Description
The purpose of this paper is to assess the financial disclosure vis-a´-vis economic reality of
research and development (R&D) expensed by Australian firms under the pre-2005 Australian
generally accepted accounting principles (A-GAAP) regime via the lens of market-to-book.
Accounting Research Journal
R&D profitability, intensity and market-to-book: evidence from Australia
Kamran Ahmed J ohn Hillier Elisabeth Tanusasmita
Article information:
To cite this document:
Kamran Ahmed J ohn Hillier Elisabeth Tanusasmita, (2011),"R&D profitability, intensity and market-to-book:
evidence from Australia", Accounting Research J ournal, Vol. 24 Iss 2 pp. 150 - 177
Permanent link to this document:http://dx.doi.org/10.1108/10309611111163691
Downloaded on: 24 January 2016, At: 21:13 (PT)
References: this document contains references to 45 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 945 times since 2011*
Users who downloaded this article also downloaded:
Astrid Schmulian, Stephen Coetzee, (2011),"Class absenteeism: reasons for non-attendance and
the effect on academic performance", Accounting Research J ournal, Vol. 24 Iss 2 pp. 178-194 http://
dx.doi.org/10.1108/10309611111163718
Kym Butcher, Graeme Harrison, J ill McKinnon, Philip Ross, (2011),"Auditor appointment in
compulsory audit tendering", Accounting Research J ournal, Vol. 24 Iss 2 pp. 104-149 http://
dx.doi.org/10.1108/10309611111163682
J acqueline Birt, Greg Shailer, (2011),"Forecasting confidence under segment reporting", Accounting
Research J ournal, Vol. 24 Iss 3 pp. 245-267http://dx.doi.org/10.1108/10309611111186993
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
R&D pro?tability, intensity
and market-to-book: evidence
from Australia
Kamran Ahmed and John Hillier
School of Accounting, Faculty of Law and Management, La Trobe University,
Melbourne, Australia, and
Elisabeth Tanusasmita
Bombardier Transportation, Milton, Australia
Abstract
Purpose – The purpose of this paper is to assess the ?nancial disclosure vis-a´-vis economic reality of
research and development (R&D) expensed by Australian ?rms under the pre-2005 Australian
generally accepted accounting principles (A-GAAP) regime via the lens of market-to-book.
Design/methodology/approach – The authors estimated ?rms’ R&D pro?t rate, measured R&D
revenue intensity and modelled the impacts of these and related economic factors, via economic and
?nancial disclosure channels, on market-to-book using data for 1988-2004.
Findings – R&D, on average, was pro?t neutral and had undetectable impacts on market-to-book
whether via equity valuation or ?nancial disclosure.
Research limitations/implications – Market-to-book’s information content is best viewed as
conditional on the reference disclosure regime. Australian ?rms’ typically at best minimal R&D
pro?tability is an international anomaly. Data limitations in terms of the generating process and
availability mean that R&D’s impact on market-to-book via ?nancial reporting is not de?nitively
determined.
Practical implications – Restrictive rules on the capitalization of intangible asset-related
expenditures under A-GAAP apparently did not adversely impact market-to-book’s economic
information. AIFRS’s more permissive rule risks compromisingmarket-to-book’s reliabilityinsucha role.
Originality/value – For Australia, the paper is anticipated to be the ?rst to estimate the pro?t rate of
R&D, measure the intensity of R&D, and model R&D’s in?uence on the market-to-book ratio. It develops
a framework for the economic and ?nancial reporting impacts of investments on a key indicator of ?rms’
?nancial standing and contributes to the debate on identi?able intangibles’ disclosure.
Keywords Australia, Listed companies, Research and development, Financial reporting,
Market-to-book, Intangibles, A-GAAP, AIFRS
Paper type Research paper
1. Aim and overview
From the late 1980s through the mid-2000s, Australia experienced a steady rise in the
number of public, listed corporations reporting research and development (R&D)
expenses in their annual ?nancial statements (Table I, Panel B), and in the average
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1030-9616.htm
The authors’ names are listed in alphabetical order. The authors acknowledge the comments of
participants at the Global Accounting and Organisational Change conference, Melbourne,
9-11 July 2008 and the Asia-Paci?c Conference on International Accounting Issues, Paris,
9-12 November 2008, the advice of Abdul Shamsuddin, and the ?nancial support of
La Trobe University.
ARJ
24,2
150
Accounting Research Journal
Vol. 24 No. 2, 2011
pp. 150-177
qEmerald Group Publishing Limited
1030-9616
DOI 10.1108/10309611111163691
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
intensity (relative to sales revenue) of such outlays (Table III and Figure 2).
Contemporaneously, the average market value of equity to book value of equity (BVEq)
(“market-to-book”) for R&D-active ?rms increased substantially (Table III and
Figure 2)[1].
Incorporating ?rms’ equity prices and the net book value of identi?able assets,
market-to-book conveys informationrelevant to ?rms’ prospects for strategies including
investments in intangible assets such as those associated with R&D engagement.
Market-to-book’s performance in such a role, relative to R&D, depends jointly on the
equity market’s assessment of R&Dexpenditure’s economic impact, and on its handling
in the ?nancial statements, and the ?nancial disclosure regime. To the extent that R&D
expenditure is viewed by the capital market as an investment, the anticipated economic
bene?ts are by de?nition re?ected in ?rms’ equity prices. However, such R&D
expenditures’ immediate write-off implies ?rms’ net book values are in part disclosed
Panel A: Industry membership
Sector Abbreviation Obs
Obs (%)
Mining and construction
a
Mincon 297 18.0
Manufacturing
a
Manuf 465 28.3
Chemicals Chem 74 4.5
Transport Incl. in TURS 5 0.3
Utilities
a
Incl. in TURS 26 1.6
Information and communications technology
a
ICT 191 11.6
Pharmaceuticals and health
a
Pharmhealth 304 18.5
Food
a
Food 129 7.8
Retail
a
Incl. in TURS 22 1.3
Services
a
Incl. in TURS 133 8.1
Total 1,646 100
Panel B: Number per year
Fiscal year n
1988 20
1989 23
1990 27
1991 31
1992 41
1993 61
1994 71
1995 77
1996 111
1997 111
1998 111
1999 125
2000 147
2001 174
2002 170
2003 177
2004 169
Total 1,646
Notes: The sectors marked with super script “a” are industry sectors consolidated from the ?ner
company analysis sectors as detailed in Note 13; TURS (transport, utilities, retail, services) re?ects a
further consolidation
Table I.
Companies reporting
R&D by industry and
frequency, 1988-2004
R&D
pro?tability
151
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
on a basis inconsistent with their market valuations, with consequences for
market-to-book’s economic information.
Framed generally, this paper investigates the ?nancial disclosure vis-a´-vis economic
reality of Australian ?rms’ expensed R&D efforts in the era of Australian generally
accepted accounting principles (A-GAAP). Market-to-book serves as a lens allowing
appraisal of this issue. If pro?tability is R&D’s main in?uence over market-to-book, the
latter conveys primarily “economic” information regarding R&D efforts. Alternatively,
if R&D expenditures’ handling in the ?nancial statements is the dominant impact then
there is potential for degradation of market-to-book’s information. Such potential is
realized by R&Dexpenditures’ expensing under increasing (or decreasing) outlays, with
the effect the systematic biasing of ?rms’ equity book values. This disclosure artefact’s
symptom is, as argued below, market-to-book’s positive relation to R&D intensity.
This paper adopts the basic rationale andmodelling approachapplied byHand(2001)
to US experience with R&D in the 1980s and 1990s. However, it offers a more explicit,
deeper theoretical framework and extends the array of statistical analyses to achieve
arguably more reliable conclusions. The modelling is based on two key dimensions of
?rms’ R&D, with market-to-book’s information on R&Danalysed via its response to the
dimensions’ channels of in?uence. Additionally, interpretation of market-to-book is
argued to be conditional on the reference (preferred) disclosure regime.
The ?rst of R&D’s above-referred dimensions is its rate of pro?tability. Derivation
is via statistical estimation with ?nancial statement data. A little-researched but key
issue in its own right, it has an (equity) market value channel of in?uence, via
market-to-book’s numerator.
R&D’s second dimension, revenue intensity, is computed as R&D relative to sales
revenue andadopted to mitigate size-of-?rmeffects inthe analysis, consistent with Hand
(2001). Crucially, R&D intensity’s level has a book value channel, via market-to-book’s
denominator. Additionally, intensity level, compounded with rate of pro?tability, has a
market value channel. Finally, R&D intensity’s ?rst difference has a market value
channel. Figure 1 shows a schema for R&D’s channels of in?uence over market-to-book.
This paper employs a dataset consisting of 1,646 Australian listed company
?rm-years with reported R&Dand supplemented by zero-R&Dobservations within two
Figure 1.
Schema for R&D’s
channels of in?uence over
market-to-book
Market value of equity
Book value of equity
Rate of profitability
1
2
2
3
4
1: Impact of rate of profitability
2: Impact of level of intensity at a given rate of profitability
3: Impact of management signal on profit-rate parameter(s)
4: Impact due to financial statement handling of expenditure
Revenue intensity
ARJ
24,2
152
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
years of succeeding or one year of preceding R&D, over 1988-2004 and across ten
industry sectors. The paper ?rst attempts to determine the on-average rate of
pro?tability of R&D. Then, the impact of such pro?tability versus the biasing effect on
equity book values due to R&D expenditures’ disclosure as expenses are evaluated in
terms of market-to-book’s information role over 1988-2004.
Firms’ on-average R&D pro?t rates are estimated for the overall sample on a
pooled cross-section/time series basis across 1989-2004 and on annual and
industry/multi-annual partition-based sub-samples for 1992-2004. That the
on-average net discounted R&D pro?t rate is zero cannot be rejected for the overall
sample and for most period and industry/multi-period sub-samples. Australian lack of
pro?tability is entirely at variance with international evidence.
Two approaches are used to model market-to-book in relation to R&D. First, the two
sets of partitioned-sample R&D pro?t rate estimates are combined with the respective
mean R&D intensities and market-to-book ratios computed from similar sub-samples
and the impacts of R&D on market-to-book estimated. No reliable evidence emerges for
either a R&D rate of pro?tability or revenue intensity effect.
Second, R&Dintensity, using ?rm-year level data over 1988-2004, is decomposed into
variables re?ecting book value and market value channels of in?uence and arrayed in
competition with lagged market-to-book and the current change in overall pro?tability to
model market-to-book. No evidence emerges for market-to-book’s economic information
being degraded by A-GAAP’s stringent R&D standard relative to either “ef?cient
contracting” or “equityvaluation”-orienteddisclosure regimes. This null result maybe an
artefact of ?rms’ typical short-durationreportedR&Dengagement, inter-?rmrather than
intra-?rm variation in R&D intensity over the era modelled, and limitations of the data.
The approach and ?ndings enhance understanding of the impact of disclosure
methods for investments in R&D and, potentially, other intangibles, on the information
conveyed via market-to-book. Implications are drawn with respect to the
now-historical A-GAAP and the currently prevailing AIFRS disclosure regime.
From this point, the paper is structured as follows: Section 2 deals with key concepts
and surveys the literature. The research method is developed in Section 3, data and
summary statistics are described in Section 4, followed by presentation of results in
Section 5. Section 6 provides a summary and implications.
2. Concepts and literature review
The literature dealing with market-to-book in general and especially the relation to
R&D is scarce, with Australian applications noticeably absent. However, the following
reviews precursors to this paper. Attention is given to market-to-book’s potential
information, R&D’s economic bene?ts and rates of pro?tability, the equity market
response to R&D pro?tability, and, R&D’s reporting under A-GAAP and its impact on
equity book values’ properties relative to measurement and to disclosure regimes and
the evidence on market-to-book’s re?ection of R&D under US-GAAP.
2.1 Market-to-book
Market-to-book is a synthetic equity market and accounting data-derived indicator,
ideally informative on ?rms’ economic standing and prospects. One interpretation sees
market-to-book as a measure of growth opportunities and thereby as a proxy for the
unobservable investment opportunity set (Kallapur and Trombley, 1999). An additional
R&D
pro?tability
153
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
role is in predicting abnormal earnings (Ohlson, 1995; Feltham and Ohlson, 1995, 1996).
Another is as a ?nancial distress risk indicator (Fama and French, 1992, 1993). Afurther
interpretation regards market-to-book as an analogue of (or even empirical substitute
for) Tobin’s Q, the market value of a ?rmdivided by its “book” value at replacement cost
(Tobin, 1969). Market-to-book has also been viewed as re?ecting the market’s valuation
of unrecognised assets (Zambon and Associates, 2003).
The market premium (Penman, 2004, p. 40) may be viewed as the sum of three
components – the market versus book value difference of recognised assets, the value
of unrecognised identi?able intangibles, and economic goodwill[2]. Market-to-book
may therefore be useful as an indicator of intellectual (or organizational) capital to the
extent that ?rst, identi?able intangible assets such as R&D investments are
unrecognised and second, that the synergistic effect re?ected in the going concern
premium is a synonymous concept[3].
2.2 R&D’s rate of pro?tability
The question of the relation between R&D expenditures and subsequent economic
bene?ts has attracted signi?cant interest in recent times, with wide recognition of R&D
activity as a factor inherent in technological change (Bosworth and Rogers, 1998). Cohen
and Levinthal (1989) delineate R&D’s role as twofold – ?rst, to generate new, applicable
knowledge and second, to develop “absorptive capacity”, the ability to recognize,
assimilate and exploit others’ knowledge, with both roles potentially contributing to
?rms’ economic capabilities. Contributions to the management and economics
literatures (Mairesse and Sassenou, 1991; Mairesse and Mohen, 1995; Bosworth and
Rogers, 1998) argue R&D tends to generate net economic bene?ts.
A typical ?nding in accounting studies is that, on average, R&D expenditure
parameters attract positive estimates for future and even contemporaneous rates of
pro?tability[4]. Key US studies include Sougiannis (1994) who reports an estimate of a
one-dollar increase in R&D leading to a two-dollar increase in pro?t over a seven-year
period. Hand (2001) estimates net present value pro?tability per dollar of R&D of 0.51,
0.45, and $0.44 at lag lengths of zero, three, and seven years, respectively. Ding et al.
(2007) adopt an international perspective, estimating R&D internal rates of return on a
country basis for ?rms in six advanced economies. Using data from 1991-2000 with
six-year lag lengths, positive returns are reported, ranging from 18 per cent for the
USA and Switzerland to 36 per cent for Japan.
2.3 R&D and equities
Widely referenced ?rm valuation models imply that equity values are increasing
functions of anticipated earnings (White et al., 2003, Ch. 19). Consequently, if R&D
typically increases future earnings, a positive, contemporaneous relationship between
equity prices and R&D or, alternatively, equity returns and growth in R&D should be
observed, consistent with previous studies (Canibano et al., 2000).
Various national studies document positive impacts of R&D expenditures on ?rms’
equity price levels and returns including, for the USA, Hirschey and Weygandt (1985)
and, for Canadian ?rms, Johnson and Pazderka (1993). Sougiannis (1994) using US data
estimates that a one-dollar increase in R&D leads to a ?ve-dollar increase in equity
market value. Additionally, for the USA, Lev and Sougiannis (1996) report estimated
net R&D investment and assets to be positively associated with contemporaneous
ARJ
24,2
154
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
equity returns and prices, respectively. For Australia, contemporaneous equity values
are found to be positively related to the stock of intangibles and intensity of R&D
expenditures (Bosworth and Rogers, 1998) and Ahmed and Falk (2006) report that
R&D expenditures are equities market value-relevant.
R&D’s in?uence on the market-to-book numerator is clearly implied by the above
effects on equity valuation. In terms of a positive impact: ?rst, an increased pro?t rate,
and second, an increased level of investment given a constant, positive pro?t rate. The
relevant market-value channels (the ?rst and second channels, respectively, in Figure 1)
convey R&D’s rate of pro?tability in the former case and R&D’s intensity, interacting
with the pro?t rate, in the latter.
2.4 R&D expenditure disclosure and implications for market-to-book
The reporting of R&D expenditures has long been a contentious issue in many
jurisdictions due to the typical long duration but highly uncertain nature of associated
economic bene?ts. Under A-GAAP, very limited managerial discretion existed with
respect to the disclosure of R&D expenditures under AASB 1011, applicable up to the
AIFRS start on January 1, 2005. R&D costs had to be expensed except where future
bene?ts were expected, beyond reasonable doubt, to equal or exceed those costs and
any future costs necessary to give rise to the future economic bene?ts, in which case
asset recognition (i.e. capitalisation and amortisation of the costs) was permitted. This
rule was presumably designed to force expensing of all R&D expenditures apart from
those at the low end of the risk spectrum and, therefore, was only a little less stringent
than the equivalent standard applying in the USA[5]. Hence this paper’s key,
policy-related issue – did A-GAAP allow market-to-book to properly re?ect the
economic impact of R&D?
To answer the preceding question, two issues are relevant, one measurement and the
other disclosure-type. The measurement issue concerns the stock-?ow dynamics of
R&D, associated revenue, and the BVEq. Owing to immediate expense and lagged
revenue recognition, an R&D project imparts a non-positive and typically negative bias
relative to the ?nal impact on BVEq at termination. The effect attenuates subsequently,
eventually fully. Worth emphasizing is that bias exists given project-generated revenue
is both non-zero and that any recognition is delayed beyond the period of expensing –
lack of complete expenditure recoupment is consistent with bias.
If the ?ow of R&D is time-invariant, the stock of BVEq will follow the path set by
the former’s parameters. Changes in the ?ow of R&D necessarily disturb such a path.
It follows from the aforementioned project-bias that, for instance, an increased level of
R&D causes an initially non-positive, typically negative and subsequently attenuating
displacement of BVEq from its path otherwise, and vice versa[6].
A displacement of BVEq’s path of either sign does not necessarily imply a similarly
signed response of the level of BVEq. The latter depends on the pre-existing
R&D-related BVEq process relative to the effects of the increase or decrease of R&D[7].
However, informal analysis and even, ideally, intuition indicates that, given the
above-described dynamics of displacement, BVEq will typically be negatively related
to R&D[8].
The above-argued BVEq-R&Dassociation extends immediately, on an “other things
equal” basis to the hypothesis of market-to-book as, on average, positively related to
R&D’s scale, represented by R&D intensity, at the individual ?rm, time series level.
R&D
pro?tability
155
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This effect of the expensing of R&D expenditure is presented as the fourth channel of
R&D’s in?uence on market-to-book in Figure 1. Extension to time series of cross-section
aggregations of ?rm data (via R&D intensity period means) is somewhat tenuous,
though central to Hand’s (2001) modelling and conclusions on “biased accounting”
(Hand, 2001, pp. 1, 3, 18). Additionally, the association’s empirical existence can be
attributed purely to R&D’s disclosure effect (via equity book value) only if the
interaction of R&D’s pro?t rate and intensity, as discussed above, can be ruled out as a
possible in?uence on average.
The above-referenced disclosure issue concerns the implications of alternate
?nancial reporting regimes for market-to-book’s economic information. In examining
these implications it is useful to distinguish two fundamentally different types of
regime with respect to asset recognition.
The ?rst regime, the hard-assets type, is consistent with support for ef?cient
contracting as the prime objective of ?nancial reporting. Recognition requirements
include well-de?ned property rights ensuring assets are controlled and separable,
saleable andthat adequate certainty onfuture cashin?ows exists (Kothari et al., 2009)[9].
Market-to-book under such a regime re?ects anticipated, discounted cash ?ows relative
to, essentially, the depreciation-revaluation-adjusted, liabilities-netted costs of assets
judged capable of reliable encashment. As economic information, it is probably not
directly useful, say, as an equity growth indicator but best interpreted through ?lters of
industry sector, life-cycle stage and similar.
With respect to R&D expenditures at least, A-GAAP was arguably close to the hard
assets end of the reporting spectrum, given the strict rule ?ltering out all investments
other than those with virtual certain recovery of outlays. US-GAAP was (and remains),
with minor exceptions, a hard-assets regime with respect to R&D activity.
The second regime, the broad-assets type, is consistent with support for
equity valuation as the prime ?nancial reporting objective (Kothari et al., 2009). This
regime recognises both hard and “soft” assets. Expenditures on the latter, say R&D
investments, are typically recognised as assets, for instance under AIFRS, based on
“probable” economic bene?ts. Under this regime, market-to-book re?ects anticipated,
discounted cash ?ows relative to, effectively, the depreciated-revalued, liabilities-netted
costs of the entire set of identi?able assets used to generate them. As economic
information, a key interpretation is that of a direct indicator of equity growth prospects.
Only relative to a broad-assets regime is the notion that due to the practice of immediate
expensing of expenditures, increased R&D“downward biases the book value of equity”
(Hand, 2001, p. 2).
In approaching the issue of the ?nancial disclosure versus economic reality of
R&D via market-to-book, the latter’s economic information content can, on the
above reasoning, be coherently evaluated only relative to reference disclosure regimes.
However, Hand (2001, p. 16), in arguing that “biased accounting for R&D and other
intangibles makes it harder for widely used price-based indicators such as
market-to-book to convey purely economic information” implicitly and arbitrarily
adopts broad-assets-type as the reference regime. In the analysis of Australian data
below, the interpretation of economic information conveyed by market-to-book
distinguishes such regimes.
At issue for the experience under A-GAAP, given the substantial increases and
?uctuations in market-to-book, is the extent any R&D association re?ects rate
ARJ
24,2
156
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
of pro?tability versus mode of ?nancial reporting. Hand (2001) is the only known
previous attempt at resolving this type of issue, being applied to US data for 1980-2000.
That study attempted to determine if R&D-engaged ?rms’ annual market-to-book
increases were related to R&D becoming more pro?table or, instead, becoming more
intensive[10]. R&Dwas found to be pro?table at all lag lengths estimated (including the
zero-lagcase), withthe rate of pro?t of R&Dtriplinginthe 1990s relative to the 1980s[11].
Further, market-to-book was positively related to both R&D’s pro?t rate and intensity
but the latter dominated. Hence “for R&D, the impact of biased accounting appears to
swamp that of improved economic pro?tability” (Hand, 2001, p. 18)[12].
3. R&D rate of pro?tability model
Essential to this study is the issue of R&D’s pro?t rate and its ?uctuation for
Australian ?rms over the 1988 to 2004 period. The approach adopted is based on the
fundamental relation between the values of assets and the earnings generated by them
(Lev and Sougiannis, 1996). Accordingly, de?ne, for ?rm i, gross operating income
(GOI
i,t
) for year t, derived as gross pro?t before tax (PBT) net of R&D, depreciation and
amortization, as a function of R&D expenses for year t, RD
i,t
, a proxy for other
unrecognized assets, market value of equity at end of year t, MVEq
i,t
, and recognized
(total) assets at end of year t, TA
i,t
:
GOI
i;t
¼ fðRD
i;b
MVEq
i;b
TA
i;t
Þ ð1Þ
Similar to the approaches taken by Lev and Sougiannis (1996), Hand (2001), and Ding
et al. (2007), it is proposed that equation (1) takes the linear form represented by
equation (2) with GOI a function of current ( j equal to zero) through j equal to K-year
lagged values of RD and MVEq, and end-of-year TA:
GOI
i;t
¼ a þ
X
K
j¼0
b
j
RD
i;t2j
þ
X
K
j¼0
g
j
MVEq
i;t2j
þuTA
i;t
þ
X
2004
q¼1989
p
q
Yr
q;i;t
þ
X
10
r¼2
l
r
In
r;i;t
þ1
i;t
ð2Þ
In equation (2) Yr
q,i,t
and In
r,i,t
are year and industry sector dummies, respectively,
with q indexing the year and r the industry and taking value unity where q is equal to t
and ?rm i is a member of industry r, respectively, and zero otherwise, and a and 1
i,t
are
constant and error terms, respectively.
Market value of equity is included to represent bene?ts of non-R&D intangibles not
recognized in the ?nancial statements[13]. Total assets comprise recognised assets,
both tangible and intangible.
The adoption of a fundamental rather than market-based approach to estimating
the pro?t rate of R&D avoids the circularities involved with inferring the pro?t rate
from market prices (Lev and Sougiannis, 1996, pp. 110-11) and using such estimates to
explain market-to-book ratios (Hand, 2001, p. 8). In contrast to Lev and Sougiannis
(1996) and Ding et al. (2007) but consistent with Hand (2001), the basic variables in
equation (2) are not de?ated by a scale factor such as revenue in order to mitigate
heteroscedasticity. Rather, this paper adopts corrected covariances for signi?cance
R&D
pro?tability
157
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
testing, as detailed below, and relies on Barth and Kallapur (1996) in regard to
inclusion of scale factors as independent variables (MVEq and TA in this case) in
preference to de?ation.
Following Hand (2001) and in contrast with other, related studies (Sougiannis, 1994;
Lev and Sougiannis, 1996; Ding et al., 2007), no structure is imposed on estimates of the
parameters of variables with current and lagged values. A lag structure is typically
imposed as an approach to the problemof high collinearity of current and lagged values
of explanatory variables making reliable estimation of individual coef?cients dif?cult.
As the goal is to estimate total impacts via summations or adjusted summations of
coef?cients, a lag structure is not required given that a suf?cient number of lags is
included (Hand, 2001). Large, positive autocorrelations of the explanatory variables will
mean little difference in the summed coef?cient values, as re?ected in the robustness of
Hand’s (2001) estimates across Kequal to zero, three and seven years for both gross and
net pro?t rates. This study employs lag lengths through seven years, consistent with
Sougiannis (1994) who concludes that R&D’s impact can extend over such a horizon.
Equation (2) is estimated via least squares applied on a pooled cross-section/time
series basis. Two forms of parameter estimates are derived from application of
equation (2). The ?rst form re?ects the gross GOI impact per dollar of R&D, market
capitalisation, and recognised assets. Gross impact per dollar of R&D, for instance, is
estimated as
P
K
j¼0
^
b
j
[14]. Net discounted pro?tability (NDP) per dollar is the second,
transformed version of parameter estimate computed. To obtain the NDP of, say, a
marginal dollar of R&D expense, the
^
bs are discounted at gross rate R and the $1
expenditure netted, giving
P
K
j¼0
^
b
j
R
2j
21. R is set equal to 1.11 to re?ect a uniform
nominal weighted average annual cost of capital for 1988-2004[15].
The model also estimates the parameter u, the current year gross incremental return
on recognised assets after allowing for the impacts of current and lagged R&D and
market capitalisation on ?rms’ GOI. To obtain the current year net incremental return on
recognised assets, NIR(TA), the current year cost of using the recognised assets is
subtracted from
^
u. Following Hand (2001), the latter cost is approximated by the median
of ?rms’ ratios of depreciation plus amortisation expense (DA) to total assets[16]:
NIRðTAÞ ¼
^
u 2med
DA
i;t
TA
i;t
ð3Þ
For estimation of equation (2), the upper 1 per cent of observations are Winsorised in
order to attenuate the in?uence of outliers.
4. Data and summary statistics
Data for ?nancial years 1988-2004 on all Australian listed companies for which R&D
data are available are sourced fromthe Thomson Reuters Company Analysis Database.
Apart from those de?ned above for equations (2) and (3), further foundational variables
are annual data on ?rms’ BVEq, PBT and sales revenue (Rev). The basic set of usable
observations consists of ?rm-years with non-zero R&D expenses and positive sales
revenue, market value of equity, BVEq and total assets.
Table I provides summary data on the ?rm-year observations available across
industry sectors and time, given the above speci?cations. There are 1,646 usable
?rm-year positive R&D observations from ten industry sectors over a 17-year span.
Panel A of Table I displays the industry sector composition of observations.
ARJ
24,2
158
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Several of the sectors are derived by consolidating the ?ner partitioning available from
the database[17]. The consolidation is primarily needed to obtain suf?cient
observations for the subsequent industry-based estimates of market-to-book as a
function of rate of pro?t and revenue intensity. The ten sectors derived are the basis for
the industry dummy variables used in the estimation of equation (2). A further
consolidation of the derived sectors of transport, utilities, retail and services (TURS) is
also made for the aforementioned industry-based estimation.
Manufacturing is the largest sector, constituting 28 per cent of the observations.
Pharmaceuticals and health and mining and construction each have about 18 per cent
while information and communications technology has 12 per cent and all other sectors
are each less than 10 per cent. Panel B of Table I indicates a continuing increase of
?rms with R&D activity. In 1988 there were 20 ?rms reporting R&D and other
required data, with a peak of 177 as at 2003.
Table II reports percentiles, means and standard deviations for the raw
observations on the dependent and independent variables entering the rate of
pro?tability modelling (GOI and RD, MVEq and TA, respectively). Similar statistics
are reported for the market-to-book modelling at the ?rm-year level (market-to-book or
MVEq/BVEq, and PBT net of R&D relative to BVEq or PBT_RD/BVEq and R&D
revenue intensity or RD/Rev, respectively) where however, the observations for each
?rm are augmented to include up to two leading, one trailing and three intermediate
zero-R&D observations in its time series, as available. Several features stand out.
First, the data for all variables apart from PBT_RD/BVEq have positive skew[18].
Second, the upper 5 per cent of ?rm-years have extraordinarily high R&D revenue
intensities. Third, overall, R&D intensity is weak, with the median ?rm expensing just
$0.62 per $100 of sales revenue, although this number re?ects 381 zero-R&D values
amongst 2,029 in total. Fourth, market-to-book for the typical ?rm is quite modest, as
indicated by the median of 1.7. However, a relatively exclusive set of ?rms has much
higher market-to-books, for instance, the upper 5 per cent 10.9 or more and the upper
Percentile GOI ($m) RD ($m) MVEq ($m) TA ($m) MVEq/BVEq PBT_RD/BVEq RD/Rev
99 3,031.5 69.198 21,986.1 20,237.8 32.34 1.00 36.4022
95 674.9 36.273 4,692.7 5,969.6 10.87 0.46 1.8929
75 86.4 6.149 652.5 701.0 2.91 0.21 0.0785
50 9.4 1.410 101.6 93.6 1.69 0.10 0.0062
25 20.3 0.335 22.1 20.1 1.08 20.14 0.0007
5 28.1 0.029 4.0 3.5 0.53 21.52 0.0000
1 263.6 0.004 1.5 1.4 0.27 25.44 0.0000
Mean 173.4 6.834 1,265.6 1,261 .2 3.47 20.27 2.19
SD 693.5 14.454 5,621.3 3,716.6 9.84 3.41 28.27
Notes: GOI, gross operating income: pro?t before tax adjusted for R&D, depreciation and
amortization; RD, research and development expenses; MVEq, market value of equity; TA, total assets
at the end of year; MVEq/BVEq, market value of equity relative to book value of equity; PBT_RD/
BVEq, pro?t before tax adjusted for R&D expenses relative to book value of equity; RD/Rev, R&D
expenses relative to revenue (R&D revenue intensity); all variables’ data are “raw” in that they are not
Winsorised; valid observations available for the estimation of R&Dpro?t rates (equation (2)) (GOI, RD,
MVEq, TA) (n ¼ 1,646); valid observations available for estimation of market-to-book’s relation with
R&D revenue intensity (equation (5)) (MVEq/BVEq, PBT_RD/BVEq, RD/Rev) (n ¼ 2,029)
Table II.
Descriptive statistics for
the modelling variables’
raw values, 1988-2004
R&D
pro?tability
159
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
1 per cent a stellar 32.3 or better. To attenuate the in?uence of outliers and the related
skewness, Winsorisation is applied in varying degrees to the data[19].
4.1 Annual data
Table III provides whole-of-sample data (1,646 ?rm-years) on the mean market-to-book
ratio and R&D revenue intensity trends over 1988-2004 based on the data as
Winsorised for the market-to-book modelling at the ?rm-year level. These data are
presented graphically in Figure 2.
Market-to-book trended upward overall over 1988-2004. This key, combined
market-accounting measure ?uctuated at modest levels through the late 1980s and
most of the 1990s, jumping abruptly in 1999, easing substantially through 2002, and
recovering partly thereafter. Growth across the sample period was 135 per cent. R&D
revenue intensity increased steadily over the study period (start-to-?nish growth of
1,417 per cent), with minor relapses in 1994, 1998-1999, and 2002.
5. Results
5.1 Overall pro?t rate of R&D, 1989-2004
Table IV reports the results for estimation of equation (2), with the gross pro?t
rates displayed in Panel A and net discounted rates in Panel B. Alternate lag lengths K
equal to 1, 4, and 7 years are used. To defend against the estimation impacts of
heteroscedasticity (and possible error term autocorrelation), the Newey-West
heteroscedasticity and autocorrelation consistent (HAC) covariances (Newey and
West, 1987) are computed as the basis for Wald tests on the coef?cient subset funcions,
Year n MVEq/BVEq RD/Rev
1988 20 1.630 0.030
1989 23 1.498 0.012
1990 27 1.399 0.009
1991 31 1.535 0.007
1992 41 1.550 0.037
1993 61 1.864 0.080
1994 71 2.167 0.062
1995 77 1.557 0.126
1996 111 1.911 0.172
1997 111 2.758 0.265
1998 111 2.808 0.262
1999 125 4.519 0.248
2000 147 4.019 0.284
2001 174 3.523 0.430
2002 170 3.097 0.410
2003 177 3.443 0.444
2004 169 3.837 0.455
Total Obs 1,646
Notes: The entire sample’s upper 2 per cent of MVEq/BVEq observations and upper 5 per cent of RD/
Rev observations are Winsorised to attenuate the in?uence of outliers, consistent with the data
averages reported in Table VI and the estimation reported in Tables VII and VIII; MVEq/BVEq,
market value of equity relative to book value of equity (“market-to-book”); RD/Rev, R&D expenses
relative to sales revenue (“R&D revenue intensity”)
Table III.
Means of market-to-book
and R&D revenue
intensity, 1988-2004
ARJ
24,2
160
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
whether in gross or net form[20]. Caveats arising fromfurther estimation diagnostics for
residual normality ( Jarque-Bera) and general speci?cation (RESET) (both signi?cant for
each lag length) lead to use of the x
2
statistics from the Wald tests to gauge the
signi?cance of the functions of the coef?cient subsets[21].
The marginal signi?cance levels of R&D’s gross pro?t rate estimates indicate that
R&D does not add to pro?tability at the horizons considered – given the null
hypothesis of the sum of b coef?cients equal to zero against the two-sided alternative
and adopted 0.05 size of test[22]. In contrast, the estimates of market capitalisation’s
gross impact on GOI are mixed, with a positive impact for the one-year lag length only.
Recognised assets’ gross incremental return estimates are positive at all lag lengths.
Net discounted or net incremental form GOI impact estimates are consistent with
gross values. R&D’s NDP estimates are not signi?cant at each lag length. The NDP of
market capitalisation estimates are consistently negative. However, recognised assets’
NIR estimates are positive for each lag length. The insigni?cant ?ndings for the
Australian sample are in marked contrast to those of Hand (2001) for the USA and
Ding et al. (2007) for six advanced economies.
On the above evidence, R&D in the A-GAAP era overall was not typically a material
in?uence on ?rms’ market-to-book via pro?tability. To extend this analysis and assess
the in?uence of R&D on market-to-book directly, a two-stage process is followed.
First, R&D pro?t rates are estimated and average R&D revenue intensity
Figure 2.
Mean market-to-book and
R&D revenue intensity,
1988-2004
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
198819891990
Notes: n = 1,646; the entire sample upper 2 per cent of the MVEq/BVEq observations and
upper 5 per cent of RD/Rev observations are Winsorised, consistent with the data averages
reported in Table VI and the estimation reported in Tables VII-IX; MVEq/BVEq: market
value of equity relative to BVEq (“market-to-book”); RD/Rev: R&D expenses relative to
revenue (“R&D revenue intensity”)
1991199219931994
MVEq/BVEq 10*RD/Rev
1995199619971998199920002001200220032004
R&D
pro?tability
161
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
G
O
I
i
;
t
¼
a
þ
X
K
j
¼
0
b
j
R
D
i
;
t
2
j
þ
X
K
j
¼
0
g
j
M
V
E
q
i
;
t
2
j
þ
u
T
A
i
;
t
þ
X 2
0
0
4
q
¼
1
9
8
9
þ
k
p
q
Y
r
q
;
i
;
t
þ
X
1
0
r
¼
2
l
r
I
n
r
;
i
;
t
þ
1
i
;
j
N
u
m
b
e
r
o
f
l
a
g
s
P
Kj
¼
0
^
b
j
P
Kj
¼
0
g
j
^
u
J
a
r
q
u
e
-
B
e
r
a
p
R
E
S
E
T
p
P
a
n
e
l
A
:
G
r
o
s
s
p
r
o
?
t
r
a
t
e
s
a
n
d
g
r
o
s
s
i
n
c
r
e
m
e
n
t
a
l
r
e
t
u
r
n
s
K
¼
1
0
.
7
8
0
.
0
2
0
.
1
0
,
0
.
0
0
1
a
,
0
.
0
0
1
b
(
0
.
4
6
6
)
(
0
.
0
4
7
)
(
,
0
.
0
0
1
)
K
¼
4
2
.
5
1
0
.
0
2
0
.
1
1
,
0
.
0
0
1
,
0
.
0
0
1
(
0
.
0
5
1
)
(
0
.
5
8
3
)
(
,
0
.
0
0
1
)
K
¼
7
2
.
9
2
2
0
.
0
1
0
.
1
2
,
0
.
0
0
1
0
.
0
3
7
(
0
.
0
9
8
)
(
0
.
7
9
4
)
(
,
0
.
0
0
1
)
N
u
m
b
e
r
o
f
l
a
g
s
P
Kj
¼
0
^
b
j
R
2
j
2
1
P
Kj
¼
0
g
j
R
2
j
2
1
^
u
2
m
e
d
ð
D
A
i
;
t
=
T
A
i
;
j
Þ
ð
A
d
j
:
R
2
=
C
o
e
f
f
:
F
ÿ
s
t
a
t
:
Þ
p
n
P
a
n
e
l
B
:
N
D
P
r
a
t
e
s
a
n
d
N
I
R
s
K
¼
1
2
0
.
4
8
2
0
.
9
6
0
.
0
5
0
.
9
2
1
,
2
5
5
(
0
.
6
3
4
)
(
,
0
.
0
0
1
)
(
0
.
0
0
9
)
,
0
.
0
0
1
c
K
¼
4
2
0
.
1
0
2
0
.
9
7
0
.
0
6
0
.
9
2
5
6
3
(
0
.
9
2
9
)
(
,
0
.
0
0
1
)
(
0
.
0
2
6
)
,
0
.
0
0
1
K
¼
7
2
0
.
4
1
2
0
.
9
8
0
.
0
7
0
.
9
2
2
6
8
(
0
.
7
7
3
)
(
,
0
.
0
0
1
)
(
0
.
0
0
3
)
,
0
.
0
0
1
N
o
t
e
s
:
G
O
I
i
,
t
–
g
r
o
s
s
o
p
e
r
a
t
i
n
g
i
n
c
o
m
e
:
p
r
o
?
t
b
e
f
o
r
e
t
a
x
a
d
j
u
s
t
e
d
f
o
r
R
&
D
,
d
e
p
r
e
c
i
a
t
i
o
n
a
n
d
a
m
o
r
t
i
z
a
t
i
o
n
f
o
r
?
r
m
i
i
n
y
e
a
r
t
;
R
D
i
,
t
–
r
e
s
e
a
r
c
h
a
n
d
d
e
v
e
l
o
p
m
e
n
t
e
x
p
e
n
s
e
s
f
o
r
?
r
m
i
i
n
y
e
a
r
t
;
M
V
E
q
i
,
t
–
m
a
r
k
e
t
v
a
l
u
e
o
f
e
q
u
i
t
y
f
o
r
?
r
m
i
a
t
t
h
e
e
n
d
o
f
y
e
a
r
t
;
T
A
i
,
t
–
t
o
t
a
l
a
s
s
e
t
s
o
f
?
r
m
i
a
t
t
h
e
e
n
d
o
f
y
e
a
r
t
;
Y
r
q
,
i
,
t
a
n
d
I
n
r
,
i
,
t
a
r
e
y
e
a
r
a
n
d
i
n
d
u
s
t
r
y
s
e
c
t
o
r
d
u
m
m
i
e
s
,
r
e
s
p
e
c
t
i
v
e
l
y
,
w
i
t
h
q
i
n
d
e
x
i
n
g
t
h
e
y
e
a
r
a
n
d
r
t
h
e
i
n
d
u
s
t
r
y
a
n
d
t
a
k
i
n
g
v
a
l
u
e
1
w
h
e
r
e
q
¼
t
a
n
d
i
n
d
u
s
t
r
y
r
h
a
s
?
r
m
i
a
s
a
m
e
m
b
e
r
a
n
d
z
e
r
o
o
t
h
e
r
w
i
s
e
;
a
n
d
a
a
n
d
1
i
,
t
a
r
e
c
o
n
s
t
a
n
t
a
n
d
e
r
r
o
r
t
e
r
m
s
,
r
e
s
p
e
c
t
i
v
e
l
y
;
D
A
i
,
t
–
d
e
p
r
e
c
i
a
t
i
o
n
p
l
u
s
a
m
o
r
t
i
z
a
t
i
o
n
e
x
p
e
n
s
e
f
o
r
?
r
m
i
i
n
y
e
a
r
t
;
T
A
i
,
t
–
t
o
t
a
l
a
s
s
e
t
s
o
f
?
r
m
i
a
t
t
h
e
e
n
d
o
f
y
e
a
r
t
;
t
h
e
u
p
p
e
r
1
p
e
r
c
e
n
t
o
f
G
O
I
,
R
D
,
M
V
E
q
a
n
d
T
A
o
b
s
e
r
v
a
t
i
o
n
s
a
v
a
i
l
a
b
l
e
f
o
r
i
n
c
l
u
s
i
o
n
a
r
e
W
i
n
s
o
r
i
s
e
d
t
o
a
t
t
e
n
u
a
t
e
t
h
e
i
n
?
u
e
n
c
e
o
f
o
u
t
l
i
e
r
s
;
e
s
t
i
m
a
t
i
o
n
u
s
e
s
N
e
w
e
y
-
W
e
s
t
H
A
C
-
c
o
n
s
i
s
t
e
n
t
c
o
v
a
r
i
a
n
c
e
s
;
n
u
m
b
e
r
s
i
n
p
a
r
e
n
t
h
e
s
e
s
b
e
l
o
w
c
o
e
f
?
c
i
e
n
t
e
s
t
i
m
a
t
e
s
a
r
e
m
a
r
g
i
n
a
l
s
i
g
n
i
?
c
a
n
c
e
l
e
v
e
l
s
o
f
t
h
e
W
a
l
d
t
e
s
t
x
2
s
t
a
t
i
s
t
i
c
s
o
n
t
h
e
r
e
l
e
v
a
n
t
s
u
m
o
f
p
a
r
a
m
e
t
e
r
s
o
r
u
p
a
r
a
m
e
t
e
r
e
q
u
a
l
t
o
z
e
r
o
;
a
m
a
r
g
i
n
a
l
s
i
g
n
i
?
c
a
n
c
e
l
e
v
e
l
o
f
t
h
e
J
a
r
q
u
e
-
B
e
r
a
(
1
9
8
1
)
s
t
a
t
i
s
t
i
c
f
o
r
n
o
r
m
a
l
i
t
y
o
f
r
e
s
i
d
u
a
l
s
;
b
m
a
r
g
i
n
a
l
s
i
g
n
i
?
c
a
n
c
e
l
e
v
e
l
o
f
t
h
e
R
a
m
s
e
y
,
(
1
9
6
9
)
R
E
S
E
T
s
t
a
t
i
s
t
i
c
,
u
s
i
n
g
o
n
e
?
t
t
e
d
t
e
r
m
(
s
q
u
a
r
e
o
f
?
t
t
e
d
v
a
l
u
e
s
)
,
f
o
r
g
e
n
e
r
a
l
s
p
e
c
i
?
c
a
t
i
o
n
(
o
m
i
t
t
e
d
v
a
r
i
a
b
l
e
s
,
f
u
n
c
t
i
o
n
a
l
f
o
r
m
,
a
n
d
c
o
r
r
e
l
a
t
i
o
n
b
e
t
w
e
e
n
i
n
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
s
a
n
d
r
e
s
i
d
u
a
l
s
)
;
c
m
a
r
g
i
n
a
l
s
i
g
n
i
?
c
a
n
c
e
l
e
v
e
l
o
f
F
-
s
t
a
t
i
s
t
i
c
o
n
a
l
l
c
o
e
f
?
c
i
e
n
t
s
(
a
p
a
r
t
f
r
o
m
t
h
e
c
o
n
s
t
a
n
t
t
e
r
m
)
e
q
u
a
l
t
o
z
e
r
o
Table IV.
OLS regressions of pro?t
rates on R&D, equity
market capitalization,
and incremental returns
on recognized assets,
1989-2004
ARJ
24,2
162
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and market-to-book are computed for annual and industry/multi-annual partition-based
sub-samples. Second, market-to-book is modelled in relation to the two R&Ddimensions
across each partition. The annual partition follows Hand (2001) whereas the
industry/multi-annual partition is an extension adopted in this paper to both enlarge
the sample for, and introduce industry variation to the market-to-book modelling.
5.2 Annual and industry/multi-annual pro?t rate of R&D, 1992-2004
Table V reports NDP estimates for R&D and market capitalisation and the estimate of
NIRs for recognised assets using a modi?ed form of equation (2) where the dummy
variables are dropped. Both annual and industry/multi-annual partition-based
sub-sample estimates are obtained, using four-year lags, with beginning year 1992
due to lack of pre-1988 data. Considering ?rst the annual sub-samples, the adjusted R
2
values (not tabulated) are equal to or exceed 0.92 for all years and the F-statistics on the
coef?cient set apart from constant term equal to zero (not tabulated) have p-values less
than 0.001 for all models. Some caveats do arise from the diagnostics, with eight
signi?cant values out of thirteen for the Jarque-Bera statistic and ten signi?cant
RESET statistics from the 13[23] yearly estimations.
R&D’s NDP estimates are signi?cant in four years, but negative in three of these
cases, with no consistent pattern over time. Estimates of market capitalisation’s NDP
are signi?cant and negative in all years. For recognised assets’ NIR, estimates are
signi?cant and positive in eight years and otherwise not signi?cant. Lack of evidence
for a trend in Australian ?rms’ R&D pro?t rate is in marked contrast to Hand (2001)
who ?nds positive annual rates of pro?t for US R&D, using three-year lag lengths,
in 16 of the years 1980 through 2000, and increasing from the 1980s to the 1990s.
Considering, second, the industry/multi-annual sub-samples, adjusted R
2
-values (not
tabulated) are equal to or exceed 0.90 for all years apart fromthe Foodindustry 1992-1999
where the value is 0.59. F-statistics on all coef?cients apart fromthe constant termequal
to zero (not tabulated) have p-values less than 0.01 for all models. The Jarque-Bera
statistic is signi?cant in seven out of 17 cases and the RESET statistic is signi?cant in
11 of the 17. The RD coef?cients are signi?cant in seven sub-samples, with four positive
values and three negative. No noticeable pattern is apparent in the distribution of
the pro?t rate of R&D, although the two signi?cant mincon industry values are positive.
The R&D’s on-average pro?t rate is unstable across the industry/multi-annual partition,
reinforcing the evidence for lack of consistent pro?tability.
5.3 Market-to-book in relation to R&D’s pro?t rate and intensity
Table VI displays mean market-to-book and revenue intensities of R&D for the annual
and industry/multi-period sub-samples consistent with the basis for the above pro?t rate
estimates. Market-to-book displays a somewhat similar pattern of annual ?uctuation to
that of the entire sample as reported in Table III, however with the peak at year 2004.
Annual mean R&D revenue intensity increased from 0.07 per cent in 1992 to
approximately 29.0 per cent in 2004. A strong overall upward trend is apparent and the
?uctuationacross the periodis fairlyconsistent withthat for the entire sample as reported
in Table III. R&D intensity was at a minimum in 1992 and at a maximum in 2003.
There are no stand-out industry sector differences in market-to-book apart
from the high values for Pharmhealth, this sector having the three greatest values
in the sample partition. Clearly, vast variation occurs in R&D intensity across sectors,
R&D
pro?tability
163
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
GOI
i;t
¼ a þ
P
K
j¼0
b
j
RD
i;t2j
þ
P
K
j¼0
g
j
MVEq
i;t2j
þuTA
i;t
þ1
i;j
Year
P
4
j¼0
^
b
j
R
2j
21
P
4
j¼0
g
j
R
2j
21
^
u 2medðDA
i;t
=TA
i;t
Þ J-B, RESET sgnf.
a
n
Panel A: Annual partition
1992 21.43 21.23 0.23 –, – 17
(0.305) (,0.001) (,0.001)
1993 0.68 20.92 0.03 –,
* *
19
(0.645) (,0.001) (0.286)
1994 6.60 21.07 0.12 –,
* *
23
(0.196) (,0.001) (0.002)
1995 26.78 21.01 0.13 –,
* *
27
(0.176) (,0.001) (0.001)
1996 210.72 20.92 0.10 –,
* *
31
(,0.001) (,0.001) (,0.001)
1997 24.22 20.94 0.09
*
,
* *
42
(0.063) (,0.001) (,0.001)
1998 3.30 21.07 0.11
*
,
* *
46
(0.142) (,0.001) (,0.001)
1999 2.33 21.13 0.18
*
,
* *
50
(0.042) (,0.001) (,0.001)
2000 20.75 20.96 0.07
*
,
* *
60
(0.398) (,0.001) (,0.001)
2001 24.66 20.93 0.04
*
,
* *
57
(0.046) (,0.001) (0.075)
2002 21.48 20.90 0.02
*
,
* *
59
(0.325) (,0.001) (0.573)
2003 21.74 20.93 0.02
*
, – 61
(0.030) (,0.001) (0.306)
2004 22.21 20.96 0.05
*
, – 71
(0.200) (,0.001) (0.062)
Total 563
Industry/
period
P
4
j¼0
^
b
j
R
2j
21
P
4
j¼0
g
j
R
2j
21
^
u 2medðDA
i;t
=TA
i;t
Þ
Jarque-Bera, RESET
sgnf. n
Panel B: Industry/multi-annual partition
Mincon 8.13 21.03 0.10 31
1992-1995 (0.008) (,0.001) (0.053) –,
* *
Mincon 21.88 21.03 0.13 30
1996-1998 (0.487) (,0.001) (,0.001) –,
* *
Mincon 212.17 21.00 0.06 39
1999-2001 (0.132) (,0.001) (0.106) –, –
Mincon 10.78 20.95 0.06 28
2002-2004 (0.021) (,0.001) (0.022) –, –
Manuf 2.26 21.07 0.10 30
1992-1995 (0.007) (,0.001) (0.006) –, –
Manuf 0.61 20.97 0.04 27
1996-1997 (0.775) (,0.001) (0.519)
*
,
* *
Manuf 23.19 21.09 0.17 32
1998-1999 (0.008) (,0.001) (0.211) –, –
Manuf 28.85 20.70 20.05 38
2000-2001 (,0.001) (,0.001) (0.473)
*
,
* *
Manuf 21.42 21.09 0.12 58
2002-2004 (0.719) (,0.001) (0.130)
*
,
* *
(continued)
Table V.
Sub-sample OLS
regressions of NDP rates
on R&D, market
capitalization, and NIRs
on recognized assets
(lag length ¼ 4 years)
ARJ
24,2
164
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Chem 1.15 20.90 20.00 45
1992-2004 (0.581) (,0.001) (0.870) –,
* *
ICT 21.47 21.03 0.06 25
1992-2004 (0.389) (,0.001) (0.590)
*
, –
Pharmhealth 1.40 21.05 0.08 28
1992-1999 (0.097) (,0.001) (0.005)
*
,
* *
Pharmhealth 1.08 21.00 0.06 35
2000-2002 (0.116) (,0.001) (0.005)
*
,
* *
Pharmhealth 23.61 20.87 20.02 33
2003-2004 (,0.001) (,0.001) (0.617) –, –
Food 16.68 21.06 20.00 25
1992-1999 (0.003) (,0.001) (0.990)
*
,
* *
Food 4.40 20.96 0.03 31
2000-2004 (0.192) (,0.001) (0.130) –,
* *
TURS 24.30 20.97 0.04 28
1992-2004 (0.146) (,0.001) (0.018) –,
* *
Total 563
Notes: GOI
i,t
– gross operating income: pro?t before tax adjusted for R&D, depreciation and
amortization for ?rmi in year t; RD
i,t
– research and development expenses for ?rmi in year t; MVEq
i,t
–
market value of equity for ?rm i at the end of year t; TA
i,t
– total assets of ?rm i at the end of year t;
DA
i,t
– depreciation plus amortization expense for ?rm i in year t; and a and 1
i,t
are constant and error
terms, respectively; DA
i,t
– depreciation plus amortization expense for ?rm i in year t; TA
i,t
– total
assets of ?rmi at the end of year t; see Table I for industry de?nitions; the upper 1 per cent of the GOI, RD,
MVEq and TAobservations available for inclusion are Winsorised to attenuate the in?uence of outliers;
a
*
Jarque-Bera marginal signi?cance less than 0.05;
* *
RESET marginal signi?cance less than 0.05 Table V.
Year (MVEq/BVEq)
t
(RD/Rev)
t
n Industry/Period (MVEQ/BVEq)
r
(RD/Rev)
r
n
1992 1.854 0.007 17 Mincon 1992-1995 1.723 0.009 31
1993 1.883 0.007 19 Mincon 1996-1998 1.681 0.008 30
1994 1.886 0.007 23 Mincon 1999-2001 1.574 0.006 39
1995 1.676 0.008 27 Mincon 2002-2004 2.153 0.005 28
1996 1.845 0.040 31 Manuf 1992-1995 1.660 0.007 30
1997 2.055 0.141 42 Manuf 1996-1997 1.799 0.043 27
1998 2.119 0.118 46 Manuf 1998-1999 1.815 0.026 32
1999 2.632 0.154 50 Manuf 2000-2001 1.708 0.073 38
2000 2.472 0.204 60 Manuf 2002-2004 1.808 0.071 58
2001 2.754 0.152 57 Chem 1992-2004 1.514 0.010 45
2002 2.333 0.287 59 ICT 1992-2004 2.612 0.115 25
2003 2.814 0.311 61 Pharmhealth 1992-1999 3.559 0.530 28
2004 2.921 0.290 71 Pharmhealth 2000-2002 5.778 0.910 35
Pharmhealth 2003-2004 5.065 0.947 33
Food 1992-1999 2.991 0.004 25
Food 2000-2004 1.284 0.004 31
TURS 1992-2004 2.914 0.244 28
Totals 563 563
Notes: (MVEq/BVEq)
n
– mean end-date market to book value of ?rms’ equity across sub-sample n;
(RD/Rev)
n
– mean R&D revenue intensity of ?rms across n; t represents annual partitioning and r
represents industry/multi-annual partitioning; the entire sample’s upper 2 per cent of MVEq/BVEq
observations and upper 5 per cent of RD/Ren
Table VI.
Sub-samples means of
market-to-book and
revenue intensity
R&D
pro?tability
165
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
with extraordinarily high values for Pharmhealth, and, to a lesser extent, TURS. Mincon
and Food are sectors relatively weak in their intensity of R&D, with values of less than
1 per cent in all sub-samples. The leading values of Pharmhealth’s market-to-book and
its exceptional R&D intensity likely re?ect the sector’s reliance on intellectual capital
and its typical engagement with R&D in the ordinary course of business.
The annual and industry/multi-annual R&D pro?t rate estimates (Table V) and
means of market-to-book and R&D revenue intensity for similar sample partitions
(Table VI) allow evaluation of the association of R&D’s pro?t rate and intensity with
market-to-book. Correlations for the annual data, extended to include Australian equity
market total returns, and for the industry/multi-annual data, are displayed in Table VII.
Across neither partition is market-to-book correlated with R&D’s pro?t rate. However,
market-to-book is positively and highly correlated with R&D intensity across both sets
of sub-samples. These results indicate that in a bivariate sense R&D is associated with
market-to-book, although the only reliable relation is via intensity.
Market-to-book’s association with R&D is clari?ed by extending the modelling to a
multivariate form, following Hand (2001). Using the above annual and
industry/multi-annual R&D pro?t rate estimates and mean intensities, and
additionally, for the former partition, Australian equity market annual total returns,
estimates of the dependence of mean market-to-book on these factors are obtained via
equation (4):
MVEq
BVEq
v
¼ a þI
1
bNDP_RD
v
þI
2
g
RD
Rev
v
þI
3
dRI
v
þ1
v
ð4Þ
In equation (4), (MVEq/BVEq)
n
is the mean of ?rms’ market-to-book based on the
Table VI sub-sample (v), NDP_RD
v
is equal to
P
4
j¼0
^
b
j
R
2j
21 (the estimate of
NDP(RD) for v, as per Table V, Panels A and B as relevant), (RD/Rev)
v
is the mean of
?rms’ R&D revenue intensities across v, drawn fromTable VI, RI
v
is Australian annual
total return on equities for (year) v[24], a and 1
v
are constant and error terms,
respectively, and, for cases (i) and (iii), I
1
and I
2
are equal to unity, and I
3
is equal to zero
and for case (ii), I
1
, I
2
, and I
3
are equal to unity.
MVEq/BVEq NDP(RD) RD/Rev
Panel A: Annual partition – 13 observations
NDP(RD) 0.10
(0.743)
a
RD/Rev 0.86 0.01
(,0.001) (0.982)
RI 0.01 20.26 20.12
(0.976) (0.396) (0.702)
Panel B: Industry/multi-annual partition– 17 observations
NDP(RD) 0.05
(0.859)
RD/Rev 0.94 20.15
(,0.001) (0.558)
Notes: MVEq/BVEq – market-to-book of ?rms’ equity; NDP(RD) – net discounted pro?tability of
R&D; RD/Ren – revenue intensity of R&D; RI – Australian annual total equity returns;
a
numbers in
parentheses below estimates are relevant marginal signi?cance levels
Table VII.
Pearson correlations for
mean market-to-book,
estimated R&D pro?t
rate, mean R&D intensity
and annual total equity
returns
ARJ
24,2
166
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
With R&D’s pro?t rate arrayed in competition with R&D intensity only (Table VIII),
the estimate of b, the coef?cient of the pro?t rate is not signi?cant in the annual
partition (case (i)). However, the estimate of g, the R&D intensity coef?cient is highly
signi?cant and positive. With the set of competitors expanded to include equity annual
total returns (case (ii)) consistent with Hand (2001) and Ryan (1995), the R&D pro?t
rate coef?cient remains insigni?cant while the R&D intensity coef?cient remains
signi?cant and positive. However, d, the coef?cient of equity returns is insigni?cant.
These results are mostly consistent with Hand (2001), where intensity is in?uential and
reduces otherwise univariately signi?cant R&D pro?tability to insigni?cance in
explaining US annual market-to-book. However, they differ due to the positive impact
of US equity returns reported by Hand (2001)[25].
For the industry/multi-annual partition (case (iii)), with R&D’ s pro?t rate and
intensityagainincompetition, the estimate of bis insigni?cant, giventhe null hypothesis
of equality to zero, the alterative of non-equality and the 0.05 signi?cance level. On this
basis, and on the insigni?cant result for R&D’s rate of pro?tability from the complete
sample pooled time-series and cross-section data presented above, it appears safe to rule
out R&D’s pro?t rate as on-average in?uential for market-to-book. By extension,
on-average in?uence on market-to-book of the compounded pro?t rate and intensity of
R&D, represented by the second channel in Figure 1, can similarly be ruled out.
Equation (4), cases (i) and (ii), does indicate market-to-book as increasing in R&D
intensity[26]. This potentially leads to an inference that typically, increasing
expenditure on R&D negatively biased BVEq, distorting the economic information
conveyed by market-to-book.
Two considerations count against the evidence from equation (4) for R&D intensity’s
impact on market-to-book. First and most critically, market-to-book and revenue
Case
^
b ^ g
^
d Adj. R
2
Coeff. F-stat. p n
ðMVEq=BVEqÞ
v
¼ a þI
1
bNDP_RD
v
þI
2
gðRD=RevÞ
v
þI
3
dRI
v
þ1
v
Cases (i) and (iii): I
1
¼ I
2
¼ 1, I
3
¼ 0; Case (ii): I
1
¼ I
2
¼ I
3
¼ 1.
Panel A: Annual partition
J-B p RESET p
(i) 0.01 3.18 – 0.69 0.001
a
13
(0.276) (,0.001) 0.782
b
0.228
c
(ii) 0.01 3.25 0.47 0.68 0.004 13
(0.105) (,0.001) (0.324) 0.428 0.555
Panel B: Industry/multi-annual partition
J-B p RESET p
(iii) 0.04 4.00 0.92 ,0.001 17
(0.038) (,0.001) 0.755 0.933
Notes: (MVEq/BVEq)
n
– mean period-end-date market-to-book value of ?rms’ equity across sub-
sample n; NDP_RD
n
– net discounted pro?tability of R&D for n; (RD/Rev)
n
– mean revenue intensity
of ?rms across n; RI
n
– Australian annual total return on equities for n; and a and 1
n
are constant and
error terms, respectively; estimation uses Newey-West HAC consistent covariances; numbers in
parentheses are marginal signi?cance levels of t-statistics on the relevant coef?cient being equal to
zero;
a
marginal signi?cance level of F-statistic on all coef?cients (apart from the constant term) equal
to zero;
b
marginal signi?cance level of Jarque-Bera statistic for normality of residuals;
c
marginal
signi?cance level of RESET statistic for general speci?cation
Table VIII.
OLS regressions of
market-to-book ratio on
annual and
industry/multi-annual
R&D pro?tability and
intensity, and total equity
returns, 1992-2004
R&D
pro?tability
167
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
intensity may be non-stationary, follow an upward trend, yet be independent (cases (i)
and (ii)). If this is so, the estimates are outcomes of spurious regressions[27]. Second, the
relevant caveat mentioned in Section 2.4 applies – time series of cross-section
aggregations (period means) of ?rms’ data may not well re?ect the underlying biasing
effect of R&D on equity book value at the individual ?rm level.
Finally, equation (4) does not separate the ?nancial statement handling (the fourth
channel in Figure 1) and equity market-value (the third channel in Figure 1) effects of
R&D intensity. It is therefore pertinent to investigate R&D intensity’s in?uence on
market-to-book and identifying the channel(s) of any such in?uence. Accordingly, we
take this study well beyond the approach taken by Hand (2001).
5.4 Market-to-book and R&D intensity
In dealing with the channels of R&D intensity’s in?uence on market-to-book this paper
adopts a suggestion of Hand (2001) and distinguishes a book value channel and a
market value channel. The former channel, the fourth in Figure 1, conveys R&D
intensity’s level, re?ecting equity book value’s typically negative relation with R&D
argued above (Section 2.4). The latter channel, the third in Figure 1, conveys the
current change in R&D intensity.
An increase, say, in R&D intensity is conjectured to be a signal of management’s
private information of an increase R&D’s pro?t rate, typically a positive indicator for
?rms’ equity market values. Translating to a hypothesis of market-to-book being, on
average, increasing in R&D intensity’s ?rst difference, this effect operates at the
individual ?rm, time-series level.
The approach adopted to analyse the impact of R&D intensity is to model
market-to-book at the ?rm-year level. R&D intensity’s level and ?rst difference are
arrayed against variables controlling for likely general in?uences on market-to-book.
Given the paucity of theory in this area, two control variables are adopted on a
heuristic basis. The ?rst control is the one-year lagged value of the dependent variable,
re?ecting an autoregressive, endogenous factor in market-to-book’s evolution[28]. Data
on changed pro?tability is the second control, in the form of the ?rst difference of
current PBT (excluding R&D) relative to equity book value. This formulation is
re?ected in equation (5) where (MVEq/BVEq)
i,t
is market-to-book for ?rm i in year t,
(PBT_RD/BVEq)
i,t
is, for ?rm i in year t, PBT excluding R&D expense on BVEq,
(RD/Rev)
i,t
is R&D revenue intensity for ?rm i in year t, and a and 1
i,t
are constant and
error terms, respectively:
MVEq
BVEq
i;t
¼ a þb
MVEq
BVEq
i;t21
þg
PBT_RD
BVEq
i;t
2
PBT_RD
BVEq
i;t21
þd
RD
Rev
i;t
þu
RD
Rev
i;t
2
RD
Rev
i;t21
þ1
i;t
ð5Þ
Estimation is via panel least squares with cross-section ?xed effects and White (1980)
cross-section coef?cient covariances. The panel-data approach is adopted to handle the
proposed individual ?rm, time series relation of market-to-book with R&D intensity’s
level and ?rst difference[29]. The dataset is based on that used to estimate equation (2)
but augmented by including, as available, up to two leading, one trailing and three
intermediate zero-R&D observations in each ?rm’s time series, so as to adequately
capture the effects of variation in R&D intensity. Results are displayed in Table IX.
ARJ
24,2
168
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The redundant ?xed effects F-statistic’s approximately-zero marginal signi?cance
level leads to rejection of the test’s null hypothesis of no ?xed effects. Asimilar value for
the marginal signi?cance of the Hausman (1978) x
2
-statistic leads to rejection of that
test’s null of a randomeffects speci?cation. Hence the panel data (?xed effects) approach
is con?rmed.
Lagged market-to-book in?uences current market-to-book, with a positive-value
estimate for b. None of the coef?cient estimates for the in?uence of the current ?rst
difference of R&D-adjustedpro?tabilitybefore taxper dollar bookvalue, g, R&Dintensity
level, d, and R&D intensity ?rst difference, u, are signi?cantly different from zero[30].
The lack of in?uence of R&D intensity’s level on market-to-book overturns the
relevant indications arising from estimation of equation (4). This result reinforces the
caveat discussed in Section 5.3 – that spurious regression likely applies to cases (i) and
(ii). The null ?nding suggests that despite trend-increasing mean R&D intensity over
1988-2004, ?rms’ equity book values were not typically substantially biased by the
expensing of R&Dexpenditures, and thereby market-to-book’s economic information is
not signi?cantly degraded. However, characteristics of the data generating process and
limitations of the dataset may disguise the real impact of R&D’s disclosure:
P1. R&D and other expenses may have a substitution association, due to either or
both R&D attractiveness in periods when other expenses are less onerous, or
misclassi?cation of R&D vis-a´-vis other expenses.
ðMVEq=BVEqÞ
i;t
¼ a þbðMVEq=BVEqÞ
i;t21
þgððPBT_RD=BVEqÞ
i;t
2ðPBT_RD=BVEqÞ
i;t21
Þ þ
dðRD=RenÞ
i;t
þuððRD=RenÞ
i;t
2ðRD=RenÞ
i;t21
Þ þ1
i;t
^ a
^
b
^ g
^
d
^
u n
2.25 0.20 20.41 20.04 20.13 1,677
(,0.001) (,0.001) (0.026) (0.902) (0.691)
Adj. R
2
Coeff. F-stat. p Jarque-Bera p Redundant ?xed effects F-stat. p Hausman
x
2
stat. p # cross-section units
0.56 ,0.001
a
,0.001
b
,0.001
c
,0.001
d
340
Notes: (MVEq/BVEq)
i,t
– market-to-book value of equity for ?rm i at end of period t; (PBT_RD/
BVEq)
i,t
– pro?t before tax adjusted for R&D expenses of ?rm i for year t to book value of equity for
?rm i at end of period t; (RD/Ren)
i,t
– R&D revenue intensity for ?rm i in year t; and a and 1
i,t
are
constant and error terms, respectively; estimation is via panel least squares with cross-section ?xed
effects; the data set is based on that used to estimate equation (2) but augmented by including, as
available, up to two leading, one trailing and three intermediate zero-R&D observations in each ?rm’s
time series, to adequately capture the effects of variation in R&D intensity; the upper 2 per cent of the
MVEq/BVEq observations, the lower and upper 2 per cent of the PBT_RD/BVEq observations and the
upper 5 per cent of the RD/Rev observations available for inclusion are Winsorised to attenuate excess
in?uence of outliers; estimation uses White (1980) cross-section coef?cient covariances; numbers in
parentheses are marginal signi?cance levels of t-statistics on the relevant coef?cient estimate being
equal to zero;
a
the RESET statistic is not applicable to panel estimation;
b
marginal signi?cance level of
F-statistic on all coef?cients (apart from constant) equal to zero;
c
marginal signi?cance level of Jarque-
Bera statistic for normality of residuals;
d
marginal signi?cance level of F-statistic for redundant ?xed
effects;
i
marginal signi?cance level of Hausman x
2
statistic for a random effects speci?cation
Table IX.
Panel regression of
market-to-book on
one-year lagged value,
the current ?rst
difference of PBT
(excluding R&D) relative
to book value of equity,
the current level of R&D
intensity, and the current
?rst difference of R&D
intensity, 1989-2004
a,b,c,d,e
R&D
pro?tability
169
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
P2. R&D may exhibit a complementation relation with overall net-of-R&D
pro?tability.
P1 and P2 imply strong positive correlation of RD/Rev with the ?rst difference of
PBT_RD/BVEq, leading to substantial imprecision of estimates for g and d. To evaluate
P1 and P2, RD/Rev’s dependence on PBT_RD/BVEq’s ?rst difference is estimated using
a panel-data approach, although in this case a cross-section randomeffects speci?cation
is statistically appropriate. The (untabulated) results, however, indicate the coef?cient
for the latter variable is insigni?cantly different from zero. P1 and P2 can therefore be
discounted. However, the typical time series of ?rms’ reported engagement in R&D
likely plays a role in the null result for R&D intensity level’s impact on market-to-book.
Firm engagement with R&D is mainly of short duration, evidenced by the
1,677 ?rm-years (including zero-R&D “wrap-around” and intermediate years) and
340 cross-section units (?rms), giving a mean ?rm time series length of 4.9 years. Quite
limited scope thereby exists for intra-?rm variation accounting for the rising average
R&D intensity and therefore the biasing of equity book values over 1989-2004. Rather,
the generally positive trend in mean R&D intensity over 1988-2004 apparently re?ects
?rms in the cross-section reporting greater intensities in their time series in later years.
Hence the vital role, as alluded to above, for zero-R&D observations to properly re?ect
the impact of increases (or decreases) in R&D. However, the data in this latter respect
display considerable limitations. In fact, of the 340 ?rms in the cross-section, 105 have
two leading zero-R&D observations available (guaranteeing one such observation for
inclusion in the estimation, due to the lagged data requirement), 97 have a trailing such
observation and 28 have intermediate-series-located similar observations. It is
conceivable that the shortage of zero-R&D observations may have a substantial impact
on the estimation of equation (5), especially of d, the R&D intensity-level coef?cient.
Finally, the non-in?uential role for market-to-book of R&D intensity-difference
suggests that changes in R&D intensity are not typically price-relevant signals for
capital market participants regarding managers’ changing views on R&D’s
pro?tability. This ?nding likely re?ects the above-reported lack of on-average
pro?tability of R&D for Australian listed ?rms.
6. Summary and implications
Market-to-book is a joint capital market and ?nancial reporting-determined indicator of
?rms’ ?nancial standing. Hand (2001) reported that for US ?rms over the period
1980-2000, the increasing intensity of R&D dominated the in?uence R&D pro?tability
growth when considering the overall increase and ?uctuation of market-to-book. The
intensity impact was attributed to R&D expenditures’ disclosure as an expense
imparting negative bias to equity book values on increasing R&D. It was concluded that
the economic information conveyed by market-to-book was compromised as a result.
In the A-GAAP era over 1988-2004, Australian R&D ?rms generally experienced an
uptrend and ultimately very substantial increase in R&D revenue intensity. Over the
same period, the mean market-to-book of such ?rms ?uctuated signi?cantly and,
overall, more than doubled. This paper estimated Australian listed ?rms’ R&D rate of
pro?tability, measured their R&D revenue intensity and estimated the impact of these
R&D dimensions on market-to-book via their channels of in?uence. This allowed
evaluation of the ?nancial disclosure vis-a´-vis economic reality of R&D.
ARJ
24,2
170
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
For the 1988-2004 ?rm-year data used in this study, it was typically inferred that
R&D had a zero on-average rate of pro?tability, whether estimated on pooled
time-series/cross-section, annual or industry/multi-annual bases. This ?nding revealed
the economics of most Australian ?rms’ R&D engagement as anomalous in the
international context. These ?rms’ experience might re?ect a lack of economies of scale
in R&D being achievable within the Australian industrial structure, in contrast
to the USA (Hand, 2003). Additionally, Australian ?rms may typically, in an
international context, be R&D “followers” rather than “leaders” (Lev et al., 2005), with
R&D intensity less than that of their industry and inferior sustained future (general)
pro?tability relative to industry leaders.
Modelling of market-to-book as a function of R&D’s rate of pro?tability and
intensity on annual and industry/multiannual bases did not yield convincing evidence
for either dimension’s in?uence, whether via market or BVEq-type channels. The paper
considered two channels in which market-to-book may be associated with R&D
intensity by extending modelling to the ?rm-year level. The ?rst, key R&D intensity
channel re?ects typical negative bias imparted to book value by an increased level of
R&D (or vice versa), due to R&D expenditures’ handling via write-off. This effect leads
to the hypothesis of on-average positive individual-?rm, time series relation of
market-to-book with R&D intensity’s level.
The second channel of R&D intensity re?ects the conjecture of, say, a positive effect
on the market-to-book numerator due to an increase in R&D intensity being interpreted
by the market as a signal of more favourable management perceptions of R&D’s pro?t
rate. Multivariate testing for the two R&D intensity channels using ?rm-year data
failed to con?rm either the expected intensity-level impact or an intensity-difference
in?uence on market-to-book.
What can be concluded concerning the market-to-book impact of A-GAAP’s
apparently restrictive rule forcing expensing except for projects with near-certain
recoupment of R&D expenditure? First, the information conveyed by market-to-book
was almost certainly consistent with the implicit intent of the (for R&D expenditures at
least) near hard assets, ef?cient contracting-oriented in-place regime.
Second, on the evidence of lack of (positive) relation of market-to-book to R&D
intensity, R&D apparently did not typically impart signi?cant negative bias to ?rms’
equity book values despite the apparently mostly positive trend in ?rms’ typical R&D
engagement over the period under study. This benign outcome probably re?ects the real
impact of cross-?rmrather than intra-?rmincreasing R&Dintensity over 1988-2004 but
likely also the limitations of the available data. With reference to broad assets, equity
valuation orientation as the reference disclosure regime, the result at face value implies
on-average immaterial adverse effects for market-to-book’s information on, say, ?rms’
growth opportunities. In consequence, A-GAAP apparently served a “best of both
worlds” role.
Finally, turning attention to the AIFRS regime, the less restrictive rule
on capitalisation of R&D expenditures ideally results in equity book values with less
systematic bias than, say, values likely under continuing A-GAAP with on-average
R&D pro?tability[31]. Market-to-book in such case typically conveys economic
information more consistent with the new regime’s orientation. However, Kothari et al.
(2009) argue that under IFRS the handling of R&D expenditure is characterised by
excess managerial discretion on asset valuations. If so, ?rms’ equity book values may
R&D
pro?tability
171
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
frequently possess signi?cant, idiosyncratic, managerial policy-related bias under, say,
increasing R&D expenditures. Consequently, for such ?rms, the market-to-book
information for growth prospects will be compromised relative to the equity valuation
objective implicit in AIFRS.
The asset recognition rule for R&D expenditures under AIFRS applies widely to
internally generated intangibles. Equity book values’ bias and related impacts
on market-to-book therebyconstitutes an issue extending beyond the scope of A-GAAP’s
traditional R&D expenditures, strengthening its appeal for future research attention.
Notes
1. Hereafter in the text, except as otherwise indicated explicitly or via the context, “R&D” is to
be taken as “research and development expenditures handled as expenses in the ?nancial
statements”.
2. See White et al. (2003, Ch. 19) for basic concepts in this area. The suggested market-book value
difference decomposition best aligns with the transaction-theoretic approach where the values
are expectations of future transaction costs (Hodgson et al., 1993). An alternative, direct
economic value-theoretic decomposition is suggested by Upton (2003) based on “market
assessments” and includes a component re?ecting unrecognized identi?able intangibles.
3. Whether market-to-book can serve as a basis for measurement of intellectual capital is
problematical as, logically, the “values” of assets such as R&D cannot be estimated via
market prices (Lev and Sougiannis, 1996, pp. 110-11).
4. Even if R&Dexpenditures are, on average, pro?table, they are often highly risky. Kothari et al.
(2002) ?nd that the variability of future earnings is greater for R&Doutlays than for property,
plant and equipment expenditures while Bens et al. (2004) report that R&D intensity is
positively associated with the volatility of forward earnings realizations. However, Amir et al.
(2007) ?nd that R&Dcontributes more to subsequent earnings variability than physical asset
capital expenditure only in relatively R&D-intensive industries. Dif?culty in predicting future
bene?ts for particular expenditures may arise partly from the inherent uncertainty of R&D
program technical outcomes and partly from the nature of markets where even technically
successful R&D ventures may not prove pro?table (Stickels, 1996).
5. In the USA, the debate among researchers has focused on the almost universally required
full-expensing of R&Dunder Statement of Financial Accounting Standards (SFAS) No. 2(FASB,
1974). This rule, along with the argued poor quality and quantity of R&D disclosure is widely
representedinthe USliterature as re?ectingmisguidedpolicy, creatinginformationasymmetries
with adverse cost-of-capital and insider trading implications (Aboody and Lev, 2000; Lev, 2001).
6. This typical effect of R&D on BVEq can be analysed most easily in the base case of a “BVEq
steady state” ?ow of R&D projects and a superimposed positive or negative R&D shock.
7. Relevant parameters of R&D-related BVEq are likely the relative levels of R&D and
associated revenues for the pre-existing process and for the increment or decrement, and,
additionally, regarding the latter, the time phasing of the revenues.
8. The impact of R&D on BVEq’s evolution is left to intuition, although it could be treated by
numerical illustration, simulation or (formal) analysis.
9. Kothari et al. (2009) note that internally-generated intangibles such as R&D typically have
minimal value in liquidation and highly uncertain cash ?ow realizations.
10. Average market-to-book approximately doubled over 1981-1999 and increased by about
50 per cent over 1980-2000 in Hand’s (2001) US ?rms. For 1980 through 2000 mean revenue
intensity of R&D spending for sample ?rms increased sevenfold.
ARJ
24,2
172
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
11. Hand (2003), extending Hand (2001) ?nds that average, net pro?tability of R&D increased
with the scale of expenditure and that this increasing returns-to-scale effect itself increased
over the 1980-2000 sample period. The inference is made of a competition imperative with
the largest R&D investments creating cost-based and/or network-derived barriers to entry
and resulting super-normal rents. This returns-to-scale ?nding may be relevant to
Australia’s R&D pro?tability experience, as referred to in Section 6 below.
12. In drawing such an inference Hand (2001) attributes the market-to-book impact of increased
R&D intensity entirely to the negative biasing of equity book values arising with increasing
R&D expensing. The analysis, however, ignores the channel of in?uence ?owing from R&D
intensity to market value of equity, given an above-zero rate of pro?tability, via anticipated
increased discounted net cash ?ows (second channel in Figure 1).
13. Equation (2) abstracts from two potentially important explanatory variables representing
possible intangibles – advertising, and general and administrative expenses. The former
category likely re?ects part of brand intangibles and the latter further brand plus personnel
intangibles (Lev and Sougiannis, 1996; Hand, 2001; Ding et al., 2007). Unfortunately, separate
data were not available for these items, creating a potential omitted variables and resultant
bias problem. Inclusion of MVEq as a proxy for non-R&D unrecognized assets is intended to
ameliorate this issue.
14. This assumes that the lag j coef?cient from GOI in year t represents adequately the response
of GOI in year t þ j to R&D in year t (Hand, 2001, p. 9). Strictly, the estimated effect of R&D
in year t should be the sum of the lag 0 coef?cient from year t, the lag 1 coef?cient from year
t þ 1, the lag 2 coef?cient from(for) year t þ 2 and so on, as per Sougiannis (1994, p. 59).
15. GOI is calculated before adjustments re?ecting the costs of debt and equity capital. Auniform
rate glosses over variation in the risk levels of recognized and unrecognized assets and
amongst the latter, differences between R&D and the non-R&D intangible assets represented
by equity market capitalization. The discount rate adopted is computed in three stages. First,
by calculating the average of quarterly Australian corporate bond yields (middle rate) over Q1
1988 through Q4 2004, obtained from Datastream (AUSCRPB). Second, calculating the
average of Australian equity market annual total returns over 1988-2004, derived from the
relevant Datastream index (TOTMKAU(RI)). Third, the weight for equity ?nancing was
approximated by calculating the average BVEq/TA of the sample, with the debt proportion
set to the complement. Thanks to Abul Shamsuddin, University of Newcastle, Australia for
advice on the nominal weighted average annual cost of capital derivation.
16. NIR(TA) is not equivalent to but rather is a component of the NDP of recognized assets. Not
estimating the NDP of recognized assets will only bias the coef?cients in equation (2) if the
difference between NDP(TA) and NIP(TA) is correlated with the explanatory variables
(Hand, 2001).
17. The ten industry sectors used were consolidated fromthe company analysis classi?cation. The
consolidation re?ected the need for suf?cient observations to support the industry/multi-annual
analysis of market-to-bookinterms of R&Dpro?tabilityversus R&Drevenue intensityreported
in Table VIII. The consolidation is as follows: “Mincon” from“Mining”, “Oil and gas”, “Forestry
and paper” and “Construction and building materials”; “Manuf” from“Aerospace and defence”,
“Automobiles and parts”,“Diversi?ed industrial”, “Engineering and machinery”, “Electronic
and electrical”, “Household goods and textiles”, “Steel and other metals” and “Packaging”;
“Utilities” from “Water”, “Electicity” and “Gas distribution”; “ICT” from “Software and
computer”, “Telecommunications” and “IThardware”; “Pharmhealth” from“Pharmaceuticals”
and “Health”; “Food” from “Food producers and processors” and “Beverages”; “Retail” from
“Food and drug retailers”, “General retail” and “Personal care and household”; “Services” from
“Distributors”, “Leisure andentertainment andhotels”, “Media andphotography”, “Real estate”,
R&D
pro?tability
173
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
“Support services”, “Investment”, “Specialty and other ?nance” and “Unknown”; other sectors
are as per the database classi?cation.
18. Positive skew is re?ected by the mean exceeding the median and vice versa for the negative
case. See, for instance, Lewis and Orav (1989, p. 153).
19. The Winsorisation involves the upper 1 per cent of the observations on GOI, RD, MVEq, and
TA for the analyses presented in Tables IV and V. For the observations underlying
the summaries presented in Tables III and VI and the analyses presented in Tables VII-IX
the Winsorisation involves, as relevant, the upper 2 per cent of MVEq/BVEq, the lower and
upper 2 per cent of PBT_RD/BVEq, and the upper 5 per cent of RD/Ren.
20. In all three lag-length cases, the error term is heteroscedastic, as indicated by the marginal
signi?cance values for the White (1980) test statistic (not tabulated) being less than 0.05.
21. First, for each lag length, the error term is not normally distributed, as indicated by the
marginal signi?cance levels of the Jarque-Bera test statistic (Bera and Jarque, 1981). The
principal feature underlying non-normality of the residuals is leptokurtosis.
Transformations (such as taking logarithms) did not correct the problem and would
obscure the direct, per dollar pro?tability interpretation of the coef?cients. However, as the
samples are relatively large, violation of normality is likely to be of minor consequence as the
relevant test statistics asymptotically follow the appropriate distributions, despite error
non-normality (see, for instance, Brooks, 2002, p. 182). Second, for each lag length, the
marginal signi?cance values of the RESET statistic (Ramsey, 1969) suggest some
shortcoming(s) in the speci?cation. The RESET is however purely diagnostic, providing no
guidance as to improving the speci?cation (Brooks, 2002, pp. 194-7). A search for a superior
model for ?rm pro?tability is beyond the scope of this paper.
22. A similar framework for inference (whether gross or net discounted functions of or single
coef?cients) is applied throughout this paper, i.e. a zero value under the null, a two-sided
(non-zero value) alternative and a 0.05 signi?cance level.
23. Via the White statistic, the error term is heteroscedastic in 1999 through 2002 only. However,
again the Newey-West HAC covariances are computed for all sub-samples considered in this
section. The Wald test x
2
statistics are used to assess signi?cance of the estimates for such
sub-samples.
24. Australian annual total returns on equities are derived from the relevant index,
TOTMKAU(RI) (Datastream).
25. Reliability of the inferences fromestimation of equation (4) are supported by the insigni?cant
values of the diagnostic statistics for residual normality, speci?cation and homoscedasticity
(the latter are not tabulated).
26. The equation 4, case (iii) estimate of R&D intensity coef?cient g is highly signi?cant and
positive but is not relevant to the question of the biasing of equity book value due to the
mainly cross-section nature of the data. In this case, intensity serves as a control. The
estimate is likely biased due to omitted in?uential variables re?ecting industry-varying
non-R&D intellectual, organisational capital-generating identi?able intangibles or
synergistic, going concern premium-yielding effects.
27. See, for instance, Brooks (2002, Ch. 7). This consideration casts doubt on Hand’s (2001)
results on market-to-book as a function of dimensions of R&D given that the US annual
market-to-book, R&D pro?tability, and R&D revenue intensity series all have the
appearance of trending upward over time.
28. Use of a lagged dependent variable violates the classical linear regression model assumption
of exogeneity of the independent variables. However, the standard properties of least
ARJ
24,2
174
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
squares will on appropriate assumptions, apply asymptotically. See, for instance, Greene
(2008, pp. 72-75).
29. Fundamentally, the panel data with cross-section ?xed effects approach is adopted to handle
inter-?rm heterogeneity due to omitted, unobserved variables correlated with those included.
The approach defends against biased and inconsistent estimates otherwise arising in
estimating the purely time series-type conjectured dependence of ?rms’ market-to-book on
R&D intensity’s level and ?rst difference. See, for instance, Greene (2008, Ch. 9).
30. Again, in regard to the estimation of equation (5), the implication of error non-normality
indicated by the marginal signi?cance of the Jarque-Bera statistic is attenuated by the
sample size.
31. Key to the more permissive rule on capitalization of R&Dunder AIFRS is that for expenditure
in the development phase only, AASB 138 Intangible Assets requires and permits only
recognition as an asset where “it is probable that the expected future economic bene?ts that
are attributable to the asset will ?owto the entity . . . ” (paragraph 21 (a), emphasis added). See
for instance, Institute of Chartered Accountants in Australia (2011, pp. 884-915).
References
Aboody, D. and Lev, B. (2000), “Information asymmetry, R&D, and insider gains”, Journal of
Finance, Vol. 55 No. 6, pp. 2747-66.
Ahmed, K. and Falk, H. (2006), “The value relevance of management’s policy choice of research
and development expenditure reporting: evidence from Australia”, Journal of Accounting
and Public Policy, Vol. 25 No. 4, pp. 231-64.
Amir, E., Guan, Y. and Livne, G. (2007), “The association of R&D and capital expenditures with
subsequent earnings variability”, Journal of Business Finance and Accounting, Vol. 34
Nos 1/2, pp. 222-46.
Barth, M.E. and Kallapur, S. (1996), “The effects of cross-sectional scale differences on regression
results in empirical accounting research”, Contemporary Accounting Research, Vol. 13
No. 2, pp. 527-67.
Bens, D.A., Hanna, J.D. and Zhang, X.F. (2004), “Research and development, risk and stock
returns”, working paper, Chicago Graduate School of Business.
Bera, A. and Jarque, C.M. (1981), “An ef?cient large-sample test for normality of observations
and regression residuals”, ANU Working Papers in Economics 40, Canberra.
Bosworth, D. and Rogers, M. (1998), “Research and development, intangible assets and the
performance of large Australian companies”, Melbourne Institute Working Paper No. 2/98.
Brooks, C. (2002), Introductory Econometrics for Finance, Cambridge University Press,
Cambridge.
Canibano, L., Garcia-Ayuso, M. and Sanchez, M.P. (2000), “Accounting for intangibles: a literature
review”, Journal of Accounting Literature, Vol. 19, pp. 102-30.
Cohen, W. and Levinthal, D. (1989), “Innovation and learning: the two faces of R&D –
implications for the analysis of R&D investment”, Economic Journal, Vol. 99, pp. 419-23.
Ding, Y., Stolowy, H. and Tenenhaus, M. (2007), “R&D productivity: an exploratory international
study”, Review of Accounting and Finance, Vol. 6 No. 1, pp. 86-101.
Fama, E.F. and French, K.R. (1992), “The cross-section of expected stock returns”, Journal of
Finance, Vol. 47 No. 2, pp. 427-65.
Fama, E.F. and French, K.R. (1993), “Common risk factors in the returns on stocks and bonds”,
Journal of Financial Economics., Vol. 33 No. 1, pp. 3-56.
R&D
pro?tability
175
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
FASB (1974), Statement of Financial Accounting Standards No. 2: Accounting for Research and
Development Costs, Financial Accounting Standards Board, Stamford, CT.
Feltham, G.A. and Ohlson, J.A. (1995), “Valuation and clean surplus accounting for operating and
?nancial activities”, Contemporary Accounting Research, Vol. 11, pp. 698-732.
Feltham, G.A. and Ohlson, J.A. (1996), “Uncertainty resolution and the theory of depreciation
measurement”, Journal of Accounting Research, Vol. 34 No. 2, pp. 209-34.
Greene, W.H. (2008), Econometric Analysis, 6th ed., Pearson Prentice-Hall, Upper Saddle River, NJ.
Hand, J.R.M. (2001), “The economic versus accounting impacts of R&D on US market-to-book
ratios”, working paper, University of North Carolina at Chapel Hill, 25 September, SSRN:
285108.
Hand, J.R.M. (2003), “The increasingreturns to scale of intangibles”, inLev, B. andHand, J.R.M. (Eds),
Intangible Assets: Values, Measures, and Risks, Oxford University Press, Oxford, pp. 303-31.
Hausman, J.A. (1978), “Speci?cation tests in econometrics”, Econometrica, Vol. 46, pp. 1251-72.
Hirschey, M. and Weygandt, J. (1985), “Amortization policy for advertising and research and
development expenditures”, Journal of Accounting Research, Vol. 23 No. 1, pp. 326-35.
Hodgson, A., Okunev, J. and Willett, R. (1993), “Accounting for intangibles: a theoretical
perspective”, Accounting & Business Research, Vol. 23 No. 90, pp. 138-50.
Institute of Chartered Accountants in Australia (2011), Financial Reporting Handbook 2011,
Wiley, Milton.
Johnson, L.D. and Pazderka, B. (1993), “Firm value and investment in R&D”, Managerial and
Decision Economics, Vol. 14 No. 1, pp. 15-24.
Kallapur, S. and Trombley, M.A. (1999), “The association between investment opportunity set
proxies and realised growth”, Journal of Business Finance and Accounting, Vol. 26,
pp. 505-19.
Kothari, S.P., Laguerre, T.E. and Leone, A.J. (2002), “Capitalization versus expensing: evidence on
the uncertainty of future earnings from capital expenditures versus R&D outlays”,
Review of Accounting Studies, Vol. 7 No. 4, pp. 355-82.
Kothari, S.P., Ramanna, K. and Skinner, D.J. (2009), “What should GAAP look like? A survey and
economic analysis”, MIT Sloan/Harvard Business/Chicago Booth Working Paper,
17 September, SSRN: 1413775.
Lev, B. (2001), Intangibles: Management, Measurement, and Reporting, Brookings Institution
Press, Washington, D.C.
Lev, B. and Sougiannis, T. (1996), “The capitalization, amortization, and value-relevance of
R&D”, Journal of Accounting and Economics, Vol. 21, pp. 107-38.
Lev, B., Radhakrishnan, S. and Ciftci, M. (2005), “The stock market valuation of R&D leaders”,
Stern School of Business, New York University, working paper December.
Lewis, P.A.W. and Orav, E.J. (1989), Simulation Methodology for Statisticians, Operations
Analysts, and Engineers, Wadsworth, Paci?c Grove, CA.
Mairesse, J. and Mohen, P. (1995), Research and Development and Productivity, A Survey of the
Economic Literature, INSEE, Paris.
Mairesse, J. and Sassenou, M. (1991), “R&D and productivity: a survey of econometric studies at
the ?rm level”, Science, Technology and Industry Review, Vol. 7, pp. 131-47.
Newey, W.K. and West, K.D. (1987), “A simple positive-de?nite heteroskedasticity and
autocorrelation-consistent covariance matrix”, Econometrica, Vol. 55, pp. 703-8.
Ohlson, J.A. (1995), “Earnings, equity book value, and dividends in equity valuation”,
Contemporary Accounting Research, pp. 661-87.
ARJ
24,2
176
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Penman, S.H. (2004), Financial statement analysis and security valuation, 2nd ed.,
McGraw-Hill/Irwin, New York, NY.
Ramsey, J.B. (1969), “Tests for speci?cation errors in classical linear least squares regression
analysis”, Journal of the Royal Statistical Society B, Vol. 31 No. 2, pp. 350-71.
Ryan, S.G. (1995), “A model of accrual measurement with implications for the evolution of the
book-to-market ratio”, Journal of Accounting Research, Vol. 33 No. 1, pp. 95-112.
Sougiannis, T. (1994), “The accounting based valuation of corporate R&D”, The Accounting
Review, Vol. 69 No. 1, pp. 44-68.
Stickels, G. (1996), “Authorities aim for the objective on R&D”, Business Review Weekly,
18 March, pp. 99-100.
Tobin, J. (1969), “A general equilibrium approach to monetary theory”, Journal of Money, Credit,
and Banking, Vol. 1, pp. 15-27.
Upton, W.S. Jr (2003), “Challenges from the new economy for business and ?nancial reporting”,
FASB Special Report, reprinted in Lev, B. and Hand, J.R.M. (eds) Intangible Assets: Values,
Measures, and Risks, Oxford University Press, Oxford, pp. 469-86.
White, H. (1980), “A heteroskedasticity-consistent covariance matrix estimator and a direct test
for eteroskedasticity”, Econometrica, Vol. 48, pp. 817-38.
White, G.I., Sondhi, A.C. and Fried, D. (2003), The Analysis and Use of Financial Statements,
Wiley, Hoboken, NJ.
Zambon, S. and Associates (2003), “Study on the measurement of intangible assets and
associated reporting practices”, prepared for Commission of the European Communities
Enterprise Directorate General.
About the authors
Kamran Ahmed is a Professor in the School of Accounting at La Trobe University. He holds a
PhD in Accounting awarded by the Australian National University (ANU), has numerous
publications in internationally refereed journals in the accounting discipline and has previously
held academic appointments at Victoria University of Wellington, University of New England
and the ANU.
John Hillier is a Research Fellowin the School of Accounting at La Trobe University. He holds
a PhD in Accounting awarded by the University of Otago, has several publications in
internationally refereed journals in the ?elds of accounting and ?nance and has previously held
academic appointments at the University of New England, University of New South Wales and
the University of Southern Queensland. John Hillier is the corresponding author and can be
contacted at: [email protected]
Elisabeth Tanusasmita graduated as BCom (Hons) from La Trobe University in 2006 and
now works as a ?nancial analyst in the private sector.
R&D
pro?tability
177
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Alireza Vafaei, Kamran Ahmed, Paul Mather. 2015. Board Diversity and Financial Performance in the Top
500 Australian Firms. Australian Accounting Review 25:10.1111/auar.2015.25.issue-4, 413-427. [CrossRef]
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_321645276.pdf
The purpose of this paper is to assess the financial disclosure vis-a´-vis economic reality of
research and development (R&D) expensed by Australian firms under the pre-2005 Australian
generally accepted accounting principles (A-GAAP) regime via the lens of market-to-book.
Accounting Research Journal
R&D profitability, intensity and market-to-book: evidence from Australia
Kamran Ahmed J ohn Hillier Elisabeth Tanusasmita
Article information:
To cite this document:
Kamran Ahmed J ohn Hillier Elisabeth Tanusasmita, (2011),"R&D profitability, intensity and market-to-book:
evidence from Australia", Accounting Research J ournal, Vol. 24 Iss 2 pp. 150 - 177
Permanent link to this document:http://dx.doi.org/10.1108/10309611111163691
Downloaded on: 24 January 2016, At: 21:13 (PT)
References: this document contains references to 45 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 945 times since 2011*
Users who downloaded this article also downloaded:
Astrid Schmulian, Stephen Coetzee, (2011),"Class absenteeism: reasons for non-attendance and
the effect on academic performance", Accounting Research J ournal, Vol. 24 Iss 2 pp. 178-194 http://
dx.doi.org/10.1108/10309611111163718
Kym Butcher, Graeme Harrison, J ill McKinnon, Philip Ross, (2011),"Auditor appointment in
compulsory audit tendering", Accounting Research J ournal, Vol. 24 Iss 2 pp. 104-149 http://
dx.doi.org/10.1108/10309611111163682
J acqueline Birt, Greg Shailer, (2011),"Forecasting confidence under segment reporting", Accounting
Research J ournal, Vol. 24 Iss 3 pp. 245-267http://dx.doi.org/10.1108/10309611111186993
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
R&D pro?tability, intensity
and market-to-book: evidence
from Australia
Kamran Ahmed and John Hillier
School of Accounting, Faculty of Law and Management, La Trobe University,
Melbourne, Australia, and
Elisabeth Tanusasmita
Bombardier Transportation, Milton, Australia
Abstract
Purpose – The purpose of this paper is to assess the ?nancial disclosure vis-a´-vis economic reality of
research and development (R&D) expensed by Australian ?rms under the pre-2005 Australian
generally accepted accounting principles (A-GAAP) regime via the lens of market-to-book.
Design/methodology/approach – The authors estimated ?rms’ R&D pro?t rate, measured R&D
revenue intensity and modelled the impacts of these and related economic factors, via economic and
?nancial disclosure channels, on market-to-book using data for 1988-2004.
Findings – R&D, on average, was pro?t neutral and had undetectable impacts on market-to-book
whether via equity valuation or ?nancial disclosure.
Research limitations/implications – Market-to-book’s information content is best viewed as
conditional on the reference disclosure regime. Australian ?rms’ typically at best minimal R&D
pro?tability is an international anomaly. Data limitations in terms of the generating process and
availability mean that R&D’s impact on market-to-book via ?nancial reporting is not de?nitively
determined.
Practical implications – Restrictive rules on the capitalization of intangible asset-related
expenditures under A-GAAP apparently did not adversely impact market-to-book’s economic
information. AIFRS’s more permissive rule risks compromisingmarket-to-book’s reliabilityinsucha role.
Originality/value – For Australia, the paper is anticipated to be the ?rst to estimate the pro?t rate of
R&D, measure the intensity of R&D, and model R&D’s in?uence on the market-to-book ratio. It develops
a framework for the economic and ?nancial reporting impacts of investments on a key indicator of ?rms’
?nancial standing and contributes to the debate on identi?able intangibles’ disclosure.
Keywords Australia, Listed companies, Research and development, Financial reporting,
Market-to-book, Intangibles, A-GAAP, AIFRS
Paper type Research paper
1. Aim and overview
From the late 1980s through the mid-2000s, Australia experienced a steady rise in the
number of public, listed corporations reporting research and development (R&D)
expenses in their annual ?nancial statements (Table I, Panel B), and in the average
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1030-9616.htm
The authors’ names are listed in alphabetical order. The authors acknowledge the comments of
participants at the Global Accounting and Organisational Change conference, Melbourne,
9-11 July 2008 and the Asia-Paci?c Conference on International Accounting Issues, Paris,
9-12 November 2008, the advice of Abdul Shamsuddin, and the ?nancial support of
La Trobe University.
ARJ
24,2
150
Accounting Research Journal
Vol. 24 No. 2, 2011
pp. 150-177
qEmerald Group Publishing Limited
1030-9616
DOI 10.1108/10309611111163691
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
intensity (relative to sales revenue) of such outlays (Table III and Figure 2).
Contemporaneously, the average market value of equity to book value of equity (BVEq)
(“market-to-book”) for R&D-active ?rms increased substantially (Table III and
Figure 2)[1].
Incorporating ?rms’ equity prices and the net book value of identi?able assets,
market-to-book conveys informationrelevant to ?rms’ prospects for strategies including
investments in intangible assets such as those associated with R&D engagement.
Market-to-book’s performance in such a role, relative to R&D, depends jointly on the
equity market’s assessment of R&Dexpenditure’s economic impact, and on its handling
in the ?nancial statements, and the ?nancial disclosure regime. To the extent that R&D
expenditure is viewed by the capital market as an investment, the anticipated economic
bene?ts are by de?nition re?ected in ?rms’ equity prices. However, such R&D
expenditures’ immediate write-off implies ?rms’ net book values are in part disclosed
Panel A: Industry membership
Sector Abbreviation Obs
Obs (%)Mining and construction
a
Mincon 297 18.0
Manufacturing
a
Manuf 465 28.3
Chemicals Chem 74 4.5
Transport Incl. in TURS 5 0.3
Utilities
a
Incl. in TURS 26 1.6
Information and communications technology
a
ICT 191 11.6
Pharmaceuticals and health
a
Pharmhealth 304 18.5
Food
a
Food 129 7.8
Retail
a
Incl. in TURS 22 1.3
Services
a
Incl. in TURS 133 8.1
Total 1,646 100
Panel B: Number per year
Fiscal year n
1988 20
1989 23
1990 27
1991 31
1992 41
1993 61
1994 71
1995 77
1996 111
1997 111
1998 111
1999 125
2000 147
2001 174
2002 170
2003 177
2004 169
Total 1,646
Notes: The sectors marked with super script “a” are industry sectors consolidated from the ?ner
company analysis sectors as detailed in Note 13; TURS (transport, utilities, retail, services) re?ects a
further consolidation
Table I.
Companies reporting
R&D by industry and
frequency, 1988-2004
R&D
pro?tability
151
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
on a basis inconsistent with their market valuations, with consequences for
market-to-book’s economic information.
Framed generally, this paper investigates the ?nancial disclosure vis-a´-vis economic
reality of Australian ?rms’ expensed R&D efforts in the era of Australian generally
accepted accounting principles (A-GAAP). Market-to-book serves as a lens allowing
appraisal of this issue. If pro?tability is R&D’s main in?uence over market-to-book, the
latter conveys primarily “economic” information regarding R&D efforts. Alternatively,
if R&D expenditures’ handling in the ?nancial statements is the dominant impact then
there is potential for degradation of market-to-book’s information. Such potential is
realized by R&Dexpenditures’ expensing under increasing (or decreasing) outlays, with
the effect the systematic biasing of ?rms’ equity book values. This disclosure artefact’s
symptom is, as argued below, market-to-book’s positive relation to R&D intensity.
This paper adopts the basic rationale andmodelling approachapplied byHand(2001)
to US experience with R&D in the 1980s and 1990s. However, it offers a more explicit,
deeper theoretical framework and extends the array of statistical analyses to achieve
arguably more reliable conclusions. The modelling is based on two key dimensions of
?rms’ R&D, with market-to-book’s information on R&Danalysed via its response to the
dimensions’ channels of in?uence. Additionally, interpretation of market-to-book is
argued to be conditional on the reference (preferred) disclosure regime.
The ?rst of R&D’s above-referred dimensions is its rate of pro?tability. Derivation
is via statistical estimation with ?nancial statement data. A little-researched but key
issue in its own right, it has an (equity) market value channel of in?uence, via
market-to-book’s numerator.
R&D’s second dimension, revenue intensity, is computed as R&D relative to sales
revenue andadopted to mitigate size-of-?rmeffects inthe analysis, consistent with Hand
(2001). Crucially, R&D intensity’s level has a book value channel, via market-to-book’s
denominator. Additionally, intensity level, compounded with rate of pro?tability, has a
market value channel. Finally, R&D intensity’s ?rst difference has a market value
channel. Figure 1 shows a schema for R&D’s channels of in?uence over market-to-book.
This paper employs a dataset consisting of 1,646 Australian listed company
?rm-years with reported R&Dand supplemented by zero-R&Dobservations within two
Figure 1.
Schema for R&D’s
channels of in?uence over
market-to-book
Market value of equity
Book value of equity
Rate of profitability
1
2
2
3
4
1: Impact of rate of profitability
2: Impact of level of intensity at a given rate of profitability
3: Impact of management signal on profit-rate parameter(s)
4: Impact due to financial statement handling of expenditure
Revenue intensity
ARJ
24,2
152
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
years of succeeding or one year of preceding R&D, over 1988-2004 and across ten
industry sectors. The paper ?rst attempts to determine the on-average rate of
pro?tability of R&D. Then, the impact of such pro?tability versus the biasing effect on
equity book values due to R&D expenditures’ disclosure as expenses are evaluated in
terms of market-to-book’s information role over 1988-2004.
Firms’ on-average R&D pro?t rates are estimated for the overall sample on a
pooled cross-section/time series basis across 1989-2004 and on annual and
industry/multi-annual partition-based sub-samples for 1992-2004. That the
on-average net discounted R&D pro?t rate is zero cannot be rejected for the overall
sample and for most period and industry/multi-period sub-samples. Australian lack of
pro?tability is entirely at variance with international evidence.
Two approaches are used to model market-to-book in relation to R&D. First, the two
sets of partitioned-sample R&D pro?t rate estimates are combined with the respective
mean R&D intensities and market-to-book ratios computed from similar sub-samples
and the impacts of R&D on market-to-book estimated. No reliable evidence emerges for
either a R&D rate of pro?tability or revenue intensity effect.
Second, R&Dintensity, using ?rm-year level data over 1988-2004, is decomposed into
variables re?ecting book value and market value channels of in?uence and arrayed in
competition with lagged market-to-book and the current change in overall pro?tability to
model market-to-book. No evidence emerges for market-to-book’s economic information
being degraded by A-GAAP’s stringent R&D standard relative to either “ef?cient
contracting” or “equityvaluation”-orienteddisclosure regimes. This null result maybe an
artefact of ?rms’ typical short-durationreportedR&Dengagement, inter-?rmrather than
intra-?rm variation in R&D intensity over the era modelled, and limitations of the data.
The approach and ?ndings enhance understanding of the impact of disclosure
methods for investments in R&D and, potentially, other intangibles, on the information
conveyed via market-to-book. Implications are drawn with respect to the
now-historical A-GAAP and the currently prevailing AIFRS disclosure regime.
From this point, the paper is structured as follows: Section 2 deals with key concepts
and surveys the literature. The research method is developed in Section 3, data and
summary statistics are described in Section 4, followed by presentation of results in
Section 5. Section 6 provides a summary and implications.
2. Concepts and literature review
The literature dealing with market-to-book in general and especially the relation to
R&D is scarce, with Australian applications noticeably absent. However, the following
reviews precursors to this paper. Attention is given to market-to-book’s potential
information, R&D’s economic bene?ts and rates of pro?tability, the equity market
response to R&D pro?tability, and, R&D’s reporting under A-GAAP and its impact on
equity book values’ properties relative to measurement and to disclosure regimes and
the evidence on market-to-book’s re?ection of R&D under US-GAAP.
2.1 Market-to-book
Market-to-book is a synthetic equity market and accounting data-derived indicator,
ideally informative on ?rms’ economic standing and prospects. One interpretation sees
market-to-book as a measure of growth opportunities and thereby as a proxy for the
unobservable investment opportunity set (Kallapur and Trombley, 1999). An additional
R&D
pro?tability
153
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
role is in predicting abnormal earnings (Ohlson, 1995; Feltham and Ohlson, 1995, 1996).
Another is as a ?nancial distress risk indicator (Fama and French, 1992, 1993). Afurther
interpretation regards market-to-book as an analogue of (or even empirical substitute
for) Tobin’s Q, the market value of a ?rmdivided by its “book” value at replacement cost
(Tobin, 1969). Market-to-book has also been viewed as re?ecting the market’s valuation
of unrecognised assets (Zambon and Associates, 2003).
The market premium (Penman, 2004, p. 40) may be viewed as the sum of three
components – the market versus book value difference of recognised assets, the value
of unrecognised identi?able intangibles, and economic goodwill[2]. Market-to-book
may therefore be useful as an indicator of intellectual (or organizational) capital to the
extent that ?rst, identi?able intangible assets such as R&D investments are
unrecognised and second, that the synergistic effect re?ected in the going concern
premium is a synonymous concept[3].
2.2 R&D’s rate of pro?tability
The question of the relation between R&D expenditures and subsequent economic
bene?ts has attracted signi?cant interest in recent times, with wide recognition of R&D
activity as a factor inherent in technological change (Bosworth and Rogers, 1998). Cohen
and Levinthal (1989) delineate R&D’s role as twofold – ?rst, to generate new, applicable
knowledge and second, to develop “absorptive capacity”, the ability to recognize,
assimilate and exploit others’ knowledge, with both roles potentially contributing to
?rms’ economic capabilities. Contributions to the management and economics
literatures (Mairesse and Sassenou, 1991; Mairesse and Mohen, 1995; Bosworth and
Rogers, 1998) argue R&D tends to generate net economic bene?ts.
A typical ?nding in accounting studies is that, on average, R&D expenditure
parameters attract positive estimates for future and even contemporaneous rates of
pro?tability[4]. Key US studies include Sougiannis (1994) who reports an estimate of a
one-dollar increase in R&D leading to a two-dollar increase in pro?t over a seven-year
period. Hand (2001) estimates net present value pro?tability per dollar of R&D of 0.51,
0.45, and $0.44 at lag lengths of zero, three, and seven years, respectively. Ding et al.
(2007) adopt an international perspective, estimating R&D internal rates of return on a
country basis for ?rms in six advanced economies. Using data from 1991-2000 with
six-year lag lengths, positive returns are reported, ranging from 18 per cent for the
USA and Switzerland to 36 per cent for Japan.
2.3 R&D and equities
Widely referenced ?rm valuation models imply that equity values are increasing
functions of anticipated earnings (White et al., 2003, Ch. 19). Consequently, if R&D
typically increases future earnings, a positive, contemporaneous relationship between
equity prices and R&D or, alternatively, equity returns and growth in R&D should be
observed, consistent with previous studies (Canibano et al., 2000).
Various national studies document positive impacts of R&D expenditures on ?rms’
equity price levels and returns including, for the USA, Hirschey and Weygandt (1985)
and, for Canadian ?rms, Johnson and Pazderka (1993). Sougiannis (1994) using US data
estimates that a one-dollar increase in R&D leads to a ?ve-dollar increase in equity
market value. Additionally, for the USA, Lev and Sougiannis (1996) report estimated
net R&D investment and assets to be positively associated with contemporaneous
ARJ
24,2
154
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
equity returns and prices, respectively. For Australia, contemporaneous equity values
are found to be positively related to the stock of intangibles and intensity of R&D
expenditures (Bosworth and Rogers, 1998) and Ahmed and Falk (2006) report that
R&D expenditures are equities market value-relevant.
R&D’s in?uence on the market-to-book numerator is clearly implied by the above
effects on equity valuation. In terms of a positive impact: ?rst, an increased pro?t rate,
and second, an increased level of investment given a constant, positive pro?t rate. The
relevant market-value channels (the ?rst and second channels, respectively, in Figure 1)
convey R&D’s rate of pro?tability in the former case and R&D’s intensity, interacting
with the pro?t rate, in the latter.
2.4 R&D expenditure disclosure and implications for market-to-book
The reporting of R&D expenditures has long been a contentious issue in many
jurisdictions due to the typical long duration but highly uncertain nature of associated
economic bene?ts. Under A-GAAP, very limited managerial discretion existed with
respect to the disclosure of R&D expenditures under AASB 1011, applicable up to the
AIFRS start on January 1, 2005. R&D costs had to be expensed except where future
bene?ts were expected, beyond reasonable doubt, to equal or exceed those costs and
any future costs necessary to give rise to the future economic bene?ts, in which case
asset recognition (i.e. capitalisation and amortisation of the costs) was permitted. This
rule was presumably designed to force expensing of all R&D expenditures apart from
those at the low end of the risk spectrum and, therefore, was only a little less stringent
than the equivalent standard applying in the USA[5]. Hence this paper’s key,
policy-related issue – did A-GAAP allow market-to-book to properly re?ect the
economic impact of R&D?
To answer the preceding question, two issues are relevant, one measurement and the
other disclosure-type. The measurement issue concerns the stock-?ow dynamics of
R&D, associated revenue, and the BVEq. Owing to immediate expense and lagged
revenue recognition, an R&D project imparts a non-positive and typically negative bias
relative to the ?nal impact on BVEq at termination. The effect attenuates subsequently,
eventually fully. Worth emphasizing is that bias exists given project-generated revenue
is both non-zero and that any recognition is delayed beyond the period of expensing –
lack of complete expenditure recoupment is consistent with bias.
If the ?ow of R&D is time-invariant, the stock of BVEq will follow the path set by
the former’s parameters. Changes in the ?ow of R&D necessarily disturb such a path.
It follows from the aforementioned project-bias that, for instance, an increased level of
R&D causes an initially non-positive, typically negative and subsequently attenuating
displacement of BVEq from its path otherwise, and vice versa[6].
A displacement of BVEq’s path of either sign does not necessarily imply a similarly
signed response of the level of BVEq. The latter depends on the pre-existing
R&D-related BVEq process relative to the effects of the increase or decrease of R&D[7].
However, informal analysis and even, ideally, intuition indicates that, given the
above-described dynamics of displacement, BVEq will typically be negatively related
to R&D[8].
The above-argued BVEq-R&Dassociation extends immediately, on an “other things
equal” basis to the hypothesis of market-to-book as, on average, positively related to
R&D’s scale, represented by R&D intensity, at the individual ?rm, time series level.
R&D
pro?tability
155
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This effect of the expensing of R&D expenditure is presented as the fourth channel of
R&D’s in?uence on market-to-book in Figure 1. Extension to time series of cross-section
aggregations of ?rm data (via R&D intensity period means) is somewhat tenuous,
though central to Hand’s (2001) modelling and conclusions on “biased accounting”
(Hand, 2001, pp. 1, 3, 18). Additionally, the association’s empirical existence can be
attributed purely to R&D’s disclosure effect (via equity book value) only if the
interaction of R&D’s pro?t rate and intensity, as discussed above, can be ruled out as a
possible in?uence on average.
The above-referenced disclosure issue concerns the implications of alternate
?nancial reporting regimes for market-to-book’s economic information. In examining
these implications it is useful to distinguish two fundamentally different types of
regime with respect to asset recognition.
The ?rst regime, the hard-assets type, is consistent with support for ef?cient
contracting as the prime objective of ?nancial reporting. Recognition requirements
include well-de?ned property rights ensuring assets are controlled and separable,
saleable andthat adequate certainty onfuture cashin?ows exists (Kothari et al., 2009)[9].
Market-to-book under such a regime re?ects anticipated, discounted cash ?ows relative
to, essentially, the depreciation-revaluation-adjusted, liabilities-netted costs of assets
judged capable of reliable encashment. As economic information, it is probably not
directly useful, say, as an equity growth indicator but best interpreted through ?lters of
industry sector, life-cycle stage and similar.
With respect to R&D expenditures at least, A-GAAP was arguably close to the hard
assets end of the reporting spectrum, given the strict rule ?ltering out all investments
other than those with virtual certain recovery of outlays. US-GAAP was (and remains),
with minor exceptions, a hard-assets regime with respect to R&D activity.
The second regime, the broad-assets type, is consistent with support for
equity valuation as the prime ?nancial reporting objective (Kothari et al., 2009). This
regime recognises both hard and “soft” assets. Expenditures on the latter, say R&D
investments, are typically recognised as assets, for instance under AIFRS, based on
“probable” economic bene?ts. Under this regime, market-to-book re?ects anticipated,
discounted cash ?ows relative to, effectively, the depreciated-revalued, liabilities-netted
costs of the entire set of identi?able assets used to generate them. As economic
information, a key interpretation is that of a direct indicator of equity growth prospects.
Only relative to a broad-assets regime is the notion that due to the practice of immediate
expensing of expenditures, increased R&D“downward biases the book value of equity”
(Hand, 2001, p. 2).
In approaching the issue of the ?nancial disclosure versus economic reality of
R&D via market-to-book, the latter’s economic information content can, on the
above reasoning, be coherently evaluated only relative to reference disclosure regimes.
However, Hand (2001, p. 16), in arguing that “biased accounting for R&D and other
intangibles makes it harder for widely used price-based indicators such as
market-to-book to convey purely economic information” implicitly and arbitrarily
adopts broad-assets-type as the reference regime. In the analysis of Australian data
below, the interpretation of economic information conveyed by market-to-book
distinguishes such regimes.
At issue for the experience under A-GAAP, given the substantial increases and
?uctuations in market-to-book, is the extent any R&D association re?ects rate
ARJ
24,2
156
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
of pro?tability versus mode of ?nancial reporting. Hand (2001) is the only known
previous attempt at resolving this type of issue, being applied to US data for 1980-2000.
That study attempted to determine if R&D-engaged ?rms’ annual market-to-book
increases were related to R&D becoming more pro?table or, instead, becoming more
intensive[10]. R&Dwas found to be pro?table at all lag lengths estimated (including the
zero-lagcase), withthe rate of pro?t of R&Dtriplinginthe 1990s relative to the 1980s[11].
Further, market-to-book was positively related to both R&D’s pro?t rate and intensity
but the latter dominated. Hence “for R&D, the impact of biased accounting appears to
swamp that of improved economic pro?tability” (Hand, 2001, p. 18)[12].
3. R&D rate of pro?tability model
Essential to this study is the issue of R&D’s pro?t rate and its ?uctuation for
Australian ?rms over the 1988 to 2004 period. The approach adopted is based on the
fundamental relation between the values of assets and the earnings generated by them
(Lev and Sougiannis, 1996). Accordingly, de?ne, for ?rm i, gross operating income
(GOI
i,t
) for year t, derived as gross pro?t before tax (PBT) net of R&D, depreciation and
amortization, as a function of R&D expenses for year t, RD
i,t
, a proxy for other
unrecognized assets, market value of equity at end of year t, MVEq
i,t
, and recognized
(total) assets at end of year t, TA
i,t
:
GOI
i;t
¼ fðRD
i;b
MVEq
i;b
TA
i;t
Þ ð1Þ
Similar to the approaches taken by Lev and Sougiannis (1996), Hand (2001), and Ding
et al. (2007), it is proposed that equation (1) takes the linear form represented by
equation (2) with GOI a function of current ( j equal to zero) through j equal to K-year
lagged values of RD and MVEq, and end-of-year TA:
GOI
i;t
¼ a þ
X
K
j¼0
b
j
RD
i;t2j
þ
X
K
j¼0
g
j
MVEq
i;t2j
þuTA
i;t
þ
X
2004
q¼1989
p
q
Yr
q;i;t
þ
X
10
r¼2
l
r
In
r;i;t
þ1
i;t
ð2Þ
In equation (2) Yr
q,i,t
and In
r,i,t
are year and industry sector dummies, respectively,
with q indexing the year and r the industry and taking value unity where q is equal to t
and ?rm i is a member of industry r, respectively, and zero otherwise, and a and 1
i,t
are
constant and error terms, respectively.
Market value of equity is included to represent bene?ts of non-R&D intangibles not
recognized in the ?nancial statements[13]. Total assets comprise recognised assets,
both tangible and intangible.
The adoption of a fundamental rather than market-based approach to estimating
the pro?t rate of R&D avoids the circularities involved with inferring the pro?t rate
from market prices (Lev and Sougiannis, 1996, pp. 110-11) and using such estimates to
explain market-to-book ratios (Hand, 2001, p. 8). In contrast to Lev and Sougiannis
(1996) and Ding et al. (2007) but consistent with Hand (2001), the basic variables in
equation (2) are not de?ated by a scale factor such as revenue in order to mitigate
heteroscedasticity. Rather, this paper adopts corrected covariances for signi?cance
R&D
pro?tability
157
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
testing, as detailed below, and relies on Barth and Kallapur (1996) in regard to
inclusion of scale factors as independent variables (MVEq and TA in this case) in
preference to de?ation.
Following Hand (2001) and in contrast with other, related studies (Sougiannis, 1994;
Lev and Sougiannis, 1996; Ding et al., 2007), no structure is imposed on estimates of the
parameters of variables with current and lagged values. A lag structure is typically
imposed as an approach to the problemof high collinearity of current and lagged values
of explanatory variables making reliable estimation of individual coef?cients dif?cult.
As the goal is to estimate total impacts via summations or adjusted summations of
coef?cients, a lag structure is not required given that a suf?cient number of lags is
included (Hand, 2001). Large, positive autocorrelations of the explanatory variables will
mean little difference in the summed coef?cient values, as re?ected in the robustness of
Hand’s (2001) estimates across Kequal to zero, three and seven years for both gross and
net pro?t rates. This study employs lag lengths through seven years, consistent with
Sougiannis (1994) who concludes that R&D’s impact can extend over such a horizon.
Equation (2) is estimated via least squares applied on a pooled cross-section/time
series basis. Two forms of parameter estimates are derived from application of
equation (2). The ?rst form re?ects the gross GOI impact per dollar of R&D, market
capitalisation, and recognised assets. Gross impact per dollar of R&D, for instance, is
estimated as
P
K
j¼0
^
b
j
[14]. Net discounted pro?tability (NDP) per dollar is the second,
transformed version of parameter estimate computed. To obtain the NDP of, say, a
marginal dollar of R&D expense, the
^
bs are discounted at gross rate R and the $1
expenditure netted, giving
P
K
j¼0
^
b
j
R
2j
21. R is set equal to 1.11 to re?ect a uniform
nominal weighted average annual cost of capital for 1988-2004[15].
The model also estimates the parameter u, the current year gross incremental return
on recognised assets after allowing for the impacts of current and lagged R&D and
market capitalisation on ?rms’ GOI. To obtain the current year net incremental return on
recognised assets, NIR(TA), the current year cost of using the recognised assets is
subtracted from
^
u. Following Hand (2001), the latter cost is approximated by the median
of ?rms’ ratios of depreciation plus amortisation expense (DA) to total assets[16]:
NIRðTAÞ ¼
^
u 2med
DA
i;t
TA
i;t
ð3Þ
For estimation of equation (2), the upper 1 per cent of observations are Winsorised in
order to attenuate the in?uence of outliers.
4. Data and summary statistics
Data for ?nancial years 1988-2004 on all Australian listed companies for which R&D
data are available are sourced fromthe Thomson Reuters Company Analysis Database.
Apart from those de?ned above for equations (2) and (3), further foundational variables
are annual data on ?rms’ BVEq, PBT and sales revenue (Rev). The basic set of usable
observations consists of ?rm-years with non-zero R&D expenses and positive sales
revenue, market value of equity, BVEq and total assets.
Table I provides summary data on the ?rm-year observations available across
industry sectors and time, given the above speci?cations. There are 1,646 usable
?rm-year positive R&D observations from ten industry sectors over a 17-year span.
Panel A of Table I displays the industry sector composition of observations.
ARJ
24,2
158
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Several of the sectors are derived by consolidating the ?ner partitioning available from
the database[17]. The consolidation is primarily needed to obtain suf?cient
observations for the subsequent industry-based estimates of market-to-book as a
function of rate of pro?t and revenue intensity. The ten sectors derived are the basis for
the industry dummy variables used in the estimation of equation (2). A further
consolidation of the derived sectors of transport, utilities, retail and services (TURS) is
also made for the aforementioned industry-based estimation.
Manufacturing is the largest sector, constituting 28 per cent of the observations.
Pharmaceuticals and health and mining and construction each have about 18 per cent
while information and communications technology has 12 per cent and all other sectors
are each less than 10 per cent. Panel B of Table I indicates a continuing increase of
?rms with R&D activity. In 1988 there were 20 ?rms reporting R&D and other
required data, with a peak of 177 as at 2003.
Table II reports percentiles, means and standard deviations for the raw
observations on the dependent and independent variables entering the rate of
pro?tability modelling (GOI and RD, MVEq and TA, respectively). Similar statistics
are reported for the market-to-book modelling at the ?rm-year level (market-to-book or
MVEq/BVEq, and PBT net of R&D relative to BVEq or PBT_RD/BVEq and R&D
revenue intensity or RD/Rev, respectively) where however, the observations for each
?rm are augmented to include up to two leading, one trailing and three intermediate
zero-R&D observations in its time series, as available. Several features stand out.
First, the data for all variables apart from PBT_RD/BVEq have positive skew[18].
Second, the upper 5 per cent of ?rm-years have extraordinarily high R&D revenue
intensities. Third, overall, R&D intensity is weak, with the median ?rm expensing just
$0.62 per $100 of sales revenue, although this number re?ects 381 zero-R&D values
amongst 2,029 in total. Fourth, market-to-book for the typical ?rm is quite modest, as
indicated by the median of 1.7. However, a relatively exclusive set of ?rms has much
higher market-to-books, for instance, the upper 5 per cent 10.9 or more and the upper
Percentile GOI ($m) RD ($m) MVEq ($m) TA ($m) MVEq/BVEq PBT_RD/BVEq RD/Rev
99 3,031.5 69.198 21,986.1 20,237.8 32.34 1.00 36.4022
95 674.9 36.273 4,692.7 5,969.6 10.87 0.46 1.8929
75 86.4 6.149 652.5 701.0 2.91 0.21 0.0785
50 9.4 1.410 101.6 93.6 1.69 0.10 0.0062
25 20.3 0.335 22.1 20.1 1.08 20.14 0.0007
5 28.1 0.029 4.0 3.5 0.53 21.52 0.0000
1 263.6 0.004 1.5 1.4 0.27 25.44 0.0000
Mean 173.4 6.834 1,265.6 1,261 .2 3.47 20.27 2.19
SD 693.5 14.454 5,621.3 3,716.6 9.84 3.41 28.27
Notes: GOI, gross operating income: pro?t before tax adjusted for R&D, depreciation and
amortization; RD, research and development expenses; MVEq, market value of equity; TA, total assets
at the end of year; MVEq/BVEq, market value of equity relative to book value of equity; PBT_RD/
BVEq, pro?t before tax adjusted for R&D expenses relative to book value of equity; RD/Rev, R&D
expenses relative to revenue (R&D revenue intensity); all variables’ data are “raw” in that they are not
Winsorised; valid observations available for the estimation of R&Dpro?t rates (equation (2)) (GOI, RD,
MVEq, TA) (n ¼ 1,646); valid observations available for estimation of market-to-book’s relation with
R&D revenue intensity (equation (5)) (MVEq/BVEq, PBT_RD/BVEq, RD/Rev) (n ¼ 2,029)
Table II.
Descriptive statistics for
the modelling variables’
raw values, 1988-2004
R&D
pro?tability
159
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
1 per cent a stellar 32.3 or better. To attenuate the in?uence of outliers and the related
skewness, Winsorisation is applied in varying degrees to the data[19].
4.1 Annual data
Table III provides whole-of-sample data (1,646 ?rm-years) on the mean market-to-book
ratio and R&D revenue intensity trends over 1988-2004 based on the data as
Winsorised for the market-to-book modelling at the ?rm-year level. These data are
presented graphically in Figure 2.
Market-to-book trended upward overall over 1988-2004. This key, combined
market-accounting measure ?uctuated at modest levels through the late 1980s and
most of the 1990s, jumping abruptly in 1999, easing substantially through 2002, and
recovering partly thereafter. Growth across the sample period was 135 per cent. R&D
revenue intensity increased steadily over the study period (start-to-?nish growth of
1,417 per cent), with minor relapses in 1994, 1998-1999, and 2002.
5. Results
5.1 Overall pro?t rate of R&D, 1989-2004
Table IV reports the results for estimation of equation (2), with the gross pro?t
rates displayed in Panel A and net discounted rates in Panel B. Alternate lag lengths K
equal to 1, 4, and 7 years are used. To defend against the estimation impacts of
heteroscedasticity (and possible error term autocorrelation), the Newey-West
heteroscedasticity and autocorrelation consistent (HAC) covariances (Newey and
West, 1987) are computed as the basis for Wald tests on the coef?cient subset funcions,
Year n MVEq/BVEq RD/Rev
1988 20 1.630 0.030
1989 23 1.498 0.012
1990 27 1.399 0.009
1991 31 1.535 0.007
1992 41 1.550 0.037
1993 61 1.864 0.080
1994 71 2.167 0.062
1995 77 1.557 0.126
1996 111 1.911 0.172
1997 111 2.758 0.265
1998 111 2.808 0.262
1999 125 4.519 0.248
2000 147 4.019 0.284
2001 174 3.523 0.430
2002 170 3.097 0.410
2003 177 3.443 0.444
2004 169 3.837 0.455
Total Obs 1,646
Notes: The entire sample’s upper 2 per cent of MVEq/BVEq observations and upper 5 per cent of RD/
Rev observations are Winsorised to attenuate the in?uence of outliers, consistent with the data
averages reported in Table VI and the estimation reported in Tables VII and VIII; MVEq/BVEq,
market value of equity relative to book value of equity (“market-to-book”); RD/Rev, R&D expenses
relative to sales revenue (“R&D revenue intensity”)
Table III.
Means of market-to-book
and R&D revenue
intensity, 1988-2004
ARJ
24,2
160
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
whether in gross or net form[20]. Caveats arising fromfurther estimation diagnostics for
residual normality ( Jarque-Bera) and general speci?cation (RESET) (both signi?cant for
each lag length) lead to use of the x
2
statistics from the Wald tests to gauge the
signi?cance of the functions of the coef?cient subsets[21].
The marginal signi?cance levels of R&D’s gross pro?t rate estimates indicate that
R&D does not add to pro?tability at the horizons considered – given the null
hypothesis of the sum of b coef?cients equal to zero against the two-sided alternative
and adopted 0.05 size of test[22]. In contrast, the estimates of market capitalisation’s
gross impact on GOI are mixed, with a positive impact for the one-year lag length only.
Recognised assets’ gross incremental return estimates are positive at all lag lengths.
Net discounted or net incremental form GOI impact estimates are consistent with
gross values. R&D’s NDP estimates are not signi?cant at each lag length. The NDP of
market capitalisation estimates are consistently negative. However, recognised assets’
NIR estimates are positive for each lag length. The insigni?cant ?ndings for the
Australian sample are in marked contrast to those of Hand (2001) for the USA and
Ding et al. (2007) for six advanced economies.
On the above evidence, R&D in the A-GAAP era overall was not typically a material
in?uence on ?rms’ market-to-book via pro?tability. To extend this analysis and assess
the in?uence of R&D on market-to-book directly, a two-stage process is followed.
First, R&D pro?t rates are estimated and average R&D revenue intensity
Figure 2.
Mean market-to-book and
R&D revenue intensity,
1988-2004
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
198819891990
Notes: n = 1,646; the entire sample upper 2 per cent of the MVEq/BVEq observations and
upper 5 per cent of RD/Rev observations are Winsorised, consistent with the data averages
reported in Table VI and the estimation reported in Tables VII-IX; MVEq/BVEq: market
value of equity relative to BVEq (“market-to-book”); RD/Rev: R&D expenses relative to
revenue (“R&D revenue intensity”)
1991199219931994
MVEq/BVEq 10*RD/Rev
1995199619971998199920002001200220032004
R&D
pro?tability
161
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
G
O
I
i
;
t
¼
a
þ
X
K
j
¼
0
b
j
R
D
i
;
t
2
j
þ
X
K
j
¼
0
g
j
M
V
E
q
i
;
t
2
j
þ
u
T
A
i
;
t
þ
X 2
0
0
4
q
¼
1
9
8
9
þ
k
p
q
Y
r
q
;
i
;
t
þ
X
1
0
r
¼
2
l
r
I
n
r
;
i
;
t
þ
1
i
;
j
N
u
m
b
e
r
o
f
l
a
g
s
P
Kj
¼
0
^
b
j
P
Kj
¼
0
g
j
^
u
J
a
r
q
u
e
-
B
e
r
a
p
R
E
S
E
T
p
P
a
n
e
l
A
:
G
r
o
s
s
p
r
o
?
t
r
a
t
e
s
a
n
d
g
r
o
s
s
i
n
c
r
e
m
e
n
t
a
l
r
e
t
u
r
n
s
K
¼
1
0
.
7
8
0
.
0
2
0
.
1
0
,
0
.
0
0
1
a
,
0
.
0
0
1
b
(
0
.
4
6
6
)
(
0
.
0
4
7
)
(
,
0
.
0
0
1
)
K
¼
4
2
.
5
1
0
.
0
2
0
.
1
1
,
0
.
0
0
1
,
0
.
0
0
1
(
0
.
0
5
1
)
(
0
.
5
8
3
)
(
,
0
.
0
0
1
)
K
¼
7
2
.
9
2
2
0
.
0
1
0
.
1
2
,
0
.
0
0
1
0
.
0
3
7
(
0
.
0
9
8
)
(
0
.
7
9
4
)
(
,
0
.
0
0
1
)
N
u
m
b
e
r
o
f
l
a
g
s
P
Kj
¼
0
^
b
j
R
2
j
2
1
P
Kj
¼
0
g
j
R
2
j
2
1
^
u
2
m
e
d
ð
D
A
i
;
t
=
T
A
i
;
j
Þ
ð
A
d
j
:
R
2
=
C
o
e
f
f
:
F
ÿ
s
t
a
t
:
Þ
p
n
P
a
n
e
l
B
:
N
D
P
r
a
t
e
s
a
n
d
N
I
R
s
K
¼
1
2
0
.
4
8
2
0
.
9
6
0
.
0
5
0
.
9
2
1
,
2
5
5
(
0
.
6
3
4
)
(
,
0
.
0
0
1
)
(
0
.
0
0
9
)
,
0
.
0
0
1
c
K
¼
4
2
0
.
1
0
2
0
.
9
7
0
.
0
6
0
.
9
2
5
6
3
(
0
.
9
2
9
)
(
,
0
.
0
0
1
)
(
0
.
0
2
6
)
,
0
.
0
0
1
K
¼
7
2
0
.
4
1
2
0
.
9
8
0
.
0
7
0
.
9
2
2
6
8
(
0
.
7
7
3
)
(
,
0
.
0
0
1
)
(
0
.
0
0
3
)
,
0
.
0
0
1
N
o
t
e
s
:
G
O
I
i
,
t
–
g
r
o
s
s
o
p
e
r
a
t
i
n
g
i
n
c
o
m
e
:
p
r
o
?
t
b
e
f
o
r
e
t
a
x
a
d
j
u
s
t
e
d
f
o
r
R
&
D
,
d
e
p
r
e
c
i
a
t
i
o
n
a
n
d
a
m
o
r
t
i
z
a
t
i
o
n
f
o
r
?
r
m
i
i
n
y
e
a
r
t
;
R
D
i
,
t
–
r
e
s
e
a
r
c
h
a
n
d
d
e
v
e
l
o
p
m
e
n
t
e
x
p
e
n
s
e
s
f
o
r
?
r
m
i
i
n
y
e
a
r
t
;
M
V
E
q
i
,
t
–
m
a
r
k
e
t
v
a
l
u
e
o
f
e
q
u
i
t
y
f
o
r
?
r
m
i
a
t
t
h
e
e
n
d
o
f
y
e
a
r
t
;
T
A
i
,
t
–
t
o
t
a
l
a
s
s
e
t
s
o
f
?
r
m
i
a
t
t
h
e
e
n
d
o
f
y
e
a
r
t
;
Y
r
q
,
i
,
t
a
n
d
I
n
r
,
i
,
t
a
r
e
y
e
a
r
a
n
d
i
n
d
u
s
t
r
y
s
e
c
t
o
r
d
u
m
m
i
e
s
,
r
e
s
p
e
c
t
i
v
e
l
y
,
w
i
t
h
q
i
n
d
e
x
i
n
g
t
h
e
y
e
a
r
a
n
d
r
t
h
e
i
n
d
u
s
t
r
y
a
n
d
t
a
k
i
n
g
v
a
l
u
e
1
w
h
e
r
e
q
¼
t
a
n
d
i
n
d
u
s
t
r
y
r
h
a
s
?
r
m
i
a
s
a
m
e
m
b
e
r
a
n
d
z
e
r
o
o
t
h
e
r
w
i
s
e
;
a
n
d
a
a
n
d
1
i
,
t
a
r
e
c
o
n
s
t
a
n
t
a
n
d
e
r
r
o
r
t
e
r
m
s
,
r
e
s
p
e
c
t
i
v
e
l
y
;
D
A
i
,
t
–
d
e
p
r
e
c
i
a
t
i
o
n
p
l
u
s
a
m
o
r
t
i
z
a
t
i
o
n
e
x
p
e
n
s
e
f
o
r
?
r
m
i
i
n
y
e
a
r
t
;
T
A
i
,
t
–
t
o
t
a
l
a
s
s
e
t
s
o
f
?
r
m
i
a
t
t
h
e
e
n
d
o
f
y
e
a
r
t
;
t
h
e
u
p
p
e
r
1
p
e
r
c
e
n
t
o
f
G
O
I
,
R
D
,
M
V
E
q
a
n
d
T
A
o
b
s
e
r
v
a
t
i
o
n
s
a
v
a
i
l
a
b
l
e
f
o
r
i
n
c
l
u
s
i
o
n
a
r
e
W
i
n
s
o
r
i
s
e
d
t
o
a
t
t
e
n
u
a
t
e
t
h
e
i
n
?
u
e
n
c
e
o
f
o
u
t
l
i
e
r
s
;
e
s
t
i
m
a
t
i
o
n
u
s
e
s
N
e
w
e
y
-
W
e
s
t
H
A
C
-
c
o
n
s
i
s
t
e
n
t
c
o
v
a
r
i
a
n
c
e
s
;
n
u
m
b
e
r
s
i
n
p
a
r
e
n
t
h
e
s
e
s
b
e
l
o
w
c
o
e
f
?
c
i
e
n
t
e
s
t
i
m
a
t
e
s
a
r
e
m
a
r
g
i
n
a
l
s
i
g
n
i
?
c
a
n
c
e
l
e
v
e
l
s
o
f
t
h
e
W
a
l
d
t
e
s
t
x
2
s
t
a
t
i
s
t
i
c
s
o
n
t
h
e
r
e
l
e
v
a
n
t
s
u
m
o
f
p
a
r
a
m
e
t
e
r
s
o
r
u
p
a
r
a
m
e
t
e
r
e
q
u
a
l
t
o
z
e
r
o
;
a
m
a
r
g
i
n
a
l
s
i
g
n
i
?
c
a
n
c
e
l
e
v
e
l
o
f
t
h
e
J
a
r
q
u
e
-
B
e
r
a
(
1
9
8
1
)
s
t
a
t
i
s
t
i
c
f
o
r
n
o
r
m
a
l
i
t
y
o
f
r
e
s
i
d
u
a
l
s
;
b
m
a
r
g
i
n
a
l
s
i
g
n
i
?
c
a
n
c
e
l
e
v
e
l
o
f
t
h
e
R
a
m
s
e
y
,
(
1
9
6
9
)
R
E
S
E
T
s
t
a
t
i
s
t
i
c
,
u
s
i
n
g
o
n
e
?
t
t
e
d
t
e
r
m
(
s
q
u
a
r
e
o
f
?
t
t
e
d
v
a
l
u
e
s
)
,
f
o
r
g
e
n
e
r
a
l
s
p
e
c
i
?
c
a
t
i
o
n
(
o
m
i
t
t
e
d
v
a
r
i
a
b
l
e
s
,
f
u
n
c
t
i
o
n
a
l
f
o
r
m
,
a
n
d
c
o
r
r
e
l
a
t
i
o
n
b
e
t
w
e
e
n
i
n
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
s
a
n
d
r
e
s
i
d
u
a
l
s
)
;
c
m
a
r
g
i
n
a
l
s
i
g
n
i
?
c
a
n
c
e
l
e
v
e
l
o
f
F
-
s
t
a
t
i
s
t
i
c
o
n
a
l
l
c
o
e
f
?
c
i
e
n
t
s
(
a
p
a
r
t
f
r
o
m
t
h
e
c
o
n
s
t
a
n
t
t
e
r
m
)
e
q
u
a
l
t
o
z
e
r
o
Table IV.
OLS regressions of pro?t
rates on R&D, equity
market capitalization,
and incremental returns
on recognized assets,
1989-2004
ARJ
24,2
162
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and market-to-book are computed for annual and industry/multi-annual partition-based
sub-samples. Second, market-to-book is modelled in relation to the two R&Ddimensions
across each partition. The annual partition follows Hand (2001) whereas the
industry/multi-annual partition is an extension adopted in this paper to both enlarge
the sample for, and introduce industry variation to the market-to-book modelling.
5.2 Annual and industry/multi-annual pro?t rate of R&D, 1992-2004
Table V reports NDP estimates for R&D and market capitalisation and the estimate of
NIRs for recognised assets using a modi?ed form of equation (2) where the dummy
variables are dropped. Both annual and industry/multi-annual partition-based
sub-sample estimates are obtained, using four-year lags, with beginning year 1992
due to lack of pre-1988 data. Considering ?rst the annual sub-samples, the adjusted R
2
values (not tabulated) are equal to or exceed 0.92 for all years and the F-statistics on the
coef?cient set apart from constant term equal to zero (not tabulated) have p-values less
than 0.001 for all models. Some caveats do arise from the diagnostics, with eight
signi?cant values out of thirteen for the Jarque-Bera statistic and ten signi?cant
RESET statistics from the 13[23] yearly estimations.
R&D’s NDP estimates are signi?cant in four years, but negative in three of these
cases, with no consistent pattern over time. Estimates of market capitalisation’s NDP
are signi?cant and negative in all years. For recognised assets’ NIR, estimates are
signi?cant and positive in eight years and otherwise not signi?cant. Lack of evidence
for a trend in Australian ?rms’ R&D pro?t rate is in marked contrast to Hand (2001)
who ?nds positive annual rates of pro?t for US R&D, using three-year lag lengths,
in 16 of the years 1980 through 2000, and increasing from the 1980s to the 1990s.
Considering, second, the industry/multi-annual sub-samples, adjusted R
2
-values (not
tabulated) are equal to or exceed 0.90 for all years apart fromthe Foodindustry 1992-1999
where the value is 0.59. F-statistics on all coef?cients apart fromthe constant termequal
to zero (not tabulated) have p-values less than 0.01 for all models. The Jarque-Bera
statistic is signi?cant in seven out of 17 cases and the RESET statistic is signi?cant in
11 of the 17. The RD coef?cients are signi?cant in seven sub-samples, with four positive
values and three negative. No noticeable pattern is apparent in the distribution of
the pro?t rate of R&D, although the two signi?cant mincon industry values are positive.
The R&D’s on-average pro?t rate is unstable across the industry/multi-annual partition,
reinforcing the evidence for lack of consistent pro?tability.
5.3 Market-to-book in relation to R&D’s pro?t rate and intensity
Table VI displays mean market-to-book and revenue intensities of R&D for the annual
and industry/multi-period sub-samples consistent with the basis for the above pro?t rate
estimates. Market-to-book displays a somewhat similar pattern of annual ?uctuation to
that of the entire sample as reported in Table III, however with the peak at year 2004.
Annual mean R&D revenue intensity increased from 0.07 per cent in 1992 to
approximately 29.0 per cent in 2004. A strong overall upward trend is apparent and the
?uctuationacross the periodis fairlyconsistent withthat for the entire sample as reported
in Table III. R&D intensity was at a minimum in 1992 and at a maximum in 2003.
There are no stand-out industry sector differences in market-to-book apart
from the high values for Pharmhealth, this sector having the three greatest values
in the sample partition. Clearly, vast variation occurs in R&D intensity across sectors,
R&D
pro?tability
163
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
GOI
i;t
¼ a þ
P
K
j¼0
b
j
RD
i;t2j
þ
P
K
j¼0
g
j
MVEq
i;t2j
þuTA
i;t
þ1
i;j
Year
P
4
j¼0
^
b
j
R
2j
21
P
4
j¼0
g
j
R
2j
21
^
u 2medðDA
i;t
=TA
i;t
Þ J-B, RESET sgnf.
a
n
Panel A: Annual partition
1992 21.43 21.23 0.23 –, – 17
(0.305) (,0.001) (,0.001)
1993 0.68 20.92 0.03 –,
* *
19
(0.645) (,0.001) (0.286)
1994 6.60 21.07 0.12 –,
* *
23
(0.196) (,0.001) (0.002)
1995 26.78 21.01 0.13 –,
* *
27
(0.176) (,0.001) (0.001)
1996 210.72 20.92 0.10 –,
* *
31
(,0.001) (,0.001) (,0.001)
1997 24.22 20.94 0.09
*
,
* *
42
(0.063) (,0.001) (,0.001)
1998 3.30 21.07 0.11
*
,
* *
46
(0.142) (,0.001) (,0.001)
1999 2.33 21.13 0.18
*
,
* *
50
(0.042) (,0.001) (,0.001)
2000 20.75 20.96 0.07
*
,
* *
60
(0.398) (,0.001) (,0.001)
2001 24.66 20.93 0.04
*
,
* *
57
(0.046) (,0.001) (0.075)
2002 21.48 20.90 0.02
*
,
* *
59
(0.325) (,0.001) (0.573)
2003 21.74 20.93 0.02
*
, – 61
(0.030) (,0.001) (0.306)
2004 22.21 20.96 0.05
*
, – 71
(0.200) (,0.001) (0.062)
Total 563
Industry/
period
P
4
j¼0
^
b
j
R
2j
21
P
4
j¼0
g
j
R
2j
21
^
u 2medðDA
i;t
=TA
i;t
Þ
Jarque-Bera, RESET
sgnf. n
Panel B: Industry/multi-annual partition
Mincon 8.13 21.03 0.10 31
1992-1995 (0.008) (,0.001) (0.053) –,
* *
Mincon 21.88 21.03 0.13 30
1996-1998 (0.487) (,0.001) (,0.001) –,
* *
Mincon 212.17 21.00 0.06 39
1999-2001 (0.132) (,0.001) (0.106) –, –
Mincon 10.78 20.95 0.06 28
2002-2004 (0.021) (,0.001) (0.022) –, –
Manuf 2.26 21.07 0.10 30
1992-1995 (0.007) (,0.001) (0.006) –, –
Manuf 0.61 20.97 0.04 27
1996-1997 (0.775) (,0.001) (0.519)
*
,
* *
Manuf 23.19 21.09 0.17 32
1998-1999 (0.008) (,0.001) (0.211) –, –
Manuf 28.85 20.70 20.05 38
2000-2001 (,0.001) (,0.001) (0.473)
*
,
* *
Manuf 21.42 21.09 0.12 58
2002-2004 (0.719) (,0.001) (0.130)
*
,
* *
(continued)
Table V.
Sub-sample OLS
regressions of NDP rates
on R&D, market
capitalization, and NIRs
on recognized assets
(lag length ¼ 4 years)
ARJ
24,2
164
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Chem 1.15 20.90 20.00 45
1992-2004 (0.581) (,0.001) (0.870) –,
* *
ICT 21.47 21.03 0.06 25
1992-2004 (0.389) (,0.001) (0.590)
*
, –
Pharmhealth 1.40 21.05 0.08 28
1992-1999 (0.097) (,0.001) (0.005)
*
,
* *
Pharmhealth 1.08 21.00 0.06 35
2000-2002 (0.116) (,0.001) (0.005)
*
,
* *
Pharmhealth 23.61 20.87 20.02 33
2003-2004 (,0.001) (,0.001) (0.617) –, –
Food 16.68 21.06 20.00 25
1992-1999 (0.003) (,0.001) (0.990)
*
,
* *
Food 4.40 20.96 0.03 31
2000-2004 (0.192) (,0.001) (0.130) –,
* *
TURS 24.30 20.97 0.04 28
1992-2004 (0.146) (,0.001) (0.018) –,
* *
Total 563
Notes: GOI
i,t
– gross operating income: pro?t before tax adjusted for R&D, depreciation and
amortization for ?rmi in year t; RD
i,t
– research and development expenses for ?rmi in year t; MVEq
i,t
–
market value of equity for ?rm i at the end of year t; TA
i,t
– total assets of ?rm i at the end of year t;
DA
i,t
– depreciation plus amortization expense for ?rm i in year t; and a and 1
i,t
are constant and error
terms, respectively; DA
i,t
– depreciation plus amortization expense for ?rm i in year t; TA
i,t
– total
assets of ?rmi at the end of year t; see Table I for industry de?nitions; the upper 1 per cent of the GOI, RD,
MVEq and TAobservations available for inclusion are Winsorised to attenuate the in?uence of outliers;
a
*
Jarque-Bera marginal signi?cance less than 0.05;
* *
RESET marginal signi?cance less than 0.05 Table V.
Year (MVEq/BVEq)
t
(RD/Rev)
t
n Industry/Period (MVEQ/BVEq)
r
(RD/Rev)
r
n
1992 1.854 0.007 17 Mincon 1992-1995 1.723 0.009 31
1993 1.883 0.007 19 Mincon 1996-1998 1.681 0.008 30
1994 1.886 0.007 23 Mincon 1999-2001 1.574 0.006 39
1995 1.676 0.008 27 Mincon 2002-2004 2.153 0.005 28
1996 1.845 0.040 31 Manuf 1992-1995 1.660 0.007 30
1997 2.055 0.141 42 Manuf 1996-1997 1.799 0.043 27
1998 2.119 0.118 46 Manuf 1998-1999 1.815 0.026 32
1999 2.632 0.154 50 Manuf 2000-2001 1.708 0.073 38
2000 2.472 0.204 60 Manuf 2002-2004 1.808 0.071 58
2001 2.754 0.152 57 Chem 1992-2004 1.514 0.010 45
2002 2.333 0.287 59 ICT 1992-2004 2.612 0.115 25
2003 2.814 0.311 61 Pharmhealth 1992-1999 3.559 0.530 28
2004 2.921 0.290 71 Pharmhealth 2000-2002 5.778 0.910 35
Pharmhealth 2003-2004 5.065 0.947 33
Food 1992-1999 2.991 0.004 25
Food 2000-2004 1.284 0.004 31
TURS 1992-2004 2.914 0.244 28
Totals 563 563
Notes: (MVEq/BVEq)
n
– mean end-date market to book value of ?rms’ equity across sub-sample n;
(RD/Rev)
n
– mean R&D revenue intensity of ?rms across n; t represents annual partitioning and r
represents industry/multi-annual partitioning; the entire sample’s upper 2 per cent of MVEq/BVEq
observations and upper 5 per cent of RD/Ren
Table VI.
Sub-samples means of
market-to-book and
revenue intensity
R&D
pro?tability
165
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
with extraordinarily high values for Pharmhealth, and, to a lesser extent, TURS. Mincon
and Food are sectors relatively weak in their intensity of R&D, with values of less than
1 per cent in all sub-samples. The leading values of Pharmhealth’s market-to-book and
its exceptional R&D intensity likely re?ect the sector’s reliance on intellectual capital
and its typical engagement with R&D in the ordinary course of business.
The annual and industry/multi-annual R&D pro?t rate estimates (Table V) and
means of market-to-book and R&D revenue intensity for similar sample partitions
(Table VI) allow evaluation of the association of R&D’s pro?t rate and intensity with
market-to-book. Correlations for the annual data, extended to include Australian equity
market total returns, and for the industry/multi-annual data, are displayed in Table VII.
Across neither partition is market-to-book correlated with R&D’s pro?t rate. However,
market-to-book is positively and highly correlated with R&D intensity across both sets
of sub-samples. These results indicate that in a bivariate sense R&D is associated with
market-to-book, although the only reliable relation is via intensity.
Market-to-book’s association with R&D is clari?ed by extending the modelling to a
multivariate form, following Hand (2001). Using the above annual and
industry/multi-annual R&D pro?t rate estimates and mean intensities, and
additionally, for the former partition, Australian equity market annual total returns,
estimates of the dependence of mean market-to-book on these factors are obtained via
equation (4):
MVEq
BVEq
v
¼ a þI
1
bNDP_RD
v
þI
2
g
RD
Rev
v
þI
3
dRI
v
þ1
v
ð4Þ
In equation (4), (MVEq/BVEq)
n
is the mean of ?rms’ market-to-book based on the
Table VI sub-sample (v), NDP_RD
v
is equal to
P
4
j¼0
^
b
j
R
2j
21 (the estimate of
NDP(RD) for v, as per Table V, Panels A and B as relevant), (RD/Rev)
v
is the mean of
?rms’ R&D revenue intensities across v, drawn fromTable VI, RI
v
is Australian annual
total return on equities for (year) v[24], a and 1
v
are constant and error terms,
respectively, and, for cases (i) and (iii), I
1
and I
2
are equal to unity, and I
3
is equal to zero
and for case (ii), I
1
, I
2
, and I
3
are equal to unity.
MVEq/BVEq NDP(RD) RD/Rev
Panel A: Annual partition – 13 observations
NDP(RD) 0.10
(0.743)
a
RD/Rev 0.86 0.01
(,0.001) (0.982)
RI 0.01 20.26 20.12
(0.976) (0.396) (0.702)
Panel B: Industry/multi-annual partition– 17 observations
NDP(RD) 0.05
(0.859)
RD/Rev 0.94 20.15
(,0.001) (0.558)
Notes: MVEq/BVEq – market-to-book of ?rms’ equity; NDP(RD) – net discounted pro?tability of
R&D; RD/Ren – revenue intensity of R&D; RI – Australian annual total equity returns;
a
numbers in
parentheses below estimates are relevant marginal signi?cance levels
Table VII.
Pearson correlations for
mean market-to-book,
estimated R&D pro?t
rate, mean R&D intensity
and annual total equity
returns
ARJ
24,2
166
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
With R&D’s pro?t rate arrayed in competition with R&D intensity only (Table VIII),
the estimate of b, the coef?cient of the pro?t rate is not signi?cant in the annual
partition (case (i)). However, the estimate of g, the R&D intensity coef?cient is highly
signi?cant and positive. With the set of competitors expanded to include equity annual
total returns (case (ii)) consistent with Hand (2001) and Ryan (1995), the R&D pro?t
rate coef?cient remains insigni?cant while the R&D intensity coef?cient remains
signi?cant and positive. However, d, the coef?cient of equity returns is insigni?cant.
These results are mostly consistent with Hand (2001), where intensity is in?uential and
reduces otherwise univariately signi?cant R&D pro?tability to insigni?cance in
explaining US annual market-to-book. However, they differ due to the positive impact
of US equity returns reported by Hand (2001)[25].
For the industry/multi-annual partition (case (iii)), with R&D’ s pro?t rate and
intensityagainincompetition, the estimate of bis insigni?cant, giventhe null hypothesis
of equality to zero, the alterative of non-equality and the 0.05 signi?cance level. On this
basis, and on the insigni?cant result for R&D’s rate of pro?tability from the complete
sample pooled time-series and cross-section data presented above, it appears safe to rule
out R&D’s pro?t rate as on-average in?uential for market-to-book. By extension,
on-average in?uence on market-to-book of the compounded pro?t rate and intensity of
R&D, represented by the second channel in Figure 1, can similarly be ruled out.
Equation (4), cases (i) and (ii), does indicate market-to-book as increasing in R&D
intensity[26]. This potentially leads to an inference that typically, increasing
expenditure on R&D negatively biased BVEq, distorting the economic information
conveyed by market-to-book.
Two considerations count against the evidence from equation (4) for R&D intensity’s
impact on market-to-book. First and most critically, market-to-book and revenue
Case
^
b ^ g
^
d Adj. R
2
Coeff. F-stat. p n
ðMVEq=BVEqÞ
v
¼ a þI
1
bNDP_RD
v
þI
2
gðRD=RevÞ
v
þI
3
dRI
v
þ1
v
Cases (i) and (iii): I
1
¼ I
2
¼ 1, I
3
¼ 0; Case (ii): I
1
¼ I
2
¼ I
3
¼ 1.
Panel A: Annual partition
J-B p RESET p
(i) 0.01 3.18 – 0.69 0.001
a
13
(0.276) (,0.001) 0.782
b
0.228
c
(ii) 0.01 3.25 0.47 0.68 0.004 13
(0.105) (,0.001) (0.324) 0.428 0.555
Panel B: Industry/multi-annual partition
J-B p RESET p
(iii) 0.04 4.00 0.92 ,0.001 17
(0.038) (,0.001) 0.755 0.933
Notes: (MVEq/BVEq)
n
– mean period-end-date market-to-book value of ?rms’ equity across sub-
sample n; NDP_RD
n
– net discounted pro?tability of R&D for n; (RD/Rev)
n
– mean revenue intensity
of ?rms across n; RI
n
– Australian annual total return on equities for n; and a and 1
n
are constant and
error terms, respectively; estimation uses Newey-West HAC consistent covariances; numbers in
parentheses are marginal signi?cance levels of t-statistics on the relevant coef?cient being equal to
zero;
a
marginal signi?cance level of F-statistic on all coef?cients (apart from the constant term) equal
to zero;
b
marginal signi?cance level of Jarque-Bera statistic for normality of residuals;
c
marginal
signi?cance level of RESET statistic for general speci?cation
Table VIII.
OLS regressions of
market-to-book ratio on
annual and
industry/multi-annual
R&D pro?tability and
intensity, and total equity
returns, 1992-2004
R&D
pro?tability
167
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
intensity may be non-stationary, follow an upward trend, yet be independent (cases (i)
and (ii)). If this is so, the estimates are outcomes of spurious regressions[27]. Second, the
relevant caveat mentioned in Section 2.4 applies – time series of cross-section
aggregations (period means) of ?rms’ data may not well re?ect the underlying biasing
effect of R&D on equity book value at the individual ?rm level.
Finally, equation (4) does not separate the ?nancial statement handling (the fourth
channel in Figure 1) and equity market-value (the third channel in Figure 1) effects of
R&D intensity. It is therefore pertinent to investigate R&D intensity’s in?uence on
market-to-book and identifying the channel(s) of any such in?uence. Accordingly, we
take this study well beyond the approach taken by Hand (2001).
5.4 Market-to-book and R&D intensity
In dealing with the channels of R&D intensity’s in?uence on market-to-book this paper
adopts a suggestion of Hand (2001) and distinguishes a book value channel and a
market value channel. The former channel, the fourth in Figure 1, conveys R&D
intensity’s level, re?ecting equity book value’s typically negative relation with R&D
argued above (Section 2.4). The latter channel, the third in Figure 1, conveys the
current change in R&D intensity.
An increase, say, in R&D intensity is conjectured to be a signal of management’s
private information of an increase R&D’s pro?t rate, typically a positive indicator for
?rms’ equity market values. Translating to a hypothesis of market-to-book being, on
average, increasing in R&D intensity’s ?rst difference, this effect operates at the
individual ?rm, time-series level.
The approach adopted to analyse the impact of R&D intensity is to model
market-to-book at the ?rm-year level. R&D intensity’s level and ?rst difference are
arrayed against variables controlling for likely general in?uences on market-to-book.
Given the paucity of theory in this area, two control variables are adopted on a
heuristic basis. The ?rst control is the one-year lagged value of the dependent variable,
re?ecting an autoregressive, endogenous factor in market-to-book’s evolution[28]. Data
on changed pro?tability is the second control, in the form of the ?rst difference of
current PBT (excluding R&D) relative to equity book value. This formulation is
re?ected in equation (5) where (MVEq/BVEq)
i,t
is market-to-book for ?rm i in year t,
(PBT_RD/BVEq)
i,t
is, for ?rm i in year t, PBT excluding R&D expense on BVEq,
(RD/Rev)
i,t
is R&D revenue intensity for ?rm i in year t, and a and 1
i,t
are constant and
error terms, respectively:
MVEq
BVEq
i;t
¼ a þb
MVEq
BVEq
i;t21
þg
PBT_RD
BVEq
i;t
2
PBT_RD
BVEq
i;t21
þd
RD
Rev
i;t
þu
RD
Rev
i;t
2
RD
Rev
i;t21
þ1
i;t
ð5Þ
Estimation is via panel least squares with cross-section ?xed effects and White (1980)
cross-section coef?cient covariances. The panel-data approach is adopted to handle the
proposed individual ?rm, time series relation of market-to-book with R&D intensity’s
level and ?rst difference[29]. The dataset is based on that used to estimate equation (2)
but augmented by including, as available, up to two leading, one trailing and three
intermediate zero-R&D observations in each ?rm’s time series, so as to adequately
capture the effects of variation in R&D intensity. Results are displayed in Table IX.
ARJ
24,2
168
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The redundant ?xed effects F-statistic’s approximately-zero marginal signi?cance
level leads to rejection of the test’s null hypothesis of no ?xed effects. Asimilar value for
the marginal signi?cance of the Hausman (1978) x
2
-statistic leads to rejection of that
test’s null of a randomeffects speci?cation. Hence the panel data (?xed effects) approach
is con?rmed.
Lagged market-to-book in?uences current market-to-book, with a positive-value
estimate for b. None of the coef?cient estimates for the in?uence of the current ?rst
difference of R&D-adjustedpro?tabilitybefore taxper dollar bookvalue, g, R&Dintensity
level, d, and R&D intensity ?rst difference, u, are signi?cantly different from zero[30].
The lack of in?uence of R&D intensity’s level on market-to-book overturns the
relevant indications arising from estimation of equation (4). This result reinforces the
caveat discussed in Section 5.3 – that spurious regression likely applies to cases (i) and
(ii). The null ?nding suggests that despite trend-increasing mean R&D intensity over
1988-2004, ?rms’ equity book values were not typically substantially biased by the
expensing of R&Dexpenditures, and thereby market-to-book’s economic information is
not signi?cantly degraded. However, characteristics of the data generating process and
limitations of the dataset may disguise the real impact of R&D’s disclosure:
P1. R&D and other expenses may have a substitution association, due to either or
both R&D attractiveness in periods when other expenses are less onerous, or
misclassi?cation of R&D vis-a´-vis other expenses.
ðMVEq=BVEqÞ
i;t
¼ a þbðMVEq=BVEqÞ
i;t21
þgððPBT_RD=BVEqÞ
i;t
2ðPBT_RD=BVEqÞ
i;t21
Þ þ
dðRD=RenÞ
i;t
þuððRD=RenÞ
i;t
2ðRD=RenÞ
i;t21
Þ þ1
i;t
^ a
^
b
^ g
^
d
^
u n
2.25 0.20 20.41 20.04 20.13 1,677
(,0.001) (,0.001) (0.026) (0.902) (0.691)
Adj. R
2
Coeff. F-stat. p Jarque-Bera p Redundant ?xed effects F-stat. p Hausman
x
2
stat. p # cross-section units
0.56 ,0.001
a
,0.001
b
,0.001
c
,0.001
d
340
Notes: (MVEq/BVEq)
i,t
– market-to-book value of equity for ?rm i at end of period t; (PBT_RD/
BVEq)
i,t
– pro?t before tax adjusted for R&D expenses of ?rm i for year t to book value of equity for
?rm i at end of period t; (RD/Ren)
i,t
– R&D revenue intensity for ?rm i in year t; and a and 1
i,t
are
constant and error terms, respectively; estimation is via panel least squares with cross-section ?xed
effects; the data set is based on that used to estimate equation (2) but augmented by including, as
available, up to two leading, one trailing and three intermediate zero-R&D observations in each ?rm’s
time series, to adequately capture the effects of variation in R&D intensity; the upper 2 per cent of the
MVEq/BVEq observations, the lower and upper 2 per cent of the PBT_RD/BVEq observations and the
upper 5 per cent of the RD/Rev observations available for inclusion are Winsorised to attenuate excess
in?uence of outliers; estimation uses White (1980) cross-section coef?cient covariances; numbers in
parentheses are marginal signi?cance levels of t-statistics on the relevant coef?cient estimate being
equal to zero;
a
the RESET statistic is not applicable to panel estimation;
b
marginal signi?cance level of
F-statistic on all coef?cients (apart from constant) equal to zero;
c
marginal signi?cance level of Jarque-
Bera statistic for normality of residuals;
d
marginal signi?cance level of F-statistic for redundant ?xed
effects;
i
marginal signi?cance level of Hausman x
2
statistic for a random effects speci?cation
Table IX.
Panel regression of
market-to-book on
one-year lagged value,
the current ?rst
difference of PBT
(excluding R&D) relative
to book value of equity,
the current level of R&D
intensity, and the current
?rst difference of R&D
intensity, 1989-2004
a,b,c,d,e
R&D
pro?tability
169
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
P2. R&D may exhibit a complementation relation with overall net-of-R&D
pro?tability.
P1 and P2 imply strong positive correlation of RD/Rev with the ?rst difference of
PBT_RD/BVEq, leading to substantial imprecision of estimates for g and d. To evaluate
P1 and P2, RD/Rev’s dependence on PBT_RD/BVEq’s ?rst difference is estimated using
a panel-data approach, although in this case a cross-section randomeffects speci?cation
is statistically appropriate. The (untabulated) results, however, indicate the coef?cient
for the latter variable is insigni?cantly different from zero. P1 and P2 can therefore be
discounted. However, the typical time series of ?rms’ reported engagement in R&D
likely plays a role in the null result for R&D intensity level’s impact on market-to-book.
Firm engagement with R&D is mainly of short duration, evidenced by the
1,677 ?rm-years (including zero-R&D “wrap-around” and intermediate years) and
340 cross-section units (?rms), giving a mean ?rm time series length of 4.9 years. Quite
limited scope thereby exists for intra-?rm variation accounting for the rising average
R&D intensity and therefore the biasing of equity book values over 1989-2004. Rather,
the generally positive trend in mean R&D intensity over 1988-2004 apparently re?ects
?rms in the cross-section reporting greater intensities in their time series in later years.
Hence the vital role, as alluded to above, for zero-R&D observations to properly re?ect
the impact of increases (or decreases) in R&D. However, the data in this latter respect
display considerable limitations. In fact, of the 340 ?rms in the cross-section, 105 have
two leading zero-R&D observations available (guaranteeing one such observation for
inclusion in the estimation, due to the lagged data requirement), 97 have a trailing such
observation and 28 have intermediate-series-located similar observations. It is
conceivable that the shortage of zero-R&D observations may have a substantial impact
on the estimation of equation (5), especially of d, the R&D intensity-level coef?cient.
Finally, the non-in?uential role for market-to-book of R&D intensity-difference
suggests that changes in R&D intensity are not typically price-relevant signals for
capital market participants regarding managers’ changing views on R&D’s
pro?tability. This ?nding likely re?ects the above-reported lack of on-average
pro?tability of R&D for Australian listed ?rms.
6. Summary and implications
Market-to-book is a joint capital market and ?nancial reporting-determined indicator of
?rms’ ?nancial standing. Hand (2001) reported that for US ?rms over the period
1980-2000, the increasing intensity of R&D dominated the in?uence R&D pro?tability
growth when considering the overall increase and ?uctuation of market-to-book. The
intensity impact was attributed to R&D expenditures’ disclosure as an expense
imparting negative bias to equity book values on increasing R&D. It was concluded that
the economic information conveyed by market-to-book was compromised as a result.
In the A-GAAP era over 1988-2004, Australian R&D ?rms generally experienced an
uptrend and ultimately very substantial increase in R&D revenue intensity. Over the
same period, the mean market-to-book of such ?rms ?uctuated signi?cantly and,
overall, more than doubled. This paper estimated Australian listed ?rms’ R&D rate of
pro?tability, measured their R&D revenue intensity and estimated the impact of these
R&D dimensions on market-to-book via their channels of in?uence. This allowed
evaluation of the ?nancial disclosure vis-a´-vis economic reality of R&D.
ARJ
24,2
170
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
For the 1988-2004 ?rm-year data used in this study, it was typically inferred that
R&D had a zero on-average rate of pro?tability, whether estimated on pooled
time-series/cross-section, annual or industry/multi-annual bases. This ?nding revealed
the economics of most Australian ?rms’ R&D engagement as anomalous in the
international context. These ?rms’ experience might re?ect a lack of economies of scale
in R&D being achievable within the Australian industrial structure, in contrast
to the USA (Hand, 2003). Additionally, Australian ?rms may typically, in an
international context, be R&D “followers” rather than “leaders” (Lev et al., 2005), with
R&D intensity less than that of their industry and inferior sustained future (general)
pro?tability relative to industry leaders.
Modelling of market-to-book as a function of R&D’s rate of pro?tability and
intensity on annual and industry/multiannual bases did not yield convincing evidence
for either dimension’s in?uence, whether via market or BVEq-type channels. The paper
considered two channels in which market-to-book may be associated with R&D
intensity by extending modelling to the ?rm-year level. The ?rst, key R&D intensity
channel re?ects typical negative bias imparted to book value by an increased level of
R&D (or vice versa), due to R&D expenditures’ handling via write-off. This effect leads
to the hypothesis of on-average positive individual-?rm, time series relation of
market-to-book with R&D intensity’s level.
The second channel of R&D intensity re?ects the conjecture of, say, a positive effect
on the market-to-book numerator due to an increase in R&D intensity being interpreted
by the market as a signal of more favourable management perceptions of R&D’s pro?t
rate. Multivariate testing for the two R&D intensity channels using ?rm-year data
failed to con?rm either the expected intensity-level impact or an intensity-difference
in?uence on market-to-book.
What can be concluded concerning the market-to-book impact of A-GAAP’s
apparently restrictive rule forcing expensing except for projects with near-certain
recoupment of R&D expenditure? First, the information conveyed by market-to-book
was almost certainly consistent with the implicit intent of the (for R&D expenditures at
least) near hard assets, ef?cient contracting-oriented in-place regime.
Second, on the evidence of lack of (positive) relation of market-to-book to R&D
intensity, R&D apparently did not typically impart signi?cant negative bias to ?rms’
equity book values despite the apparently mostly positive trend in ?rms’ typical R&D
engagement over the period under study. This benign outcome probably re?ects the real
impact of cross-?rmrather than intra-?rmincreasing R&Dintensity over 1988-2004 but
likely also the limitations of the available data. With reference to broad assets, equity
valuation orientation as the reference disclosure regime, the result at face value implies
on-average immaterial adverse effects for market-to-book’s information on, say, ?rms’
growth opportunities. In consequence, A-GAAP apparently served a “best of both
worlds” role.
Finally, turning attention to the AIFRS regime, the less restrictive rule
on capitalisation of R&D expenditures ideally results in equity book values with less
systematic bias than, say, values likely under continuing A-GAAP with on-average
R&D pro?tability[31]. Market-to-book in such case typically conveys economic
information more consistent with the new regime’s orientation. However, Kothari et al.
(2009) argue that under IFRS the handling of R&D expenditure is characterised by
excess managerial discretion on asset valuations. If so, ?rms’ equity book values may
R&D
pro?tability
171
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
frequently possess signi?cant, idiosyncratic, managerial policy-related bias under, say,
increasing R&D expenditures. Consequently, for such ?rms, the market-to-book
information for growth prospects will be compromised relative to the equity valuation
objective implicit in AIFRS.
The asset recognition rule for R&D expenditures under AIFRS applies widely to
internally generated intangibles. Equity book values’ bias and related impacts
on market-to-book therebyconstitutes an issue extending beyond the scope of A-GAAP’s
traditional R&D expenditures, strengthening its appeal for future research attention.
Notes
1. Hereafter in the text, except as otherwise indicated explicitly or via the context, “R&D” is to
be taken as “research and development expenditures handled as expenses in the ?nancial
statements”.
2. See White et al. (2003, Ch. 19) for basic concepts in this area. The suggested market-book value
difference decomposition best aligns with the transaction-theoretic approach where the values
are expectations of future transaction costs (Hodgson et al., 1993). An alternative, direct
economic value-theoretic decomposition is suggested by Upton (2003) based on “market
assessments” and includes a component re?ecting unrecognized identi?able intangibles.
3. Whether market-to-book can serve as a basis for measurement of intellectual capital is
problematical as, logically, the “values” of assets such as R&D cannot be estimated via
market prices (Lev and Sougiannis, 1996, pp. 110-11).
4. Even if R&Dexpenditures are, on average, pro?table, they are often highly risky. Kothari et al.
(2002) ?nd that the variability of future earnings is greater for R&Doutlays than for property,
plant and equipment expenditures while Bens et al. (2004) report that R&D intensity is
positively associated with the volatility of forward earnings realizations. However, Amir et al.
(2007) ?nd that R&Dcontributes more to subsequent earnings variability than physical asset
capital expenditure only in relatively R&D-intensive industries. Dif?culty in predicting future
bene?ts for particular expenditures may arise partly from the inherent uncertainty of R&D
program technical outcomes and partly from the nature of markets where even technically
successful R&D ventures may not prove pro?table (Stickels, 1996).
5. In the USA, the debate among researchers has focused on the almost universally required
full-expensing of R&Dunder Statement of Financial Accounting Standards (SFAS) No. 2(FASB,
1974). This rule, along with the argued poor quality and quantity of R&D disclosure is widely
representedinthe USliterature as re?ectingmisguidedpolicy, creatinginformationasymmetries
with adverse cost-of-capital and insider trading implications (Aboody and Lev, 2000; Lev, 2001).
6. This typical effect of R&D on BVEq can be analysed most easily in the base case of a “BVEq
steady state” ?ow of R&D projects and a superimposed positive or negative R&D shock.
7. Relevant parameters of R&D-related BVEq are likely the relative levels of R&D and
associated revenues for the pre-existing process and for the increment or decrement, and,
additionally, regarding the latter, the time phasing of the revenues.
8. The impact of R&D on BVEq’s evolution is left to intuition, although it could be treated by
numerical illustration, simulation or (formal) analysis.
9. Kothari et al. (2009) note that internally-generated intangibles such as R&D typically have
minimal value in liquidation and highly uncertain cash ?ow realizations.
10. Average market-to-book approximately doubled over 1981-1999 and increased by about
50 per cent over 1980-2000 in Hand’s (2001) US ?rms. For 1980 through 2000 mean revenue
intensity of R&D spending for sample ?rms increased sevenfold.
ARJ
24,2
172
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
11. Hand (2003), extending Hand (2001) ?nds that average, net pro?tability of R&D increased
with the scale of expenditure and that this increasing returns-to-scale effect itself increased
over the 1980-2000 sample period. The inference is made of a competition imperative with
the largest R&D investments creating cost-based and/or network-derived barriers to entry
and resulting super-normal rents. This returns-to-scale ?nding may be relevant to
Australia’s R&D pro?tability experience, as referred to in Section 6 below.
12. In drawing such an inference Hand (2001) attributes the market-to-book impact of increased
R&D intensity entirely to the negative biasing of equity book values arising with increasing
R&D expensing. The analysis, however, ignores the channel of in?uence ?owing from R&D
intensity to market value of equity, given an above-zero rate of pro?tability, via anticipated
increased discounted net cash ?ows (second channel in Figure 1).
13. Equation (2) abstracts from two potentially important explanatory variables representing
possible intangibles – advertising, and general and administrative expenses. The former
category likely re?ects part of brand intangibles and the latter further brand plus personnel
intangibles (Lev and Sougiannis, 1996; Hand, 2001; Ding et al., 2007). Unfortunately, separate
data were not available for these items, creating a potential omitted variables and resultant
bias problem. Inclusion of MVEq as a proxy for non-R&D unrecognized assets is intended to
ameliorate this issue.
14. This assumes that the lag j coef?cient from GOI in year t represents adequately the response
of GOI in year t þ j to R&D in year t (Hand, 2001, p. 9). Strictly, the estimated effect of R&D
in year t should be the sum of the lag 0 coef?cient from year t, the lag 1 coef?cient from year
t þ 1, the lag 2 coef?cient from(for) year t þ 2 and so on, as per Sougiannis (1994, p. 59).
15. GOI is calculated before adjustments re?ecting the costs of debt and equity capital. Auniform
rate glosses over variation in the risk levels of recognized and unrecognized assets and
amongst the latter, differences between R&D and the non-R&D intangible assets represented
by equity market capitalization. The discount rate adopted is computed in three stages. First,
by calculating the average of quarterly Australian corporate bond yields (middle rate) over Q1
1988 through Q4 2004, obtained from Datastream (AUSCRPB). Second, calculating the
average of Australian equity market annual total returns over 1988-2004, derived from the
relevant Datastream index (TOTMKAU(RI)). Third, the weight for equity ?nancing was
approximated by calculating the average BVEq/TA of the sample, with the debt proportion
set to the complement. Thanks to Abul Shamsuddin, University of Newcastle, Australia for
advice on the nominal weighted average annual cost of capital derivation.
16. NIR(TA) is not equivalent to but rather is a component of the NDP of recognized assets. Not
estimating the NDP of recognized assets will only bias the coef?cients in equation (2) if the
difference between NDP(TA) and NIP(TA) is correlated with the explanatory variables
(Hand, 2001).
17. The ten industry sectors used were consolidated fromthe company analysis classi?cation. The
consolidation re?ected the need for suf?cient observations to support the industry/multi-annual
analysis of market-to-bookinterms of R&Dpro?tabilityversus R&Drevenue intensityreported
in Table VIII. The consolidation is as follows: “Mincon” from“Mining”, “Oil and gas”, “Forestry
and paper” and “Construction and building materials”; “Manuf” from“Aerospace and defence”,
“Automobiles and parts”,“Diversi?ed industrial”, “Engineering and machinery”, “Electronic
and electrical”, “Household goods and textiles”, “Steel and other metals” and “Packaging”;
“Utilities” from “Water”, “Electicity” and “Gas distribution”; “ICT” from “Software and
computer”, “Telecommunications” and “IThardware”; “Pharmhealth” from“Pharmaceuticals”
and “Health”; “Food” from “Food producers and processors” and “Beverages”; “Retail” from
“Food and drug retailers”, “General retail” and “Personal care and household”; “Services” from
“Distributors”, “Leisure andentertainment andhotels”, “Media andphotography”, “Real estate”,
R&D
pro?tability
173
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
“Support services”, “Investment”, “Specialty and other ?nance” and “Unknown”; other sectors
are as per the database classi?cation.
18. Positive skew is re?ected by the mean exceeding the median and vice versa for the negative
case. See, for instance, Lewis and Orav (1989, p. 153).
19. The Winsorisation involves the upper 1 per cent of the observations on GOI, RD, MVEq, and
TA for the analyses presented in Tables IV and V. For the observations underlying
the summaries presented in Tables III and VI and the analyses presented in Tables VII-IX
the Winsorisation involves, as relevant, the upper 2 per cent of MVEq/BVEq, the lower and
upper 2 per cent of PBT_RD/BVEq, and the upper 5 per cent of RD/Ren.
20. In all three lag-length cases, the error term is heteroscedastic, as indicated by the marginal
signi?cance values for the White (1980) test statistic (not tabulated) being less than 0.05.
21. First, for each lag length, the error term is not normally distributed, as indicated by the
marginal signi?cance levels of the Jarque-Bera test statistic (Bera and Jarque, 1981). The
principal feature underlying non-normality of the residuals is leptokurtosis.
Transformations (such as taking logarithms) did not correct the problem and would
obscure the direct, per dollar pro?tability interpretation of the coef?cients. However, as the
samples are relatively large, violation of normality is likely to be of minor consequence as the
relevant test statistics asymptotically follow the appropriate distributions, despite error
non-normality (see, for instance, Brooks, 2002, p. 182). Second, for each lag length, the
marginal signi?cance values of the RESET statistic (Ramsey, 1969) suggest some
shortcoming(s) in the speci?cation. The RESET is however purely diagnostic, providing no
guidance as to improving the speci?cation (Brooks, 2002, pp. 194-7). A search for a superior
model for ?rm pro?tability is beyond the scope of this paper.
22. A similar framework for inference (whether gross or net discounted functions of or single
coef?cients) is applied throughout this paper, i.e. a zero value under the null, a two-sided
(non-zero value) alternative and a 0.05 signi?cance level.
23. Via the White statistic, the error term is heteroscedastic in 1999 through 2002 only. However,
again the Newey-West HAC covariances are computed for all sub-samples considered in this
section. The Wald test x
2
statistics are used to assess signi?cance of the estimates for such
sub-samples.
24. Australian annual total returns on equities are derived from the relevant index,
TOTMKAU(RI) (Datastream).
25. Reliability of the inferences fromestimation of equation (4) are supported by the insigni?cant
values of the diagnostic statistics for residual normality, speci?cation and homoscedasticity
(the latter are not tabulated).
26. The equation 4, case (iii) estimate of R&D intensity coef?cient g is highly signi?cant and
positive but is not relevant to the question of the biasing of equity book value due to the
mainly cross-section nature of the data. In this case, intensity serves as a control. The
estimate is likely biased due to omitted in?uential variables re?ecting industry-varying
non-R&D intellectual, organisational capital-generating identi?able intangibles or
synergistic, going concern premium-yielding effects.
27. See, for instance, Brooks (2002, Ch. 7). This consideration casts doubt on Hand’s (2001)
results on market-to-book as a function of dimensions of R&D given that the US annual
market-to-book, R&D pro?tability, and R&D revenue intensity series all have the
appearance of trending upward over time.
28. Use of a lagged dependent variable violates the classical linear regression model assumption
of exogeneity of the independent variables. However, the standard properties of least
ARJ
24,2
174
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
squares will on appropriate assumptions, apply asymptotically. See, for instance, Greene
(2008, pp. 72-75).
29. Fundamentally, the panel data with cross-section ?xed effects approach is adopted to handle
inter-?rm heterogeneity due to omitted, unobserved variables correlated with those included.
The approach defends against biased and inconsistent estimates otherwise arising in
estimating the purely time series-type conjectured dependence of ?rms’ market-to-book on
R&D intensity’s level and ?rst difference. See, for instance, Greene (2008, Ch. 9).
30. Again, in regard to the estimation of equation (5), the implication of error non-normality
indicated by the marginal signi?cance of the Jarque-Bera statistic is attenuated by the
sample size.
31. Key to the more permissive rule on capitalization of R&Dunder AIFRS is that for expenditure
in the development phase only, AASB 138 Intangible Assets requires and permits only
recognition as an asset where “it is probable that the expected future economic bene?ts that
are attributable to the asset will ?owto the entity . . . ” (paragraph 21 (a), emphasis added). See
for instance, Institute of Chartered Accountants in Australia (2011, pp. 884-915).
References
Aboody, D. and Lev, B. (2000), “Information asymmetry, R&D, and insider gains”, Journal of
Finance, Vol. 55 No. 6, pp. 2747-66.
Ahmed, K. and Falk, H. (2006), “The value relevance of management’s policy choice of research
and development expenditure reporting: evidence from Australia”, Journal of Accounting
and Public Policy, Vol. 25 No. 4, pp. 231-64.
Amir, E., Guan, Y. and Livne, G. (2007), “The association of R&D and capital expenditures with
subsequent earnings variability”, Journal of Business Finance and Accounting, Vol. 34
Nos 1/2, pp. 222-46.
Barth, M.E. and Kallapur, S. (1996), “The effects of cross-sectional scale differences on regression
results in empirical accounting research”, Contemporary Accounting Research, Vol. 13
No. 2, pp. 527-67.
Bens, D.A., Hanna, J.D. and Zhang, X.F. (2004), “Research and development, risk and stock
returns”, working paper, Chicago Graduate School of Business.
Bera, A. and Jarque, C.M. (1981), “An ef?cient large-sample test for normality of observations
and regression residuals”, ANU Working Papers in Economics 40, Canberra.
Bosworth, D. and Rogers, M. (1998), “Research and development, intangible assets and the
performance of large Australian companies”, Melbourne Institute Working Paper No. 2/98.
Brooks, C. (2002), Introductory Econometrics for Finance, Cambridge University Press,
Cambridge.
Canibano, L., Garcia-Ayuso, M. and Sanchez, M.P. (2000), “Accounting for intangibles: a literature
review”, Journal of Accounting Literature, Vol. 19, pp. 102-30.
Cohen, W. and Levinthal, D. (1989), “Innovation and learning: the two faces of R&D –
implications for the analysis of R&D investment”, Economic Journal, Vol. 99, pp. 419-23.
Ding, Y., Stolowy, H. and Tenenhaus, M. (2007), “R&D productivity: an exploratory international
study”, Review of Accounting and Finance, Vol. 6 No. 1, pp. 86-101.
Fama, E.F. and French, K.R. (1992), “The cross-section of expected stock returns”, Journal of
Finance, Vol. 47 No. 2, pp. 427-65.
Fama, E.F. and French, K.R. (1993), “Common risk factors in the returns on stocks and bonds”,
Journal of Financial Economics., Vol. 33 No. 1, pp. 3-56.
R&D
pro?tability
175
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
FASB (1974), Statement of Financial Accounting Standards No. 2: Accounting for Research and
Development Costs, Financial Accounting Standards Board, Stamford, CT.
Feltham, G.A. and Ohlson, J.A. (1995), “Valuation and clean surplus accounting for operating and
?nancial activities”, Contemporary Accounting Research, Vol. 11, pp. 698-732.
Feltham, G.A. and Ohlson, J.A. (1996), “Uncertainty resolution and the theory of depreciation
measurement”, Journal of Accounting Research, Vol. 34 No. 2, pp. 209-34.
Greene, W.H. (2008), Econometric Analysis, 6th ed., Pearson Prentice-Hall, Upper Saddle River, NJ.
Hand, J.R.M. (2001), “The economic versus accounting impacts of R&D on US market-to-book
ratios”, working paper, University of North Carolina at Chapel Hill, 25 September, SSRN:
285108.
Hand, J.R.M. (2003), “The increasingreturns to scale of intangibles”, inLev, B. andHand, J.R.M. (Eds),
Intangible Assets: Values, Measures, and Risks, Oxford University Press, Oxford, pp. 303-31.
Hausman, J.A. (1978), “Speci?cation tests in econometrics”, Econometrica, Vol. 46, pp. 1251-72.
Hirschey, M. and Weygandt, J. (1985), “Amortization policy for advertising and research and
development expenditures”, Journal of Accounting Research, Vol. 23 No. 1, pp. 326-35.
Hodgson, A., Okunev, J. and Willett, R. (1993), “Accounting for intangibles: a theoretical
perspective”, Accounting & Business Research, Vol. 23 No. 90, pp. 138-50.
Institute of Chartered Accountants in Australia (2011), Financial Reporting Handbook 2011,
Wiley, Milton.
Johnson, L.D. and Pazderka, B. (1993), “Firm value and investment in R&D”, Managerial and
Decision Economics, Vol. 14 No. 1, pp. 15-24.
Kallapur, S. and Trombley, M.A. (1999), “The association between investment opportunity set
proxies and realised growth”, Journal of Business Finance and Accounting, Vol. 26,
pp. 505-19.
Kothari, S.P., Laguerre, T.E. and Leone, A.J. (2002), “Capitalization versus expensing: evidence on
the uncertainty of future earnings from capital expenditures versus R&D outlays”,
Review of Accounting Studies, Vol. 7 No. 4, pp. 355-82.
Kothari, S.P., Ramanna, K. and Skinner, D.J. (2009), “What should GAAP look like? A survey and
economic analysis”, MIT Sloan/Harvard Business/Chicago Booth Working Paper,
17 September, SSRN: 1413775.
Lev, B. (2001), Intangibles: Management, Measurement, and Reporting, Brookings Institution
Press, Washington, D.C.
Lev, B. and Sougiannis, T. (1996), “The capitalization, amortization, and value-relevance of
R&D”, Journal of Accounting and Economics, Vol. 21, pp. 107-38.
Lev, B., Radhakrishnan, S. and Ciftci, M. (2005), “The stock market valuation of R&D leaders”,
Stern School of Business, New York University, working paper December.
Lewis, P.A.W. and Orav, E.J. (1989), Simulation Methodology for Statisticians, Operations
Analysts, and Engineers, Wadsworth, Paci?c Grove, CA.
Mairesse, J. and Mohen, P. (1995), Research and Development and Productivity, A Survey of the
Economic Literature, INSEE, Paris.
Mairesse, J. and Sassenou, M. (1991), “R&D and productivity: a survey of econometric studies at
the ?rm level”, Science, Technology and Industry Review, Vol. 7, pp. 131-47.
Newey, W.K. and West, K.D. (1987), “A simple positive-de?nite heteroskedasticity and
autocorrelation-consistent covariance matrix”, Econometrica, Vol. 55, pp. 703-8.
Ohlson, J.A. (1995), “Earnings, equity book value, and dividends in equity valuation”,
Contemporary Accounting Research, pp. 661-87.
ARJ
24,2
176
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Penman, S.H. (2004), Financial statement analysis and security valuation, 2nd ed.,
McGraw-Hill/Irwin, New York, NY.
Ramsey, J.B. (1969), “Tests for speci?cation errors in classical linear least squares regression
analysis”, Journal of the Royal Statistical Society B, Vol. 31 No. 2, pp. 350-71.
Ryan, S.G. (1995), “A model of accrual measurement with implications for the evolution of the
book-to-market ratio”, Journal of Accounting Research, Vol. 33 No. 1, pp. 95-112.
Sougiannis, T. (1994), “The accounting based valuation of corporate R&D”, The Accounting
Review, Vol. 69 No. 1, pp. 44-68.
Stickels, G. (1996), “Authorities aim for the objective on R&D”, Business Review Weekly,
18 March, pp. 99-100.
Tobin, J. (1969), “A general equilibrium approach to monetary theory”, Journal of Money, Credit,
and Banking, Vol. 1, pp. 15-27.
Upton, W.S. Jr (2003), “Challenges from the new economy for business and ?nancial reporting”,
FASB Special Report, reprinted in Lev, B. and Hand, J.R.M. (eds) Intangible Assets: Values,
Measures, and Risks, Oxford University Press, Oxford, pp. 469-86.
White, H. (1980), “A heteroskedasticity-consistent covariance matrix estimator and a direct test
for eteroskedasticity”, Econometrica, Vol. 48, pp. 817-38.
White, G.I., Sondhi, A.C. and Fried, D. (2003), The Analysis and Use of Financial Statements,
Wiley, Hoboken, NJ.
Zambon, S. and Associates (2003), “Study on the measurement of intangible assets and
associated reporting practices”, prepared for Commission of the European Communities
Enterprise Directorate General.
About the authors
Kamran Ahmed is a Professor in the School of Accounting at La Trobe University. He holds a
PhD in Accounting awarded by the Australian National University (ANU), has numerous
publications in internationally refereed journals in the accounting discipline and has previously
held academic appointments at Victoria University of Wellington, University of New England
and the ANU.
John Hillier is a Research Fellowin the School of Accounting at La Trobe University. He holds
a PhD in Accounting awarded by the University of Otago, has several publications in
internationally refereed journals in the ?elds of accounting and ?nance and has previously held
academic appointments at the University of New England, University of New South Wales and
the University of Southern Queensland. John Hillier is the corresponding author and can be
contacted at: [email protected]
Elisabeth Tanusasmita graduated as BCom (Hons) from La Trobe University in 2006 and
now works as a ?nancial analyst in the private sector.
R&D
pro?tability
177
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Alireza Vafaei, Kamran Ahmed, Paul Mather. 2015. Board Diversity and Financial Performance in the Top
500 Australian Firms. Australian Accounting Review 25:10.1111/auar.2015.25.issue-4, 413-427. [CrossRef]
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_321645276.pdf