ACCOUNTING CHOICE DECISIONS AND UNLEVERED FIRMS: FURTHER EVIDENCE ON DEBT/EQUITY HYPOTHESIS
Journal Of
Financial And Strategic Decisions
Volume 7 Number 3 Fall 1994
ACCOUNTING
CHOICE DECISIONS
AND
UNLEVERED FIRMS: FURTHER
EVIDENCE ON
DEBT/EQUITY HYPOTHESIS
V. Gopalakrishnan*
Abstract
This study
extends the accounting choice literature by empirically examining a set of
firms that do not have long-term debt (unlevered) in their capital
structure. Currently, evidence on how
these firms make accounting choice decisions is scarce. Empirical evidence on this issue is important
for two reasons. First, the case of
unlevered firms serves as an additional testing ground for the positive theory
of accounting choice and the findings are likely to enhance our understanding
of accounting choice decisions per se. Second, it offers light in the area of
generalizability of debt/equity and political cost hypotheses, particularly to
smaller firms. The results indicate that
unlevered firms tend to choose income-increasing accounting methods more than
their levered counterparts. This is
particularly true in the case of inventory method choice. It appears that even without the presence of
long-term debt, leverage, measured as total short-term liabilities over equity,
is a significant determinant of accounting choice. Finally, political cost hypothesis does not
seem to apply to smaller firms.
INTRODUCTION
A number of researchers have
attempted to model what motivates managers in choosing accounting methods
(Hagerman and Zmijewski [11]; Zmijewski and Hagerman [23]), why managers lobby
before standard setting agencies (Francis [7]), and why the stock market reacts
to mandated accounting changes (Leftwich [14]; Lys [16]).1 Within the accounting choice literature,
researchers have used the following perspectives to identify the determinants
of accounting choice: the opportunistic behavior, efficient contracting and
information perspectives (Holthausen [13]).
It has been well documented that 'leverage' and 'political visibility'
are important determinants of accounting choice.
However, the prior researchers in general examined firms
that were generally large and more importantly, levered. This study extends the accounting choice
literature by examining the depreciation and inventory method choices for a set
of firms that do not have long-term debt in their capital structure. Empirical evidence on this issue is important
for two reasons. First, the case of
unlevered firms serves as an additional testing ground for the positive theory
of accounting choice. The accounting choice literature concludes that managers
of levered firms are likely to adopt "asset-increasing" and
"income-increasing" accounting choices. One may argue that managers of levered and
unlevered firms do not share similar incentives to choose an income increasing
accounting method and therefore one would expect a different set of
determinants of accounting choice for unlevered firms. In short, this study is likely to enhance our
understanding of accounting choice decisions per se. Second, it offers light in the area of
generalizability of contracting and political cost hypotheses, particularly to
smaller firms. The prior researchers,
based on their analysis of mostly large firms found firm size, a proxy for
political visibility, to be negatively correlated with accounting choice. However, it is not clear whether for managers
of smaller firms,
*
George Mason
University
The author acknowledges the helpful comments of P. R. Chandy,
Joseph Cheung, Mohinder Parkash and participants at the 1992 Annual Meetings of
the Decision Sciences Institute held at San Francisco.
33
minimizing political
visibility is a concern. By focusing on
a set of smaller firms, this study examines the question whether the political
visibility argument explains the accounting choice decisions.
This paper is organized as follows. The next section presents the hypotheses,
develops the empirical models, describes the sample selection process and
offers evidence on the types of accounting-based covenants present in
short-term credit agreements. Section
three presents the results followed by conclusions.
HYPOTHESES, MODEL AND SAMPLE SELECTION
Hypotheses
Two hypotheses have been extensively
tested in the accounting choice literature (Watts and Zimmerman [20], Ch. 9 and
10). They are (1) 'debt/equity
hypothesis' : ceteris paribus, the larger a firm's debt/equity ratio, the more
likely the firm's manager is to select accounting procedures that shift reported
earnings from future periods to the current period; and (2) 'size hypothesis'
(also known as political cost hypothesis) : ceteris paribus, the larger the
firm, the more likely the manager is to choose accounting procedures that defer
reported earnings from current to future periods.
Consistent with the extant literature, I postulate the
following four hypotheses to examine how unlevered firms choose their
depreciation and inventory accounting methods.
Hypothesis 1:
Ceteris paribus, the higher
the leverage, the greater the likelihood that a firm will choose straight-line
depreciation method.
Hypothesis 2:
Ceteris paribus, the higher
the leverage, the greater the likelihood that a firm will choose FIFO inventory
method.
Here, leverage is defined as total current liabilities
divided by market value of equity. If
short-term liabilities were to influence the choice of depreciation and
inventory methods, one would expect a positive association between leverage and
an income-increasing accounting method.
Hypothesis 3:
Ceteris paribus, the larger
the firm, the lesser the likelihood, it will choose straight-line depreciation
method.
Hypothesis 4:
Ceteris paribus, the larger
the firm, the lesser the likelihood, it will choose FIFO inventory method.
Hypotheses 3 and 4 test whether or not the political cost
hypothesis apply to unlevered firms, which are generally smaller than their
levered counterparts. Descriptive
statistics for the sample firms are presented in a later section.
Model
I propose the following logit models to
test hypothesis 1 through 4:2
Equation 1 log[Psl,i/(1-Psl,i)] = b0
+ b1SIZEi + b2LEVERAGEi + b3PROFITi where:
Psl,i
|
=
|
Probability firm i will choose the Straight-Line
(SL) method of depreciation
|
SIZE
|
=
|
Log of total sales
|
LEVERAGE
|
=
|
Total current liabilities
divided by market value of equity
|
PROFIT
|
=
|
Operating income before depreciation and exclusive of
changes in LIFO reserve divided by market value of equity and total current
liabilities
|
Here, SIZE
and LEVERAGE are the variables of
interest. Both leverage and size are
generally found significant in studies of accounting choice.3 However, Lilien, Mellman and Pastena [15]
provide evidence that unsuccessful firms are more likely to choose income
increasing accounting procedures than successful firms. Thus, PROFIT
is included as a control variable. Thus,
consistent with the extant literature, both b1 and b3 are
expected to be < 0 and b2 > 0.
Model (1) will be estimated to test hypotheses 1 and 3 that
deal with the choice of depreciation methods.
To test the remaining two hypotheses (2 and 4), the following model will
be estimated:
Equation 2 log[Pfifo,i/(1-Pfifo,i)] = b0 + b1SIZEi + b2LEVERAGEi + b3PROFITi + b4TAXRATEi
where Pfifo,i
is the probability firm i
will choose the FIFO (First-In-First-Out) method and TAXRATE is the firm's tax rate calculated by dividing tax
expense by net income. Since inventory
method choices could be subject to tax incentives, TAXRATE is included as a second control variable. Once again, b1, b3 and
b4 are expected to be < 0 and b2 > 0.
Sample Selection
To identify firms that do not have any long-term debt in
their capital structure, I first searched the Compustat database for firms with a long-term debt of zero.4 Second, I excluded regulated and non-service
firms from the sample.5
Third, I excluded firms that did not report depreciation method or did
not have inventory and firms that used methods other than LIFO or FIFO. Firms for which data on independent
variables, particularly, closing market price and operating income before
depreciation were not available were deleted.
I repeated these steps for the years 1983 through 1987. My objective was to obtain a sufficient set
of firms, for a consecutive five year period to estimate the logit model. The final sample include 727 and 690 firms
reporting depreciation and inventory method choices, respectively. Table 1 contains the descriptive statistics
for the sample on size measure, leverage and profitability ratios and dependent
variables.
TABLE 1
Descriptive Statistics
For A Sample Of 727 Non-Regulated, Non-Service
And Unlevered Firms
For The Years 1983 Through 1987
|
MEAN
|
S.D.
|
MIN
|
MAX
|
LEVERAGE
|
0.32
|
0.81
|
0.00
|
10.79
|
PROFIT
|
0.02
|
0.16
|
-1.74
|
1.00
|
SIZE1
|
2.49
|
2.25
|
-6.91
|
7.89
|
TABLE 1
Descriptive Statistics
For A Sample Of 727 Non-Regulated, Non-Service
And Unlevered Firms
For The Years 1983 Through 1987
(CONT’D)
|
MEAN
|
S.D.
|
MIN
|
MAX
|
SIZE2
|
2.64
|
1.66
|
-1.22
|
7.39
|
TAXRATE
|
0.31
|
1.11
|
0.00
|
0.83
|
DEPRECIATION
|
0.84
|
0.35
|
0
|
1
|
INVENTORY
|
0.85
|
0.36
|
0
|
1
|
MVE$
|
97.51
|
332.40
|
0.15
|
7123.97
|
BVE$
|
38.20
|
94.52
|
-266.06
|
1073.70
|
SALES$
|
90.03
|
253.32
|
0.01
|
2661.07
|
ASSETS$
|
59.26
|
154.56
|
0.29
|
1624.90
|
CURRENT
LIABILITIES$
|
13.44
|
42.30
|
0.02
|
551.20
|
LEVERAGE = total current liabilities divided
by market value of equity;
PROFIT = operating
income before depreciation and exclusive of changes in LIFO reserve divided by
market value of equity and total current liabilities;
SIZE1 = log of
sales;
SIZE2 = log of
total assets;
TAXRATE = tax expense divided by net
income;
DEPRECIATION = coded as 1 for straight-line and 0 for accelerated
methods; INVENTORY = coded as 1
for FIFO and 0 for LIFO.
MVE = market value of equity
BVE = book value of equity
$ in
millions.
The mean for sales and market value
of equity was about $90 million and $98 million respectively. The mean current liabilities over the same
period was about $13 million. It appears
that the sample firms are much smaller than firms used in prior studies. For example, the mean sales in Hagerman and
Zmijewski [11] was $1,760 million.
Similarly, the mean market value of equity and sales in Press and
Weintrop [18] was $775 million and $1,911 million, respectively. The accounting choice literature indicates
(Watts and Zimmerman [19]) that the political cost hypothesis is generally
applicable to large firms like the ones used in Watts and Zimmerman [19] and in
subsequent studies. The descriptive
statistics in Table 1 underscores the differences in firm size between the
sample firms and firms examined in prior studies. This difference allows us to test whether or
not the political cost hypothesis apply to the sample firms.
Table 1 shows that the sample firms were barely making
profits. The mean return on investments
was about 2%. Given this low
profitability ratio, it is likely that profitability could influence the choice
of an accounting method. Another note
worthy finding from Table 1 is the tendency of the sample firms to choose
incomeincreasing accounting methods, particularly, FIFO inventory method. Table 2 reports the distribution of
depreciation and inventory choices for the sample firms.
TABLE 2
Distribution Of
Depreciation And Inventory Accounting
Policy Choices For The
Sample Firms During 1983-1987
|
Depreciation
|
Inventory
|
Income-Increasing1
|
611
|
587
|
Income-Decreasing2
|
116
|
103
|
|
_____
|
_____
|
Total Number Of Firms
|
727
|
690
|
|
_____
|
_____
|
Percent
|
84 85
|
1 Income-Increasing = Straight-Line Method
(Depreciation) And FIFO (Inventory).
2 Income-Decreasing = Accelerated Method
(Depreciation) And LIFO (Inventory).
Percent = Percentage Of
Firms Choosing An Income-Increasing Method (SL/FIFO).
The percentage of sample firms choosing straight-line
depreciation method during was 84%. This
is consistent with evidence reported in Hagerman and Zmijewski [11] and Press
and Weintrop [18] for levered firms.
However, there is a significant difference between unlevered firms and
levered firms in the case of chosen inventory methods. The percentage of firms
choosing FIFO over LIFO was 50% in Hagerman and Zmijewski [11] and 41% in Press
and Weintrop [18]. For the sample firms,
the mean percentage of firms choosing FIFO was 85%. This is interesting given that for smaller
firms, where maximizing cash flow is likely to be an important objective, tax
based incentives generally lead to a preference in favor of LIFO. An empirical test of whether or not taxes influence
inventory method choice is presented in a later section.
Presence Of Restrictive Covenants For
Sample Firms
To find out whether or not the sample firms are subject to
restrictive covenants as part of short-term debt agreements, I searched the
annual reports to locate the nature and extent of binding covenants if any.6 Generally, I found that firms disclose the
existence of covenants either in management discussion and analysis section or
in footnotes to the financial statements section. A description of the identified covenants
appears in Table 3.
TABLE 3
A Description Of
Restrictive Covenants Found For
A Sample Of 31
'All-Equity' Firms1
|
|
Number
|
|
Type Of Covenant
|
Of Firms
|
1.
|
Maintenance of a minimum
tangible net worth.
|
17
|
2.
|
Maintenance of a minimum
working capital.
|
14
|
3.
|
Total number of firms where the only information
reported was "...the agreement requires the company to maintain certain
financial ratios".
|
14
|
TABLE 3
A Description Of
Restrictive Covenants Found For
A Sample Of 31
'All-Equity' Firms1
(CONT’D)
|
|
Number
|
|
Type Of Covenant
|
Of Firms
|
4.
|
Prohibition or restriction
on payment of cash dividends.
|
13
|
5.
|
Ceilings on total
liabilities to equity ratio.
|
9
|
6.
|
Maintenance of a minimum
current ratio.
|
5
|
7.
|
Restrictions on investments
and acquisitions.
|
4
|
8.
|
Restrictions on incurring
additional indebtedness.
|
3
|
9.
|
Prohibition or restriction on purchasing, redeeming
or retiring any capital stock.
|
2
|
10.
|
Maintenance of a minimum
profitability ratio.
|
2
|
11.
|
Restrictions on incurring
additional capital expenditures.
|
2
|
12.
|
Restrictions on the ability of the firm to encumber
its assets or engage in certain transactions outside the normal course of
business.
|
2
|
13.
|
Restrictions on pledging of
certain assets.
|
2
|
14.
|
Maintenance of a minimum
quick ratio.
|
1
|
15.
|
Maintenance of cash flow.
|
1
|
1. Source:
Company Annual Reports
The fifteen types of covenants identified for a sub-sample
of 31 firms relate exclusively to short-term obligations. Typically, these covenants were part of the
short-term line-of-credit agreements.
These covenants are very similar to covenants identified for levered
firms (Frost and Bernard [8]; Healy and Palepu [12]; and Press and Weintrop
[18]). The most common covenant was
maintenance of a minimum dollar amount of tangible net worth. I also find that
the levels and ratios expressed in the covenants, typically used accounting
numbers. However, this sub-sample of 31
firms may or may not be representative of the entire sample. At the least, the above evidence suggests
that the managers of unlevered firms and levered firms face similar incentives
to choose accounting methods that relax covenants found in the lending
agreements.
Univariate Analyses
Tables 4 and 5 present the results of univariate
analyses. Spearman rank correlations
among the dependent and independent variables are contained in Table 4.
It appears that both PROFIT
and SIZE seem to be associated with the
choice of depreciation methods. The
correlation between DEPRECIATION
and SIZE is negative as predicted and
significant at the 0.01 level.
Similarly, PROFIT is
negatively correlated with DEPRECIATION
as expected and the correlation is significant at the 0.05 level. As expected, LEVERAGE and DEPRECIATION are positively correlated but the correlation is
not statistically significant.
In the case of inventory method choice, all the four predictor
variables LEVERAGE, PROFIT and SIZE
are correlated with INVENTORY at the
0.01 level. As hypothesized earlier, PROFIT, SIZE and TAXRATE
are negatively correlated with INVENTORY. Similarly, as predicted, LEVERAGE is positively correlated with INVENTORY. Thus, in summary,
the correlation analysis offers support that both SIZE and PROFIT
drive the choice of both depreciation and inventory method choices. On the other hand, LEVERAGE appears to drive only the choice of inventory method
not the depreciation. Overall, the
correlation analysis lend a strong support in favor of the political cost
hypothesis and a weak support for the debt/equity hypothesis. Also, there is no evidence of serious
multicollinearity among the independent variables.
TABLE 4
Correlation Matrix[1]
Panel A: Depreciation Choice
|
DEPRECIATION
|
LEVERAGE
|
PROFIT
|
SIZE1
|
DEPRECIATION
|
1.00
|
0.07
|
-0.10**
|
-0.16***
|
LEVERAGE
|
|
1.00
|
0.22***
|
0.37***
|
PROFIT
|
|
|
1.00
|
0.64***
|
SIZE1
|
|
|
|
1.00
|
Panel B: Inventory Choice
|
INVENTORY
|
LEVERAGE
|
PROFIT
|
SIZE1
|
TAXRATE
|
INVENTORY
|
1.00
|
0.18***
|
-0.28***
|
-0.48***
|
-0.26***
|
LEVERAGE
|
|
1.00
|
0.20***
|
0.36***
|
0.16***
|
PROFIT
|
|
|
1.00
|
0.65***
|
0.60***
|
SIZE1
|
|
|
|
1.00
|
0.60***
|
TAXRATE
|
|
|
|
|
1.00
|
|
DEPRECIATION = coded as 1 for straight-line and 0 for accelerated
methods;
INVENTORY
|
= coded as 1
for FIFO and 0 for LIFO;
|
LEVERAGE
|
= total
current liabilities divided by market value of equity;
|
PROFIT
|
= operating
income before depreciation and exclusive of changes in LIFO;reserve divided
by market value of equity and total current liabilities;
|
SIZE1
|
= log of
sales;
|
TAXRATE
|
= tax
expense divided by net income.
|
TABLE 5
Results Of Univariate
Analysis Of Firms That Made
Income-Increasing And
Income-Decreasing Accounting Choices During 1983-1987
Panel A: Depreciation Choice
|
|
MEAN
|
|
|
MEDIAN
|
|
SL
|
ACC
|
t-Stat
|
SL
|
ACC
|
Chi-Sq
|
|
LEVERAGE
|
0.54
|
0.28
|
1.72*
|
0.12
|
0.11
|
0.04
|
SIZE1
|
2.35
|
3.47
|
4.43***
|
2.46
|
3.24
|
6.43***
|
PROFIT
|
0.02
|
0.06
|
1.96*
|
0.05
|
0.07
|
7.54***
|
Panel B: Inventory Choice
|
|
MEAN
|
|
|
MEDIAN
|
|
FIFO
|
LIFO
|
t-Stat
|
FIFO
|
LIFO
|
Chi-Sq
|
|
LEVERAGE
|
0.47
|
0.27
|
1.69*
|
0.18
|
0.09
|
15.7***
|
SIZE1
|
1.96
|
5.06
|
18.74***
|
2.10
|
5.07
|
89.3***
|
PROFIT
|
0.01
|
0.11
|
6.79***
|
0.03
|
0.11
|
52.6***
|
TAXRATE
|
0.24
|
0.37
|
3.24***
|
0.31
|
0.45
|
49.5***
|
SL and ACC
are straight-line and accelerated depreciation methods, respectively.
*, ** and
*** denote statistical significance in two-tailed tests, at the 0.10, 0.05 and
0.01 levels, respectively. The
t-statistic is for the significance of the differences between the means of the
two groups. The chi-square test is for
the equality of the median of the two groups.
Table 5 reports the results of more univariate analysis of
depreciation and inventory method choices made by unlevered firms. These results, like correlation analysis,
indicate how the predictor variables when considered individually, are related
to accounting choice. For each
accounting choice, the sample firms are partitioned into income-increasing
(straight- line or first-in-first-out) and income-decreasing (accelerated or
last-in-first-out) methods.7
I then applied univariate tests (two-sample t and median tests) to
examine whether there is a systematic difference between firms that chose
income-increasing and income-decreasing methods with respect to LEVERAGE, SIZE1, PROFIT
and TAXRATE.
For the depreciation method choice, the findings are
consistent with the size hypothesis documented in prior studies of accounting
choice. Firms that chose accelerated
depreciation methods were larger than those that chose straight-line
depreciation. In other words, larger
firms, as expected, chose income-decreasing accounting methods compared to
smaller firms. The differences in SIZE1 are statistically significant
at the 0.01 level. Consistent with the
debt/equity hypothesis, the mean LEVERAGE
for firms choosing straight-line depreciation are higher compared to firms
choosing accelerated methods. These
differences are significant at the 0.10 level.
However, the differences in the median LEVERAGE are not significant at the 0.10 level even though
the median LEVERAGE for firms that chose
straight-line depreciation is higher.
Finally, both the mean and median PROFIT
for the firms that chose straight-line depreciation is lesser than firms that
chose accelerated methods. This is
consistent with Lilien, Mellman and Pastena [15] that poor performers are more
likely to choose income increasing methods than successful performers.
The findings regarding inventory method choice is
supportive of both the size and debt/equity hypotheses. Both the mean and the median LEVERAGE for firms that chose FIFO is
higher compared to firms that chose LIFO and these differences are
statistically significant. Similarly,
LIFO firms are larger than FIFO firms and these differences are significant at
the 0.01 level. The mean and median PROFIT is higher for LIFO firms and
this is consistent with the notion that PROFIT
influences inventory method choice.
Finally, there is a significant difference (at the 0.01 level) in the
tax rates of LIFO firms compared to FIFO firms.
As expected, both the mean and median tax rates are higher for LIFO
firms.
Overall, the results presented in Table 5 are consistent
with the extant literature that both leverage and size are determinants of
accounting choice. The results of the
multivariate models are discussed in the next section.
RESULTS OF MULTIVARIATE ANALYSES
The empirical results of models (1) and (2) are contained
in Table 6. I discuss the results
pertaining to depreciation method choice first.
TABLE 6
Results Of The Logit
Models Of Depreciation
And Inventory Method
Choice1
Variable
& Predicted Sign
|
Depreciation
|
Inventory
|
INTERCEPT
(?)
|
-2.59
(0.000)
|
-5.87
(0.000)
|
LEVERAGE (+)
|
0.25
(0.013)
|
0.40
(0.014)
|
SIZE1 (-)
|
0.25
(0.000)
|
1.05
(0.000)
|
PROFIT (-)
|
0.03
(0.961)
|
0.67
(0.626)
|
TAXRATE (-)
|
----
|
-0.03
(0.937)
|
-2*log
L
|
573.47
|
346.58
|
Model
Chi-Square
|
30.54
|
161.66
|
df
|
3
|
4
|
TABLE 6
Results Of The Logit
Models Of Depreciation
And Inventory Method
Choice1
(CONT’D)
p-Value
|
0.0001
|
0.0001
|
Correctly
Classified2
|
86%
|
90%
|
Sample
Size
|
727
|
690
|
LEVERAGE = total current liabilities divided by market value of
equity;
PROFIT = operating income before depreciation and exclusive of
changes in LIFO; reserve divided by market value of equity and total current
liabilities;
SIZE1 = log of
sales;
TAXRATE = tax
expense divided by net income.
1 Dependent
variable is coded as 1 for income-increasing methods (SL or FIFO) and 0 for
income-decreasing methods (accelerated or LIFO). A positive coefficient indicates a higher
probability of choosing an income-increasing method.
2 A naive
prediction of 84% (85% for inventory) can be achieved based on the proportion
of the more commonly selected alternative.
p-Values in parentheses.
Depreciation Method Choice
Several noteworthy points emerge from the findings. First, the chi-square statistic of 30.54
relating to the joint significance of the three explanatory variables in model
(1) is highly significant (at the 0.0001 level). The value of chi-square statistic reported in
Table 6 is higher than values reported in prior research. Second, the overall predictability of the
model is high. In prior studies of
accounting choice (Zmijewski and Hagerman [23], Elliot and Kennedy [6] and
Press and Weintrop [18]), the classification rate ranged from 33% to about 73%.8 Here, for depreciation method choice, the
classification rate is 86% which is higher than a naive prediction rate of 84%
based on the proportion of the more commonly selected alternative as shown in
Table 6. Both in models (1) and (2), LEVERAGE and SIZE are the variables of interest. Results of univariate analyses presented in
Tables 4 and 5 indicate that both PROFIT
and TAXRATE (for inventory) could
influence accounting choice. Therefore,
these are included as control variables in the logit models. If LEVERAGE
and SIZE are key determinants of
accounting choice, then both of them should be statistically significant after
controlling for PROFIT and TAXRATE (for inventory). Of the explanatory variables, only LEVERAGE coefficient has the
predicted sign. LEVERAGE is positive as hypothesized and is highly
significant (at less than 0.02 level). SIZE coefficient is positive and is
significant at the 0.001 level. The
coefficient for PROFIT is close
to zero and is statistically insignificant.
Finally, the intercept term is significant, suggesting possible
measurement error in the model. This is
consistent with the prior research. Hagerman and Zmijewski [11], Zmijewski and Hagerman [23]
and Press and Weintrop [18] all report significant intercept terms.
Overall, the above evidence offers
support for the hypothesis that the higher the leverage, the greater the
likelihood that a firm will choose straight-line depreciation method. This finding is consistent with the extant
literature on accounting choices. However, the interesting about this finding is
the role of leverage as a significant determinant of accounting choice even in
the extreme case of firms with no long-term debt. In other words, the above findings along with
the findings documented in the extant literature may be interpreted as
follows. As long as there are
constraints present in the lending agreements, whether short-term or long-term,
managers are likely to relax the tightness of those constraints by choosing
income-increasing accounting methods.
The above findings do not support the
hypothesis that the larger the firm, the lesser the likelihood, it will choose
straight-line depreciation method. On
the contrary, it appears that, larger the firm, the greater the likelihood of
adopting straight-line depreciation method.
This result is not surprising for the following reasons. First, Watts
and Zimmerman [19] posit that political costs which are operationalized through
the SIZE variable, are important only for
very large firms. In other firms, they
argue that there is a threshold effect.
Minimizing political visibility is not likely to be a major concern for
smaller firms compared to larger firms.
Second, empirical evidence based on a sub-sample of low
political cost firms, presented in Zmijewski and Hagerman [23] (Table 5) does
not offer support for the political cost hypothesis. In Zmijewski and Hagerman, the SIZE coefficient was positive and not
significant. Similarly, Bowen, Noreen
and Lacey [3] found that outside the oil and gas industry, larger firms were
more likely to capitalize interest, an
income-increasing policy. More evidence
of a positive relationship between firm size and accounting choice is found in
Gray [10]. Gray reports that 72% of the
largest industrial and commercial banks examined chose to adopt SFAS No. 8 in
1981 which increased reported earnings.9
Third, Ball and Foster [2] point out that the political
sensitivity is not the only factor for which SIZE could proxy. They
suggest that SIZE could
represent the competitive disadvantages of disclosure, information production
costs and management ability and advice.
The information production costs are likely to be important, particularly
for smaller firms and therefore smaller firms, may be motivated to choose
accounting methods that are inexpensive (such as straight-line depreciation) to
implement. Therefore, to the extent the SIZE variable captures the
information production costs, for smaller firms, the SIZE variable is likely to be positively related to an
inexpensive accounting method.
Inventory Method Choice
Results of model (2) concerning inventory method choice are
also contained in Table 6. Once again,
the chisquare statistic relating to the joint significance of the four
predictor variables is highly significant (at the 0.0001 level). The chi-square statistic values are much
higher for inventory choice model compared to depreciation method choice. The predictability rate of the model was
90%. This compares favorably to a naive
prediction rate of 85% (587 firms out of 690 chose FIFO) and 58% reported in
Hagerman and Zmijewski [11]. Once again,
the coefficient for LEVERAGE
has the predicted sign and is highly significant.
SIZE
is positive and remains highly significant.
This is consistent with the results reported earlier for depreciation
choice. To the extent the SIZE represents information
production costs, for smaller firms, the SIZE variable
is likely to be positively related to an inexpensive accounting method such as
FIFO. The administrative burden
associated with the adoption of LIFO is well-known. If a firm maintains its inventory records
based on FIFO method and if it wants to reduce taxes by adopting LIFO, then
under the LIFO conformity requirement, it must also use LIFO for external
reporting. Compliance with this
requirement generally increases book-keeping cost. Based on a survey of chief financial officers
of firms that did not use LIFO, Granof and Short [9] found that 20% of the
respondents cited the presence of high administrative cost for rejecting LIFO
in favor of FIFO.
The PROFIT variable is positive and is not significant at the
0.10 level. The second control variable,
TAXRATE is negative as predicted and
is not statistically significant.
Overall, results reported in Table 6 offer support for the hypothesis
that higher the leverage, the greater the likelihood that a firm will choose
FIFO inventory method. This is
consistent with the results based on model (1) reported earlier. However, the findings do not support the
hypothesis that larger the firm, the lesser the likelihood, it will choose FIFO
inventory method.
The above findings are also
consistent with evidence reported in Agrawal and Nagarajan [1]. Based on an analysis of a sample of unlevered
firms, Agrawal and Nagarajan report that unlevered firms maintain a large
cushion of liquid assets and generally appear to be averse to debt, both
long-term and short-term. They conclude
that the decision to shun long-term debt is consistent with the managerial
concern for default risk.
In summary, the following two conclusions emerge based on
the results contained in Table 6. First,
consistent with the debt/equity hypothesis, it appears that depreciation and
inventory methods choice decisions made by unlevered firms are driven by
leverage. Results based on the
multivariate models after controlling for profitability and tax strongly
support the notion that higher the leverage, the greater the likelihood of
choosing straight-line depreciation and or FIFO. Second, the so called political cost
hypothesis documented in extant literature does not seem to apply to smaller
firms examined in this study.
Alternate Specifications
It is possible that the empirical results presented in
Table 6 could be subject to specification error particularly with respect to SIZE1. I estimated models (1) and (2) again with an
alternative proxy, SIZE2 (log of
total assets). This measure of size has been used in Hagerman and Zmijewski
[11]. Findings based on alternate
specifications confirm my earlier findings and fully supports the two
conclusions reached earlier. The
classification rates remain unchanged and the overall significance of the
models are much stronger. For both
depreciation and inventory choice, the chi-square values were higher than
comparable values reported in Table 6. I
also estimated models (1) and (2) with an alternate specification for LEVERAGE1. Consistent with Chow [4] and Press and
Weintrop [18], I defined LEVERAGE2
as total current liabilities divided by market value of equity plus total
current liabilities. These extensions
did not change the conclusions reached earlier.10
As an additional analysis, I also examined the accounting
choice decisions of the sample firms when they acquire long-term debt. Particularly, I was examining whether firms
that used an income-decreasing accounting method (LIFO or accelerated
depreciation methods) change to an income-increasing method after acquiring
longterm debt. I found that of the few
firms that did acquire long-term debt, there was no change in their accounting
choice decisions, because these firms already chose an income-increasing method
prior to acquiring long-term debt. One
possible explanation for this could be that the firms planning to seek
long-term financing prepare themselves by adopting income-increasing accounting
methods, to mitigate contracting costs imposed by debt covenants.
SUMMARY AND CONCLUSIONS
A plethora of research studies
emerged during the eighties as a result of a seminal work by Watts and
Zimmerman [19]. Of particular interest
to the researchers was the linkage between debt covenants and accounting
choice. Debt covenants typically
restrict the ability of the managers to engage in certain kinds of activities
and often impose a variety of costs on the firm. Therefore, managers often tend to choose
accounting methods that moves the firm away from the restrictions imposed by
the debt covenants. It has been well
documented in the accounting choice literature that leverage and political
costs are key determinants of accounting choice. However, the above evidence was essentially
based on firms carrying long-term debt in their capital structure.
This study extends the accounting
choice literature by examining the depreciation and inventory method choices
for a sample of unlevered firms. An
examination of the accounting choice decisions made by the unlevered firms is
motivated by the opportunity to test a well established theory in an extreme
case. The findings lead to the following
conclusions. First, it appears that
unlevered firms tend to choose income-increasing accounting methods more than
their levered counterparts. This
particularly true in the case of inventory method choice.
Second, an analysis of actual debt
covenants for a small sub-sample reveals the presence of a variety of
affirmative and negative covenants similar to the ones documented by prior
researchers for levered firms. This
suggests that the managers of unlevered firms and levered firms face similar
incentives to choose accounting methods to mitigate costs imposed by
restrictive covenants.
Third, it appears that, even without the presence of
long-term debt, leverage, measured as total short-term liabilities over equity,
is a determinant of accounting choice.
This is consistent with the debt/equity hypothesis documented in the
extant literature.
Fourth, political cost hypothesis does not seem to apply to
smaller firms. The coefficient for the SIZE variable, a proxy for political
visibility, was consistently positive.
In other words, the findings confirm the presence of a threshold effect
discussed in the literature (Zimmerman [22]).
To the extent, the SIZE
variable proxy for information production cost, it appears that there is a
positive association between firm size and accounting choice. This is
consistent with the notion that managers, particularly in very small firms, to
keep information production costs low, are likely to adopt accounting methods
that are 'inexpensive' (such as straight-line method or FIFO) to implement.
Fifth, LEVERAGE
and SIZE factors influence accounting
choice decisions even after controlling for tax and profitability factors. There is no evidence of managers being
influenced by profitability and tax factors in setting accounting
policies. This is surprising given the
evidence in Lilien, Mellman and Pastena [15] and the fact the sample firms were
barely making profits during the time period studied.
The case of unlevered firms holds promise in further
enhancing our understanding of how and why firms make accounting choice
decisions. For example, Watts and
Zimmerman [21] suggest that accounting choice decisions could be correlated
with investment opportunity set, financial policy and organizational
structure. Thus, a comparative study of
accounting choices made by both levered and unlevered firms could throw light
on the linkage between capital structure choice and accounting choice.
ENDNOTES
1. See
Watts and Zimmerman [20], [21] for a review.
2. Prior
researchers have used both probit (Zmijewski and Hagerman [23] and Press and
Weintrop [18]) and logit (Malmquist [17]) models. However, Elliot and Kennedy [6] while
replicating Zmijewski and Hagerman [23] found that logit model generally
outperformed probit model.
3. For
example, Zmijewski and Hagerman [23] report that in addition to leverage and
size, management compensation and concentration ratio were significant in
explaining accounting choices. However,
subsequent researchers (Press and Weintrop [18]) have not been successful in
replicating their results.
4. The
Compustat CD-ROM defines ‘long-term’
debt to include capitalized lease obligations.
I reviewed the annual reports of more than 100 of the sample firms to
confirm the absence of long-term debt.
5. This
criterion excluded firms with the following four digit SIC codes: > 6000
(financial, insurance, real estate, hospitals and services; these firms are not
likely to have inventory or incur significant depreciation cost) 4811, 4890,
4911, 4922-4924, 4931 and 4932 (communications, gas and electric utilities.)
6. I
also searched Moody’s Industrial
manuals to locate covenants but I did not find any covenants on short-term obligations.
7. My
categorization of straight-line depreciation and FIFO as income-increasing and
accelerated depreciation and LIFO as income-decreasing is consistent with the
prior research (Hagerman and Zmijewski [11]; Zmijewski and Hagerman [23]; and
Press and Weintrop [18]) and affords a direct comparison of my findings with
the extant literature. However, it is
important to recognize that this categorization is at best a
simplification. Although, LIFO is
generally income-decreasing if inventory prices and quantities are rising, it
is not income decreasing if inventory prices are falling. For example, some companies in the
electronics industry have experienced declining costs. For these companies, using FIFO is considered
more conservative than using LIFO.
However, without additional firmspecific information it is difficult to
fine tune this categorization.
8. In
Hagerman and Zmijewski [10], the classification rate for depreciation was
85.33%. But based on a naive prediction
rule, one would expect 85% of the firms to choose straight-line
depreciation, For example, 255 (85%) out
of 300 firms chose straight-line depreciation.
9. For
1981, firms could select either SFAS No. 8 or SFAS No.52. Under SFAS No. 8, translation gains and
losses are included in net income. On
the other hand, SFAS No.52 requires these gains and losses to be taken directly
to the balance sheet.
10. For
model (1) (depreciation), -2*log L and the chi-square statistic were 566.75 and
39.04 (significant at 0.0001), respectively.
Both LEVERAGE and SIZE were positive and significant at
the 0.02 level. PROFIT was not significant. For model (2), -2*log L and the
chi-square values were 422.70 and 153.25 (significant at the 0.0001
level). Once again, both LEVERAGE and SIZE were positive and highly significant.
REFERENCES
[1]
Agrawal, A., and N. Nagarajan, "Corporate
Capital Structure, Agency Costs, and Ownership Control: The Case of All-Equity
Firms," Journal of Finance XLV, 1990, pp. 1325-1331.
[2]
Ball, R., and G. Foster, "Corporate
Financial Reporting: A Methodological Review of Empirical Research," Journal
of Accounting Research 20, 1982, pp. 161-234.
[3]
Bowen, R. M., E. W. Noreen and J. M. Lacey,
"Determinants of the Corporate Decision to Capitalize Interest," Journal
of Accounting and Economics 3, 1981, pp. 151-179.
[4]
Chow, C., "The Demand for External
Auditing: Size, Debt and Ownership Influences," The Accounting Review 57,
1982, pp. 272-291.
[5]
Christie, A., "Aggregation of Test
Statistics: An Evaluation of the Evidence on Contracting and Size
Hypotheses," Journal of Accounting and Economics 12, 1990, pp. 15-36.
[6]
Elliot, J., and D. Kennedy, "Estimation and
Prediction of Categorical Models in Accounting Research," Journal
of Accounting Literature 7, 1988, pp. 202-242.
[7]
Francis, J., "Lobbying Against Proposed
Accounting Standards: The Case of Employers' Pension Accounting," Journal
of Accounting and Public Policy 6, 1987, pp. 35-57.
[8]
Frost C., and V. Bernard, "The Role of Debt
Covenants in Assessing the Economic Consequences of Limiting Capitalization of
Exploration Costs," The AccountingReview 64, 1989, pp.
788-808.
[9]
Granof, M., and D. Short, "Why Do Companies
Reject LIFO?," Journal of Accounting, Auditing and Finance 7, 1984, pp.
323-333.
[10] Gray,
D., "Corporate Preferences for Foreign Currency Accounting
Standards," Journal of Accounting Research 22, 1984, pp. 760-764.
[11] Hagerman,
R., and M. Zmijewski, "Some Economic Determinants of Accounting Policy
Choice," Journal of Accounting and Economics 1, 1979, pp. 141-161.
[12] Healy,
P., and K. Palepu, "Effectiveness of Accounting-Based Dividend
Covenants," Journal of Accounting and Economics 12, 1990, pp. 97-123.
[13] Holthausen,
R., "Accounting Method Choice: Opportunistic Behavior, Efficient
Contracting and Information Perspectives, " Journal of Accounting and
Economics 12, 1990, pp. 207-218.
[14] Leftwich,
R., "Evidence of the Impact of Mandatory Changes in Accounting Principles
on Corporate Loan Agreements," Journal of Accounting and Economics
3, 1981, pp. 3-36.
[15] Lilien,
S., M. Mellman and V. Pastena, " Accounting Changes: Successful versus
Unsuccessful Firms," The Accounting Review 63, 1988, pp.
642-656.
[16] Lys,
T., "Mandated Accounting Changes and Debt Covenants: The Case of Oil and
Gas Accounting," Journal of Accounting and Economics
6, 1984, pp. 39-65.
[17] Malmquist,
D., "Efficient Contracting and the Choice of Accounting Method in the Oil
and Gas Industry," Journal of Accounting and Economics
12, 1990, pp. 173-205.
[18] Press,
E., and J. Weintrop, "Accounting-based Constraints in Public and Private
Debt Agreements: Their Association with Leverage and Impact on Accounting
Choice," Journal of Accounting and Economics 12, 1990, pp. 65-95.
[19] Watts,
R., and J. Zimmerman, "Towards a Positive Theory of the Determination of
Accounting Standards," The Accounting Review 53, 1978, pp.
112-134.
[20] Watts,
R., and J. Zimmerman, Positive accounting theory. Prentice Hall, Englewood Cliffs,
NJ, 1986.
[21] Watts,
R., and J. Zimmerman, "Positive Accounting Theory: A Ten Year
Perspective," The Accounting Review 65, 1990, pp. 131-156.
[22] Zimmerman,
J., "Taxes and Firm Size," Journal of Accounting and Economics
5, 1983, pp. 119-149.
[23] Zmijewski,
M., and R. Hagerman, "An Income Strategy Approach to the Positive Theory
of Accounting Standard Setting/Choice," Journal of Accounting and
Economics 3, 1981, pp. 129-149.
[1] Spearman Rank
Correlations.
*** significant at the
0.01 level, two-tailed test. ** significant at the 0.05 level, two-tailed test.
No comments