8/4/2019 Classificatory Income Smoothing
1/50
Working Paper Series
Classificatory income smoothing: the impact of FRS3
Vasiliki AthanasakouNorman StrongMartin Walker
Manchester Business School Working Paper No. 502June 2006
Manchester Business SchoolCopyright 2006, Athanasakou, Strong and Walker. All rights reserved.Do not quote or cite without permission from the author.
Manchester Business School
The University of ManchesterBooth Street WestManchester M15 6PB
+44(0)161 306 1320http://www.mbs.ac.uk/research/working-papers/default.aspxISSN 0954-7401
The working papers are produced by The University of Manchester - Manchester Business School and areto be circulated for discussion purposes only. Their contents should be considered to be preliminary. Thepapers are expected to be published in due course, in a revised form and should not be quoted withoutthe authors permission.
8/4/2019 Classificatory Income Smoothing
2/50
2
Author(s) and affiliation
Vasiliki AthanasakouCentre for Analysis of Investment RiskManchester Business SchoolThe University of Manchester
Crawford HouseBooth Street EastManchester, M13 9PLTel: +44(0) 161 275 0227Fax: +44(1) 161 275 0227E-mail: [email protected]: http://www.mbs.ac.uk
Norman Strong Martin WalkerAccounting & Finance Division Accounting & Finance DivisionManchester Business School Manchester Business SchoolThe University of Manchester The University of ManchesterCrawford House Crawford House
Booth Street East Booth Street EastManchester, M13 9PL Manchester, M13 9PLTel: +44(0) 161 275 4006 Tel: +44(0) 161 275 4008Fax : +44(0) 161 275 4023E-Mail : [email protected] Email : [email protected] :http://www.mbs.ac.uk Web : http://www.mbs.ac.uk
JEL ClassificationM41
Abstract
Financial Reporting Standard No 3 (FRS3) has regulated the reporting of financial performance by UKfirms since 1993. FRS3 effectively outlawed extraordinary items, but provided for a clearer distinctionbetween recurring and transitory income by giving firms considerable discretion over classifications ofexceptional items and disclosures of alternative measures of EPS, thereby increasing the scope forclassificatory income smoothing to highlight persistent profitability. We examine the impact of FRS3 onclassificatory smoothing by UK firms and document a significant increase in this practice post-FRS3.Even though this increase coincides with a rise in the magnitude of negative non-recurring items, FRS3removed the asymmetry in the distribution of non-recurring items that existed pre-FRS3, when negativeitems were more likely to be classified below the line of basic earnings. More importantly, we find thatdeviations of net income from expected earnings induce a significantly higher level of classificatorysmoothing post-FRS3 and that removing non-recurring items results in more persistent earnings post-FRS3 at the pre-exceptional level. Our results suggest greater use of classificatory choices to highlightsustainable profitability after the change in performance reporting regime.
How to quote or cite this document
Athanasakou, Vasiliki, Strong, Norman & Walker, Martin. (2006). Classificatory income smoothing: theimpact of FRS3. Manchester Business School Working Paper, Number 502, available:http://www.mbs.ac.uk/research/working-papers.aspx.
8/4/2019 Classificatory Income Smoothing
3/50
3
Classificatory income smoothing: the impact of FRS3
Vasiliki Athanasakou,*
Norman Strong, Martin Walker
Manchester Business School, The University of Manchester
June, 2006
JEL classification: M41
*Corresponding author: Centre for Analysis of Investment Risk, Manchester Business School, University of Manchester,
Manchester M15 6PB, UK. Tel: ++ (0)161 2750227. Email: [email protected].
We gratefully acknowledge the comments of Professor Kenneth Peasnell and Dr. Steven Young (Lancaster University).
8/4/2019 Classificatory Income Smoothing
4/50
4
Classificatory income smoothing: the impact of FRS3
Abstract
Financial Reporting Standard No 3 (FRS3) has regulated the reporting of financial performance by
UK firms since 1993. FRS3 effectively outlawed extraordinary items, but provided for a clearer distinction
between recurring and transitory income by giving firms considerable discretion over classifications of
exceptional items and disclosures of alternative measures of EPS, thereby increasing the scope for
classificatory income smoothing to highlight persistent profitability. We examine the impact of FRS3 on
classificatory smoothing by UK firms and document a significant increase in this practice post-FRS3. Even
though this increase coincides with a rise in the magnitude of negative non-recurring items, FRS3 removed
the asymmetry in the distribution of non-recurring items that existed pre-FRS3, when negative items were
more likely to be classified below the line of basic earnings. More importantly, we find that deviations of
net income from expected earnings induce a significantly higher level of classificatory smoothing post-
FRS3 and that removing non-recurring items results in more persistent earnings post-FRS3 at the pre-
exceptional level. Our results suggest greater use of classificatory choices to highlight sustainable
profitability after the change in performance reporting regime.
8/4/2019 Classificatory Income Smoothing
5/50
5
Classificatory income smoothing: the impact FRS3
1. Introduction
We explore the impact of Financial Reporting Standard No3 (FRS3): Reporting Financial
Performance on the practice of income smoothing via classifications of transitory items. FRS3 introduced
a radical change in how UK firms report their financial performance. Before FRS3, SSAP3 required firms
to report basic EPS on ordinary income, while SSAP6 offered a flexible definition of extraordinary items.1
This encouraged inconsistent classifications of items below ordinary income and classificatory smoothing
through extraordinary items. It raised concerns among UK authorities that companies were increasingly
classifying transitory items as exceptional if they were debits, and as extraordinary if they were credits.
FRS3 mandated calculation of basic EPS on net income, disabling classifications of extraordinary
items to smooth or increase EPS. At the same time, FRS3 contained new provisions for exceptional items,
allowing for more informative distinctions between recurring and non-recurring income and between
operating and non-operating income. It further allowed firms to disclose alternative EPS as long as they
reconciled any alternative EPS to the basic figure, disclosed it consistently over time, and gave it no greater
prominence in the annual report than basic EPS. These provisions enabled firms to smooth profit and loss
sub-totals via classifications of non-recurring items, and had several implications for practice. First,
FRS3s wide definition of exceptional items encouraged managerial judgment in classifying them in line
with the nature of the firms activities. Second, firms could report alternative EPS on earnings before all
exceptional items, not just those formerly classified as extraordinary. Third, FRS3s disclosure
requirements increased the transparency of classificatory choices. In all, the increased flexibility and
transparency of classificatory choices under FRS3 allowed the provision of a more accurate measure of
recurrent profitability.
To predict the impact of FRS3 on classificatory smoothing, we focus on the information
perspective of accounting choice (Holhausen and Leftwich, 1983) and assess the implications of FRS3s
1FRS3, issued by the Accounting Standards Board (ASB), amended Statement of Standard Accounting Practice No 3
(SSAP3), issued by the ASBs predecessor, the Accounting Standards Committee (ASC), and superseded SSAP6.
8/4/2019 Classificatory Income Smoothing
6/50
6
provisions on classificatory smoothing as a way of highlighting a better indicator of future profitability.
Pre-FRS3 we examine classificatory smoothing via extraordinary items. Post-FRS3, in line with the
gradual increase in the number of firms reporting alternative EPS on earnings before exceptional items
(Choi, Lin, Walker, and Young 2005), we examine classificatory smoothing via exceptional items. We
refer to extraordinary and exceptional items as classification items (CIs). Consistent with predictions, our
results reveal an increase in the practice of classificatory smoothing post-FRS3. The rise is due to absolute
unexpected earnings inducing a higher level of classificatory smoothing through exceptional items post-
FRS3. As a consequence, earnings before exceptional items appear significantly more persistent in the
post-FRS3 period. These results are robust to alternative measures, test specifications, and explanations.
Our study makes several contributions to the earnings management literature. First, we provide
valuable information for accounting standard setters by shedding light on the impact of regulatory
intervention on firms income smoothing practices. Consistent with our hypotheses, we find that FRS3
brought about an increase in the more transparent and less costly practice of classificatory income
smoothing. Second, we extend the growing literature on disclosures of adjusted earnings (Bhattacharya,
Black, Christensen, and Larson 2003, Choi et al. 2005). We find that within the UK, wider flexibility over
classifications of exceptional items, enhanced disclosure of such items, and the opportunity to disclose
alternative measures of performance, increased the practice of classificatory smoothing. Third, we extend
prior research on the information content of adjusted earnings (Elliot and Hanna 1996, Bradshaw and Sloan
2002, Eames and Sepe 2003). Consistent with evidence on the superior value relevance of earnings before
non-recurring items, we find that earnings adjusted for extraordinary and exceptional items are more
persistent than net income, especially post-FRS3. Our results suggest that FRS3 enhanced the role of
classificatory choices in highlighting permanent earnings. Fourth, we add to prior research on
classificatory smoothing, which in contrast to that on abnormal working capital accruals, focuses mainly on
small datasets (Beattie et al. 1994, Godfrey and Jones 1999). We extend Beattie et al. (1994), who examine
classificatory choices within an incentives-based framework, by providing consistent evidence based on a
large sample. We also make a methodological contribution by developing a measure of the magnitude of
classificatory smoothing that relaxes the assumption of a single smoothing object across firms. This is
8/4/2019 Classificatory Income Smoothing
7/50
7
important given that disclosure of alternative EPS is optional and firms may choose any level of profit as an
alternative to net income.
Exploring the impact of FRS3 on classificatory smoothing is timely in relation to current policy
issues. In their current convergence project for reporting financial performance, the International
Accounting Standards Board (IASB) and the US Financial Accounting Standards Board (FASB) have
decided that financial statements should include either a single statement of comprehensive income, or two
statementsan income statement and a statement of recognised income and expense (IASB 2005). The
Boards have yet to determine a common approach for disaggregating financial information and the totals
and subtotals to report in the income statement. At the same time, US regulators are concerned about
increasing investor reliance on adjusted earnings (e.g. pro forma, Street earnings) disclosed in earnings
announcements. In the light of these current developments, FRS3 offers a timely policy experiment.
2. Prior literature, institutional overview, and development of hypotheses
2.1 Literature review
The income smoothing literature begins with the income smoothing hypothesis of Gordon (1964).
Within his framework income smoothing arises as rational behaviour based on assumptions that: a)
managers maximize their utility; b) earnings fluctuations and unpredictability of earnings determine market
risk measures; c) the dividend payout ratio determines firm value; and d) managers utility depends on firm
value. This literature makes the implicit assumption that the market is inefficient, i.e. that it functionally
fixates on bottom line earnings, irrespective of the accounting choices involved in determining income.
However, it is not necessary for the market to be inefficient for income smoothing behaviour to exist. It is
sufficient that managers believe the market is inefficient. In either case, managers smooth income to
reduce the actual or perceived riskiness of the firm arising from a cause-and-effect relation between
earnings fluctuations and market risk (Moses 1987, p.366). Lev and Kunitzky (1974) provide evidence that
the extent of income smoothing is correlated with both total and systematic risk.
In contrast to this original income smoothing hypothesis, positive accounting theory (PAT),
assumes the market is efficient and can detect smoothing behaviour (Watts and Zimmerman 1978, 1979).
PAT rationalizes earnings management through positive contracting costs. Performance-related contracts
8/4/2019 Classificatory Income Smoothing
8/50
8
induce managers to make accounting choices to maximize their own wealth or to minimize agency costs
with contracting parties (shareholders or debtholders). Alternatively, managers make accounting choices to
reduce political costs associated with public visibility or regulation. The earliest smoothing study adopting
a positive accounting framework is Moses (1987). Moses documents a significant association between the
extent of smoothing and proxies for political costs and agency costs of equity.
Holthausen and Leftwich (1983) propose the information theory of accounting choice as an
alternative to the contracting theory. The information theory predicts that managers choose accounting
techniques to convey information about expected future cash flows. The literature on the information
theory identifies smoothing as an explicit strategy managers use to reveal their expectations (Schipper
1989). Thus, under the information theory, for income smoothing to be effective investors must observe
managers accounting choices. We focus on the information perspective of accounting choice as it links to
the regulatory framework for reporting financial performance. This is because the former addresses the
issue of information asymmetry and limited communication between managers and the market (Fields, Lys,
and Vincent 2001), while the latter determines the channels of communication.
How do firms smooth income? Ronen and Sadan (1981) show that managers can smooth income
either inter-temporally, by timing actual transactions to influence the income of particular periods, or via
classifications of non-recurring items. Smoothing via classificatory items is feasible when the earnings
figure managers seek to smooththe smoothing objectis a level of profit other than net income. Barnea,
Ronen and Sadan (1976) and Ronen and Sadan (1981) argue that ordinary income is a desirable smoothing
object, as transitory income components leave it unaffected and it dominates net income in predicting
future profitability. Michelson, JordanWagner and Wootton (1995) find that firms that smooth ordinary
income have significantly lower betas and higher equity market values. Using an incentives-based
framework, Beattie et al. (1994) examine classificatory smoothing via extraordinary items by UK firms.
Consistent with Moses (1987), they find a significant association between the extent of classificatory
smoothing, agency costs, and accounting risk. Godfrey and Jones (1999) verify the link between political
costs and classificatory income smoothing. They use the redefinition of ordinary activities for Australian
firms in 1989 to construct a measure of classificatory smoothing through extraordinary items and find that
8/4/2019 Classificatory Income Smoothing
9/50
9
managers of companies with high labour related political costs (high employee union membership) attempt
to smooth net operating profit through classifications or recurring gains and losses.
Recent research investigates adjusted earnings disclosures as a way of increasing the predictive
power of reported earnings for future performance. Such disclosures enable a form of classificatory
smoothing that is less restrictive than classificatory smoothing through extraordinary items, as managers
may exclude any income components they consider transitory from the adjusted earnings figure. Evidence
from Bradshaw and Sloan (2002) and Bhattacharya et al. (2003) suggests that pro forma earnings
disclosures by US firms are more informative and more persistent than GAAP earnings. However, the ad-
hoc nature of pro forma disclosures and the fact that they are outside the audited profit and loss statement
have raised concerns about opportunistic classificatory choices. Doyle, Lundholm and Soliman (2003) and
Landsman, Miller and Yeh (2006) substantiate these concerns with evidence that US firms remove value
relevant items from pro forma earnings.
The main difference in adjusted earnings reporting between the UK and the US is that FRS3
regulates UK practice. We consider this difference in examining classificatory smoothing by UK firms
post-FRS3. Similar to Beattie et al. (1994) and Godfrey and Jones (1999), we use an incentives-based
framework to examine classificatory smoothing. Unlike these two studies, which are constrained by the
use of small datasets due to research design limitations,2 we use a large panel of UK firms, as we seek to
examine the impact of FRS3 on the overall practice of classificatory smoothing. Apart from extending
prior research, our study is the first to examine the impact of a change in financial performance reporting
regime on classificatory income smoothing.
2 Beattie et al. (1994) use hand collected data from annual reports for 163 companies included in the 198990 survey
of UK published accounts and reporting extraordinary and exceptional items. Godfrey and Jones (1999) use a sample
of 58 Australian listed firms that restate their 1989 extraordinary item figure in their 1990 accounts, and have
available data for 19801990.
8/4/2019 Classificatory Income Smoothing
10/50
10
2.2 FRS3 and classificatory choices
By redefining ordinary activities3
to include the effects of any event irrespective of its frequency or
unusual nature, FRS3 effectively outlawed extraordinary items. In effect, all items previously classified as
extraordinary became exceptional items. The disclosure of exceptional items also changed substantially.
Instead of being aggregated in one heading as under SSAP6, FRS3 required firms to distinguish between
operating and non-operating exceptionals and to disclose operating exceptionals either on the face of the
profit and loss statement or in a note. For each item, firms had to provide an adequate description to allow
users to understand its nature (FRS319). FRS3 required disclosure of non-operating exceptionals under
separate headings after operating profit and before interest. Non-operating exceptional items include
profits or losses on the sale or termination of an operation, costs of a fundamental reorganization or
restructuring and profits or losses on the disposal of fixed assets.
The effect of FRS3 on exceptional items extended beyond these disclosure requirements. By
allowing firms to disclose alternative EPS on other levels of profit, FRS3 enabled firms to remove
exceptional items from alternative EPS and to highlight a more meaningful measure of earnings. FRS3
widened the scope for classificatory choices further through its broader definition of exceptional items.
FRS3 defined exceptional items as any material items which derive from events or transactions that fall
within the ordinary activities of the reporting entity and need to be disclosed by virtue of their size or
incidence if financial statements are to give a true and fair view (FRS35). Apart from the three non-
operating exceptional items, this definition of exceptional items allowed UK firms to classify as
exceptional not only highly unusual items falling within the ordinary activities of the firm but also items
arising regularly, but being unusually large in the current year.4
In practice, firms disclose a variety of
items as operating exceptional. An abbreviated list includes: reorganization costs (e.g. redundancy costs),
restructuring costs, provisions for permanent diminution in the value of property and land, provisions for
3Ordinary activities are any activities, which are undertaken by a reporting entity as part of its business and such
related activities in which the reporting entity engages in furtherance of, incidental to, or arising from, these activities.
They include the effects on the reporting entity of any event in the various environments in which it operates,
including the political, regulatory, economic and geographical environments irrespective of the frequency or unusual
nature of the events (FRS32).4
In the latter case the exceptional effect on the results is not the whole amount of the item, but only the excess over a
normal amount. For example, where a firm revises the estimated useful life of an asset, the firm can classify the
excess amount of the revised depreciation charge over the old charge, if material, as exceptional.
8/4/2019 Classificatory Income Smoothing
11/50
11
business rationalization, provisions for environmental liabilities, current asset write downs, provisions for
losses in contracts, and goodwill written-off. Instead of a uniform list of exceptionals, FRS3 provided a
definition of exceptional items that allowed managers to classify items according to the nature of the firms
operations.
2.3 Development of hypotheses
Classificatory choices have three advantages over inter-temporal income smoothing devices. First,
they do not affect net income and are less likely to be challenged by auditors or trigger GAAP violations.
Second, they do not flow through the accounting system and do not affect future period income. Third,
they do not have tax implications.
UK firms have been able to engage in classificatory smoothing since the introduction in 1974 of
SSAP6, which required the separate disclosure of extraordinary items in the profit and loss statement after
ordinary income. As EPS was calculated on ordinary income, UK firms could classify items below the line
of basic EPS to smooth reported earnings. Empirical evidence documents the widespread use of
extraordinary items as an income-smoothing device (Smith 1992, Beattie et al. 1994). Pope and Walker
(1999) report evidence of firms using extraordinary items to classify bad news earnings components,
mainly in the form of write-offs of large transitory losses. Consistent with this evidence, Peasnell, Pope
and Young (2000) report that pre-FRS3 the majority of extraordinary items reported by UK firms had the
effect of increasing income, with their mean value approximating 2.7 million.
FRS3 increased the flexibility of classificatory smoothing. Evidence that firms exploited this
flexibility comes from Choi et al. (2005), who document a gradual increase in the number of firms
reporting alternative EPS on pre-exceptional earnings levels. They find the main reason for disclosures of
alternative EPS is the low information content of net income for sustainable performance. Furthermore,
they document a gradual increase in the magnitude of negative exceptional items post-FRS3, due mainly to
the growth of negative operating exceptionals. This evidence is consistent with an increase in total CIs
post-FRS3 resulting from the desire to report a better indicator of sustainable profitability.
In addition to increasing flexibility, FRS3s disclosure requirements increased the transparency of
classificatory smoothing, allowing investors to identify and assess the adjustments made to arrive at the
8/4/2019 Classificatory Income Smoothing
12/50
12
alternative earnings figure. Higher disclosure helps investors and analysts identify core profitability and
make more accurate forecasts of future performance (Penman 2003). Acker, Horton, and Tonks (2002)
examine the effect of FRS3 on analyst forecast errors based on pre-exceptional levels of earnings.5
They
find that after the first year of FRS3, pre-exceptional earnings are associated with lower forecast errors.
Lin (2002) examines whether analyst forecasts reflect the post-FRS3 information in unexpected earnings
components. He finds that analysts revise their current and future earnings forecasts to impound
information contained in non-recurring items and to improve the accuracy of their predictions.
A significant implication of higher transparency is that it constrains opportunistic classificatory
choices, as it increases the likelihood that investors can detect attempts to remove income components
opportunistically. Opportunistic firms may prefer less transparent income smoothing devices. However,
for firms wishing to convey a more sustainable measure of performance, classificatory smoothing through
FRS3 provisions offers an efficient communication channel.
Higher flexibility and disclosure of classificatory choices post-FRS3 allow for an increase in
classificatory smoothing and the provision of a better metric of persistent profitability. Accordingly, we
predict that FRS3 increased the practice of classificatory smoothing by UK firms to highlight recurrent
profitability and to signal future performance. We form two hypotheses, the first on the overall effect of
FRS3 on the practice of classificatory smoothing, the second on the incentive underlying this effect, which
stems from a consequence of income smoothing,
Hypothesis 1: FRS3 increased the practice of classificatory smoothing by UK firms.
Hypothesis 2: FRS3 increased the use of classifications of non-recurring items to highlight
persistent earnings.
5Analysts forecast earnings before exceptional items post FRS3, in contrast to the pre-FRS3 regime in which analysts
excluded only extraordinary items.
8/4/2019 Classificatory Income Smoothing
13/50
13
3. Research design
3.1 Measuring classificatory smoothing
We measure classificatory smoothing as the extent to which removing extraordinary and
exceptional items (CIs) reduces the variability of reported earnings. We start by constructing an index of
income smoothing following Eckel (1981) as,
REV
EARNINGSSI
=
where EARNINGS is the standard deviation of earnings and REV is the standard deviation of sales
revenue. Scaling by sales volatility controls for economic performance across firms.6
Low values of SI
indicate that, ceteris paribus, managers smooth reported earnings. We calculate the ratio at the firm level
and separately for pre- and post-FRS3 periods.
We follow the framework of Ronen and Sadan (1981) on the interrelation between smoothing
objects and smoothing dimensions to measure the level of classificatory income smoothing. In this
framework, firms can smooth net income only inter-temporally, but can smooth levels of earnings other
than net income by a combination of inter-temporal and classificatory smoothing (see the Appendix).
Based on this framework we calculate the smoothing index on net income )(EARN and on alternative
levels of earnings that are potential smoothing objects. We calculate a classificatory smoothing index
)(CSI by subtracting the smoothing index of the smoothing object from the smoothing index of net
income. Positive values of CSI capture smoothing of reported earnings through classifications of
extraordinary and exceptional items. We calculate CSI separately for pre- and post-FRS3 periods.
Pre-FRS3, consistent with prior research (Beattie et al. 1994), we assume the smoothing object is
ordinary income before extraordinary items ),(EARNbXI communicated to investors through basic EPS.
We calculate smoothing indexes of EARN )( 1SI and EARNbXI ).( 2SI Table 1 describes the
calculations. CSI is the difference between1
SI and the minimum of1
SI and2
SI , and captures the
magnitude of smoothing resulting from classifications of extraordinary items.
6We repeat the analysis using cash sales (total revenue minus change in receivables) as the denominator to subtract
the revenue component that may be subject to artificial smoothing (through accruals). Our main results remain
unaltered using this alternative specification.
8/4/2019 Classificatory Income Smoothing
14/50
14
Identifying the smoothing object pre-FRS3 is straightforward, but this is not the case post-FRS3.
The option to disclose alternative EPS on levels of earnings other than net income creates the possibility of
multiple smoothing objects. We examine two smoothing objects besides :EARN earnings before non-
operating exceptional items )(EARNbNOEI and earnings before all (both non-operating and operating)
exceptional items ).(EARNbEI For the post-FRS3 period we consider two scenarios. In the first, we
assume firms smooth ,EARN ,EARNbNOEI or .EARNbEI We identify the least volatile figure as the
smoothing object. We calculate smoothing indices for EARN )(1
SI , EARNbNOEI )(3
SI and EARNbEI
).(4
SI CSI is the difference between1
SI and the minimum of ,1
SI ,3
SI and4
SI and captures the
magnitude of smoothing resulting from classifications of either non-operating exceptional items when the
smoothing object is ,EARNbNOEI or all exceptional items, when the smoothing object is .EARNbEI This
scenario accommodates evidence by Choi et al. (2005) of a simultaneous increase in the frequency of
alternative EPS disclosures from 1994 to 1996 and a gradual increase in operating exceptional charges.
In the second scenario, we assume firms smooth either EARN or .EARNbNOEI We use the same
method to identify the smoothing object amongst the two alternatives. CSI is the difference between1
SI
and the minimum of1
SI and3
SI and captures the magnitude of smoothing resulting from classifications of
non-operating exceptional items when EARNbNOEI is the smoothing object. EARNbNOEI is similar to
headline earnings that the IIMR (Institute of Investment Management and Research) introduced in 1993 as
an earnings metric that excludes all exceptional items formerly classified as extraordinary and exceptionals
that relate to capital values. To this extent, the second scenario is mostly plausible for the initial period
following the implementation of FRS3.
3.2 Research design for classificatory smoothing
To test the impact of FRS3 on the level of classificatory smoothing we use descriptive statistics and
the following multivariate specification,
21,p;N,...,iuLIa
PIaPIaSIZEaLEVaRISKaPERMLOSSa
GAINLOSSaDIVERGENCE3FRSaDIVERGENCEa3FRSaaCSI
pipi
pipipipipipi
pipipipipi
==++
++++++
++++=
1,
)(
,,11
,10,9,8,7,6,5
,4,3,2,10,
(1)
8/4/2019 Classificatory Income Smoothing
15/50
15
In equation (1), p = 1 refers to the pre-FRS3 period, p = 2 to the post-FRS3 period. 3FRS is a GAAP
regime indicator, taking the value 0 when p = 1 and 1 when p = 2. Equation (1) is therefore a panel
regression involving a cross-section ofN firms, each with two observations (pre- and post-FRS3). We
discuss the remaining regressors below. We estimate equation (1) for each scenario of the post-FRS3
period separately. We calculate all explanatory variables annually and use pre- and post-FRS3 averages.
Our predictions focus on the information perspective of income smoothing. By smoothing income,
managers produce a stable and predictable earnings stream to facilitate forecasts of future profitability. In
this signalling framework, Moses (1987) argues that managers seek to report earnings that are closer to
expectations and that incentives to smooth income increase with the divergence between actual and
expected earnings. Accordingly, we expect a positive association between CSI and the extent to which
pre-managed earnings diverge from expectations. We introduce DIVERGENCE as the absolute value of
unexpected earnings to capture this effect.7,8
To the extent high deviations from expected earnings induced
classificatory smoothing pre-FRS3, we expect2
a to be positive. If UK firms use classifications of non-
recurring items to a greater extent post-FRS3 to report earnings closer to expectations and highlight
persistent profitability, we expect the positive association between DIVERGENCE and CSI to be more
pronounced post-FRS3. We add an interaction term, ,3 DIVERGENCEFRS to capture this structural
shift. Consistent with our second hypothesis, we expect3
a to be positive in both scenarios.
The practice of classificatory smoothing relies on the firms propensity to highlight adjusted
earnings metrics. We add two proxies to capture this. Lougee and Marquardt (2004) find that firms
reporting GAAP losses are more likely to disclose pro forma earnings. This incentive is particularly keen
when adjusted earnings are positive, as managers wish to highlight an indicator of the firms earnings
generating ability that is unaffected by large transitory losses. Choi et al. (2005) document a positive
association between the likelihood of alternative EPS disclosure and the frequency of firms with net income
losses and positive adjusted earnings. As a result, we add an indicator, ,GAINLOSS of cases where net
income is negative while adjusted earnings are positive as a proxy for the inclination to highlight pre-
7
Section 4.2 gives the precise definitions of all variables.8DIVERGENCE also serves as a control, as deviations of pre-managed earnings from expectations affect the
remaining smoothing incentives. Moses (1987) provides evidence that larger deviations from expectations are
necessary to induce certain smoothing incentives (e.g. political costs).
8/4/2019 Classificatory Income Smoothing
16/50
16
exceptional earnings. We expect a positive association between GAINLOSS and .CSI Choi et al. (2005)
further argue that firms are unlikely to disclose an adjusted earnings metric when this metric is negative as
it is less likely to be informative about their future prospects. Consistent with their claim, they find that the
probability of alternative EPS disclosure is lower when adjusted earnings are negative. Accordingly, as a
second proxy we add an indicator of negative pre-exceptional earnings, ,PERMLOSS and expect a
negative association between PERMLOSS and .CSI
In addition to dampening deviations from expected earnings to signal future performance,
managers smooth income to reduce actual or perceived riskiness resulting from the relationship between
earnings volatility and risk (Moses 1987, Lev and Kunitzky 1974, Beattie et al. 1994, Michelson et al.
1995). Accordingly, we include market risk ))(( RISK in equation (1). We expect a positive association
between CSI and ),(RISK as riskier firms have greater incentives to smooth earnings. Consistent with
PAT, we add gearing )(LEV and size )(SIZE as proxies for agency costs of equity and political costs.
More highly levered firms are more likely to smooth ordinary (or operating) income to avoid fluctuations
that might breach debt covenants, especially in relation to the interest cover ratio. Even in the absence of
interest cover constraints, managers may smooth income to create an impression of financial stability.
Therefore, we expect a positive association between CSI and .LEV Larger firms have higher incentives
to dampen earnings fluctuations and minimize the costs of potential external intervention, thus we expect a
positive association between CSI and .SIZE Following Ashari, Koh, Tan and Wong (1994), we control
for firm profitability (profitability index, ).PI Fluctuations in income have a more severe impact on low
profitability firms giving them a stronger motive to smooth income. We also control for the change in
profitability ( )PI in response to Whites (1970) evidence that firms with declining profitability tend to
smooth income.
Finally, removing a classification item alters both the level and the variability of reported earnings.
Moses (1987) provides evidence that managers are concerned about the joint effect of their accounting
choices on earnings levels and variability. Since managers are concerned about both effects and some of
the factors in our model could motivate managers to adjust the level of earnings, we add a control for the
impact that removing CIs has on the level of reported income ).(LI Adding LI is particularly important in
8/4/2019 Classificatory Income Smoothing
17/50
17
view of the evidence of Beattie et al. (1994) that incentives to smooth depend on the size of the effect of
CIs on the level of earnings (i.e. this effect should reach a specific level to trigger smoothing behaviour).
Including LI also controls for the ability to engage in classificatory smoothing. This is because
classifications of non-recurring items are to some extent determined by events giving rise to their
occurrence (e.g. restructurings, divestitures, events triggering fair value adjustments and changes in
accounting estimates etc.).
Because CSI takes only zero and positive values, equation (1) belongs to the class of censored
regression models. Even though CSI is a continuous variable over strictly positive values, it takes the
value zero with positive probability, in the sense that for some managers the optimal choice may be zero
classificatory smoothing. This type of response variable is a corner solution outcome (Wooldridge 2002).
Accordingly, we estimate the coefficients of equation (1) using a corner solution Tobit model.
3.3 Earnings persistence tests
To explore whether an increase in classificatory smoothing post-FRS3 increases the persistence of
pre-exceptional earnings and highlights a superior indicator of sustainable profitability, we complement our
multivariate analysis with earnings persistence tests. First, we assess the extent to which removing
classification items increases the persistence of reported earnings in the pre- and post-FRS3 periods.
Second, we test whether this effect increases as a result of the introduction of FRS3. We initially use
earnings persistence regressions for different levels of earnings )(EARNADJ separately for pre- and post-
FRS3 periods and then add an interaction term between contemporaneous earnings and 3FRS to capture
any structural break in earnings persistence after the enforcement of FRS3
titiEARNADJEARNADJ
,101, +=
+(2)
titititi vEARNADJFRSEARNADJEARNADJ ,,2,101, 3 +++=+ (3)
We use three measures of :EARNADJ
8/4/2019 Classificatory Income Smoothing
18/50
18
Consistent with the second hypothesis, we expect2
to be positive for all three measures.
4. Sample and definition of variables
4.1 Sample
We use data for all UK (dead and live) non-financial listed firms from Datastream for the period
1986 until the enforcement of FRS10 in December 1998. Choosing FRS10 as the sample period cut-off is
important due the potential effect of FRS10 on classificatory smoothing. FRS10 required UK firms to
charge amortization of goodwill and intangibles assets to the income statement, providing an additional
motive to managers to disclose alternative EPS on pre-exceptional earnings in order to deduct the
substantial charge of amortization. As our objective is to examine the effect of FRS3, the sample period
ends with the introduction of FRS10 (23/12/1998).9
We employ a balanced sample to assess whether there is a structural break in the practice of
classificatory smoothing caused by the introduction of FRS3. With a balanced sample design, each firm
has at least one observation in both the pre- and post-FRS3 period. This way, each sample firm serves as
its own control, minimising the effects of temporal differences in sample composition on the results
(Peasnell et al. 2000). The balanced sample includes 9,459 observations for 993 firms. To calculate firm
and period specific classificatory income smoothing indices we keep firms with at least two observations in
both the pre- and post-FRS3 periods. The final sample is 914 firms and 9,222 firmyear observations.
9To the extent firms anticipated FRS10, this may have encouraged them to release an alternative EPS on earnings
before all exceptional items before FRS10. Thus, there is a risk of capturing part of the FRS10 effect on the practice
of classificatory smoothing. As FRS10 was issued in December 1997, the risk relates mainly to 1998 observations.
Our main results are robust to removing these observations.
Earnings level Definition
1. 1EARNADJ Pre-FRS3: Earnings before extraordinary items )(EARNbXI
Post-FRS3: Earnings before non-operating exceptional items )(EARNbNOEI
2. 2EARNADJ Pre-FRS3: Earnings before extraordinary items )(EARNbXI
Post-FRS3: Earnings before all exceptional items )(EARNbEI
3. 3EARNADJ Pre-FRS3: Earnings before extraordinary and exceptional items )&( EIEARNbXI
Post-FRS3: Earnings before all exceptional items )(EARNbEI
8/4/2019 Classificatory Income Smoothing
19/50
19
4.2 Definition of variables
For the empirical analysis we use five earnings measures: earnings after extraordinary items
),(EARN earnings before extraordinary items ),(EARNbXI earnings before extraordinary and exceptional
items ),&( EIEARNbXI earnings before non-operating exceptional items ),(EARNbNOEI and earnings
before all exceptional items ).(EARNbEI10
DIVERGENCE is the absolute value of the difference between
pre-managed earnings and expected earnings scaled by total sales (DS104). We use EARN as pre-
managed earnings, since classificatory choices do not affect net income. For expected earnings, we assume
a random walk model and choose the earnings level in line with the smoothing object. Pre-FRS3 expected
earnings is lagged .EARNbXI Post-FRS3, expected earnings is either lagged EARNbNOEI or lagged
.EARNbEI
Pre-FRS3 GAINLOSS equals 1 when the firm reports net income losses )0( EARNbXI 0 otherwise. Post-FRS3 GAINLOSS equals 1 when the firm reports net income
losses )0( EARNbNOEI or ),0>EARNbEI 0
otherwise. Pre-FRS3 PERMLOSS equals 1 if ,0
8/4/2019 Classificatory Income Smoothing
20/50
20
lagged assets. For the change in profitability index )( PI we use decile portfolios based on the annual
change in net income scaled by lagged assets.
We measure the level impact of classification items )(LI as,
We winsorize all variables at the 0.5% and 99.5% percentiles.
5. Results
5.1 The effect of FRS3 on classificatory income smoothing
5.1.1 Descriptive statistics
Table 2, Panel A reports descriptive statistics for extraordinary items )(XI and exceptional items
)(EI for the pre-FRS3 period and for non-operating exceptional items )(NOEI and operating exceptional
items )(OEI for the post-FRS3 period. Results show that pre-FRS3 mean XI (0.004) and EI (0.003)
are both negative and significant. Contrary to the perceptions of regulators, the results do not show a
dominance of negative extraordinary items and positive exceptional items pre-FRS3. Similar to the pre-
FRS3 period, mean NOEI (0.007) and OEI (0.006) post-FRS3 are negative and significant but are
larger in magnitude. Since FRS3 treats most former XI as ,NOEI we calculate their mean difference as
11Datastream reports exceptional and extraordinary items as negative when they are costs or losses, and positive when
they are revenues or profits.12
We repeat the main tests using deciles portfolios based on LI, to account for the negative skewness of classification
items. The core results remain using this alternative measure.
Pre-FRS3: The income-increasing (decreasing) effect of negative (positive) extraordinary items
).( XI11
XI is DS193.
Post-FRS3: Scenario A: The income-increasing (decreasing) effect of either negative (positive) non-
operating exceptionals )( NOEI or negative (positive) exceptionals )( EI in line with
the smoothing object. EI is DS194.
Scenario B: The income-increasing (decreasing) effect of negative (positive) non-operating
exceptional items ).( NOEI NOEI is D1097.DS10941083 DS
LI is scaled by lagged assets and is zero when the classificatory smoothing index )(CSI is zero.12
8/4/2019 Classificatory Income Smoothing
21/50
21
well as the mean difference between EI pre-FRS3 and OEI post-FRS3. Both differences are significant,
indicating an increase in the magnitude of negative exceptional items post-FRS3.
To provide information on variation over time, Table 2, Panel B reports annual statistics for XI
and EI pre-FRS3 and for NOEI and OEI post-FRS3. Analysis of mean XI indicates increasingly
negative values starting in 1990 and reaching 0.011 in the 92/93 pre-FRS3 observations. These results are
consistent with prior evidence on the income-increasing role of XI reported by UK firms (Smith 1992,
Pope and Walker 1999, Peasnell et al. 2000). As mean earnings before extraordinary items in the 92/93
pre-FRS3 observations (not tabulated) is 0.031, the income-increasing effect of XI for this period is
substantial. EI displays a similar pattern to ,XI despite the pre-FRS3 position of exceptional items above
ordinary income. Negative values of mean EI start in 1990 and reach 0.010 in the 92/93 pre-FRS3
observations.13
Results show that NOEI and OEI are negative throughout the post-FRS3 period, totalling
0.015 immediately after the implementation of FRS3 and peaking at 0.022 in 1998. Negative averages
for both NOEI and OEI post-FRS3 are consistent with evidence of firms disclosing alternative EPS
reported by Choi et al. (2005). Results further indicate that even though directly following the
implementation of FRS3 the ratio of OEI to NOEI is 0.50 (0.005/0.010), the ratio is 2 (0.006/0.003)
in 1997 and 1 (0.011/0.011) in 1998. This suggests an increase in the relative magnitude of negative
,OEI as Choi et al. (2005) also document.
Table 2, Panel C reports the frequency of positive, negative, and zero XI and EI for the pre-FRS3
period and ofNOEI and OEI for the post-FRS3 period. In Panel D we repeat the analysis by year. For
both XI and ,NOEI the majority of observations are zero in both periods (55 and 53 percent), indicating
that a substantial proportion of firms do not report these items. For firms disclosing such items, the
majority (67 percent) ofXI are negative. A binomial (two-sided) test for the difference in frequencies of
positive and negative extraordinary items conditional on non-zero disclosures is highly significant,
confirming the predominance of extraordinary losses. This holds for most pre-FRS3 accounting periods.
In contrast, NOEI has roughly equal frequencies of positive and negative values. A binomial test for the
13The economic recession prevailing in the UK in the early 1990s affects trends for this period. The average growth
rate of real GDP for 19901993 was approximately 0.59 percent compared to 4.44 percent for 19861989.
8/4/2019 Classificatory Income Smoothing
22/50
22
overall post-FRS3 period does not reject equality of the two frequencies ).805.0( =p Annual analysis
provides evidence of asymmetry in the distribution ofNOEIonly in the first period reporting under FRS3
and in 1997.
For EI and OEI the frequency of zero observations is 19 and 17 percent, suggesting that only a
small proportion of the population do not disclose such items. Where firms report EI pre-FRS3, the
majority are positive (57 percent). The binomial test is significant, confirming EI is positively skewed.
Therefore, focusing on the signs of the items, the evidence appears to justify regulatory concerns over the
prevalence of negative XI and positive EI pre-FRS3. Analysis by year in Panel D shows that positive
skewness is not evident in the distribution ofEI in the two periods preceding the introduction of FRS3.
The shift in frequencies might be due to restructuring costs, since in 1991 the ASB required UK firms to
classify these costs as exceptional rather than extraordinary. Finally, similar to the pattern of ,NOEI OEI
appears equally distributed between positive and negative values post-FRS3. A binomial test for the
overall post-FRS3 period does not reject equality of the two frequencies ).332.0( =p FRS3 therefore
appears to have successfully removed the skewness in the distribution of classification items. Analysis by
year, however, shows that in 1997 and 1998 the majority of OEI are negative. The gradual rise in the
frequency of negative operating exceptions might reflect managerial attempts to release an alternative EPS
on earnings before all exceptional items in anticipation of FRS10.
Moving to the core empirical analysis for the first hypothesis, Table 3 reports the mean and median
values of the classificatory smoothing index )(CSI partitioned into the pre- and post-FRS3 periods. Pre-
FRS3, CSI measures income smoothing from classifications of .XI Post-FRS3, the index varies with the
scenario we examine. In Scenario A, CSI measures income smoothing from classifications of either
NOEI or both NOEI and .OEI In Scenario B, CSI measures income smoothing from classifications of
.NOEI For the pre-FRS3 period, Table 3 shows a significant mean and median CSI of 0.063 and 0.004.
For the majority of firms (527/914 = 58%) CSI is positive. These results confirm the smoothing property
of XI for ordinary income documented by Beattie et al. (1994). Post-FRS3 in Scenario A, mean and
median CSI increase to 0.157 and 0.023. The difference across periods is highly significant for both the
mean and the median. The number of uncensored observations )0( >CSI increases to 715, indicating a
8/4/2019 Classificatory Income Smoothing
23/50
23
substantial rise in the frequency of firms engaging in classificatory smoothing via all exceptional items
post-FRS3. In Scenario B, mean and median CSI increase to 0.102 and 0.006. The increase in mean is
significant. The frequency of uncensored observations )0( >CSI in this scenario increases to 550. These
results provide evidence consistent with our first hypothesis of an increase in classificatory income
smoothing post-FRS3, especially in Scenario A.
Table 4 reports descriptive statistics for the control variables of equation (1) partitioned into the
pre- and post-FRS3 periods. Post-FRS3 there is evidence of a significant decrease in DIVERGENCE,
especially in Scenario A, the frequency of firms with negative pre-exceptional earnings
,(PERMLOSS Scenario A) and ),(RISK and of a significant increase in the frequency of firms with pre-
exceptional earnings and net income losses ,(GAINLOSS Scenario A). The impact of CIs on the level of
reported earnings )(LI is significantly positive, reflecting the income-increasing effect of CIs in both
periods and in both scenarios. Calculated in line with Scenario A, there evidence of a significant rise in the
income-increasing effect of CIs post-FRS3.
5.1.2 Multivariate analysis
Table 5 reports Tobit regression results for the effect of FRS3 on .CSI Model 1 contains the
results of estimating equation (1) in Scenario A. As is common for corner solution applications we also
report the partial effects (economic significance) of the explanatory variables. DIVERGENCE is positive
and significant ),41.2,278.0( =t suggesting a positive association between deviations of net income from
lagged ordinary income and classificatory smoothing via extraordinary items pre-FRS3.
DIVERGENCEFRS 3 is positive and significant ),882,702.0( .t= increasing the coefficient on
DIVERGENCE to 0.980 post-FRS3. This increase suggests that deviations of net income from lagged
earnings before all exceptional items induce a significantly higher level of classificatory smoothing post-
FRS3. ,GAINLOSS ,PERMLOSS ),(RISK ,LEV ,SIZE PI and PI are all significant and in
accordance with predicted signs. LI is positive and significant )046,863.3( .t= , indicating that increasing
8/4/2019 Classificatory Income Smoothing
24/50
24
adjusted earnings through income-increasing CIs (e.g. restructuring costs, losses from disposals of assets,
exceptional operating costs) also result in smoother earnings.14
Model 2 in Table 5 reports the results of estimating equation (1) in Scenario B.
DIVERGENCEFRS 3 is positive and significant ),152,373.0( .t= suggesting that deviations of net
income from lagged earnings before non-operating exceptional items induce a higher level of classificatory
smoothing post-FRS3. The magnitude of the coefficient on DIVERGENCEFRS 3 is substantially lower
than in Model 1.
In summary, the results in Table 5 provide evidence in favour of our second hypothesis of an
increase in classificatory income smoothing post-FRS3 to report earnings closer to expectations and
highlight recurrent profitability. The increase is evident where we allow a combination of smoothing
objects between EARN and EARNbNOEI in Scenario B and is substantially higher when we allow for the
likelihood that some firms smooth EARNbEI in Scenario A.
5.2 Additional analyses
5.2.1 Refining the estimation ofCSI
For the initial calculation of CSI we require firms to have at least two observations in both the pre-
and post-FRS3 periods. This is the minimum requirement to calculate standard deviations. For more
efficient calculation of standard deviations, we repeat the analysis of Table 5 on a subset of 609 firms
approximately 67 percent of the original samplewith at least five observations in each period. Table 6
reports the regressions results. The coefficients on DIVERGENCEFRS 3 are positive and significant in
both scenarios ( 24.3,410.1 =t in Scenario A and 412,754.0 .t= in Scenario B), and higher than those in
Table 5. Consistent with our main predictions, the results provide strong evidence of a significant rise in
the level of classificatory smoothing post-FRS3 to reduce deviations of net income from expectations.
14By construction, CSI is left censored so we eliminate cases where classificatory choices magnify the volatility of
reported earnings. As a result, we are unable to capture the effect of income-increasing CIs that increase the
variability of earnings. This biases results against finding a negative association betweenLI and CSI. To ensurerobustness of this result, we redefine CSIto include negative values and define LIaccordingly. We then repeat the
regression analysis using OLS instead of censored Tobit. The result of a positive association between CSIand LI
remains. All other explanatory variables, with the exception of ),(RISK remain significant and in accordance with
predicted signs.
8/4/2019 Classificatory Income Smoothing
25/50
25
5.2.2 Other structural shifts
To test whether structural breaks for any of the remaining drivers of variation in classificatory
smoothing confound our core findings, we allow for interaction terms between these drivers and .3FRS
Table 7 reports the results. DIVERGENCEFRS 3 remains positive and significant under both scenarios
46.2,667.0( =t in Scenario A and 70.1,329.0 =t in Scenario B). There is also evidence of a structural
shift in .LI In Scenario A the coefficient on LI increases from 2.056 pre-FRS3 to 4.515 )459.2056.2( +
post-FRS3. The shift is similar in Scenario B and is more pronounced when we repeat the analysis for
firms with at least five observations in both the pre- and post-FRS3 period. The increase in the positive
association between LI and CSI suggests that firms use CIs to a greater extent post-FRS3 to increase and
smooth reported earnings simultaneously.
5.2.3 Alternative explanation for the increase in classificatory smoothing
The main problem in measuring classificatory smoothing is that the magnitude and incidence of
CIs is primarily due to business fundamentals. Failure to control for these factors can result in
measurement error in the proxy for classificatory smoothing. For example, an increase in business
restructurings in the post-FRS3 period could lead to an increase in both the magnitude of negative non-
operating exceptional items and the extent to which these items reduce the variability of alternative
measures of performance. Thus, an apparent increase in the magnitude of classificatory income smoothing
could reflect an increase in the frequency of real events. Elliott and Hanna (1996) document a dramatic
increase in the frequency of negative special items for US firms for the period 19801994, despite the same
reporting standard governing the disclosure of special items throughout the period. The authors point out
that the rising trend coincides with an increasing frequency of corporate restructurings during the 1980s.
To limit the risk of an increase in the frequency of real events driving our results, we scale earnings
volatility by revenue volatility in calculating the smoothing indices. Gore, Pope and Singh (2002) argue
that revenues also reflect the effect of events such as mergers, acquisitions, and other structural business
changes, which give rise to non-operating exceptional items. The fact that the documented increase in
negative exceptional items post-FRS3 results mainly from gradual growth in the relative magnitude of
8/4/2019 Classificatory Income Smoothing
26/50
26
operating exceptional items further mitigates this risk. Events giving rise to operating exceptional items are
even more likely to be reflected in revenue. As for restructuring charges, which are a major component of
non-operating exceptional items, prior evidence in the UK documents a decreased frequency of corporate
restructurings in the years following the introduction of FRS3 up until 1998 (Mak 2002).
As an additional robustness check on our control for operating performance, we recalculate CSI
based on smoothing indices that scale the standard deviation of earnings by the standard deviation of
operating cash flows. Controlling for cash flows captures the effect of events giving rise to exceptional
items reflected in costs instead of revenue. We repeat the core tests with the alternative CSI measure and
find consistent results in both scenarios.
5.2.4 Repeating the analysis for firms disclosing alternative EPS
In calculating CSI for the post-FRS3 period we assume the least volatile earnings number is the
firms smoothing object. This might be a misleading criterion if the firm does not disclose this number to
investors as an alternative EPS.15
On the other hand, lack of an alternative EPS does not necessarily mean
that a firms smoothing object is basic EPS. Ronen and Sadan (1981) argue that the smoothing object is the
variable managers perceive to have the greatest impact on investors actions. The original intention of the
ASB when issuing FRS3 was to provide users with a range of important components of performance to
enable them to convert net income to a more sustainable performance indicator and to form better informed
expectations about future results and cash flows. To the extent managers are confident that analysts and
investors make the necessary adjustments to arrive at core earnings, they may smooth a profit and loss
subtotal without reporting alternative EPS on this number. Shortly after the enforcement of FRS3
(September 1993) the IIMR introduced headline earnings, an income figure focusing on a companys
trading performance and less volatile than net income. For the initial period following the introduction of
FRS3, the EPS that analysts forecasted was headline earnings (Acker et al. 2002). In two post-FRS3
15 This is a plausible scenario especially for the period directly following implementation of FRS3. As disclosing
alternative EPS was not mandatory under FRS3, and as it was inevitably not an established policy in the initial period,
some firms may have been reluctant to make such disclosures immediately. Possible reasons are confusion over the
new definitions, uncertainty over the disclosure choices of other firms, and the time needed to choose an alternative
level to disclose consistently over time.
8/4/2019 Classificatory Income Smoothing
27/50
27
accounting periods, 1993 and 1996, Lin and Walker (2000) find that headline earnings were more value
relevant for stock prices than basic EPS.
To further investigate this issue, we repeat the core tests of the first hypothesis on a subset of firms
that disclose alternative EPS post-FRS3. We use data on alternative EPS disclosures in 1996 for the 500
largest (based on market capitalization) UK listed non-financial firms.16
From the initial number of 254
disclosers of alternative EPS,17
197 satisfy the criteria for the original sample. Table 8 reports the mean
and median CSI partitioned into the pre- and post-FRS3 periods. Pre-FRS3, mean and median CSI are
0.071 and 0.016. Post-FRS3, mean and median CSI increase to 0.190 and 0.068 in Scenario A, and 0.127
and 0.026 in Scenario B. While the increase in mean CSI is significant in both scenarios, the increase in
median CSI is significant only in Scenario A. The frequency of firms with positive CSI in Scenario A
rises by 13% )197/)139164(( post-FRS3. The respective increase in Scenario B is only 4%, suggesting
that a great proportion of firms disclosing alternative EPS smooth income through all and not just non-
operating exceptional items.
Regression results, reported in Table 9, show that DIVERGENCEFRS 3 is positive and
significant in both scenarios. The coefficients on DIVERGENCEFRS 3 (4.274, t= 5.74 in Scenario A
and 2.122, t= 2.52 in Scenario B) substantially exceed those in Table 5 (0.702 and 0.373). In Table 10 we
extend the analysis to allow for interactions between the remaining drivers of the variation in CSI and
.3FRS DIVERGENCEFRS 3 remains positive and significant in both scenarios 70.6,115.4( =t in
Scenario A and 2.195, t= 2.41 in Scenario B). We find evidence of structural shifts in SIZE (only in
Scenario A) and .LI The coefficient on SIZE increases from 0.008 pre-FRS3 to 0.041 )033.0008.0( +
post-FRS3, consistent with a more pronounced positive association between size and classificatory
smoothing post-FRS3. The coefficient on LI switches from negative (2.677) pre-FRS3, to positive
)799.5476.8677.2( =+ post-FRS3. The shift is similar in Scenario B. So, while pre-FRS3 the level and
smoothing effect of removing CIs on reported earnings work in opposite directions, post-FRS3 income-
increasing CIs also result in smoother earnings.
16We thank Dr Young-Soo Choi (Lancaster University) for providing his data on disclosures of alternative EPS.
17This is approximately 50% of the top 500 UK firms. We repeat the analysis for firms disclosing alternative EPS in
the first post-FRS3 financial statements published in 1993/1994 (approximately 38% of the top 500 UK firms). The
core results remain.
8/4/2019 Classificatory Income Smoothing
28/50
28
To test whether absolute unexpected earnings induce a greater increase in classificatory smoothing
post-FRS3 for firms disclosing alternative EPS in 1996 relative to the remaining firms in the sample, we
run tests on the full sample adding an interaction term between an indicator of disclosure of alternative EPS
)(DIS and .3 DIVERGENCEFRS Table 11 reports the regression results. While DIVERGENCEFRS 3
remains significant in both scenarios ,27.3,730.0( =t in Scenario A and ,26.2,393.0 =t in Scenario B)
DIVERGENCEFRSDIS 3 is also significant ,60.3,403.2( =t in Scenario A and 84.1,425.1 =t in
Scenario B), indicating that absolute unexpected earnings induce an incremental rise in classificatory
smoothing post-FRS3 for disclosure firms relative to the remaining firms in the sample.
In summary, results in Table 911 provide useful insights. They provide evidence of a greater
increase in classificatory income smoothing post-FRS3 to reduce deviations of net income from
expectations for firms disclosing an alternative EPS in 1996 relative to the remaining firms in the sample in
both scenarios. Even though this result outlines the importance of FRS3s option to disclose alternative
EPS, evidence on the increase remaining within non-disclosure firms suggests that our core findings are not
solely attributable to this option. This might be because firms that did not disclose alternative EPS still
smoothed pre-exceptional earnings in the belief that investors would focus on these levels post-FRS3. The
structural break for size is consistent with larger firms engaging in a higher level of classificatory
smoothing post-FRS3 through all exceptional items. Finally unlike pre-FRS3, post-FRS3 removals of
negative non-recurring items by disclosers of alternative EPS yield smoother earnings.
5.3 The effect of FRS3 on earnings persistence
We initially examine the extent to which removing CIs increases the persistence of reported
earnings in the pre- and post-FRS3 periods and then test whether there is an increase in the persistence of
pre-exceptional earnings post-FRS3. Panel A of Table 12 reports the results of earnings persistence
regressions for ,EARN ,EARNbXI EIEARNbXI& pre-FRS3 and ,EARN ,EARNbNOEI and EARNbEI
post-FRS3. The panel reports t-tests for the significance of the differences in the persistence coefficients
within each period. Pre-FRS3 the persistence coefficient increases from 0.52 for EARN to 0.59 for
.EARNbXI The increase results from the removal of XI and is marginally significant (p = 0.098). Even
though the incremental growth in persistence of EIEARNbXI& to 0.64 is not significant, the overall
8/4/2019 Classificatory Income Smoothing
29/50
29
increase in persistence from EARN to EIEARNbXI& pre-FRS3 (0.12) is significant. Post-FRS3, the
persistence coefficient increases from 0.49 for EARN to 0.64 for EARNbNOEI and to 0.73 for .EARNbEI
The incremental increases of 0.15 and 0.09 are significant, indicating that earnings before NOEI are more
persistent than net income and earnings before EI are in turn more persistent than earnings before .NOEI
The overall increase in persistence from EARN to EARNbEI post-FRS3 is 0.24, which is double the
increase in the pre-FRS3 period.
Table 12, Panel B reports the results of estimating equation (2) for four earnings levels: a) ,EARN
b) ,1EARNADJ c) ,2EARNADJ and d) .3EARNADJ The interaction term captures the change in
persistence of the profit levels post-FRS3. There is no evidence of a significant change in the persistence
ofEARN post-FRS3. Even though net income is communicated to investors through basic EPS, it appears
to be as persistent as in the pre-FRS3 period. The coefficient on 13EARNADJFRS is positive and
marginally significant (0.057, t= 1.91), indicating that EARNbNOEI is more persistent than EARNbXI of
the pre-FRS3 period. 23EARNADJFRS and 33EARNADJFRS are positive and highly significant
49.5,132.0( =t and ),15.4,090.0 =t suggesting that EARNbEI is substantially more persistent than both
EARNbXI and EIEARNbXI& pre-FRS3.
In sum, results in Table 12 are consistent with our second hypothesis of an increase in the
persistence of pre-exceptional profit levels post-FRS3. The results indicate that removing CIs increases the
persistence of earnings mainly in the post-FRS3 period. Moreover, excluding all and not just non-operating
exceptional items results in more persistent earnings post-FRS3. This evidence is in line with FRS3s
provisions enhancing the role of classificatory choices in identifying recurring earnings.
6. Conclusion
We examine the effect of FRS3 on the practice of classificatory income smoothing. Results based
on a large balanced sample provide evidence of an overall increase in the practice of classificatory
smoothing post-FRS3. The increase coincides with growth in the magnitude of CIs and greater use of
income-increasing CIs (e.g. transitory losses, exceptional costs) to smooth pre-exceptional earnings levels.
We also find that unlike extraordinary items pre-FRS3, exceptional items post-FRS3 are equally distributed
between positive and negative values, indicating that FRS3 succeeded in removing the asymmetry in the
8/4/2019 Classificatory Income Smoothing
30/50
30
distribution of CIs. Results from multivariate analysis further show that deviations of net income from
target earnings (lagged adjusted earnings) induce a higher level of classificatory smoothing post-FRS3.
Consistent with this evidence, earnings persistence tests show that pre-exceptional earnings levels are
significantly more persistent than net income especially post-FRS3. Our results suggest that FRS3
increased the use of classificatory choices to highlight an indicator of sustainable earnings.
Our results shed light on managerial smoothing practices following the introduction of FRS3,
which have important implications for both investors and accounting standard setters. For investors the
main implication is the greater transparency of classificatory smoothing post-FRS3. Investors can identify
and assess all adjustments made to basic EPS to arrive at the alternative figure. Even when firms report
basic EPS as the sole performance indicator, greater disclosure of exceptionals helps investors extract their
preferred measure of persistent profitability. Higher persistence of pre-exceptional earnings post-FRS3
further facilitates investors assessments of firms future prospects.
With regard to regulators, FRS3 facilitated communication between managers and the market by
enabling classificatory smoothing as a means to convey information about sustainable performance
necessary for security valuation. The transparency requirements helped users ascertain whether firms
reasonably remove income components, making the practice of classificatory smoothing especially
challenging for opportunistic managers. Moreover, by enabling the less costly practice of classificatory
smoothing, FRS3 reduced the costs of income smoothing.
FRS3 is a reference point for developing a global model for reporting financial performance. In
line with FRS3s approach, the IASB, the FASB and the ASB are considering deemphasizing reliance on a
single performance indicator (IASB 2005). Our results indicate that FRS3s requirements for a more
informative distinction between permanent and transitory income components along with the option to
disclose alternative performance measures encouraged the practice of classificatory smoothing. To the
extent firms distinguish non-recurring items from permanent income they assist users in evaluating firms
future prospects. Evidence on the increased persistence of pre-exceptional earnings appears consistent with
this notion. To the extent, however, that managers remove value relevant items from the alternative figure
of firm performance to increase the perceived value of the firms ongoing earnings potential (Doyle et al.
2003) then users may be misled and firm valuations become tentative.
8/4/2019 Classificatory Income Smoothing
31/50
31
In our analysis we concentrate on the signalling role of smoothing practices. However, despite the
increased level of transparency the possibility of opportunistic classificatory choices remains. Although the
wide definition of exceptionals in FRS3 allowed for a more precise measure of core profitability, it may
have encouraged opportunistic classifications of recurring items. Our research design is not appropriate for
capturing classificatory choices motivated by opportunism. This requires a setting where managerial
incentives for engaging in income-increasing earnings management are particularly strong. Empirical
evidence suggests that meeting or exceeding earnings benchmarks is a strong asset pricing motivation for
inflating reported earnings (Burgstahler and Dichev 1997, Degeorge, Patel and Zeckhauser 1999, Bartov,
Givoly and Hayn 2002). To the extent UK firms simultaneously increase and smooth alternative earnings
through classification items, especially post-FRS3, they may move items below the line of alternative EPS
in an attempt to inflate earnings and meet or beat analyst forecasts. The latter benchmark provides strong
incentives for income-increasing earnings management, offering a suitable context for identifying
opportunistic classifications.
References
Acker, D., J. Horton and I. Tonks, 2002, Accounting standards and analysts forecasts: the impact of FRS3
on analysts ability to forecast EPS, Journal of Accounting and Public Policy 21, 193217.
Ashari, N., H.C. Koh, S.L. Tan and W.H. Wong, 1994, Factors affecting income smoothing among listed
companies in Singapore, Accounting and Business Research 24, 291301.
Barnea, A., J. Ronen and S. Sadan, 1976, Classificatory smoothing of income with extraordinary items, The
Accounting Review 51, 110122.
Bartov, E., D. Givoly and C. Hayn, 2002, The rewards to meeting or beating earnings expectations, Journal
of Accounting and Economics 33, 173204.
Beattie, V., S. Brown, D. Ewers, B. John, S. Manson, D. Thomas and M. Turner, 1994, Extraordinary items
and income smoothing: a positive accounting approach, Journal of Business Finance and Accounting 21,
791811.
8/4/2019 Classificatory Income Smoothing
32/50
32
Bhattacharya, N., E.L. Black, T.E. Christensen and C.R. Larson, 2003, Assessing the relative
informativeness and permanence of pro forma earnings and GAAP operating earnings, Journal of
Accounting and Economics 36, 285319.
Bradshaw, M.T. and R.G. Sloan, 2002, GAAP versus the Street: an empirical assessment of two alternative
definitions of earnings, Journal of Accounting Research 40, 4166.
Burgstahler, D. and I. Dichev, 1997, Earnings management to avoid earnings decreases and losses, Journal
of Accounting and Economics 24, 99126.
Choi, Y.S., S. Lin, M. Walker and S. Young, 2005, Voluntary disclosure of alternative earnings per share
numbers, Working Paper, Lancaster University.
Degeorge, F., J. Patel and R. Zeckhauser, 1999, Earnings management to exceed thresholds, Journal of
Business 72, 133.
Doyle, J. T., R. J. Lundholm and M.T. Soliman, 2003, The predictive value of expenses excluded from pro
forma earnings, Review of Accounting Studies 8, 145174.
Eames M.J. and J. Sepe, 2003, The valuation of special items, Working Paper, Santa Clara University.
Eckel, N., 1981, The income smoothing hypothesis revisited, Abacus 17, 2840.
Elliott, J.A. and J.D. Hanna, 1996, Repeated accounting write-offs and the information content of earnings,
Journal of Accounting Research 34 Supplement, 135155.
Fields, T.D., T.Z. Lys and L. Vincent, 2001, Empirical research on accounting choice, Journal of
Accounting and Economics 31, 255307.
Godfrey, J.M. and K.L. Jones, 1999, Political cost influences on income smoothing via extraordinary item
classification, Accounting and Finance 39, 229254.
Gore, P., P.F. Pope and A.K. Singh, 2002, Earnings management and the distribution of earnings relative to
targets: UK Evidence, Working Paper, University of Lancaster.
Gordon, M.J., 1964, Postulates, principles and research in accounting, The Accounting Review 39, 251
263.
Holthausen, R.W. and R.W. Leftwich, 1983, The economic consequences of accounting choice:
Implications of costly contracting and monitoring, Journal of Accounting and Economics 5, 77117.
8/4/2019 Classificatory Income Smoothing
33/50
33
International Accounting Standards Board (IASB), 2005, Performance Reporting Project Update (Revised
December 8th
2005). http://www.iasb.org/current/active_projects.asp
Landsman, W.R., B.L. Miller and S. Yeh, 2006, Implications of components of income excluded from pro
forma earnings for future profitability and equity valuation, Working Paper (University of North
Carolina).
Lev, B. and S. Kunitzky, 1974, On the association between smoothing measures and the risk of common
stocks, The Accounting Review 49, 259270.
Lin, S. and M. Walker, 2000, FRS3 earnings, Headline earnings, and accounting-based valuation models,
Accounting and Business Research 30, 299306.
Lin, S., 2002, The association between analysts forecast revisions and earnings components: The Evidence
of FRS 3, British Accounting Review 34, 126.
Lougee, B. and C. Marquardt, 2004, Earnings quality and strategic disclosure: An empirical examination of
pro forma earnings, The Accounting Review 79, 769795
Mak, C.Y., 2002, Effects of corporate refocusing on UK listed industrial companies valuations, reported
earnings and financial analysts forecasts, Unpublished thesis, University of Manchester.
Michelson, S.E., J. JordanWagner and C.W. Wootton, 1995, A market based analysis of income
smoothing, Journal of Business Finance and Accounting 22, 11791193.
Moses, O.D., 1987, Income smoothing and incentives: Empirical tests using accounting changes, The
Accounting Review 62, 358377.
Peasnell, K.V., P.F. Pope and S. Young, 2000, Accrual management to meet earnings targets: UK evidence
pre- and post-Cadbury, British Accounting Review 32, 415445.
Penman, S.N., 2003, Financial statement analysis and security valuation, Second Edition (McGraw-
Hill/Irwin, New York).
Pope, P.F. and M. Walker, 1999, International differences in the timeliness, conservatism, and
classification of earnings, Journal of Accounting Research 37 Supplement, 5387.
Ronen, J. and S. Sadan, 1981, Smoothing income numbers, objectives, means and implications (Addison-
Wesley Publishing Company, Reading).
Schipper, K., 1989, Commentary on earnings management, Accounting Horizons 3, 91102.
8/4/2019 Classificatory Income Smoothing
34/50
34
Smith, T., 1992, Accounting for Growth (Century Business, London).
Watts, R.L. and J.L. Zimmerman, 1978, Towards a positive theory of the determination of accounting
standards, The Accounting Review 53, 112134.
Watts, R.L. and J.L. Zimmerman, 1979, The demand for and supply of accounting theories: The market for
excuses, The Accounting Review 54, 273305.
White, G.E., 1970, Discretionary accounting decisions and income normalization, Journal of Accounting
Research 8, 260273.
Wooldridge, J.M. (2002). Econometric analysis of cross section and panel data. Cambridge Massachusetts:
The MIT Press.
8/4/2019 Classificatory Income Smoothing
35/50
35
Appendix
Adapted from Figure 3.2 of Ronen and Sadan (1981, p.44) on the relation between smoothing objects and
smoothing dimensions. When the smoothing object is operating or ordinary income, managers can smooth
either inter-temporally from above or from below through classificatory items (e.g. classifications of
depreciation, discretionary fixed charges and other charges, and non-recurring and non-operating items).
Firms smooth net income only inter-temporally through all items above it.
Sales
Cost of sales
Gross Margin
Operating discretionary expenses
Operating Income
Depreciation
Discretionary fixed charges
Other charges
Non-recurring and non-operating items
Income statement items Smoothing dimensions
Ordinary income
Net income
Inter-temporal
less
less
less
less
less
lessClassificatory
SSmmooootthhiinngg oobbjjeeccttss aanndd rreellaatteedd ddiimmeennssiioonnss
8/4/2019 Classificatory Income Smoothing
36/50
36
Table 1
Measures of the overall magnitude of classificatory income smoothing
Period Level of earnings Smoothing index )(SI Classificatory smoothing index )(CSI
Pre-FRS3 EARNREV
EARNSI
=1
EARNbXI REV
EARNbXISI
=2 ),min( 211 SISISICSI =
Post-FRS3 EARNREV
EARNSI
=1
EARNbNOEIREV
EARNbNOEISI
=3
Scenario A:
),,min( 4311 SSSISICSI =
EARNbEIREV
EARNbEISI
=4
Scenario B:
),min( 311 SSISICSI =
EARN is earnings after extraordinary items. EARNbXI is earnings before extraordinary items. EARNbNOEI is earnings before non-
operating exceptional items. EARNbEI is earnings before all exceptional items.
8/4/2019 Classificatory Income Smoothing
37/50
37
Table 2
Panel A: Descriptive statistics for XI and EI pre-FRS3, NOEI and OEI post-FRS3.
Period N Variable Mean Std. dev. Median
Pre-FRS3 5,677 XI(a) 0.004*** 0.036 0.000
EI(b) 0.003*** 0.030 0.000
Total 0.007*** 0.049 0.000
Post-FRS3 4,969 NOEI(c) 0.007*** 0.044 0.000OEI(d) 0.006*** 0.031 0.000
Total 0.013*** 0.060 0.000
Diff (c)(a)(p-value)
0.003(
8/4/2019 Classificatory Income Smoothing
38/50
38
Panel C: Frequencies of positive, negative, and zero XI and EI pre-FRS3, NOEI and OEI post-FRS3.
Pre-FRS3 Post-FRS3
Variable Freq (%)c Freq (%)d p-value Variable Freq (%)c Freq (%)d p-value
XI> 0 15.04 33.06 NOEI> 0 23.78 50.15
XI< 0 30.46 66.94 0 46.54 57.45 OEI> 0 40.57 49.16
EI< 0 34.47 42.55 0 XI < 0 XI = 0 p-value EI > 0 EI < 0 EI = 0 p-value
1986 14.64% 35.53% 49.84%
8/4/2019 Classificatory Income Smoothing
39/50
39
Table 3
The overall magnitude of classificatory income smoothing (classificatory smoothing indexCSI)
Pre-FRS3 Post-FRS3
SO = (EARNbXI)
Mean/(Median)
Scenario A:SO = (EARN,EARNbNOEI,
EARNbEI)Mean/
(Median)
Scenario B:SO = (EARN,
EARNbNOEI)Mean/
(Median)
),min( 211 SISISICSI = 0.063***(0.004)***
1 1 3 4min( , , )CSI SI SI SI SI = 0.157***(0.023)***1 1 3min( , )CSI SI SI SI =
0.102***
(0.006)***
N
914
C: 387
UC: 527
914
C: 199
UC: 715
914
C: 364
UC: 550
Diff. (pre post).0.094
(0.019)
0.039
(0.002)
p-valuea
8/4/2019 Classificatory Income Smoothing
40/50
40
Table 4
Descriptive statistics on key variables affecting the variation of the classificatory smoothing index (CSI)
Variable StatisticsPre-FRS3(N=914)
Post-FRS3(N=914) p-valuea
DIVERGENCE Mean 0.070*** 0.057*** 0.012
(Scenario A) Std. dev. 0.155 0.125
Median 0.026 0.024 0.064
DIVERGENCE Mean 0.070*** 0.058*** 0.057
(Scenario B) Std. dev. 0.155 0.127
Median 0.026 0.025 0.161
GAINLOSS Mean 0.032*** 0.053***
8/4/2019 Classificatory Income Smoothing
41/50
41
Table 5
Regressions of the classificatory smoothing index (CSI) on an FRS3 indicator, divergence from expected earnings,
and a set of control variables.
Scenario A
Model 1
Scenario B
Model 2
Variables
Predicted
Sign
Coefficient
(t-statistic) Partial effects
Coefficient
(t-statistic) Partial Effects
Intercept0.060
(1.45)
0.058
(1.48)
FRS30.072***
(4.68)0.040
0.015
(1.17)0.007
DIVERGENCE +0.278**
(2.41)0.155
0.264**
(2.43)0.130
FRS3 DIVERGENCE + 0.702***(2.88) 0.390 0.373**(2.15) 0.184
GAINLOSS +0.320***
(3.13) 0.178 0.413***(4.09) 0.203PERMLOSS
0.103*
(1.82) 0.057 0.079*(1.70) 0.039RISK() +
0.030*
(1.66) 0.0170.027*
(1.65) 0.013
LEV +0.236***
(3.11)0.131
0.246***
(3.69)0.121
SIZE +0.014***
(4.54)0.008
0.012***(4.26)
0.006
PI 0.010**
(2.32)0.004
0.008*
(1.85)0.004
PI 0.020***
(2.80)0.011
0.019***
(3.01)0.010
LI +3.863***
(6.04)2.146
3.397***
(5.59)1.672
N1,828
(C:586 /UC:1242)
1,828
(C:751/UC:1077)
Chi-square 788.66 534.10
p-value
8/4/2019 Classificatory Income Smoothing
42/50
42
Table 6
Regressions of the classificatory smoothing index (CSI) on an FRS3 indicator, divergence from expected earnings
and a set of control variables for firms with at least five observations in each of the pre and post-FRS3 periods.
Scenario A
Model 1
Scenario B
Model 2
VariablesPredictedSign
Coefficient(t-statistic) Partial effects