The question of whether government regulation is, on aggregate, helpful or harmful has been widely studied in economics. However, an equally important question is whether or not regulation is selectively enforced, and this has received substantially less attention from academic researchers. Because regulation can hurt voters as well as help them, the vigorous enforcement of government regulation can be costly to politicians and other government officials, including the regulators themselves. Thus there are often a set of incentives that can result in politicians and other political actors being reluctant to enforce regulations among politically important constituents.
To examine whether or not regulation is selectively enforced in the U.S., we identify a set of regulations related to the enforcement of an economically important statute that applies uniformly to the 50 U.S. states: the Clean Water Act of 1972. We also identify a source (which varies over time) of the state-level political importance of voters: whether or not a state is a “battleground” or “swing” state in presidential elections. We find that the Environmental Protection Agency (EPA) is much less likely to find facilities in battleground states to be in violation of the CWA than facilities in non-battleground states, consistent with regulation being less stringently enforced among important voters.
The CWA requires that facilities which discharge water directly into the surface waters of the United States obtain a permit to do so, and these permits are tracked by the EPA’s National Pollutant Discharge Elimination System (NPDES). The NPDES data is publicly available and goes back to at least 1976. In addition to tracking permit information, the NPDES data also records several different categories of permit-related CWA violations (e.g., a facility emits water with pollutant levels above those allowed by their permit).
We hypothesize that, because of the importance of these states in presidential elections, regulatory agencies (whose leadership is typically appointed by the president) are more likely to exhibit leniency in the enforcement of government regulation in battleground states. Thus, we wish to determine whether facilities in these states are less likely to be found in violation of the CWA by the EPA. A key difficulty in these kinds of studies is establishing causality. Specifically, we wish to show whether being located in a battleground state causes a reduction in violation rates.
A key feature of our data allows us to do this. Specifically, the NPDES data includes the latitude and longitude of each facility that is registered with the system. This locational data allows us to match facilities in battleground states with similar facilities that are located nearby in non-battleground states (and vice versa). Thus we create a “border” sample of 11,398 unique and similar facilities that are located along the border of battleground and non-battleground states. Aside from being located in different states, and since rivers and streams often serve as the border between states, this frequently means that these otherwise similar facilities differ only by being located across from each other along a river or stream (and thus are likely subject to similar economic and demographic trends and, importantly, ostensibly similar treatment under the CWA). Thus, we argue, we are able to isolate effects that are driven by changes to a state’s battleground status.
In the most basic analysis, we find that, for any given year between 1976 and 2014, facilities have only an 11.3 percent chance of being found in violation of the CWA if they are located in a state that was a battleground state in the most recent presidential election, while facilities in non-battleground states have a 23.7 percent chance. That is, facilities in battleground states experience a 51.9 percent (or 12.3 percentage point) decrease in the likelihood that they will be found to violate the CWA in a given year. This effect is large and statistically significant. Unless becoming a battleground state causes facilities to be more effective in their compliance with environmental regulation, this finding is more consistent with regulations being more loosely enforced.
We also examine the changes to violation rates among facilities that are located in states that change battleground state status after an election. For example, after an election where a non-battleground state becomes a battleground state, the facilities in that state subsequently experience lower violation rates. The reverse holds true for facilities located in battleground states that revert to non-battleground state status after an election.
We also examine this battleground state effect in several multivariate regressions where we control for additional factors that might affect violation rates. In these regressions we include explanatory variables like the number of electoral votes allocated to the state in which the facility is located, state-level unemployment rates as well as state per-capita income levels. We also include “fixed effects” to control for any other constant state effects (such as whether the state is generally Republican or Democrat), time effects (which capture any effects that vary by year, such as stock market returns, business cycle effects, etc.), and industry-by-year effects (such as regulatory changes that affect each industry differently or other industry-level trends). We find results that are similar to those previously outlined: facilities located in battleground states have violation rates that are as much as 6.9 percentage points lower than similar nearby facilities located in non-battleground states.
We run a number of additional tests to examine the particulars of this effect. Does it vary depending upon the party of the president? We find that it does not; whether the president is a Republican or Democrat makes little difference. Does it vary by the year in the elections cycle? Perhaps the increased scrutiny of battleground states results in their being more compliant (and therefore experience lower violation rates) in election years. We find that the effect is evenly spread out across a four-year presidential election cycle, and is not concentrated by effects in an election year only. We also repeat the analysis using the full sample of 288,490 unique facilities (i.e., we include all NPDES facilities, and not just those located along state borders) and look at different types of NPDES violations. We find that the general result holds in the full sample as well as across violation types: facilities in battleground states are significantly less likely to be found in violation of the CWA, and across different categories of NPDES violations.
Finally, we examine how likely it is that this effect is driven by our method of determining battleground states. To do this, we run simulations where we randomly assign battleground state status to the fifty states and repeat some of our analysis. We find that, in 10,000 simulations, between 0.0 to 1.5 percent of randomly assigned battleground states yield results similar to those we find. It is thus very unlikely that this effect is driven by our method for assigning battleground state status.
Why would the EPA lightly enforce the CWA in battleground states? We do not answer this question definitively, though we speculate it is related to the political incentives incident to presidential elections. In particular, we argue that, because the perceived state of the economy plays a large role in the re-election of incumbent politicians (or their party), presidents have incentive to decrease regulatory burdens when they may have a negative economic impact, particularly on important constituents (which, in the Electoral College, includes voters in battleground states). Consistent with this, we offer supportive (but limited) evidence that decreased CWA enforcement results in increased state-level votes to the party of the incumbent president.
We are also quick to observe that our study does not touch on the efficacy of environmental law. Whether the lighter enforcement of the CWA results in a positive effect to the general welfare (e.g., avoidance of unnecessary regulatory costs) or a net negative effect (e.g, greater negative health and other environmental outcomes) is beyond the scope of our study.
These caveats notwithstanding, we demonstrate that, whatever the mechanism or the effect on public welfare, government regulation is selectively enforced among politically important constituents, this effect is large, and it is persistent. We argue that the political debate over government regulation should not include discussions only about its anticipated effects but also whether it will be uniformly enforced.
This post comes to us from Professor Huseyin Gulen of Purdue University’s Krannert School of Management and Professor Brett W. Myers of Texas Tech’s Rawls College of Business. It is based on their paper, “The Selective Enforcement of Government Regulation: Battleground States and the EPA,” available here.