Deals Not Done: Sources of Failure in the Market for Ideas

Transcription

1 Deals Not Done: Sources of Failure in the Market for Ideas Ajay Agrawal, Iain Cockburn, Laurina Zhang September 9, 2013 IPO/ESRC Conference on Patent Use London 1

2 Motivation The market for ideas is economically significant Facilitates vertical specialization Allocates resources and directs R&D effort Potentially high social and private gains Volume of transactions in technology is large and rapidly growing Global market for technology is worth $100 billion (Arora and Gambardella, 2010) but susceptible to market failure Information asymmetry problems Disclosure problems (Arrow, 1962) Moral hazard problems (Arora, 1996) Hold-up problems (Pisano, 1991) Opportunism Lack of liquidity Limited empirical research Typically we look for information on prices and quantities and compare them to benchmarks Data on the market for ideas is poor (with few exceptions e.g., universities) Not clear what the benchmark is to evaluate the efficiency of technology transactions Objective: Shed light on sources of failure in the market for ideas using novel survey data; identify where failure occurs 2

3 High variability of deals not done in the market for ideas LES survey, 2006 N = 461 3

4 Research Questions How does the relationship between deal failure and market features vary across different stages of the licensing process? What inferences can we make about the nature of market frictions? 4

5 Market Design (Roth, 2008) Market features associated with efficient market operation: 1) Thickness A large enough proportion of potential buyers and sellers to produce a satisfactory outcome for both sides of the transaction 2) Non-congestion 3) Safety Participants have enough time to make satisfactory choices when faced with a variety of alternatives Participants do not have incentives for misrepresentation or strategic action that undermine the ability of others to evaluate potential trades 5

6 Gans and Stern (2010) Applies market design concepts to market for ideas context Specific characteristics of ideas/technologies hinder efficient functioning of the market Ideas complementarity => Lack of market thickness User reproducibility => Lack of market safety Value rivalry => Congestion Repugnance Social constraints that prevent exchange from taking place - Organ exchange - Trolls? 6

7 Discrete Stages of the Licensing Process How do market features differentially impact the rate of deal success in each stage? Stage 1: Identifying potential partners Stage 2: Starting negotiations Stage 3: Reaching an agreement Lack of Market Thickness Lack of Market Safety Congestion/ Bargaining Frictions Data: LES (2006) 7

8 Data 2006 annual survey conducted by the authors in collaboration with Licensing Executives Society (LES). Directed towards organizations that create IP and technology directly Central theme in 2006 survey is impediments to licensing Participation: 600 out of approximately 3000 organizations in LES Not unrepresentative of LES membership Five industries energy, healthcare, software & electronics, transportation, university Response rate varies across questions (~70%) Not every respondent answers every question No obvious difference between responders and non-responders (within the survey) in terms of industry affiliation and firm size Caveat Data limitations preclude a causal analysis 8

9 Descriptive Statistics: Firm Size Annual revenue (millions) Annual R&D budget (millions) # Obs Mean Std. Dev. Min Max # Employees # Licensing professionals

10 Firm Size Annual revenues LES survey, 2006 N =

11 Firm Size - # Employees LES survey, 2006 N =

12 Descriptive Statistics: Industry # Obs Mean Std. Dev. Min Max Energy Healthcare Software & Electronics Transportation University

13 Method (Deal Success) i = α + β 1 (Lack of Market Thickness) i + β 2 (Lack of Market Safety) i + β 3 (Bargaining Frictions) i + X i +ε i Unit of analysis: organization Probit models and robust standard errors Controls for: Demand for IP, organization size (annual revenue, annual R&D budget, # employees, # licensing professionals), industry (energy, healthcare, software & electronics, transportation, university) Results robust to: Linear probability models, logit, ordered probit with known thresholds, joint estimation with oprobit with random respondent effect, logged dependent variables 13

14 Dependent Variables Deal Success Stage 1: Identifying potential partners The level of unlicensed IP Thinking with at least about one intellectual potential property that licensor/licensee could have been licensed in the last 12 months but wasn t, for what percentage was your organization able to identify at least one potential licensee/licensor? Lack of Market Thickness Lack of Market Safety Congestion/ Bargaining Frictions 14

15 Stage 1: Identifying potential partners 40% of unlicensed IP has an identifiable buyer/seller 15

16 Dependent Variables Deal Success Stage 1: Identifying potential partners The level of unlicensed IP with at least one potential licensor/licensee Stage 2: Starting negotiations The level of negotiations Where started potential once potential licensees/licensors were licensors/licensees identified, in what are percentage of found cases were substantive negotiations ever started? Lack of Market Thickness Lack of Market Safety Congestion/ Bargaining Frictions 16

17 Stage 2: Starting negotiations 50% of negotiations were started, where potential buyers/sellers were found 17

18 Dependent Variables Deal Success Stage 1: Identifying potential partners The level of unlicensed IP with at least one potential licensor/licensee Stage 2: Starting negotiations The level of negotiations started once potential licensors/licensees are found Stage 3: Reaching an agreement The level of agreements Of all the times you entered completed into substantive once licensing negotiations in began the last 12 months, what percentage resulted in a successful agreement? Lack of Market Thickness Lack of Market Safety Congestion/ Bargaining Frictions 18

19 Stage 3: Reaching an agreement 56% of negotiations reach an agreement, where negotiations started % Stage 1: Level of unlicensed IP with a potential licensor/licensee Stage 2: Level of negotiations Stage 3: Level of agreements started reached Probability of a successful transaction = 0.4 x 0.5 x 0.56? 19

20 Explanatory Variables Lack of Market Thickness Survey question: Compare a $10M IP licensing transaction with one involving a tangible asset of similar dollar value (for example, leasing a piece of real estate or contracting for use of a specialized production facility). In your experience: There are usually fewer potential buyers/sellers for the IP [relative to tangible assets] Equals 1 if respondents agree to the statement and 0 otherwise. 20

21 Main Results Lack of Market Thickness Probit model DV: Rate of Deal Success (1) (2) (3)* Stage 1 Stage 2 Stage 3 Lack of market thickness ** (0.095) (0.104) (0.089) Lack of market safety ** (0.001) (0.001) (0.001) Bargaining frictions/congestion: Due diligence is difficult/costly (0.100) (0.111) (0.095) Inability to arrive at mutually acceptable financial terms * (0.001) (0.001) (0.001) Too many parties at the table ** (0.003) (0.004) (0.004) Legal/regulatory problems ** (0.005) (0.004) (0.004) Other bargaining frictions Yes Yes Yes Controls Yes Yes Yes Observations R-squared *restricted sample + p<0.10, *p<0.05, **p<

22 Main Results Lack of Market Thickness Interpretation Lack of market thickness is associated with lower deal success in the first stage Not surprising Lack of market thickness is not correlated with deal success in the latter stages Licensing negotiations often occur under conditions of bilateral monopoly. Gans and Stern (2010): Detailed negotiations over the precise terms and conditions of a license take place in a bilateral rather than multilateral environment Each potential buyers value may depend on whether other buyers have had access to the technology or not (since rival access would allow competitors to expropriate some portion of the value by imitating technology) Although negotiations are influenced by the shadow of competition, it is less relevant in the market for ideas context since the value of the idea declines if the seller negotiates with multiple buyers Bilateral negotiations do not seem to be influenced by the shadow of competition. 22

23 Explanatory Variables: Lack of Market Safety Survey question: Of the IP that your organization would like to license, but cannot, approximately what fraction would you say is not effectively protectable by patents, trade secrets, etc.? Use mid-point of response category ranges (0%, 1-5%, 5-25%, %, 100%) 23

24 Main Results - Lack of Market Safety Probit model (1) (2) (3)* DV: Rate of deal success Stage 1 Stage 2 Stage 3 Lack of market thickness ** (0.095) (0.104) (0.089) Lack of market safety ** (0.001) (0.001) (0.001) Bargaining frictions/congestion: Due diligence is difficult/costly (0.100) (0.111) (0.095) Inability to arrive at mutually acceptable financial terms * (0.001) (0.001) (0.001) Too many parties at the table ** (0.003) (0.004) (0.004) Legal/regulatory problems ** (0.005) (0.004) (0.004) Other bargaining frictions Yes Yes Yes Controls Yes Yes Yes Observations R-squared *restricted sample + p<0.10, *p<0.05, **p<

25 Main Results Lack of Market Safety Interpretation Firms only engage in legal due diligence after they have determined the general feasibility of reaching an agreement Negotiating parties are less likely to reach an agreement when sellers are hesitant to provide full disclosure due to expropriation risk Firms do not seem to anticipate safety issues not correlated with deal success in first two stages 25

26 Explanatory Variables: Bargaining Frictions Information costs Equals 1 if respondents agree to the statement: Due diligence will be much more difficult/costly for the IP deal [relative to tangible assets], and 0 otherwise. For the IP deal, negotiations with a specific buyer/seller will be more difficult to bring to closure. Contracting costs Over the past 12 months, when substantive licensing negotiations have failed to reach an executed agreement, in what percentage of cases was the breakdown due to inability to arrive at mutually acceptable financial terms inability to arrive at mutually acceptable non-financial terms inability to agree on the appropriate scope of IP to be included in the agreement (patents, know-how, or other key IP assets) too many parties at the table (multiple licensors/licensees) due diligence revealed problems with enforceability/validity of IP better alternatives emerged for one or more parties delay (ie. clock ran out) legal and regulatory problems poor negotiating skills lack of trust ego Use mid-point of response category ranges (0%, 1-5%, 5-25%, %, 100%) 26

27 Explanatory Variables: Bargaining Frictions 27

28 Main Results Bargaining Frictions Probit model (1) (2) (3)* DV: Rate of deal success Stage 1 Stage 2 Stage 3 Lack of market thickness ** (0.095) (0.104) (0.089) Lack of market safety ** (0.001) (0.001) (0.001) Bargaining frictions/congestion: Due diligence is difficult/costly (0.100) (0.111) (0.095) Inability to arrive at mutually acceptable financial terms * (0.001) (0.001) (0.001) Too many parties at the table ** (0.003) (0.004) (0.004) Legal/regulatory problems ** (0.005) (0.004) (0.004) Other bargaining frictions Yes Yes Yes Controls Yes Yes Yes Observations R-squared *restricted sample + p<0.10, *p<0.05, **p<

29 Main Results Bargaining Frictions Interpretation Firms do not seem to anticipate these problems frictions are not significant in stage 1 No evidence of other frictions correlating with deal success: i.e., poor negotiating skills, lack of trust, and ego 29

30 Selection Low Quality Deals Probit model (1) (2) (3) Stage 1 Stage 2 Stage 3 DV: Rate of deal success Lack of market thickness ** (0.097) (0.108) (0.089) Lack of market safety ** (0.001) (0.001) (0.001) Bargaining frictions/congestion: Due diligence is difficult/costly (0.100) (0.115) (0.095) Inability to arrive at mutually acceptable financial terms (0.001) (0.001) (0.001) Too many parties at the table ** (0.004) (0.004) (0.004) Legal/regulatory problems ** (0.004) (0.004) (0.004) Controls Yes Yes Yes Observations R-squared p<0.10, *p<0.05, **p<0.01. Lower deal success in the first stage not due to low quality IP IP that the firm is willing and able to license Limit the sample to firms that have at least 5-25% of negotiations that have reached an agreement 30

31 University vs. Firms Stage 1: Identifying potential partners Stage 2: Starting negotiations Stage 3: Reaching an agreement University (28% of sample) Interpretation University deals are more likely to fail in the first stage embryonic technologies (Thursby and Thursby, 2001) Conditional on finding a partner, university deals are more likely to reach agreement Different incentives i.e., royalties and research lab support (Lach and Schankerman, 2008) Do not compete in downstream product market 31

32 Conclusion Stage 1: Identifying potential partners Stage 2: Starting negotiations Stage 3: Reaching an agreement Lack of Market Thickness Lack of Market Safety Congestion/ Bargaining Frictions 32