Structural Breaks in Public Infrastructure Investment in the U.S.

This paper investigates the possible existence and timing of structural breaks in public infrastructure investment in the US. Results are obtained using both the conventional Chow tests as well as the multivariate Bai, Lumsdaine and Stock tests which are more appropriate for finding breaks in the context of a VAR structure. All tests consider aggregate public investment as well as nine different disaggregated categories of public investment. In addition, multivariate tests include private output, employment, and investment, the variables usually considered in the literature. Empirical results suggest that a break in the mean of public investment occurred in the late 1970s or early 1980s. This is true at the aggregate level as well as for most types of public infrastructures. The measurement of the economic effects of public investment is a critical piece of information to understand the effects of any fiscal stimulus package. The empirical evidence in the literature is somewhat inconclusive at least partially because structural breaks have been largely ignored. We regard the evidence presented in this paper as the first step in the process of revisiting the analysis of the effects of such investments in a VAR framework while accounting for the presence of structural breaks.