Association for Public Policy Analysis & Management

Floods are part of the earliest recorded history of the Mississippi. Since the twentieth century, an increasing number of significant flood events have occurred in the Mississippi. As a means of minimizing flood vulnerability, a levee is a man-made structure, usually an earthen embankment, to contain, control, or divert the flow of water to reduce risk from flooding. While the purpose of levee building is to protect farmland and properties behind levees from flooding, levee building has gradually been recognized to cause more frequent and severe floods on the Mississippi River, as levees force the water that would otherwise spread across the area to go through a narrower passageway. As the passageway gets narrower, the water flows faster and higher. In addition, when one levee is taller than the other the extra levee height can shift some large floods to the other side or downstream of the river, inducing negative externality.

This paper provides both theoretical and empirical evidence on levee building in response to flooding. In specific, this paper seeks to address the following two questions. First, how does a jurisdiction’s optimal level of levee building respond to exogenous flooding shocks when there is negative externality from its neighbors’ levee building? Second, how large is the negative externality when levees are overbuilt? A conceptual framework that derives the optimal level of levee building generates two relationships of interest: (1) the optimal level of levee building increases with the probability of flooding; and (2) the difference in the optimal level of levee building between the case when there is influence from neighbors’ levee building and the case when there is no such influence also increases with the probability of flooding. Taking the Great Flood of 1993 in the Upper Mississippi River as a natural experiment, we empirically test for the competitive levee building in response to flooding and further examine the magnitude of the negative externality from neighbors’ levee building. Using a triple-difference identification strategy that exploits geographic variation in flood coverage, county variation in levee building, and temporal variation in pre- and post-flood, we estimate the economic impacts of levee building on agricultural development.

Preliminary results from this study show that consistent with our model predictions, the Great Flood of 1993 increased levee heights in flooded counties. Heterogeneity in levee height increase is substantial, however, depending on the locality of the levee: levees located in counties that are downstream of counties with levees had the highest height increase. We also find that farm numbers and acres of cultivated cropland in flooded counties decreased relative to similar non-flooded counties, suggesting a negative impact of floods on agricultural development. More importantly, flooded counties without levee systems but adjacent to counties with levees had the highest decrease in agricultural development. This result implies that levee building induces negative externalities across jurisdictional boundaries by increasing the damage from floods to neighbors. Without considering the negative externality from levee building, policymakers may overestimate the cost-effectiveness of levees in flood control.