Widely disparate results have flowed from various attempts to analyze the impact of public investment in tangible infrastructure on measures of economic activity. The author takes the substantial body of data put together by Munnell and Cook for 48 states over the years 1970 to 1986 and uses the data in pooled time series regressions, in pooled cross sections, and finally in distributed-lag investment functions.

The author's results support Munnell's finding that states that have more capital have greater output, even after taking into account both their amounts of labor (nonagricultural employment) and private capital. The author notes that serious questions remain, however, as to which is the cause and which is the effect.