Fleeing a Lame Duck: Policy Uncertainty and Manufacturing Investment in US States

Abstract

It is found that electorally induced policy uncertainty decreases manufacturing investment in US states. In a state with average partisan polarization, the elasticity of election-year investment to a specific measure of policy uncertainty is −0.027. When the incumbent governor is term limited, there is greater uncertainty over the outcome, providing an instrument to demonstrate this effect is causal, not simply coincidental. Moreover, manufacturing investment does not rebound following the election. Rather, own-state uncertainty is associated with a large and significant coincident rise in neighboring states’ investment. These findings suggest that policy uncertainty at the subnational level drives investment to alternate sites. (JEL D25, D72, E22, G31, L60, R11)

The Real Effects of Checks and Balances: Policy Uncertainty and Corporate Investment

“…A number of articles have used national elections (e.g. Julio and Yook, 2012;Durnev, 2010;Kelly, Pástor, and Veronesi, 2016) as well as U.S.gubernatorial elections (Gao and Qi, 2013;Colak, Durnev, and Qian, 2017;Jens, 2017;Falk and Shelton, 2018) Alabama, Alaska, Arizona, Arkansas, California, Colorado, Connecticut, Florida, Georgia, Hawaii, Idaho, Illinois, Iowa, Kansas, Maine, Maryland, Massachusetts, Michigan, Minnesota, Nebraska, Nevada, New Mexico, New York, Ohio, Oklahoma, Oregon, Pennsylvania, Rhode Island, South Carolina, South Dakota, Tennessee, Texas, Wisconsin and Wyoming hold gubernatorial elections in even numbered years non concurrent with the Presidential election. Delaware, Indiana, Missouri, Montana, North Carolina, North Dakota, Utah, Washington, and West Virginia hold their gubernatorial elections in the same year as presidential elections.…”

“…In addition, Compustat firms are generally large multi-state firms that may be less sensitive to their headquarters-state policy than smaller single-state firms. To address this concern I also report results in table 5 from similar regressions for the smallest firms, which are more likely to be geographically concentrated in their home state (Jens, 2017 (Falk and Shelton, 2018). I adopt a different identification strategy to address the last two points, using a regression discontinuity design.…”

“…First, there could be an omitted variable bias such that unobserved state-level economic conditions or anticipated state-level economic conditions are driving both the election outcomes and the investment cycle. Second, the difference-in-differences specification may not capture the full effect of switching when transitions are correctly and easily anticipated leading firms to take corrective actions ex-ante such as shifting investment from one state to another, biasing firm-level result towards zero (Falk and Shelton, 2018). The RD design exploits the pseudo-random variation regarding which party wins elections generated by the uncertainty of close races.…”

“…If anticipated economic conditions are not fully captured by the covariates (controlling for past economic conditions), then the documented drop in investment may just reflect a negative state economic outlook. Second, the DD does not capture the full effect of the switch, as many transitions are correctly and easily anticipated leading firms to take corrective actions ex-ante such as shifting investment from one state to another(Falk and Shelton, 2018) biasing firm-level result towards zero.27 …”

“…Falk and Shelton, 2018 show how businesses respond to electoral uncertainty during gubernatorial election years by decreasing investment in their own state and increasing investment in neighboring states.…”