We examine whether economic growth is generated endogenously or exogen
ously, and estimate the externality effects due to private and public
capital respectively. Applying a multivariate stochastic cointegration
method to US data, we find that the evidence is unfavourable to the e
ndogenous growth model with public infrastructure. The estimated elast
icity of output with respect to public capital is 0.11, smaller than t
ypical values obtained in single-equation regression studies, On the o
ther hand, if the shave of capital income is taken to be one-third, th
en the spillover effect due to private capital is positive but may be
as low as 0.10.