Audit delay and the timeliness of corporate

reporting*
ROBERT H. ASHTON Duke University

PAUL R. GRAUL Eastern Washington University

JAMES D. NEWTON University of Alberta

Abstract. We examine the determinants of "audit delay," the number of calendar days
from fiscal year-end to the audit report date. A descriptive model of audit delay is tested
on a sample of 465 companies listed on the Toronto Stock Exchange from 1977 to 1982.
Although several variables included in the model are statistically significant, the propor-
tion of variability in audit delay explained by the variables is low. Descriptive data are
presented for variables consistently associated with audit delay over the six-year period -
auditor size, industry classification, existence of extraordinary items, and sign of net
income. Some directions for friture research are also suggested.

Resume. Les auteurs examinent les determinants du «deiai de verification», soit le

nombre de jours francs qui s'ecoulent entre la fin de l'exercice financier et la publication
du rapport des vdHficateurs. Us v6rifient un module descriptif du deiai de verification stir
un echantillon de 465 entreprises cot^es k la bourse de Toronto, de 1977 a 1982. Bien que
plusieurs variables incluses dans le module soient statistiquement significatives, la propor-
tion de la variation du deiai de verification expliquee par ces variables est mince. Les
auteurs pr^sentent des donnees descriptives pour les variables associees de fagon uniforme
au deiai de verification, sur la periode de six ans - la taille du cabinet de verification, la
classification du secteur, l'existence de postes extraordinaires et le caract^re positif ou
ndgatif du resultat net. Us sugg^rent dgalement certaines pistes de recherche k envisager.

Introduction
This study examines some determinants of "audit delay," the length of time from
a company's fiscal year-end to its audit report date. Better understanding of the
determinants of audit delay may facilitate inferences conceming the structure and
function of the auditing profession. It may also help to explain "reporting delay,"
the length of time from fiscal year-end to the public release of eamings informa-
tion, which has been shown to be related to the market's reaction to the informa-
tion disclosed. For example, late announcements of eamings are more often

* Support provided to the first author by the Arthur Andersen Fund, Fuqua School of Business,
Duke University is gratefully acknowledged. We are indebted to Christine Fouillard and Kathy
West for assistance in data collection, and to Andy Hsu for assistance in data analysis. We also are
indebted to participants in the Accounting Workshop at the University of Alberta, and to the
editor Haim Falk, and two anonymous reviewers, for helpful comments.

Contemporary Accounting Research Vol. 5 No. 2 pp. 657-673

658 R.H. Ashton P.R. Graul J.D. Newton

associated with lower (often negative) abnormal retums than are early announce-
ments (Chambers and Penman (1984); Givoly and Palmon (1982), Kross (1982),
Kross and Schroeder (1984)). Since audit delay can affect the timeliness of
eamings releases, research on its determinants may improve our understanding
of market reactions to such releases. In addition to examining Canadian data on
audit delay for the first time, the study incorporates several methodological
refinements that distinguish it from prior studies.
The next section reviews existing research on reporting delay and audit delay.
Then a model of audit delay is presented, and the data used to test the model are
described. The results follow, and a concluding section discusses some direc-
tions for further research.

Prior research
The question of why reporting delay varies across firms has motivated several
studies. It has been suggested that management has incentives to exercise discre-
tion over the timeliness of reporting (e.g., Givoly and Palmon (1982); Pastena
and Ronen (1979); Patell and Wolfson (1982); Penman (1984); Ronen (1977);
and Verrechia (1983)). In particular, it has been hypothesized that bad news is
released later than good news, and empirical research strongly supports this
contention (Chambers and Penman (1984); Courtis (1976); Dodd et al. (1984);
Elliott (1982) Givoly and Palmon (1982); Kross (1981); Lawrence (1983); Lurie
and Pastena (1975); Niederhoffer and Regan (1972); Ohlson (1980); Pastena and
Ronen (1979); Patell and Wolfson (1982); Whittred (1980a); and Whittred and
Zimmer (1984)). These results have been interpreted as supporting the hypothe-
sis that management exercises discretion over the timeliness of reporting.
Givoly and Palmon (1982) suggested that variability in the length of the annual
extemal audit is a factor that explains variability in reporting delay. They main-
tained that the "single most important determinant of the timeliness of the eam-
ings announcement is the length of the audit" (p. 491). Givoly and Palmon's
argument distinguishes between reporting delay, the length of time from fiscal
year-end to the public announcement of eamings, and audit delay, the length of
time from year-end to the audit report date. Although reporting delay and audit
delay are likely to be highly correlated, empirical evidence suggests that they are
not identical. Garsombke (1981) has found a mean difference of 2.4 days between
the reporting delays and audit delays of a sample of U.S. firms, while other
researchers have found more sizable differences for samples of Australian firms
(e.g., Davies and Whittred (1980); Dyer and McHugh (1975); and Whittred
(1980a, 1980b)).
To determine whether audit delay is a predictable phenomenon, Givoly and
Palmon examined the relation between audit delay (measured in calendar days)
and three explanatory variables - total sales and two measures of firm complex-
ity. Dummy variables to control for good and bad news were also included. The
sample was composed of COMPUSTAT industrial firms in 1973 (n = 142) and
1974 (n = 149). Regression results showed that bad news was associated with
 Timeliness of Corporate Reporting 659

greater delay, as expected. Of the three size and complexity variables, however,
only one of the complexity measures was significantly associated with audit
delay, and then in only one of the two years. The overall R^ was 0.26 for 1973
and 0.19 for 1974.
Ashton, Willingham and Elliott (1987) also examined the relation between
audit delay and a set of explanatory variables. They examined 14 variables from
488 U.S. clients of Peat, Marwick, Mitchell & Co. in 1981-82, and their sample
included both public and nonpublic clients from six industries. The variables
were total revenues, four measures of firm complexity, industry classification,
public/nonpublic status, month of fiscal year-end, quality of intemal control, the
relative mix of audit work performed at interim and final dates, the length of time
the company had been a client of the auditor, two measures of profitability, and
the type of audit opinion issued. Regression results indicated that five variables
were significantly associated with the natural logarithm of audit delay - total
revenues, one of the complexity measures, intemal control quality, the mix of
interim and final work, and whether or not the company was publicly traded. The
R^ was 0.265 for the overall sample, but was higher for financial and public
subsamples (0.310 and 0.388, respectively).

Sample, data and model of audit delay

The present sample of companies differs from any that has been analyzed pre-
viously in studies of audit delay or reporting delay: it is composed entirely of
Canadian companies which were audited by Canadian auditors.' Specifically, it
includes all companies listed on the Toronto Stock Exchange (TSE) for 1977-1982
for which we were able to observe eight potential explanatory variables in each
of the six consecutive years. Data were available for 465 of 588 companies, or
79.1 percent of the population.
The remaining companies were excluded because data were not available for
one or more of the six years (82 companies) or because they were audited by
non-Canadian auditors (41 companies). We excluded audits by non-Canadian
auditors because we were specifically interested in the Canadian audit environ-
ment. One aspect of this environment was the prohibition by the Canadian
Institute of Chartered Accountants (CICA) of the subject-to opinion in November
1980, a prohibition which applied only to Canadian auditors (see e.g., Thomton
(1983, pp. 104-105)). The prohibition against subject-to opinions meant that
contingencies that previously were highlighted in audit opinions would hence-
forth be disclosed by management in footnotes to the financial statements. Audi-
tors would then give clean opinions if the disclosures were adequate. We discuss
some results related to this reporting change later.
The dependent variable was audit delay (DLY), defined as the number of
calendar days from the financial statement date to the audit report date. Data

1 Other studies of audit delay have involved companies in the United States (Garsombke (1981);
Givoly and Palmon (1982); Ashton, Willingham and Elliott (1987)), Australia (Davies and
Whittred (1980); Whittred (1980a, 1980b)), and New Zealand (Courtis (1976); Gilling (1977)).
660 R.H. Ashton P.R. Graul J.D. Newton

were collected on eight variables which had been shown (or suggested) to be
important in prior studies and which, a priori, we also believed to be important:
(1) company size, (2) industry classification, (3) month of year-end, (4) audit
firm, (5) sign of net income, (6) extraordinary items, (7) contingencies, and (8)
type of audit opinion.^
The direction of the association that should be expected between each of these
variables and audit delay is not clear. Consequently, we consider this study
descriptive. In considering each of the explanatory variables below, we discuss
potential reasons why either a positive or negative (or no) relation between each
explanatory variable and audit delay might be expected. Where applicable, we
compare our results with those of prior research.

Company size
Total assets (AST) was used as the measure of company size. Data on both assets
and revenues were collected as possible surrogates for size, but our analyses
indicated that assets provided greater explanatory power. An additional reason
for using assets is that the TSE lists mining companies in the development stage.
For such companies, revenues are treated as cost recoveries, or offsets, and
deducted from accumulated development costs, instead of being reported as
revenues in the income statement. Therefore, the use of revenues as a measure of
size would have decreased the number of companies in our sample.^
A positive relation between AST and DLY might be expected if it is thought
that increased time is required to audit larger companies; however, an increase in
audit work might not lead to longer audit delay, because the auditor has flexibil-
ity in timing the audit work. More work could be done prior to the financial
statement date, or it could be done after year-end by assigning more staff or
working more overtime. Moreover, larger companies may choose to implement
stronger internal controls, thus allowing the auditor to place more reliance on
interim compliance tests than on substantive tests of year-end balances. In addi-
tion, managements of larger companies may have incentives to reduce both audit
delay and reporting delay since larger companies may be monitored more closely
by investors, unions and regulatory agencies, and thus face greater external
pressure to report earlier (Dyer and McHugh (1975)). Other studies that have
used assets as a measure of company size have found a negative (though typically
weak) relation with audit delay (Courtis (1976); Davies and Whittred (1980);
Garsombke (1981); and Gilling (1977)).

Industry
Industry classification (IND) was employed as an explanatory variable. The TSE
2 Audit opinions were classified as either unqualified or "other." See footnote 7 for details.
3 There are 17 development-stage mining companies in the sample, but not all of them were in the
development stage for the entire six years. In fact, there are only 62 company-years (of a total
of 2,790 company-years) of this type. Our LOSS variable (described later) was coded in the usual
way for these company-years; that is, we grouped observations having zero income (including
these 62) together with those having positive income. In addition, none of these company-years
involved extraordinary items (see EXTR, also described later).
 Timeliness of Corporate Reporting 661

industry classification scheme, which involves 14 major categories, was used

because all companies were listed on the TSE. For our purposes, the 14 catego-
ries were combined into two groups - (1) financial services companies, and (2)
all others. The financial (nonfinancial) companies were assigned a 0 (1) for data
analysis purposes. Prior studies have found that financial companies have shorter
audit delays than companies in other industry classifications (e.g., Courtis (1976);
Ashton, Willingham and Elliott (1987)), which suggests that, dollar for dollar,
financial assets are easier to audit than nonfinancial assets.

Year-end
A third explanatory variable was month of fiscal year-end (MON). A large
percentage of TSE companies report on a calendar year basis. Thus, the "busy
season" is likely to include January and February. To capture the potential effect
of the busy season on audit delay, we distinguished between those companies
with year-ends in December or January (assigned a 1) and those with year-ends
in any of the other ten months (assigned a 0)."*
Performing audits during the busy season could result in either increased or
decreased audit delay, depending on whether the increased workload is handled
by increased overtime or more audit staff. In Australia, Davies and Whittred
(1980) found longer audit delays for companies with June year-ends, the most
common year-end for companies listed on the Sydney Stock Exchange. In the
United States, Garsombke (1981) found longer audit delays for January through
March year-ends. However, Ashton, Willingham and Elliot (1987), also in the
United States, found the opposite to that of Garsombke, although the effect was
weak.

Auditor
Another explanatory variable employed was the company's auditor (AUD). For
analysis purposes, AUD was classified into two groups - the Canadian "Big
Nine" and all other auditors. During the years in the study, the Big Nine group
was composed of seven of the traditional "Big Eight" international firms (exclud-
ing Arthur Young) and the two largest Canadian firms, Clarkson Gordon (affili-
ated with Arthur Young) and Thome Riddell (affiliated with Klynveld, Main,
Goerdler). Only audit reports issued by Canadian offices of the firms were
included.^
It may be reasonable to expect that larger audit firms would complete audits on
a more timely basis because of their experience in auditing companies listed on
4 Thefinancialstatement dates of companies that use a 52/53 week year may fall in different
months in different years. For example, thefinancialstatement date of a January-year-end com-
pany could be January 30 in one year and February 2 in the following year. For the purpose of
studying the importance of busy season (MON), such a company was coded as "January" in both
years. However, when computing DLY as the number of daysfromfinancialstatement date to
auditreportdate, the actual financial statement date was used.
5 It is common practice in Canada that three auditors are appointed for banks. One performs the
audit each year, while the other two rotate annually. Thus, the audit of a bank is the joint product
of two auditors each year. We randomly designated one of these two as the bank's auditor. Only
seven companies in our sample are baijcs.
662 R.H. Ashton P.R. Graul J.D. Newton

the TSE. Large audit firms may be able to audit such companies more efficientiy
than small audit firms. Therefore, since Big Nine (smaller) firms were assigned a
1 (0) in our data set, one might expect to find a negative relation between AUD
and DLY. We note, however, that Garsombke (1981) found no differences in
audit delay for Big Eight vs. other auditors in the U.S., and that Davies and
Whittred (1980) found no differences among the Big Eight in Australia.

Other explanatory variables

The preceding four variables represent characteristics of the sample companies
that are not likely to change significantiy from year to year. Although some firms
did change auditors, these situations were infrequent.^ In contrast, the remaining
four variables relate specifically to the particular year that is being audited. First,
the sign of net income (designated LOSS) can be considered one indication of
whether good news or bad news resulted from the year's activities. Second, the
presence of extraordinary items (EXTR) which, by definition, refiect nonrecur-
ring events that arise from something other than the company's normal opera-
tions, may be expected to require additional audit time. The same can be said of
contingencies (CTNG) which require disclosure in the financial statements. Fin-
ally, the length of the audit could be affected if something other than a "clean"
opinion (OPIN) is issued.
For purposes of analysis, these variables were assigned a 1 when a negative
net income occurred, when an exfraordinary item was reported, when a contin-
gency was reported, and when the audit opinion was other than clean - and a 0
otherwise. (In the case of OPIN, consistency exceptions were freated as clean
opinions.) Coded in this way, a positive relation between each of these variables
and DLY might be expected (also see Davies and Whitfred (1980); Whitti-ed
(1980a); Ashton, Willingham and Elliot (1987)). On the other hand, it is possible
that these variables could have no effect on DLY if the problem (e.g., the
exfraordinary item or the contingency) was anticipated at the beginning of the
audit.

Results
The results are presented in three parts. First, summary statistics that describe the
dependent variable and the eight explanatory variables are presented. Second,
the results from six cross-sectional multiple regressions of DLY on the explana-
tory variables are presented. Finally, variables that are consistently important
(over the six years) in explaining audit delay are analyzed in more detail.

Descriptive statistics
Table 1 presents descriptive statistics for all variables. Notice that three of the
explanatory variables (AST, AUD and CTNG) increased steadily over the six-
6 There were only 22 switches during the six-year period between our Big Nine and non-Big Nine
categories, 18 of which were from smaller firms to the Big Nine. In contrast, there were 54
switches within our two categories, 52 of which were within the Big Nine.
 00 '^ O 00 r^ ^^

ll

oo oo P-- O tN
00 fo ON o ^^
00 r-- oo c^ en

13;

oo oo ^ ri r-; 00 <
00 o »n .-i ri o
00 f**^ oo ^^ f^ f^

S-.

00 cs
00 •—" • wS Q od o\
00 r ^^ r^ (S

eQ.2 »
€
it
^1
I
>
664 R.H. Ashton P.R. Graul J.D. Newton

year period. In fact, mean AST more that doubled from 1977 to 1982, as did the
variability of AST. Of course, the increase in asset size is partially related to
infiation (which we consider later). The values of AST are skewed to the right
because of inclusion in the sample of some extremely large companies, as sug-
gested by a comparison of the mean and median AST values in Table 1. The
steady increase in AUD suggests that a growing proportion of TSE companies
are being audited by one of Canada's Big Nine, and the increase in CTNG
reveals that the proportion of companies reporting contingencies grew from 1977
to 1982.
Further examination of Table 1 shows that the proportion of companies report-
ing negative net incomes doubled from 1980 to 1981 and increased by another 75
percent in 1982 (see LOSS). In that year, losses were reported by more than
one-third of the firms in the sample. Table 1 also reveals that the proportion of
companies receiving qualified audit opinions decreased sharply after 1979 (see
OPIN). The timing of this decrease coincides with the decision of the CICA to
prohibit the issuance of subject-to qualifications. (This decision was rendered in
August 1980, and applied to audit reports dated on or after November 1, 1980.)
Descriptive statistics for the remaining explanatory variables (IND, MON and
EXTR) change little from year to year.
Table 1 reveals that mean audit delay (DLY) is stable at 55 days in each year.
Mean delay is substantially shorter than mean delays found in previous studies
conducted in Australia (Dyer and McHugh (1975); Whittred (1980a, 1980b)),
New Zealand (Courtis (1976); Giiling (1977)), and the U.S. (Ashton, Willing-
ham and Elliott (1987)). There is substantial variability within any particular
year, as shown by the standard deviations of 26 to 32 days.

Regression results
The results reported below focus on the multivariate relation among the variables
within each of the years covered by the study. This approach, which involved
regression modeling, was also followed by Givoly and Palmon (1982) and Ash-
ton, Willingham and Elliot (1987). Because of the positive skew in DLY, the
natural logarithm of the dependent variable was used in the regression analyses.
Thus, the regression coefficients reported in Table 2 can be interpreted as the
percentage change in DLY associated with changes in the independent variables.
The natural logarithm of AST was also used, and asset values were deflated using
the GNP deflator.
Multicollinearity among the explanatory variables, as well as the assumptions
of constant variance and normality of residuals, were examined. The simple
correlations among the explanatory variables were low. In fact, of the 168
correlations computed (28 each year times six years), only six were above 0.30
and the largest was 0.39. Consequently, we do not consider multicollinearity a
significant problem in interpreting the regression results reported below. Visual
inspection of residuals plots revealed no heteroscedasticity problems in any of
the six years. Finally, Kolmogorov-Smimov tests for normality of residuals
indicated that the normality assumption was violated only in 1979 (p < 0.05).
 I cicS
II

p < s — — — (Sp—'
If OOOOOOOO
I I I I

O I (N —' f«^ r^ O (N
I II II

OOOOOOOO
I I I I I

I I I I

pp
OOOOOOOO
I I I I

o
o

p<soqp<spp ——
OOOOOOOO -*0
I

'S .S

OOOOOOOO
I I I I
666 R.H. Ashton P.R. Graul J.D. Newton

The results of regressing DLY on the set of eight explanatory variables are
shown in Table 2. Four variables have statistically significant explanatory power
in at least four of the six years: IND and EXTR are significant in all six years,
while LOSS is significant in five years, and OPIN in four. In addition, the signs
of IND, EXTR and LOSS are consistent over the entire six-year period, while the
signs for OPIN are consistent for the four years in which it is significant.^ The
signs are also consistent for AST, MON and AUD, but they reach significance
for three or fewer of the six years. Finally, CTNG is characterized by weak
significance results and inconsistent signs.
These results indicate that company size as measured by AST is inversely
related to audit delay. They also indicate that financial services companies, as
well as companies with ye£U"-ends in their "busy season," have shorter delays.
Moreover, the results for LOSS and EXTR reveal longer delays for companies
with negative net incomes and for companies that report extraordinary items.
Finally, the AUD variable indicates that Big Nine auditors are consistently
associated with shorter audit delays than are smaller auditing firms.
The explanatory power of the entire set of variables is estimated by the adjusted
R^ figures shown at the bottom of Table 2. These numbers range from 0.126 in
1978 to 0.088 in 1982. Although these values are statistically significant, it is
clear that only a small amount of the total variiance in audit delay is explained by
this set of eight variables. In fact, these ^^s are smaller than those reported by
Ashton, Willingham and Elliott (1987), who also used the natural logarithm of
DLY as the dependent variable.

Further analyses
LOSS, EXTR, IND and AUD are examined in more detail below. We do not
consider further the other variables because of their weaker significance levels
and, in some cases, inconsistent signs.

Negative net income (LOSS): Detailed information for the LOSS variable is
reported in Table 3, which shows that mean audit delay is greater for companies
reporting negative net incomes than for companies reporting zero or positive net
incomes. The difference ranges from 12.1 days (1982) to 28.5 days (1979), and
the mean difference for the six-year period is 19.2 days. The differences in mean
delay between companies reporting negative versus zero or positive net incomes
are significant via t tests at the 0.01 level for each of the six years.
These results suggest that, on average, audit delay for companies reporting
negative net incomes is more than one-third longer than audit delay for other
7 Audit opinions were classified as either unqualified or "other," where the other category includes
opinions that are subject-to, except for, adverse, and denials. Recall, however, that from
November 1980 the CICA prohibited auditors from giving subject-to opinions. Consequently,
our "other" category contains subject-to opinions for 1977 through part of 1980, but does not con-
tain subject-to opinions for the remainder of 1980 through 1982. As a check on our results, we
used alternative coding schemes to determine if the results were sensitive to the scheme used,
e.g., subject-to opinions were coded separately. No differences in results were found.
 Timeliness of Corporate Reporting 667

TABLE 3
Effect of sign of net income on audit delay

Audit delay

Direction of Number of Standard Coefficient of

Year net income companies Mean deviation variation

1977 Positive" 391 51.7 24.8 0.5

Negative 74 75.0 43.1 0.6
1978 Positive 413 53.7 27.7 0.5
Negative 52 72.6 41.0 0.6
1979 Positive 418 52.0 24.2 0.5
Negative 47 80.5 66.3 0.8
1980 Positive 417 52.9 23.8 0.4
Negative 48 66.8 31.5 0.5
1981- Positive 372 51.3 23.8 0.5
Negative 93 69.8 40.7 0.6
1982 Positive 303 50.8 29.2 0.6
Negative 162 62.9 33.8 0.5

'"Positive" indicates positive or zero net income.

companies. Note, however, that while mean delay is almost constant over the six
years for companies with zero or positive net incomes, it tends to decrease over
time for companies with negative net incomes. Moreover, the standard devia-
tions of delay are greater each year, and less consistent from year to year, for
companies with negative net incomes. Finally, examination of the coefficients of
variation (standard deviation -r- mean) reveals greater coefficients for companies
with negative net incomes for all years except 1982.

Extraordinary items (EXTR): The effect of exfraordinary items (EXTR) on audit

delay is shown in Table 4, where the sample is broken down into three categories -
companies with extraordinary items that increase or decrease net income, and
companies with no extraordinary items. Table 4 shows that extraordinary items
that increase net income outnumber those that decrease net income by a factor of
almost 3:1 over the six-year period. This varies from 1.6:1 for 1982 to 4.1:1 for
1979. However, the direction of effect on net income does not have a consistent
impact on audit delay. Firms with extraordinary items that increase net income
are associated with longer audit delays than all other firms in three of the six
years, while firms with extraordinary items that decrease net income are associ-
ated with longer audit delays than all other firms in the other three years. How-
ever, the existence of extraordinary items, regardless of their income effect, is
associated with longer audit delays for each of the six years. The differences in
mean delay between companies reporting extraordinary items of either sign
versus no exfraordinary items are significant via /-tests at the 0.01 level for all
years except 1979, for which the difference is significant at the 0.05 level. On
668 R.H. Ashton P.R. Graul J.D. Newton

TABLE 4
Effect of extraordinary items on audit delay

Audit delay

Effect of EXTR Number of Standard Coefficient of

Year on net income companies Mean deviation correlation

1977 Increase 99 60.9 22.8 0.4

Decrease 44 68.9 39.1 0.6
No EXTR 322 51.9 29.4 0.6
1978 Increase 121 66.3 32.4 0.5
Decrease 31 61.2 23.4 0.4
No EXTR 313 51.2 28.6 0.6
1979 Increase 126 58.0 22.4 0.4
Decrease 31 66.1 30.9 0.5
No EXTR 308 52.5 35.3 0.7
1980 Increase 108 63.6 27.3 0.4
Decrease 33 60.7 27.5 0.5
No EXTR 324 50.6 24.2 0.5
1981 Increase 101 59.4 22.9 0.4
Decrease 44 66.3 40.0 0.6
No EXTR 320 52.1 28.3 0.5
1982 Increase 85 65.6 45.3 0.7
Decrease 53 60.9 25.4 0.4
No EXTR 327 51.3 26.8 0.5

average, audit delay is approximately 11 days longer when extraordinary items

of either sign are reported. Like the association between audit delay and direction
of extraordinary items, however, neither the standard deviation nor the coeffi-
cient of variation of delay is consistently related to either the direction or exis-
tence of extraordinary items.

Industry (IND): Recall that although the companies in our sample represent 14
industry classifications used by the TSE, we divided industries into two catego-
ries (financial and nonfinancial) for purposes of analysis. This classification
scheme obscures differences that exist within the nonfinancial category. Addi-
tional detail is provided in Table 5.
Mean audit delay for financial services companies is 43.3 days over the six-
year period, compared to 56.5 days for nonfinancial companies. Examination of
each of the six years separately reveals that the differences in mean delay between
financial and nonfinancial companies are significant via r-tests at the 0.01 level
for all years except 1982, which is not significant at conventional levels. There is
considerable variation, however, among the nonfinancial industries. For exam-
ple, utilities and paper and forest products companies have delays that are shorter
than average, while transportation and oil and gas companies have delays that are
longer than average. In addition, the variability in audit delay over the six years
 Timeliness of Corporate Reporting 669

TABLE 5
Effect of industry on audit delay

Audit delay

Industry companies" Mean" Minimum'^ Maximum

Financial services 52 43.3 41 48

Utilities 17 38.3 37 41
Paper and forest prxxlucts 18 42.4 40 44
Pipelines 8 42.7 39 52
Communication and media 15 51.7 49 54
Golds 24 53.8 52 55
Metals and minerals 76 54.7 52 58
Industrial products 65 54.8 53 57
Consumer products 51 58.1 55 61
Management companies 9 59.5 57 61
Transportation 13 60.2 53 68
Merchandising 40 61.7 57 73
Real estate and construction 16 63.7 58 73
Oil and gas 61 66.2 64 69

' The number of companies in each industry was constant over the six-year period.
*" Mean delay over the six-year period.
° Minimum (Maximum) "= smallest (largest) average annual delay during the six years.

differs substantially among the nonfinancial industries, as can be seen by examin-

ing the "minimum" and "maximum" columns of Table 5.*

Auditor (AUD): It may also be of interest to examine more closely the effect of
auditor on audit delay. Although the auditor (AUD) variable reached significance
in the regressions for only one of the six years, the signs on the coefficients were
consistently negative, indicating shorter audit delays for Big Nine auditors. Over
the six-year period, mean delay for the Big Nine (non-Big Nine) was 54.2 (65.2)
days. Examination of each of the six years separately reveals that differences in
mean delay between Big Nine and non-Big Nine auditors are significant via
f-tests at the 0.05 level for all years except 1978 and 1982, which are significant
at the 0.01 level.
The aggregate mean delay of 54.2 days, however, obscures a wide range of
differences in delay among the Big Nine. More detailed information is presented
in Table 6. Mean audit delay ranges from 44.5 days to 61.2 days for clients of
Big Nine firms. However, even the longest mean delay among clients of the Big
Nine is shorter than the mean delay for clients of smaller auditors. Variability in
mean delay over the six years also differs considerably among clients of the Big
Nine. For example, clients of Arthur Andersen, Price Waterhouse, and Touche
Ross are characterized by low inter-year variations, while clients of Peat, Mar-
wick, Mitchell; Thome Riddell, and Deloitte, Haskins & Sells show much
higher variations.
8 The range varies from three days for the Golds Industry to 16 days for the Merchandising
Industry.
670 R.H. Ashton P.R. Graul J.D. Newton

TABLE 6
Effect of auditor on audit delay

Audit delay

Audit firm companies" Mean" Minimum*^ Maximum

Arthur Andersen 18 44.5 43 46

Price Waterhouse 68 47.8 47 49
Coopers & Lybrand 50 49.4 46 54
Clarkson Gordon 71 52.2 51 58
Peat Marwick & Mitchell 40 54.9 49 65
Thome Riddell 82 57.6 47 62
Touche Ross 46 58.9 57 61
Emst & Whinney 7 60.6 54 66
Deloitte, Haskins & Sells 23 61.2 53 72
All other firms 60 65.2 63 75

"Mean number of companies audited over the six-year period.

•"Mean delay over the six-year period.
'^Minimum (Maximum) = smallest (largest) average annual delay during the six years.

Tables 5 and 6 show both industry effects and auditor effects on audit delay.
We examined the possibility that these two types of effects are related. In
particular, we considered it possible that companies in industries with long
(short) mean delays are heavily concentrated in audit firms having long (short)
mean delays. A median split was used to divide both the industries and the
auditors into two categories with long and short audit delays, resulting in four
industry/auditor "cells." For each cell, we counted the number of times an
auditor had more than ten percent of the companies in an industry. If a systematic
relation exists between industry delays and auditor delays, we should find larger
numbers in the long/long and short/short cells than in the other two cells. This
does not occur: The numbers are 15,14,18 and 16 for the long/long, long/short,
short/long and short/short cells, respectively. We repeated this analysis using
cutoffs of 15, 20 and 25 percent, with no change in the results. Thus, there does
not appear to be a systematic tendency for companies in long-delay (short-delay)
industries to be clients of auditors having long (short) delays. In other words, the
effects on audit delay of industry and auditor are not redundant.

Summary and directions for further research

This study examined cross-sectional variability in audit delay. A model based on
several factors often hypothesized to affect the length of an audit was developed,
and the model was tested on 465 Toronto Stock Exchange companies in each of
six years. Four such factors were significantly related to audit delay for at least
four of the six years, and others had consistent signs in every year but were
significant in three or fewer years.
Our measure of company size - total assets - was inversely related to audit
delay, as was the size of the company's auditor. Financial services companies
had shorter audit delays them compeuiies in other industries. Companies with
 Timeliness of Corporate Reporting 671

negative net incomes had longer audit delays than companies with positive or
zero net incomes, and the existence of extraordinary items, regardless of whether
they increased or decreased net income, was associated with longer audit delays.
Finally, audits conducted during the "busy season" had shorter delays than those
conducted during other months.
While these variables are significant in a statistical sense, little of the cross-
sectional variability in audit delay for this particular sample is explained by the
model. The adjusted R^ is relatively low in each of the six years, suggesting that
audit delay depends on factors outside our model. It is possible that management
discretion is an important underlying factor. Additional research that focuses on
different data sets, different time periods and, especially, different explanatory
variables is needed to significantly improve our understanding of audit delay. We
describe below four possible avenues for further research.
First, consider the LOSS variable. Our analysis provides strong support for the
notion that the audit report is delayed when a loss is reported. One possible
reason is that a loss reflects difficult times for the company, requiring more audit
work. However, we argued earlier that audits of larger companies need not take
longer since the schedtiling of the audit or the assignment of staff is under the
control of the audit firm, and this argument also applies to companies reporting
losses. A second possibility is that management delays the reporting of bad news
by delaying the audit. This might involve, for example, being "slow" to respond
to the auditor's inquiries, or engaging in lengthy negotiations with the auditor
about valuation or disclosure options, materiality limits, or the appropriate type
of opinion to be issued. Another possibility is that auditors are risk averse in the
face of a loss, since the possibility of lawsuits might increase when a company
appears nearer to failure, and a loss might signal impending failure. In this case,
auditors, instead of management, could be responsible for the increased delay
because of additional testing. These possibilities may be good candidates for
further research.^
Second, note that in 1982 the audit delays of companies reporting negative net
incomes averaged 12.1 days longer than audit delays of other companies, while
in 1977 through 1981 the mean difference was 20.6 days. Since audit delay
remained almost constant for companies with nonnegative net incomes from
1977 to 1982, this difference of 8.5 days can be attributed largely to a decrease in
audit delay for companies reporting losses in 1982. In addition, the percentage
of companies reporting losses was substantially higher in 1982 than in 1977-1981.
Perhaps the 1982 decrease in audit delay for companies reporting losses was due
to the fact that losses were no longer viewed as unusual. Another possibility is

9 We performed several analyses to try to determine which of these possibilities is a better explana-
tion for ourresultsconcerning LOSS. For example, werefinedthe "bad news" criterion by
incorporating various expectations measures (e.g., annual companyresultsversus annual industry
results, and annual company results versus long-run company results), and investigated the
possibility that, in the presence of these various definitions of bad news, audit delay would be
longer. However, these efforts did not enable us to determine the underlying cause(s) of increased
delay.
672 R.H. Ashton P.R. Graul J.D. Newton

that auditing firms shifted a disproportionately large amount of resources to the

audits of such companies. These conjectures could be pursued by further research.
Third, recall that there were differences in audit delay not only between Big
Nine and non-Big Nine auditors, but also among the Big Nine themselves. It is
possible that auditing-firm policies play an important role in audit delay. Exam-
ples could be policies related to the mix of work done at interim and final dates,
policies related to overtime scheduling, policies related to specific industries in
which the auditing firm is a "specialist," and so one. Further research that
focuses on the auditing firm may improve our understanding of audit delay.
Finally, the disclosure of a contingency was associated with longer audit
delays in 1977 through 1980, and the coefficient of CTNG was significant in
1977 and 1979. Canadian auditors have been prohibited since November 1980
from issuing subject-to opinions due to contingencies. Our results show that
CTNG was not significant in any of the years 1980 through 1982, nor did it have
a consistent sign. Moreover, the number of contingencies disclosed in 1980-1982
was greater than the number disclosed in 1977-1979. The increase in contin-
gency disclosures may have refiected the shift in responsibility for disclosure of
some uncertainties from auditors (through subject-to opinions) to management
(through footnotes to the financial statements). Possibly the significance of CTNG
in the earlier period indicates that the contingencies then disclosed by manage-
ment involved residual uncertainty that could not be resolved by an equivocal
audit opinion, and which was therefore relatively more important in infiuencing
audit delay. More research is needed to substantiate this hypothesis.

References
Ashton, R.H., J.J. Willingham and R.K. Elliott, "An Empirical Analysis of Audit
Delay," Journal of Accounting Research (Autumn 1987) pp. 275-292.
Chambers, A.E. and S.H. Penman, "Timeliness of Reporting and the Stock Price Reac-
tion to Eamings Announcements," Journal of Accounting Research (Spring 1984)
pp. 21-47.
Courtis, J.K., "Relationships Between Timeliness in Corporate Reporting and Corporate
Attributes," Accounting and Business Research (Winter 1976) pp. 45-56.
Davies, B. and G.P. Whittred, "The Association Between Selected Corporate Attributes
and Timeliness in Corporate Reporting: Further Analysis," Abacus (June 1980)
pp. 48-60.
Dodd, P., N. Dopuch, R. Holthausen and R. Leftwich, "Qualified Audit Opinions and
Stock Prices: Information Content, Announcement Dates, and Concurrent Disclostires,"
Journal of Accounting and Economics (April 1984) pp. 3-38.
Dyer, J.D. and A.J. McHugh, "The Timeliness of the Australian Annual Report,"
Journal of Accounting Research (Autumn 1975) pp. 204-219.
Elliott, J.A., " 'Subject to' Audit Opinions and Abnormal Security Returns - Outcomes
and Ambiguities," Journal ofAccounting Research (Autumn 1982 Part II) pp. 617-638.
Garsombke, H.P., "The Timeliness of Corporate Financial Disclosure," in: J.K. Courtis
(ed.). Communication via Annual Reports, AFM Exploratory Series No. 11 (Armidale,
N.S.W.: University of New England, 1981) pp. 204-218.
Gilling, D.M., "Timeliness of Corporate Reporting: Some Further Cotnment," Account-
ing and Business Research (Winter 1977) pp. 34-36.
 Timeliness of Corporate Reporting 673

Givoly, D. and D. Palmon, "Timeliness of Annual Earnings Announcements: Some

Empirical Evidence," The Accounting Review (July 1982) pp. 486-508.
Kross, W., "Eamings and Announcement Time Lags," Journal of Business Research
(September 1981) pp. 267-281.
, "Profitability, Earnings Announcement Time Lags, and Stock Prices," Journal
of Business Finance and Accounting (Autumn 1982) pp. 313-328.
• and D. A. Schroeder, "An Empirical Investigation of the Effect of Quarterly
Earnings Announcement Titning on Stock Retums," Journal of Accounting Research
(Spring 1984) pp. 153-176.
Lawrence, E.C., "Reporting Delays for Failed Firms," Journal of Accounting Research
(Autumn 1983) pp. 606-610.
Lude, A. and V. Pastena, "How Promptly do Corporations Disclose Their Problems?,"
Financial Analysts Journal (September-October 1975) pp. 55-61.
Niedeihoffer, V. and P. Regan, "Eamings Changes, Analysts' Forecasts and Stock
Prices," Financial Analysts Journal (May-June 1972) pp. 65-71.
Ohlson, J.A., "Financial Ratios and the Probabilistic Prediction of Bankmptcy," Journal
of Accounting Research (Spring 1980) pp. 109-131.
Pastena, V. and J. Ronen, "Some Hjrpotheses on the Pattern of Management's Informal
Disclosures," Journal of Accounting Research (Autumn 1979) pp. 550-564.
Patell, J.M. and M.A. Wolfson, "Good News, Bad News, and the Intraday Timing of
Corporate EWsclosures," The Accounting Review (July 1982) pp. 509-527.
Penman, S.H., "Abnonnal Retums to Investment Strategies Based on the Timing of Earn-
ings Reports," Journal of Accounting and Economics (December 1984) pp. 165-183.
Ronen, J., "The Effect of Insider Trading Rules on Information Generation and Dis-
closure by Corporations," The Accounting Review (April 1977) pp. 438-449.
Thornton, D.B., Tlie Financial Reporting of Contingencies arui Uncertainties: Theory
and Practice, Research Monograph No. 5 (Vancouver: Canadian Certified General
Accountants' Research Foundation, 1983).
Venecchia, R.E., "Discretionary Disclosure," Journal cf Accounting and Economics
(December 1983) pp. 179-194.
Whittred, G.P., "Audit Qualification and the Timeliness of Corporate Annual Reports,"
The Accounting Review (October 1980a) pp. 563-577.
, "The Timeliness of the Australian Annual Report: 1972-1977," Journal of
Accounting Research (Autumn 1980b) pp. 623-628.
Whittred, G. and I. Zitnmer, "Timeliness of Financial Reporting and Financial Distress,"
The Accounting Review (April 1984) pp. 287-295.