You are on page 1of 35

Licensing Out as a Real Activity to engage in Myopic Management

Goretti Cabaleiro Cervio


Universidad Carlos III de Madrid
gcabalei@emp.uc3m.es

(Early draft: Please Do Not Distribute Without Permission)

This paper empirically tests that managers, under the pressure to attain analysts
forecast, can license out their intellectual property as a mean to increase short-term
earnings. However, as licensing contracts always imply a trade-off, I expect that these
companies will increase their benefits in the short term (Revenue Effect) but they will
also harm their market share in the long run (Dissipation Effect). Results confirm that 1)
companies are more likely to license out their intellectual property when they were not
able to achieve analysts forecasts and 2) companies that have increased the number of
licensing out contracts with respect to the previous year present a decreasing market
shares trend in the following two years, 3) this decreasing market shares trend is
stronger for companies that license out their technology and were not able to achieve
analysts forecast than for companies that overcome the earnings threshold.


Key words: Licensing, Intellectual Property, Real Activities, Myopic Management.


















2

1. INTRODUCTION.

Over the last two decades licensing agreements have experienced an unprecedented
growth and its management had become a core competence of high-tech companies
(Zuniga & Guellec, 2008, Lichtenthaler and Einst, 2009; Kamiyama et al. 2006). In fact,
in order to facilitate knowledge transfer, companies are establishing their own licensing
department and/or publishing in their web their technology available to license
1
.
The main reason that explains this increasing trend is the revenue that licensing
generates. In a survey conducted by Zuniga & Guellec (2009) 51% of the European
companies and 53.6% of Japanese companies recognized that their main motivation to
license-out their technology in the previous three years was the licensing revenue.
However, licensing also has a negative counterpart. That is, even though companies
increase their benefits by the licensing revenues (net of transaction costs), they also
could reduce their market share or their price-cost margin because the additional
competition in the product market (Arora & Fosfuri, 2003; Fosfuri, 2006). Therefore,
licensing decision has to be taken with caution: balancing the short-term earnings
against the possible negative long-term consequences. Nevertheless, evidence shows
that companies sometimes put at stake their competitive position underestimating this
long-term effect
2
.
Similarly, over the last decades financial analysts are increasingly influencing
companies strategies. The consequences of missing their forecast, even by a small
quantity, have been disproportionate
3
. It has put extra pressure on managers and has
created incentives to manipulate current earnings. In a survey conducted by Graham et
al. (2005) CFOs admit that they put much attention on meeting the earnings thresholds.

1
See, for example:
a. Dow Chemical: http://www.dow.com/licensing/
b. Kimberly Clark: http://www.merck.com/licensing/home.html
c. Merck & Co: http://www.merck.com/licensing/home.html a. Dow Chemical: http://www.dow.com/licensing/
b. Kimberly Clark: http://www.merck.com/licensing/home.html
c. Merck & Co: http://www.merck.com/licensing/home.html

2
For instance, Hitachi before 2003 was one of the companies that license-out more technology. In fact, in 2002 the company
presented licensing revenues of JPY43 billion. However, as a result of its aggressive licensing strategy, licensees in China and
Korea rapidly improved their technology and threatened its competitive advantage. In 2003 Hitachi had to restrict its licensing
policy if it did not want to be overcome by licensees (Kamiyama et al. 2006).

S
For instance, Oracle on December 1997 had seen how its stock price declined by 29% as a consequence of not achieving the
analysts forecast by 0.04$ (even this result was 4% above EPS for the same quarter in last year). (Skiner & Sloan, 2002). Also
Procter & Gamble has seen how its stock price was reduced by 30% when they warned that the company would not beat the
analysts forecast in the first quarter of 2000. The same warning before the second quarter of that year generated an additional
reduction on the stock price of 10% and the CFOs dismissal (Duncan, 2001).
S
They recognized that they are willing to inflate current earnings in order to achieve
them and that they prefer to manipulate earnings using real activities rather than using
accruals.
Motivated by these two apparently unrelated trends this paper wants to shed light on the
relationship between the companys financial situation and its licensing strategy. In
particular, my objective is to examine 1) whether managers license-out their technology
to inflate current earnings and 2) what are the long-term consequences for the
companies that took the licensing decision under pressure to meet analysts forecast. I
argue that managers that feel the pressure to attain the analysts forecast will have
incentives to inflate current earnings and, thus, engage in myopic management. Since
myopic managers put more emphasis on the short term than in the long run, I expect
that, at the time to make the licensing decision, managers will overestimate the revenue
effect (short-term effect) while underestimate the dissipation effect (long-term effect).
An example of the overestimation of the revenue effect is the declaration of Daniel M.
McGavock, managing director of intellectual property consulting firm Intercap: On
one hand, you dont want to abandon your patents ability to exclude competitors from
your market. But, on the other hand, you could be talking about hundreds of millions of
dollars in new revenue from strategic licensing, not to mention a host of strategic
benefits(Kline, 2003).
This distortion will generate that, under pressure for inflating short-term earnings,
managers license out more technology than the optimal and/or license it under not
appropriate situations and/or license out inappropriate technology and this, in turn, will
imply negative long-term consequences.
I test my hypotheses using a panel of 107 U.S. high-tech companies during 1998-2009
(1,281 observations). Licensing data is one of the strength points of this paper. It has
been collected from four different sources: Prompt Database, Google, Highbeam
Research and SDC Platinum. This extensive search has delivered 1,729 licensing
agreements
4
of which 840 correspond to license in, 716 to license out and 173 to
cross licensing.
To test the first hypothesis I followed the same procedure as in Bushee (1998). I
estimated a logit model explaining the probability of increasing the number of licensing
out contracts from period t-1 to period t using as main independent variable a dummy
that captures whether the company has or has not achieved analysts forecasts in period

4
0nly 1S4 licensing agieements weie collecteu fiom SBC Platinum uuiing the peiiou of stuuy foi this sample.
4
t-1. Results confirm that companies are more likely to license out their intellectual
property when they are not able to achieve analysts forecasts.
To provide evidence about the second hypothesis I compared the market shares
evolution of a) companies that were not able to achieve analysts forecast in the
previous year with b) companies that were not able to achieve analysts forecast and
have increased the number of licensing out contracts with respect to the previous year
and with c) companies that have increased the number of licensing out contracts and
that were able to achieve analysts forecast in the previous year. Results show that 1)
companies that have increased the number of licensing out contracts with respect to the
previous year present a decreasing market shares trend in the following two years, 2)
this decreasing market shares trend is stronger for companies that license out their
technology and were not able to achieve analysts forecast than for companies that
overcome the earnings threshold.
This paper contributes to innovation literature that examines the strategic drivers of
technology licensing. Prior research has identified economic as well strategic
motivations for licensing (Shepard, 1987; Katz & Shapiro,1985; Gallini, 1984;
Rockett,1990; Lichtenthaler, 2007). I extend this research showing that the pressure to
achieve analysts forecasts is also a potential determinant to license out technology.
I also contribute to the Myopic Management Theory. I proposed a new real activity to
inflate current earnings. Prior research had identified several activities that managers
use to engage in myopic management (Aaker ,1991; Roychowdhury, 2006; Moorman &
Spencer, 2008; Chapman & Steenburgh, 2009). I extended this literature showing that
licensing-out technology can be used as a real activity that increases current earnings at
the expense of reducing market share in the incoming years.
The remainder of this paper is organized as follows. Section 2 presents the theoretical
background regarding Myopic Management and Licensing. Section 3 develops the
hypotheses to be tested. Section 4 describes the methodological analysis. Section 5
reports and discusses the results. Section 6 concludes.






S
2. THEORICAL BACKGROUND

2.1. Licensing Theory


Over the last decades companies have moved from protecting aggressively their
knowledge to license it (Vishwasrao, 2004). Nowadays, licensing is the most important
way of technology transfer and companies are increasingly working on developing an
efficient corporate structure to facilitate knowledge transfer (Arora et al. 2011)
The most important motivation for licensing out technology is the revenue it generates.
That is, the present value of the fixed fee and/or the royalties that the licensee has to pay
to the licensor. Surveys conducted by Gambardella (2005), Robbins (2008) and Zuniga
& Guellec (2009) corroborate that earnings revenue is, by far, the main motivation for
companies to license-out technology.
However, in order to really generate profits from licensing, three important points
should be considered before taking the decision. First, managers should know what are
the downstream assets needed to exploit the innovation. For companies that do not own
the marketing and distributions capabilities would be more difficult to benefit from
their innovations. Hence, license out their technology would be a good idea (Teece,
1986). Second, managers should be conscious of what are the transaction costs related
with the licensing contract. If the transaction costs generated with the buying of
downstream assets are greater than the transaction costs related with the licensing of
technology the best strategy would be to license out technology (Arora & Fosfuri,
2003).
5
Finally, managers should pay attention to the extent of the profit-dissipation
effect. This effect refers to the reduction in the licensors profit as a consequence of the
additional competition in the product market or of an existing firm becomes more
aggressive. Therefore, a company should license its technology only if the revenue
effect (net of transaction costs) is greater than the profit-dissipation effect.
In order to limit the extent of the latter effect researchers found that it is better to license
out technology when the strength of the patent protection is high (Cohen et al., 2000;
Arora and Ceccagnoli, 2006), when intellectual property refers to general technologies

5 In general, these contracts are distinguish by: 1) higher search costs resulting from looking for suitable licensees and/or licensors,
2) the existence of information asymmetries between the parties and the consequent incomplete contracts, 3) the bargaining
difficulties due to the risk of giving information before signing the contract and 4) the lack of an established mechanism for pricing
technologies.

6
(Bresnahan and Gambardella, 1998), when intellectual knowledge are based on
scientific knowledge (Arora & Gambardella, 1994), when the market share of the
company is small (Fosfuri, 2006), when intellectual property are related to non-core
technologies and to geographically separated markets (Patel and Pavitt, 1997;
Granstrand et al. 1997) and when the competition in the product market is high (Arora
& Fosfuri, 2003).
Moreover, prior research has also shown that in some situations strategic incentives are
more important than the mere licensing revenues. First, Gallini (1984) demonstrates that
licensing could be used to guarantee technology leadership. If the established company
license out its technology to potential entrants, it will reduce their incentives to develop
its own, maybe better technology. Second, Katz & Shapiro (1985) show that licensing
could be used to achieve a collusive agreement. If the licensor choose an appropriate
royalty rate for the licensee, it will increase price and stimulate the formation of a cartel
(Shapiro, 2001). Third, Farrel & Gallini (1988) demonstrated that licensing could be
used to create a second source mechanism and, thus, to encourage purchase. That is, if
the new technology is complex and is only produced by one company, potential buyers
will be reluctant to buy it for fear that this company cannot meet the demand for several
circumstances. Fourth, Rockett (1990) found that through licensing a company could
choose its competitors. If the established firm licenses its technology to a weak rival, it
will crowd the market and block the entry by a stronger competitor. Fifth, Lichtenthaler
(2007) shows that if a established company license out its technology to the most of the
competitors could create market standards. Finally, Lichtenthaler (2007) demonstrated
that through licensing and cross licensing agreements companies have more freedom to
operate: they do not block each other and avoid going to court or ceasing production.










7
2.2. Myopic Management.

Effective management requires to be focused on the long-term and to take projects that
generate the highest expected net present value (Mizik, 2009). However, the importance
that the market gives to current earnings forces managers to put more emphasis on
strategies that result in immediate pay-offs (Dechow, 1994; Degeorge et al. 1999).
Usually managers compensation as well their evaluation are based on the companys
stock price (Mizik, 2009) and this, in turn, depends on the achievement of three
earnings benchmarks: zero earnings, the prior comparable periods earnings and the
analysts forecasts (Degeorge et al. 1999). The pressure to meet these thresholds gives
managers incentives to manipulate results and, thus, inflate current earnings. In fact,
evidence has shown that this pressure has already changed the distribution of earnings
reported: only few firms report losses while so many report small profits (Dechow,
Richardson & Tuna, 2003). In a survey conducted by Graham et al. (2005) financial
executives declared that in order to avoid negative surprises they are willing to inflate
current earnings and that they prefer to do so by using real activities manipulation rather
than using accruals. Even though the objective under both strategies is to inflate current
earnings their implications and costs differ greatly. Namely, manipulating discretionary
accruals supposes to change the time at which earnings are recognized, not to modify
neither the quantity nor the temporal flow of economic profits. However, real activities
manipulation always implies engaging in Myopic Management. Basically, it consists on
using some activities to inflate current earnings at the expense of sacrificing the long-
term firm value. Then, for attaining the same objective, managers prefer to use the
strategy that implies worse consequences.
Prior research on Myopic Management has mainly concentrated on the reduction of
R&D investment as well on the factors that influences this practice. In particular,
evidence has shown that managers reduce R&D expenditures when they cannot ensure
positive earnings for the next year (Baber et al.,1991), when the time of retirement is
close (Dechow and Sloan, 1991), when the presence of institutional ownership is not
high (Bushee, 1998) and when managers have to repurchase stock to avoid EPS
dilution.
However, managers are also using other activities to inflate current earnings. In
particular, Aaker (1991) found that companies reduce marketings expenditures used to
enhance brand value while increasing the ones oriented to inflate temporarily the
8
results. Bartov (1993) and Herrmann et al. (2003) show that managers increase the sell
of fixed assets; Roychowdhury (2006) demonstrated that managers use price discounts
and zero financing strategies, overproduce and reduce discretionary expenses and
Moorman and Spencer (2008) show that managers delay the introduction of innovations
in order to not reduce earnings.
Although the evidence of myopic management is more established, only few studies
have quantified the financial future impact of engaging in this practice. Pauwels et al.
(2004) demonstrated that sales promotions imply negative long-term effects on !rm
value. Gunny (2005) found that myopic practices are associated with lower ROA in the
subsequent year. Mizik and Jacobson (2007) found that two years after reducing the
marketing expenditure the company began to have negative earnings and that in the fifth
year the market value of the company was reduced by 25%. In the same vein, Mizik
(2009) found that companies that have reduced the marketing expenditure present
greater negative abnormal returns in the future than companies that did not. Chapman &
Steenburgh (2009) show that companies can use marketing to increase quarterly net
income by up to 5% but that this strategy will be reflected in a 7,5% reduction of the
next period quarterly net income.

3. HYPOTHESIS DEVELOPMENT.

Financial analysts are increasingly influencing companies strategies. The severe
consequences of missing their forecast, even by a small quantity, have caused that
managers modify the normal business of their companies (Skinner & Sloan, 2002). In
order to avoid stock prices reduction, to maintain their job and to enhance their
reputation they are even willing to engage in not efficient projects that put at stake the
long-term firm performance (Degeorge et al., 1999). No doubt, licensing out could be
one of these inefficient projects because the trade-off that it implies. Although the main
motivation to license is the revenue it generates, companies only will take benefits from
it if the profit-dissipation effect is lower than the revenue effect. However, if managers
are facing pressure to beat analysts forecast they will put more emphasis on the short
term and, thus, they will have incentives to inflate current earnings even at the expense
of long-term firm performance.
Stein (1989) argues that for managers interested in manipulate short-term earnings the
easier is to reduce intangible assets expenditures because 1) they are not separately
9
recorded in the balance sheet and 2) they are not directly related to production. These
two characteristics also can be applied to licensing agreements. Usually, these contracts
are private and confidential and accounting rules do not force companies to recognize
licensing revenues as a separate item in corporate reports. Therefore, if a company
receives licensing revenues external observers only will perceive an increase in earnings
but they cannot know if the reported earnings are a valid proxy of the firm future
performance or if these earnings are coming at the expense of future profits. On the
other hand, licensing out intellectual property does not affect to production in the short
term. Even if companies license out their core technology to competitors, it will take
some time after they observe the reduction in the market share. This inability to
immediately identify the licensing out practice provides managers with the
opportunity to inflate current earnings and to benefit from this for some time.
Therefore, I expect that companies that are under pressure to beat the analysts
forecasts, will overestimate the revenue effect and wont take the efficient licensing
decision. This distortion will lead to license out without any control and without taking
into account the negative long-term consequences.

H1. Companies are more likely to license out their intellectual property when they were
not able to achieve analyst forecasts in previous year.

While many researchers have studied the motivations to license-out technology, few of
them have focused on its consequences. Theoretically, researchers agree upon the
existence of a negative long-term effect: Dissipation Effect (Arora & Fosfuri, 2003;
Gambardella et al. 2006; Zuniga & Guellec, 2009). However, as best of my knowledge,
it was never shown empirically.
In general, when companies license out their technology they are increasing their own
competition and putting at stake their reputation. Thus, it can significantly erode their
market share and their price cost margin (Arora & Fosfuri, 2003; Fosfuri, 2006). Even
though prior research had proposed some strategies to limit the extent of this profit-
dissipation effect (Granstrand et al., 1997; Patel & Pavitt, 1997; Arora & Fosfuri, 2003),
additional competition always can be a threat for the company.
Accordingly, I hypothesize that managers that took the decision to license out its
technology under the pressure to inflate current earnings will face the consequences of
underestimating the dissipation effect in the next years. In particular, we expect that
1u
companies that have engaged in myopic management will present a reduction in their
market share in comparison with the companies that do not.

H2. Companies that license out technology under a pressure situation show a shrinking
market share in the next years compared to companies that do not engage in such
behavior.

4. METHODOLODY

4.1. SAMPLE.

The empirical analysis is based on a sample of innovative U.S. companies. The
operative criterion to select the sample was the following. First, I focused on the
companies (140) with the larger number of granted patents at the U.S.P.T.O. during the
period 1990-2009
6
. Licensing is not a common established practice in many companies
and sectors. As the main objective of this paper is to analyze if companies license out
their technology to inflate current earnings I focus on companies that own the raw
material to do it: companies with technological assets. Second, because of their rich
information environment and because the size of their market for technology, I
narrowed the initial sample to U.S. companies. Licensing data is per se difficult to find,
however, this search process would be even worse in countries where the information
about companies is less accessible and where markets for technology are no so large.
Third, I used annual data. Since a number of companies present quarterly losses
because the intrinsic seasonality related with their business, I prefer to focus on yearly
analysts forecast. From my point of view, it imposes more pressure on managers and,
thus, more incentives to manage earnings.


4.2. DATA

Licensing data were obtained from four sources: Prompt database, HighBeam Research,
Google and SDC Platinum. The first three sources are based on press news while the
last one is an established licensing database. In Prompt, Google News and HighBeam

6
I choose a peiiou of almost 2u yeais because of the noimal length of a gianteu patent. Theiefoie, if 0.S.P.T.0. gianteu
patents to a company in 199u, these patents can still be valiu at the enu the peiiou of stuuy.
11
Research I looked for licensing agreements using key words. In particular, I have
always introduced the term licensing agreement plus the company name. For Prompt
and HighBeam Research, I read all the resulting news, in Google I checked them until
the 20
th
page. After reading the news, I codified these agreements as licensing out,
licensing in or cross licensing. Afterwards, I matched these licensing agreements
with the ones that I obtained from SCD Platinum (For my sample and for the period of
study I found 154 licensing agreements). The final licensing output was 1,729 licensing
agreements of which 840 correspond to license in, 716 to license out and 173 to
cross licensing.
Afterwards, I have matched the licensing data above with the Compustat financial data
and with the analysts forecast data offered by DataStream. These matches reduce the
sample to 107 companies
7
during the period 1998-2009. (1,281
8
observations)

4.3. METHODOLOGY.

To test the first hypothesis, I follow the same procedure as in Bushee (1998). First, in
order to create a model for observing changes I took differences for all variables with
respect to the previous year. Second, I converted these changes in dummies: they are
equal to one if they increased with respect to the previous year, equal to zero otherwise.
3) I use a logit
9
model to predict the probability of observing an increase in the number
of licensing agreements.
To check the second hypothesis I compared the market shares evolution of the
following groups: a) companies that were not able to achieve analysts forecast and that
have increased the number of licensing out contracts with the rest of the sample b)
companies that were not able to achieve analysts forecast in the last year with the rest
of the sample c) companies that have increased the number of licensing out contracts
and that have overcome the analysts thresholds in the previous year with the rest of the
sample.



7
The names of the companies aie iepoiteu in Table 1.
8
1,281 is the iesult of 1uS companies*12 yeai peiiou +1 company*11 yeai peiiou + 1 company*1u yeai peiiou.

9
I ueciueu to use a logit insteau a piobit because two main ieasons: 1) The pseuuo R2 is highei unuei the logit mouel anu 2)
The kuitosis aftei the estimation is highei than S.

12
4.4. VARIABLES DESCRIPTION.

Dependent variable:

The dependent variable that I use (INCLICOUT) is a dummy variable that equals one if
the firm increases the number of licensing contracts relative to the prior year, and zero if
the firm maintains or decreases them. I decided to use a binary variable because of two
main reasons. First, I consider that is unlikely that the magnitude of the change in the
number of licensing agreements be a linear function of difference between actual
earnings per share and the analysts forecast earnings per share. Second, because in
order to inflate current earnings, the fact that a company increases the number of
licensing contracts by seven does not mean that it is increasing more their current
benefits than a company that just increases the number of licensing out contracts by one.
Everything depends on the economic condition of the contracts. Therefore, in this case,
the magnitudes change could be meaningless
10
.

Independent variable

The independent variable (EARNINGS_PRESSURE) is an indicator variable equal to
one if the company has not beat the analysts forecast in the previous year (or if it
presents exactly the same results as the analysts predict) and equal to zero if the
company has surpassed the analysts forecast in the previous year.
This variable was developed in the following way. First I calculated the difference
between the actual earnings per share and the mean of the consensus of analysts
forecast during the fiscal year before the results presentation
11
. I considered that if this
difference is equal to zero or negative, managers will suffer more pressure to attain the
analysts forecast next year. Second, I create a dummy variable that reflect the latter
argument. This variable is equal to one if the difference between the actual earnings per
share and the mean of the consensus of analysts forecast is zero or negative, and zero
otherwise. Evidence has shown that the consequences of missing analysts have been

1u
As iobustness checks I use the magnituue of the change in the numbei of licensing out contiacts ielative to pievious yeai
as uepenuent vaiiable. Results weie no so goou as using binaiy vaiiable, but they weie still significant (Table 17).
11
Item numbei # in BataStieam
1S
disproportionate and that the important distinction is between achieving vs not
achieving them.
Third, I measure earnings pressure in previous year because writing and executing a
reliable licensing contract requires time. If a company wants to license out its
technology it has to discover who could be interested on it, understand what is the
technological base of the licensee, figure out how to implement its technology and try to
negotiate in a context where there are asymmetric information, lack of experience,
difficulties to describe a technology and to value it. Prior literature has demonstrated
that establishing a licensing contract is not an easy task. Razgaitis (2004) show that the
75% of the companies that want to license out their technology are not able to find
licensees. In the same line, Gambardella et al. (2007) found that the 7% of the
technologies available to license remain unlicensed even though companies want to do
it. In an OECD survey, patenting companies recognized that they want to license out
more but that is much difficult to achieve a successful licensing agreement (Zuniga &
Guellec, 2009). Recently, Ali & Cockburn (2011) show that, once licensing
negotiations have began, only the 75% of companies successfully sign the licensing
contract. Given these difficulties, I think that have not sense to propose licensing
contracts as a real activity that has an immediate impact on earnings. Therefore, I
considered that it is reasonable to think that although managers could began with the
licensings negotiations when they realize that they will not be able to attain the
analysts forecast, the licensing contract will not be established until the next year.

Control variables

In order to control for the company size I followed Bushee (1998) and I use the
logarithm of the market value (LOGMV)
12
. In general, size proxies for the amount of
information available about the firm and for the likelihood the firm faces cash
constraints. Following Bushee I expect that larger companies will face fewer
opportunities to successfully manage earnings because of the richer information
environment and less cash flow shortages that force the company to engage in myopic
management. To take into account the economic fluctuations I defined INCGDPUS as a
dummy variable that reflects the change in the U.S. Gross Domestic Product (GDPUS)

12
Results weie invaiiant using the logaiithm of the numbei of employees.
14
with respect to the previous year. This variable is equal to one if the U.S. GDP has
increased from previous year and zero otherwise
13
. Following Bushee (1998) changes in
Gross Domestic Product measure growth in the overall economy and proxy for
increases in the level of technological progress in the economy. Therefore, it is expected
that if the GDP is increasing (decreasing) firms will have more (fewer) opportunities to
license out their technology. Based on previous literature regarding Myopic
Management I defined INCCL as a dummy variable that reflects the change in the
current liabilities (CL) with respect to the prior year. This variable is equal to one if the
current liabilities
14
increase with respect to the previous year, or zero otherwise. It is
supposed that when the ability to pay credits and other short-term liabilities is at stake,
managers will be more worry about the negative reaction of suppliers. Therefore,
following Roychowdhury (2006) I would expect that companies that increase (decrease)
the liabilities with short-term suppliers will have more (less) incentives to inflate current
earnings and, thus, engage in myopic management. I also defined INCGO as a dummy
variable that reflect the change in the companys growth opportunities (GO) with
respect to the previous year. This variable is equal to one if the opportunities to growth
for a company has increased from previous year, zero otherwise. Following Skinner &
Sloan (2002), Hribar et al. (2004) and Roychowdhury (2006) this variable is defined as
the ratio between the market value of equity
15
and the book value of the equity
16
. Since
companies with higher opportunities to growth are more punished by financial markets
if they fail to meet the expected objectives, I would expect that companies with higher
opportunities to growth have more incentives to engage in myopic management. In
order to control for the investments in R&D I defined DECRDINT as a dummy variable
that reflects the change in the R&D intensity with respect to previous year. RD intensity
(RDINT) is defined as the ratio between the R&D expenditure
17
and the sales
18
. This
variable is equal to one if the R&D intensity has decreased from previous year and zero
otherwise. I create seven sector dummy variables (SIC). Six of them correspond to the
more common specific SIC-2 codes in the data and the other one includes the rest of the
SIC2 codes. Table 3 provides a detailed description of the more important SIC codes in
terms of licensing. Finally YEART corresponds to a year trend.

1S
I maue a test to analyze if theie exists multicollineaiity between the !"##$%&'( anu )*(+,. Results show that theie is
not.
14
Compustat item n5
1S
Compustat item n199 * Compustat item n25
16
Compustat item n60 * Compustat item n25
17
Compustat item n46
18
Compustat item n12
1S
5. RESULTS
***INSERT TABLE 2 HERE***

Table 2 shows the main descriptive statistics and correlations of the variables.
Companies establish by mean 0.502 licensing-out contracts per year and they are
involved, as maximum, in 13 licensing out contracts by year.
Table 3 and Table 4 present more detailed information of the licensing out contracts in
the sample. In particular, Table 3 shows that the most of the companies (73.77%) are
not involved in any licensing-out contract per year while the 10.62% of the companies
are involved in less than six per year. This corroborates previous findings that show that
Markets for Technologies are still underdeveloped (Gambardella et al. 2006; Arora et
al. 2010). Table 4 provides evidence that companies belonging to the Electronic &
Other Electrical Equipment, Machinery and Chemical & Allied Products SIC
codes are the ones that license out more technology. Anand & Khanna (2000) have
already shown that these sectors were the three more important in terms of licensing.
However, the proportions of licensing contracts in each sector are quite different. In my
database Electronic & Other Electrical Equipment is the sector more important in terms
of licensing (represents the 22.33% vs 18.15% in Anand & Khanna) followed by
Machinery (19.59% vs 24.2%) and by Chemical (14.05% vs 38.5%). These differences
could be due to a change in the licensing trend over the last 10 years in which the
Chemical sector lost importance.

***INSERT TABLE 3 HERE***

***INSERT TABLE 4 HERE***

***INSERT TABLE 5 HERE***

Table 5 presents a more detailed description of the difference between actual earnings
per share and the mean of the analysts forecast during the year before to the result
presentation (EARNINGS_PRESSURE in magnitude terms). Companies that surpass the
analysts forecast report, as maximum, earnings per share 4.82 higher than the threshold
while companies that fail to beat them report, as maximum, earnings per share 13.7
lower than the analysts forecast. This asymmetry makes that the mean of this variable
be negative (-0.1121) while the median is positive (0.01). Its negative skewness (-6.7)
shows that the mass of the distribution is concentrated on the right and that there are
16
relatively few low values. This corroborates the finding of Dechow, Richarson & Tuna
(2003) that show that so many firms report small profits while only few present losses.
In Table 6 we can see that the 55.89% present positive results while 44.11% report
losses.
*** INSERT TABLE 6 HERE***

To describe better the data I divided the sample in four categories attending to two main
variables in the dummy version: INCLICOUT and EARNINGS_PRESSURE.

***INSERT TABLE 7 HERE ***

Table 7 shows that the percentage of companies that have increased the number of
licensing out contracts is higher for the companies that did not attain analysts forecast
in the previous year (29.3%) than for the companies that have meet them (23.10%).
This intuition is corroborated by the logit regression in Table 8. Model 1 reports the
results of the logit regression only with the control variables and Model 2 reports the
results of the logit regression with the introduction of the earnings pressure variable.
Most of the control variables are significantly associated with the probability of
increasing licensing out. In particular, the coefficient of R&D intensity (DECRDINT) is
significantly negative. This seems not to have so much sense because, in general, higher
R&D intensity would imply higher licensing out (because of a higher number of
technologies available to license). However, if we pay attention to Myopic
Managements literature, there is a potential explanation of this negative sign. Previous
research has shown that cutting R&D expenditures is the most common way to engage
in myopic management. Therefore, in order to inflate current earnings, increasing
licensing out contracts and reducing R&D intensity could be complementary strategies
with different temporal effects
19
. Increasing current liabilities (INCCL) with respect to
the previous year is positively associated with the probability of increase the number of
licensing out contracts. This would be according with the findings of Roychowdhury
(2006) that claim that, when current liabilities with short-term suppliers increases,
managers have more incentives to engage in myopic management. Increasing growth
opportunities (INCGO) with respect to previous year is also positively related with the

19
In oiuei to unueistanu the negative sign I iegiess a logit mouel to analyze if when theie is eainings piessuie manageis
ieuuce R&B intensity. Results coiioboiates pievious finuings.

17
probability of increasing licensing out. This is in accordance with the Skinner & Sloans
findings (2004) that argues that companies with higher opportunities to growth have
more incentives to engage in myopic management. Contrary to Bushees findings, the
logarithm of the market value (LOGMV) is positively and significantly associated with
the probability of increasing licensing-out contracts
20
. Therefore, in our case companies
with higher market value are more likely to increase the number of licensing out
contracts. Gross Domestic Product (INCGDPUSA) has a positive relation with the
probability of licensing-out, hence, when GDP is increasing companies will have more
probabilities to license out technology. Finally, over the years it is more likely to
increase the licensing-out of technology. This corroborates that licensing has increased
over the last years
21
.
Hypothesis 1 proposes that earnings pressure in the previous year makes more likely
that companies increase the number of licensing out contracts during this year. Model 2
support this hypothesis: the coefficient of earnings pressure is positive and significant at
1%. The probability of increasing licensing-out technology increases from 22.6%
(without earnings pressure) to 30% (with earnings pressure)
22
. The introduction of this
variable also generates an increased in the Pseudo R2 from 0.126 to 0.138.

*** INSERT TABLE 8 HERE***

*** INSERT TABLE 9 HERE***

In order to understand the negative sign of the variable Decrease R&D Intensity I run
a logit regression in which this variable is the dependent one. The only difference with
the Increase Licensing Out regression is that I measure earnings pressure in the
current year and not in the previous one. Previous findings show that when managers
suspect that they will not be able to attain earnings thresholds, one of the first things
that they do is to cut discretionary expenses (Stein, 1998; Mizik, 2009). Therefore, as
cutting R&D expenditures has an immediate effect in earnings, I expect that managers
decreased R&D intensity in order to try to inflate current earnings and, thus, meet
analysts forecast. In other words, I expect that they cut R&D intensity before failing to
meet analyst forecasts.

2u
If I use the logaiithm of the numbei of employees as a pioxy of size, the association is still positive. The coefficient is
u.127 ** unuei Nouel 1 anu u.1SS** unuei Nouel 2.

21
In iegiess the same mouel contiolling foi licensing out in the pievious yeai, Tobin's q, anu with the numbei accumulateu
of patents gianteu. These vaiiables weie insignificant anu the mouel was invaiiant without them. Theiefoie, I uiop them.

22
The maiginal effects aftei logit is u.2264 anu the uisciete change of *(+"!"#'%+*''&+* fiom u to 1 is equal to u.u74.
18
*** INSERT TABLE 10 HERE***
*** INSERT TABLE 11 HERE***

Table shows that the coefficient of earnings pressure is positive and significant at the
5%. The probability of decreasing R&D intensity increases in 0.0636 when the variable
earnings pressure is introduced. That is, the probability of decreasing R&D intensity
goes from 45,6% to 52%. This suggest that, even reducing R&D intensity and
increasing licensing out have different temporal effects, managers could use both in
order to inflate current earnings and try to improve the financial appearance of the
company.
Regarding the second hypothesis, as the main objective is to compare the evolution of
the companies that have suffered earnings pressure and have increased the number of
licensing out contracts I distinguished the observations that satisfy these two conditions
SUSPECT- from the rest of the sample -NO_SUSPECT-. Table 12 shows the existing
differences between the two subsamples attending to several variables that describe the
size of the companies, their investments in R&D and the financial situation. Looking to
this table we can pointed out several things: 1) suspect firms are in mean larger than the
no suspect ones, 2) suspect firms invest more in R&D and have more patents available
to license, 3) suspect firms as maximum establish 5 licensing out agreements while no
suspect firms establish as maximum 13, 4) suspect firms in mean present lower market
value, book value, EBIT, total assets, net income, ROA, ROE, ROI, common equity,
cash flow from operations and long term debt than no suspect firms, 5) suspect firms are
characterized because in mean present higher current liabilities and short term debt than
the no suspect ones. Previous description is in accordance with prior findings that show
that larger companies are the ones that more invest in own research and the ones that
have a lower rate of licensing out (their market share is higher and thus, the profit
dissipation effect could be worse). On the other hand, no suspect firms are the ones that
present a better financial situation, higher market value, higher cash flow from
operations, while suspect companies are the ones that have higher short-term
liabilities.
*** INSERT TABLE 12 HERE ***

19
Taking into account this distinction, I compared the evolution of the market shares
23
of
these two subsamples in the next 2 years.

*** INSERT TABLE 13 HERE***

Table 13 shows that the percentage of no-suspect companies that decrease their market
share is constant over the period (57%) while this percentage is increasing in the group
of suspect companies (from 52% to 61%). In period t seems that suspect companies
perform even better than the non-suspect ones: only 52% of the suspect companies
decrease their market share with respect to previous year while 57% of the non-suspect
ones do so. In period t+1 suspect companies began to perform worse than the non-
suspect: 58,6% against 57,1%. Finally, in period t+2, suspect firms behave much worse
than the non-suspect ones: 61% of suspect companies decrease their market share while
only 57% of the non-suspect companies do so.
In order to analyze if this decreasing trend in the market share is caused by previous
financial problems, and not caused by licensing out, I compared the evolution of the
market share for those companies that were not able to achieve analysts forecast and
for those that have achieved them. Table 14 shows that the percentage of companies that
decrease their market share is stable and similar for both groups (57% for companies
that were able to meet analysts forecast and 58% for the ones that were not able). It
suggests that the decrease in the market share is independent of the financial situation of
the company. Finally, I compared the evolution of the market share for the companies
that have increased the number of licensing out contracts and have achieved the
analysts forecast in previous year (INC_NOPRESSURE=1) with the rest of the sample.
Table 15 shows that those companies present a decreasing market share trend over the
three-year period (from 44% to 53%) while this trend is quite stable (58%) for the rest
of the sample. Thus, from the latter comparisons, we can observe that 1) companies that
have increased the number of licensing out contracts with respect to the previous year
present a decreasing market shares trend in the following two years, 2) the percentage
of companies that decrease market share is greater (8% in the three years) for the
companies that license out technology and were not able to achieve analysts forecast
than for companies that overcome the earnings threshold 3) missing analysts forecast in

2S
In oiuei to compute maiket Shaie I iuentifieu the sectoi using the S uigit SIC coue. BECREASENARKETSBARE is uefineu
as a uummy vaiiable equal to one if the maiket shaie have uecieaseu fiom pievious yeai anu equal to zeio if not.
2u
previous year seems to be independent of the decrease in the market share. Graph 1
compares the evolution of the market share of each group for the next two years.

***INSERT GRAPH 1 HERE***

ROBUSTNESS CHECKS

As a robustness checks I also estimated the model using two alternative methods. First,
using the binary variable, results remain significant under the conditional logit model
(Table 16). Second, using as dependent variable the magnitude of the change in the
number of licensing agreements, the earnings pressure variable was also significant
under the ordinary least squares method (Table 17)
24
.



6. SUMMARY & CONCLUSION.

The main objective of this paper was to shed light on the relationship between the
company financials situation and its licensing strategy. Over last years financial
analysts have changed the way in which managers run the business. The consequences
of missing analysts expectations have been so severe that managers are increasingly
becoming short-term minded. Accordingly, they have improved their creativity and
have employed several real activities to inflate current earnings even at the expense of
long term performance (Aaker, 1991; Bartov, 1993; Herman et al., 2003;
Roychowdhury, 2006; Moorman & Spencer, 2008). From my point of view, licensing
out technology could be one of these real activities because the trade off that it implies.
On one hand, companies will increase their current benefits by the licensing revenues
(net of transaction costs) but, on the other hand, companies could reduce their market
share or their price-cost margin because they are creating their own competition in the
product market. Since myopic managers by definition put more emphasis on the short
term (revenue effect) than in the long term (dissipation effect), I expect that managers
under the pressure to beat analysts forecast license out their technology just taking into
account the revenue effect. Stein (1989) argue that the easier way to inflate short term
earnings is to reduce intangible assets expenditures because they are not separately

24
I founu that the eainings piessuie vaiiable was significant unuei fixeu effect mouel. This iegiession is available fiom the
autoi upon iequest.
21
recorded in the balance sheet and because they are not directly related to production.
Licensing out technology also satisfies these two conditions: 1) companies have not the
obligation of recording licensing revenues as a separate item (then, external observers
just perceive an increase in the earnings) and 2) they do not affect to production in the
short term. Accordingly, from my point of view, licensing out intellectual property
could provide managers the opportunity to inflate current earnings and to take
advantage of them for a period. Nevertheless, companies that engage in such practices
will face negative consequences in the long term.
Based on the latter argument I proposed that 1) managers are more likely to license out
technology when they were not able to attain analysts forecast in the previous year and
that 2) companies that license out technology under a pressure situation will reduce their
market share in the incoming years.
I test these two hypotheses using a panel of 107 U.S. companies during a twelve year
period (1,281 observations). Results confirm that 1) when companies were not able to
achieve analysts forecast in previous years are more likely to increase the number of
licensing out contracts and that 2) companies that license out their technology under a
pressure situation will show a shrinking market share in the next two years compared to
companies that did not. Results also corroborates previous findings that show that
managers decrease R&D expenditures when they suspect that they will not be able to
achieve earnings thresholds in next period.
This paper contributes to innovation literature in two ways. First, because it shows that
the pressure to achieve analysts forecasts is also a potential determinant to license out
technology. Second, because it is the first time that the negative long-term
consequences of licensing (dissipation effect) have been empirically tested (dissipation
effect). This paper also contributes to the Myopic Management Theory in two ways.
First, because it shows that licensing out technology could be used as a real activity that
increases current earnings at the expense of reducing market share in the incoming
years. Second, because it amplifies the temporal horizon of the real activities. As the
best of my knowledge, researchers have focused on the strategies that managers took in
the period previous to the negative earnings surprise. However, this paper suggests that
managers can also follow strategies that have not an immediate effect.
This paper has also several limitations. The first one is related with the earnings
pressures proxy. I considered that companies are supporting earnings pressure if they
have failed to meet analysts forecast (or just matched them) in previous year. However,
22
this proxy is not very precise. Ideally, I would have to know the exact moment at which
managers realized that they will not be able to achieve analysts forecast because it is in
this moment when they will begin to take decisions in order to inflate current earnings.
Indeed, it is quite likely that firms that miss analysts forecast have been manipulating
results and dealing with financial difficulties for some time. Nevertheless, it is
impossible to know the exact moment at which companies decide to behave myopically
and to differenciate real earnings from the inflated ones. There are also two limitations
regarding licensing data. First, even though I suspect that the most of the licensing
contracts are related with a patent (patent protection incentivates licensing) I cannot
ensure that it is true. Therefore, licensing agreements could be related with know how,
copyrights, trade secrets or patents. Second, I treated any increase in the number of
licensing out agreements as suspect without knowing specific information about the
contract. In particular, I have defined suspect companies as the ones that have missed
(or have matched) analysts forecast in the previous year and the ones that, at the same
time, have increased the number of licensing contracts from previous years. However, I
cannot ensure that these new licensing contracts are dangerous for the companies
because I do not know if contracts are related with core technology, if entail direct
competitors or if, for instance, both parties operates in the same geographical area. In
the part of future research, I propose to introduce some measures to avoid this problem.
Finally, the results of this paper cannot be generalized to the whole population of
companies. Licensing out is still not a common practice in all the industries and
countries. Companies in the sample are characterized by four main things: 1) they have
the intellectual assets needed to trade in the Markets for Technology, 2) they belong to
industries in which licensing agreements are frequently established, 3) they operate in
the most developed environment (U.S.) for Markets for Technology and 4) they operate
in a country where analysts forecast imposes much pressure. Therefore, results cannot
be applied to other companies that do not satisfy the latter characteristics.
Regarding future research, my plan is the following. First of all, I would like to add to
my data the economic conditions of licensing contracts. Myopic management can also
be reflected in the way in which the fixed fee and the royalties are established. In
particular, I would expect that firms engaged in myopic management establish greater
fixed fee and lower royalties than optimal in order to borrow earnings from the future.
Second, it would be interesting to know the sector at which belongs each company
involved in the licensing contract or the technological proximity among them. It would
2S
allow to me to analyze deeper the cautions that managers have in order to limit the
dissipation extent. I would expect that companies that are supporting earnings pressure
do not be cautious and license out technology to companies that operates in the same
sector or that are technologically close. Finally, I am interested in analyzing the
application of grant back clauses in licensing contracts. Under these clauses the licensee
is required to disclose and transfer all the improvements made in the licensed
technology to the licensor. No doubt, those provisions limit the extent of the dissipation
effect and motivate companies to license out technology. However, licensees are not
conformable with their application and try to avoid it as far as they can. Accordingly, I
would expect that companies that are supporting the pressure to meet earnings
benchmarks will want to negotiate fast and, thus, will not engage in discussions
regarding rant back clauses.
From a practical point of view, this paper gives some insights that companies should
take into account. First, it is needed to educate managers about the potential long-term
consequences of licensing. It is important that they analyze the decision to license with
caution, focusing on the net benefits of the strategy. Second, it is required to change the
way in which managerial compensation is established in order to incentive managers to
engage in projects that maximize the sum of discounted future profits. Third, the results
also highlight the negative consequences of a decentralized licensing structure. If
companies had an independent licensing department in charged on taking those
decisions and whose incentives be different from the ones of the economic department,
managers could not license out their technology just to benefit from the inflated current
earnings. Finally, it would be good that society meditate about the negative
consequences of imposing earnings pressure to managers. Clearly, this pressure do not
allow managers to focus on long term strategies putting at stake the long term
productivity of companies and, in turn, the one of whole society.








24
TABLES
TABLE 1. COMPANIES IN THE SAMPLE
1 3M CENTER 55 HONEYWELL INTERNATIONAL INC
2 ADOBE SYSTEMS INC 56 HUMAN GENOME SCIENCES
3 ADVANCED MICRO DEVICES 57 ILLINOIS TOOL WORKS
4 AIR PRODUCTS AND CHEMICALS INC 58 IMATION CORPORATION
5 ALCOA INC 59 INCYTE CORPORATION
6 ALLERGAN INC 60 INTEL CORPORATION
7 ALTERA CORPORATION 61 INTL BUSSINES MACHINES CORPORATION
8 AMGEN INC 62 INTL FLAVORS & FRAGANCES
9 ANALOG DEVICES 63 ISIS PHARMACEUTICALS INC
10 APPLE INC 64 ITT CORPORATION
11 APPLIED MATERIALS INC 65 JDS UNIPHASE CORPORATION
12 AT&T INC 66 KIMBERLY CLARK CORP
13 AVERY DENNISON CORP 67 KLA-TENCOR CORPORATION
14 BAKER HUGHES INCORPORATED 68 LAM RESEARCH CORPORATION
15 BARD (CR) INC 69 LATTICE SEMICONDUCTOR CORPORATION
16 BECTON DICKINSON & CO 70 LEAR CORPORATION
17 BOEING CO 71 LEXMARK INTL INC
18 BORGWARNER INC 72 LOCKHEED MARTIN CORPORATION
19 BOSTON SCIENTIFIC CORPORATION 73 LUBRIZOL CORPORATION
20 BRISTOL MYERS SQUIBB CORPORATION 74 MATTEL INC.
21 BROADCOM CORPORATION 75 MEDTRONIC INC.
22 BRUNSWICK CORPORATION 76 MERCK & CO.
23 CABOT CORPORATION 77 MICRON TEHCNOLOGY INC
24 CADENCE DESIGN SYSTEMS INC 78 MICROSOFT CORPORATION
25 CALLAWAY GOLF COMPANY 79 MOLEX INCORPORATED
26 CATERPILLAR INC 80 NATIONAL INSTRUMENTS CORPORATION
27 CIENA CORPORATION 81 NATIONAL SEMICONDUCTOR CORP
28 CIRRUS LOGIC, INC. 82 NCR CORPORATION
29 CISCO SYSTEMS INC 83 NIKE INC -CL B
30 COLGATE PALMOLIVE CO 84 NORDSON CORPORATION
31 CONEXANT SYSTEMS INC 85 NORTEL NETWORKS CORP
32 CORNING INC 86 NORTHROP GRUMMAN CORPORATION
33 CYPRESS SEMICONDUCTOR CORPORATION 87 NOVELLUS SYSTEMS INC
34 DANA HOLDING CORPORATION 88 NVIDIA CORPORATION
35 DEERE & CO 89 ORACLE CORPORATION
36 DELL INC 90 PFIZER INC
37 DOW CHEMICAL CO 91 PPG INDUSTRIES INC
38 EASTMAN CHEMICAL COMPANY 92 PRAXAIR INC
39 EASTMAN KODAK CO 93 QUALCOMM INC
40 EATON CORPORATION 94 RAMBUS INC
41 EI DU PONT DE NEMOURS 95 RAYTHEON CO
42 ELI LILLY & CO 96 ROCKWELL AUTOMATION
2S
43 EMC CORPORATION 97 SANDISK CORPORATION
44 EMERSON ELECTRIC CO 98 TERADYNE INC
45 EXXON MOBIL CORPORATION 99 TEXAS INSTUMENTS INCORPORATED
46 FMC CORPORATION 100 TRIMBLE NAVIGATION LTD
47 FORD MOTOR CO 101 UNISYS CORPORATION
48 GENERAL ELECTRIC COMPANY 102 UNITED TECHNOLOGIES CORPORATION
49 GENTEX CORPORATION 103 WESTERN DIGITAL CORPORATION
50 GOODRICH CORPORATION 104 WEYERHAEUSER COMPANY
51 GOODYEAR TIRE & RUBBER 105 WHEATERFORD INTL LTC
52 HALLIBURTON CO 106 XEROX CORPORATION
53 HARRIS CORPORATION 107 XILINX INC
54 HEWLETT-PACKARD CO





TABLE 3. LICENSING OUT
AGREEMENTS BY YEAR
!"#$%&
(%& )%*& +&%,- .%&/%01 2"#-
0 945 73.77 73.77
1 194 15.14 88.91
2 68 5.31 94.22
3 35 2.73 96.96
4 21 1.64 98.59
5 8 0.62 99.22
6 4 0.31 99.53
7 1 0.08 99.61
9 2 0.16 99.77
10 1 0.08 99.84
12 1 0.08 99.92
13 1 0.08 100
TABLE 2. DESCRIPTIVE STATISTICS & CORRELATIONS
MEAN SD MAX MIN
1 2 3 4 5 6 7
1.LICOUT
0.502 1.173 13 0 1.000

2.EARNINGS_PRESSURE
-0.112 0.857 4.820 -13.700 0.005 1

3.RDINT
45.691 70.457 659.468 0.004 -0.099 -0.0111 1

4.CL
4,516.180 8,721.251 141,579 12.751 0.173 0.0223 -0.232 1

5.GO
5.957 45.140 1,575 -118.636 0.148 0.0014 -0.0197 -0.013 1

6.LOGMV
9.235 1.641 13.139 1.282 0.148 0.1517 -0.1175 0.5355 0.0142 1

7.GDPUSA
155.319 5.467 163 145 -0.022 -0.0288 0.0246 -0.0818 0.0051 0.037 1
26












TABLE 6. EARNINGS_PRESSURE
DISTRIBUTION

Freq. Percent Cum.
POSITIVE 716 55.89 55.89
NEGATIVE 565 44.11 100
TOTAL 1,281 100




TABLE 4. LICENSING AGREEMENTS BY SIC-2
SIC2 Definition Freq. Percent Cum.
36 Electronic & Other Electrical Equipment 286 22.33 66.28
35 Machinery 251 19.59 43.95
28 Chemical & Allied Products 180 14.05 19.67
37 Transportation Equipment 132 10.3 76.58
38 Instruments & Related Products 132 10.3 86.89
73 Business Services 84 6.56 100
39 Miscelaneous Manufacturing Industries 48 3.75 90.63
25 Furniture & Fixtures 24 1.87 3.75
26 Paper & Allied Products 24 1.87 5.62
30 Rubber & Miscelaneous Plastic Products 24 1.87 22.48
67 Holding & Other Investments Offices 24 1.87 93.44
1 Agriculture 12 0.94 0.94
24 Lumber & Wood Products 12 0.94 1.87
29 Petroleum & Coal Products 12 0.94 20.61
32 Stone, Clay & Glass Products 12 0.94 23.42
33 Primary Metal Industries 12 0.94 24.36
48 Communications 12 0.94 91.57

TABLE 5. EARNINGS_PRESSURE
Percentiles Smallest
1% -3.43 -13.7
5% -0.97 -11.59

10% -0.51 -7.17 Obs 1,281
25% -0.14 -6.78


50% 0.01

Mean -0.1121

Largest Std. Dev. 0.86
75% 0.08 3.05

90% 0.27 3.35 Variance 0.73
95% 0.47 4.15 Skewness -6.7
99% 1.51 4.82 Kurtosis 89.77
27
TABLE 7. SAMPLE DIVIDED IN 4 CATEGORIES


EARNINGS_PRESSURE

NO YES
LICOUTt+1-LICOUTt<0
449 (76.90%) 417 (70.70%)
LICOUTt+1-LICOUTt>0

135 (23.10%) 173 (29.30%)

.
TABLE 8. LOGIT REGRESSION.
INCREASE LICENSING OUT
(1) (2)
INCLICOUT
NO EARNINGS
PRESSURE
EARNINGS
PRESSURE

DECRDINT -0.552
***
-0.582
***

(0.148) (0.154)

INCCL 0.332 0.382
*

(0.170) (0.174)

INCGO 0.281 0.339
*

(0.156) (0.163)

INCGDPUSA 0.993
***
0.880
***

(0.127) (0.132)

LOGMVE 0.157
**
0.153
**

(0.0549) (0.0582)

YEART 0.164
***
0.172
***

(0.0208) (0.0251)

SIC28 0.140 0.0640
(0.288) (0.303)

SIC35 -0.127 -0.0647
(0.256) (0.270)

SIC36 -0.241 -0.351
(0.268) (0.281)

SIC37 -0.607 -0.672
*

(0.322) (0.301)

SIC38 -0.141 -0.0212
(0.312) (0.319)

SIC73 -0.0930 -0.222
(0.356) (0.356)

EARNINGSPRESSURE 0.418
**

(0.141)

_cons -3.951
***
-4.156
***

(0.589) (0.685)
N 1163 1163
pseudo R
2
0.126 0.138

Standard errors in parentheses
* p < 0.05, ** p < 0.01, *** p < 0.001


28

TABLE 9. MARGINAL EFFECTS AFTER LOGIT.
INCREASE LICENSING OUT
(1) (2)
INCLICOUT
NO EARNINGS
PRESSURE
EARNINGS
PRESSURE

DECRDINT (d) -0.0954
***
-0.102
***

(0.0250) (0.0264)

INCCL (d) 0.0567
*
0.0659
*

(0.0283) (0.0290)

INCGO (d) 0.0492 0.0602
*

(0.0271) (0.0286)

INCGDPUSA (d) 0.181
***
0.165
***

(0.0218) (0.0247)

LOGMVE 0.0275
**
0.0271
**

(0.00959) (0.0102)

YEART 0.0287
***
0.0305
***

(0.00313) (0.00381)

SIC28 (d) 0.0253 0.0115
(0.0530) (0.0550)

SIC35 (d) -0.0217 -0.0113
(0.0431) (0.0468)

SIC36 (d) -0.0407 -0.0589
(0.0437) (0.0447)

SIC37 (d) -0.0925
*
-0.102
**

(0.0424) (0.0395)

SIC38 (d) -0.0240 -0.00373
(0.0514) (0.0559)

SIC73 (d) -0.0159 -0.0373
(0.0596) (0.0566)

EARNINGSPRESSURE (d) 0.0740
**

(0.0249)
N 1163 1163
pseudo R
2
0.126 0.138

Marginal effects; Standard errors in parentheses
(d) for discrete change of dummy variable from 0 to 1
*
p < 0.05,
**
p < 0.01,
***
p < 0.001









29

TABLE 10. LOGIT REGRESSION
DECREASE RD INTENSITY
(1) (2)
DECRDINT
NO EARNINGS
PRESSURE
EARNINGS
PRESSURE

INCCL -0.00618 0.0242
(0.126) (0.127)

INCGO 0.0186 0.00531
(0.116) (0.118)

INCGDPUSA 0.0880 0.104
(0.116) (0.118)

LOGMVE 0.00321 0.0139
(0.0405) (0.0401)

YEART -0.0754
***
-0.0679
***

(0.0152) (0.0155)

SIC28 0.252 0.272
(0.203) (0.203)

SIC35 0.0581 0.0713
(0.183) (0.182)

SIC36 0.0716 0.0938
(0.156) (0.153)

SIC37 -0.0488 -0.0303
(0.192) (0.192)

SIC38 0.354
*
0.394
**

(0.144) (0.144)

SIC73 0.0158 0.0302
(0.207) (0.199)

EARNINGSPRESSURE_T 0.256
*

(0.127)

_cons 0.0823 -0.217
(0.408) (0.409)
N 1170 1170
pseudo R
2
0.015 0.018

Standard errors in parentheses
*
p < 0.05,
**
p < 0.01,
***
p < 0.001









Su


TABLE 11. MARGINAL EFFECTS AFTER LOGIT.
DECREASE RD INTENSITY
(1) (2)
DECRDINT
NO EARNINGS
PRESSURE
EARNINGS
PRESSURE

INCCL (d) -0.00154 0.00600
(0.0312) (0.0315)

INCGO (d) 0.00461 0.00132
(0.0288) (0.0292)

INCGDPUSA (d) 0.0219 0.0258
(0.0287) (0.0293)

LOGMVE 0.000797 0.00345
(0.0101) (0.00995)

YEART -0.0187
***
-0.0169
***

(0.00379) (0.00384)

SIC28 (d) 0.0629 0.0678
(0.0507) (0.0507)

SIC35 (d) 0.0144 0.0177
(0.0455) (0.0453)

SIC36 (d) 0.0178 0.0233
(0.0387) (0.0380)

SIC37 (d) -0.0121 -0.00753
(0.0474) (0.0477)

SIC38 (d) 0.0883
*
0.0983
**

(0.0358) (0.0357)

SIC73 (d) 0.00393 0.00751
(0.0514) (0.0496)

EARNINGSPRESSURE_T
(d)
0.0636
*

(0.0313)
N 1170 1170
pseudo R
2
0.015 0.018

Marginal effects; Standard errors in parentheses
(d) for discrete change of dummy variable from 0 to 1
*
p < 0.05,
**
p < 0.01,
***
p < 0.001









S1


TABLE 13. DECREASE MARKET SHARE
NO_SUSPECT SUSPECT
T 0.575812274 0.520231214
T+1 0.571028037 0.586538462
T+2 0.57378741 0.612244898

TABLE 14. DECREASE MARKET SHARE
NO_EARNINGSPRESSURE EARNINGSPRESSURE
T 0.561643836 0.583050847
T+1 0.573260073 0.581573896
T+2 0.574736842 0.583505155





2S
All these vaiiables weie collecteu fiom Compustat except Licensing0ut (that comes fiom the 4 souices explaineu befoie)
anu Patentsuianteu (collecteu fiom 0.S.P.T.0). AccPatentsuianteu was computeu using Patentuianteu with a uepieciation
factoi of 1S%.
TABLE 12. STATISTICS DESCRIPTIVE OF SUBSAMPLES
25

VARIABLE
343.526 !78343.526

95:! 3; < 7=3 9>! 9:? 95:! 3; < 7=3 9>! 9:?
59.@7A553 !"#$%&#' )*++"&," %)% $! *++$#+ $!$,!&)' ")%*#&!, %#+' + *++$#+
3:@53 +'")+&,* '#!*"!&% %)* %#%)&+#) +!#)++ ,#)*+&+, %+'+"'&% %%#, "+&$"*% ++,#+*
B; '!#!++&+ '+#!%!&, %") %#!# ++)### '%,"$%&" '',",%&) %#)$ "+ ++%,)*
.:65!6CB:65; '$$&"%"* $$#&#!+' %)' # ',"' %,'&#$% *)%&+'!, %#+) # *+,*
:22.:65!63CB:!65; $$$&+*,) )+#&%!*" %)* %#&' !$*)&! *'#&!'*! "!$&+)* %%#, # "+$$&$
@>25!3>!C746 #&$!""$)$ #&+%)+'! %)* # ! #&!#+#'!* %&'#,!"' %%#, # %*
=77DE:@45 %#,*!&*% )$""&,") %)* -),'%&%%$ %%)'+*&, '%'"%&!" %!#!$&$" %%#" -%)*%%&#' %'%)"'&$
9:BD56E:@45 $###,&$% **$$! %)* %%&$!,)$ *,"$#'&% )%#!"&+) "!"!$&* %%#" *&"#$, !#,*'"&'
5=>6 +,),+&!+ ',*",,&" %)* -)#'### +,,)## %""!%#&, ',%##$ %%#, -+",### ++,"##
676:@:33563 $#)"%&"* ''!'"&$, %)* %,*&'!" )+!**) %%!'#"&$ "%#")&,' %%#, +"&#)$ )+))"+
!56>!2795 %,#)&"+" %'!!&$* %)* -''*"&!, '''#, $#'!&'** $'#%&,!% %%#, -!"%'%&+ $!''#
B564B!:33563 #&,'"#',+ *&$"$$'" %)* -%##&+'' '$&*$, %$&!)*,% '%&""'$+ %%#, -$!,&*% $,&%!*
B564B!35F4>6A '&)"$#+ -%#&**"%% %") -'),&)$! "'&)*! *,&,+!') $,+&'$%, %#,' -%!"*%&*% *')&*+)
B564B!>!E53695!6 '&!$$+,' '%&*),*" %)# -%*!&+#' $,&)'! !&!'##$* *!&#++)% %%#* -!'*&!!+ +*&%%)
27997!5F4>6A %%#*!&"! )$!)&%%* %)* -),'# %%)'+% '%'''&'! %!#*+&%' %%#, -%)*%% %'%)"'
2:3G7.5B:6>7!3 $%))&,*+ '""+&+*" %$$ -%#%! $!+") )""*&++) "%!!&*%, +$, -*++% !+)'!
@7!C65B9;5=6 %#!*'&#) $$#%&%') %)* # *))%*, $'+!%&*, %,)''&+' %%#) # **##")
24BB5!6@>:=>@>6>53 !,"*&)++ +*#)&'%+ %"! %$&"*" !'#"% $*%#&$,! ,"%$&*'+ %#,% %'&)!% %'%!)+
3G7B65B9;5=6 $,,%&)%) '*,*'&%+ %)* # %+!%#% '*,$&''! %$#"%&,* %%#, # %+*"+!
S2

TABLE 15. DECREASE MARKET SHARE
INC_NOPRESSURE=0 INC_NOPRESSURE=1
T 0.582897033 0.444444444
T+1 0.578458682 0.505154639
T+2 0.581632653 0.528735632

GRAPH 1. MARKET SHARE EVOLUTION




TABLE 16. ROBUSTNESS CHECKS.
CONDITIONAL LOGIT. INCREASE LICENSING OUT.
(1) (2)
INCLICOUT

NO EARNINGS PRESSURE

EARNINGS PRESSURE

DECRDINT -0.609
***
-0.644
***

(0.153) (0.163)

INCCL 0.438
**
0.438
*

(0.164) (0.172)

INCGO 0.247 0.313
(0.158) (0.166)

INCGDPUSA 0.975
***
0.865
***

(0.151) (0.159)

LOGMVE 0.0796 0.114
(0.122) (0.133)

YEART 0.165
***
0.174
***

(0.0222) (0.0267)

EARNINGSPRESSURE 0.470
**

(0.171)
N 1163 1163
pseudo R
2
0.159 0.180
Standard errors in parentheses
*
p < 0.05,
**
p < 0.01,
***
p < 0.001



SS
TABLE 17. OLS, USING THE MAGNITUDE CHANGE IN THE
NUMBER OF LICENSING OUT CONTRACTS AS DEPENDENT VARIABLE
(1) (2)
NO EARNINGS PRESSURE EARNINGS PRESSURE
DECRDINT 0.00123 -0.00901
(0.0790) (0.0857)

CUTCL -0.0000430 -0.0000477
(0.0000232) (0.0000257)

CUTGO 0.00193
**
0.00197
**

(0.000586) (0.000602)

CUTGDPUSA 0.00888 0.00120
(0.0149) (0.0166)

LOGMVE 0.00987 0.00677
(0.0255) (0.0278)

YEART -0.0262
*
-0.0155
(0.0130) (0.0152)

SIC28 -0.0181 -0.0633
(0.143) (0.154)

SIC35 -0.00294 0.0150
(0.134) (0.144)

SIC36 -0.0281 -0.0470
(0.129) (0.138)

SIC37 -0.0121 -0.00610
(0.161) (0.173)

SIC38 -0.0249 0.00255
(0.156) (0.168)

SIC73 -0.0112 -0.0562
(0.181) (0.194)

EARNINGSPRESSURE 0.178
*

(0.0898)

_cons 0.100 -0.0241
(0.262) (0.308)
N 1128 1027
pseudo R
2
















S4
REFERENCES (Incomplete)

Arora, A., Fosfuri, A. & Gambardella, A. (2001), Markets for Technology and Their
Implications for Corporate Strategy. Industrial and Corporate Change, Vol.10,
Issue 2, pp. 419-451.
Arora, Ashish & Andrea Fosfuri, 2003. Licensing the Market for Technology. Journal
of Economics Behavior & Organization, Vol. 52, pp. 277-295.
Arora, Ashish & Alfonso Gambardella, 2010. Ideas for Rent: An Overview of Markets
for Technology. Industrial and Corporate Change, Vol. 19, No.3, pp. 775-803.
Benassi, M. & Di Minin, A. (2009). Playing in between: patent brokers in markets for
technology. R&D Management, 39, 1,
Cesaroni, F. (2001). Technology Strategies in the Knowledge Economy. The Licensing
Activity of Himont. LEM Working Paper Series. St. Anna School of Advanced
Studies, Pisa, Italy.
Degeorge et al. (1999). Earnings Management to Exceed Thresholds. Journal of
Business, 1999, vol. 72, no.1
European Patent Office (EPO), Japan Patent Office (JPO) and United States Patent and
Trademark Office (USPTO), 2004. Trilateral Statistical Report 2003 Edition.
www.jpo.go.jp/torikumi_e/kokusai_e/tws/tsr2003/ch2.pdf.
Farrel, Joseph & Nancy Gallini, 1988. Second-Sourcing as a Commitment: Monopoly
Incentives to Attract Competition. The Quarterly Journal of Economics, Vol.103,
No.4, pp. 673-694.
Fosfui, A. (2006), The licensing dilemma: understanding the determinants of the rate of
technology licensing. Strategic Management Journal. Vol. 27, Issue 12, pp.
1141-1158.
Gallini, Nancy, 1984. Deterrences by Market Sharing: A Strategic Incentive for
Licensing. The American Economic Review, Vol. 74, No. 5, pp. 930-941.
Gambardella, A., Guiri, P. & Luzzi, A. The Market for Patents in Europe.(2006). LEM
Papers Series 2006/04, Laboratory of Economics and Management (LEM), Sant
Anna School of Advanced Studies, Pisa, Italy.
Japan Patent Office (JPO), 2004. Results of the Survey on Intellectual Property-Related
Activities, 2003. JPO, Tokyo.
Kamiyama, S., J. Sheehan and C. Martinez (2006), Valuation and Exploitiation of
Intellectual Property, OECD Science, Technology and Industry Working
Papers, 2006/2005, OECD Publishing.
SS
Katz, Michael & Carl Shapiro. On the Licensing of Innovations. The RAND Journal of
Economics, Vol. 16, No. 4
Lichtenthaler, Ulrich, 2007. The Drivers of Technology Licensing: An Industry
Comparison. California Management Review, Vol. 49, No.4, pp. 67.
Pluvia Zuniga, M. and D. Guellec (2009), Who Licenses out Patents and Why? Lesssons
from a Business Survey, OECD Science, Technology and Industry Working
Papers, 2009/5, OECD Publishing.
Shapiro, Carl. 1985. Patent Licensing and R&D Rivalry. The American Economic
Review, Vol. 75, No. 2, pp. 25-30.
Shapiro, Carl. 2001. Navigating the Patent Thicket: Cross Licenses, Patent Pools, and
Standard Setting. Innovation Policy and the Economy, Vol. 1, pp. 119-150.
Shepard, Andrea. 1987. Licensing to Enhance Demand for New Technologies. RAND
Journal of Economics., Vol. 18, No.3, Autumm.
Vishwasrao, Sharmila. 2007. Royalties vs. Fees: How do Firms Pay for Foreign
Technology?. International Journal of Industrial Organization, Vol. 25, pp. 741-
759.
Yanagisawa, T. & Guellec, D. The Emerging Patent Marketplace (2009). STI Working
Paper 2009/9. Statistical Analysis of Science, Technology and Industry