This paper implements an improved methodology for estimating the effec
ts of infrastructure spending on aggregate output and shows the result
s of previous studies to be biased end imprecise. previous studies hav
e found the elasticity of aggregate output with respect to infrastruct
ure spending to be in the 0.3 to 0.6 range. Literally hundreds of news
paper articles have referenced these specific results in discussing th
e need for Increased spending. Using a transfer function model which a
ccounts for nonstationarity and serial correlation. we find the aggreg
ate output elasticity of net nonmilitary public investment is about 0.
03.