Knowledge Spillovers /2

Question 1

Are knowledge spillovers really unintentional and “in the air”?

Answer 1

Studies have investigated that they are not and that they are driven by specific mechanisms.

Question 2

Talk about the Almeida, Kogut 1999 study

Answer 2

‘Localization of knowledge and the Mobility of Engineers in Regional Networks’.

Idea and Hypothesis: ideas spread, at least in part, through the local mobility of knowledge employees.

Structure of the study:
1. Analysis of the variation in the localization of knowledge across regions (Jaffe et al. 1993 replication) in a specific sector (semiconductor industry).

2. Analysis of inventors’ mobility to show that knowledge is transferred through regional labour market.

Method: 
Step 1. replication of JTH experiments, thus construction of citing sample and control sample based on same technological class and application date. 

Similar and expected results, evidence that Citation matching is higher than the control matching.

Step 2. Look at the Affiliation (firm or university) of the 483 inventors in the major patent applications, and traced whether that inventor changed affiliation through their patenting career.
Check if the new affiliation was within the same old region or outside the region.

Look at the difference between the intra-regional moves and the inter-regional moves.
Negative = people move more outside the region.
Positive = people move more inside the region.

They perform correlational analysis.
Dependent variable: co-location of originating and citing patents.
Key covariate: interregional mobility + intraregional mobility + control variables.

Correlation between regional differences in the mobility of inventors and localisation of knowledge flows!

Question 3

What are further explanation other than mobility, to the localisation of knowledge spillovers?

Answer 3

Social ties and networks.

Breschi, Lissoni, “mobility of skilled workers and co-invention networks: an autonomy of localised knowledge flows”.

Research question: what remains of localisation effects (unintended knowledge spillovers) once we take into account the mobility and “formal” social ties?

Method:
Take 2 patents: a cited patent P and a citing patent M assigned to different organisations.
The link between P’s and M’s inventors can be:

• At least one inventor is both in P and M –> mobility of the inventor.
• There is a link between at least an inventor in P and in M.
• There is no connection

Data from EPO.

Method:
JTH methodology with subsamples of observation.
Constructed the three samples:
- Originating patents
- Citing patents
- Control patents, similar to the citing patents, but with no citation link

For each cited-citing patent they construct:
• Co-location measure = 1 if at least one inventor from the citing sample is located in the same metropolitan area as the cited patent
• The type of linkage suggested earlier, “mobility”, “connected” or “not connected”.

Findings:
Mobility for citing-cited is higher than control sample, so when there is a citation link, there is also a higher probability that the inventors are mobile.

Co-invention network is also higher for citing-cited than for control.

Hence, both mobility and social ties can explain the higher probability of receiving local citations.

Conclusion: knowledge spillover exist and are explained by formal mechanisms. So geography matters for knowledge exchange, since mobility and networks are bounded in space.