mergers and innovation in big pharma carmine ornaghi university of southampton
DESCRIPTION
Mergers and Innovation in Big Pharma Carmine Ornaghi University of Southampton Toulouse, January 2008. Outline. 1 - M&As and Innovation: Limitations of the Literature 2 - Objectives of this Work 3 - Theoretical Insights 4 - Empirical Models 5 - Data and Variables 6 - Main Findings - PowerPoint PPT PresentationTRANSCRIPT
Mergers and Innovation in Big Pharma
Carmine OrnaghiUniversity of Southampton
Toulouse, January 2008
Outline
1 - M&As and Innovation: Limitations of the Literature
2 - Objectives of this Work
3 - Theoretical Insights
4 - Empirical Models
5 - Data and Variables
6 - Main Findings
7 - Mergers and Innovation: A Competition Policy perspective
1. Mergers and Innovation: Limitations of the Literature
• Empirical studies on M&As have found contradictory results about
their effects on firms’ performance: economists are still divided on
whether mergers enhance economic efficiency or increase market
power or neither of the two (e.g. managers’ interests).
• Main features of most of these studies:
Based on data of different industries.
Focused on assessing the short-run effects on sales and profits
(Guegler et all., 2003) and market value abnormal returns around
the announcement day (Fuller et all., 2002).
• Limitations of this literature:
Recent empirical findings show the existence of industry clustering in merger activity (Andrade et al., 2001): mergers as a response to exogenous changes in industry structure → Cross-industry studies can give inconclusive results.
The post-merger performance of the merged entities is likely to depend on the “relatedness” of the merging parties → Hardly considered in the literature.
In R&D intensive industry, relevant dimension of competition is innovation rather than price → Short-run analysis on sales and profits is not suitable.
1. Mergers and Innovation: Limitations of the Literature
• This work tries to overcome these limitations by studying the effects of M&As on innovation in a single industry.
• Analysis conducted for the case of large mergers in the Pharmaceutical Industry
• Research questions:
(1) Do mergers have a positive effect on the innovative ability of the firms involved, as their proponents often claim?
(2) Is there any relationship between the ex-ante technological and product relatedness of merging parties and the ex-post effects?
2 - Objectives of the Work
• M&As affect optimal R&D through different channels:
1. Avoidance of duplication of fixed costs (eg. library, labs, …) → decrease in R&D inputs
2. Economies of scope and knowledge synergies → increase in R&D inputs and outputs
3. Internalization of spillovers, reduction in the number of competitors and higher barriers to entry → increase of R&D inputs and outputs
4. … But knowledge is embodied in scientists and mergers usually imply a reduction of the employees. Moreover, cultural dissonances might disrupt innovation outcomes → decrease in R&D output
• It is not possible to define clear predictions on the net effects of these forces: Empirical evidence is needed
3 – Theoretical Insights: Effects of M&As on Research
• Most of the effects above are driven by forces whose magnitude depends on the ex-ante technology relatedness (TR) of the merged parties (e.g. synergies due to cross fertilization of ideas or elimination of useless duplication).
• Product relatedness (PR) might also have an indirect effect on innovation through changes in the market equilibria for approved drugs
• An empirical questions arise:
Can TR and PR explain differences in post-merger results of consolidated companies and competitors?
3 – Theoretical Insights : Technology and Product Relatedness
4 – Empirical Model
where the dependent variable measures the percentage change in R&D inputs/outputs, M0, M1, M2 and M3 are dummy variables that take on value of 1 if the firm goes through a merger in period t, in
period t-1 (i.e. one-year ago), in t-2 or in t-3, respectively. T is a complete set of time dummies for the period 1988-2004.
• M0 represent a difference-in-difference estimate of the changes in Y due to the merger, and the other dummies assess whether there are lagged effects of consolidation in the following years.
• To access the effects of mergers (up to 3 years after the deal), I use a dummy variable model:
uTMMMMY 3210% 3210
4 – Empirical Model: Problem of Endogeneity
• Endogeneity of the merger decision can lead to a (spurious) correlation between the merger dummies and the outcome for reasons unrelated to the causal effect we are interested.
Example: It has been found that firms with important drugs coming off patents are more likely to pursue a merger. As patent expirations affect future revenues, we would find a negative correlation between mergers and growth of revenues even in the absence of a causal effect of the first on the second.
• I account for the selection problems in two ways:
Propensity score: each acquirer and target is matched with firms with the closest probability of merging
Technological relatedness: exogenous technological shocks are likely to hit firms with similar research activities
4 – Empirical Model: Relatedness
• To assess the role of TR and PR in post-merger effects, I estimate the model:
uXPRTRY )(% 21
where λ(Xβ) is the inverse Mills ratio which controls for selection problems (Heckman “two-step” procedure).
• New dataset containing publicly traded pharmaceutical firms constructed using three main data sources:
- Financial Data (sales, stock market values, R&D expenditures) from Compustat and Osiris
- Patents Data from the US Patent Office (patent class and citation)
- Merger transactions data for 1988-2004 from Mergers Year Book.
• All observations double checked and completed with sources in the internet (mainly, web pages of firms and www.sec.gov)
• Our sample represents the universe of big pharma companies (excluding large generic producers such as Teva or Mylan) and the consolidations that they have been involved
5 – Data and Variables
• Technological and Product Relatedness:
Correlation of Patent Classes (PatCr) – Jaffe (1986)
A similar measure has been constructed for Product Classes
Overlapping between Cited Patents
5 – Data and Variables
areasy technolog in the patents of vector theis )S,...,(S S where
)'()'(
'
k1
2/12/1
k
SSSS
SSPatCr
TTAA
TA
T
TA
BPatents inNumber of
B BPatents inNumber of Over
BA (BT) is the set of Patents cited by the
patent portfolio of acquirer (target)
6 – Main Empirical Findings
• EFFECTS OF MERGERS (DUMMY VARIABLE MODEL):
Negative signs for R&D inputs, output and productivity.
Market value growth below the other non-merging firms.
Results similar when accounting for endogeneity and selectivity issues (only the negative sign for Market Value growth is no longer statistically significant)
6 – Main Empirical Findings
• THE ROLE OF TECHNOLOGICAL RELATEDNESS:
Results suggest that product relatedness has a positive effect on post-merger outcomes while technological relatedness seems to have detrimental impact
Most interesting finding concerns the change in stock market value: positive and statistically significant coefficient for PR and negative and statistically significant coefficient for TR.
7 - Competition Policy Implications
• “Efficiencies are easy to promise, yet may be difficult to deliver''. Lawrence White
• Our results cast some doubts on the actual materialisation of the efficiency gains in R&D commonly claimed by merging firms to defend consolidations.
• Mergers between firms with large technological relatedness are found to deliver worse outcomes.
• The importance of the questions here analysed and the difficulty involved in the empirical analysis impose extreme cautions in drawing any radical conclusions for competition policy.
• Relate ex-post effects to ex-ante characteristics is an important task for future research agenda.