International Review of Financial Analysis 32 (2014) 132–142

Contents lists available at ScienceDirect

International Review of Financial Analysis

Innovation efficiency and asymmetric timeliness of earnings: Evidence

from an emerging market
Ann L.-C. Chan
Department of Accounting, National Chengchi University, 64 Sec. 2 Zhi-Nan Road, Taipei 11605, Taiwan

a r t i c l e i n f o a b s t r a c t

Article history: This study investigates the impact of intellectual capital output on the asymmetric income timeliness of firms in
Received 17 June 2013 Taiwan, which is an emerging economy with a high concentration of research and development (R&D)-intensive
Received in revised form 26 January 2014 firms but poor investor protection. The higher growth opportunities and increased risk faced by R&D firms make
Accepted 29 January 2014
them more vulnerable to shareholder litigation. Specifically, I examine whether companies that are less efficient
Available online 7 February 2014
at transforming new ideas or techniques into granted patents are more timely in their recognition of losses and
JEL classification:
less timely in their recognition of gains in response to higher expected litigation costs. Indeed, I observe a
M41 negative relationship between innovation efficiency and asymmetric income timeliness and this is more evident
among highly R&D-intensive firms. In a parallel analysis, I also find that the observed greater earnings conservatism
Keywords: of low-innovation-efficient firms is more evident following the introduction of the Securities and Futures Investor
Innovation efficiency Protection Act in the year 2003, when a class action litigation mechanism was established. Together, these findings
Asymmetric income timeliness highlight the role of accounting conservatism in mitigating the potential litigation risk faced by high-tech industries.
Information asymmetry © 2014 Elsevier Inc. All rights reserved.
Litigation risk

1. Introduction
This study examines whether non-financial information based on in-
novation efficiency reduces asymmetric information and litigation risk.
Research and development (R&D) is one of the most important corpo-
rate strategies that firms have to take, despite the fact that investments
in R&D do not always result in immediate product innovation. Barth and
Kasznik (1999) argue that firm investment in intangible assets induces
greater information asymmetry than does investment in tangible assets.
A firm's R&D expenditure is a major contributor to information asym-
metry and the outcome of such innovation inputs plays a key role in
mitigating shareholders' concerns about managerial expropriation. A
less innovation-efficient firm may be more vulnerable to shareholder
litigation than a high innovation-efficient firm that is more capable of
transforming new ideas into granted patents.
Specifically, I examine whether innovation efficiency affects conser-
vative reporting, manifested through less timely gain recognition and
more timely loss recognition. Earnings conservatism facilitates efficient
contracting and could increase firm value by reducing contracting and
litigation costs (Watts, 2003a,b). Recent literature confirms the eco-
nomic benefits of timely loss recognition in financial reporting by
documenting that conservative accounting choices mitigate informa-
tion asymmetry and reduce the costs of capital (e.g. Garcia Lara, Garcia
Osma, & Penalva, 2011). Since R&D and other internally developed in-
tangibles are generally expensed, investors have little information on
E-mail address: lcchan@nccu.edu.tw.
the value and productivity of firms' R&D investments. Prior studies
show that management may disclose private information through con-
ference calls to reduce R&D-related asymmetric information (Tasker,
1998a,b). Unlike these previous studies, I examine the role of patents,
which are a direct output from R&D activities, in mitigating information
asymmetries and reducing litigation risks.
Conservative reporting defers earnings and generates lower net as-
sets and emerges as a defensive response by managers and auditors to
reduce the likelihood of being sued in a shareholder lawsuit (Watts,
2003a). Technology firms are likely to have greater conservative income
as they are subject to both shareholder litigation risk and conservative
accounting practices (Chandra, 2011). Taiwan is a natural laboratory
for studying this issue since it has an economy dominated by high-
tech industries, where investors see R&D as a crucial determinant of
firm value but are poorly protected by regulation. Shareholders are es-
pecially sensitive to the agency problems that ensue from R&D expendi-
tures in intellectual capital-intensive firms as a result of the high
correlation between innovation input (which enhances competitive-
ness) and proprietary information (which reduces disclosure incen-
tives). Such agency problems can easily lead to corporate litigation,
especially shareholder litigation. Given the relative lack of transparency
of Taiwanese financial reporting and the potential inability of the public
to understand the details released in conference calls about firms'
innovation efforts, patents are important as they certify the credibility
of otherwise hard-to-verify voluntary disclosures regarding R&D
successes. The introduction of the Securities and Futures Investor
Protection Act in Taiwan in 2003 also provides a natural choice in
which to observe how an exogenously induced change in the shareholder

1057-5219/$ – see front matter © 2014 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.irfa.2014.01.014
 A.L.-C. Chan / International Review of Financial Analysis 32 (2014) 132–142
litigation environment affects asymmetric income timeliness, especially
among innovation-intensive firms. By evaluating the impact of this
regulation, my study also substantiates the view that litigation risk is
one of the main driving forces behind earnings conservatism, especially
in high-tech sectors.
Some argue that it is costly for outside investors to acquire R&D-
related information since it requires a significant investment in gather-
ing scientific knowledge, analyzing financial statements, and participat-
ing in conference calls (Aboody & Lev, 2000; Hirshleifer, Hsu, & Li,
2013). Hirshleifer et al. (2013) indicate that information about new
technologies is difficult for investors to process as it requires knowledge
of how the economic fundamentals are changing as well as an analysis
of the innovation process – from the generation of a new idea to the re-
lease of the final product on the market – since the expected profit is
highly uncertain. Investors may be unable to respond fully and immedi-
ately to the arrival of relevant public information on innovative activi-
ties.1 Levitas and Mcfadyen (2009) also argue that new knowledge of
innovation may not be easily diffused to external investors, resulting
in knowledge asymmetries between management and external
suppliers of capital. In the light of these studies, investments in R&D in-
crease a firm's intangible resources but also contribute to the informa-
tion asymmetry between management and outside investors.
Existing accounting research documents that conservatism in finan-
cial statements facilitates efficient contracting between managers and
outside investors in the presence of agency problems (e.g. LaFond &
Watts, 2008; Watts, 2003a,b). Watts (2003a) provides four main
explanations for earnings conservatism – namely, contracting incen-
tives, taxation, litigation and regulation – and states that conservative
reporting provides economic benefits from both the contracting and
litigation perspectives. Following the registration of patents, the asym-
metric information is alleviated as investors acquire wealth benefits
from these investments, which also reduces the probability of share-
holder litigation lawsuits. As firms differ in their abilities to convert
R&D into tangible outputs and future performance, I investigate the
extent to which asymmetric income timeliness is conditional on the
success of a firm's R&D efforts.
Prior studies show that the asymmetric timeliness of earnings is
higher in periods when shareholder litigation is more prevalent in the
US (Basu, 1997; Lobo & Zhou, 2006). Emerging markets generally have
less transparency in their information environments and weaker inves-
tor protection than mature markets. In Taiwan, the Securities and Fu-
tures Investor Protection Act, which went into effect on January 1,
2003, aims to ensure the protection of investors by establishing a class
action litigation mechanism. Since the introduction of the act, firms in
Taiwan have faced a greater probability of shareholder class action law-
suits, which is likely to have induced them to report economic losses in a
more timely fashion. I exploit this unique setting of an investor protec-
tion regime change in an emerging market to examine the extent to
which mandatory patent granting could alleviate the asymmetric infor-
mation and expected litigation costs arising from R&D, thereby reducing
litigation-driven earnings conservatism. The sample comprises firms
listed on the Taiwan Stock Exchange from 2000 to 2008. Compared to
the world level, business innovation in Taiwan is quite active (Chin,
Lee, Wang, & Kleinman, 2007). According to the report ‘How technology
sectors grow: Benchmarking IT industry competitiveness 2008’, IT compet-
itiveness in Taiwan is ranked second in the world, slightly behind that of
the US (Yeo, 2008). The country's strong performance in R&D is the main
contributor to improvements in IT competitiveness, in an emerging mar-
ket that is strongly connected to the global network of production and
1
As investors may place less weight on information that is more difficult to process, the
limited attention investors pay to innovation activities could lead to return predictability.
Indeed, Hirshleifer et al. (2013) find that innovation efficiency strongly predicts subse-
quent stock returns in the US. They argue that investors underreact to information
concerning innovative activities due to the difficulty in evaluating their economic implica-
tions, and that those innovation-efficient firms may be undervalued by investors.
133
innovation (Ernst, 2010). Moreover, the nation has been ranked top in
the field of patents by generating one patent for every 2000 persons
(Yeo, 2008). These statistical facts make the findings of this study even
more interesting and provide out-of-sample evidence regarding the
role of conservatism in mitigating asymmetric information in an
environment with poor investor protection.
To measure the degree of earnings conservatism, this study applies
the Basu (1997) model by regressing reported earnings on stock returns
and allowing the return coefficient to change with the sign of the return.
The quantity of patents granted indicates the non-financial attributes of
a firm's innovation process, provides valuable information to share-
holders, and reduces information asymmetry, agency costs, and expect-
ed litigation risk. The results show that innovation efficiency, measured
as patent count scaled by R&D capital, is negatively associated with
asymmetric income timeliness. Specifically, companies that are less effi-
cient in transforming new ideas into granted patents are more timely in
recognizing losses and less timely in recognizing gains. Such evi-
dence is more pronounced among R&D-intensive firms, which face
severe information asymmetry and greater expected litigation
costs. In parallel, I seek evidence of the impact of an exogenous
change in the shareholder litigation environment on conservative
reporting. I find that the asymmetric income timeliness is more evident
in the period since the introduction of the Investor Protection Act, when
the shareholder class action litigation mechanism became effective. The
effect of this is more pronounced among less innovation-efficient firms
that are more vulnerable to shareholder class action lawsuits. The
findings of this study contribute to the accounting literature on the
beneficial role of conservatism in financial reporting and highlight the
importance of strengthening investor protection mechanisms in devel-
oping economies.
The paper is organized as follows. Section 2 reviews the literature
and develops hypotheses. Section 3 describes the methodology, sample
and data. Section 4 reports the empirical findings. Conclusions are given
in Section 5.
2. Literature review and hypothesis development
2.1. Institutional environment in Taiwan
2.1.1. Patent disclosures
Companies in Taiwan seldom disclose innovative activities in their
financial statements, due to concerns over proprietary costs, and if
they do disclose any information about patents, the majority of that
information is concerned with patent infringements. Thus, the Taiwan
Intellectual Property Office's (TIPO) announcement of the number of
patents granted each year complements the hard information in finan-
cial statements and verifies the soft information of voluntary disclosures
through conference calls, press releases or conversations with analysts.
According to the annual statistics provided by TIPO, the majority of pat-
ents are granted in the electronics sector.2 Industrial development in
Taiwan is focused on made-to-order manufacturing services and
researching and developing manufacturing processes so as to attract or-
ders from large international companies and improve the efficiency of
producing high-tech products. Taiwanese companies are proficient in
original design and are dominant manufacturers in certain areas, such
as desktop personal computers, notebooks, LCD screens, and mother-
boards (Einhorn, 2005). In recent years, several companies have pur-
sued an aggressive strategy of filing protective patents in the face of
increasing global competition. Although there has been rapid growth
in the quantity and quality of innovation carried out by companies in
Taiwan, the high patent count is highly concentrated in terms of
2
For instance, Taiwanese companies are major global suppliers of made-to-order chips
(e.g., Taiwan Semiconductor Manufacturing Co.) and manufacturers of DRAM memory
chips (e.g. Mosel Vitelic Inc. and Winbond Eletronics Co.).
134 A.L.-C. Chan / International Review of Financial Analysis 32 (2014) 132–142
products and patent holders (Ernst, 2010). Moreover, most patents are
used as tools for self-protection or cross-licensing, and generally have
weak patent citations, science linkages, and technological capabilities
(Liu, 2001).
Public corporations in emerging markets generally have low disclo-
sure quality and their ownership is dominated by pyramid and cross-
holding structures which exacerbates agency conflicts between the
controlling owners and the minority shareholders (Lee, 2007).3 While
companies in Taiwan engage heavily in innovation activities, the lack
of accounting recognition and other appropriate disclosures reduce
the informativeness of financial statements. According to La Porta,
Lopez-de-Silanes, Shleifer, and Vishny (1998), Taiwan can be catego-
rized as part of the Germanic, civil-law group of nations, for which the
transparency of the information environment and investor protection
are lower than in common-law countries. The lack of clarity or suitable
regulatory standards and the weak legal enforcement of the financial
market and the regulatory reporting agencies in Taiwan result in a rel-
atively opaque information environment. To communicate information
to outside investors, traditionally, companies have relied on informal
networking or personal relationships. As the financial market becomes
more developed, an institutional framework involving external codified
regulations has appeared, replacing these more informal networking
strategies (Chin et al., 2007).
2.1.2. Investor protection
The Securities and Futures Investor Protection Act, which went into
effect on January 1, 2003, authorized the government to establish an
institution for investor protection. This institution is the Securities and
Futures Investors Protection Center (SFIPC). The services provided by
the SFIPC include consultation and mediation, handling investor com-
plaints, filing class-action lawsuits on behalf of investors, and managing
an investor compensation fund. Its role is to strengthen the investor
protection mechanism and to promote the sound development of the fi-
nancial markets in Taiwan. For any securities investment trading fraud
involving twenty or more victims, the SFIPC may file lawsuits or an ar-
bitration application under its own name as an authorized representa-
tive of the victims, in accordance with the Investor Protection Act and
the SFIPC Charter, which could reduce the total litigation cost and
avoid congestion of the court's trial calendars.
The Securities and Exchange Law in Taiwan aims to protect investor
rights and maintain the stock market as a well-functioning platform for
financing. The law contains articles that forbid certain types of activities
that would jeopardize fairness in trading and specifies four major types
of crime: making false financial statements, producing false prospectus-
es, influencing share prices illegally, and insider trading. The SFIPC ana-
lyzes cases brought by prosecutors to determine whether they involve
any of the aforementioned crimes and brings class-action suits on behalf
of investors seeking compensation. Since the introduction of the
Investor Protection Act in 2003, the center has won more than twenty
cases and several are ongoing.4 Among the winning cases, half relate
to misleading prospectus or financial statements, and there is a greater
incidence of class action lawsuits among technology firms than other
firms. If, as has been hypothesized in prior research, earnings conserva-
tism is driven by shareholder litigation, I would expect to observe
higher earnings conservatism in the period since the introduction of
the Investor Protection Act, when the class action lawsuit mechanism
was established.
3
East Asian firms, including those in Taiwan, are typically majority owned and family
controlled. Prior studies document that this type of corporate ownership structure reduces
the credibility of reported earnings and is associated with lower earnings informativeness
(e.g., Fan & Wong, 2002).
4
Information is available from the SFIPC at http://www.sfipc.org.tw.
2.2. Innovation activity and information asymmetry
Investments in innovation activities generate knowledge, which is
an important input into a firm's production function (Pandit, Wasley,
& Zach, 2011). Firms with higher rates of success in their innovations
are likely to exhibit superior future performance. However, innovating
is a highly uncertain process. Kothari, Laguerre, and Leone (2002) find
that R&D investments generate more uncertain future benefits than
investments in fixed assets. Amir, Guan, and Livne (2007) further
show that the positive relation between R&D and the volatility of future
earnings described in Kothari et al. (2002) is stronger in R&D-intensive
industries. Chan, Lakonishok, and Sougiannis (2001) find a positive
association between R&D spending and stock return volatility, due to
investor uncertainty about future payoffs. From the shareholder's view-
point, the success of R&D activities is indicated by the registration of
patents. This enables them to predict the future benefits of R&D.
Corporate investments generate information asymmetry as insiders
have better information than external investors and can continually ex-
amine the changes in the potential of individual investments, while out-
side investors can only gather limited information. The extent of
information asymmetry associated with intangible resources is higher
than that with tangibles due to their unique nature and accounting
treatments of them (Aboody & Lev, 2000). Information asymmetry be-
tween management and outside investors creates agency costs that, in
turn, impair equity value. All parties involved in the firm have an incen-
tive to reduce agency costs by finding mechanisms to make the infor-
mation supplied to investors credible. Examples of such mechanisms
include better corporate governance, accounting policy choices, and
voluntary disclosure (e.g. Gelb, 2002).
Prior studies (e.g. Brown, Hillegeist, & Lo, 2004) confirm that
companies may voluntarily disclose information, through confer-
ence calls and press releases for example, to reduce information
asymmetries. Barth, Kasznik, and McNichols (2001) report that,
due to the private nature of information regarding R&D, analyst cover-
age is significantly greater for R&D-intensive firms than for firms with
lower levels of R&D or none at all. Furthermore, empirical evidence
shows that R&D-intensive companies hold more conference calls with
analysts than do firms engaged in low levels of R&D (Tasker, 1998a),
implying that investors have a strong demand for information about
companies' intangibles, and that the majority of questions raised by an-
alysts in conference calls are R&D and innovation-related (Tasker,
1998b).
The lack of recognition of R&D outlays as accounting assets in the fi-
nancial statements causes differences in beliefs between managers and
outside investors. The investing public is unable to directly observe the
process of innovation or the success or failure of R&D implementation.
Given the relative lack of transparency of Taiwanese financial reporting
and the difficulty for investors in understanding the details of firms' in-
novation efforts, the use of conference calls appears to be a medium for
communicating with investors. More innovative firms appear to hold
conference calls more frequently (Chin et al., 2007). However, the even-
tual output that is indicated by the number of patents granted further
certifies the credibility of the information released through these volun-
tary disclosures regarding R&D successes.
2.3. The role of accounting conservatism
Previous studies identify two types of accounting conservatism: un-
conditional and conditional. Unconditional conservatism is related to
the balance sheet and reflects the understatement of book values of
net assets (e.g., the immediate expensing of R&D or adopting accelerat-
ed depreciation and shorter asset lives). It is unrelated to changes in
future cash flows (news independent). On the other hand, conditional
conservatism is news dependent and earnings related. According to
Basu (1997), conditional conservatism is manifested through the asym-
metric response of earnings to economic gains and losses. Unconditional
 A.L.-C. Chan / International Review of Financial Analysis 32 (2014) 132–142
conservatism could limit the degree of conditional conservatism, since
firms with lower net asset values have less scope to write down im-
paired assets (Beaver & Ryan, 2005). Moreover, conditional conserva-
tism involves a greater degree of managerial discretion (e.g.,
restructuring charges, gains and losses on the sale of assets, and the ef-
fects of discontinued operations) as firms may use different verifiability
criteria to recognize economic gains and losses. Garcia Lara, Garcia
Osma, and Mora (2005) show that, in institutional contexts where
there is weaker investor protection and a less dispersed ownership
structure, earnings management is a significant factor in conditional
(earnings) conservatism.
Earnings conservatism has significant benefits for financial reporting
as it addresses moral hazard problems and constrains managers' oppor-
tunistic behavior in reporting biased accounting measures. LaFond and
Watts (2008) indicate that conservative reporting is a rational response
to the information asymmetry between managers and outside inves-
tors, aimed at lowering the agency costs associated with possible man-
agerial expropriation and helping investors to assess alternative sources
of information regarding equity values. As managers are reluctant to re-
port less-verifiable losses, a lower verification criterion for losses is like-
ly to elicit information that managers are unwilling to report, implying
that reported loss information is more reliable. Empirical evidence gen-
erally supports the arguments put forward by Watts (2003a) and
LaFond and Watts (2008) that conservatism is an effective mechanism
for reducing agency and litigation costs (e.g. Ahmed, Billings, Morton,
& Stanford-Harris, 2002).
The litigation explanation puts forth the idea that conservative ac-
counting, deferring earnings and generating lower net assets emerge
as a defensive response by managers and auditors to reduce the likeli-
hood of being sued in shareholder litigation (Watts, 2003a). Empirical
studies provide evidence of a greater asymmetric timeliness of earnings
in periods when shareholder litigation is more prevalent in the US
(Basu, 1997; Lobo & Zhou, 2006). At the firm level of analysis, Chandra
(2011) argues that technology firms are likely to be more conservative
in their reporting than other firms because they are subject to both
higher shareholder litigation risk and conservative accounting practices.
These firms display a higher unconditional conservatism due to the re-
quirement to immediately expense R&D, which is manifested through
reduced operating cash flows. On the other hand, conditional conserva-
tism of R&D-intensive firms is mainly driven by litigation risk and is
manifested through timely loss recognition, such as asset write-
downs, restructuring charges and other loss accruals.
2.4. Hypothesis development
As the expected litigation costs of the firm increase, liability rules
granting shareholders the right to sue for financial statement misrepre-
sentation generate incentives to recognize economic losses in earnings
in a more timely fashion (Watts, 2003a,b). The higher growth opportu-
nities and increased risk faced by R&D-intensive firms make them more
vulnerable to shareholder litigation (Francis, Philbrick, & Schipper,
1994; Johnson, Kasznik, & Nelson, 2001; Skinner, 1997). Chandra
(2011) shows that technology firms in the US are more conservative
in their reporting, as a result of lower operating cash flows that are
due to the expense of R&D outlays and more income-decreasing ac-
cruals in response to litigation risk.
Investors benefit from examining patent-based indicators that re-
flect various aspects of innovation activities. Levitas and Mcfadyen
(2009) indicate that patents provide positive signals to external parties,
which could increase investors' valuations of a firm and make external
capital more accessible. Pandit et al. (2011) combine the use of R&D ex-
penditure and non-financial measures such as patent counts to shed
further light on the informational role of R&D and how firms differ in
their ability to transform R&D into innovation outputs. They argue
that their finding that more innovation-efficient firms have a higher
market value is not only due to the real option value embedded in the
135
patents, but also to the predicted superior future performance. As
firms differ in their ability to transform R&D into innovation outputs,
following Hirshleifer et al. (2013), I define innovation efficiency as num-
ber of patents (patent count) over R&D capital.
The accounting practice of immediately expensing R&D limits the
degree of conditional conservatism. However, the threat of shareholder
litigation may induce managers to accrue losses in a timely fashion. I
predict that firms that are less capable of transforming their R&D inputs
into granted patents (i.e., those with lower R&D success or that are less
innovation-efficient) are likely to recognize losses in a timely manner,
primarily because of accounting choices made in response to the litiga-
tion risk that is associated with the highly uncertain future economic
benefits of R&D outlays. This leads to the first hypothesis:
H1. Less innovation-efficient firms are associated with greater asymmetric
income timeliness and this is especially evident among highly R&D-
intensive firms.
In Taiwan, domestic individual investors perform the majority of all
trades. Individual investors tend to be short-term speculators without
the resources to undertake fundamental firm research (Chiao, Wang,
& Lai, 2009). These individual investors are in an economically inferior
position when they encounter any conflict against securities firms or se-
curities issuers. Once their rights or interests are violated, individual in-
vestors seldom pursue judicial procedures to protect themselves, due to
inadequate information, limited personal capabilities, or high litigation
costs. Class action litigation is commonly exercised as a solution in the
US, where victims of the same fraud case are allowed to seek protection
collectively. Other countries, such as the UK, Japan and Korea, have
established specialized institutions or funds in charge of investor pro-
tection services.
The Investor Protection Act in Taiwan, announced in 2003,
established a class action litigation mechanism and an investor protec-
tion fund, making reference to foreign practices and related consumer
protection rules. As this act has strengthened the investor protection
mechanism, expected litigation costs are likely to increase due to the in-
creased probability of shareholder lawsuits. As shareholder class action
litigation becomes more prevalent, companies are likely to report more
conservatively and this phenomenon should be more pronounced
among innovative firms that are less efficient in transforming new
ideas into patents. This leads to the second hypothesis:
H2. Less innovation-efficient firms are associated with greater asymmetric
income timeliness and this is especially evident since the introduction of the
Investor Protection Act.
3. Methodology, sample and data
To measure the degree of conservative reporting, this study applies
the Basu (1997) framework to capture the asymmetric timeliness of
earnings:
NIit ¼ α 1 þ α 2 DUMit þ α 3 Rit þ α 4 Rit DUMit þ εit ð1Þ
where NI is earnings per share scaled by the price at the beginning of the
fiscal year-end, R is the annual return of firm i for year t obtained from
the monthly returns over the fiscal year, so as to capture any news
occurring over the accounting period and DUM is a dummy variable tak-
ing the value one when R is negative and zero otherwise. The sign of the
return (R) is a proxy for economic gains (good news) or losses (bad
news). From Eq. (1), the degree of earnings conservatism depends on
the extent to which reported earnings contemporaneously reflect
value-relevant news. The intercept (α1) captures the cost of equity
and the effect of the previous year's news and α2 has an expected
value of zero. The slope coefficient α3 indicates the earnings response
to good news and α4 is the incremental earnings response to bad
136 A.L.-C. Chan / International Review of Financial Analysis 32 (2014) 132–142

news. Unconditional conservatism resulting from the immediate period into before (2000–2002) and after (2003–2008) the Investor
expensing of R&D will exert a negative effect on income that is indepen- Protection Act. As the litigation risk increases after the introduction of
dent of economic news, which will reduce the intercept α1. As fewer as- the act in 2003, and investors have more protection, I would expect
sets are capitalized for R&D-intensive firms, this reduces the scope for the change in the shareholder litigation environment to cause compa-
future impairment, resulting in a lower α4. However, the higher expect- nies to accrue losses in a more timely fashion and for this to be more
ed litigation costs faced by firms with lower R&D success are associated pronounced among the less innovation-efficient firms (i.e. those with
with a greater α4 when managers take asset write-downs or recognize low R&D success).
other losses as a defensive strategy for reducing litigation risk. The control variables are firm size, leverage and the beginning
I report pooled regression results and t-statistics based on Huber– market-to-book ratio. Firm size (SIZE) is measured as the natural
White standard errors. Although the Basu (1997) model has been criti- log of total assets; leverage (LEV) captures firm risk and is calculated
cized in several studies (e.g. Dietrich, Muller, & Riedl, 2007; Givoly, as total liabilities divided by total assets; and the market-to-book
Hayn, & Natarajan, 2007), the asymmetric timeliness of earnings is the ratio (MB) proxies for the investment opportunity set and the degree
most direct indicator of earnings conservatism and Basu's (1997) of past conservatism. I expect LEV to be positively associated with
return-based test has been widely applied in empirical studies. Ball, earnings conservatism due to the demand for efficient debt
Kothari, and Nikolaev (2013) counter the criticisms by showing a formal contracting (LaFond & Watts, 2008; Watts, 2003a). MB reflects both
econometric analysis of Basu's (1997) return-based model. Previous net asset conservatism and firm growth. Net asset conservatism
studies indicate that the market-to-book ratio is associated with the (higher MB) constrains the asymmetric timeliness of earnings,
asymmetric timeliness of earnings (Roychowdhury & Watts, 2007) while greater firm growth is associated with greater information
and that the ratio is an alternative measure of asset recognition that is asymmetry and timely loss recognition (LaFond & Watts, 2008).
correlated with economic leverage (Beaver & Ryan, 2009). To address There is no direct relation between SIZE and timely loss recognition.
the concerns over the use of the return-based test, I control for the Larger firms may report more conservative earnings due to greater
market-to-book ratio, leverage and firm size. political costs. However, information asymmetry may be lower for
Efficiency of innovation activities is measured as the number of pat- larger firms as they are generally more mature and have established
ents (PATNUM) relative to R&D capital in fiscal year t. Following Chan disclosure policies (Ozkan & Ozkan, 2004). Consequently, I do not
et al. (2001) and Hirshleifer et al. (2013), I compute the R&D capital as make a specific prediction about the relation between timely loss
the five-year cumulative R&D expenditure, assuming an annual amorti- recognition and either SIZE or MB.
zation rate of 20%: I collect accounting and market data for companies listed on the
Taiwan Stock Exchange during the period 2000 to 2008, taken from
R&D capital ¼ RDit þ 0:8  RDit−1 þ 0:6  RDit−2 þ 0:4  RDit−3 the Taiwan Economic Journal database. Patent data is collected
þ 0:2  RDit−4 : ð2Þ from the Intellectual Property Office of the Ministry of Economic Af-
fairs. Panel A of Table 1 describes the sample selection. I exclude
I take the natural log of the patent-related measure, as its distri- companies in the financial and utility industries as they are highly
bution is highly skewed and its value is often zero. Thus, the innova- regulated sectors and adopt different accounting practices and
tion efficiency (IE) is calculated as ln[1 + (PATNUM/R&D capital)].5 reporting rules from other sectors. After excluding observations
To examine the impact on asymmetric income timeliness from the with missing accounting data, market data or the innovation
information asymmetry and expected litigation risks arising from R&D
investments, I use earnings before R&D expenditure but after the
imputed R&D amortization expenses scaled by beginning share price, Table 1
assuming an annual amortization rate of 20% (NI⁎). I augment Eq. (1) Sample selection.
to incorporate innovation efficiency and to control for determinants of Panel A: Sample selection criterion
earnings conservatism.
Sample selection criterion Firm-year observations

NI it ¼ β1 þ β2 DUMit þ β3 Rit þ β4 Rit DUM it þ β5 LIEDUM Original sample (all listed Taiwanese 6082
companies 2000–2008)
þβ6 LIEDUMit DUMit þ β7 LIEDUMit Rit þ β8 LIEDUM it Rit DUMit ð3Þ Less: financial and utility companies (494)
Less: missing accounting and market- (811)
þControls þ εit
based data (earnings, annual stock returns,
total liabilities, share prices, total assets,
LIEDUM is a dummy variable equal to one if innovation efficiency market-to-book ratio)
(IE) is less than the yearly median or is equal to zero, and zero other- Less: missing innovation efficiency measure (1367)
Final sample 3410
wise. To test H1, I separate the firms into groups with high and low
R&D intensity based on the yearly median values and perform multivar- Panel B: Industry distribution
iate regression analysis. In response to shareholder litigation, I expect Industry No. of firms No. of obs.
less innovation-efficient firms to be less timely than more innovation-
Automobiles and parts 3 27
efficient firms in recognizing gains, as indicated through a negative β7,
Chemicals and biotechnology 31 206
and more timely in recognizing losses, as indicated through a positive Construction and materials 12 74
β8. This effect will be more pronounced for firms with greater innova- Electrical and mechanical engineering 34 256
tion inputs (i.e. highly R&D-intensive firms) as they are subject to great- Electrical appliances and telecommunications 11 80
Electronic manufacturer 306 1953
er uncertainty over their future performance and have higher expected
Food producers 14 124
litigation costs than other companies. To test H2, I separate the sample General retailers 3 11
Glass and ceramic 5 37
Industrial transportation 5 17
5 Metal 17 109
The measure of innovation efficiency is missing if either the number of patents is miss-
Paper 6 38
ing or R&D capital is missing or zero. As it is difficult to precisely identify the length of the
Plastics 18 129
lags between R&D outlays and successful patent applications, I also consider different al-
Rubber 7 58
ternative deflators. Assuming a two-year lag in the processing of applications, I apply ro-
Textile and fibers 37 291
bustness tests using patent counts scaled by three-year or five-year cumulative R&D
Total 509 3410
expenditures ending in the year t-2. These lead to similar conclusions.
 A.L.-C. Chan / International Review of Financial Analysis 32 (2014) 132–142 137

efficiency measure, the final sample includes 3410 firm-year observa-
tions for 509 firms. Panel B shows the distribution of the sample by in-
dustry. The majority of the firms are from the electronics sector. This is
supported by the annual statistics provided by TIPO, which show that
most of the top ten organizations in terms of number of patents granted
are manufacturers of electrical and electronic appliances.
Table 2 presents summary statistics (Panel A) and correlation analy-
ses (Panel B). To control for outliers, all variables except patent count
and innovation efficiency are winsorized at the top and bottom one per-
cent of observations. The mean (median) values of NI⁎ and R are 0.066
(0.077) and 0.046 (−0.029) respectively. The average sample firm has
a MB of 1.494 and LEV of 0.371. The mean R&D intensity (R&D over
sales) of the sample firms is 3.097%, the mean patent count (PATNUM)
is 13.527 and the mean value of innovation efficiency (IE) is 0.018.
The mean value of the dummy variable for low innovation efficiency
(LIEDUM) is 0.518. The summary statistics indicate a high variability in
the number of patents granted to the sample firms and there are a
few companies with a large number of patents.
Panel B of Table 2 reports the Pearson product‐moment and Spear-
man rank-order correlations between the variables. The findings from
the Pearson and Spearman correlations are generally similar. To facili-
tate the discussion, I focus on the Spearman correlations. NI⁎ is signifi-
cantly positively correlated with R, PATNUM, IE, SIZE and MB, and
negatively correlated with LIEDUM and LEV. R&D-intensive firms are
likely to generate more granted patents (PATNUM), and are associated
with higher innovation efficiency (IE) and a higher MB, while these
firms are negatively associated with annual share returns, and are likely
to be smaller firms with a lower degree of leverage. Moreover, firms
that are more efficient in transforming R&D inputs into patents make
greater R&D outlays and are associated with better accounting and
market-based performance measures.
4. Empirical results
4.1. Innovation efficiency and asymmetric timeliness of earnings
Table 3 reports the impact of innovation efficiency on asymmetric
income timeliness. Before considering a firm's innovation activity,
the results of the earnings–return regression generally confirm
those of prior studies. The coefficients reflecting the response of
earnings to economic gains (R) and the incremental response of
earnings to economic losses (R × DUM) are positive and significant.
LIEDUM captures the different abilities of companies to transform
new ideas into patents and equals one for firms with low innovation
efficiency, and zero otherwise. The adjusted R-square of the results,
after taking into account innovation efficiency and controlling for
several firm-specific characteristics, is 26.60%. The findings confirm
the prediction that less innovation-efficient firms are generally
more conditionally (earnings) conservative. Relative to highly
innovation-efficient firms, firms with low efficiency recognize
good news in a less timely manner (IEDUM × R = − 0.036, t =
− 2.57), while they are more timely in recognizing bad news
(IEDUM × R × DUM = 0.071, t = 1.85). The greater asymmetric in-
come timeliness reported by less innovation-efficient firms is likely
to be driven by the higher expected litigation risk. Turning to the
control variables, I find that larger companies recognize losses in a
less timely fashion as they are subject to lower information asymmetry.
In addition, firms with higher leverage report more conservative
earnings, as evidenced by the significantly positive coefficient of
R × DUM × LEV, supporting the claim by Watts (2003a,b) that con-
servatism facilitates debt contracting. Roychowdhury and Watts
(2007) show the importance of controlling for the composition
of equity values, through the use of the market-to-book ratio in

Table 2
Descriptive statistics and correlation analyses.

Panel A: Descriptive statistics

Variables Mean Median Max Min Stdev

NI⁎ 0.066 0.077 0.436 −0.442 0.142

R 0.046 −0.029 1.972 −0.768 0.532
PATNUM 13.527 0.500 1399.000 0.000 66.128
RD(%) 3.097 1.755 547.744 0.000 10.178
IE 0.018 0.000 3.053 0.000 0.090
LIEDUM 0.518 1.000 1.000 0.000 0.500
SIZE 15.718 15.508 19.603 13.569 1.272
LEV 0.371 0.369 0.762 0.078 0.150
MB 1.494 1.170 6.370 0.260 1.124

Panel B: Correlation analyses

NI⁎ R PATNUM RD(%) IE LIEDUM SIZE LEV MB

NI⁎ 1 0.479 0.095 0.017 0.110 −0.094 0.076 −0.162 0.511

R 0.414 1 0.016 −0.056 0.046 −0.015 0.014 −0.049 0.461
PATNUM 0.045 0.001 1 0.392 0.913 −0.910 0.349 −0.031 0.281
RD(%) −0.029 −0.034 0.040 1 0.283 −0.345 −0.143 −0.237 0.273
IE 0.034 0.030 0.075 −0.020 1 −0.923 0.180 −0.028 0.236
LIEDUM −0.086 −0.010 −0.210 −0.075 −0.212 1 −0.229 0.037 −0.236
SIZE 0.078 0.001 0.347 −0.059 0.028 −0.234 1 0.134 0.093
LEV −0.200 −0.046 −0.001 −0.120 0.015 0.045 0.098 1 −0.112
MB 0.350 0.425 0.181 0.120 0.025 −0.213 0.114 −0.140 1

The overall sample size is 3410 firm-year observations. The table reports summary statistics (Panel A) and correlation analyses (Panel B). All firms are listed on the Taiwan Stock Exchange
from 2000 to 2008 and the data are collected from the Taiwan Economic Journal (TEJ) database and the Intellectual Property Office of the Ministry of Economic Affair. NI⁎ is earnings before
R&D expenses but after the imputed R&D amortization expenses scaled by beginning share price, assuming an annual amortization rate of 20%; R is the annual stock return for firm i for
year t obtained by monthly returns over the fiscal year; PATNUM is the number of patents granted in the year; RD is R&D expenditures over total sales; IE is innovation efficiency calculated
as ln [1 + (PATNUM/R&D capital)]; LIEDUM is a dummy variable equal to one if innovation efficiency (IE) is less than the yearly median or is equal to zero, zero otherwise; SIZE is firm size
measured as natural log of total assets; LEV is leverage measured by the ratio of total liabilities to total assets at the fiscal-year end; and MB is the market-to-book ratio. In Panel B, the lower
left-hand section reports Pearson product-moment correlations while the upper right-hand section reports Spearman rank-order correlations. Bold text indicates significant at 5% level and
italic text indicates significant at 10% level.
138 A.L.-C. Chan / International Review of Financial Analysis 32 (2014) 132–142

Table 3
Innovation efficiency and asymmetric timeliness of earnings.

Predicted sign Coeff. (t-Stat.) Coeff. (t-Stat.) Coeff. (t-Stat.)

Intercept 0.083 (8.74)⁎⁎⁎ 0.084 (6.75)⁎⁎⁎ 0.008 (0.09)

DUM −0.011 (−1.04) −0.007 (−0.36) −0.007 (−0.04)
R + 0.075 (4.52)⁎⁎⁎ 0.095 (7.27)⁎⁎⁎ 0.098 (0.91)
R × DUM + 0.086 (2.72)⁎⁎⁎ 0.040 (1.07) 0.168 (1.69)⁎⁎
LIEDUM −0.001 (−0.10) 0.014 (1.24)
LIEDUM × DUM −0.008 (−0.38) −0.006 (−0.27)
LIEDUM × R − −0.037 (−2.94)⁎⁎⁎ −0.036 (−2.57)⁎⁎⁎
LIEDUM × R × DUM + 0.088 (1.77)⁎⁎ 0.071 (1.85)⁎⁎
SIZE 0.003 (0.50)
SIZE × DUM 0.002 (0.17)
SIZE × R + 0.004 (0.71)
SIZE × R × DUM −/+ −0.014 (−2.08)⁎⁎
LEV −0.060 (−1.77)⁎
LEV × DUM −0.055 (−1.26)
LEV × R − −0.142 (−1.35)⁎
LEV × R × DUM + 0.240 (1.63)⁎
MB 0.027 (3.23)⁎⁎⁎
MB × DUM −0.003 (−0.39)
MB × R − −0.012 (−2.16)⁎⁎
MB × R × DUM −/+ 0.016 (1.00)
Adj R-square (%) 18.13 18.93 26.60

Variables are defined in the notes to Table 2. DUM is a dummy variable taking the value one when R is negative, or zero otherwise. t-statistics in parentheses are based on Huber–White
standard errors clustered by firm and year.
⁎⁎⁎, ⁎⁎, ⁎ indicate statistical significance at 1%, 5%, and 10% level respectively for one-tailed t-tests of coefficients with predicted signs and two-tailed t-tests otherwise.

conservatism tests. The coefficient MB × R is significantly negative, I predict that evidence of firms with low innovation efficiency
consistent with the predicted effect of MB as a proxy for uncondi- being more earnings conservative should appear mainly among
tional (balance sheet) conservatism. those companies with greater investment in R&D. Table 4 reports

Table 4
Innovation efficiency and asymmetric timeliness of earnings, high and low R&D intensity.

Panel A: Median values, high and low R&D intensity

Low R&D intensity High R&D intensity Difference

Variables (Obs. = 1707) (Obs. = 1703) (High–Low)

NI⁎ 0.077 0.076 −0.001

R 0.000 −0.061 −0.060⁎⁎⁎
IE 0.000 0.006 0.006⁎⁎⁎
SIZE 15.639 15.352 −0.287⁎⁎⁎
LEV 0.395 0.340 −0.055⁎⁎⁎
MB 1.000 1.350 0.350⁎⁎⁎

Panel B: Regression analyses

Low R&D intensity High R&D intensity

Predicted sign Coeff. (t-Stat.) Coeff. (t-Stat.) Coeff. (t-Stat.) Coeff. (t-Stat.)

Intercept 0.120 (0.96) 0.119 (0.91) 0.037 (0.36) 0.005 (0.05)

DUM −0.205 (−1.26) −0.222 (−1.22) 0.110 (0.65) 0.125 (0.65)
R + −0.149 (−1.07) −0.090 (−0.54) 0.142 (0.97) 0.217 (1.47)⁎
R × DUM + 0.405 (1.29)⁎ 0.300 (0.91) 0.301 (1.08) 0.049 (0.20)
LIEDUM 0.001 (0.05) 0.017 (1.10)
LIEDUM × DUM 0.007 (0.36) −0.015 (−0.75)
LIEDUM × R − −0.032 (−1.93)⁎⁎ −0.039 (−2.10)⁎⁎
LIEDUM × R × DUM + 0.051 (1.22) 0.093 (2.08)⁎⁎
SIZE −0.002 (−0.29) −0.002 (−0.27) −0.001 (−0.11) 0.001 (0.11)
SIZE × DUM 0.012 (1.09) 0.013 (1.08) −0.004 (−0.35) −0.004 (−0.37)
SIZE × R + 0.019 (1.70)⁎⁎ 0.016 (1.31)⁎ 0.001 (0.24) −0.003 (−0.44)
SIZE × R × DUM −/+ −0.027 (−1.58) −0.023 (−1.23) −0.021 (−1.19) −0.007 (−0.44)
LEV −0.144 (−3.16)⁎⁎⁎ −0.146 (−3.09)⁎⁎⁎ 0.029 (0.40) 0.025 (0.36)
LEV × DUM 0.047 (0.90) 0.049 (0.99) −0.168 (−3.13)⁎⁎⁎ −0.166 (−3.11)⁎⁎⁎
LEV × R − −0.107 (−0.78) −0.104 (−0.76) −0.172 (−1.03) −0.157 (−1.01)
LEV × R × DUM + 0.315 (1.88)⁎⁎ 0.311 (1.87)⁎⁎ 0.210 (0.96) 0.183 (0.85)
MB 0.045 (4.51)⁎⁎⁎ 0.045 (4.55)⁎⁎⁎ 0.022 (2.90)⁎⁎⁎ 0.023 (3.03)⁎⁎⁎
MB × DUM −0.008 (−0.60) −0.008 (−0.58) −0.001 (−0.10) −0.002 (−0.22)
MB × R − −0.022 (−2.03)⁎⁎ −0.025 (−2.23)⁎⁎ −0.008 (−1.55)⁎ −0.009 (−1.90)⁎⁎
MB × R × DUM −/+ 0.023 (0.86) 0.027 (0.98) 0.018 (0.99) 0.018 (0.98)
Adj R-square (%) 31.05 31.09 24.69 24.90

Variables are defined in the notes to Table 2. DUM is a dummy variable taking the value one when R is negative, or zero otherwise. High (low) R&D intensity consists of companies with R&D
expenditures over net sales higher (lower) than the yearly median. t-statistics in parentheses are based on Huber–White standard errors clustered by firm and year. ⁎⁎⁎, ⁎⁎, ⁎ indicate
statistical significance at 1%, 5%, and 10% level respectively for one-tailed t-tests of coefficients with predicted signs and two-tailed t-tests otherwise.
 A.L.-C. Chan / International Review of Financial Analysis 32 (2014) 132–142
the findings for sub-samples partitioned by R&D intensity. Firms with
R&D intensity greater than the yearly median are included in the highly
R&D-intensive group, which consists mainly (around 91% of the sub-
sample) of companies from the biotechnology, electronics and electrical
equipment sectors. Panel A shows the median values of the main vari-
ables. The results indicate that more R&D-intensive firms are generally
more efficient in getting patents granted and have higher market-to-
book ratios, while incurring significantly lower share returns and
being smaller and lower leveraged.
Panel B of Table 4 reports the regression analyses. I find that, be-
fore taking innovation efficiency into account, the response of
earnings to bad news is significant for the highly R&D-intensive
firms (R + R × DUM = 0.142 + 0.301 = 0.443, p-value = 0.06)
but insignificant for the less R&D-intensive firms. This supports
the conjecture that R&D-intensive firms are likely to be more earn-
ings conservative in response to greater litigation risks. After con-
sidering the firms' different capabilities in transforming new ideas
into patents, the results indicate that highly innovation-efficient
firms in both sub-samples do not display asymmetric income time-
liness. However, probably because they face greater litigation costs,
highly R&D-intensive but less innovation-efficient firms are shown
by the results to be more timely in recognizing losses and less timely in
recognizing gains. The coefficient of LIEDUM × R is significantly negative
(−0.039, t = −2.10) and the coefficient LIEDUM × R × DUM is signifi-
cantly positive (0.093, t = 2.08) in the highly R&D-intensive group. I
also find that among the less R&D-intensive firms, lower innovation ef-
ficiency is associated with less timely gain recognition but I find no ev-
idence of timely loss recognition. These findings suggest that less
innovation-efficient firms are likely to be more conservative in their
reporting, through less timely gain and more timely loss recognition,
and this is more evident in the highly R&D-intensive firms, which sup-
ports the litigation explanation. Of the control variables, LEV is the main
determinant of timely loss recognition among low R&D companies.
Fig. 1 compares the incremental response of earnings to economic
losses in high and low R&D-intensive groups. The percentage difference
between high and low innovation efficiency is calculated using the coef-
ficients from Panel B of Table 4 (i.e., LIEDUM × R × DUM/R × DUM). It can
be observed that the incremental response of earnings to bad news
among less innovation-efficient firms is greater than that of highly
innovation-efficient firms in both high and low R&D groups, and the dif-
ference is more pronounced among highly R&D-intensive companies.
This indicates that firms that are less capable of transforming their
R&D inputs into granted patents are more likely to recognize losses in
a timely manner, especially among those with high R&D intensity.
200%
180%
160%
140%
120%
100%
Percentage difference
80%
of the incremental
60% earnings response to
40% bad news
20%
0%
Low R&D intensity High R&D intensity
Fig. 1. Differential bad news effect between less innovation-efficient and highly innovation-
efficient firms: comparison of high and low R&D intensity. Percentage difference in the de-
gree of earnings conservatism between high and low innovation efficiency is calculated
using the coefficients from Panel B of Table 4. The coefficient of R × DUM and the sum of
the coefficients of R × DUM and LIEDUM × R × DUM capture the incremental response of
earnings to bad news for highly innovation-efficient and less innovation-efficient firms,
respectively.
139
Table 5 reports the effects of the exogenous increase in litigation risk
caused by the introduction of the Investor Protection Act in 2003. Panel A
reports the median values of the main variables in the sub-periods. The
results indicate that the market-to-book ratio is higher while leverage
is lower in the period since the introduction of the act. I do not find
any difference in innovation efficiency before and after the change in
the shareholder litigation regime. Panel B shows that firms are more
timely in their recognition of losses in the period when litigation risk is
higher.6 Before taking innovation efficiency into account, the coefficient
capturing the incremental response of earnings to bad news (R × DUM)
is significantly positive (0.457, t = 4.53). After considering the impact
of innovation activities, I find that more innovation-efficient firms are
more timely in recognizing losses only in the period after the introduction
of the Investor Protection Act (R × DUM = 0.233, t = 1.93). However, the
asymmetric income timeliness is more prevalent among firms with
greater information asymmetry (i.e. less innovation-efficient firms) in
that period. In the post-act period, the coefficient of LIEDUM × R is signif-
icantly negative (−0.031, t = −2.05), that of LIEDUM × R × DUM is sig-
nificantly positive (0.098, t = 2.13), and the sum of the coefficients of
LIEDUM × R and LIEDUM × R × DUM is positive and significant, indicating
that less innovation-efficient firms are associated with less timely gain
recognition and more timely loss recognition than more innovation-
efficient firms. In addition, the findings show that leverage is associated
with timely loss recognition and that larger firms report less conservative
earnings in the post-act period.
Fig. 2 compares the incremental response of earnings to economic
losses before and after the change in the shareholder litigation regime.
The percentage difference between high and low innovation efficiency
is calculated using the coefficients of R × DUM and LIEDUM × R × DUM
from Panel B of Table 5. The result shows that the degree of earnings con-
servatism among less innovation-efficient firms is higher than that of
highly innovation-efficient firms in both sub-periods, and the difference
is more pronounced in the post-act period when litigation costs are ex-
pected to be higher. This is consistent with the conjecture that less
innovation-efficient firms are likely to recognize bad news in a more
timely fashion, especially when they are subject to higher litigation risk.
4.2. Additional analyses
4.2.1. Conference calls
It has been documented that management discloses R&D-related in-
formation via conference calls in an attempt to reduce asymmetric in-
formation between themselves and investors (e.g. Chin et al., 2007;
Tasker, 1998a,b). I thus also examine the role of conference calls in mit-
igating the asymmetry of R&D-related information. The frequencies
with which the companies hold conference calls are hand-collected
from China Times. On average, there are conference calls held relating
to 21% of the firm-year observations, with the maximum frequency
being eight times a year. I further augment Eq. (3) with a dummy vari-
able equal to one if firms hold at least one conference call (CALLDUM).
Table 6 shows that less R&D-intensive firms that hold conference calls
are associated with the more timely recognition of gains but less timely
recognition of losses. However, I do not find such evidence among
6
One could argue that other confounding factors, such as the revised patent law in 2004
and changes in accounting rules, may also contribute to the results. However, the conclu-
sions of this study are not seriously affected by these factors for the following reasons.
First, I do not find any significant difference in the median values of innovation efficiency
before and after the revised regulations for patents became effective. Second, the process
of gradual convergence to IFRS in Taiwan began in 1999, and so overlaps the entire test pe-
riod of this study. It is possible that, since the convergence to IFRS, management and audi-
tors have been more conservative in their reporting as a defensive strategy to cope with
the rapid changes in accounting rules and so reduce the possibility of violating these ac-
counting rules and the expected litigation risks. If this is the case, then earnings conserva-
tism would be steadily increasing over the years. My findings not only show higher
earnings conservatism in the period after the year 2003 but further demonstrate that in-
novation efficiency plays a role in determining asymmetric income timeliness, mainly in
the period when the shareholder class action lawsuit had become effective.
140 A.L.-C. Chan / International Review of Financial Analysis 32 (2014) 132–142

Table 5
Innovation efficiency and asymmetric timeliness of earnings, before and after the Securities and Futures Investor Protection Act.

Panel A: Median values, before and after the Securities and Futures Investor Protection Act

Pre-act Post-act Difference

Variables (Obs. = 857) (Obs. = 2553) (Post–Pre)

NI⁎ 0.071 0.077 0.006

R −0.117 −0.006 0.111⁎⁎⁎
IE 0.000 0.002 0.002
SIZE 15.541 15.501 −0.041
LEV 0.388 0.362 −0.025⁎⁎⁎
MB 1.010 1.210 0.200⁎⁎⁎

Panel B: Regression analyses

Pre-act Post-act

Predicted sign Coeff. (t-Stat.) Coeff. (t-Stat.) Coeff. (t-Stat.) Coeff. (t-Stat.)

Intercept 0.336 (1.12) 0.284 (1.04) 0.004 (0.05) −0.012 (−0.11)

DUM −0.312 (−2.29)⁎⁎ −0.228 (−1.10) −0.001 (−0.01) −0.025 (−0.14)
R + 0.026 (0.09) 0.052 (0.18) −0.094 (−0.75) −0.022 (−0.14)
R × DUM + 0.127 (0.17) 0.089 (0.13) 0.457 (4.53)⁎⁎⁎ 0.233 (1.93)⁎⁎
LIEDUM 0.027 (1.37) 0.006 (0.63)
LIEDUM × DUM −0.041 (−0.88) 0.012 (0.60)
LIEDUM × R − −0.009 (−0.51) −0.031 (−2.05)⁎⁎
LIEDUM × R × DUM + 0.005 (0.06) 0.098 (2.13)⁎⁎
SIZE −0.017 (−0.81) −0.015 (−0.75) 0.005 (0.75) 0.005 (0.80)
SIZE × DUM 0.026 (2.40)⁎⁎ 0.022 (1.53) −0.001 (−0.12) 0.000 (−0.01)
SIZE × R + 0.015 (0.83) 0.014 (0.77) 0.010 (1.35)⁎ 0.007 (0.76)
SIZE × R × DUM −/+ −0.009 (−0.18) −0.007 (−0.14) −0.028 (−4.35)⁎⁎⁎ −0.017 (−2.63)⁎⁎⁎
LEV −0.091 (−0.76) −0.101 (−0.98) −0.071 (−2.69)⁎⁎⁎ −0.072 (−2.63)⁎⁎⁎
LEV × DUM −0.229 (−1.79)⁎ −0.215 (−1.69)⁎ 0.014 (0.35) 0.015 (0.42)
LEV × R − −0.453 (−12.43)⁎⁎⁎ −0.449 (−12.80)⁎⁎⁎ 0.016 (0.46) 0.020 (0.63)
LEV × R × DUM + 0.232 (1.55)⁎ 0.230 (1.77)⁎⁎ 0.211 (1.86)⁎⁎ 0.205 (1.89)⁎⁎
MB 0.021 (1.98)⁎⁎ 0.023 (2.16)⁎⁎ 0.024 (2.32)⁎⁎ 0.024 (2.43)⁎⁎
MB × DUM −0.002 (−0.21) −0.005 (−0.54) 0.001 (0.06) 0.001 (0.06)
MB × R − 0.017 (1.79) 0.018 (1.59) −0.007 (−1.28) −0.009 (−1.70)⁎⁎
MB × R × DUM −/+ −0.024 (−1.02) −0.026 (−1.31) 0.018 (0.76) 0.023 (0.87)
Adj R-square (%) 30.59 30.48 26.96 27.11

Variables are defined in the notes to Table 2. DUM is a dummy variable taking the value one when R is negative, or zero otherwise. Pre-act period is 2000–2002 and post-act period is
2003–2008. t-statistics in parentheses are based on Huber–White standard errors clustered by firm and year. ⁎⁎⁎, ⁎⁎, ⁎ indicate statistical significance at 1%, 5%, and 10% level respectively
for one-tailed t-tests of coefficients with predicted signs and two-tailed t-tests otherwise.

highly R&D-intensive firms. For those firms, the coefficient of LIEDUM ×
R (LIEDUM × R × DUM) remains significantly negative (positive) after
taking conference calls into account. These findings support the results
in Table 4 and substantiate the importance of the granting of patents
in certifying the credibility of voluntary disclosures made via conference
calls, especially among R&D-intensive firms.7
non-operating accruals scaled by average total assets across the whole
sample is −0.016. I regress non-operating accruals on LIEDUM and con-
trol for size, leverage and mark-to-book ratio. Untabulated results con-
firm that less innovative firms are associated with more earnings
conservatism, measured by industry-adjusted non-operating accruals
or three-year average non-operating accruals. Further, such evidence is

4.2.2. Corporate governance

Prior studies show that governance mechanisms affect both R&D ac- 45%
tivities (Baysinger, Kosnik, & Turk, 1991) and earnings conservatism 40%
(Ahmed and Duellman, 2007). I thus also include several corporate
35%
governance variables in Eq. (3) as a robustness check. The findings are
robust to board independence, measured by the proportion of indepen- 30%
dent directors on the board, the ownership structure (e.g., institutional 25%
shareholdings, and directors' and supervisors' shareholdings), and dual- 20%
Percentage difference
ity of CEO and President. 15% of the incremental
earnings response to
10%
4.2.3. Alternative measures of earnings conservatism bad news
To address concerns over the use of a return-based test, I use an 5%
accrual-based conservatism proxy. Non-operating accruals are calculat- 0%
ed as total accruals (i.e., net income less cash flow from operations) Pre-act period Post-act period
less operating accruals scaled by total assets, where operating accruals
are estimated by changes in non-cash current assets less changes in cur- Fig. 2. Differential bad news effect between less innovation-efficient and highly
innovation-efficient firms: comparison of pre- and post-act periods. Percentage difference
rent liabilities less depreciation and amortization. The mean value of
in the degree of earnings conservatism between high and low innovation efficiency is
calculated using the coefficients from Panel B of Table 5. The coefficient of R × DUM and
7
Untabulated results show that more innovation-efficient firms are more likely to hold the sum of the coefficients of R × DUM and LIEDUM × R × DUM capture the incremental
conference calls, which implies that the tendency to hold conference calls depends on a response of earnings to bad news for highly innovation-efficient and less innovation-
firm's ability to transform R&D into innovation outputs. efficient firms, respectively.
 A.L.-C. Chan / International Review of Financial Analysis 32 (2014) 132–142 141

Table 6
Asymmetric timeliness of earnings and conference calls.

Low R&D intensity High R&D intensity

Predicted sign Coeff. (t-Stat.) Coeff. (t-Stat.) Coeff. (t-Stat.) Coeff. (t-Stat.)

Intercept 0.103 (0.86) 0.096 (0.75) 0.068 (0.57) 0.037 (0.29)

DUM −0.202 (−1.21) −0.213 (−1.11) 0.100 (0.53) 0.115 (0.56)
R + 0.002 (0.01) 0.085 (0.57) 0.155 (0.97) 0.224 (1.35)⁎
R × DUM + 0.204 (0.73) 0.082 (0.29) 0.271 (0.86) 0.047 (0.16)
LIEDUM 0.003 (0.30) 0.017 (1.07)
LIEDUM × DUM 0.004 (0.24) −0.014 (−0.72)
LIEDUM × R − −0.042 (−2.38)⁎⁎⁎ −0.038 (−2.04)⁎⁎
LIEDUM × R × DUM + 0.059 (1.33)⁎ 0.092 (2.23)⁎⁎
CALLDUM 0.003 (0.18) 0.002 (0.09) 0.015 (0.79) 0.016 (0.81)
CALLDUM × DUM −0.013 (−0.38) −0.012 (−0.34) −0.002 (−0.05) −0.001 (−0.03)
CALLDUM × R + 0.057 (1.29)⁎ 0.061 (1.39)⁎ 0.012 (0.52) 0.010 (0.39)
CALLDUM × R × DUM − −0.092 (−2.02)⁎⁎ −0.094 (−2.11)⁎⁎ −0.016 (−0.30) −0.003 (−0.06)
SIZE −0.001 (−0.15) −0.001 (−0.11) −0.003 (−0.36) −0.001 (−0.16)
SIZE × DUM 0.012 (1.02) 0.012 (0.97) −0.003 (−0.26) −0.003 (−0.29)
SIZE × R + 0.009 (0.79) 0.006 (0.42) 0.000 (0.02) −0.004 (−0.41)
SIZE × R × DUM −/+ −0.014 (−0.96) −0.009 (−0.56) −0.019 (−0.92) −0.006 (−0.33)
LEV −0.141 (−3.11)⁎⁎⁎ −0.143 (−3.07)⁎⁎⁎ 0.032 (0.44) 0.029 (0.40)
LEV × DUM 0.041 (0.77) 0.045 (0.87) −0.169 (−3.34)⁎⁎⁎ −0.167 (−3.32)⁎⁎⁎
LEV × R − −0.121 (−0.85) −0.117 (−0.84) −0.160 (−0.93) −0.147 (−0.92)
LEV × R × DUM + 0.321 (1.85)⁎⁎ 0.317 (1.86)⁎⁎ 0.194 (0.87) 0.173 (0.79)
MB 0.043 (4.00)⁎⁎⁎ 0.044 (4.22)⁎⁎⁎ 0.021 (2.58)⁎⁎⁎ 0.022 (2.67)⁎⁎⁎
MB × DUM −0.005 (−0.47) −0.005 (−0.47) −0.001 (−0.07) −0.002 (−0.19)
MB × R − −0.030 (−2.39)⁎⁎⁎ −0.034 (−3.09)⁎⁎⁎ −0.009 (−1.50)⁎ −0.010 (−1.73)⁎⁎
MB × R × DUM −/+ 0.034 (1.25) 0.039 (1.49) 0.018 (0.80) 0.017 (0.78)
Adj R-square (%) 30.34 30.50 24.76 24.94

Variables are defined in the notes to Table 2. DUM is a dummy variable taking the value one when R is negative, or zero otherwise. CALLDUM is a dummy variable equal to one if
firms hold at least one conference call in year t, zero otherwise. High (low) R&D intensity consists of companies with R&D expenditures over net sales higher (lower) than the
yearly median. t-statistics in parentheses are based on Huber–White standard errors clustered by firm and year. ⁎⁎⁎, ⁎⁎, ⁎ indicate statistical significance at 1%, 5%, and 10% level
respectively for one-tailed t-tests of coefficients with predicted signs and two-tailed t-tests otherwise.

more pronounced in the post-act period, when the shareholder class ac-
tion mechanism is effective, which confirms my conclusions.
4.2.4. Economic consequences of innovation efficiency
I further analyze the economic consequences of innovation efficien-
cy. Specifically, I consider the extent to which different capabilities, in
terms of converting new ideas into granted patents, influence market li-
quidity, as measured by the bid–ask spread, which is also commonly
used as a proxy for information asymmetry. I estimate the following re-
gression to investigate the extent to which lower innovation efficiency
is associated with information asymmetry and its impact on market
liquidity:
logðBidask spreadit Þ ¼ γ0 þ γ1 LIEDUMit þ γ 2 logðTURNOVERit−1 Þ
þγ 3 logðPRICEit−1 Þ þ γ4 logðRETVOLi
The daily quoted spread is calculated as the difference between the
closing bid and ask prices divided by the midpoint. Bid_ask spread is
the median daily spread over the fiscal year. The mean and median
values of the bid–ask spread are 0.54% and 0.46%. Following Leuz and
Verrecchia (2000), I control for the previous year's share price
(PRICE), share turnover (TURNOVER) measured as annual NT$ trading
volume divided by the market value of outstanding equity, and return
volatility (RETVOL), calculated as the annual standard deviation of
monthly share returns. I take the natural log of the bid–ask spread
and the control variables. Table 7 shows that the coefficient of LIEDUM
is significantly positive, suggesting that less innovation-efficient firms
see a significant decrease in market liquidity relative to more
innovation-efficient firms. Untabulated results indicate that such evi-
dence exists in both highly and less R&D-intensive firms. The model
controlling for industry and year fixed effects explains 41.95% of the
variation in the bid-ask spreads. Regarding the control variables, the co-
t−1 Þ þ εit : efficients of share turnover and market price are both negative and sig-
ð4Þ nificant, and the coefficient of return volatility is significantly positive,
supporting the findings in prior studies. These results also confirm the
conjecture that innovation efficiency is associated with information
asymmetry.

Table 7
Economic consequences of innovation efficiency.

Dependent variable: log(bid_ask spread)

Predicted sign Coeff. (t-Stat.) Coeff. (t-Stat.)

Intercept −0.984 (−11.44)⁎⁎⁎ −0.611 (−6.20)⁎⁎⁎

LIEDUM + 0.108 (4.44)⁎⁎⁎ 0.123 (5.26)⁎⁎⁎
log(TURNOVER) − −0.141 (−8.49)⁎⁎⁎ −0.180 (−9.34)⁎⁎⁎
log(PRICE) − −0.232 (−12.95)⁎⁎⁎ −0.214 (−9.84)⁎⁎⁎
log(RETVOL) + 0.344 (12.22)⁎⁎⁎ 0.245 (7.40)⁎⁎⁎
Industry fixed effect YES
Year fixed effect YES
Adj R-square (%) 27.37 41.95

log(bid_ask spread) is calculated as the natural log of median daily spread over the fiscal year; LIEDUM is a dummy variable equal to one if innovation efficiency (IE) is less than the yearly
median or is equal to zero, zero otherwise; log(TURNOVER) is natural log of share turnover calculated as annual NT$ trading volume divided by market value of outstanding equity;
log(PRICE) natural log of share price; and log(RETVOL) is natural log of return volatility calculated as the standard deviation of monthly share returns. The control variables are lagged
by one year. After excluding missing values, the sample includes 3342 firm-year observations. t-statistics in parentheses are based on Huber–White standard errors clustered by firm
and year. ⁎⁎⁎, ⁎⁎, ⁎ indicate statistical significance at 1%, 5%, and 10% level respectively for one-tailed t-tests of coefficients with predicted signs and two-tailed t-tests otherwise.
142 A.L.-C. Chan / International Review of Financial Analysis 32 (2014) 132–142
5. Conclusions
Earnings conservatism facilitates efficient contracting and serves as a
natural defense against litigation. R&D activities cause intangible assets
to grow but are also likely to induce asymmetric information between
the management and the outside investors and lead to potential share-
holder litigation. The granting of patents conveys valuable information
about successful scientific and technological developments and may
mitigate R&D-related information asymmetry and reduce expected liti-
gation risks.
The empirical evidence of this study implies that the providers and
users of financial statements recognize the verifiability of accounting in-
formation, and further confirms that conservatism is an equilibrium
mechanism for mitigating the value reductions and agency costs due
to information asymmetry. The findings indicate that earnings conser-
vatism in response to R&D-related asymmetric information and litiga-
tion risk exists mainly in less innovation-efficient firms. This evidence
is more pronounced among highly R&D-intensive firms. Despite the
fact that R&D outlays are generally expensed, which preempts the
level of conditional conservatism through lower recorded assets, my
findings imply that less innovation-efficient firms recognize losses in a
timely manner (e.g., through larger asset write-offs on recorded assets,
or restructuring charges) as a defense against litigation. Additional anal-
yses controlling for conference calls and corporate governance mecha-
nisms confirm the main findings. In parallel, I provide evidence
showing that asymmetric income timeliness exists in the period since
the shareholder class action mechanism was established and that such
evidence is more pronounced for less innovation-efficient firms. These
results may be generalizable to other emerging markets where the in-
formation environment tends to be less transparent. Overall, the find-
ings of this study further substantiate the importance of earnings
conservatism in facilitating efficient contracting and risk reduction, par-
ticularly in an emerging market characterized by high-technology but
with weaker investor protection.
Acknowledgments
I would like to thank Edward Lee, Cheng-Lung Chin, Steve Lin, and
participants at the European Accounting Association 2010 annual meet-
ing and American Accounting Association 2010 annual congress for
their valuable comments. I gratefully acknowledge the financial support
provided by the National Science Council of Taiwan.
References
Aboody, D., & Lev, B. (2000). Information asymmetry, R&D, and insider gains. Journal of
Finance, 55(6), 2747–2766.
Ahmed, A., Billings, B., Morton, R., & Stanford-Harris, M. (2002). The role of accounting
conservatism in mitigating bondholder–shareholder conflicts over dividend policy
and in reducing debt costs. The Accounting Review, 77(4), 867–890.
Ahmed, A., & Duellman, S. (2007). Accounting conservatism and board of director charac-
teristics: An empirical analysis. Journal of Accounting and Economics, 43(2–3),
411–437.
Amir, E., Guan, Y., & Livne, G. (2007). The association of R&D and capital expenditures
with subsequent earnings variability. Journal of Business Finance and Accounting,
34(1/2), 222–246.
Ball, R., Kothari, S. P., & Nikolaev, V. (2013). Econometrics of the Basu asymmetric timeli-
ness coefficient and accounting conservatism. Working paper (Available at: http://
papers.ssrn.com/sol3/papers.cfm?abstract_id=999710).
Barth, M. E., & Kasznik, R. (1999). Share repurchase and intangible assets. Journal of
Accounting and Economics, 28(2), 211–241.
Barth, M. E., Kasznik, R., & McNichols, M. F. (2001). Analyst coverage and intangible assets.
Journal of Accounting Research, 39(1), 1–34.
Basu, S. (1997). The conservatism principle and the asymmetric timeliness of earnings.
Journal of Accounting and Economics, 24(1), 3–37.
Baysinger, B.D., Kosnik, R. D., & Turk, T. A. (1991). Effects of board and ownership struc-
ture on corporate R&D strategy. Academy of Management Journal, 34(1), 205–214.
Beaver, W., & Ryan, S. (2005). Conditional and unconditional conservatism: concepts and
modeling. Review of Accounting Studies, 10(2/3), 269–309.
Beaver, W. H., & Ryan, S. G. (2009). Risky debt, mixed-attribute accounting, and the iden-
tification of conditional conservatism. Working paper (Available at: http://papers.
ssrn.com/sol3/papers.cfm?abstract_id=1477528).
Brown, S., Hillegeist, S., & Lo, K. (2004). Conference calls and information asymmetry.
Journal of Accounting and Economics, 37(3), 343–366.
Chan, L., Lakonishok, J., & Sougiannis, T. (2001). The stock market valuation of research
and development expenditures. Journal of Finance, 56(6), 2341–2456.
Chandra, U. (2011). Income conservatism in the U.S. technology sector. Accounting
Horizons, 25(2), 285–314.
Chiao, C., Wang, Z., & Lai, H. (2009). Order submission behaviors and opening price be-
haviors: Evidence from an emerging market. Review of Quantitative Finance and
Accounting, 33(2), 253–278.
Chin, C. L., Lee, P., Wang, P. W., & Kleinman, G. (2007). Does innovation matter to the con-
ference calls? IEEE Transactions on Engineering Management, 54(4), 699–715.
Dietrich, D., Muller, K., & Riedl, E. (2007). Asymmetric timeliness tests of accounting con-
servatism. Review of Accounting Studies, 12(1), 95–124.
Einhorn, B. (2005,May 16). Why Taiwan matters: The global economy couldn't function
without it. Business Week, 76–81.
Ernst, D. (2010). Upgrading through innovation in a small network economy: insights from
Taiwan's IT industry. Economics of Innovation and New Technology, 19(4), 295–324.
Fan, P. H., & Wong, T. J. (2002). Corporate ownership structure and the informativeness of
accounting earnings in East Asia. Journal of Accounting and Economics, 33(3), 401–425.
Francis, J., Philbrick, D., & Schipper, K. (1994). Shareholder litigation and corporate disclo-
sures. Journal of Accounting Research, 32(2), 137–164.
Garcia Lara, J. M., Garcia Osma, B., & Mora, A. (2005). The effect of earnings management on
the asymmetric timeliness of earnings. Journal of Business Finance and Accounting,
32(3/4), 691–726.
Garcia Lara, J. M., Garcia Osma, B., & Penalva, F. (2011). Conditional conservatism and cost
of capital. Review of Accounting Studies, 16(2), 247–271.
Gelb, D. S. (2002). Intangible assets and firms' disclosures: An empirical investigation.
Journal of Business Finance and Accounting, 29(3/4), 457–476.
Givoly, D., Hayn, C., & Natarajan, A. (2007). Measuring reporting conservatism. The
Accounting Review, 82(1), 65–106.
Hirshleifer, D., Hsu, P. H., & Li, D. (2013). Innovation efficiency and stock returns. Journal of
Financial Economics, 107(3), 632–654.
Johnson, M., Kasznik, R., & Nelson, K. (2001). The impact of securities litigation reform on
the disclosure of forward-looking information by high technology firms. Journal of
Accounting Research, 39(2), 297–327.
Kothari, S. P., Laguerre, T., & Leone, A. (2002). Capitalization versus expensing: Evidence
on the uncertainty of future earnings from capital expenditures versus R&D outlay.
Review of Accounting Studies, 7(4), 355–382.
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. W. (1998). Law and finance.
Journal of Political Economy, 106(6), 1113–1155.
LaFond, R., & Watts, R. L. (2008). The information role of conservative financial state-
ments. The Accounting Review, 83(2), 447–478.
Lee, K. (2007). Corporate voluntary disclosure and the separation of cash flow rights from
control rights. Review of Quantitative Finance and Accounting, 28(4), 393–416.
Leuz, C., & Verrecchia, R. E. (2000). The economic consequences of increased disclosure.
Journal of Accounting Research, 38, 91–124 (Suppl.).
Levitas, E., & Mcfadyen, M.A. (2009). Managing liquidity in research-intensive firms: sig-
naling and cash flow effects of patents and alliance activities. Strategic Management
Journal, 30(6), 659–678.
Liu, S. J. (2001). Competing in the knowledge game: Intellectual property rights. In C. Y.
Chang, & P. L. Yu (Eds.), Made by Taiwan: Booming in the information technology era
(pp. 213–234). New Jersey: World Scientific.
Lobo, G. J., & Zhou, J. (2006). Did conservatism in financial reporting increase after the
Sarbanes–Oxley Act? Initial evidence. Accounting Horizons, 20(1), 57–73.
Ozkan, A., & Ozkan, N. (2004). Corporate cash holdings: An empirical investigation of UK
companies. Journal of Banking and Finance, 28(9), 2103–2134.
Pandit, S., Wasley, C. E., & Zach, T. (2011). The effect of R&D inputs and outputs on the re-
lation between the uncertainty of future operating performance and R&D expendi-
tures. Journal of Accounting, Auditing and Finance, 26(1), 121–144.
Roychowdhury, S., & Watts, R. L. (2007). Asymmetric timeliness of earnings, market-
to-book and conservatism in financial reporting. Journal of Accounting and
Economics, 44(1/2), 2–31.
Skinner, D. J. (1997). Earnings disclosures and stockholder lawsuits. Journal of Accounting
and Economics, 23(3), 249–282.
Tasker, S. (1998). Bridging the information gap: Quarterly conference calls as a
medium for voluntary disclosure. Review of Accounting Studies, 3(1/2),
137–167.
Tasker, S. (1998). Technology company conference calls: A small sample study. Journal of
Financial Statement Analysis, 4(1), 6–14.
Watts, R. L. (2003). Conservatism in accounting Part I: Explanations and implications.
Accounting Horizons, 17(3), 207–221.
Watts, R. L. (2003). Conservatism in accounting part II: Evidence and research opportunities.
Accounting Horizons, 17(4), 287–301.
Yeo, V. (2008). Taiwan is world's No.2 in IT competitiveness, ZDNet Asia, 16 September 2008.
Available at: http://www.zdnetasia.com/news/business/0,39044229,62046168,00.htm