Working Paper: NBER ID: w13338
Authors: Rachel Griffith; Sokbae Lee; John Van Reenen
Abstract: We examine the home bias of international knowledge spillovers as measured by the speed of patent citations (i.e. knowledge spreads slowly over international boundaries). We present the first compelling econometric evidence that the geographical localization of knowledge spillovers has fallen over time, as we would expect from the dramatic fall in communication and travel costs. Our proposed estimator controls for correlated fixed effects and censoring in duration models and we apply it to data on over two million citations between 1975 and 1999. Home bias declines substantially when we control for fixed effects: there is practically no home bias for the more modern sectors such as pharmaceuticals and information/communication technologies.
Keywords: knowledge spillovers; patent citations; home bias; econometric models
JEL Codes: F23; O32; O33
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| models without fixed effects (C23) | strong home bias (G15) |
| foreign inventors citing domestic patents more slowly (O34) | home bias (F23) |
| coefficients for foreign citations become closer to zero (C29) | reduced home bias (F29) |
| unobserved patent characteristics, such as quality (L15) | apparent home bias observed in naive estimators (C51) |
| modern sectors like pharmaceuticals and information and communication technologies (L86) | little to no evidence of home bias (F29) |
| home bias has declined over time (F61) | faster knowledge transfer across geographic boundaries (O36) |
| decline in home bias (F29) | comparison of rejection rates across time periods (C52) |
| controlling for unobserved heterogeneity (C21) | evidence for home bias in patent citations (F23) |
| fixed effects introduced (C23) | evidence for home bias weakens considerably (F29) |