Knowledge characteristics and the dynamics of technological alliances in pharmaceuticals: empirical evidence from Europe, US and Japan

Abstract

The paper investigates the co-evolutionary patterns of the dynamics of technological alliances and of the structure of the knowledge base in the pharmaceutical sector. The main hypothesis under scrutiny is that technological alliances represent a key resource for firms in knowledge intensive sectors to cope with dramatic changes in the knowledge base, marked by the introduction of discontinuities opening up new technological trajectories. Using patent information and data on technological alliances drawn from the CATI-MERIT database, we compare the evidence concerning the so-called triad regions, i.e. United States, Europe and Japan. The empirical results support the existence of a life cycle in biotechnology affecting the pharmaceutical industry. Furthermore, the dynamics of alliances is found to depend on (i) the phase of the biotechnology life cycle, (ii) the strength of the region in biotechnology and (iii) the general features of the economic environment of the region.

Appendices

Appendix A

1.1 A.1 Knowledge variety

We decided to measure technological variety by using the information entropy index. Entropy measures the degree of disorder or randomness of the system, so that systems characterized by high entropy will also be characterized by a high degree of uncertainty (Saviotti 1988). Differently from common measures of variety and concentration, the information entropy has some interesting properties (Frenken and Nuvolari 2004). An important feature of the entropy measure is its multidimensional extension. Consider a pair of events (X_l, Y_j), and the probability of co-occurrence of both of them p _lj . A two dimensional total variety (TV) measure can be expressed as follows:

$$ KV\equiv H(X,Y)=\sum\limits_{l} {\sum\limits_{j} {p_{lj} \log_{2} \left({\frac{1}{p_{lj}} } \right)}} $$

(A1)

If one considers p _lj to be the probability that two technological classes l and j co-occur within the same patent, then the measure of multidimensional entropy focuses on the variety of co-occurrences of technological classes within regional patents applications.

Moreover, the total index can be decomposed in a “within” and a “between” part anytime the events to be investigated can be aggregated into a smaller numbers of subsets. Within-entropy measures the average degree of disorder or variety within the subsets, while between-entropy focuses on the subsets measuring the variety across them. Frenken et al. (2007) refer to between- and within- group entropy respectively as unrelated and related variety.

It can be easily shown that the decomposition theorem holds also for the multidimensional case. Hence if one allows l ∈ S _g and j ∈ S _z (g = 1,…,G; z = 1,…, Z), we can rewrite H(X, Y) as follows:

$$ KV=H_{Q} +\sum\limits_{g=1}^{G} {\sum\limits_{z=1}^{Z} {P_{gz} H_{gz}} } $$

(A2)

Where the first term of the right-hand-side is the between-entropy and the second term is the (weighted) within-entropy. In particular:

$$ UKV\equiv H_{Q} =\sum\limits_{g=1}^{G} {\sum\limits_{z=1}^{Z} {P_{gz} \log_{2} \frac{1}{P_{gz}} }} $$

(A3)

$$\begin{array}{@{}rcl@{}} RKV\equiv \sum\limits_{g=1}^{G} {\sum\limits_{z=1}^{Z} {P_{gz} H_{gz}} } \notag\\ P_{gz} =\sum\limits_{l\in S_{g}} {\sum\limits_{j\in S_{Z}} {p_{lj}} } \notag\\ H_{gz} =\sum\limits_{l\in S_{g}} {\sum\limits_{j\in S_{z}} {\frac{p_{lj}} {P_{gz}} \log_{2} \left({\frac{1}{p_{lj} /P_{gz}}} \right)}} \end{array} $$

(A4)

We can therefore refer to between- and within-entropy respectively as unrelated technological variety (UTV) and related technological variety (RTV), while total information entropy is referred to as general technological variety.

A.2 Knowledge coherence

Knowledge coherence measures the degree of complementarity among technologies. We expect it to provide us with an indication of the difficulty, or cost, a firm has to face to learn a new type of knowledge. Typically a firm needs to combine, or integrate, many different pieces of knowledge to produce a marketable output. Thus, in order to be competitive a firm not only needs to learn new ’external’ knowledge but it needs to learn to combine it with other, new and old, pieces of knowledge. We can say that a knowledge base in which different pieces of knowledge are well combined, or integrated, is a coherent knowledge base. The technologies contained in the knowledge base are by definition complementary in that they are jointly required to obtain a given outcome. For this reason, we turned to calculate the coherence of the knowledge base, defined as the average relatedness of any technology randomly chosen within the sector with respect to any other technology (Nesta and Saviotti 2005, 2006; Nesta 2008).

To yield the knowledge coherence index, a number of steps are required. In what follows we will describe how to obtain the index at the sector level. First of all, one should calculate the weighted average relatedness WAR _l of technology l with respect to all other technologies present within the sector. Such a measure builds upon the measure of technological relatedness τ _lj (see Nesta and Saviotti 2005, for details). Following Teece et al. (1994), WAR _l is defined as the degree to which technology l is related to all other technologies j ∈ l in the sector, weighted by patent count P _jt :

$$ WAR_{lt} =\frac{\sum\nolimits_{j\ne l} {\tau_{lj} P_{jt}} } {\sum\nolimits_{j\ne l} {P_{jt}} } $$

(A5)

Finally the coherence of knowledge base within the sector is defined as weighted average of the WAR _lt measure:

$$ COH_{t} =\sum\limits_{l\ne j} {WAR_{lt} \times \frac{P_{lt}} {\sum\nolimits_{l} {P_{lt}} } } $$

(A6)

It is worth stressing that such index implemented by analysing co-occurrences of technological classes within patent applications, measures the degree to which the services rendered by the co-occurring technologies are complementary to one another. The relatedness measure τ _lj indicates indeed that the utilization of technology l implies that of technology j in order to perform specific functions that are not reducible to their independent use. This makes the coherence index appropriate for the purposes of this study.

1.1 A.3 Cognitive distance

We need a measure of cognitive distance (Nooteboom 2000) able to express the dissimilarities amongst different types of knowledge. A useful index of distance can be derived from the measure of technological proximity. Originally proposed by Jaffe (1986, 1989), who investigated the proximity of firms’ technological portfolios. Subsequently Breschi et al. (2003) adapted the index in order to measure the proximity, or relatedness, between two technologies. The idea is that each firm is characterized by a vector V of the k technologies that occur in its patents. Knowledge similarity can first be calculated for a pair of technologies l and j as the angular separation or un-cented correlation of the vectors V _lk and V _jk . The similarity of technologies l and j can then be defined as follows:

$$ S_{lj} =\frac{\sum\nolimits_{k=1}^{n} {V_{lk} V_{jk}} } {\sqrt{\sum\nolimits_{k=1}^{n} {V_{lk}^{2}}} \sqrt{\sum\nolimits_{k=1}^{n} {V_{jk}^{2}} } } $$

(A7)

The idea underlying the calculation of this index is that two technologies j and l are similar to the extent that they co-occur with a third technology k. The cognitive distance between j and l is the complement of their index of the similarity:

$$ d_{lj} =1-S_{lj} $$

(A8)

Once the index is calculated for all possible pairs, it needs to be aggregated at the industry level to obtain a synthetic index of technological distance. This can be done in two steps. First of all one can compute the weighted average distance of technology l, i.e. the average distance of l from all other technologies.

$$ WAD_{lt} =\frac{\sum\nolimits_{j\ne l} {d_{lj} P_{jit}} } {\sum\nolimits_{j\ne l} {P_{jit}} } $$

(A9)

Where P _j is the number of patents in which the technology j is observed. Now the average cognitive distance at time t is obtained as follows:

$$ CD_{t} =\sum\nolimits_{l} {WAD_{lit} \times \frac{P_{lit}} {\sum\nolimits_{l} {P_{lit}} }}, $$

(A10)