Choice of Law in Federal Courts: From Erie and Klaxon to Cafa and Shady Grove
Roosevelt, Kermit, Northwestern University Law Review
ABSTRACT-Erie is one of our canonical cases. Everyone agrees that it is important, but that is about all they agree on. Different understandings of Erie abound, and Erie analysis is notoriously complex and confusing. This Article offers a simpler way to approach the Erie problem: a choice-of-law methodology I call the two-step model. It shows how Erie appears somewhat more straightforward when viewed from the choice-of-law perspective, and how the two-step model solves some of the most notorious Erie problems. Beyond the payoffin terms of clarifying choice-of-law analysis in federal courts, the model's success in this context should serve as an advertisement for its use in choice of law more generally.
Everyone agrees that Erie Railroad v. Tompkins is an important case.1 And at a high level of generality, everyone also agrees on what the case stands for: federal courts exercising diversity jurisdiction apply federal procedure but state substantive law.2 Dig any deeper than that, though, and the agreement dissolves.3
In fact, Erie analysis is notorious for the puzzles it has produced. Does the case have any import for state courts? What happens when a federal court transfers a case to a district in another state? How should federal common law created to fill gaps in the Federal Rules of Civil Procedure interact with state law deemed substantive for Erie purposes? Does the Constitution require federal courts to follow state choice-of-law rules? Or might they have the power to modify such rules in the context of a class action?
The Supreme Court has offered answers to some of these questions,4 but the Justices have usually differed amongst themselves, and their work has received mixed reviews.5 My aim in this Article is to provide a better way of thinking about the Erie problem.
That better way is a conceptual framework developed in the choice-oflaw context, what I will call the "two-step model." It is appropriate for Erie analysis because, I will argue, Erie is in fact best understood as a choice-oflaw case. Part I of this Article will present that argument by engaging in a close reading of Erie to show that its conceptual structure is the same as that of the classic choice-of-law problem. Parts II and III then demonstrate that the understanding of Erie achieved by viewing it through the two-step model gives us new insight into some difficult problems, both classic Erie puzzles and more recent choice-of-law conundrums.
But solving those problems is not my sole or final goal. I have argued in other publications that this model is the correct way to conceptualize choice-of-law problems, but it is by no means universally accepted in that context.6 I hope that its success in resolving some of these problems will serve as an advertisement for its use in choice of law more generally. I hope to show, that is, that Erie is fundamentally a choice-of-law case, that we can deal with the problems it creates by using the two-step model, and that the model's success suggests that it is indeed the correct way to think about choice of law.
A. The Lower Courts and the General Law
One dark night, Pennsylvania resident Harry Tompkins was walking alongside the tracks of the Erie Railroad in Hughestown, Pennsylvania when he was struck and knocked down by an object-probably the swinging door of a refrigerator car-protruding from a passing train.7 As he fell, his right arm went under the wheels of the train.8 Most of the arm was severed, and the remainder was later amputated by doctors.9 Tompkins sued the railroad in federal district court in New York, claiming the railroad had been negligent in allowing the door to swing open.10
A crucial question was the duty the railroad owed to Tompkins, who was a trespasser on its right-of-way. Pennsylvania courts had ruled that for pedestrians on a path parallel to the tracks (as opposed to a crossing), the railroad had a duty only "to refrain from willful or wanton injury. …