Arbitrary Modeling = Arbitrary Decision

Sometimes a political decision shouldn't be dressed up to look like anything else.

The story reads like a textbook.  Politician scores highway pork the old fashioned way, sneaking it into an appropriations logroll nobody ‘took charge of.’  Said bill puts a road project on the work program which nobody much cared for or thought necessary.  Sheboygan-City-SignTo hear the court describe it, nobody really even wanted to know about this project—let alone actually to build it. But the zero-sum politicking of federal highway subsidies waits for no one.  And the Federal Highway Administration (FHWA) and Wisconsin Department of Transportation (WisDOT) know that score.  The court described the dynamics:

The Highway 23 project has a long history, and the administrative record contains several draft and final impact statements that FHWA and WisDOT prepared over the years. In the late 1990s, state legislators with districts in the area of Sheboygan and Fond du Lac began to pressure WisDOT to schedule the expansion of Highway 23 to four lanes for the entire distance between the two cities. . . . In 1998, WisDOT informed one of the legislators that it had no plans to commence such an expansion project soon, and that the project could not be planned until the state’s Transportation Projects Commission (a commission composed of state elected officials and private citizens) recommended the project for “enumeration” in the state’s next budget. . . .  However, in 1999, without waiting for a recommendation from the Transportation Project Commission, the state legislature enumerated the project in a budget bill, with the result that WisDOT was required [under state law] to proceed with construction.

1000 Friends of Wisconsin v. U.S. Dept. of Transportation (E.D. Wisc.2015) (Slip Opinion) at 2 (internal citations omitted).  So began a ground war to widen Highway 23 between Plymouth (Sheboygan) and Fond du Lac, a highway nobody seems to think really needs widening.  At a cost of about $150 million.

Because FHWA was involved, a NEPA environmental impact statement was required.  That was back in 2004.  Once this EIS got underway, things got interesting.  The EIS stated that

the purpose of the project was ‘to provide additional capacity to serve existing and projected traffic volumes and to improve operational efficiency and safety for local and through traffic.’  As part of the NEPA process, various state and federal agencies submitted written comments on the draft statement. Members of the public also commented on the project. Many comments focused on whether it was really necessary to expand the highway to four lanes. These commentators questioned whether traffic levels on the highway would require additional capacity at any time in the near future, and they wondered whether any needed additional capacity could be supplied by making less drastic changes to the highway, such as by adding passing lanes in certain areas.

Slip Opinion at 2.  In short, people—including plaintiffs 1000 Friends of Wisconsin (1KFriends)—started asking hard questions.  And the EIS helped them do so.

An Unjustified Project is Hard to Justify without the Right Tools

The Highway 23 case might’ve ended with a district court deferring to political actors making a political decision—a road here instead of somewhere else that needs it more—if it hadn’t been for the way WisDOT and FHWA tried to seal things up.  In order to show that Highway 23 really ought to be widened, they decided to use some modeling techniques they thought could make their case.

The problem is that plaintiffs today are often sophisticated enough to cut through phony modeling.  And they can explain it to a generalist judge who is willing to hear them out.  And that’s what happened here.  After making their case to FHWA while the EIS was being developed, 1KFriends went to court and then made their case to Judge Lynn Adelman.  They challenged the methods FHWA and WisDOT used to forecast traffic volumes on the section of 23 in question.  They argued that the flawed methods then rendered “deficient the defendants’ evaluation of reasonable alternatives to the proposed action.” Slip Opinion at 2.  (If you think there’ll be more traffic than there’ll really be, you’ll probably botch the alternatives analysis, too.)  Lastly, 1KFriends argued that the defendants gave inadequate consideration to the impact of “induced travel,” something other agencies and members of the public had raised.

The fascinating part is how easy Judge Adelman made it look cutting through FHWA and WisDOT’s use of biased sampling and model selection—something that prevailing arbitrariness review doctrine would’ve allowed the judge simply to ignore.  Cf. Baltimore Gas & Electric Co. v. Natural Resources Defense Council, 462 U.S. 87, 103 (1983) (observing that an agency making predictions within its special area of expertise which may be “at the frontiers of science” is entitled to special deference from a reviewing court).  He described the choices in these terms (simplified to Model A versus Model B):

[Model A] uses historical traffic counts to project by how much traffic volumes can be expected to grow in the future. It does not take other factors into account, such as expected changes in the population of the area, the expected locations of businesses and residences, or the number of lanes of the roadway. . . . In contrast, [Model B] is a model that attempts to account for factors other than historical traffic counts. It incorporates information about road networks, land use, demographics, and economic conditions, “analyze[s] future land use development scenarios to predict how and where future roadway traffic will go,” and “accounts for anticipated changes in population and employment in specific locations.”

Slip Opinion at 3-4.


Math is Hard?

Even if the agencies average the values generated by Model A and Model B, if Model A is just plain wrong because the future won’t look like the past, then averaging A and B will be wrong, too.  Simplified to its essence, the modeling isn’t hard to understand. And inexplicable choices become, well, inexplicable.  Indeed, this was exactly what 1KFriends alleged, and to stinging effect:

A related problem is that the defendants have not adequately explained how recently updated demographic data might affect the traffic projections that appear in the impact statement. When WisDOT built [Model B], it used [outdated] population projections . . . to estimate growth in the region through the year 2035. Shortly before the defendants released the final impact statement in March 2014, [updated] population projections [were released, showing] a lower growth rate than was reflected in the Department’s 2035 projections. The plaintiff contends, and the defendants do not dispute, that the updated population projections . . . showed that the population in the area of Highway 23 would grow only about one-third as quickly [as had been] previously projected.

Slip Opinion at 6 (internal citations omitted).  And that means that previously dismissed alternatives of a lesser development intensity perhaps shouldn’t have been so summarily dismissed.

Modeling Done Well is Justifiable. Modeling Done to Paper Over Bad Decisions is a Sham

Environmental lawyers have grown a lot more familiar with modeling in the last decade.  More work is still to be done.  The advocacy in this case that got a district judge to look past the easy way out, i.e., Baltimore Gas’s dictum about ‘prediction within the agency’s peculiar area of expertise,’ was exemplary.  And Judge Adelman denied most of the plaintiffs’ claims.  But the claim that the models were either intentionally jiggered with stale data or at the very least used to render a skewed analysis opaque was not.  More work like this showing (1) impenetrable methods and/or inferences in the agency’s forecasting, and (2) the flawed alternatives analysis that flowed therefrom, is needed—and desperately.

Judge Adelman’s remand to FHWA ordered the agency to fix its EIS, putting the Highway 23 project on hold unless Wisconsin can fund it without federal funds.  But he denied an injunction, citing Monsanto v. Geertson Seed Farms, 130 S.Ct. 2743 (2010).  highway funniesStill, with the federal highway trust fund ailing like it is, this might be the perfect opportunity to kick the Highway 23 project.  At the very least, someone’s going to have to explain in plain English why this project is needed, why it will serve that need better than other alternatives, and do so without cooking up more expected growth in southeastern Wisconsin’s future than responsible demographers actually project.  Good luck with that.

I teach environmental, natural resources, and administrative law at Penn State Law. Before teaching I was an enforcement lawyer at U.S. EPA. Along the way I've done work for environmental nonprofits and written a fair bit about NEPA.
No Comment

Leave a Reply