Some panel cointegration models of international R&D spillovers

Coe and Helpman (1995) estimated a relationship between TFP and levels of domestic and foreign R&D capital, but couldn't provide compelling evidence of the panel cointegration needed to support their estimation strategy. This paper uses Pedroni's (1997, 1998) tests for panel cointegration in both the Coe-Helpman setup and in a framework with more cross-section heterogeneity. Criticisms of the Coe-Helpman approach are also tested. Coe and Helpman's models exhibit cointegration, but when more heterogeneity is allowed, coefficient estimates appear less robust. The elasticity of productivity with respect to foreign R&D is unstable across alternative estimation strategies.