R

1
RD-Intensity, Mispricing, and Stock Returns in
Taiwan Stock Market
2
Literatures review properties of high
RD-intensive firms

• Barron et al (2001) and Barth, Kasznink, and
  Mcnichols (1999) observe that analysts forecast
  error is negatively associated with a firm level
  of intangible assets including RD and others.
• Aboody and Lev (2000) suggest that RD
  expenditures generate information asymmetry and
  insider gain
• Kothari et al. (2002) report that the earnings
  volatility associated with RD expenditures is
  three times larger than that associated with
  tangible investment in property, plant and
  equipment
• Barron, Byard, Kile, and Riedl (2001) find that
  lower levels of analyst consensus are associated
  with the RD expenditures of high-technology
  manufacturing companies.

3
Motivations

• The high RD intensity firms are hard to be
  correctly evaluated.
• RD are increasingly importantly under Taiwans
  economic development.
• Given the increasing importance of RD, are their
  valuations for RD different from those in the
  U.S. stock market recorded in previous
  literature?

4
The three essential issues

• Does the stock market correctly value firms RD
  intensity?
• Is the valuation meaningfully attributable to the
  fundamental dimensions of return and associated
  risks or characteristics of firms? Does
  mispricing associated with the RD intensity
  exist?
• Is the valuation, if existing, subject to
  numerous conditions, such as seasonality, market
  conditions, industrial structures, and even the
  development process of Taiwans Economics?

5
Recent evidence

• Chan, Lakonishok and Sougiannis (2001) find
  evidence that returns on firms involved in RD
  are equivalent to the returns on firms without
  RD.
• Many researchers find a positive relationship
  between RD expenditures and subsequent stock
  returns (e.g., Hirschey and Weygandt, 1985
  Cockburn and Griliches, 1988 Bublitz and
  Ettredge, 1989 Lev and Sougiannis, 1996)
• A recent study by Titman, Wei, and Xei (2001)
  provide evidence that firms with high capital
  expenditures yield lower benchmark-adjusted stock
  returns, a result primarily driven by the
  over-investing problem.

6
Samples and Methodology

• Data
• All the data are obtained from the Taiwan
  Economic Journal (TEJ).
• The sample period covers from July 1988 to June
  2002. There are totally 168 observations for each
  stock.
• Following Chan, Lakonishok and Sougiannis (2001),
  we use RD expenditures relative to sales
  (RD/Sales) and RD expenditures relative to
  market value (RD/ME) as the two measures of RD
  intensity.

7
The economic interpretations of two RD
intensity measures

• The first measure (RD/Sales) is used as an
  indicator of how much money, proportional to
  their sales, firms spend in RD activities or how
  enthusiastically firms devotes their resources to
  RD activities.
• The second measure of RD intensity (RD/ME),
  unlike the first measure, not only depends on
  firms spending in RD, but also explicitly
  accommodates the market price or the valuation of
  investors. The essence of this measure plausibly
  keeps up with many financial indicators widely
  used in financial research, such as the ratios of
  BM, EP, and cash flows to price that embed the
  stock price information.

8
Methodology

• Follow Chan and Chen (1991) to derive the
  size-adjusted portfolio returns.
• Follow Carhart (1997) to derive the abnormal
  returns.
• The four-factor model is explicitly defined as
• where ap is the intercept term of the regression
  or so-called the abnormal return of portfolio p
  after controlling for the risks mimicked by MOM,
  SMB, and HML.

9
Preliminary results (Table 1)

• There is a remarkable upward trend in the
  aggregate RD intensity.
• Compared to those in other countries, RD
  intensity in Taiwan was low but grew strongly.
• Firms in the Electronics industry are
  enthusiastic in RD activities.
• The two measures of RD intensity, RD/Sales and
  RD/ME, may be very different in essence.
• Given the high correlation coefficient between
  RD/ME and BM as well as size, we expect that
  RD/ME is able to predict future returns than
  RD/Sales.

10
Table 1 Descriptive Statistics of RD intensity
11
Table 1 Descriptive Statistics of RD intensity
(cont.)
12
The findings based on RD/Sales (Table 2)

• The returns before formation generally increase
  with respect to the RD/Sales .
• The returns after formation monotonically are
  increasing with RD/Sales.
• The zero-cost portfolio returns reported in the
  last column in years 1, 2, and 3 after formation
  get enlarged over time and are 0.77, 0.99, and
  0.73, with t-value 1.49, 1.87, and 1.33,
  respectively.
• These observations indicate that firms having
  invested more on RD are not only the prior
  winners but also the future winners, and the
  spread between the portfolio returns with the
  highest and lowest RD intensity widens.
• There is a pronounced difference between the
  returns on stocks with highest RD/Sales and
  stocks without RD expenditures.

13
The findings of portfolios based on RD/Sales
(cont.)

• The results above are inconsistent with
  Lakonishok, Shleifer, and Vishny (1994) and Chan,
  Lakonishok and Sougiannis (2001).
• The evidence suggests that firms that the return
  spread between firms with different levels of RD
  intensity is not caused by size effect.
• Investors undervalue the RD-intensive firms and
  overvalue the non-RD-intensive firms.
• These preliminary results seem to support
  mispricing hypothesis that the market could not
  correctly values the future benefits from the RD
  spending.

14
Table 2 Monthly returns () and characteristics
of portfolios based on RD/Sales

15
Table 2 Monthly returns () and characteristics
of portfolios based on RD/Sales (cont.)
16
The findings based on RD/ME (Table 3)

• The relationship between returns and RD/ME are
  much stronger than between returns and RD/SALES.
• The mispricing of stocks not only exists but also
  is persistent.
• Size effect cannot explain the anomaly.
• It is obvious that mispricing phenomenon is even
  stronger using RD/ME as the measure of the RD
  intensity than using RD/Sales.

17
Table 3 Monthly returns () and characteristics
of portfolios based on RD/ME
18
Table 3 Monthly returns () and characteristics
of portfolios based on RD/ME (cont.)
19
Abnormal returns (Table 4)

• The four-factor model works rather well for
  RD/Sales-based portfolio returns, but not for
  RD/ME-based portfolio returns.
• All results provide further evidence that
  mispricing does exist and persist, especially
  when applying RD/ME as the measure of RD
  intensity.
• The relatively low adjusted R2s of the
  regressions on the RD-intensive portfolio
  returns support that the uncertainty associated
  with the RD activities cannot be captured by the
  four well-known factors.

20
Table 4 Abnormal returns
21
Explanatory power of the RD intensity in stock
returns

• To investigate the relationship between stock
  returns and competing characteristics such as
  firm size, BM, RD/Sales, and RD/ME, we employ
  the following empirical model the estimate the
  stock returns

22
The results (Table 5)

• Consistent with the results of Chan et al. (1998)
  and Chui and Wei (1997), no evidence of the BM
  effect exists in Taiwan stock market.
• The explanatory power of the RD intensity
  measured by RD/ME is stronger than that measured
  by RD/Sales.
• The result suggests that, in Taiwan stock market,
  the RD intensity may be more informative than
  firm size and BM.

23
Table 5 Average coefficients of the regression
results of stock returns on ln(ME), ln(BM),
ln(RD/SALES), and ln(RD/ME)
24
Figure 1 Cumulative returns on the market index,
Electronic index, and zero-cost portfolios from
July 1988 to June 2002
25
Table 8 Returns and characteristics of
portfolios based on RD/ME in the Electronic
industry and non-Electronic industries
26
Table 9 Average coefficients of regressions of
stock returns on ln(ME), ln(BM), and ln(RD/ME)
for all firms, firms in the Electronic industry,
and firms in non-Electronic industries before and
after 07/1996