Sources of appropriation capacity in licensing agreements

Technovation xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Technovation journal homepage: www.elsevier.com/locate/technovation

Sources of appropriation capacity in licensing agreements Goretti Cabaleiro Universidad Alberto Hurtado, Departamento de Gestión y Negocios, Chile

A R T I C LE I N FO

A B S T R A C T

JEL codes: O32 O33 L24

This article analyzes how a licensor's market value varies at the time that it announces a licensing agreement. It posits that licensors’ appropriation capacity is a function of their bargaining power (determined by their financial situation and information asymmetries when signing the contract) and the potential cost of imitation that the licensors face (determined by their position in the sector). As bargaining power in each situation should determine licensors’ success, it also should enhance their appropriation capacity in terms of market value. However, the cost of imitation should have a negative effect on licensors’ appropriation capacity. This study shows that companies with cash constraints appropriate fewer benefits from licensing than do companies that have a ready cash flow (0.87% vs. 2.84%). If companies license out under low information asymmetries (same sector), they also appropriate more benefits from licensing than do companies facing high information asymmetries (different sector) (7.27% vs. 1.53%). Finally, licensors that are leaders appropriate fewer benefits than those that are followers (0.61% vs. 3.86%).

Keywords: Technology licensing Bargaining power Appropriation capacity

1. Introduction With the proliferation of markets for technology, licensing has attracted increasing attention in the management literature (Fosfuri, 2006; Bianchi et al., 2011; Arora et al., 2013; Conti et al., 2013; Natalicchio et al., 2014). However, previous research puts little emphasis on the factors that influence the appropriation capacity of the parties involved in the agreement. Therefore, little is known about the abilities of the parties to take possession and extract value from licensing. The main reason for this gap in the literature is the lack of uniform licensing data: licensing agreements are confidential, and companies are not required to report licensing revenues as a separate item in their income statements. As a consequence, it is quite difficult to access the specific economic conditions of the agreements (such as fixed fee and royalties) to quantify the well-known Revenue Effect: “rents earned by the licensee which will accrue to the patent holder in the form of licensing payments” (Arora and Fosfuri, 2003, p. 278). Given this limitation, the few studies that focus on the appropriation capacity of the parties rely on data from surveys (Bianchi et al., 2014; Bianchi and Lejarraga, 2016) or on cumulative abnormal returns (CARs) as a market-based measure of performance (Walter, 2012). This study analyzes the determinants of licensors’ appropriation capacity—measured as the variation in the licensors’ market value—at the moment that they announce a licensing agreement. In particular, the paper argues that licensors’ appropriation capacity is a function of their bargaining power and the potential cost of imitation. Specifically, the paper posits that

bargaining power is determined by the existence of financial constraints in the company and the presence of information asymmetries between parties. The previous literature notes that financial constraints erode companies’ bargaining power (Aghion and Tirole, 1994; Lerner and Merges, 1998). Thus, this study hypothesizes that, as a consequence of their weak bargaining power, companies with cash constraints appropriate fewer benefits from licensing than do companies that are not cash-constrained. The previous research also demonstrates that incomplete information reduces companies’ bargaining power (Chatterjee and Samuelson, 1983). Therefore, the hypothesis here is that licensors are better able to appropriate benefits from licensing when they license out in the same sector (characterized by lower information asymmetries) than when they license out to another sector (characterized by higher information asymmetries). Since bargaining power in each situation determines licensors’ success and the level of rewards in the agreement (Kim, 1997; Mannix, 1993a; Pinkley et al., 1994), one would expect it to increase licensors’ appropriation capacity in terms of market value. In addition, this paper argues that the potential cost of imitation has a negative influence on licensors’ appropriation capacity. Since the threat of imitation increases with licensors’ greater market share (Arora and Fosfuri, 2003), the next hypothesis is that licensors that lead a specific sector (i.e., have the highest market share) appropriate fewer benefits from licensing than those that are followers. This empirical question is important for the following reasons. First, in an efficient market, the current share price is the best available estimate of a company's true value (Akhtar and Oliver, 2009). Second,

E-mail address: [email protected]. https://doi.org/10.1016/j.technovation.2018.07.006 Received 8 July 2017; Received in revised form 16 July 2018; Accepted 19 July 2018 0166-4972/ © 2018 Published by Elsevier Ltd.

Please cite this article as: Cabaleiro, G., Technovation (2018), https://doi.org/10.1016/j.technovation.2018.07.006

Technovation xxx (xxxx) xxx–xxx

G. Cabaleiro

2000a; Hagedoorn, 2002; Somaya et al., 2010). Despite this surge in popularity, establishing a licensing agreement is not an easy task. Markets for technology are characterized mainly by asymmetric information between parties, difficulties in describing and valuing the technology, uncertainty about the validity and applicability of the traded technology, and the risk of opportunistic behavior by licensees (Arora et al., 2001). Thus, it is reasonable to expect that parties do not equally appropriate the total profits generated by the agreement. Therefore, the output of any negotiation depends on the specific characteristics of the company, sector, and industry in which each licensing agreement is signed. However, the existing research focuses on analyzing licensing more as a value-created mechanism (Anand and Khanna, 2000b)—understood as an agreement that generates common benefits to the parties—than as a value appropriation mechanism—understood as an agreement from which the parties distribute the common benefits unequally (Latvie, 2007). In that sense, the existing research thoroughly analyzes the strategic advantages of establishing these contracts (Telesio, 1979; Katz and Shapiro, 1985; Shepard, 1987; Gallini, 1984; Farrel and Gallini, 1988; Rockett, 1990b; Lei and Slocum, 1991; Lowe and Taylor, 1998; Shapiro, 2001). It also tends to focus either on explaining, through the use of surveys, that licensing revenues are a main motivation for companies to license out (Gambardella et al., 2007; Robbins, 2006; Zuniga and Guellec, 2009) or on quantifying the extent of markets for technology at an aggregate level (Arora et al., 2001; Sheehan et al., 2004; Robbins, 2006; Athreye and Cantwell, 2007). Nevertheless, even though most firms fail to capture monetary value from licensing (Bianchi and Lejarraga, 2016), research on the ability of the parties to take possession of and extract value from this practice is scarce. As stated earlier, Walter (2012), Bianchi et al. (2014) and Bianchi and Letarraga (2016) appear to offer the only analyses of the appropriation capacity of the involved parties. Specifically, analyzing 11 years of licensing in the U.S. computer and pharmaceutical sectors, Walter (2012) uses abnormal stock market returns to distinguish the impact of licensing on the licensor from that on the licensee. The results show that the average cumulative abnormal returns for licensors are almost double those for licensees (2.00% vs. 1.06%). In addition, Walter points out four determinants that have an impact on the stock market: firm size (positive effect); R&D intensity (positive); business relatedness (negative); and industry (varying). On the other hand, Bianchi et al. (2014), using data from the Spanish Business Strategy Survey (SBSS), analyze a 2003–2007 dataset of 733 manufacturing firms; they conclude that the high intensity of marketing (measured as the ratio of money invested in advertising and promotion over total sales) and relational resources (measured as the firm's participation in technological collaborations with universities, clients, suppliers and competitors) reinforce the positive impact of technological resources (measured as the ratio of the number of euros invested in R&D over total sales) on licensors’ revenues. Also using data from the SBSS, Bianchi and Lejarraga (2016) analyze a ten-year (1998–2007) panel of 2014 Spanish firms belonging to 16 different manufacturing sectors. Their results show that prior experience in licensing positively affects the capacity to extract licensing revenues at a decreasing rate. Moreover, they find that the workforce's skills positively moderate the previous relationship, and, therefore, learning effects are larger for firms with a higher proportion of highly skilled employees. Considering the strategic corporate implications of knowing how to capture monetary value from licensing, it seems very important to more deeply analyze the factors that determine appropriation levels at the time of establishing a licensing agreement. Such knowledge would help managers take better advantage of their technologies and be conscious of the impact of announcing a licensing agreement. Accordingly, this paper analyzes how licensors’ appropriation capacity varies under different contingencies at the time that they announce a licensing agreement. This appropriation capacity should depend on two main factors: bargaining power and the potential cost of

previous research demonstrates that the initial stock market response to a key event is positively correlated with the company's long-term performance and the value of the event (Kale et al., 2002). Third, managers’ compensation is usually linked to the company's market value. As a result, in order to maintain their jobs and their reputations and to receive higher salaries, managers will not make any strategic decision that could negatively influence this variable. Fourth, it is necessary for companies to understand the signal that the market perceives at the time of licensing their technology: if a company licenses out its technology, does the market view that as a signal of superior technology or as a signal that the company is incapable of commercializing the technology by itself? This study uses a sample of 260 licensing agreements between U.S. high-tech public companies from 1998 to 2009 to test its hypotheses. The licensing data come from four different sources: PROMT, Google, Highbeam Research, and SDC Platinum. An event study with the software Eventus—with a two-day event window (−1,0) around the announcement dates—is used to analyze the impact on licensors’ market value at the time of announcing the contract. The study follows Flammer's (2013) procedure to obtain the average CARs, computing abnormal returns using the market model; estimating the coefficients with ordinary least squares based on 200 trading days [− 240, − 41]; and employing the daily return of the equally weighted market portfolio as a reference. To determine whether the proposed contingencies have different impacts on licensors’ appropriation capacity, six subsamples are designated. The (−1,1) and the (−3,3) event windows under the market model are used for robustness checks, and the marketadjusted model and the original model with a different reference portfolio—the value-weighted index portfolio—are also applied to check whether the results were influenced by the choice of the model or the choice of the reference portfolio. In addition, as the second hypothesis provides insights regarding the choice of the sector and the third one regarding the relative position of the company in the sector, the study explores their combined effect on CARs, considering four additional cases. The results of this study show that: (1) companies with cash constraints appropriate fewer benefits from licensing than do companies with no cash problems in terms of market value (0.87% vs. 2.84%); (2) companies that license out under low information asymmetries (same sector) benefit more from licensing than do companies that license out under high information asymmetries (different sector), in terms of market value (7.27% vs. 1.53%); (3) leaders in the sector benefit much less from licensing, in terms of market value, than do companies that follow (0.61% vs. 3.86%); (4) in the specific context of licensing out in the same sector, leaders appropriate much less value than do companies that follow (0.73% vs. 8.46%); and (5) the difference between being a leader and being a follower is less important when a company licenses out to a different sector (0.59% vs. 2.19%). The remainder of this paper is organized as follows. Section 2 presents the literature review and hypothesis development. Section 3 describes the data, variables and methodology. Section 4 reports the results and robustness checks, and Section 5 provides a discussion of the results and their managerial implications. Section 6 presents the study's conclusions. 2. Literature review and hypothesis development A new way to manage R&D innovations has emerged over the last decades. With strong competition in the product market, shorter product life cycles, and robust growth in information and communication technologies, companies cannot produce everything by themselves (Zuniga and Guellec, 2009). Increasingly, they must trust networks, new entrants, and technology-based firms if they want to remain efficient and competitive. Thus, licensing agreements are much more common and are now the most important method for commercializing and diffusing new technologies outside the firm (Anand and Khanna, 2

Technovation xxx (xxxx) xxx–xxx

G. Cabaleiro

imitation. The former should be a function of the existence of financial constraints in the company and the presence of information asymmetries between the parties. Since, with more bargaining power, the licensor can demand more rewards in the agreement (Kim, 1997; Mannix, 1993a; Pinkley et al., 1994), one might expect this contingency to increase the licensor's appropriation capacity. However, because the cost of imitation is a function of the licensor's relative position in a sector (leader vs. follower), it is more reasonable to expect that the contingency will hinder the licensor's appropriation capacity in terms of market value.

licensors and investors should, then, have more confidence in the selection decision because they possess good information about the technology, the company, and the sector in which they operate (Balakrishna and Koza, 1993). However, if companies license out to a company in a different sector, then licensing represents a way to enter a new market. With this option, companies can recover returns on their innovative effort without having to invest additional time or money to develop new sector-specific assets and knowledge. Yet in this case, licensors’ bargaining power is also much lower, on average. They face greater information asymmetries and uncertainty because they do not know the new sector, do not fully understand how the technology will be used, and lack any previous experience to help them anticipate opportunistic behavior. Therefore, when companies function in different sectors, and the licensor seeks to expand its technology to another market, a shift in bargaining power occurs and licensors become dependent on licensees to generate returns. Their other option is more time-consuming and requires a greater monetary investment. Thus, following the Bargaining Power Theory—which states that the higher the information, the higher the bargaining power at the time of the negotiation (Inkpen and Beamish, 1997)—licensors should appropriate more value from licensing out when they deal with a company within the same sector. If they deal with a different sector, they face uncertainty and information asymmetries in relation to the sector, the technology, and the company. In that case, licensors depend more on the partner's efforts to extract some benefit from their technology. Accordingly, Hypothesis 2 states that, because of their greater bargaining power, companies that license out to a company in the same sector appropriate more value than do companies that license out to a company in a different sector.

2.1. Bargaining power: cash constraints Consider a company that has successfully developed a product innovation. Under “normal” circumstances, this company can choose from at least two options: entering the product market (commercializing the final product) or licensing out the innovation to another company. Thus, companies should compare the benefits of developing the final product on their own (e.g., revenues from sales in the product market, less the costs of acquiring complementary assets, hiring personnel, and buying necessary raw materials) against the benefits of licensing out (e.g., licensing payments, less a potential decrease in market share by creating more competition in the product market). The optimal decision is the one that maximizes the net benefits resulting from each option (Gans et al., 2003). However, companies with cash constraints usually do not have these two options. Since they lack the necessary financial resources, they can rarely develop the product on their own, and, thus, licensing out is often their only option for recovering their investments in R&D. As the company's bargaining power relates positively to its number of options (Fisher and Ury, 1981), and a company with cash constraints has lower bargaining power at the time of negotiation (Lerner and Merges, 1998), a company that has just one option because it is cashconstrained would be expected to have less bargaining power. In particular, the assumption is that, under such weak bargaining power, TIOLI (“Take It Or Leave It”) offers by the other party are common; and, as licensors do not have enough power to enforce the economic conditions, they end up accepting the conditions imposed by the licensee. Therefore, based on the latter arguments, companies facing cash problems are likely to suffer from less bargaining power, and, as the stock market reflects their exact position in the negotiation, they appropriate less value from licensing than do companies without cash constraints.

H2. All else being equal, licensors that license out their technology to companies in the same sector capture greater abnormal market returns from licensing out technology than do licensors that license out to companies in other sectors. However, even though licensing out to a company in the same sector could imply a higher appropriation capacity in the short term, it is important to note that there is also ample evidence that warns about the potential long-term negative effect of licensing out technology in the same sector. This effect, known as the Rent Dissipation Effect, refers to the reduction of market share as a consequence of the additional competition in the product market (Arora and Fosfuri, 2003; Fosfuri, 2006). The previous research suggests some strategies to limit the extent of this latter effect such as: licensing out technology based on scientific knowledge (Arora and Gambardella, 1994); licensing out intellectual property related to non-core technologies and targeted toward geographically separated markets (Granstrand et al., 1997); licensing out intellectual property that refers to general technologies (Bresnahan and Gambardella, 1998); licensing out when patent protections are strong (Arora and Ceccagnoli, 2006; Cohen et al., 2000); when competition in the product market is high (Arora and Fosfuri, 2003); and when the firm's market share is small (Fosfuri, 2006). Despite these possible strategies, additional competition can still be a real threat to the company. Therefore, when companies license out in the same sector, they should, more than ever, balance two different forces that form the licensing trade-off. On the one hand, the licensor might appropriate more benefits in the short term because it has greater bargaining power as a consequence of the greater information, and this will be reflected in the negotiation and in the market value. On the other hand, in the long term, the licensor might experience a reduction in the market share because of the additional competition in the product market, and the stock market will reflect this negative effect on the company's market value. As these two forces are implicit in the establishment of licensing agreements, is important to check the data to determine which effect predominates.

H1. All else being equal, licensors with short-term cash problems capture lower abnormal market returns from licensing out technology than do licensors that face no cash constraints. 2.2. Bargaining power: information asymmetries Markets for technology are characterized by asymmetric information between parties, difficulties describing and valuing the focal technology, and uncertainty about the validity and applicability of the traded technology (Arora and Gambardella, 2010; Hu et al., 2015). However, the influence of asymmetric information on the licensor's bargaining power varies when the licensing agreement includes companies that belong to the same sector (low information asymmetry) or different sectors (high information asymmetry). Thus, researchers generally analyze these two cases separately (e.g., Walter, 2012; Müllez- Seitz, 2012). The rationale for this distinction is that, on average, asymmetric information diminishes significantly if the licensor belongs to the same sector as the licensee. First, companies in the same sector can more readily detect opportunistic behavior, understand the agreement and its potential, and establish and communicate a common objective (Koh and Venkatraman, 1991). Second, the similarity between companies in a sector suggests lower uncertainty and lower transaction costs. Both 3

Technovation xxx (xxxx) xxx–xxx

G. Cabaleiro

assets to license. The second step—considering the easy access to information and the size of the market for technology—was to narrow the initial sample to U.S. companies. Licensing data per se are difficult to find, and this search process becomes nearly impossible in countries in which information about companies is less accessible and markets for technology are smaller. Third, because the stock market value of the company is used as a proxy for appropriation capacity, I retain only those agreements in which both companies are public. Fourth, because they do not fit with the theoretical focus of this paper and could have totally different impacts on market value, cross-licensing agreements, agreements regarding standards and agreements enforced by a settlement are eliminated. Unfortunately, these features exactly characterize many innovative U.S. companies, which means that the results are less generalizable. However, due to the scarce evidence on this topic, this setting is the right one to improve the understanding of an important and unexplored relationship: how licensors’ appropriation capacity varies in different situations. The licensing data come from four sources: Predicast Overview of Markets for Technology (PROMT) database; HighBeam Research; Google News; and SDC Platinum. The first three sources are search engines that offer news from business and trade journals, industry newsletters, newspapers, market research studies, and news releases, while the last is an established licensing database. To extract licensing agreements from the PROMT, Google News and HighBeam Research databases the following key words were used: “licensing agreement” plus the company name or “licens” plus the company name. All of the results from PROMPT and HighBeam Research were examined; however, the results from Google News were relevant only up to page 20 because, after that, all the licensing agreements were repeated, and the search did not provide any new information. Once it was corroborated that each announcement referred to a license agreement, the information was collected and codified. The announcement date (if the same agreement listed different dates, the earliest one was selected), the licensor's name, and the licensee's name could be extracted from these stories. These licensing agreements were next appended to those obtained from SCD Platinum, which provided information about joint ventures, marketing agreements, and licensing agreements, among other contract forms. As Anand and Khanna (2000a) show, information from SCD regarding contract dates is highly inaccurate, so each date had to be checked manually, using the names of the parties involved, with a Google search for “name of company 1” “name of company 2” “license agreement.” The extent of inaccuracy in the sample is not as great as that reported by Anand and Khanna (2000a). The date variance is from one to ten days, probably due to the different criteria used to select dates. However, since, in an event study, data accuracy is critical and the window used is very narrow, to avoid any biases, the first possible date was always chosen. Next, corresponding firm-level identifiers (gvkey and permno) were sought, using the licensor's and licensee's names. Then, licensing data were matched with Compustat financial data and CRSP stock market data. The final licensing output featured 260 licensing agreements during 1998–2009. A potential concern about this data collection method is the possibility that it did not capture the entire universe of licensing agreements. In general, licensing agreements are private and confidential. Companies do not have to report licensing agreements on their income statements, and even in countries with regulatory reporting requirements, they refer only to cross-border transactions. Therefore, the collection method might not have captured all of the licensing agreements established by these companies. However, focusing on companies from the same country and information environment, with the same characteristics (public companies with the most patents granted), reduced the chances of obtaining more news from one company than from another.

2.3. Costs of imitation: leader/follower Through their licensing agreements, licensors grant access to their proprietary technology and allow licensees to use it. After understanding and internalizing how the licensed technology works, the licensees can invent around the technology, imitate licensors, and compete directly with them in the product market. This is called “Learning by Licensing” (LBL) (Wang et al., 2015). Such additional competition reduces the licensors’ benefits by an amount equal to the cost of imitation—that is, the difference between licensors’ benefits in the product market with and without imitation. The importance of this cost depends on the presence of the licensor in the product market. Arora and Fosfuri (2003) show that when market share before licensing out is small, the reduction implied by imitation is almost insignificant because each company internalizes just a small loss. In contrast, when companies have higher market shares, the impact of imitation can be very strong: they suffer a significant reduction in the market share or in the pricecost margin due to the additional competition in the product market. Therefore, the potential cost of imitation is higher for leaders than for followers in a specific sector. Leaders usually have the largest market share in the sector, along with accumulated experience in their business and huge investments in fixed assets. Because they often own complementary assets, leaders can benefit from economies of scale in R &D, achieve faster learning curves, and understand how to commercialize the technology. In this sense, a significant portion of company benefits depends on leaders’ activity in the product market, so the impact of imitation can be very negative in terms of competitive advantage. Followers, instead, tend to have small market shares and lack both the downstream assets to commercialize the final product (e.g., distribution channel, marketing, manufacturing) and the legitimacy in the marketplace to leverage incurred R&D expenditures. Therefore, licensing may represent the only way to recover their previous investments (Teece, 1986; Kollmer and Dowling, 2004), establish relationships with large companies, and enhance their reputation (Teece, 1986; Stuart et al., 1999; Shane and Venkataraman, 2000). Followers have incentives to license out their technology since they have much to gain from licensing and little to lose in the product market. Because leaders have more to lose from licensing than followers do, the prediction here is that they will appropriate fewer benefits from licensing, in terms of market value, than followers will. H3. All else being equal, licensors that lead a specific sector capture fewer cumulative abnormal returns from licensing out technology than do licensors that are followers. 3. Data, variables and methodology 3.1. Sample selection and data The empirical analysis is based on a sample of innovative U.S. companies, selected according to several criteria. The first step was to identify 140 companies with the most patents granted by the U.S. Patent and Trademark Office (USPTO) during the almost-20-year period from 1990 to 2009. The length of the period was selected to ensure that a patent granted to a company by the USPTO in 1990 would still be valid at the end the study period. Eighty percent of the companies selected belong to the following sectors: Electronic and Other Electrical Equipment Except Computer Equipment (SIC2:36); Industrial and Commercial Machinery and Computer Equipment (SIC2:35); Chemicals & Allied Products (SIC2:28); Measuring, Analyzing and Controlling Instruments/Photographic, Medical and Optical Goods/Watches and Clocks (SIC2:38); Business Services (SIC2:73); and Rubber & Miscellaneous Plastics Products (SIC2:30) (Table 1 provides more detailed information.) Because the main objective of this paper is to analyze how licensors’ appropriation capacity varies in terms of market value, the focus is on potential licensors—namely, companies with technological 4

Technovation xxx (xxxx) xxx–xxx

G. Cabaleiro

Table 1 SIC-2 of companies in the sample and patents granted by SIC-2 during the 20 years. Firms by SIC

Patents granted (In the 20 years, by SIC)

SIC-2

Sector

Percentage

Total patents

Percentage of patents granted by SIC

Mean patents granted by SIC

36 35 28 38

Electronic and other electrical equipment except computer equipment Industrial and commercial machinery and computer equipment Chemicals & allied products Measuring, analyzing and controlling instruments; photographic, medical and optical goods; watches and clocks Business services Rubber & miscellaneous plastics products Other sectors (29, 32, 34,37, 39, 48 AND 67)

22,9% 19,5% 18,4% 8,1%

117,433 24,675 26,576 5165

49,1% 10,3% 11,1% 2,2%

512.808 126.279 144.592 64.161

6,9% 3,5% 20,8%

20,769 12,168 4819

8,7% 5,1% 2,0%

301.000 352.696 23.191

73 30 29

Rˆ it = αi + βi *Rmt .

licensing agreements is used, and firms with a net income above and those with a net income below the median firm are separated: ni_previousyear_undermedian and ni_previousyear_abovemedian. Companies with a net income below the median are more financially constrained than companies with a net income above it. A dummy variable (SSIC) distinguishes between licensing agreements within the same or across different sectors; it is equal to 1 if the four-digit standard industrial classification (SIC) of the licensor is identical to the four-digit SIC of the licensee, and 0 otherwise. The fourdigit SIC code, acquired from the Compustat database, accounts for the division, the major group, and the industry group of each company. Finally, to classify each licensor as a leader or a follower in a specific sector, its market share is computed as the ratio between the licensor's net sales and total net sales in the licensor's industry. Industry net sales represent the sum of all net sales by companies operating in the same four-digit SIC code in year t. Following Giroud and Mueller (2011), the study includes all available Compustat firms in the same SIC code but excludes firms for which net sales are missing. Once the market share of each licensor in each year is computed, a dummy variable is used to distinguish between companies with high (LEADER) and low (FOLLOWER) market shares. This variable equals 1 if the licensor is one of the three companies with the highest market shares in the four-digit SIC code at year t, and 0 otherwise.

Next, the abnormal daily return (AR) of company i on day t is calculated as follows:

3.3. Methodology

3.2. Variable description 3.2.1. Dependent variable: cumulative abnormal returns (CARs) Following Kale et al. (2003) and Gulati et al. (2009) for each firm i the abnormal returns are computed using the market model (Fama et al., 1969), which assumes a stable linear relationship between market returns and returns on the financial instrument; accounts for market trends and firm risk; and improves the chances of isolating the effect of specific events (Campbell et al., 1997). To estimate the coefficients αi (average return of the firm compared with the market average) and βi (sensitivity of its return to the market return or risk of the stock), ordinary least squares (OLS) are used with the 200 trading days in the estimation, which correspond to the interval [− 240, − 41] according to daily return data from CRSP. Formally:

Rit = αi + βi *Rmt + eit where Rit is the return on the stock of company i on day t; αi is the intercept; βi is the systematic risk of stock I; Rmt is the daily return of the equally weighted CRSP market portfolio; and eit is the daily risk-adjusted residual for firm i. The corresponding estimated return on the stock for firm i on day t is given by:

ARit = Rit − Rˆ it .

3.3.1. Event study An event study is used to determine the stock price reaction to licensing announcements and to analyze the average CARs during the event window. These CARs capture how much the stock price deviates from its expected value on the day of the licensing announcement. This methodology relies on the assumption that stock markets are efficient and that prices perfectly reflect all public information related to the company's prospects. Thus, the effect of a specific event should be reflected almost immediately in the stock market. That is, when an event occurs, the market updates its forecast, causing a shift in market value. To avoid the inclusion of unrelated events that might influence stock returns, the event window must be sufficiently narrow (Gulati et al., 2009). A common approach sets the event day (day 0) as the day of the announcement and also considers the possibility that the event might have happened on the previous day, before the closing of the stock market (day −1) (MacKinlay, 1997). Previous research indicates that a two-day window is more effective than longer windows for capturing stock market reactions (Crutchley et al., 1991). However, a robustness check also considers the (−1,1) and (−3,3) windows. Prior studies of licensing (Anand and Khanna, 2000b; Walter, 2012), alliances (Merchant and Schendel, 2000; Kale et al., 2002; Gulati et al., 2009), and joint ventures (Balakrishna and Koza, 1993; Park and Kim, 1997; Reuer and Koza, 2000) employ CARs as an effective, marketbased measure of firm performance. Moreover, prior research finds a high correlation (around 40%) between this variable and long-term

The final computation—of the CARs for each time interval—is performed by summing up the abnormal returns within the specific time window [− 1,0].

⎛ CAR ⎜−1, ⎝

⎞ 0⎟ = ⎠

0

∑

ARt

t =−1

3.2.2. Key variables to define subsamples The study uses two variables to classify companies with cash constraints. The first, following the previous literature, is the current ratio (Ahuja, 2000; Bergh, 1997; Bergh and Lawless, 1998; Bromiley, 1991; Bolton, 1993; Chang and Singh, 1999; Dailey, 1995; Dailey and Dalton, 1994; Hambrick et al., 1996; Hitt et al., 1996; Hoskisson and Johnson, 1992). The current ratio indicates the company's ability to meet shortterm debt obligations, computed as current assets divided by current liabilities. If the firm's current assets are more than twice its current liabilities, the company is financially strong. If current liabilities exceed current assets, the company may have problems meeting its short-term obligations. Then, the observations are divided at the median into current ratio_undermedian and current ratio_abovemedian. The second variable is the measure proposed by Lerner and Malmendier (2010) to identify firms that are financially constrained. In particular, the net income of companies in the year prior to the establishment of the 5

Technovation xxx (xxxx) xxx–xxx

G. Cabaleiro

firm performance and the value of the event (Koh and Venkatraman, 1991; Kale et al., 2002). Thus, to examine whether the various contingencies proposed here have different impacts on the licensor's market value, the sample is divided into six cases, and the CARs for each are computed separately: licensors with and without cash constraints (H1); licensing agreements between companies that belong to the same or different sectors (H2); and licensors that are leaders or followers (H3). Next, the CARs are computed for four interactions in order to provide a deeper analysis: (1) licensing agreements between companies that belong to the same sector + leaders; (2) licensing agreements between companies that belong to the same sector + followers; (3) licensing agreements between companies that belong to different sectors + leaders; and (4) licensing agreements between companies that belong to different sectors + followers.

window equals 0.85. However, this does not present a problem because they are not used simultaneously. Table 4 contains a summary of the results for each daily return and the whole population of licensing agreements (N = 260), and Fig. 1 represents them graphically. On the day of the announcement (day 0), mean ARs are positive (2.46%) and statistically significant at the 1% level for all tests. By day 1, the mean of ARs remains positive (0.88%) and is statistically significant in the parametric tests. However, no significant effects emerge in the days immediately before or after. As Table 5 shows the mean CARs variable, for the three event windows that contains the day of the licensing agreement—(−1,0), (−1,1) and (−3,3) — are positive and statistically significant for all tests at the 1% level. However, for the event window that does not contain the day of the announcement—(2, 30)—the mean CARs is positive but not significant. Table 6 contains the resulting mean CARs and test statistics for each subsample, using an event window of (−1,0). The precision weighted cumulative average abnormal return (CAAR), defined as a weighted average of the original CARs, provides a means to report an average standardized of cumulative abnormal return (average SCAR). For each group, the number of securities with positive and negative average abnormal returns is also indicated. In line with H1, regarding the effect of cash constraints, the mean CARs resulting from the announcement of a licensing agreement are lower when the licensor has cash limitations than when it does not, whether using the current ratio, “undermedian/abovemedian” (1.33% vs. 2.42%), or the net income from the previous year, “undermedian/ abovemedian” (0.87% vs 2.84%). The resulting mean CARs also are statistically significant in all the tests at least at the 5% level. There is also support for same-sector benefits, as predicted in H2. Specifically, the mean CARs resulting from announcing a licensing agreement increase when both companies belong to the same sector rather than to different sectors (7.27% vs. 1.53%). The positive mean CARs are statistically significant for both the Patell Z and CDA tests at the 0.1% level. For the Generalized Sign Z, announcements between companies in the same SIC are significant at the 0.1% level, and the mean CARs resulting from announcements between companies from different sectors are significant at the 1% level. Finally, the data corroborate H3, in that the mean CARs resulting from the announcement of licensing agreements are greater when licensors are followers instead of leaders in the sector (0.61% vs. 3.86%). For followers, the mean CARs are statistically significant under the Patell Z and CDA tests at the 0.1% level and under the Generalized Sign Z test at the 1% level. For leaders, they are statistically significant at the 5% level (Patell Z and Generalized Sign Z) or the 10% level (CDA). Table 7 provides the Student's T-test results to determine whether the difference in means between the groups is statistically significant. All of the mean differences are statistically significant at the 10% level.

3.3.2. Abnormal tests Three different tests determine whether the resulting mean CARs differ significantly from zero. Two of them are parametric—the Patell Z test (Patell, 1976) and the Crude Dependence Adjustment, CDA, (Brown, 1980; Brown and Warner, 1985)—and one is non-parametric—the Generalized Sign Z (Cowan, 1992). The parametric tests rely on the assumption that a firm's CARs are normally distributed, while the non-parametric test acknowledges that daily stock data are not normally distributed. Details regarding the tests are in the Appendix A. 3.3.3. Significance of the differences between groups After analyzing the significance of the mean CARs, a necessary next step is to discern whether the differences between the mean CARs of the compared groups are significant. The T-Statistic is used to compare the means between two groups. The description of this test is detailed in the Appendix A. The process is repeated for event windows equal to (−1,1) and (−3,3). 4. Results 4.1. Descriptive statistics and main results Table 1 provides information regarding the industries of the companies in the sample. Note that the table shows that the 80% of the companies selected belong to six different sectors: 1) the Electronic and Other Electrical Equipment Except Computer Equipment Sector (SIC2:36); 2) the Industrial and Commercial Machinery and Computer Equipment Sector (SIC2:35); 3) Chemicals & Allied Products (SIC2:28); 4) Measuring, Analyzing and Controlling Instruments/Photographic, Medical and Optical Goods/Watches and Clocks (SIC2:38); 5) Business Services (SIC2:73); 6) and Rubber & Miscellaneous Plastics Products (SIC2:30). Table 2 provides the descriptive statistics for the variables of interest. The table shows that the mean of CARs using a three-day event window (−1,1) is greater than the mean of CARs using a two-day event window (−1,0). It also demonstrates that the mean of market share equals 9.69%, showing that most of the companies do have a market share lower than 10%. Table 3 includes their correlations: note that the correlation between CARs using a two-day event and a three-day event

4.2. Robustness check As a robustness check, in Tables 8 and 9, the mean CARs for each subgroup are computed for two additional event windows, (−1,1) and (−3,3). The main results are again corroborated. However, one notable change is that the mean CARs for leaders are not significant in either the CDA test or the generalized sign Z test (previously significant at the 10% and the 5% level, respectively). These changes appear in both Tables 8 and 9. Also, different models are run in order to show that the results are not influenced either by the choice of the model or by the choice of the reference portfolio. These models are the Market Adjusted Model and the original model with a different reference portfolio—the Value Weighted Index Portfolio. Table 10 reports the results of the event study for each subsample using the Market Adjusted Returns Model and, as in the previous case, the Equally Weighted Index. Under this scenario, CARs for each subsample are positive and statistically significant under

Table 2 Descriptive statistics. Variable

Obs

Mean

Std. Dev.

Min

Max

CAR (−1,0) % CAR (−1,1) % Market Share Same SIC Current Ratio Net Income Previous Year

260 260 259 260 245 245

2.926858 3.689724 9.690217 0.1992337 3.935711 1839.909

15.96793 19.14492 19.09417 0.4001915 4.101579 4673.721

− 26.61 − 39.15 0 0 0.3871703 − 4244

227.13 225.84 92.13502 1 30.08364 36130

6

Technovation xxx (xxxx) xxx–xxx

G. Cabaleiro

Table 3 Correlations.

CAR (−1,0)% CAR (−1,1) % Market Share Same Sic Dummy Current Ratio Net Income

CAR (−1,0)%

CAR (−1,1) %

Market Share

Same SIC Dummy

Current Ratio

Net Income

1.0000 0.8547 − 0.0626 0.0957 − 0.0747 − 0.0724

1.0000 − 0.0754 0.0790 0.0391 − 0.0677

1.0000 − 0.1833 − 0.2984 0.2794

1.0000 0.0790 − 0.0724

1.0000 − 0.2164

1.0000

Table 4 Daily mean abnormal returns and test statistics for licensing announcements (N = 260). Day

N

Mean Abnormal Return

Positive: Negative

Patell Z

Portfolio Time-Series (CDA) t

Generalized Sign Z

− 30 − 29 − 28 − 27 − 26 − 25 − 24 − 23 − 22 − 21 − 20 − 19 − 18 − 17 − 16 − 15 − 14 − 13 − 12 − 11 − 10 −9 −8 −7 −6 −5 −4 −3 −2 −1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 259 259 259 259 259 259 259 259 259 259

− 0.13% 0.59% 0.04% − 0.11% − 0.08% 0.05% − 0.24% 0.44% 0.07% − 0.17% − 0.22% − 0.05% 0.59% − 0.02% 0.58% − 0.32% − 0.04% − 0.09% 0.32% − 0.05% 0.29% 0.33% 0.24% 0.34% − 0.12% 0.18% − 0.21% 0.35% − 0.37% 0.22% 2.46% 0.88% − 0.07% − 0.13% 0.08% − 0.13% 0.16% 0.08% 0.34% −0.53% 0.09% 0.24% 0.12% − 0.11% 0.16% 0.12% − 0.03% 0.05% 0.16% 0.18% − 0.04% − 0.07% − 0.04% − 0.09% − 0.01% 0.15% 0.14% 0.14% − 0.04% −0.25% −0.25%

121:139 135:125) 127:133 119:141 123:137 134:126) 115:145 132:128 128:132 121:139 122:138 122:138 123:137 125:135 135:125) 119:141 119:141 120:140 140:120 > 130:130 138:122 > 134:126) 122:138 127:133 120:140 125:135 123:137 133:127 107:153 < 126:134 154:106 > > > 118:142 126:134 119:141 121:139 116:144 125:135 110:150( 128:132 112:148( 143:117 > > 121:139 136:124) 130:130 128:132 135:125) 134:126) 120:140 120:140 128:132 135:125) 115:144 124:135 120:139 131:128 122:137 132:127 122:137 112:147( 116:143 118:141

− 0.362 1.941 * 0.678 − 1.009 − 0.186 1.322$ − 1.872 * 2.313 * − 0.118 − 2.176 * − 0.659 0.318 2.544 * * − 0.321 2.480 * * − 0.669 − 0.242 − 0.92 1.051 − 0.392 1.485$ 1.942 * 0.989 1.086 − 1.128 1.063 − 1.2 2.186 * − 1.654 * 0.937 9.615 * ** 3.210 * ** 0.496 − 0.974 0.458 − 0.009 0.502 0.323 2.869 ** −2.974 ** 1.514$ 1.472$ 1.194 − 0.35 − 0.309 1.607$ 0.338 0.205 1.054 1.232 0.086 − 0.067 − 0.543 0.552 1.136 0.689 0.882 0.586 0.467 −0.486 −0.97

− 0.518 2.379 * * 0.173 − 0.443 − 0.327 0.219 − 0.976 1.775 * 0.284 − 0.692 − 0.913 − 0.22 2.401 * * − 0.079 2.366 * * − 1.284$ − 0.182 − 0.357 1.286$ − 0.184 1.191 1.355$ 0.994 1.384$ − 0.5 0.735 − 0.843 1.404$ − 1.493$ 0.897 10.004 * ** 3.577 * ** − 0.299 − 0.523 0.315 − 0.537 0.657 0.34 1.398$ −2.153 * 0.346 0.979 0.481 − 0.461 0.634 0.498 − 0.113 0.213 0.663 0.735 − 0.146 − 0.29 − 0.178 − 0.373 − 0.057 0.621 0.559 0.556 −0.175 −1.037 −1.025

− 0.25 1.489$ 0.495 − 0.499 − 0.002 1.365$ − 0.995 1.116 0.619 − 0.25 − 0.126 − 0.126 − 0.002 0.247 1.489$ − 0.499 − 0.499 − 0.374 2.110 * 0.868 1.861 * 1.365$ − 0.126 0.495 − 0.374 0.247 − 0.002 1.24 − 1.989 * 0.371 3.849 * ** − 0.623 0.371 − 0.499 − 0.25 − 0.871 0.247 −1.617$ 0.619 −1.368$ 2.483 ** −0.25 1.613$ 0.868 0.619 1.489$ 1.365$ − 0.374 − 0.374 0.619 1.489$ − 0.938 0.182 − 0.316 1.053 − 0.067 1.177 − 0.067 −1.312$ −0.814 −0.565

7

Technovation xxx (xxxx) xxx–xxx

G. Cabaleiro

the three significance tests. Table 11 reports the results of the event study for each subsample, using, as in the first case, the Market Model and, as a novel addition, the Value Weighted Index. In this scenario, CARs are also positive and statistically significant under the three significance tests. Therefore, the results are not affected by the choice of either the model or the reference portfolio.

now they lack any bargaining power. Licensors that are leaders suffer both the information asymmetry and the potential cost of imitation. This scenario is relatively unlikely, as licensees incur massive investments to compete directly with licensors, but licensees can plausibly imitate licensors, which reduces the licensors’ benefits. Table 12 shows that, following the Market Model and using the Equally Weighted Index, the mean CARs of leaders are much lower than those of followers when they license in the same sector (0.73% vs. 8.46%), though the impact is statistically significant only when licensors are followers. Furthermore, the difference between the mean CARs when licensors are leaders versus followers is much lower if they belong to different sectors (0.59% and 2.19% vs. 0.73% and 8.46%). Therefore, the best situation in which to appropriate benefits, in terms of market value, is when licensees are followers and belong to the same sector as licensors (CARs=8.46%). On the contrary, appropriation is very low when licensor and licensee belong to different sectors and licensors are leaders (CARs=0.73%). This conclusion is still corroborated under two additional specifications: 1) the Market Adjusted Returns Model, using the Equally Weighted Index Model; and 2) the Market Model, using the Value Weighted Index.

4.3. Interaction effects

5. Discussion, implications and limitations

Results regarding Hypotheses 2 and 3 show that companies that license out to a company in the same sector appropriate more (H2) and that leaders appropriate less than followers (H3). However, as the former hypothesis offers insights into the choice of the sector and the latter one into the relative position of the company in the sector, these two supported hypotheses are not mutually exclusive. Rather, together, they represent the licensing trade-off that Fosfuri (2006) describes. On the one hand, companies that license out technology in their own sector appropriate more than companies that license out to a different sector because of the lower information asymmetries. On the other hand, the potential cost of imitation is higher if the companies are in the same sector: given the similarities of companies in the same sector, it is easier for companies to invent around the licensed technology and to imitate the licensor in the product market, thus reducing the licensor's appropriation capacity. Therefore, the appropriation capacity depends not only on the licensee's sector, but also on the licensee's sector and the licensor's relative position in the product market. Four cases are considered to explore their combined effect on CARs. First, among companies that license to same-sector companies, the expectation is that followers appropriate more value from licensing than leaders appropriate. Recall that followers have little to lose due to imitation by competitors, and they already have some bargaining power because they are negotiating in a context marked by low information asymmetries. Although leaders have similar bargaining power due to low information asymmetries, the cost of imitation can be high, so they likely appropriate fewer benefits from licensing. Second, for companies that license out to different sectors, the expectation is, again, that followers appropriate more value than leaders do. However, this difference should not be as pronounced as in the previous case because all licensors confront high information asymmetries and depend on their licensees to earn benefits from their previous investments. Followers still have little to lose if licensees decide to imitate them, but

This paper advances the analysis of the appropriation capacity of the parties involved in licensing agreements by analyzing how a licensor's market value varies at the time that it announces a licensing agreement under different company, sector and industry factors. Based on an event study, the resulting CARs lead to the following results: (1) companies with cash constraints appropriate fewer benefits from licensing than do companies with no cash problems in terms of market value (1.10% vs. 2.65%); (2) companies that license out under low information asymmetries (same sector) benefit more from licensing than do companies that license out under high information asymmetries (different sector), in terms of market value (7.27% vs. 1.53%); (3) leaders in the sector benefit much less from licensing, in terms of market value, than do companies that follow (0.61% vs. 3.86%); (4) in the specific context of licensing out in the same sector, leaders appropriate much less value than do companies that follow (0.73% vs. 8.46%); and (5) the difference between being a leader and being a follower is less important when a company licenses out to a different sector (0.59% vs. 2.19%).

Fig. 1. Mean Abnormal Returns from 30 days before the licensing announcement until 30 days after the licensing announcement.

5.1. Implications for theory This study sheds new light on some important and relatively unexplored topics. In particular, it provides more evidence on the value of appropriation in licensing agreements. The previous research in licensing focuses more on the common benefits that licensing generates to the parties than on their relative distribution. Based on a literature review, it appears that only Walter (2012), Bianchi et al. (2014) and Bianchi and Lejarraga (2016) study how the influence of different determinants changes the appropriation capacity of either or both parties involved in licensing agreements. Furthermore, the current study also explicates bargaining power in the context of licensing agreements by proposing and confirming that it is a function of two factors: cash

Table 5 Mean cumulative abnormal returns and test statistics for licensing announcement (N = 260): market model, equally weighted index. Days

N

Mean Cumulative Abnormal Return

Precision Weighted CAAR

Positive: Negative

Patell Z

Portfolio Time-Series (CDA) t

Generalized Sign Z

(−1,0) (−1,+1) (−3,+3) (+2,+30)

260 260 260 260

2.68% 3.56% 3.34% 0.40%

1.60% 2.09% 2.10% 1.66%

151:109 > > > 155:105 > > > 148:112 > > > 136:124)

7.462 * ** 7.946 * ** 5.222 * ** 2.040 *

7.709 * ** 8.359 * ** 5.128 * ** 0.303

3.476 * ** 3.973 * ** 3.104 * ** 1.613$

Notes: The symbols $,* ,* *, and * ** denote statistical significance at the 0.10, 0.05, 0.01, and 0.001 levels, respectively, using a generic one-tailed test. The symbols (, < or), > etc. correspond to $,* and show the direction and significance of the generalized sign test. 8

Technovation xxx (xxxx) xxx–xxx

G. Cabaleiro

Table 6 Mean cumulative abnormal returns and test statistics for licensing announcement divided into groups: market model, equally weighted index (−1,0). GROUP

N

Mean Cumulative Abnormal Return

Precision Weighted CAAR

Positive: Negative

Patell Z

Portfolio Time-Series (CDA) t

Generalized Sign Z

Total Licensors Current Ratio undermedian Current Ratio abovemedian NI Previous Year undermedian NI Previous Year abovemedian Same Sic Diff Sic Leader Follower

260 123 122 123

2.68% 1.33% 2.42% 0.87%

1.60% 0.96% 2.01% 0.76%

151:109 > > > 75:48 > > 68:54 > 75:48 > >

7.462 * ** 4.030 * ** 4.437 * ** 3.265 * **

7.709 * ** 3.791 * ** 3.986 * ** 2.351 * *

3.476 * ** 2.896 * * 2.031 * 2.930 * *

122

2.84%

2.43%

69:53 >

2.910 * *

4.798 * **

2099 *

52 208 93 166

7.27% 1.53% 0.61% 3.86%

3.74% 1.09% 0.52% 2.73%

34:18 > > 117:91 > > 53:40 > 97:69 > >

7.506 * ** 4.591 * ** 2.082 * 7.777 * **

9.117 * ** 4.326 * ** 1.567$ 7.905 * **

2.633 * ** 2.57 * * 1.65 * 3.035 * *

Notes: The symbols $,* ,* *, and * ** denote statistical significance at the 0.10, 0.05, 0.01, and 0.001 levels, respectively, using a generic one-tailed test. The symbols (, < or), > etc. correspond to $,* and show the direction and significance of the generalized sign test.

context in which the company has weak bargaining power; thus, the market anticipates the consequences of the deal and barely increases the market value—as compared with a situation in which the licensor is not cash-constrained. Therefore, managers should realize that licensing decisions in cash-constrained situations might generate short-term licensing revenues (i.e., lump-sum payments, upfront fees, etc.), but that they will appropriate fewer benefits in terms of market value. Second, managers should know that the stock market responds better to a licensing decision if the licensee belongs to the same sector. Another same-sector company's desire to license the technology signals the superiority of the licensor, the inability of the licensee to compete with the technology, and/or the likely standardization of this technology throughout the sector. Therefore, companies can increase their benefits by licensing in the same sector, earning immediate revenues and achieving a heightened stock market value. However, this significant stock market increase occurs only if the risk of imitation from licensing is not too high. For managers of firms that are leaders in their sector, this study reveals that they might increase short-term benefits with licensing revenues (i.e., lump-sum fees); however, since the stock market anticipates that, in the long term, companies might suffer reduced benefits due to the cost of imitation, the impact of licensing on the stock market is almost nonexistent. Therefore, for companies that are leaders, it is better not to license out unless the licensing payments exceed the potential cost of imitation. But for managers of companies that are followers, this paper suggests licensing out their proprietary technologies, as followers that are able to license out are sending a signal of reputation, quality and prospects for growth. Hence, the managers of companies that are followers will increase their benefits not only through the licensing revenues established in the contract, but also through an increase in market value. In addition, as followers usually have small market shares, the potential cost of imitation is almost nonexistent. In conclusion, this paper reveals that the company that is most likely to appropriate benefits in terms of market value is a follower that licenses out its technology to a company in the same sector.

constraints and information asymmetries. No previous study appears to provide empirical evidence of the relationship between bargaining power and the licensing's outcomes. In addition, this article highlights another potential explanation of why large firms are not really active in the market for technology. In particular, the previous literature that explores the relationship between firm size and the propensity to license presents controversial results: while some studies find a negative relationship between the two variables (Arora and Fosfuri, 2003; Gambardella et al., 2007), others describe a U-shaped relationship (Fosfuri, 2006; Zuniga and Guellec, 2009 and Kani and Motohashi, 2012). This paper provides an explanation other than the lack/ownership of complementary assets—namely, the small effect on their market value obtained from licensing out (CARs = 0.66% for the leaders with a high market share versus 3.86% for the followers). Finally, this study provides empirical evidence—through market value variations—of how both effects of the licensing trade-off interact. Specifically, by observing licensors’ market value, it was possible to determine how their appropriation capacity increases with contingencies that determine the bargaining power but declines when there is a high potential cost of imitation. 5.2. Implications for practice This study provides some strategic insights for managers. In recent years, companies’ stock market measures have increased in importance. External observers use them to proxy for firms’ future performance, and boards of directors use them to evaluate and compensate managers. Therefore, in addition to maintaining their jobs, earning higher salaries, and enhancing their reputations, managers also want to make decisions that improve stock market measures. This paper's analysis of the situations that generate stronger/weaker impacts on the stock market offers managers some guidance in making strategic decisions. First, they should know that the stock market efficiently captures potential strategic reasons for licensing. For instance, a company that is cashconstrained might decide to license out its technology to increase its short-term earnings (e.g., by requiring upfront fees or lump sum payments). However, the market is conscious of the company's cash-constrained situation and recognizes that the negotiation takes place in a

5.3. Limitations This paper does have some limitations. From an empirical point of

Table 7 Cumulative abnormal returns: differences of means. two-sample T-statistic.

Current Ratio Undermedian - Current Ratio Abovemedian NI PreviousYear Undermedian - NI PreviosYear Abovemedian Same Sic - Diff Sic Follower - Leader

Mean(Diff)

Std. Err. (Diff)

95% Conf. Interval (Diff)

T-statistic (Diff)

0.226 2.684 3.693 − 3.524

0.0539073 1.081807 1.559646 1.229588

0.0763293 − 0.3195773 − 0.63827 − 6.937883

4.1924 * 2.4810 * 2.3672 * − 2.8660 *

The symbols $,* ,* *, and * ** denote statistical significance at the 0.10, 0.05, 0.01, and 0.001 levels. 9

0.3756707 5.687577 8.022227 − 0.1101171

Technovation xxx (xxxx) xxx–xxx

G. Cabaleiro

Table 8 Mean Cumulative Abnormal Returns and Test Statistics for Licensing Announcement Divided into Groups: Market Model, Equally Weighted Index (−1,1). GROUP

Mean Cumulative Abnormal Return

Precision Weighted CAAR

Positive: Negative

Patell Z

Portfolio Time-Series (CDA) t

Generalized Sign Z

Total Licensors Current Ratio undermedian Current Ratio abovemedian NI Previous Year undermedian NI Previous Year abovemedian Same Sic Diff Sic Leader Follower

3.56% 1.52% 4.23% 0.97% 4.96% 8.52% 2.32% 0.47% 5.30%

2.09% 1.10% 3.44% 0.86% 3.81% 4.84% 1.43% 0.46% 3.79%

155:105 > > > 75:48 > > 75:47 > > > 77:46 > > > 73:49 > > 38:14 > > > 117:91 > > 50:43:00 104:62 > > >

7.946 * ** 3.766 * ** 6.205 * ** 3.235 * ** 3.472 * ** 7.924 * ** 4.923 * ** 1.499$ 8.808 * **

8.359 * ** 3.536 * ** 5.702 * ** 2.177 * * 4.080 * ** 8.721 * ** 5.347 * ** 0.984 8.873 * **

3.973 * ** 2.869 * * 3.302 * ** 3.134 * ** 3.162 * ** 3.745 * ** 2.57 * * 1.028 4.124 * **

Notes: The symbols $,* ,* *, and * ** denote statistical significance at the 0.10, 0.05, 0.01, and 0.001 levels, respectively, using a generic one-tailed test. The symbols (, < or), > etc. correspond to $,* and show the direction and significance of the generalized sign test.

common; (3) operate in the most developed (U.S.) environment for markets for technology; and (4) are public companies, subject to the pressures of the stock market. The results do not apply to companies that do not satisfy these characteristics. From a theoretical point of view, there are other potential explanations besides bargaining power that can justify the significant difference in CARs. For instance, the fact that companies with cash constraints have lower CARs when licensing out than do companies without such restraints may be related to the potential of the licensed technology in the actual company. That is, companies without cash constraints license out technology only when outside companies can exploit the technology better than they can. However, it is possible that companies with cash constraints license out under both situations: when they can exploit the technology inside their company and when other companies can do it better. The assumption in this paper is that the market reflects all public information that is important for the prospects of the company, and, thus, that should reflect the fact that the company can better exploit the technology on its own instead is licensing it out. Hence, this lower impact on the stock market may be the consequence of licensing out technology that should not be licensed rather than the consequence of lower bargaining power. In addition, the fact that licensing out in the same sector versus a different sector has a greater impact on the stock market might be explained by the presence/ absence of complementary assets and capabilities. Since companies that belong to the same sector have similar capabilities and complementary assets, the market can assume that the technology will be better exploited when it is licensed out in the same sector. Therefore, the market can react more positively because of the higher “potential” of the deal and not because of the bargaining power involved. However, despite the existence of alternative causes, this study is not able to isolate both effects.

view, the licensor's appropriation capacity measure does not capture all the benefits that result from licensing. In particular, it is measured as the percentage increase in the licensors’ market value. That measurement relies on the assumptions that markets are totally efficient and that the market value properly reflects all of the public information available about each company. However, licensing usually creates other benefits that the licensors also appropriate: lump-sum fees, royalties, and so on. As this kind of economic information is not available, it is not possible in this study to capture all of the benefits appropriated by the licensor. Furthermore, a potential concern regarding the data collection is the possibility that the data do not capture the entire history of licensing agreements for all the companies selected. However, one might expect the paper's focus on companies from the same country and with the same characteristics (public companies with most patents granted) to reduce the chances of obtaining more news from one company than from another. As Schilling (2009, p. 258) claimed, “Even though each database only captures a sample of alliance activity, it may yield reliable results for many—if not all—purposes.” However, the fact that it is not possible to observe all the licensing agreements creates limits for this study. Specifically, if not all of the licensing agreements can be observed, then neither can all the increases in the stock market. As a consequence, the normal returns used to make the prediction should be greater than in the case of identifying the whole population, and, therefore, obtaining significant abnormal returns should be more difficult. In addition, having no control sample (companies that could have licensed out but did not) can also introduce some biases. Finally, with regard to the generality of the results, they cannot be applied to wider populations of companies. Licensing has still not spread to all industries or countries. Rather, the companies for which these findings hold (1) have more patents to trade in markets for technology; (2) belong to industries in which licensing agreements are

Table 9 Mean Cumulative Abnormal Returns and Test Statistics for Licensing Announcement Divided in Groups: Market Model, Equally Weighted Index (−3,3). GROUP

N

Mean Cumulative Abnormal Return

Precision Weighted CAAR

Positive: Negative

Patell Z

Portfolio Time-Series (CDA) t

Generalized Sign Z

Total Licensors Current Ratio undermedian Current Ratio abovemedian NI Previous Year undermedian NI Previous Year abovemedian Same Sic Diff Sic Leader Follower

260 123 122 122

3.34% 2.27% 3.12% 1.51%

2.10% 1.51% 2.81% 1.16%

148:112 > > > 76:47 > > 66:56 > 73:50 > >

5.222 * ** 3.384 * ** 3.320 * ** 2.873 * *

5.128 * ** 3.457 * ** 2.748 * * 2.221 *

3.104 * ** 3.05 * * 1.668 * 2.413 * *

122

4.35%

3.41%

71:51 > >

2.805 * *

3.888 * **

2.421 * *

52 208 93 166

7.74% 2.24% 0.12% 5.11%

4.72% 1.47% 0.38% 3.88%

32:20 > 116:92 > > 52:41) 95:71 > >

5.055 * ** 3.312 * ** 0.809 5.898 * **

5.188 * ** 3.371 * ** 0.172 5.6 * **

2.078 * 2.431 * * 1.443$ 2.724 * *

Notes: The symbols $,* ,* *, and * ** denote statistical significance at the 0.10, 0.05, 0.01, and 0.001 levels, respectively, using a generic one-tailed test. The symbols (, < or), > etc. correspond to $,* and show the direction and significance of the generalized sign test. 10

Technovation xxx (xxxx) xxx–xxx

G. Cabaleiro

Table 10 Mean cumulative abnormal returns and test statistics for licensing announcement divided into groups: market adjusted returns, equally weighted index (−1,0). GROUP

N

Mean Cumulative Abnormal Return

Precision Weighted CAAR

Positive: Negative

Patell Z

Portfolio Time-Series (CDA) t

Generalized Sign Z

Total Licensors Current Ratio undermedian Current Ratio abovemedian NI Previous Year undermedian NI Previous Year abovemedian Same Sic Diff Sic Leader Follower

260 123 122 123

2.68% 1.26% 2.53% 0.74%

2.12% 1.16% 2.43% 0.65%

147:113 > > > 69:54 > 70:52 > > 73:50 > >

7.061 * ** 3.652 * ** 4.344 * ** 2.707 * *

7.549 * ** 3.515 * ** 4.159 * ** 1.940 *

3.155 * ** 1.892 * 2.500 * * 2.664 * *

122

2.99%

2.47%

65:57)

2.917 * *

4.933 * **

1.587$

52 208 93 166

7.40% 1.50% 0.55% 3.89%

5.12% 1.40% 0.55% 3.39%

31:21 > 116:92 > > 52:41) 94:72 > >

7.397 * ** 4.197 * ** 1.711 * 7.552 * **

9.104 * ** 4.066 * ** 1.375$ 7.861 * **

1.858 * 2.599 * * 1.604$ 2.668 * *

Notes: The symbols $,* ,* *, and * ** denote statistical significance at the 0.10, 0.05, 0.01, and 0.001 levels, respectively, using a generic one-tailed test. The symbols (, < or), > etc. correspond to $,* and show the direction and significance of the generalized sign test. Table 11 Mean cumulative abnormal returns and test statistics for licensing announcement divided into groups: market model, value weighted index (−1,0). GROUP

N

Mean Cumulative Abnormal Return

Precision Weighted CAAR

Positive: Negative

Patell Z

Portfolio Time-Series (CDA) t

Generalized Sign Z

Total Licensors Current Ratio undermedian Current Ratio abovemedian NI Previous Year undermedian NI Previous Year abovemedian Same Sic Diff Sic Leader Follower

260 123 122 123

2.85% 1.29% 2.81% 0.88%

1.62% 0.92% 2.18% 0.77%

155:105 > > > 78:45 > > > 70:52 > > 81:42 > > >

7.785 * ** 4.062 * ** 4.839 * ** 3416 * **

7.976 * ** 3.672 * ** 4.522 * ** 2.422 * *

3.843 * ** 3.313 * ** 2.331 * * 3.856 * **

122

3.20%

2.60%

69:53 >

3.115 * **

5.348 * **

2.059 *

52 208 93 166

7.71% 1.63% 0.73% 4.05%

3.79% 1.11% 0.60% 2.75%

34:18 > > 121:87 > > 56:37 > 98:68 > > >

7.776 * ** 4.817 * ** 2.525 * * 7.842 * **

9.547 * ** 4.579 * ** 1.994 * 7.994 * **

2.619 * * 2.987 * * 2.167 * 3.109 * **

Notes: The symbols $,* ,* *, and * ** denote statistical significance at the 0.10, 0.05, 0.01, and 0.001 levels, respectively, using a generic one-tailed test. The symbols (, < or), > etc. correspond to $,* and show the direction and significance of the generalized sign test. Table 12 Interaction effects: mean cumulative abnormal returns and test statistics for each interaction group. Market Model, Equally Weighted Index (−1,0) GROUP N Mean Cumulative Abnormal Return Leader+Same Sic 7 0.73% Leader+ Diff Sic 86 0.59% Follower+Same Sic 44 8.46% Follower+Diff Sic 122 2.19% Market Adjusted Returns, Equally Weighted Index (−1,0) GROUP N Mean Cumulative Abnormal Return Leader+Same Sic 7 0.95% Leader+ Diff Sic 86 0.51% Follower+Same Sic 44 8.94% Follower+Diff Sic 122 2.20% Market Model, Value Weighted Index (−1, GROUP N Mean Cumulative Abnormal Return Leader+Same Sic 7 0.87% Leader+ Diff Sic 86 0.71% Follower+Same Sic 44 8.94% Follower+Diff Sic 122 2.28%

Precision Weighted CAAR

Positive: Negative

Patell Z

Portfolio Time-Series (CDA) t

Generalized Sign Z

0.38% 0.54% 4.99% 1.82%

5:2 48:38) 28:16 > 69:53 >

0.542 2.01 * 7.935 * ** 4.307 * **

0.867 1.462$ 9.323 * ** 4.133 * **

1.248 1.36$ 2.206 * 2.215 *

Precision Weighted CAAR

Positive: Negative

Patell Z

Portfolio Time-Series (CDA) t

Generalized Sign Z

0.50% 0.55% 5.08% 2.23%

4:3 48:38) 29:15 > > 68:54 >

0.618 1.603$ 8.135 * ** 4.134 * **

1.095 1.231 9.728 * ** 3.988 * **

0.473 1.533$ 2.502 * * 2.106 *

Precision Weighted CAAR

Positive: Negative

Patell Z

Portfolio Time-Series (CDA) t

Generalized Sign Z

0.48% 0.61% 5.08% 1.80%

4:3 52:34 > 29:15 > > 69:53 >

0.744 2.414 * * 8.135 * ** 4.263 * **

1.181 1.858 * 9.728 * ** 4.206 * **

0.477 2.117 * 2.502 * * 2.123 *

Notes: The symbols $,* ,* *, and * ** denote statistical significance at the 0.10, 0.05, 0.01, and 0.001 levels, respectively, using a generic one-tailed test. The symbols (, < or), > etc. correspond to $,* and show the direction and significance of the generalized sign test.

6. Conclusions and future research

market value, in significant and distinct ways. First, in line with previous research indicating that financial constraints erode companies’ bargaining power (Lerner and Merges, 1998; Aghion and Tirole, 1994), this study shows that licensors’ appropriation capacity is lower when companies face cash constraints than when they do not. Also, in accordance with the research demonstrating that information asymmetries reduce companies’ bargaining power (Chatterjee and Samuelson,

This article proposes that the licensor's appropriation capacity is a function of two aspects: the licensor's bargaining power (determined by its cash constraints and the existence of information asymmetries between parties) and the potential cost of imitation. The data confirm that these factors influence the licensor's appropriation capacity, in terms of 11

Technovation xxx (xxxx) xxx–xxx

G. Cabaleiro

1983), this paper finds that licensors can better appropriate benefits from licensing when they license out in the same sector (lower information asymmetries) than when they license out to another sector (higher information asymmetries). Furthermore, this study concludes, in light of the reported findings, that the impact of imitation grows stronger when the licensor's market share increases (Arora and Fosfuri, 2003). To support this conclusion, the analysis demonstrates that licensors that are followers appropriate more benefits from licensing than do licensors that are leaders. Additionally, the evidence proves that the best situation in which to appropriate benefits from licensing, in terms of market value, is when the licensees belong to the same sector as the licensors and when they are followers. Finally, appropriation is very low when the licensor and the licensee belong to different sectors and the licensors are leaders. Further research should cross-compare the licensors’ and licensees’

market value and analyze additional organizational and industrial characteristics that might affect the distribution of bargaining power in licensing agreements, as well the consequences of this distribution. One possible extension might be to determine the consequences of low bargaining power for decisions about control over the company after the agreement. For instance, do companies with less bargaining power accept key conditions, such as investments in equity, grant-back clauses or votes on the board of directors?

Acknowledgements I would like to express my sincere thanks to two anonymous reviewers, Andrea Fosfuri, Elena Novelli, Juan Santaló and Eduardo Melero for their valuable suggestions. All errors are my responsibility.

Appendix A Tests The Patell Z Test (or standardized abnormal return test) estimates a separate standard error for each security event, assuming cross-sectional independence. The standardization ensures that each AR offers the same variance. The Patell test statistic is asymptotically N(0,1) distributed under the null hypothesis (see Linn and McConnell, 1983; Schipper and Smith, 1986; Haw et al., 1990). The Patell (1976) test statistic is defined as:

n (L1 − 4) SCA R τ , L1 − 2

tpatell =

where L1 = T1 − T0 is the length of the estimation period, and

SCA R (τ1, τ2) =

CARi (τ1, τ2) SCARi (τ1, τ2)

is the average standarized CAR. This test has been proven to be powerful when the condition of cross- sectional independence of abnormal returns is not violated. The time-series standard deviation test, or Crude Dependence Adjusted Test (CDA) (Brown and Warner, 1980, 1985), compensates for potential dependence of returns by estimating the standard deviation using the time series of sample mean returns from the estimation period. If the estimated abnormal returns are normally independent and identically distributed, this test statistic is approximately standard normal under the null hypothesis (see Dopuch et al., 1986; Brickley et al., 1991). The CDA test for day zero is defined as:

tCDA = ut / s (ut ), where ut is the equal-weighted portfolio mean abnormal return on day t—i.e., N ut = (1/ N ) ∑i = 1 uit , and the standard deviation of ut is −40

−40

s (ut ) = (1/200) ∑t =−240 (ut − u) where u = (1/201) ∑t =−240 ut . The Generalized Sign Test (Cowan, 1982) uses a normal approximation of the binomial distribution and adjusts for the fraction of positive abnormal returns in the estimation period instead of assuming 0.5. The null hypothesis is that the fraction of positive returns is the same as in the estimation period (Sanger and Peterson, 1990; Singh et al., 1991; Chen et al., 1991). The Generalized Sign Test examines whether the number of stocks with positive cumulative abnormal returns in the event window exceeds the number expected in the absence of abnormal performance. The number expected is based on the fraction of positive abnormal returns in the 200-day estimation period,

pˆ =

1 n

n

∑ j=1

1 200

E200

∑

Sjt ,

t = E1

where

1 if ARjt > 0 Sjt = ⎧ ⎨ ⎩ 0 otherwise This test statistic uses the normal approximation to the binomial distribution with parameter pˆ . Define w as the number of stocks in the event window for which the cumulative abnormal return CARj (D1, Dd) is positive. The Generalized Sign Test statistic is:

ZG =

w − npˆ . [npˆ (1 − pˆ)]1/2

The T-test, defined in the following way, is used to study whether the mean CAR of the compared groups is significant:

12

Technovation xxx (xxxx) xxx–xxx

G. Cabaleiro

t=

X1 − X2 S x 1 x2

2 n

,

where X1 − X2 is the difference between the means of the two groups; S x1 x2 =

1 (S x21 2

+ S x22 ) is the pooled standard deviation of group one and group

two; and S x21 and S x22 are the estimators of the variances of the two samples. The null hypothesis states that the difference between the means is zero. Therefore, if this null hypothesis can be rejected, there must be a significant difference between the two samples.

and bankruptcy reorganization outcomes. J. Manag. 21 (6), 1041–1056. Dopuch, N., Holthausen, R.W., Leftwich, R., 1986. Abnormal stock returns associated with media disclosures of “subject to” qualified audit opinions. J. Account. Econ. 8 (2), 93–118. Fama, E.F., Fisher, L., Jensen, M.C., Roll, R., 1969. The adjustment of stock prices to new information. Int. Econ. Rev. 10 (1), 1–21. Flammer, C., 2013. Corporate social responsibility and shareholder reaction: the environmental awareness of investors. Acad. Manag. J. 56 (3), 758–781. Farrel, J., Gallini, N., 1988. Second sourcing as a means of commitment: monopoly incentives to attract Competition. Q. J. Econ. 108, 673–694. Fisher, R., Ury, W., 1981. Getting to Yes: Negotiating Agreement Without Giving, 1st ed. Penguin, New York. Fosfuri, A., 2006. The licensing Dilemma: understanding the determinants of the rate of technology licensing. Strateg. Manag. J. 27 (12), 1141–1158. Gallini, N.T., 1984. Deterrence through market sharing: a strategic incentive for licensing. Am. Econ. Rev. 74, 931–941. Gambardella, A., Giuri, P., Luzzi, A., 2007. The market for patent in Europe. Res. Policy 36 (8), 1163–1183. Gans, J., Hsu, D., Stern, S., 2003. The product market and the market for ideas: commercialization strategies for technology entrepreneurs. Res. Policy 32, 333–350. Giroud, X., Mueller, H.M., 2011. Corporate governance, product market competition and equity prices. J. Financ. 66 (2), 563–600. Granstrand, O., Patel, P., Pavitt, K., 1997. Multi-technology corporations: why they have "distributed" rather than "distinctive core" competencies. Calif. Manag. Rev. 39 (4), 8. Gulati, R., Lavie, D., Singh, H., 2009. The nature of partnership experience and the gains from alliances. Strateg. Manag. J. 30, 1213–1233. Hagedoorn, J., 2002. Inter-firm R&D partnerships: an overview of major Trends and patterns since 1960. Res. Policy 31 (4), 477–492. Hambrick, D.C., Cho, T.S., Chen, M.J., 1996. The influence of top management team heterogeneity on firms' competitive moves. Adm. Sci. Q. 41 (4), 659. Haw, I., Pastena, V., Lilien, S., 1990. Market manifestations of nonpublic information prior to merges: the effect of ownership structure. Account. Rev. (April), 432–451. Hitt, M.A., Hoskisson, R.E., Johnson, R.A., Moesel, D.D., 1996. The market for corporate control and firm innovation. Acad. Manag. J. 39 (5), 1084. Hoskisson, R.O., Johnson, R.A., 1992. Corporate restructuring and strategic change: the effect on diversification strategy and R&D intensity. Strateg. Manag. J. 13 (8), 625–634. Hu, Y., McNamara, P., McLoughlin, D., 2015. Outbound open innovation in bio-pharmaceutical out-licensing. Technovation 35, 46–58. Inkpen, Andrew C., Beamish, Paul W., 1997. Knowledge, bargaining power, and the instability of International joint ventures. Acad. Manag. Rev. 22, 177–202 (January 1). Katz, M., Shapiro, C., 1985. On the licensing of innovations. RAND J. Econ. 16, 504–520. Kale, P., Dyer, J., Singh, H., 2002. alliance capability, stock Market response and Longterm alliance success: the role of the alliance function. Strateg. Manag. J. 23 (8), 747–767. Kale, P., Dyer, J., Singh, H., 2003. Value creation and success in Strategic alliances. Eur. Manag. J. 19 (5), 463–471. Kani, M., Motohashi, K., 2012. Understanding the technology Market for Patents: new insights from a licensing survey of Japanese firms. Res. Policy 41 (1), 226–235. Kim, P.H., 1997. Strategic timing in group negotiations: the implications of forced entry and forced exit for negotiations with unequal power. Organ. Behav. Human. Decis. Process. 71, 263–286. Kollmer, H., Dowling, M., 2004. Licensing as a commercialization strategy for new technology-based firms. Res. Policy 33 (8), 1141–1151. Koh, J., Venkatraman, N., 1991. Joint venture formations and stock market reactions: an assessment in the information technology sector. Acad. Manag. 34 (4), 869–892. Latvie, D., 2007. Alliance portfolios and firm performance: a study of value creation and appropriation in the U.S. software industry. Strateg. Manag. J. 28 (12), 1187–1212. Lei, Slocum, 1991. Global strategic alliances: payoffs and pitfalls. Organ. Dyn., Winter 19 (3), 44. Lerner, J., Merges, R.P., 1998. The control of technology alliances: an empirical analysis of the biotechnology industry. J. Ind. Econ. 46 (2), 125–156. Lerner, J., Malmendier, U., 2010. Contractibility and the design of research agreements. Am. Econ. Rev. 100 (1), 214–246. Linn, S.C., McConnell, J.J., 1983. An empirical investigation of the impact of “Antitakeover” amendments on common stock prices. J. Financ. Econ., Elsevier 11 (1–4), 361–399. Lowe, J., Taylor, P., 1998. R&D and technology purchase Through license agreements: complementary strategies and complementary assets. RD Manag. 28 (4), 263–278. Mannix, E., 1993a. Organizations as resource dilemmas: the effects of power balance on coalition formation in small groups. Organ. Behav. Human. Decis. Process. 55 (1), 1–22. MacKinlay, A.C., 1997. Event studies in economics and finance. J. Econ. Lit. 35 (1), 13–39. Merchant, H., Schendel, D., 2000. How do international joint ventures create shareholder

References Aghion, P., Tirole, J., 1994. The management of innovation. Q. J. Econ. 109 (4), 1185–1209. Ahuja, G., 2000. Collaboration networks, structural holes, and innovation: a longitudinal study. Adm. Sci. Q. 45 (3), 425. Akhtar, F., Oliver, B., 2009. Capital Market Efficiency. McGraw-Hill Australia Pty Ltd. /PPTs t/a Business Finance 10e by Peirson (Chapter 16). Anand, B.N., Khanna, T., 2000a. The structure of licensing contracts. J. Ind. Econ. 48, 103–135. Anand, B.N., Khanna, T., 2000b. Do firms learn to create value? The case of alliances. Strateg. Manag. J. 21, 295–315. Arora, A., Gambardella, A., 1994. Evaluating technological information and utilizing it: Scientific knowledge, technological capability, and external linkages in biotechnology. J. Econ. Behav. Organ. 24 (1), 91–114. Arora, A., Fosfuri, A., Gambardella, A., 2001a. Markets for Technology: The economics of Innovation and Corporate Strategy. The MIT Press, Cambridge MA. Arora, A., Fosfuri, A., 2003. Licensing the market for technology. J. Econ. Behav. Organ. 52, 277–295. Arora, A., Ceccagnoli, M., 2006. Patent protection, complementary assets, and firms' incentives for technology licensing. Manag. Sci. 52 (2), 293–308. Arora, A., Gambardella, A., 2010. Ideas for rent: an overview of markets for technology. Ind. Corp. Change, Oxf. Univ. Press 19 (3), 775–803. Arora, A., Fosfuri, A., Roende, T., 2013. Managing licensing in a market for technology. Manag. Sci. 59 (5), 1092–1106. Athreye, S., Cantwell, J., 2007. Creating competition? Globalisation and the emergence of new technology producers. Res. Policy 36, 209–226. Balakrishna, S., Koza, M.P., 1993. Information asymmetry, adverse selection and joint ventures. J. Econ. Behav. Organ. 20, 99–117. Bergh, D.D., 1997. Product with repeated measures analysis: Demonstration with a study of the diversification and Divestiture Relationship. Acad. Manag. J. 38, 1692–1708. Bergh, D.D., Lawless, M.W., 1998. Portfolio restructuring and limits to hierarchical governance: the effects of environmental uncertainty and diversification strategy. Inst. Oper. Res. Manag. Sci. Bianchi, M., Chiaroni, D., Chiesa, V., Frattini, F., 2011. Organizing for external technology commercialization: evidence from a multiple case study in the pharmaceutical industry. RD Manag. 41 (2), 120–137. Bianchi, M., Frattini, F., Lejarraga, J., Di Minin, Alberto, 2014. Technology exploitation paths: combining technological and complementary resources in new product development and licensing. J. Product. Innov. Manag. 31 (S1), 146–169. Bianchi, M., Lejarraga, J., 2016. Learning to license technology: the role of experience and workforce's skills in Spanish manufacturing firms. RD Manag. 46 (S2), 671–705. Bolton, M.K., 1993. Organizational innovation and substandard performance: when is necessary the mother of innovation? Organ. Sci. 4 (1), 57–75. Bresnahan, T., Gambardella, A., 1998. The division of inventive labor and the extent of the market. In: Helpman, E. (Ed.), General Purpose Technologies and Economic Growth. MIT Press, Cambridge, MA. Brickley, J.A., Dark, F.H., Weisbach, M.S., 1991. An agency perspective on franchising. Finantial Manag. 20 (1), 27–35. Bromiley, P., 1991. Testing a causal model of corporate risk taking and performance. Acad. Manag. J. 34 (1), 37–59. Brown, S., 1980. Measuring security price performance. J. Financ. Econ. 8 (3), 205–258. Brown, S., Warner, J., 1985. Using daily stock returns: the case of event studies. J. Financ. Econ. 14, 3–31. Campbell, J.Y., Lo, A.W., MacKindley, A.C., 1997. The Econometrics of Financial Markets. Princeton University Press. Chang, S.J., Singh, H., 1999. The impact of modes of entry and resource fit on modes of exit by multibusiness firms. Strateg. Manag. J. 20 (11), 1019–1035. Chatterjee, K., Samuelson, W., 1983. Bargaining under Incomplete Information. Oper. Res. 31 (5), 835–851. Chen, H., Hu, M.Y., Shieh, J.C.P., 1991. The Wealth Effect of International Joint Ventures: the Case of U.S. Investment in China. Financ. Manag. 20, 31–41. Cohen, W.M., Nelson, R.R., Walsh, J.P., 2000. Protecting Their Intellectual Assets: Appropriability Conditions and Why U.S. Manufacturing Firms Patent (or Not). Working paper series (7552). Conti, R., Gambardella, A., Novelli, E., 2013. Research on markets for inventions and implications for R&D allocation strategies. Acad. Manag. Ann. 7 (1), 717–774. Cowan, A.R., 1992. Nonparametric event study tests. Rev. Quant. Financ. Account. 2 (4), 343–358. Crutchley, C.E., Guo, E., Hansen, R.S., 1991. Stockholder benefits from Japanese-U.S. joint ventures. Financ. Manag. 20 (4), 22–30. Dailey, C., Dalton, D., 1994. Bankruptcy and corporate governance: the impact of board composition and structure. Acad. Manag. J. 37 (6), 1603. Dailey, C.M., 1995. The relationship between board composition and leadership structure

13

Technovation xxx (xxxx) xxx–xxx

G. Cabaleiro

Acad. Manag. Rev. 25 (1), 217–226. Shapiro, C., 2001. Navigating the patent Thicket: cross licenses, patent pools, and standard setting. Innov. Policy Econ. 1, 119–150. Sheehan, J. Martínez, C., Guellec, D., 2004. Understanding Business Patenting and Licensing: Results of a Survey. Chapter 4. Patents, Innovation and Economic Performance. In: Proceedings of and OECD Conference, OCDE, Paris. Shepard, A., 1987. Licensing to enhance demand for new technologies. RAND J. Econ. 18 (3). Singh, A.K., Cowan, A.R., Nayar, N., 1991. Underwritten calls of convertible bonds. J. Financ. Econ. 29, 173–196. Somaya, D., Kim, Y., Vonortas, N.S., 2010. Exclusivity in licensing alliances: using Hostages to support technology Commercialization. Strateg. Manag. J. 32, 159–186. Stuart, T.E., Hoang, H., Hybels, R.C., 1999. Interorganizational Endorsements and the performance of entrepreneurial. Adm. Sci. Q. 44 (2), 35. Teece, D.J., 1986. Profit. Technol. Innov.: Implic. Integr., Collab., Licens. Public Policy 15 (6), 285–305. Telesio, P., 1979. Technology Licensing and Multinational Enterprises. Praeger, New York. Walter, J., 2012. The influence of firm and industry characteristics on returns from technology licensing deals: evidence from the US computer and pharmaceutical sectors. RD Manag. 42 (5), 435–454. Wang, Y., Zhou, Z., Ning, L., Chen, J., 2015. Technology and external conditions at play: a study of learning by licensing practices in China. Technovation 43–44, 29–39. Zuniga M.P., Guellec, D., 2009. Who Licenses Out Patents and Why? Lessons From a Business Survey. STI Working Paper 2009/5. Statistical Analysis of Science, Technology and Industry.

value? Strateg. Manag. J. 21 (7), 723–737. Müllez- Seitz, G., 2012. Absorptive and desorptive capacity-related practices at the network level: the case of SEMATECH. RD Manag. 42, 90–99. Natalicchio, A., Petruzzelli, A.M., Garavelli, A.C., 2014. A literature review on markets for ideas: emerging characteristics and unasnwered questions. Technovation 34, 65–76. Park, S.H., Kim, D., 1997. Market valuation of joint ventures: joint venture characteristics and wealth gains. J. Bus. Ventur. 12, 83–108. Patell, J.M., 1976. Corporate forecast of earnings per share and stock price behavior: empirical test. J. Account. Res. 14 (2), 246–276. Pinkley, R.L., Neale, M.A., Bennett, R.J., 1994. The impact of alternatives to settlement in dyadic negotiation. Organ. Behav. Human. Decis. Process. 57 (1), 97–116. Reuer, J., Koza, M.P., 2000. Asymmetric information and joint venture performance: evidence from domestic and international joint ventures. Strateg. Manag. J. Strateg. Manag. J. 21 (1), 81–88. Robbins, 2006. Measuring Payments for the Supply and Use of Intellectual Property.NBER Chapter in International Trade in Services and Intangibles in the Era of Globalization. Rockett, K., 1990b. Choosing the competition and patent licensing. RAND J. Econ. 21, 161–171. Sanger, G.C., Peterson, J.D., 1990. An empirical analysis of common stock delistings. J. Financ. Quant. Anal. 25 (2), 261–272. Schipper, K., Smith, A., 1986. A comparison of equity carve-outs and seasoned equity offerings share price effects and corporate restructuring. J. Financ. Econ. 15, 153–186. Schilling, M.A., 2009. Understanding the alliance data. Strateg. Manag. J. 30 (3), 233–260. Shane, S., Venkataraman, S., 2000. The promise of entrepreneurship as a field of research.

14