% in Model 4 to 18.53% in Model 1. Tbe mean annual return for
tbe smootbers varies from 13.20% in Models 3 and 4 to 14.81% in Model
1. Tbe difference is significant at tbe 0.05 level for Models 1 and 4, and
significant at tbe 0.01 level for Models 2 and 3. Note tbat non smootbers bave
a significantly bigber mean annual return tban smootbers. Additionally, as
our definition of smootbing becomes more rigorous, tbe mean annual return
for smootbers decreases. Our findings suggest tbat investors do not give

© Blackwell Publishers Ltd. 1995

 ANALYSIS OF INCOME SMOOTHING 1187

Table 3
Model 1 — Difference of Means Test on Ten Year Annualized Return
Firms are classified as smoothers if any of the four variables are between 0
and 1. Classification variables are OIADP, PI, IB, and NI.

Classification No. of Firms Percent Mean Return Std. Deviation

Non Smoother 154 60 0,1853'' 0 ,1136

Smoother 102 40 0.148l'> 0 ,1298

Note:
*" Difference significant at the 0.05 level.

Table 4
Model 2 — Difference of Means Test on Ten Year Annualized Return
Firms are classified as smoothers if three out of the four variables are between
0 and 1. Classification variables are OIADP, PI, IB, and NI.

Classification No. of Firms Percent Mean Return Std. Deviation

Non Smoother 195 76 0.1821'= 0 ,1175

Smoother 61 24 0.1331^ 0 ,1274

Note:
*= Difference significantat the 0,01 level.

Table 5
Model 3 — Difference of Means Test on Ten Year Annualized Return
Firms are classified as smoothers if three out of the four variables are between
0 and 1, additionally firms with classification variables between 0.9 and 1.1
are eliminated. Classification variables are OIADP, PI, IB, and NI.

Classification No. of Firms Percent Mean Return Std. Deviation

Non Smoother 162 76 0,1797"= 0 ,1140

Smoother 50 24 0,1320'= 0 .1311

Note:
*= Difference significant at the 0,01 level.

© Blackwell Publishers Ltd. 1995

1188 MICHELSON, JORDAN-WAGNER AND WOOTTON

Table 6
Model 4 — Difference of Means Test on Ten Year Annualized Return
Firms are classified as smoothers if three out of the four variables are between
0 and 1, additionally firms with classification variables between 0.9 and 1.1
are eliminated, and the sample sizes are equalized by deleting the lower range
of non-smoothers. Classification variables are OIADP, PI, IB, and NI.

Classification No. of Firms Percent Mean Return Std. Deviation

Non Smoother 50 50 0.1775'' 0.0850

Smoother 50 50 0.1320'' 0.1311

Note:
'' Difference significant at the 0.05 level.

preference to smoother income streams and that smoothing does not increase
the market value of a firm.
Studies have suggested one reason for income smoothing is to reduce the
actual or perceived riskiness of a firm (e.g., Beidleman, 1973; and Lev and
Kunitzky, 1974). To test the hypothesis of a difference in riskiness between
smoothing and non smoothing firms, we examine the beta of the firms in our
sample. Tables 7, 8, 9, and 10 present the results of the difference of means
test on the beta of our sample of firms for Models 1 through 4. The beta of
the non smoothing firms varies from 1.1400 in Model 1 to 1.1569 in Model
4. The beta of the smoothing firms varies from 1.0660 in Models 3 and 4 to
1.0881 in Model 1. This difference is marginally significant (at the O.IO level)
for Model I. Betas for firms that smooth income appear to be lower than betas
for non smoothing firms. Additionally, as our definition of smoothing becomes
stricter, the mean beta for smoothers decreases. Even though we do not find
strong significance, the consistently lower betas for smoothing firms provides
additional corroborating evidence for our returns data. We find significantly

Table 7
Model 1 — Difference of Means Test on Beta
Firms are classified as smoothers if any of the four variables are between 0
and 1. Classification variables are OIADP, PI, IB, and NI.

Classification No. of Firms Percent Mean Beta Sid. Deviation

Non Smoother 154 60 1.1532" 0.2629

Smoother 102 40 1.088P 0.3021

Note:
" Difference significant at the 0.10 level.

© Blackwell Publishers Ltd. 1995

 ANALYSIS OF INCOME SMOOTHING 1189

Table 8
Model 2 — Difference of Means Test on Beta
Firms are classified as smoothers if three out of the four variables are between
0 and 1. Classification variables are OIADP, PI, IB, and NI.

Classification No. of Firms Percent Mean Beta Std. Deviation

Non Smoother 195 76 1.1400 0.2569

Smoother 61 24 1.0867 0.3432

Table 9
Model 3 — Difference of Means Test on Beta
Firms are classified as smoothers if three out of the four variables are between
0 and 1, additionally firms with classification variables between 0.9 and 1.1
are eliminated. Classification variables are OIADP, PI, IB, and NI.

Classification No. of Firms Percent Mean Beta Std. Deviation

Non Smoother 162 76 1.1506 0.2583

Smoother 50 24 1.0660 0.3426

Table 10
Model 4 — Difference of Means Test on Beta
Firms are classified as smoothers if three out of the four variables are between
0 and 1, additionally firms with classification variables between 0.9 and 1.1
are eliminated, and the sample sizes are equaJized by deleting the lower range
of non-smoothers. Classification variables are OIADP, PI, IB, and NI.

Classification No. of Firms Percent Mean Beta Std. Deviation

Non Smoother 50 50 1.1569 0.2256

Smoother 50 50 1.0660 0.3426

lower returns, as well as lower betas (risk) for smoothing firms.

It has been hypothesized that larger firms may have a greater incentive to
smooth income than smaller firms because larger firms are subject to greater
scrutiny by the government and general public (Moses, 1987). It has also been
hypothesized that larger firms may have a lower tendency to smooth income
than smaller firms because larger firms are analyzed and reviewed much more

© Blackwell Publishers Ltd. 1995

1190 MICHELSON, JORDAN-WAGNER AND WOOTTON

critically (Albrecht and Richardson, 1990), To test the hypothesis of a difference

in firm size between smoothing and non smoothing firms, we examine the
market value of equity ofthe firms in our sample.^ Tables 11, 12, 13 and 14
present the results of the difference of means test on market value of equity
for Models 1 through 4. The market value of equtiy for non smoothing firms

Table 11
Model 1 — Difference of Means Test on Market Value of Equity
Firms are classified as smoothers if three out of the four variables are between
0 and 1. Classification variables are OIADP, PI, IB, and NI.

Classification No. of Firms Percent Mkt. Equity Std. Deviation

(000)

Non Smoother 154 60 6,083,419 9,892,218

Smoother 102 40 8,572,141 15,068,027

Table 12
Model 2 — Difference of Means Test on Market Value of Equity
Firms are classified as smoothers if three out of the four variables are between
0 and 1, Classification variables are OIADP, PI, IB, and NI.

Classification No. of Firms Percent Mkt. Equity Std. Deviation

(000)

Non Smoother 195 76 6,188,153 9,590,667

Smoother 61 24 9,910,084 18,163,901

Table 13
Model 3 — Difference of Means Test on Market Value of Equity
Firms are classified as smoothers if three out of the four variables are between
0 and 1, additionally firms with classification variables between 0.9 and 1.1
are eliminated. Classification variables are OIADP, PI, IB, and NI.

Classification No. of Firms Percent Mkt. Equity Std. Deviation

(000)

Non Smoother 162 76 6,375,301 10,240,522

Smoother 50 24 10,439,315 18,836,076

© Blackwdl Publishers Ltd. 1995

 ANALYSIS OF INCOME SMOOTHING 1191

Table 14

Model 4 — Difference of Means Test on Market Value of Equity

Firms are classified as smoothers if three out of the four variables are between
0 and 1, additionally firms with classification variables between 0.9 and 1.1
are eliminated, and the sample sizes are equalized by deleting the lower range
of non-smoothers. Classification variables are OIADP, PI, IB, and NI.

Classification No. of Firms Percent Mkt. Equity Std. Deviation

(000)

Non Smoother 50 50 7,032,270 12,196,458

Smoother 50 50 10,439,315 18,836,076

varies from $6,083,419 in Model 1 to $7,032,270 in Model 4, (Note that the

market value of equity is presented in $1000's.) The market value of equity
for smoothing firms varies from $8,572,141 in Model 1 to $10,439,315 in
Models 3 and 4. The differences are not significant, but note that smoothing
firms are consistently larger in size than non smoothing firms. This tendency
increases from Model 1 to Model 3, as our definition of smoothing becomes
more stringent. These results are consistent with Moses (1987) and Albrecht
and Richardson (1990) that smoothing has a greater association with larger
companies than with smaller companies. This provides additional evidence to
our findings of lower returns, lower risk, and now, larger firms' sizes for
smoothing firms.

SUMMARY AND CONCLUSIONS

Previous research has investigated the effects of income smoothing in many
different ways. Studies have focused on issues such as the following: do firms
smooth income, the smoothing ability of different accounting techniques,
conditions under which smoothing is effective, the objectives of smoothing, the
objects of smoothing, the dimensions of smoothing, and smoothing variables.
Several studies have hypothesized that companies smooth to increase the nnarket
value of the firm or to reduce the actual or perceived riskiness of the firm.
However, to date no research has actually tested the reaction in the marketplace
to income smoothing. This paper investigated the relationship between income
smoothing and the return on a firm's common stock.
This study examined a sample of 358 firms from an initial sample of the
Standard & Poor's 500. We employed four models of income smoothing utilizing
the coefficient of variation of four smoothing variables: (1) operating income
after depreciation, (2) pretax income, (3) income before extraordinary items,
and (4) net income with respect to the coefficient in variation in sales. We found
that firms that smooth income have a significantly lower mean annualized return

© Blackwell Publishers Lid. 1995

1192 MICHELSON, JORDAN-WAGNER AND WOOTTON

than firms that do not smooth income. We also observe that smoothing firms
have lower betas and higher market value of equity. Additionally, as
corroborating evidence, these results become stronger from Model 1 to Model
4, as our definition of smoothing becomes more rigorous. We propose that these
results are reasonable and consistent when viewed together. We find lower
returns, lower risk, and larger firm sizes for smoothing firms. This indicates
that income smoothing lowers the actuzJ or perceived riskiness of the firm, which
in turn would lead to lower returns to those investing in the lower risk firms.
Additionally, the lower risk, more stable firms are generally identified as the
larger firms with more consistent earnings. Our findings, therefore, seem
appropriate when viewed in this context of primarily large firms with stable
net incomes having lower overall risk and return.

NOTES

We also examined the twelve year period from 1980 through 1991. The results for this period
are essentially the same as the ten year period.
This calculation of returns over extended time periods essentially follows the methodology of
Jeter and Chaney (1992), Strong (1992), Thome (1991) and O'Hanlon (1991).
This model is stricter in that we eliminate firms that are marginally classified as either smoothers
or non smoothers by deleting all firms near the cut-off point of 1.0. Therefore firms marginally
classified as smoothers (between 0.9 and 1.0) and marginally classified as non smoothers (between
1.0 and 1.1) are removed from the sample.
We develop this model in response to the problems created by having unequal sample sizes
between the two groups. This model allows us to equalize the sample sizes, while selecting
the firms that are more rigorously classified as non smoothers (i.e. the highest ratio of
/CV
Model 2 was selected for presentation of the SIC data to allow us to incorporate the full sample
(256 firms), but yet utilize the stricter definition of smoothing (between Model 1 and Model 2).
It may be noted that since size is proxied by the market vjJue of equity, there is a direct
relationship between size and the market price of the shares. Therefore, the firm size results
may be biased by this association. For example, as share price increases, there will be a
corresponding increase in firm size. In consideration of this potential problem, we also utilized
sales and total assets as alternative measures of firm size. The results, while not as strong, are
essentially the same as those presented. We thank an anonymous referee for the suggestion
and a detailed discussion of this issue.

REFERENCES

Albrecht, W.D. and F.M. Richardson (1990), 'Income Smoothing by Economy Sector',
of Business Finance & Accounting Vol. 17, No. 5 (Winter), pp. 713-730.
Barnea, A., J. Ronen and S. Sadan (1976), 'Classificatory Smoothing of Income with Extraordinary
I t e m s ' , The Accounting Review Qannary), pp. 110—122.
Beidleman, C.R. (1973), 'Income Smoothing: The Role of Management', The Accounting Review
(October), pp. 653-667.
Belkaoui, A. and R.D. Picur (1984), 'The Smoothing of Income Numbers: Some Empirical Evidence
on Systematic Differences Between Core and Periphery Industrial Sector', Journal of Business
Finance (^Accounting, Vol. U (Winter), pp. 527-545.
Copelamd, R . M . (1968), 'Income Smoothing', Journal of Accounting Research, Supplement to Volume
6, Empirical Research in Accounting: Selected Studies, Vol 6, p p . 101—116.

© Blackwell Publishers Ltd. 1995

 ANALYSIS OF INCOME SMOOTHING 1193

Dascher, P,E, and R,E, Malcom (1970), 'A Note on Income Smoothing in the Chemical Industry',
Joumat of Accounting Research ( A u t u m n ) , pp, 253—259,
Eckel, N, (1981) 'The Smoothing Hypothesis Revisited', Abacus CJune), pp, 28-40,
Gordon, M,J, (1964) 'Postulates, Principles and Research in Accounting', The Accounting Review
(April), pp, 251-263,
Hepworth, S,R, (1953), 'Smoothing Periodic Income', The Accounting Review Qanxiaxy), pp, 32—39,
Horwitz, B,N, (1977), 'Comments on Income Smoothing: A Review by J, Ronen, S, Sadan and
C, Snow', Accounting Joumal (Spring), pp, 27—29,
Imhoff, E,A,, Jr, (1981), 'Income Smoothing: An Analysis of Criticzil Issues', Quarterly Review
of Economics and Business ( A u t u m n ) , pp, 23—42,
(1977), 'Income Smoothing — A Case for Doubt', j4«oun/inf yourna/(Spring), pp, 85—101,
Jeter, D,C, and P,K, Chaney (1992), 'An Empirical Investigation of Factors Affecting the Earnings
Associa.tionCot{iicient', Joumat of Business Finance & Accounting, Vol, 19, No, 6 (November),
pp, 839-863,
Jordan-Wagner, J, and C,W, Wootton (1993), 'An Analysis of Earnings in Oil Related Industries',
Petroleum Accounting and Financial Management Joumal (Spring), pp, 110—123,
Lev, B, and S, Kunitzky (1974), 'On the Association Between Smoothing Measures and the Risk
of Common Stock', The Accounting Review (April), pp, 259—270,
Mason, R,D, and D,A, Lind (1993), Statistical Techniques in Business and Economics, 8th ed,
(Homewood, Irwin), pp, 136—137,
Moses, C D , (1987), 'Income Smoothing and Incentives: Empirical Tests Using Accounting
Changes', The Accounting Review (April), pp, 358—377,
O'Hanlon, J, (1991), 'The Relationship in Time Between Annual Accounting Returns and Annual
Stock Market Returns in the UK', Joumal of Business Finance (^Accounting, Vol, 18, No, 3
(April), pp, 305-314,
Ronen, J , and S, Sadan (1981), Smoothing Income Numbers, Objectives, Means, and Implications (Keading,
MA, Addison Wesley),
(1975), 'Classificatory Smoothing: Altemative Income Models',youma/ of Accounting Research
(Spring), pp, 133-149,
Strong, N, (1992) 'Modelling Abnormal Returns: A Review Article', youma/ of Business Finance
& Accounting, Vol, 19, No, 4 G"ne), pp, 533-553,
Thorne, D, (1991) 'The Information Content ofthe Trend Between Historic Cost Earnings and
Current Cost Earnings (United States of America)', yourna/ of Business Finance & Accounting,
Vol, 18, No, 3 (April), pp, 289-303,
Zmijewski, M,E, and R,L, Hagerman (1981), 'An Income Strategy Approach to the Positive Theory
of Accounting Standard Setting/Choice', Joumal of Accounting and Economics (August), pp,
129-149,

© Blackwell Publishers Ltd, 1995