Are there International RandD Spillovers Among Randomly Matched Trade Partners? A Response to Keller

Abstract

Keller (1998) reexamines Coe and Helpman's (1995) analysis ofinternational R&D spillovers focusing on the weights used to define the foreign R&D capital stock. Keller creates "random" weights and shows that they give rise to positive estimates of international R&D spillovers, casting doubts on the robustness of Coe and Helpman's findings. We show that Keller's "random" weights are essentially simple averages with a random error. We derive alternative random weights and present regressions showing that when they are used to define the foreign R&D capital stock, the estimated international R&D spillover estimates are nonexistent, as would be expected.