Bifurcating Settlements

GW Law Faculty Publications & Other Works Faculty Scholarship

2018

Bifurcating Settlements Bifurcating Settlements

Michael B. Abramowicz

George Washington University Law School

, abramowicz@law.gwu.edu

Sarah Abramowicz

Wayne State University

Follow this and additional works at: https://scholarship.law.gwu.edu/faculty_publications

Part of the Law Commons

Recommended Citation Recommended Citation

Abramowicz, Michael B. and Abramowicz, Sarah, Bifurcating Settlements (2018). Bifurcating Settlements,

86 Geo. Wash. L. Rev. 376 (2018) (with Sarah Abramowicz).; GWU Law School Public Law Research Paper

No. 2019-16; GWU Legal Studies Research Paper No. 2019-16. Available at SSRN: https://ssrn.com/

abstract=3366631

This Article is brought to you for free and open access by the Faculty Scholarship at Scholarly Commons. It has

been accepted for inclusion in GW Law Faculty Publications & Other Works by an authorized administrator of

Scholarly Commons. For more information, please contact spagel@law.gwu.edu.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 1 18-JUN-18 16:24

Bifurcating Settlements

Michael Abramowicz*

Sarah Abramowicz**

BSTRACT

In settling a lawsuit, parties agree on their obligations to one another, but

they need not separately address each issue, claim, or remedy that a trial court

would have confronted. The legal system, however, can bifurcate the settle-

ment process, requiring separate resolution of components of a settlement. Bi-

furcation can protect third parties, for example, by preventing divorcing

parents from trading child custody for money. In addition to identifying a

wide range of contexts in which preventing trade-offs may be desirable, this

Article shows that bifurcation will generally have only modest (and sometimes

beneficial) effects on settlement rates.

ABLE OF

ONTENTS

NTRODUCTION

................................................. 377

I. M

ODELING THE

FFECT OF

IFURCATION ON

ETTLEMENT

ATES

.................................... 381

A. The Effects of Separate Negotiation ................. 382

B. Bifurcating Independent Claims ..................... 385

1. Bifurcated vs. Aggregated Settlements .......... 385

2. Severable Settlements .......................... 389

II. P

ROTECTING

HILDREN IN

IVORCE

ASES

............ 391

A. The Problem of Custody Trade-Offs ................ 393

1. Custody Trade-Offs and Harm to Children ..... 394

2. Custody Trade-Offs and Harm to the State ..... 396

3. Custody Trade-Offs and Improper Leverage .... 397

B. Previous Approaches to Custody Trade-Offs ........ 398

1. Changes to the Substantive Law ................ 398

2. Judicial Review ................................. 401

3. Changes in Terminology ........................ 401

C. Bifurcating Custody Settlements ..................... 402

1. Costs of Bifurcation ............................ 406

* Professor of Law, George Washington University.

** Associate Professor of Law, Wayne State University.

For helpful comments, the authors thank Kirsten Carlson, David Fontana, and participants

in workshops at the George Washington University Law School and the Wayne State University

Law School. For excellent research assistance, the authors thank Komal Shah. All errors are our

own.

March 2018 Vol. 86 No. 2

376

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 2 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 377

a. The Solomonic Problem and Settlement

Accuracy .................................... 406

b. Potential Adverse Effects on Settlement ..... 408

2. Obstacles to Bifurcation ........................ 410

a. De Facto Trade-Offs ........................ 411

b. Modifiability of Custody Arrangements ..... 412

3. Design of Bifurcation ........................... 413

III. P

ROTECTING

HIRD

ARTIES

’ F

INANCIAL

NTERESTS

... 415

A. Protecting the Government: Punitive vs.

Compensatory Damages ............................ 416

B. Protecting Consumers: Damages vs. Market

Exclusion ........................................... 423

C. Protecting Class Members: Attorneys’ Fees vs.

Damages ............................................ 430

ONCLUSION

................................................... 436

NTRODUCTION

Typically, a civil settlement disposes of the parties’ claims without

separately resolving each underlying legal issue.

A suit sounding in

tort and contract, for instance, might settle for a lump sum, without

allocating the amount between the different claims.

When the plain-

tiff brings multiple claims or seeks multiple remedies—injunctive re-

lief and damages,

or punitive and compensatory damages

—the

parties may make trade-offs. These trade-offs are to the parties’ mu-

tual advantage and are thus usually benign. But in some contexts, such

trade-offs may have adverse effects on third parties. In a divorce case,

for example, the parties might agree to give a mother more custody in

Settlements generally bar reconsideration even of claims that had not been litigated.

See, e.g., Nashville, Chattanooga & St. Louis Ry. Co. v. United States, 113 U.S. 261, 266 (1885).

Parties, however, can voluntarily choose to settle only some claims. See, e.g., 15A C.J.S. Compro-

mise & Settlement § 5 (2017) (“Although the law favors full settlements over partial settlements,

the parties may settle any part of a controversy and leave the rest for litigation.” (footnote

omitted)).

See, e.g., Jon O. Shields, Note, Exclusion of Damages Derived from Personal Injury

Settlements: Tax-Planning Considerations in Light of McKay v. Commissioner, 56 M

ONT

. L. R

603, 603 (1995) (relaying a problem resulting from a lawsuit “grounded in both tort and con-

tract,” where the “settlement agreement . . . did not specify which claims were satisfied by the

payment”). There is ordinarily no need to allocate damages among claims.

It is common for a party seeking injunctive relief to accept monetary relief instead. Cf.

Ian Ayres & Kristin Madison, Threatening Inefficient Performance of Injunctions and Contracts,

148 U. P

. L. R

. 45, 47 (1999) (“A potential plaintiff who is owed a duty may, at times, seek

inefficient injunctive relief instead of damages merely to induce a defendant (the person owing

the duty) to pay an amount higher than expected court-awarded damages.”).

See infra Section III.A.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 3 18-JUN-18 16:24

378 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

exchange for less financial support than the parties would expect a

court to award,

potentially to the children’s detriment.

The law might seek to prevent such trade-offs by forcing the par-

ties to negotiate specified issues separately. Much as civil trials are

sometimes bifurcated so that some issues are tried separately from

others,

so too might the law insist on settlement bifurcation. The sim-

plest way of achieving this would be to require the parties to finalize

the settlement of certain issues sequentially. Although parties could

jointly discuss all the issues in the lawsuit, an agreement on an issue to

be determined later in the sequence would be unenforceable if signed

at the same time as or before an issue designated for earlier resolu-

tion. In the divorce context, a mother could still make a concession on

financial support in the hope that her soon-to-be-ex-husband would

follow through on a promise to make a concession in a later agree-

ment on child custody. But if she did not entirely trust him, she would

not be willing to do so. Settlement bifurcation seeks to take advantage

of the mistrust that often pervades litigation, in commercial and tort

contexts as well as family law, to thwart trade-offs between different

components of a settlement.

Especially if parties are required in final-

izing the first part of a deal to attest that they understand that the

other issues in the lawsuit remain open to negotiation, many litigants

will hesitate to make a concession in exchange for an unenforceable

promise of later reciprocation.

The idea that the law might require some issues to be settled

before others has received only fleeting recognition in the literature:

for example, a suggestion at the end of a study on custody litigation

and a brief mention in an article on reverse patent settlements.

The

courts appear to have considered the possibility in just one context—

See infra Part II.

See, e.g., Steven S. Gensler, Bifurcation Unbound, 75 W

ASH

. L. R

. 705, 710–11 (2000)

(arguing that bifurcation of issues at trial can improve the accuracy and the efficiency of the

justice system); see also John P. Rowley III & Richard G. Moore, Bifurcation of Civil Trials, 45

U. R

ICH

. L. R

. 1 (2010) (focusing specifically on bifurcation under Virginia law).

The more parties trust each other, the less effective bifurcation is. One can distinguish

between “calculative trust”—i.e., trust based on recognition of another’s incentives—and “emo-

tional trust.” See Christopher R. Leslie, Trust, Distrust, and Antitrust, 82 T

. L. R

. 515,

529–32 (2004). Bifurcation effectively reduces calculative trust by furnishing economic incentives

to break informal agreements incorporating trade-offs. Emotional trust, however, may thwart

bifurcation.

See Scott Altman, Lurking in the Shadow, 68 S. C

. L. R

. 493, 527 (1995) (recogniz-

ing the application of bifurcated settlements to custody negotiations); see also discussion infra

Part II.

See C. Scott Hemphill, An Aggregate Approach to Antitrust: Using New Data and

Rulemaking to Preserve Drug Competition, 109 C

OLUM

. L. R

. 629, 686 (2009) (recognizing a

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 4 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 379

class actions—where attorneys’ fees and damages for the class are

usually jointly negotiated.

One case insists on bifurcation,

but other

cases take the opposite position, and the Supreme Court has rejected

settlement bifurcation in a similar context.

A potential reason that the few recognized available forms of set-

tlement bifurcation have gained little traction is that scholars and

judges do not want to discourage settlement. This Article shows, how-

ever, that significant reductions in settlement are unlikely.

Indeed,

when litigation is relatively expensive, bifurcation may enhance the

likelihood of settlement. As litigation stakes rise in comparison to trial

costs, parties are much less likely to settle because they have more to

gain from strategic bargaining. It thus may be more than twice as diffi-

cult to settle a claim for a million dollars than to settle a claim for half

a million, holding litigation costs roughly constant. Breaking a settle-

ment negotiation into smaller chunks through bifurcation may thus

promote settlement. At the very least, any reduction in the probability

of settlement attributable to bifurcation is likely small. Thus, bifurca-

tion is worth considering if there are sufficient benefits.

With the recognition that bifurcation is affordable comes a need

to theorize about different contexts in which it may be beneficial. Ab-

sent a clearly identified benefit, bifurcation is inappropriate because

allowing trade-offs will often promote the most efficient allocation of

resources between the parties. Consider a nuisance case seeking both

damages and injunctive relief. Partial abatement of a nuisance might

be impossible,

so any compromise must involve exchange of money.

Or, in a trademark case, a settlement might allow continued use in

exchange for a royalty payment from the defendant, or disallow it in

exchange for a compensation payment from the plaintiff.

When two

potential solution to the reverse payments problem in patent law); see also discussion infra Sec-

tion III.B.

See infra Section III.C.

See Prandini v. Nat’l Tea Co., 557 F.2d 1015, 1021 (3d Cir. 1977).

See White v. N.H. Dep’t of Emp’t Sec., 455 U.S. 445, 452 (1982) (discussing fee determi-

nations in civil rights cases).

See infra Part I.

Partial abatement may be possible where a party can reduce its nuisance activity with-

out eliminating it completely. See, e.g., Thomas W. Merrill, Golden Rules for Transboundary

Pollution, 46 D

UKE

L.J. 931, 1014–15 (1997) (discussing polluter’s agreement for partial

abatement).

See Charles J. Faruki, Litigation Involving Trademarks: Preparing the Trademark Case

for Trial, 16 U. D

AYTON

L. R

. 85, 125 (1990) (“Payment to the defendant for a change of

name may be possible, and indeed may be the only way to avoid litigation when each side has an

investment in its own trademark.”).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 5 18-JUN-18 16:24

380 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

parties reach an out-of-court settlement, they are presumably satisfied

that the settlement is in their mutual interest. Ordinarily, the legal

system’s only interest is in enforcing the deal that the parties have

struck.

Sometimes, however, preventing parties from making trade-offs

or insisting that parties disclose how they resolved individual issues is

beneficial, particularly when the settlement affects a third party. For

example, punitive damages negotiations may affect the government,

either because such damages are taxable

(but compensatory dam-

ages are not) or because of a statute requiring partial forfeiture of

punitive damages to the government.

Bifurcating settlements may

thwart the litigants’ ability to allocate the portion of a settlement re-

flecting expectation of punitive damages to compensatory damages.

Similarly, consumers have an interest in settlements of intellectual

property disputes,

yet litigants may use settlements to collude to

keep prices high.

Bifurcation would prevent private litigants from

making trade-offs at the expense of consumers. In shareholder and

class action litigation, shareholders or class members have an interest

in assuring that representatives do not prioritize attorneys’ fees over

other forms of relief, and bifurcating attorneys’ fees from other issues

may promote this interest.

These examples highlight that scholars

and policymakers should consider bifurcation in any context in which

parties may make trade-offs at the expense of third parties.

Part I scrutinizes the assumption that bifurcation will decrease

the likelihood of settlement. Bifurcation may actually increase the

likelihood of settlement when legal costs are relatively high. Part II

offers a detailed analysis of bifurcation in the family law context. Di-

See, e.g., 26 U.S.C. § 104(a)(2) (2012) (allowing deductibility of “any damages (other

than punitive damages) received . . . on account of personal physical injuries”); Comm’r v. Glen-

shaw Glass Co., 348 U.S. 426 (1955) (holding punitive damages to be taxable income).

See generally Paul F. Kirgis, Note, The Constitutionality of State Allocation of Punitive

Damage Awards, 50 W

ASH

. & L

L. R

. 843 app. (1993) (collecting statutes). Many such

statutes have been repealed, perhaps in part because they were so easily evaded. See, e.g., C

OLO

. S

TAT

. A

. § 13-21-102(4) (West 1989) (repealed) (requiring 33% of punitive damages

awards to be paid to the state).

See infra Section III.A.

Nonetheless, they may not have standing to assert claims. See Einer Elhauge & Alex

Krueger, Solving the Patent Settlement Puzzle, 91 T

. L. R

. 283, 324 (2012) (“[P]atent law

ordinarily does not allow buyers to sue to prevent the anticompetitive exclusion of rivals through

invalid patents.”).

See generally C. Scott Hemphill, Collusive and Exclusive Settlements of Intellectual

Property Litigation, 2010 C

OLUM

. B

. L. R

. 685 (discussing the danger that collusive settle-

ments may adversely affect consumers of intellectual property goods).

See infra Section III.C.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 6 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 381

vorce is a rich context for analyzing bifurcation, not only because it

illustrates the potential benefits of a bifurcation regime in preventing

trade-offs that may harm the interests of third parties, but also be-

cause it highlights potential challenges of bifurcation. One concern is

that bifurcation may reduce the likelihood that parents will reach the

custody arrangement that best promotes their children’s interests by

limiting the ability of parents who care more about custody to sacrifice

financial well-being for assurance of custody. This concern, which we

term “the Solomonic problem,” illustrates both the information-pro-

ducing function of trade-offs and the corresponding role of trade-offs

in promoting settlement accuracy. Another problem is that parties

may make de facto trade-offs after a judgment is rendered. Finally,

Part III canvasses other contexts in which settlement bifurcation may

be useful if the costs are sufficiently low, including punitive damages,

reverse payments in patent law, and class action attorneys’ fees. In

some of these contexts, scholars have offered imperfect substantive-

law tweaks to problems without considering the procedural solution of

bifurcation. In others, there has been occasional recognition that set-

tlement might profitably be conducted sequentially, but this possibility

has received little consideration.

Overall, this Article demonstrates that settlement bifurcation

may be a useful tool when settlements may have negative effects on

unrepresented third parties. There is a risk that bifurcation may de-

crease settlement, but that risk is not as great as might initially appear.

Bifurcation prevents parties from making trade-offs in settlements at

the expense of third parties.

I. M

ODELING THE

FFECT OF

IFURCATION

ETTLEMENT

ATES

Section I.A demonstrates that bifurcation has multiple effects,

some of which increase the risk of trial and some of which decrease it.

Section I.B assesses the combination of these competing effects with a

simple simulation model. The case for bifurcation strengthens as liti-

gation costs increase relative to the stakes. With sufficiently high liti-

gation costs, bifurcation may even increase the likelihood of

settlement. Even where settlement becomes less likely, the costs of

bifurcation may be quite small. The benefits of bifurcation—identified

in Parts II and III—thus need not be high to justify the costs.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 7 18-JUN-18 16:24

382 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

A. The Effects of Separate Negotiation

Sometimes, it may be beneficial for settlement negotiators to dis-

cuss specifically their separate expectations of two issues, such as com-

pensatory and punitive damages, rather than just exchange gross

damages amounts. But there are currently no rules mandating this

practice. Perhaps a rule bifurcating settlements may help structure ne-

gotiation discussions in the same way that a mediator can serve that

role.

A rule requiring parties to agree formally on two components

of some aggregate number, rather than only on the aggregate number,

adds at least one variable they must resolve and thus increases the

potential for some additional challenge to the settlement process.

Once the parties settle one issue, they must still settle the other issue.

A prime driver of settlements is the parties’ expectation that they will

save on continuing litigation and on trial. But if settling one issue

merely reduces the probability of trial—because trial will still occur if

another issue fails to settle—then the benefits of settlement are re-

duced, and the incentive to settle is reduced as a result. On the other

hand, it may be easier to settle the first issue rather than the entire

case because a single issue is simpler and involves lower stakes than

the case as a whole.

To analyze more rigorously the effects of bifurcating, we can be-

gin with simple observations about the causes of settlement failure.

Because trial is costly, there will always be a range of settlement val-

ues better for both parties than the ultimate case outcome.

For ex-

ample, if a plaintiff would receive $1,000,000 at trial and each party

would pay $100,000 in trial costs, then any settlement between

$900,000 and $1,100,000 would be mutually advantageous. Even if

both parties knew the trial outcome with certainty, however, a case

may not settle because the parties engage in strategic bargaining,

seeking a relatively favorable value within this range. The more signif-

icant contributor to settlement failure is that parties may predict dif-

See Evan Slavitt, Using Risk Analysis as a Mediation Tool, D

ISP

. R

ESOL

. J., Nov.

2005–Jan. 2006, at 18 (highlighting the mediator’s role in breaking down issues).

“In the standard model of litigation, settlement range is defined as the set of settlement

offers that both the plaintiff and defendant would prefer to going to trial.” Tai-Yeong Chung,

Settlement of Litigation Under Rule 68: An Economic Analysis, 25 J. L

EGAL

TUD

. 261, 267

(1996) (emphasis omitted). Absent uncertainty about trial outcomes, a settlement range will

always exist. See Russell Korobkin & Chris Guthrie, Psychological Barriers to Litigation Settle-

ment: An Experimental Approach, 93 M

ICH

. L. R

. 107, 112 (1994) (“As long as the costs of

trial are higher than the costs of settlement, and as long as both sides make an identical estimate

of the likely outcome of the trial, the case should settle.” (footnote omitted)).

See Robert Cooter et al., Bargaining in the Shadow of the Law: A Testable Model of

Strategic Behavior, 11 J. L

EGAL

TUD

. 225, 226, 245–46 (1982).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 8 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 383

ferent trial outcomes.

These considerations are related. Each party

will want to convince the other that its offer reflects its expectation of

the litigation result, even when it is really engaging in strategic

bargaining.

There are two contingencies that make settlement failure more

likely. First, a particular case will be less likely to settle if the parties

are mutually optimistic about the outcome—i.e., the plaintiff’s expec-

tation of its probability of success is greater than the defendant’s ex-

pectation of the plaintiff’s probability of success and vice versa. This

could be the result of a cognitive optimism bias,

or it could be the

result of the parties having different information or different good-

faith interpretations of the same information. Second, settlement is

less likely when trial costs are relatively low in comparison to claim

stakes. For example, a plaintiff who expects to receive $1,000,000 at

trial would settle for no less than $900,000 (90% of claim value) if its

trial costs would be $100,000. But if the plaintiff expects to receive

$500,000 with the same trial costs, the plaintiff would be willing to

settle for no less than $400,000 (just 80% of claim value). A billion-

dollar case that would cost a million dollars to litigate will go to trial if

the parties are even a bit mutually optimistic; a thousand-dollar case

that would cost a million dollars to litigate will almost never go to

trial, even if the parties are highly mutually optimistic.

With this background, we can analyze the effects of bifurcation

more rigorously. Suppose a plaintiff has two claims against a defen-

dant, Claim 1 and Claim 2. These claims, we will assume for now, are

both drawn from the same distribution—Claim 1 is equally likely to

be larger or smaller than Claim 2—but are independent. In the tradi-

tional settlement regime, the parties settle Claim 1 and Claim 2 aggre-

gated together, so each party estimates each claim, sums these

estimates, and negotiates with the other party about the total pay-

ment. With what we will call “seriatim bifurcation,” the parties start

by negotiating Claim 1, and only if that is successful, negotiate Claim

2. An alternative approach to bifurcation that we call “severable set-

tlements” is discussed below.

To begin our analysis of bifurcation, assume that Claim 1 has set-

tled in accordance with the game-theoretic strategy of backward in-

See, e.g., J.J. Prescott & Kathryn E. Spier, A Comprehensive Theory of Civil Settlement,

91 N.Y.U. L. R

. 59, 61–62 (2016).

See, e.g., Korobkin & Guthrie, supra note 23, at 166. R

See infra Section I.B.2.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 9 18-JUN-18 16:24

384 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

duction. Game theory generally proceeds by backward induction,

because an earlier stage (for us, evaluation of Claim 1) depends on

what is expected to happen at the end of the game (for us, evaluation

of Claim 2 if Claim 1 is settled). Will it be easier to negotiate Claim 2

alone than to negotiate Claim 1 plus Claim 2? The likelihood of settle-

ment will be greater when the ratio of trial costs to stakes is higher, so

we must consider the numerator and denominator of this ratio. A trial

on Claim 2 alone may be cheaper than a trial on Claim 1 plus Claim 2.

Thus, the numerator of the ratio will be lower when negotiating Claim

2 alone. This will tend to make Claim 2 more difficult to negotiate

than the sum of Claim 1 plus Claim 2. It seems unlikely that this effect

will be large, however. A trial has high fixed costs, and if there is some

factual connection between the claims, the marginal savings from not

having to litigate Claim 1 are likely to be small. Meanwhile, the social

cost of any increase in the incidence of trial resulting from lower trial

costs will be reduced by the fact that such trials would be less

expensive.

The denominator effect is likely to be greater. The stakes will be

lower when negotiating Claim 2 alone. Because the stakes will likely

decrease more than trial costs, Claim 2 is likely to be easier to settle

than Claim 1 and Claim 2 in the aggregate. This brings us to the analy-

sis of Claim 1. This is more complicated because we cannot be sure

when settling Claim 1 whether Claim 2 will be settled, with the ex-

penses of trial averted. Thus, we must consider the expected trial cost

savings of the Claim 1 negotiation relative to the stakes of Claim 1.

When calculating the reduced trial costs that are the benefit of settle-

ment, we thus must multiply the trial costs by the probability that

Claim 2 settles, because Claim 1’s resolution will avoid trial only if

Claim 2 settles as well. The benefit of settling Claim 1 is then less than

the benefit of settling Claim 2. Continuing our assumption that a trial

involves fixed costs regardless of how many claims are at issue, Claim

1 alone should on average therefore be more difficult to negotiate

than Claim 2 alone. The probability of successfully settling Claim 1

and then Claim 2, meanwhile, will be the product of these probabili-

ties. This probability ultimately must be compared to the probability

of negotiating the two claims when aggregated.

It might seem that negotiating two small claims will be much

harder than negotiating a single larger claim. With sufficiently low

trial costs, this would be true. If litigation is costless, then a negotia-

See, e.g., M

ARTIN

J. O

SBORNE

& A

RIEL

UBINSTEIN

, A C

OURSE IN

AME

HEORY

99–100 (1994).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 10 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 385

tion will fail as a result of mutual optimism half of the time, whenever

the plaintiff’s estimate of the probability of liability is higher than the

defendant’s. Thus, with bifurcation, regardless of the claim value,

there will be a 75% trial rate, as settlement occurs only where both of

the individual settlements succeed. This highlights that the more items

there are to be negotiated, the more potential sources of settlement

failure exist.

The countervailing point is that once trial costs are sufficiently

large and settlement becomes very common, settlement rates will be

much higher with each small claim than with the larger claim. As soon

as trial costs are positive, the probability of settling a small claim will

be higher than the probability of settling a large claim. The higher the

stakes, the less important the trial costs, and thus the greater the

chance of settlement failure. Critically, the discrepancy in settlement

rates between each small claim and the larger claim becomes ever

larger as trial costs increase. For example, if there is a 5% chance that

parties would be so mutually optimistic that they would fail to settle

the large claim, there might be only a 1% chance that parties would be

so mutually optimistic that they would fail to settle one of the small

claims. As trial costs increase, the probability of settlement failure ap-

proaches zero twice as fast with the small claim as with the larger one.

Thus, settlement failure becomes much more likely with the larger

claim, and this is enough to compensate for the challenges associated

with settling two smaller claims.

B. Bifurcating Independent Claims

1. Bifurcated vs. Aggregated Settlements

A simple numerical analysis makes this intuitive explanation

more precise, providing a foundation for a later, more elaborate simu-

lation analysis. The law-and-economics literature assessing the effects

of different rules on the likelihood of settlement and trial relies largely

on formal mathematical modeling,

but our approach is consistent

with that of other works using mathematical analysis augmented by

simple examples

or simulations.

We use a simple numerical ap-

See, e.g., Lucian Arye Bebchuk, Litigation and Settlement Under Imperfect Information,

15 RAND J. E

CON

. 404 (1984) (analyzing the effect of one-sided information); A. Mitchell Polin-

sky & Daniel L. Rubinfeld, Does the English Rule Discourage Low-Probability-of-Prevailing

Plaintiffs?, 27 J. L

EGAL

TUD

. 519 (1998) (analyzing the effect of fee shifting).

See, e.g., Steven Shavell, Suit, Settlement, and Trial: A Theoretical Analysis Under Alter-

native Methods for the Allocation of Legal Costs, 11 J. L

EGAL

TUD

. 55 (1982) (offering an

analysis of different fee shifting rules).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 11 18-JUN-18 16:24

386 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

proach here because modeling two-sided imperfect information ana-

lytically is challenging,

even when parties are seeking to resolve only

a single claim. Perhaps in part because of the challenges of mathemat-

ical modeling, the law-and-economics literature has generally focused

on situations involving only a single claim, and even there has gener-

ally assumed either perfect information or one-sided information (i.e.,

one party knows the result with certainty, but the other party knows

only the distribution of possible results).

We must analyze continu-

ous two-sided information for each of two claims.

Ignore for the moment whether a party has an incentive to give

up. Rather, simply assume, as is common in the literature,

that a

claim will settle so long as there is a settlement range, that is, so long

as the plaintiff’s valuation exceeds the defendant’s by no more than

the sum of the expected costs that settlement would save. We imple-

mented the model under this assumption through Monte Carlo tech-

niques,

simulating a settlement negotiation 1,000,000 times for each

scenario. Each iteration of the simulation works as follows:

• For each of Claim 1 or Claim 2, a claim value is drawn at ran-

dom from a uniform distribution between $0 and $1000.

• Then, each party receives a separate signal of claim quality

that is equal to the actual claim value plus a random noise

value, drawn from a normal distribution with a standard

deviation of $100.

• Each party then makes a valid Bayesian inference about the

expected claim value based on this inference. If Claims 1 and

2 are aggregated, each party sums its valuations of the two

claims.

See, e.g., George L. Priest & Benjamin Klein, The Selection of Disputes for Litigation, 13

J. L

EGAL

TUD

. 1, 22–28 (1984) (providing simulations to bolster an analytical model).

See, e.g., Daniel Friedman & Donald Wittman, Litigation with Symmetric Bargaining

and Two-Sided Incomplete Information, 23 J.L. E

CON

. & O

. 98, 99 (2007) (offering a complex

mathematical model).

See, e.g., A. Mitchell Polinsky & Daniel L. Rubinfeld, Optimal Awards and Penalties

When the Probability of Prevailing Varies Among Plaintiffs, 27 RAND J. E

CON

. 269, 270–71

(1996).

See, e.g., Steven Shavell, Alternative Dispute Resolution: An Economic Analysis, 24 J.

EGAL

TUD

. 1, 11 (1995) (“[T]he difference between the plaintiff’s expected judgment and the

defendant’s expected judgment must exceed the sum of their trial costs for there to be a trial;

otherwise they will settle to save trial costs.”).

See generally RiskAMP, What Is Monte Carlo Simulation?, https://www.riskamp.com/

files/RiskAMP%20-%20Monte%20Carlo%20Simulation.pdf [https://perma.cc/C8Z3-KAL4].

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 12 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 387

• A negotiation is assumed to fail when the plaintiff’s valuation

exceeds the defendant’s by more than a posited settlement

range.

For each value of litigation costs, there were three distinct scenarios:

Claims 1 and 2 aggregated, Claim 2 alone, and Claim 1 alone (dis-

counting the settlement range by the probability that Claim 2 will be

negotiated successfully).

Figure 1 summarizes the results. The x-axis represents the size of

the settlement range. This is the combined expenditure on litigation

costs that settlement will avoid. For each settlement range, the

leftmost two bars represent, respectively, the probability of settlement

failure of Claim 2 on the assumption that Claim 1 has succeeded, and

the probability of settlement failure on Claim 1 without knowing the

result of Claim 2. As explained above,

the probability of settlement

failure is higher for Claim 1 than for Claim 2 because the possibility

that Claim 2 might fail to settle reduces the benefit of settling Claim 1.

Meanwhile, the third bar shows the probability of trial with bifurca-

tion (that is, the probability either that Claim 1 will not be settled or

that, after Claim 1 is resolved, Claim 2 will fail to settle). Finally, the

fourth bar shows trial rates when Claims 1 and 2 are aggregated

together.

The data suggest that any increase in the likelihood of trial attrib-

utable to bifurcation is not likely to be severe—and surprisingly, that

bifurcation might produce more settlements than aggregated settle-

ments. When the settlement range is only $50—that is, the expected

cost of litigation is only one-tenth of the average case value—then the

trial rate with bifurcated settlements is considerably higher than when

claims are aggregated: 57.4% vs. 37.9%. This gap narrows as the set-

tlement range becomes greater. Once the settlement range is above

$200, bifurcated settlements produce a lower trial rate (10.5%) than

aggregated settlements (11.6%). The gap becomes more pronounced

as the settlement range continues to rise. The reason for this is that

the rate of settlement failure for each claim individually—the two bars

on the left of each group—becomes much lower than the rate of set-

tlement failure for the two claims when aggregated. It is fairly unlikely

that mutual optimism will block settlement when $1000 is at stake and

trial costs are $300, but it is extremely unlikely when only $500 is at

stake and trial costs are still $300.

See supra Section I.A.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 13 18-JUN-18 16:24

388 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

IGURE

1. N

EGOTIATION

AILURE

ATES FOR

EPARATE

LAIMS

WITH

IFURCATED AND

GGREGATED

ETTLEMENTS

10%

20%

30%

40%

50%

60%

70%

$50 $100 $150 $200 $250 $300

Negotiation Failure Rate

Settlement Range

Settlement failure (Claim 2

alone, assuming Claim 1

success)

Settlement failure (Claim 1

alone)

Trial rate (bifurcated, Claim

1 then Claim 2)

Trial rate (aggregated)

This analysis does not justify a mandatory regime of bifurcation

in the absence of some other consideration.

After all, there is noth-

ing to stop parties from negotiating Claim 1 before Claim 2 even if the

law does not require them to negotiate.

In practice, however, parties

rarely reach partial settlements and continue negotiating, so creating a

mandatory regime might be helpful. In any event, the analysis does

provide some evidence that if there is an affirmative reason to insist

on bifurcation, the costs of doing this may be quite low or nonexistent.

A legislature should, however, tread carefully before adopting

any bifurcation regime based on this analysis, ideally implementing

bifurcation as a pilot project subject to rigorous examination.

There

are three reasons that caution is appropriate. First, the model does not

take into account the possibility of deviations from rational behavior.

Second, the model is quite stylized: it does not take into account struc-

tural factors, such as the plaintiff’s option to drop a claim.

Third, the

model focuses on asymmetric information as the source of bargaining

failure, but strategic bargaining may also lead to bargaining failure.

Cases in which both parties are aggressive bargainers for reasons

other than their valuations are less likely to settle. Importantly, bar-

See, e.g., infra Parts II, III.

See 15A C.J.S., supra note 1. R

Cf. Laurens Walker, Perfecting Federal Civil Rules: A Proposal for Restricted Field Ex-

periments, 51 L

& C

ONTEMP

. P

ROBS

. 67, 67 (1988) (urging greater experimentation in civil

procedure).

See Peter H. Huang, Lawsuit Abandonment Options in Possibly Frivolous Litigation

Games, 23 R

. L

ITIG

. 47, 50 (2004) (exploring the implications of option theory for modeling

litigation).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 14 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 389

gaining aggressiveness seems likely to be correlated across different

claims. This consideration, however, reduces the risks of bifurcation. If

settlement failure for Claims 1 and 2 are correlated, then the fact that

Claim 1 settles suggests that the parties are not being especially ag-

gressive, meaning that Claim 2 is highly likely to settle as well.

2. Severable Settlements

An alternative to seriatim bifurcation is to use severable settle-

ments. This approach would permit an initial comprehensive settle-

ment but allow either party to back out of a portion of it. The

provisions of the settlement agreement would thus be “severable,”

much like statutory provisions that stand even if courts void others on

judicial review.

Withdrawal from agreement on an issue would re-

quire further negotiation or trial. A party might be required to notify

a court or a third-party information escrow agent

if it wished to back

out of part of an agreement. Critically, the court or agent would in-

form the other party only after the withdrawal deadline, so a party

backing out of one part of an agreement would not worry that doing

so would trigger the other party to back out of the other part. An

advantage of severable settlements relative to seriatim bifurcation is

that they require greater trust among the parties to circumvent. Seria-

tim bifurcation will fail so long as a party making a concession on

Claim 1 trusts its opponent, because once a concession is made on

Claim 1, the concession is irreversible; severable settlements will fail

only if each party trusts its opponent, because either party may defect

from cooperation at any time.

We can model severable settlements by imagining separate law-

yers negotiating Claim 1 and Claim 2, each recognizing that the other

negotiation may fail. The benefits of a successful negotiation are thus

discounted in each case by the probability of a successful negotiation

in the other case. If there is only an 80% chance that the other case

will settle, then the settlement range is 80% as large. Because Claim 1

and Claim 2 are drawn from the same distribution, we must find a

value p so that when one other claim will settle with probability p, the

claim being negotiated will also settle with probability p. By zeroing in

on this equilibrium, we can determine the probability value with sev-

See generally John Copeland Nagle, Severability, 72 N.C. L. R

. 203 (1993) (discussing

judicial doctrine concerning severability provisions).

See Ian Ayres & Cait Unkovic, Information Escrows, 111 M

ICH

. L. R

. 145, 150 (2012)

(defining an “information escrow” as “a mechanism of conditional, intermediated

communication”).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 15 18-JUN-18 16:24

390 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

erable settlements. That is, for a particular settlement-range value pa-

rameter and a particular assumption about the settlement rate for the

other claim, the simulation iterates 1,000,000 times to determine the

settlement rate for this claim; it then adjusts p and repeats, until the

two values are arbitrarily close to one another. Figure 2 illustrates the

result, alongside the trial rates for bifurcation and for aggregated

claims.

IGURE

2. N

EGOTIATION

AILURE

ATES FOR

EVERABLE

ERIATIM

AND

GGREGATED

ETTLEMENTS

10%

20%

30%

40%

50%

60%

70%

$50 $100 $150 $200 $250 $300

Negotiation Failure Rate

Settlement Range

Trial rate (severable)

Trial rate (bifurcated, Claim

1 then Claim 2)

Trial rate (aggregated)

Trial rates with severable settlements are slightly above the trial

rate with seriatim bifurcation. With bifurcation, Claim 1 is more diffi-

cult to negotiate because of the possibility that the parties may fail to

resolve Claim 2. With severable claims, each claim is more difficult as

a result of the possibility that the other may not be resolved. A severa-

ble claim is more difficult to resolve than Claim 2 would be because

the expected costs savings must be multiplied by the probability of

negotiation success. Moreover, a severable claim will be more difficult

to resolve than Claim 1 would be because when negotiating Claim 1,

the expected costs savings are multiplied by the probability of negotia-

tion success with Claim 2, which is the easiest of the claims to resolve.

Thus, negotiating two severable claims will be more difficult than ne-

gotiating two bifurcated claims seriatim—but not much more difficult.

As illustrated in Figure 2, the negotiation failure rates are only slightly

less, especially with larger settlement ranges. As before, once the set-

tlement range becomes sufficiently high (around $200), severable set-

tlements produce lower trial rates than aggregated settlements.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 16 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 391

The comparison between bifurcated and severable settlements

provides some argument in favor of the latter, but psychology may be

as important as economics in choosing between these regimes. A

downside of severable settlements is the risk that each party may

worry, sometimes with good reason, that the other party is agreeing to

settle one claim (or one aspect of a claim) as a gesture that the other

party will then repudiate. This is, of course, the point—the goal is for

each party to negotiate each part of the settlement with the expecta-

tion that the other might still be negotiated—but it could increase

feelings of mistrust between the parties. On the other hand, perhaps

some deception about the acceptability of one part of a settlement will

facilitate settlement of another part of a claim. In addition, parties

might feel more comfortable agreeing to resolution of one part of a

settlement if they have what they believe is at least a tentative agree-

ment on another. Seriatim bifurcation imposes a structure to the set-

tlement process, thus largely removing parties’ need to determine

whether the other party is only pretending to agree to a settlement.

On the other hand, addressing one issue before another may be artifi-

cial, preventing the parties from considering the issues in the order

that may seem most natural given the particular facts.

II. P

ROTECTING

HILDREN IN

IVORCE

ASES

Scholars have long held up custody settlements as a paradigmatic

instance of negotiations vulnerable to improper trade-offs.

While the

empirical frequency of trade-offs is debated,

many scholars contend

that the parent more invested in custody will often trade away marital

property, spousal support, or child support for a greater share of cus-

See Robert H. Mnookin & Lewis Kornhauser, Bargaining in the Shadow of the Law:

The Case of Divorce, 88 Y

ALE

L.J. 950, 964–65, 979–80 (1979) (observing that “a parent may,

over some range, trade custodial rights for money,” and that the uncertainty of prevailing child

custody standards has the “ironic and tragic result” of disadvantaging the parent more invested

in custody); see also, e.g., Richard Neely, The Primary Caretaker Parent Rule: Child Custody and

the Dynamics of Greed, 3 Y

ALE

L. & P

’

. 168, 179 (1984) (“The everyday occurrence of

children being traded for money should be sufficient in and of itself to prompt a reevaluation of

a system that turns custody awards into bargaining chips.”); Elizabeth S. Scott, Pluralism, Paren-

tal Preference, and Child Custody, 80 C

ALIF

. L. R

. 615, 651 (1992); Jana B. Singer, The Priva-

tization of Family Law, 1992 W

. L. R

. 1443, 1550.

See, e.g., E

LEANOR

E. M

ACCOBY ET AL

., D

IVIDING THE

HILD

: S

OCIAL AND

EGAL

ILEMMAS OF

USTODY

160 (1992) (“[C]ontrary to popular perception, most divorce decrees do

not reflect a trade-off between custody and money issues.”); Margaret F. Brinig & Michael V.

Alexeev, Trading at Divorce: Preferences, Legal Rules and Transactions Costs, 8 O

HIO

. J.

ISP

. R

ESOL

. 279, 292 (1993) (“[C]ontrary to the accepted wisdom, tradeoffs between custody

and property may not always be present.”).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 17 18-JUN-18 16:24

392 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

tody.

Some claim that the less-invested parent may threaten to initi-

ate custody litigation to pressure the other parent to make such trade-

offs.

We start with custody settlements as our initial case study of

bifurcation not because this is the strongest or most straightforward

possible application. To the contrary, the complexity of divorce nego-

tiations means that settlement bifurcation in the context of child cus-

tody will face obstacles that may not exist for some other applications

of bifurcation. Our aim is to explore settlement bifurcation, not to

advocate it as a solution to every problem, and so we start with an

application in which bifurcation’s benefits and costs may both be high.

We will accordingly devote more attention to this application of bifur-

cation than to others, even though the case for some of the other ap-

plications may be stronger.

We are not the first to recognize that the legal system could re-

quire resolution of child custody before other issues. Prior commenta-

tors, however, have been too dismissive either of bifurcation or of its

downsides, and none has devoted more than brief attention to bifurca-

tion of custody settlements or to the broader question of when bifur-

cation might be justified.

The mediation literature mentions the

possibility of requiring mediators to address custody separately from

property and support.

This literature, however, largely rejects such

proposals on the assumption that custody trade-offs are inevitable and

even desirable, insofar as they increase client satisfaction with the

work of private mediators,

and it does not recognize the possibility

See, e.g., Altman, supra note 8, at 499, 501 (reporting result of survey of California R

attorneys “confirm[ing] that negotiating tactics aimed at trading custodial time for financial

terms are widespread though hardly universal”); Margaret F. Brinig, Penalty Defaults in Family

Law: The Case of Child Custody, 33 F

. S

. U. L. R

. 779, 806–07 (2006) (finding some

evidence of trade-offs); Martha Fineman, Dominant Discourse, Professional Language, and Le-

gal Change in Child Custody Decisionmaking, 101 H

ARV

. L. R

. 727, 761 (1988); Neely, supra

note 43, at 179; Scott, supra note 43, at 647, 651; Singer, supra note 43, at 1550. R

See Altman, supra note 8, at 495–510; Howard S. Erlanger et al., Participation and Flexi- R

bility in Informal Processes: Cautions from the Divorce Context, 21 L

& S

’

. 585, 597

(1987) (“[A] number of women report that they accepted poor settlement terms because their

husbands were threatening custody battles . . . .”); Neely, supra note 43, at 177–79; Scott, supra R

note 43, at 647. R

See, e.g., Craig A. McEwen et al., Bring in the Lawyers: Challenging the Dominant Ap-

proaches to Ensuring Fairness in Divorce Mediation, 79 M

INN

. L. R

. 1317, 1340–42 (1995)

(discussing attempts to require divorce mediators to resolve economic issues separately from

custody).

See id. at 1341.

See id. at 1340–42 (finding that “statutes that purport to separate economic from cus-

tody/visitation issues probably do not succeed in actually severing these issues or in preventing

parties from linking them” during mediation, and citing studies finding that costs are lower and

client satisfaction higher when mediators address custody together with financial issues).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 18 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 393

that the courts might enforce bifurcation by approving parts of settle-

ments seriatim. One commentator, Scott Altman, does recognize this

possibility, briefly considering bifurcation at the end of a study report-

ing the results of a survey of California lawyers about their experience

with custody negotiations.

Altman suggests that “[s]ettlement agree-

ments should be submitted in stages.”

Altman, however, dismisses

without elaboration some of the most significant potential downsides

to the proposal, including the possibility that settling custody sepa-

rately might increase litigation rates,

or that it might adversely affect

negotiated custody outcomes.

No jurisdiction appears to have

adopted Altman’s proposal.

A. The Problem of Custody Trade-Offs

Concern about potential trade-offs between divorcing parents

was brought to the fore in the seminal 1979 article written by Robert

Mnookin and Lewis Kornhauser on the role of background legal rules

in shaping settlement negotiations.

Divorcing parents negotiating

custody over their children must also resolve issues of marital prop-

erty division, spousal support, and child support. Mnookin and Korn-

hauser observed that the parent more invested in custody, even if

more likely to be awarded custody at trial, might relinquish rights to

property or support rather than risk even a small possibility of “sub-

stantially diminish[ing] his or her relationship with the child.”

The

less-invested parent, by contrast, has nothing to lose from threatening

to bring an action for primary custody.

According to Mnookin and

Kornhauser, this dynamic, in conjunction with the indeterminate best-

interests standard governing child-custody disputes, has the “ironic

and tragic result” of weakening the bargaining position of the “good

See Altman, supra note 8, at 527. R

Id.

Altman acknowledges that settling custody separately might incentivize litigation. See

id. at 529. But he focuses only on the scenario of a parent who litigates custody to obtain a

greater share than he desires, with a view to trading it off for financial concessions—a prospect

he rejects as unlikely to occur with any frequency because the costs of litigation will typically

outweigh financial gains. See id. Altman does not explore the possibility that bifurcating settle-

ments might be useful even when both parents genuinely want custody.

See id. (failing to address this possibility).

See Mnookin & Kornhauser, supra note 43. R

Id. at 979; see also Neely, supra note 43, at 177 (contending that “husbands will threaten R

custody fights . . . as a means of intimidating wives into accepting less child support and ali-

mony,” and that “[b]ecause women are usually unwilling to accept even a minor risk of losing

custody, such techniques are generally successful”).

See Mnookin & Kornhauser, supra note 43, at 964–65, 969–71, 978–79. R

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 19 18-JUN-18 16:24

394 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

parent” in divorce negotiations.

While courts deciding custody mat-

ters often have the option of appointing a guardian ad litem to re-

present children’s interests,

when parents settle the matter of child

custody, courts rarely second-guess the parents’ arrangement.

1. Custody Trade-Offs and Harm to Children

By reducing the share of marital wealth available to custodial

parents, custody trade-offs diminish the material well-being of the

children who reside with those parents, thus potentially contributing

to the high rates of postdivorce impoverishment of caretakers and

their children.

The primary harm to children that commentators

have identified as arising from custody trade-offs is financial.

Chil-

dren are affected not only by child-support awards (which are paid

directly to the custodial parent), but also by both the division of their

parents’ marital property and the presence or absence of an award of

spousal support. When a primary caretaker, fearful of losing custody,

makes a financial sacrifice in exchange for a desirable custody ar-

rangement, the caretaker’s children suffer from a lower standard of

living.

Children can also be harmed if the primary custodial parent

must work more as a result of this trade-off and thus spend less time

with them day to day. While spousal support is awarded only in a

Id. at 979; see also Fineman, supra note 45, at 761 (“[M]any women bargain away R

needed property and support benefits to avoid the risk of ‘losing’ their children.”).

See, e.g., C

ONN

. G

. S

TAT

. A

. § 46b-54(a) (West 2017); M

ICH

. C

OMP

. L

AW S

§ 722.24(4)(2) (2017) (“If, at any time in [a] proceeding [involving dispute of a minor child’s

custody], the court determines that the child’s best interests are inadequately represented, the

court may appoint a lawyer-guardian ad litem to represent the child.”); see also Barbara Ann

Atwood, Representing Children: The Ongoing Search for Clear and Workable Standards, 19 J.

. A

CAD

. M

ATRIM

. L

. 183, 192 (2005) (noting that most states allow courts to appoint a

guardian ad litem in custody disputes).

See Brian H. Bix, Private Ordering and Family Law, 23 J. A

. A

CAD

. M

ATRIM

. L

249, 262 (2010) (“[C]ourts tend to rubber-stamp [separation] agreements . . . even for child-

related provisions.”).

Scott, supra note 43, at 651 (“When parents are motivated to exchange custody rights R

for property . . . , such an exchange will result in an unequal property distribution—an outcome

that usually threatens the child’s future economic welfare.”); see also Singer, supra note 43, at R

1550 (“Substantial evidence suggests that the common divorce bargaining practice of a parent

trading off financial claims for custody assurances has contributed both to inadequate child sup-

port agreements and to the impoverishment of children and their custodial parents after

divorce.”).

See, e.g., Scott, supra note 43, at 651 (emphasizing financial harm to children as a result R

of custody trade-offs); Singer, supra note 43, at 1550 (arguing that custody trade-offs contribute R

to child poverty); see also Altman, supra note 8, at 512–13 (same). R

See, e.g., Neely, supra note 43, at 178 (arguing that because custody trade-offs reduce R

both the spousal support and the child support paid to custodial parents, “children are forced to

grow up poor, or at least poorer than they should be”).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 20 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 395

small fraction of cases, one factor that courts consider in awarding

such support is whether a caretaker should be awarded an amount of

support that would enable her to continue staying home with young

children.

Where a parent forgoes an award of spousal support in a

custody settlement, this may mean that she will return to work sooner

than otherwise.

Financial trade-offs also could lead to suboptimal custody alloca-

tion, for instance, by detrimentally reducing a child’s contact with a

noncustodial parent.

They could shape other aspects of custodial ar-

rangements as well, potentially in ways adverse to children’s interests.

Custody settlements commonly address such issues as whether the

custodial parent may relocate with the children, how the children shall

be raised, and who has the right to make educational, medical, and

religious decisions about the children’s upbringing.

Given the notori-

ous difficulty of assessing which custodial arrangement is in a child’s

best interests, it is hard to measure to what extent these custody ar-

rangements are better or worse than they would be absent trade-offs.

Given the assumption that custody should be determined primarily

with reference to children’s interests, however, a custody settlement

that is shaped in part by trade-offs might be considered less optimal.

Trade-offs between custody and financial matters also could in-

flict expressive harm on children. Margaret Radin argues that the ex-

change of money for parental rights commodifies children and

demeans children’s dignity.

While Radin made this point in the con-

text of surrogacy agreements,

similar arguments may apply to ex-

changing money for custody.

Indeed, custody law since the early

See, e.g., C

. F

. C

ODE

§ 4320(g) (West Supp. 2017) (instructing courts ordering

spousal support to consider, inter alia, “[t]he ability of the supported party to engage in gainful

employment without unduly interfering with the interests of dependent children in the custody

of the party”).

See Altman, supra note 8, at 512. R

See Sarah Abramowicz, Contractualizing Custody, 83 F

ORDHAM

L. R

. 67, 82–83,

86–88 (2014) (describing provisions that have been included in custody agreements). While

courts often defer to such agreements, they are not bound by them and are particularly reluctant

to enforce provisions governing children’s religious education, which raise First Amendment

concerns. See id. at 79–88.

See Margaret Jane Radin, Market-Inalienability, 100 H

ARV

. L. R

. 1849, 1909–11,

1925–28 (1987).

See id.

See, e.g., Katharine B. Silbaugh, Marriage Contracts and the Family Economy, 93 N

U. L. R

. 65, 126–28 (1998) (noting the widespread view that “commodifying the child . . . [is]

unacceptable”). But see, e.g., Martha M. Ertman, What’s Wrong with a Parenthood Market? A

New and Improved Theory of Commodification, 82 N.C. L. R

. 1, 3–5 (2003) (contending that

there are positive aspects to parenthood markets, including enabling family formation on the

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 21 18-JUN-18 16:24

396 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

nineteenth century has insisted that parental agreements exchanging

custody for financial concessions (such as agreements in which a fa-

ther forfeits his right to visitation in exchange for release from his

child-support obligations)

are unenforceable because children are

not “chattel” to be bought and sold by their parents.

The law’s ban

on explicit trade-offs suggests a perception of harm that would inhere

in implicit trade-offs too.

2. Custody Trade-Offs and Harm to the State

The state also has an interest in preventing parental trade-offs.

As parens patriae,

the state has a responsibility to protect children.

Moreover, given the connection between the conditions of children’s

upbringing and children’s capabilities as adults, the state has an inde-

pendent interest in ensuring that children are raised in conditions that

will render them productive, law-abiding adult citizens.

Additionally,

family law scholars often characterize family law (somewhat critically)

as motivated by a state interest in minimizing the likelihood that par-

ents and their children will require public assistance.

While the judi-

ciary in many states can reject marital property and support

agreements that will render a former spouse eligible for public sup-

basis of “intention and function”); Elisabeth M. Landes & Richard A. Posner, The Economics of

the Baby Shortage, 7 J. L

EGAL

TUD

. 323, 346 (1978) (suggesting that “the benefits of free baby

selling might . . . outweigh the costs”).

See, e.g., Blisset v. Blisset, 526 N.E.2d 125, 128 (Ill. 1988) (refusing to enforce such an

agreement on the basis that “[p]arents may not bargain away their children’s interests”). But see

Ferguson v. McKiernan, 940 A.2d 1236, 1238 (Pa. 2007) (upholding agreement terminating pa-

rental rights and obligations of known sperm donor).

See, e.g., Chapsky v. Wood, 26 Kan. 650, 652 (1881) (“[A] child is not in any sense like a

horse or any other chattel, subject-matter for absolute and irrevocable gift or contract.”); People

ex rel. Barry v. Mercein, 3 Hill 399, 410 (N.Y. Sup. Ct. 1842) (rejecting right of father to “violate

his duty by selling his children”). Courts used the rhetoric of “chattel” to justify refusing to

enforce any contractual transfers of parental rights, including those that did not involve mone-

tary exchange. See, e.g., Chapsky, 26 Kan. at 657–58 (refusing to enforce uncompensated transfer

of parental rights to adoptive parent); Mercein, 3 Hill at 408 (refusing to enforce separation

agreement allocating custody to wife).

See Naomi Cahn, State Representation of Children’s Interests, 40 F

. L.Q. 109, 112

(2006) (characterizing the parens patriae doctrine as “the basis for any state intervention on

behalf of children in the family based on a belief that the state has a duty to act in the best

interest of the child”).

See Anne C. Dailey, Developing Citizens, 91 I

OWA

L. R

. 431, 432–34 (2006) (demon-

strating how the conditions in which children develop shape their ability to function as rational

and deliberative adult citizens).

See, e.g., Melissa Murray, Family Law’s Doctrines, 163 U. P

. L. R

. 1985, 2013 (2015)

(identifying “privatizing dependency” as a central commitment of family law).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 22 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 397

port,

there are less obvious ways in which a financial settlement be-

tween divorcing parents may impose costs on the state.

3. Custody Trade-Offs and Improper Leverage

Much of the literature on custody trade-offs expresses concern

with the harm these trade-offs inflict, not only on children, but also on

the parent who relinquishes financial resources in exchange for cus-

tody.

Commentators have long noted that the parent less invested in

custody will often make a disingenuous threat to litigate custody—

that is, a threat to request a custodial allocation that the parent

neither desires nor expects to receive—in order to extract financial

concessions from a parent fearful of even a small risk of losing pri-

mary custody.

Scholars and judges have characterized such tactics as

“extortionate bargaining”

and “blackmail.”

It may be tempting to frame this concern with harm to the parent

as paternalistic. After all, both parents must freely consent to any set-

tlement, and, if a parent willingly trades money or property for cus-

tody, why not respect her own preferences on the matter? The

scholarly tendency is to characterize the disingenuous threat to litigate

custody as coercive,

which, by implicitly characterizing the

threatened parent as vulnerable and thus in need of special protec-

tions, raises the specter of paternalism. But a disingenuous threat to

litigate custody is problematic for reasons unrelated to the power dy-

namic between divorcing parents. Where a parent threatens to litigate

See, e.g., U

NIF

. P

REMARITAL

GREEMENT

§ (6)(b) (N

’

ONFERENCE OF

OMM

’

RS ON

NIF

. S

TATE

AW S

1983).

For instance, the state may ultimately pay more in medical costs if a former spouse is

rendered unable to afford preventive medical care. See J. Thomas Oldham, Changes in the Eco-

nomic Consequences of Divorces, 1958–2008, 42 F

. L.Q. 419, 446 (2008) (discussing how di-

vorce increases eligibility for Social Security and Medicare costs).

See, e.g., Fineman, supra note 45, at 760–61 (contending that prevailing custody alloca- R

tion process “produces bad decisions for many women and children”); Singer, supra note 43, at R

1549 (“[L]egal rules that grant unfettered discretion to private individuals to structure the pro-

cess of marital dissolution or that place dissolution-related disputes outside the ‘shadow of the

law’ may end up empowering economically stronger family members at the cost of economically

weaker ones . . . .”).

See, e.g., Altman, supra note 8, at 494–510; Neely, supra note 43, at 177–79; Scott, supra R

note 43, at 647. R

See, e.g., Neely, supra note 43, at 171 (characterizing custody litigation threats as “extor- R

tionate bargaining”).

See, e.g., Jana B. Singer & William L. Reynolds, A Dissent on Joint Custody, 47 M

. L.

. 497, 516 (1988) (characterizing disingenuous threats to request joint custody as “custody

blackmail”).

See, e.g., Altman, supra note 8, at 513–14 (evaluating whether it is coercive to threaten R

custody litigation in order to extract a better financial settlement).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 23 18-JUN-18 16:24

398 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

custody solely in order to extract a more advantageous financial set-

tlement, such a threat is a form of nuisance claim—a threat to bring a

claim that has only a low probability of success, solely for the purpose

of extracting a settlement. The legal system seeks to prevent nuisance

claims, recognizing that the use of such claims to extract settlements is

improper and adverse to social welfare.

Where litigants append nui-

sance claims to legitimate ones, as a parent does by making a disingen-

uous threat to litigate custody in order to extract a desirable

settlement on property allocation and support, the leverage that this

creates similarly exerts improper pressure and skews settlement

results.

B. Previous Approaches to Custody Trade-Offs

Although family law scholars have recognized the potential dan-

gers of trade-offs in custody negotiations, aside from Altman’s brief

treatment,

neither scholars nor courts have considered settlement bi-

furcation. Instead, they have considered a variety of other approaches

to reducing trade-offs. Similar strategies might be applied beyond

family law in other contexts where trade-offs affect third parties, and

examining them highlights the comparative simplicity of settlement

bifurcation.

1. Changes to the Substantive Law

The tendency to look to background legal rules to address the

problem of custody trade-offs originated with Mnookin and Korn-

hauser.

They argued that the indeterminate “best interests of the

child standard” that currently prevails in family law makes child-cus-

tody outcomes especially difficult to predict.

This uncertainty, they

argued, contributed to the risk aversion

of “good parents” so fearful

of even the smallest chance of losing custody that they might trade

away property and support to guarantee a desirable custody out-

come.

Mnookin and Kornhauser hypothesized that making custody

See Randy J. Kozel & David Rosenberg, Solving the Nuisance-Value Settlement Prob-

lem: Mandatory Summary Judgment, 90 V

. L. R

. 1849, 1851–52 (2004) (“The resulting settle-

ments decrease social welfare by vexing and taxing the victimized party, encouraging the

misallocation of legal resources, and diminishing public confidence in the civil liability system.”).

See Altman, supra note 8, at 527. R

See Mnookin & Kornhauser, supra note 43, at 977–80. R

Id. (emphasis omitted).

See id. at 979 (characterizing the parent who significantly prefers the certainty of a mod-

erate amount of custody to a risk of losing substantial custody as “risk-averse”).

See id.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 24 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 399

rules more determinate would reduce strategic bargaining and thereby

help to prevent trade-offs.

They also noted that rules that are both

more determinate and more favorable to the primary caretaker parent

(such as a maternal preference, a version of which preceded the more

open-ended best-interests standard) would have the additional effect

of strengthening the bargaining position of the more invested parent.

In subsequent decades, both lawmakers and courts, as well as

scholars, have continued to consider how default rules might change

to prevent custody trade-offs that work to children’s disadvantage. In

1981, in Garska v. McCoy,

Justice Robert Neely of the West Virginia

Supreme Court crafted a new child custody standard intended to ad-

dress this problem. “Our experience,” Justice Neely explained, “in-

structs us that uncertainty about the outcome of custody disputes

leads to the irresistible temptation to trade the custody of the child in

return for lower alimony and child support payments.”

Garska thus

replaced the best-interests-of-the-child standard with the more deter-

minate “primary caretaker” presumption, according to which, in cus-

tody disputes involving children of “tender years,”

it is ordinarily in

a child’s best interests to have custody awarded to the parent who has

spent the most time caring day to day for that child.

Only three states adopted the primary caretaker presumption,

and all three eventually replaced it with other approaches.

One criti-

cism was that, while formally gender neutral, the primary caretaker

presumption unfairly favored mothers.

A related reason for the fail-

See id. 979–80. Scholars have also suggested preventing custody trade-offs by making

other subjects of divorce bargaining, such as marital property division and spousal support, more

determinate. See, e.g., Herma Hill Kay, No-Fault Divorce and Child Custody: Chilling Out the

Gender Wars, 36 F

. L.Q. 27, 42 (2002) (arguing that by making spousal support less discre-

tionary, the rules proposed in the American Law Institute’s 2002 Principles of the Law of Family

Dissolution “lessen[ ] the incentive to use children as a bargaining chip”).

Mnookin & Kornhauser, supra note 43, at 977–78. R

278 S.E.2d 357 (W. Va. 1981).

Id. at 360.

Id. at 363.

See id. at 360, 362–63. Garska thus revived, in gender-neutral terms, the presumption of

maternal custody for children of tender years that West Virginia courts had applied prior to state

legislation mandating that custody be awarded on the basis of “the best interest of the children

based upon the merits of each case” rather than on a presumption that “either the father or the

mother should be awarded custody.” Id. at 360.

See Katharine T. Bartlett, U.S. Custody Law and Trends in the Context of the ALI Prin-

ciples of the Law of Family Dissolution, 10 V

. J. S

. P

’

& L. 5, 17 (2002); Barbara Bennett

Woodhouse, Child Custody in the Age of Children’s Rights: The Search for a Just and Workable

Standard, 33 F

. L.Q. 815, 822–23 (1999).

See Woodhouse, supra note 93, at 823. R

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 25 18-JUN-18 16:24

400 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

ure of the primary caretaker presumption was the increasing popular-

ity of the view, promoted initially by fathers’-rights advocates and

later by some child-rearing experts as well, that children are better

served by a close and continuing relationship with both parents than

by the sole custody/visitation paradigm under which children spend

the vast majority of time with one custodial parent.

While the best-

interests assessment still typically determines how custody is allocated

between the parents, some states have created a presumption of joint

custody, and many others have revised their custody statutes to pro-

vide that courts assessing best interests should consider both the harm

to children from disruption of the bond with a primary caretaker, and

the benefit to children of maintaining a close and continuing relation-

ship with both parents.

Preventing trade-offs between custody and money was also a mo-

tivating factor behind a more recently proposed replacement of the

best-interests standard: the ALI approximation rule, according to

which custody is awarded in proportion to the amount of time each

parent spent caring for the child during the intact relationship.

promoting the approximation rule, Elizabeth Scott cited the literature

arguing that the indeterminacy of the best-interests standard encour-

ages custody-financial trade-offs.

The hope was that the approxima-

tion rule would be so determinate, and align so closely with the

parties’ preferences, that parents would settle rather than litigate cus-

tody. The only state to adopt the approximation rule, however, has

been West Virginia, the same state that initially adopted the primary

caretaker presumption.

Presumably, proposals for more determinate substantive law

have failed to gain traction in part because of a perception that some-

thing would be lost if courts were constrained not to consider other

factors. The debate is thus an incarnation of the more general juris-

prudential choice between rules and standards. Rules are necessarily

overinclusive and underinclusive, and this must be balanced against

the benefits of greater determinism. Any approach to increasing pre-

See, e.g., David L. Chambers, Rethinking the Substantive Rules for Custody Disputes in

Divorce, 83 M

ICH

. L. R

. 477, 550–57 (1984).

See, e.g., F

. S

TAT

. A

. § 61.13(2) (West Supp. 2017).

See P

RINCIPLES OF THE

AW O F

AMILY

ISSOLUTION

: A

NALYSIS AND

ECOMMENDA-

TIONS

§ 2.08 (A

. L

NST

. 2002).

See Scott, supra note 43, at 643–56. R

W. V

. C

ODE

. § 48-9-206(a) (LexisNexis 2015); see Mary Jean Dolan & Daniel J.

Hynan, Fighting over Bedtime Stories: An Empirical Study of the Risks of Valuing Quantity over

Quality in Child Custody Decisions, 38 L

& P

SYCHOL

. R

. 45, 56 (2014).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 26 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 401

dictability by making the substantive law more rule-like invites the

objection that relevant circumstances are ignored. Settlement bifurca-

tion is compatible with either rules or standards and is thus a procedu-

ral alternative to deviating from what otherwise might be the optimal

substantive-law approach.

2. Judicial Review

Another procedural mechanism for preventing custody trade-offs

is judicial review. With respect to ensuring that trade-offs do not pro-

duce a custody arrangement adverse to children’s interests, such a

mechanism is already, nominally, in place: in most U.S. jurisdictions,

parental custody settlements can only be enforced, or incorporated

into a court order or judicial decree, if a judge determines the custody

arrangement to be in the best interests of the child or children.

100

It is

rare, however, for courts to reject parental custody agreements in the

name of children’s interests.

101

Moreover, it is difficult for courts to

discern when custody has been traded for money, and agreements re-

garding marital property division and spousal support are typically

binding upon the courts unless unconscionable or extremely unfair.

102

These problems with judicial review extend beyond the family law

context. With any settlement that may affect third parties, judicial re-

view is a potential solution, but it will always be difficult for judges to

determine whether a settlement is unfair to unrepresented parties.

3. Changes in Terminology

In addition to replacing the best-interests-of-the-child standard

with the more determinate approximation rule, the ALI sought to

conceptually distinguish child-custody settlements from those con-

cerning monetary issues by requiring parents to present their custody

agreements in the form of a separate agreement labeled a “parenting

plan.”

103

A number of states now employ “parenting plan” or similar

terminology to characterize custody settlements,

104

both to signal the

goal of parental cooperation

105

and to encourage parents to work out a

100

See Abramowicz, supra note 65, at 80–81. R

101

See id. at 80; see also, e.g., Mnookin & Kornhauser, supra note 43, at 993 (“Courts R

typically rubber stamp an agreement reached by the parties.”).

102

See Penelope Eileen Bryan, Women’s Freedom to Contract at Divorce: A Mask for Con-

textual Coercion, 47 B

UFF

. L. R

. 1153, 1238 (1999).

103

See Kay, supra note 87, at 41. R

104

See, e.g., M

INN

. S

TAT

. A

. § 518.1705 (West 2017); O

. R

. S

TAT

. A

. § 107.102

(West 2017).

105

See Marsha Garrison, Promoting Cooperative Parenting: Programs and Prospects, 9 J.L.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 27 18-JUN-18 16:24

402 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

detailed custody arrangement that facilitates an ongoing relationship

with both parents.

106

Herma Hill Kay has characterized the ALI’s change in terminol-

ogy as attempting to “build a firewall” between custody agreements

and agreements regarding financial aspects of divorce.

107

She acknowl-

edges, though, that “one cannot put too much weight on this effort to

build a firewall between the parenting plan and the financial negotia-

tions; obviously, uneven bargains still can and probably will be

struck.”

108

Without a mechanism in place to prevent trade-offs be-

tween custody and property, jurisdictions can try to nudge parents to-

ward isolating their negotiations regarding custody from their

negotiations regarding the other incidents of divorce. They cannot,

however, prevent such trade-offs altogether.

C. Bifurcating Custody Settlements

Bifurcating settlements can build a stronger “firewall” between

custody settlements and other aspects of divorce negotiations without

altering substantive child custody law in ways that might not otherwise

be desirable. This section addresses the potential benefits and costs of

bifurcating settlements in the context of child custody, how best to

implement bifurcating settlements in this context, and complicating

factors that are specific to family law but may shed light on other con-

texts as well.

The context of child-custody settlements makes clear how bifur-

cation can prevent trade-offs that harm third parties. Children are par-

adigmatic unrepresented third parties, deeply affected by their

parents’ agreements, but with no say in them.

109

This is especially the

case for divorce settlements, which determine both where and with

whom the children will live and how much money will be contributed

to their support.

110

If parents could not make financial concessions in

exchange for a desired custody arrangement, children would no longer

& F

. S

TUD

. 265, 270 (2007) (“[S]ome states now use the term ‘parenting plan’ instead of

‘custody and visitation’ . . . to reduce the sense that there are custody ‘winners’ and ‘losers’ and

thus to promote parental cooperation.”).

106

See id.

107

Kay, supra note 87, at 41. R

108

Id.

109

Cf. Sarah Abramowicz, Beyond Family Law, 63 C

ASE

W. R

. L. R

. 293, 336 (2012)

(noting that while children may be profoundly affected by parental contracts, outside of the

family law context neither courts nor scholars consider how parental contracts affect children’s

interests).

110

While many jurisdictions permit or require courts to consider children’s preferences in

resolving custody disputes, courts typically are not required to allocate custody according to

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 28 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 403

suffer the diminished material well-being or the expressive harm that

such trade-offs may inflict. Parents who forgo their custodial rights in

such trade-offs would no longer be driven by financial incentives to

accept a suboptimal custody arrangement, for instance one that detri-

mentally reduces the child’s contact with a noncustodial parent.

Meanwhile, if bifurcation can eliminate or reduce strategic threats to

litigate custody, this may make the settlement process less adversarial

and more cooperative, which commentators have linked to more har-

monious coparenting postdivorce.

111

The mechanism by which bifurcation reduces trade-offs is

straightforward. Suppose divorcing parents would like to give a

mother more custody than she could expect at trial in exchange for

less money from the father. They might still attempt this even in a

world of bifurcated settlement, but if custody is resolved first, the fa-

ther might worry that after custody is resolved, the mother will renege

on any financial promises.

112

If the father trusts the mother not to re-

nege on an implicit deal, they might still make trade-offs, however.

This is a more general limitation of bifurcated settlements; parties that

trust each other may still make trade-offs.

113

Similarly, opposing law-

yers might enter into implicit deals with one another, though the

mother’s lawyer would be duty bound to remind the mother that she

is not bound by any implicit deal. In any event, the possibility that

trust may allow evasion of bifurcated settlements does not argue

against such a settlement regime. It merely shows that in some per-

centage of cases, where trust is sufficiently high, bifurcation will have

neither significant costs nor significant benefits. Thus, a normative

analysis of bifurcation must depend largely on the remaining cases in

which parties do not trust one another sufficiently to thwart settle-

ment bifurcation.

To appreciate the degree to which bifurcating settlements reduces

custody-financial trade-offs, we must clarify the economic explanation

for why such trade-offs exist. Some of the literature on child-custody

negotiations worries about disparate bargaining power,

114

which is

those preferences. See Barbara A. Atwood, The Child’s Voice in Custody Litigation: An Empiri-

cal Survey and Suggestions for Reform, 45 A

RIZ

. L. R

. 629, 640 (2003).

111

See, e.g., Scott, supra note 43, at 649 (“Anger and bitterness generated in conflictual R

negotiation or adjudication can poison the prospects of future cooperation between parents.”).

112

See Altman, supra note 8, at 528 (“Even if one party obtained the custody agreement by R

agreeing to bad financial terms, she could now change her mind, and seek reasonable financial

terms without fear of losing custody or visitation.”).

113

See supra note 7 and accompanying text. R

114

See, e.g., Bryan, supra note 102, at 1180–91 (enumerating reasons that women are often R

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 29 18-JUN-18 16:24

404 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

driven in part by asymmetric risk aversion.

115

The party that is more

risk averse will generally be willing to make more concessions to the

other party to avoid a trial. Some claim that women are more risk

averse than men and that this affects women adversely in male-female

custody negotiations.

116

But risk aversion by itself seems unlikely to

lead to trade-offs between custody and financial issues. The party that

is more risk averse would be expected to make greater concessions

with respect to each issue, not necessarily to make trade-offs between

issues.

But if parents have asymmetric preferences

117

with regard to two

issues, trade-offs become more likely. Suppose, for example, that a

mother cares much more about custody relative to money than a fa-

ther. An extra increment in custody might be worth a sacrifice of

$10,000 in income to the mother but only $1000 in income to the fa-

ther. If that is so, then the parents are likely to trade custody for

money. Moreover, while the trade-off maximizes the joint interests of

the parents, it may seem unbalanced. We can rate settlements relative

to the expected distribution of judicial decisions, characterizing a

tenth-percentile outcome from the mother’s perspective as one that

she would expect to do better than 90% of the time. The mother

might accept such an outcome for a sixtieth-percentile outcome on

child custody, since she is willing to give up a relatively large amount

of money for a small improvement in custody. If both parties cared

much more about custody than money, this might seem like a bal-

anced agreement (placing the child’s interests aside), but the mother’s

hypothesized greater concern with custody leads to an unbalanced

outcome.

Loss aversion may further contribute to trade-offs. The behav-

ioral economics literature recognizes that what an actor experiences as

a loss may matter a great deal more than what the actor experiences

at a disadvantage in divorce negotiations with men, including lack of legal representation or

inadequate legal representation, less aggressive styles of conflict resolution, and lesser status and

education); Trina Grillo, The Mediation Alternative: Process Dangers for Women, 100 Y

ALE

L.J.

1545, 1601–05 (1991) (suggesting that the “ethic of care” may make women more conciliatory

than men, putting women at a disadvantage in divorce and custody mediation).

115

See Mnookin & Kornhauser, supra note 43, at 979–80 (discussing how having relatively R

greater risk aversion than men can put women at a disadvantage in custody bargaining).

116

See generally Margaret F. Brinig, Does Mediation Systematically Disadvantage Women?,

2 W

. & M

ARY

J. W

OMEN

& L. 1 (1995) (evaluating claim that women are disadvantaged in

divorce negotiations because they are more risk averse than men).

117

See generally Brinig & Alexeev, supra note 44 (finding evidence of asymmetric prefer- R

ences regarding custody).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 30 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 405

as a gain.

118

Thus, a parent who has been the primary caregiver may be

highly sensitive to the possibility of losing primary custody. Margaret

Brinig and Michael Alexeev note that litigating custody may have

greater “downside risk” for the primary caregiver.

119

Loss aversion in-

teracts with asymmetric stakes: while loss aversion will affect how par-

ents react to prospective losses in financial position, asymmetric

stakes again may lead to compromises. Elizabeth Scott writes that

“the psychological response of loss aversion combines with the realis-

tic assessment of asymmetrical stakes in the outcome to reinforce the

tendency of mothers facing adjudication under the best interests stan-

dard to settle out of court.”

120

In other words, economic theory rein-

forces the unsurprising point that someone who is deeply connected to

her child may be so concerned about a potential loss of custody that

she sacrifices a great deal financially, even though that causes a loss

too.

Bifurcation would not eliminate distortions in settlement attribu-

table to loss aversion, but it would help ensure that asymmetric pref-

erences do not lead to trade-offs. If a primary caregiver is loss averse

with respect to any reduction in custody, then that caregiver might be

especially resolved not to give up even a small amount of custody to

the secondary caregiver. This could lead to litigation, or it could lead

the other parent to agree to receiving less custody. On the other hand,

a primary caregiver might not be so loss averse with respect to a small

reduction in custody, but very loss averse with respect to the prospect

of receiving only visitation. In this case, the primary caregiver would

likely make greater concessions on custody than if trade-offs were

permitted, and might be more likely to accept restrictions, such as

agreeing not to relocate, as a condition of retaining custody. It is thus

not clear how bifurcation would affect the allocation of custody, and,

given the notorious difficulty of assessing children’s best interests in

the context of custody, even less clear whether the custody allocation

produced by bifurcation would promote or work against those inter-

ests. What is clear is that the more involved parent would be less likely

to accept a financial compromise adverse to the child’s interests.

118

See, e.g., Daniel Kahneman et al., Anomalies: The Endowment Effect, Loss Aversion,

and Status Quo Bias, 5 J. E

CON

. P

ERSP

. 193, 193, 199 (1991).

119

See Brinig & Aleexev, supra note 44, at 288. R

120

Scott, supra note 43, at 654. R

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 31 18-JUN-18 16:24

406 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

1. Costs of Bifurcation

Bifurcating custody settlements from the financial aspects of di-

vorce has at least two potential costs: first, preventing “good parents”

from sacrificing their material well-being for the sake of obtaining a

desired custody arrangement may lead to worse custody allocations;

and second, bifurcation might increase litigation rates and worsen the

settlement process. These costs would fall in part on children,

121

thus

inflicting harm on the same third parties that bifurcating custody

would seek to protect.

a. The Solomonic Problem and Settlement Accuracy

In the biblical story of custody adjudication, King Solomon was

confronted with two women who both claimed to be the mother of the

same infant.

122

Solomon decreed that because he could not determine

which was the true mother, the baby would be cut in half and divided

between the claimants accordingly. In response, one woman immedi-

ately cried out that King Solomon should award the child to the other

rather than carry out his judgment. This reaction revealed the objector

to be the true mother precisely because she was willing to sacrifice her

own happiness for the sake of her child’s well-being, and so King Solo-

mon gave the child to her.

123

Bifurcating custody settlements from financial ones creates the

potential of what we will call “the Solomonic problem.”

124

The Solo-

monic problem stems from the fact that, in the context of custody

trade-offs, parental sacrifice can potentially help to produce the opti-

mal custody allocation. Trade-offs play an informational function in

that they enable parents to demonstrate the extent to which they

value their children’s custody. Trade-offs can at the same time help

the parent who cares the most about custody, and thus is putatively

the better parent, to obtain a greater share of custody. The Solomonic

problem illustrates a significant potential downside of bifurcating set-

tlements more generally: What if, by eliminating trade-offs, we reduce

settlement accuracy? In the context of child custody, where the pri-

mary goal is promoting children’s interests, we can define settlement

121

Studies have found that litigating custody diminishes children’s well-being by elevating

parental conflict, reducing parental cooperation postdivorce, and increasing the likelihood of

postdivorce litigation. See Scott, supra note 43, at 648–49. R

122

1 Kings 3:16–28.

123

Id.

124

Cf. Garska v. McCoy, 278 S.E.2d 357, 362 (W. Va. 1981) (coining the term “Solomon

syndrome” to characterize the “phenomenon . . . that the parent who is most attached to the

child will be most willing to accept an inferior bargain”).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 32 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 407

accuracy as the settlement outcome that is most beneficial to the chil-

dren whose custody is at issue. Eliminating trade-offs may improve

children’s material well-being. But this material benefit must be

weighed against the possibility that bifurcation may affect negotiated

custody outcomes in ways detrimental to children.

Of course, it is not inevitable that eliminating trade-offs would, in

fact, affect negotiated arrangements to children’s detriment. First, it is

not clear that eliminating trade-offs would necessarily affect the custo-

dial outcome. Empirical studies that have attempted to assess the

presence and extent of custody trade-offs have found that custody set-

tlements tend to remain stable when background custody entitlements

change, and that what changes, if anything, is the financial settlement,

as well as the extent of litigation.

125

This suggests that eliminating cus-

tody trade-offs may have no effect on custody outcomes, while poten-

tially improving the financial position of primary custodial parents.

Some commentators, moreover, have hypothesized that eliminating

trade-offs would improve custody outcomes, as well as the settlement

process. Elisabeth Scott, in advocating building a firewall between

custody settlements and other aspects of divorce, has argued that

where parents work out the details of custody without bringing in

other aspects of the divorce settlement, they are less likely to engage

in zero-sum bargaining and more likely to work cooperatively to come

up with a custody arrangement that maximizes both parents’ prefer-

ences as well as their children’s interests.

126

A related question is whether the possible effect on custody set-

tlements of eliminating trade-offs would necessarily be detrimental to

children. Even if the willingness to forgo money for custody indicates

a greater desire for custody, a child will not always be better off with

the parent who places a higher value on custody. Moreover, even if

parental willingness to sacrifice financial well-being to obtain custody

is a proxy for good parenting, this does not mean that a child will

always benefit from spending more time with this “better” parent and

less with the “worse” parent. The prevailing view is that children ben-

efit by establishing a close and continuing relationship with both par-

125

See Brinig & Alexeev, supra note 44, at 283 (finding that “regardless of the legal rule, R

the outcomes in terms of ultimate child custody arrangements are very similar” and that “the

difference appears in the share of wealth received by the divorcing wife or in her ability to

receive long-term spousal support”); see also Brinig, supra note 45, at 806–07 (finding that intro- R

duction of legal standards more favorable to joint custody was followed by an increase in joint

custody, but that this increase was consistent with earlier trends and “might have occurred even

without the change in statute”).

126

See Scott, supra note 43, at 644–52. R

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 33 18-JUN-18 16:24

408 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

ents, and that the ideal custodial outcome will ensure sufficient

contact to maintain or create such a relationship with both parents.

127

To the extent that eliminating trade-offs would increase time with

noncustodial parents, this could be to children’s benefit.

If, in fact, custody trade-offs produce the optimal custody out-

come, then the question that follows is whether the detriment to chil-

dren from a less beneficial custody arrangement (as a result of

eliminating such trade-offs) is outweighed by the benefit to children of

the greater financial security (to the extent that eliminating custody-

money trade-offs would improve the financial well-being of children

and custodial parents). The answer is likely no. The prevailing view of

family law scholars and judges is that material wealth matters less to

children’s well-being than parental care and love.

128

Even moderate

economic deprivation can adversely affect children. As long as a

child’s basic needs are met, however, the quality of the child’s rela-

tionship with a parent, and the degree of emotional, psychological,

and developmental well-being that this relationship produces for the

child, is considered more important than material resources to the

child’s “best interests.”

129

The Solomonic problem may be specific to family law, but it illus-

trates a broader point with relevance to other potential applications of

settlement bifurcation: if settlement succeeds in forcing independent

negotiation of an issue, this will benefit third parties only on the as-

sumption that the best outcome for third parties is the one that the

negotiating parties would agree to in the absence of trade-offs.

b. Potential Adverse Effects on Settlement

Another possible downside of bifurcation is the potential to in-

crease litigation rates. The context of child custody brings out a re-

lated, but distinct, possible downside: its effect on the settlement

127

See, e.g., M

ICH

. C

OMP

. L

AW S

. § 722.23(j) (West Supp. 2017) (providing that custody

courts assessing children’s best interests must consider, inter alia, “[t]he willingness and ability of

each of the parties to facilitate and encourage a close and continuing parent-child relationship

between the child and the other parent or the child and the parents”); Solangel Maldonado,

Beyond Economic Fatherhood: Encouraging Divorced Fathers to Parent, 153 U. P

. L. R

. 921,

949–60 (2005) (describing educational, psychological, and emotional benefits to children of

greater engagement by divorced fathers).

128

See Burchard v. Garay, 724 P.2d 486, 491 (Cal. 1986) (in bank) (“[T]here is no basis for

assuming a correlation between wealth and good parenting or wealth and happiness.” (alteration

in original) (quoting Ramsay Laing Klaff, The Tender Years Doctrine: A Defense, 70 C

ALIF

. L.

. 335, 350 (1982))); Carolyn J. Frantz, Note, Eliminating Consideration of Parental Wealth in

Post-Divorce Child Custody Disputes, 99 M

ICH

. L. R

. 216 (2000) (collecting sources).

129

See sources cited supra note 128. R

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 34 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 409

process itself. Where a custody settlement is reached, but is preceded

by a drawn-out and adversarial negotiation, the emotional and rela-

tional costs to the parties and their children are similar to those of

litigation.

130

The relational costs of an adversarial settlement process

are potentially relevant in any settlements where, as in child custody,

the parties will have a continuing, long-term relationship even after

the settlement ends, such that they benefit from a settlement process

that facilitates cooperation and trust.

131

The economic model of Part I showed that when litigation costs

are especially high, bifurcation will tend to reduce litigation.

132

Indeed,

the divorce context can provide some intuition underlying the model.

Litigation costs may be high in divorce proceedings, especially if the

emotional costs of such litigation are factored in. Once child custody is

resolved, then only financial issues will be in play, and the cost of liti-

gating those issues will be very high relative to the stakes. Thus, finan-

cial issues, once isolated, are quite likely to be settled. Anticipating

this increases the payoff to resolving child custody. Even though

under bifurcation two settlements must be reached in a divorce case

instead of one, the stakes of each settlement will be lower than if par-

ties were simultaneously negotiating a single settlement encompassing

both child custody and financial issues. With sufficiently high litigation

costs, two moderate-stakes settlements will be easier to achieve than a

single high-stakes settlement. Nonetheless, as long as trade-offs be-

tween issues remains possible, it is unlikely that the parties will volun-

tarily decide issues seriatim.

This model assumes rational actors and assumes away psychologi-

cal complications. Parties act strategically in divorce litigation, and so

we believe this model has considerable resonance even in the divorce

context. The divorce context, however, highlights other considerations

that may push in the opposite direction. In the emotionally fraught

context of divorce, trade-offs between custody and other terms of a

divorce settlement could play a positive emotional role in ways that

would increase parental satisfaction with settlements and reduce liti-

gation rates. Were bifurcation to preclude such trade-offs, we could

predict that the result in such cases might be an increase in litigation

rates.

Consider, for instance, a parent who feels betrayed by and bitter

about the divorce, and has a psychological need to win some aspect of

130

See Scott, supra note 43, at 649. R

131

See id. at 648.

132

See supra Part I.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 35 18-JUN-18 16:24

410 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

the divorce settlement.

133

Where trade-offs are available, such a par-

ent might be satisfied by an arrangement in which he secures a slightly

more advantageous financial arrangement implicitly or explicitly in

exchange for not demanding a greater share of custodial time. This

slight victory might fill the parent’s need to feel that he was not com-

pletely taken advantage of in the divorce settlement. Absent the op-

tion of trade-offs, it is possible that such a parent, unable to obtain

even a small advantage over the other parent, might feel emotionally

driven to litigate instead of settling.

134

We acknowledge that this ac-

count is speculative, but our point is simply to illustrate that the eco-

nomic model does not take into account all relevant psychological

factors.

A related issue that our model in Part I does not address is the

effect of bifurcating settlements on the duration and tenor of the set-

tlement process itself. Encouraging parents to work out a custody ar-

rangement in isolation from other issues may ameliorate the

settlement process by making it more cooperative and child centered.

On the other hand, it is also possible that bifurcation could make cus-

tody settlements more adversarial and costly. Bifurcation would in-

crease the number of times that parents must appear before a judge—

first to gain approval of the custody settlement, and later to gain ap-

proval of their settlement on other issues stemming from the divorce.

Relatedly, by breaking the settlement process into stages, bifurcation

may prolong the complexity and duration of the settlement process,

which in turn could exacerbate the emotional turmoil of the divorce as

well as increase legal fees. It is difficult, of course, to weigh these con-

siderations against the benefits of settlement bifurcation. The analysis,

in any event, highlights that the benefits of settlement bifurcation may

be reduced in litigation in which emotion plays a relatively large role.

2. Obstacles to Bifurcation

Some issues specific to family law might complicate the imple-

mentation of bifurcation. By addressing some of these issues, we can

133

Cf. J

UNE

ARBONE

, F

ROM

ARTNERS TO

ARENTS

: T

ECOND

EVOLUTION IN

AM-

ILY

193 (2000) (“Custody battles have become ground zero in the gender wars because they

are among the few remaining family law disputes where courts judge adult behavior.”).

134

A number of commentators have noted the role of spite, anger, and similar emotions in

child-custody bargaining. See Altman, supra note 8, at 529 (“The strength of spite and anger as R

motives should not be ignored.”); Mnookin & Kornhauser, supra note 43, at 974 (acknowledging R

that parents might decide to litigate custody rather than settle out of spite and a desire to pun-

ish); Scott, supra note 43, at 646–47 (“[S]pite . . . is a familiar aspect of divorce negotiations.”). R

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 36 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 411

consider some of the complexities of real-world settlement and litiga-

tion that any attempt to bifurcate will implicate.

a. De Facto Trade-Offs

Perhaps the greatest obstacle to bifurcating settlements in the

custody context is that there is nothing to prevent parents from mak-

ing de facto trade-offs down the road. A parent intent on making

trade-offs could always request a greater amount of custody than he

desires, either at settlement or at trial, and then, once the court pro-

ceedings are concluded, offer to relinquish actual custody in exchange

for financial compensation from the other parent. The law could, of

course, prevent a formal modification of the settlement. Still, courts

have no mechanism for policing divorced parents’ day-to-day arrange-

ments. More generally, when negotiating parties engage in repeated

interactions following settlement, there may be opportunities for the

parties to make trade-offs unmonitored by the court.

An additional complication in child custody is that nominal cus-

tody can be used to avoid child-support obligations. Child support is

typically paid by the noncustodial parent to the custodial parent, with

deductions for the amount of time that the child spends with the non-

custodial parent.

135

When a parent declines to exercise the full extent

of visitation that he has been allocated, but receives a deduction in

support obligations on the basis of that allocation, the custodial parent

receives less child support than she should be entitled to on the basis

of the de facto custodial arrangement.

136

The custodial parent may

nonetheless refrain from requesting additional support, either because

she does not want to provide the noncustodial parent with the incen-

tive to fully exercise his rights to parenting time, or because she lacks

the resources to request a modification of child support consistent

with the de facto arrangement.

137

Nominal custody would likely have

an expanded role in a bifurcation regime because it allows ex post

135

See Sanford L. Braver et al., Public Sentiments About the Parenting Time Adjustment in

Child Support Awards, 49 F

. L.Q. 433, 434–35 (2015) (surveying approaches to child-support

adjustments on the basis of time spent with the noncustodial parent).

136

See Altman, supra note 8, at 519 (“[I]t is often said that . . . noncustodial parents seek R

substantial visitation in order to reduce their child support obligations, and then do not even

exercise their visitation rights.”). The flip side of this problem is that primary custodial parents

will at times refuse visitation to noncustodial parents who are delinquent in paying support.

Although fulfillment of support obligations is supposed to be decoupled from visitation rights,

this form of self-help is similarly difficult to prevent. See Mnookin & Kornhauser, supra note 43, R

at 964–66.

137

See, e.g., Karen Syma Czapanskiy, The Shared Custody Child Support Adjustment: Not

Worth the Candle, 49 F

. L.Q. 409, 426–28 (2015).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 37 18-JUN-18 16:24

412 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

custody-financial trade-offs that substitute for ex ante negotiated

ones.

b. Modifiability of Custody Arrangements

Custody arrangements can be modified at any time until a child

reaches the age of majority. There typically needs to be some sort of

change in the child’s circumstances to justify consideration of modifi-

cation,

138

but even the inevitable fact of the child maturing has been

found to constitute such a change.

139

For example, developmental

problems a child experiences may prompt such consideration.

140

Most

jurisdictions have held that parents cannot by agreement govern or

eliminate future modifications to custody.

141

In the absence of bifurcated settlements, the modifiability of cus-

tody arrangements may discourage trade-offs. Were a parent to trade

custody for money, this would do nothing to prevent the parent who

agreed to relinquish custody from later bringing a motion to modify

custody. Thus, the parent who relinquished marital property or sup-

port in exchange for custody could possibly lose custody nonetheless,

in the context of modification.

At the same time, modifiability may make it difficult to com-

pletely bifurcate custody from other issues. Even once parents have

agreed to a custody arrangement, the potential threat of modification

remains. Scott Altman has observed that parents negotiating financial

terms after resolving custody might feel pressured to make conces-

sions rather than risk provoking a future modification request.

142

Alt-

man proposes for this reason that bifurcation be accompanied by

making modification more difficult to obtain.

143

On the other hand,

138

The majority of states require a showing of a substantial change in circumstances affect-

ing the child’s welfare in order for a court to consider whether the child’s best interests require a

change in custody. See, e.g., O

. R

. S

TAT

. § 107.135(1) (2015); W

ASH

. R

. C

ODE

§ 26.09.260(1) (West 2017). Some states, however, do not require such a showing, allowing modi-

fication upon a showing that a change in custody is in the child’s best interests. See, e.g., N

. S

TAT

. A

. § 125.510(1) (LexisNexis 2010).

139

See, e.g., Morales v. Lincoln, 367 S.W.3d 174, 179 (Mo. Ct. App. 2012) (finding child’s

entry into kindergarten constituted a change in circumstances sufficient to modify joint custody

arrangement).

140

See, e.g., Brewer v. Whitney, 666 N.Y.S.2d 354, 355 (App. Div. 1997) (finding change in

circumstances sufficient to change from joint to sole custody where father better equipped to

handle child’s newly diagnosed speech and developmental delays).

141

See Abramowicz, supra note 65, at 85–88. R

142

See Altman, supra note 8, at 528. R

143

See id.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 38 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 413

giving in to financial demands would not remove the risk of a future

modification request.

While making modification more difficult to obtain might help to

more completely sever custody settlements from settlements regard-

ing other terms, bifurcation would nonetheless provide its intended

benefit of reducing trade-offs even under current modification stan-

dards. Most current custody modification standards favor continuing

the status quo, such that the initial custody allocation is a powerful

buffer against modification.

144

A parent who has already negotiated

and finalized a desired custody allocation should feel more comforta-

ble about negotiating financial terms without risking loss of custody

than if custody and financial terms were negotiated together.

145

3. Design of Bifurcation

A potential argument against bifurcation is that it is impractical

because issues are intertwined. Divorce settlements have been de-

scribed as entailing a “mosaic” of closely related issues, many of which

cannot be considered in isolation from the rest.

146

Matters of marital

property division, spousal support, and, most obviously, child support

are often contingent on custody outcomes.

147

When a parent is

awarded primary custody, this may affect whether that parent is

awarded the marital home (so that the children can continue living

there), as well as how much marital property each parent is entitled

to. Spousal support may sometimes be awarded on the basis that a

parent will stay home to take care of young children, so it would be

difficult to determine the proper amount of spousal support without

144

See, e.g., O

. R

. S

TAT

. § 107.135(1) (2015) (requiring a showing of a substantial

change in circumstances affecting the child’s welfare in order for a court to consider modifying

custody). The preference for the status quo is reinforced by the best-interests standard that the

court applies once a change in circumstances has been established; in most jurisdictions, either

statute or caselaw dictate that it is typically in a child’s best interest to maintain continuity. See,

e.g., id. § 107.137(1)(c) (directing courts assessing best interests to factor in “[t]he desirability of

continuing an existing relationship”); Deyo v. Deyo, 658 N.Y.S.2d 153, 155 (App. Div. 1997)

(describing “the general preference for maintaining stability by continuing the existing residen-

tial arrangement when appropriate”).

145

Moreover, where custody settlements are bifurcated, trade-offs would be less likely to

shape the original custody allocation than under the current approach, where custody and other

terms are settled in the aggregate.

146

See Taff v. Bettcher, 703 A.2d 759, 760 n.2 (Conn. 1997) (“[O]rders relating to custody

and support are part of a carefully crafted mosaic such that a change to one will necessarily

create a change to the other.”).

147

See In re Marriage of Jordan, 203 N.W.2d 314, 316 (Iowa 1972) (finding that property

settlement and child support were “influenced considerably by the award of custody”).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 39 18-JUN-18 16:24

414 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

first knowing the custody outcome.

148

And, under the federally man-

dated child-support guidelines in place in all states,

149

child support is

directly contingent on the allocation of custody.

The most straightforward design is thus seriatim bifurcation with

custody settled first, followed by settlement of marital property,

spousal support, and child support. This way, custody could be deter-

mined before the issues that tend to be contingent on custody. We

suspect that this is the best approach to bifurcation on balance, but it

is important to recognize arguments for alternatives. One alternative

would be to decide financial issues first. A benefit to this is that finan-

cial issues will often be easier to resolve. The parties may have a bet-

ter chance of resolving custody issues if they know that such

resolution will definitively resolve the divorce proceedings. To enable

financial issues to be resolved first, the financial agreement could in-

clude terms contingent upon different possible custody outcomes.

Another alternative is severable settlements. Parties would nego-

tiate child custody and financial issues simultaneously, but either party

would be able to back out of either portion of the settlement. The

disadvantage of severable settlements is that interdependencies be-

tween provisions must be resolved. Severable settlements have two

advantages over seriatim bifurcated settlements, however. First, sever-

able settlements may reduce the number of court appearances to just

one.

150

In most cases, neither party will back out of an agreed-upon

settlement. That does not mean that bifurcation is irrelevant. The

prospect that the other party might back out will tend to reduce trade-

offs in the initial agreement, so actually exercising the power to back

out of a portion of a settlement is not necessary. Second, each party

must trust the other to evade the severable settlements regime be-

cause either party may defect from the strategy of cooperation,

whereas seriatim bifurcated settlements can be evaded even where

trust only runs one way.

151

Suppose, for example, that the parties

would like to make a trade-off where a husband receives more money

and a wife receives more custody than a court would be expected to

award. If custody is negotiated first, then the parties will be able to

negotiate this trade-off if the husband trusts the wife to stick to a fi-

148

See, e.g., O

HIO

. C

ODE

. § 3105.18(C)(1)(f) (West 2017) (providing that a court

awarding spousal support must consider, inter alia, “[t]he extent to which it would be inappropri-

ate for a party, because that party will be custodian of a minor child of the marriage, to seek

employment outside the home”).

149

See 42 U.S.C. § 667(a) (2012) (mandating state adoption of child-support guidelines).

150

See supra Section II.C.

151

See supra Section II.C.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 40 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 415

nancial deal. But with severable settlements, parties will be willing to

risk a trade-off only if each party trusts the other not to renege.

Especially with severable settlements, though also with seriatim

bifurcated settlements, the court should, prior to approving a custody

agreement, ensure to its satisfaction that each parent understands the

bifurcation mechanism. With severable settlements, the court should

stress that regardless of the position of the attorneys, each party re-

tains the ability to back out of either part of the agreement by notify-

ing the court confidentially by a set date. With seriatim bifurcation,

meanwhile, parents might negotiate other issues at the same time that

they negotiate custody, but they could not be bound by such agree-

ments until after determination of custody (or financial issues, if those

are to be resolved first). The possibility of an asymmetry in legal infor-

mation and representation makes it critical that the court verify par-

ties’ understanding; otherwise, one party could easily take advantage

of the other’s ignorance.

III. P

ROTECTING

HIRD

ARTIES

’ F

INANCIAL

NTERESTS

Negotiating litigants will structure settlements to benefit them-

selves at third parties’ expense. Even if mutually antagonistic, both

parties can profit from settlements that reallocate value from third

parties to one party because the other party can share in the benefit

through some other concession. This Part describes situations in which

bifurcation might combat such value extraction. Our primary contri-

bution is not the idea of bifurcating settlements in these areas. Indeed,

for several of these potential applications, at least one scholar has

noted or suggested the possibility of bifurcating settlements. Strik-

ingly, though, the possibility of bifurcation has been mentioned only

in passing and usually quickly dismissed. We suspect that settlement

bifurcation has received so little attention in these contexts because it

is alien to our legal culture. Scholars and courts have thus devoted far

more attention to substantive-law workarounds that address third-

party effects indirectly. Our primary contribution in this Part is to

make settlement bifurcation seem less strange by treating together

what might initially seem like unrelated problems. By classifying value

extraction from negotiated third-party settlements as a general prob-

lem and identifying settlement bifurcation as a solution, we identify at

least one context in which settlement bifurcation does not seem to

have received any consideration.

152

Meanwhile, by taking settlement

152

See infra Section III.A.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 41 18-JUN-18 16:24

416 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

bifurcation seriously, we identify a wider range of benefits and costs to

bifurcation in particular contexts than prior commentators have

considered.

A. Protecting the Government: Punitive vs. Compensatory Damages

When entering into pretrial settlements in tort cases involving

physical injuries, plaintiffs and defendants have strong incentives to

characterize any damages that they agree to as compensatory rather

than as punitive. Under the Internal Revenue Code, plaintiffs who are

physically injured may exclude compensatory damages from gross in-

come.

153

Meanwhile, defendants may deduct both compensatory and

punitive damages.

154

Payments of compensatory damages thus reduce

the government’s tax receipts, while payments of punitive damages

are revenue neutral, assuming that the parties have the same marginal

tax rate.

This asymmetric tax treatment means that given any deal that

would characterize some damages as punitive, the parties can always

benefit by recharacterizing the damages as compensatory. Suppose

the plaintiff pays taxes at a 20% marginal rate and that the parties

both valued the claim at $1,000,000, believing (with certainty, to keep

the example simple

155

) that the jury would find $500,000 in compensa-

tory damages and $500,000 in punitive damages. Assuming the courts

will defer to the parties’ characterization of the damages, as they gen-

erally do,

156

the parties would anticipate that an agreement honestly

characterizing damages payments as equally divided between compen-

153

I.R.C. § 104(a)(2) (2012).

154

See Rev. Rul. 80-211, 1980-2 C.B. 57, 1980 WL 130077. See generally Kimberly A. Pace,

The Tax Deductibility of Punitive Damage Payments: Who Should Ultimately Bear the Burden

for Corporate Misconduct?, 47 A

. L. R

. 825 (1996) (arguing against the current treatment

of punitive damages).

155

The analysis would be the same if the parties both expected a 50% chance of $1,000,000

in compensatory damages and $1,000,000 in punitive damages, or a 100% chance of $500,000 in

compensatory damages and a 10% chance of $5,000,000 in punitive damages.

156

Byrne v. Comm’r, 90 T.C. 1000, 1007 (1988) (“[T]he most important fact in determining

the purpose of the payment is ‘express language [in the agreement] stating that the payment was

(or was not) made on account of personal injury.’” (alteration in original) (quoting Metzger v.

Comm’r, 88 T.C. 834, 847 (1987))). But see M. Sean Sullivan, Comment, Achieving the Best Tax

Treatment for Your Physical Injury Client’s Settlement Award, 66 M

ISS

. L.J. 579, 587 (1997) (cit-

ing Robinson v. Comm’r, 102 T.C. 116 (1994), aff’d in part, rev’d in part, 70 F.3d 34 (5th Cir.

1995)) (“[A] court will in proper circumstances disregard the terms of the agreement and allo-

cate the damages on its own.”).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 42 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 417

satory and punitive damages would lead the plaintiff to pay $100,000

in taxes.

157

In contrast, a settlement agreement characterizing the $1,000,000

payment entirely as compensatory damages for physical injury would

benefit the plaintiff and leave the defendant no worse off. Similarly, a

settlement agreement for $900,000, characterized entirely as compen-

satory damages, would benefit the defendant while leaving the plain-

tiff no worse off. Thus, if $1,000,000 would have been the settlement

value in a world without taxes, the parties will likely agree to a settle-

ment value somewhere between $900,000 and $1,000,000, perhaps at

$950,000,

158

characterized entirely as compensatory damages. Unsur-

prisingly, given this calculus, leading attorneys report that they never

specifically allocate any portion of damages to punitive damages in

settlement agreements.

159

Insurance generally does not cover punitive

damages,

160

providing a further basis to characterize damages as com-

pensatory rather than punitive.

Much of the scholarly debate has focused on whether the deduct-

ibility of punitive damages interferes with the objectives of tort law. In

the leading article on this topic, Gregg Polsky and Dan Markel noted

that because juries making punitive damage determinations often as-

sume that punitive damages are not deductible, or fail to take this

issue directly into account, “the true cost of a punitive damages award

is often substantially less than the nominal amount of the award.”

161

This has led some policymakers to argue for making punitive damages

nondeductible.

162

Polsky and Markel defend the alternative policy of

making jurors aware of the tax implications of the deductibility of pu-

nitive damages.

163

Central to Polsky and Markel’s argument is that de-

fendants could circumvent a nondeductibility rule by entering into

157

The $500,000 in compensatory damages would be untaxed, and 20% of the remaining

$500,000 in punitive damages is $100,000.

158

The literature often assumes that parties settle at the midpoint of the settlement range.

See, e.g., Samuel Issacharoff & George Loewenstein, Second Thoughts About Summary Judg-

ment, 100 Y

ALE

L.J. 73, 98 (1990).

159

See Tom Baker, Transforming Punishment into Compensation: In the Shadow of Puni-

tive Damages, 1998 W

. L. R

. 211, 218 (reporting the result of interviews with personal injury

lawyers in which none of the interviewees had settled “a case for an amount that included a

portion identified as ‘punitive damages’ ”); Gregg D. Polsky & Dan Markel, Taxing Punitive

Damages, 96 V

. L. R

. 1295, 1334 (2010) (“[S]ettlement agreements routinely and expressly

allocate the entire amount to compensatory damages.”).

160

Cf. Catherine M. Sharkey, Revisiting the Noninsurable Costs of Accidents, 64 M

. L.

. 409, 413, 452 (2005) (proposing to allow insurability of some punitive damages).

161

Polsky & Markel, supra note 159, at 1297. R

162

Id. at 1298–99 & n.5 (collecting sources).

163

Id. at 1299, 1302–24.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 43 18-JUN-18 16:24

418 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

settlements that would recharacterize punitive damages as compensa-

tory.

164

If jurors were aware of the nondeductibility of punitive dam-

ages and were to “gross up” damages awards as a result, then

settlement negotiations would take into account this anticipated in-

crease in damages, and the goals of tort law would be met even in

cases that are settled.

165

This policy recommendation may improve the legal system’s abil-

ity to achieve its tort goals, but it fails to advance the government’s tax

policy goals. Settling parties would continue to characterize damages

as compensatory rather than punitive, and those damages would thus

be deductible to the defendant and yet excludable from the plaintiff’s

income. The increase in the damages award designed to counteract

deductibility would redound to the plaintiff, rather than to the govern-

ment. Unless plaintiffs have inadequate incentives to bring punitive

damages claims, this produces a windfall.

166

Meanwhile, if the alterna-

tive policy were adopted in which punitive damages were made non-

deductible, then the parties would have an even greater incentive to

characterize damages as compensatory (though because they have

plenty enough incentive already, this would not make things substan-

tially worse). And if we assume that, absent implementation concerns,

punitive damages in theory ought to be taxable, then making punitive

damages nondeductible will not have the desired effect in settled

cases.

If it were possible to induce the parties to divide any agreed-upon

damages honestly into the compensatory and punitive categories,

based on the parties’ expected value of compensatory and punitive

damages, then the legal system would have flexibility to achieve both

its tort and tax goals. If punitive damages properly should be nonde-

ductible, then a simple policy change could accomplish the goal, be-

cause Polsky and Markel’s concerns about recharacterization of

punitive damages as compensatory would vanish with an honest

breakdown. If punitive damages properly should be deductible, as

they are now, then it might well make sense to adopt the Polsky-Mar-

kel approach of instructing juries to increase the punitive damages

award based on tax concerns, thus improving optimization of the tort

system. The government would still be able to tax the punitive dam-

164

Id. at 1296 (“[D]efendants could easily circumvent the nondeductibility rule by disguis-

ing punitive damages as compensatory damages . . . .”).

165

Id. at 1304, 1307.

166

See Catherine M. Sharkey, Punitive Damages as Societal Damages, 113 Y

ALE

L.J. 347,

370 (2003).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 44 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 419

ages. Indeed, if the parties further honestly broke down punitive dam-

ages based on their anticipation of the tax gross up, the government

could decide to tax the gross-up portion of punitive damages (that is,

the portion designed to ensure that the defendant does not benefit

from deductibility) at a higher rate to minimize the plaintiff’s windfall

from the gross up.

The literature’s focus on incomplete solutions to the punitive

damages taxation problem reveals that commentators assume that

parties will continue to be able to recharacterize punitive damages as

compensatory. The same assumption can be seen in the literature on

whether plaintiffs should be required to turn over a large portion of

punitive damages to the government. Some commentators have ar-

gued that punitive damages claims produce a windfall even if taxed at

ordinary rates, and so plaintiffs should be permitted to keep only

some small percentage of punitive damage awards.

167

The result would

be similar to whistleblower suits, such as those under the False Claims

Act,

168

where the prevailing plaintiff keeps a small percentage of dam-

ages and the government keeps the rest. Parties could evade such a

rule by characterizing damages as compensatory, just as they already

do now.

The assumption that parties can reallocate compensatory and pu-

nitive damages reflects that no one has considered whether there

might be some means of distinguishing the compensatory and punitive

portions of negotiated damages. But bifurcated or severable settle-

ments could resolve this issue easily. The simplest approach is to have

the parties negotiate compensatory damages first and punitive dam-

ages second. If the settlement is submitted to a court,

169

then the court

would insist that it approve the compensatory damages award before

the punitive damages award. In this case, the court would not honor

an agreement on punitive damages unless it were dated after the

court’s approval of the compensatory damages. Of course, this can

work without court supervision as well. All that is required is that the

punitive damages agreement be signed later than the compensatory

damages agreement. If there is a concern that the attorneys cannot be

167

See, e.g., Andrew F. Daughety & Jennifer F. Reinganum, Found Money? Split-Award

Statutes and Settlement of Punitive Damages Cases, 5 A

. L. & E

CON

. R

. 134, 158–59 (2003).

168

31 U.S.C. §§ 3729–3733 (2012).

169

Under existing law, judicial approval of settlements is not generally required. See gener-

ally Sanford I. Weisburst, Judicial Review of Settlements and Consent Decrees: An Economic

Analysis, 28 J. L

EGAL

TUD

. 55 (1999) (providing an economic theory of when judicial approval

is and is not required).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 45 18-JUN-18 16:24

420 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

trusted to date their agreements properly,

170

further precautions could

be added, such as a requirement that settlement agreements be nota-

rized, or a requirement that the parties themselves sign a document

attesting that they have previously agreed to compensatory damages

and understand that they may choose not to sign an agreement on

punitive damages. A waiting period of, say, one month also might be

used to thwart joint negotiations.

This approach makes it harder for the parties to trade compensa-

tory for punitive damages. Suppose, as in the example above, that it

were in the mutual interest of the parties to agree to a judgment for

$950,000 in compensatory damages with no punitive damages, even

though the parties believed that the expected value for compensatory

and punitive damages would be $500,000 each. Assuming the parties

do not trust each other, the defendant would not be willing to agree to

compensatory damages of $950,000 because it would worry that after

that agreement were reached, the plaintiff would continue to trial on

the punitive damages claim. If punitive damages were first, the plain-

tiff would not agree to $0 in punitive damages because then the defen-

dant would have no reason to agree to pay $950,000 in compensatory

damages later. Thus, the parties will generally negotiate compensatory

damages based on their actual expectation of such damages, also tak-

ing into account considerations of litigation cost and other variables

affecting their relative bargaining power. If an agreement on compen-

satory damages were reached, the parties would then negotiate puni-

tive damages. They might well negotiate both simultaneously, of

course, but the defendant would only agree to the compensatory dam-

ages award if it thought the agreement was based on the expectation

of those damages.

It might seem that the parties could easily evade the settlement

bifurcation system by agreeing to both compensatory and punitive

damages in advance. The defendant, however, would be willing to

agree to the $950,000 compensatory award only if it trusted the plain-

tiff not to renege on any implicit deal.

171

The bifurcation of settle-

ments rules out an explicit deal. But this seems unlikely, at least in the

typical tort case. The parties are not repeat players, and the legal sys-

tem would be structured to encourage plaintiffs in this situation to re-

nege on any implicit agreement. The lawyers are repeat players, but

170

It is unethical for lawyers to backdate documents to countermand a legal requirement.

See Jeffrey L. Kwall & Stuart Duhl, Backdating, 63 B

. L

. 1153, 1157–59 (2008). Presumably,

it would also be unethical to postdate a document.

171

See supra note 7 and accompanying text. R

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 46 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 421

their duty is to their clients, and so, for example, the plaintiff’s lawyers

would be duty bound to explain to the plaintiff that it could choose

not to follow through on an agreement on punitive damages that is not

in its ex post interest.

172

It would generally be in the lawyers’ interest

to point this out anyway, whether the lawyers are paid by the hour or

on contingency. There is thus at least a substantial chance that the

plaintiff or defendant would renege on any implicit deal that trades off

punitive for compensatory damages.

The goal of a settlement bifurcation system is not to thwart settle-

ment, but to ensure that the parties will settle the different issues sep-

arately. Ordinarily, it seems likely that the parties will reach

settlements on both issues and these settlements will be stable in the

sense that neither party has an incentive to back out. But in designing

the system, we must plan for the possibility that the parties will some-

times agree on one component of damages without agreeing on the

other. What should the court do if the parties agree on compensatory

damages but fail to agree on punitive damages? In this case, there

would still be a trial and, critically, the defendant would still be able to

contest liability. A jury ideally would not be informed of the amount

of the compensatory damages in the settlement, and a judge acting as

factfinder would be encouraged to ignore this information.

173

A jury

might be instructed not to render a compensatory damages award, be-

cause the parties have agreed to that, but would still consider punitive

damages. If that seems awkward, for example because punitive dam-

ages and compensatory damages are linked,

174

the jury might produce

a full verdict, potentially without even being informed of the compen-

satory damages settlement. But any compensatory award would be

disregarded in favor of the negotiated compensatory award.

It might seem that it would make more sense at the trial to in-

struct the jury that the parties have already agreed that liability is con-

ceded and that their only task is to determine the amount of punitive

damages. Perhaps it would also seem then that the jury should even be

172

Comprehensive settlements already present lawyers with duties to ensure that clients

have provided informed consent. See Howard M. Erichson, The Trouble with All-or-Nothing

Settlements, 58 U. K

. L. R

. 979, 1017–20 (2010).

173

This may not be easy to do. See generally Andrew J. Wistrich et al., Can Judges Ignore

Inadmissible Information? The Difficulty of Deliberately Disregarding, 153 U. P

. L. R

. 1251

(2005) (concluding that it is generally not easy for judges to avoid being influenced by inadmissi-

ble, yet relevant, information of which they become aware).

174

See, e.g., Philip Morris USA v. Williams, 549 U.S. 346, 351 (2007) (finding that punitive

damages are more likely to meet due process requirements when they are less than ten times

compensatory damages).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 47 18-JUN-18 16:24

422 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

informed of the compensatory damages as an anchor on the punitive

damages. The problem with this approach is that the negotiating par-

ties would then recognize that any settlement of compensatory dam-

ages would be viewed as an admission of liability, thus increasing the

chance of a substantial judgment on punitive damages.

175

This could

make negotiation on compensatory damages considerably more diffi-

cult. Even if the parties agreed, for example, that there is a 50%

chance of liability, the defendant would hesitate to agree to a settle-

ment for 50% of anticipated compensatory damages because that

would worsen the defendant’s position on punitive damages. Thus, to

ensure that the parties negotiate each damages component based on

its expected value, a trial on just one portion of the damages must still

consider whether liability is appropriate.

This approach may seem awkward given the linkage between

compensatory and punitive damages. A claim for compensatory dam-

ages and a claim for punitive damages are not independent; ordinarily,

one cannot receive the latter without the former.

176

Moreover, courts

review punitive damages in part by considering the ratio of punitive to

compensatory damages.

177

They do this only in cases in which a jury

has already found liability, so there is no need to worry that the dam-

ages are being discounted. If a partial settlement of compensatory

damages reflects a discount for uncertain liability, this may prevent

the court from determining the true ratio in the exercise of its power

of judicial review.

A possible solution would be for the parties to negotiate punitive

damages first. This may seem to put the cart before the horse, but

there would be no need for judicial review of the punitive damages.

Moreover, the jury instructions would be straightforward; the punitive

175

See Victor E. Schwartz & Christopher E. Appel, Putting the Cart Before the Horse: The

Prejudicial Practice of a “Reverse Bifurcation” Approach to Punitive Damages, 2 C

HARLESTON

L. R

. 375, 380–81 (2008).

176

But see, e.g., Abner v. Kan. City S. R.R. Co., 513 F.3d 154, 165 (5th Cir. 2008) (allowing

punitive damages even absent a finding of compensatory damages). The Due Process Clause

does prevent states from using punitive damages to punish a defendant for injuries to nonparties.

See Williams, 549 U.S. at 346; see also Thomas B. Colby, Clearing the Smoke from Philip Morris

v. Williams: The Past, Present, and Future of Punitive Damages, 118 Y

ALE

L.J. 392 (2008) (ex-

ploring the implications of this holding and arguing that using punitive damages as punishment

for public wrong substitutes for the criminal law and thus violates the Constitution).

177

See, e.g., State Farm Mut. Auto. Ins. Co. v. Campbell, 538 U.S. 408, 425 (2003) (“Our

jurisprudence and the principles it has now established demonstrate, however, that, in practice,

few awards exceeding a single-digit ratio between punitive and compensatory damages, to a

significant degree, will satisfy due process.”); BMW of N. Am., Inc. v. Gore, 517 U.S. 559, 580–83

(1996) (finding a punitive-compensatory damages ratio of 500 to be beyond the constitutionally

acceptable range).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 48 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 423

damages instructions would simply be excluded, as the jury would

need to calculate only compensatory damages. This is an example of

how order of bifurcated settlements may matter, and why it might

make sense to bifurcate settlements rather than allow parties to enter

into severable settlements, where it will not be possible to predict

which issue might be resolved before a trial.

A variation on this approach would require the parties to negoti-

ate not the amount of punitive damages, but the punitive damages

multiplier itself. For example, suppose that the parties negotiated a

multiplier of 5.0, but then failed to negotiate a compensatory damages

amount. A jury would be asked to determine compensatory damages

only, and perhaps even explicitly instructed that punitive damages are

not available in this proceeding. The punitive damages would then be

calculated based on the earlier agreed-upon ratio. This would further

facilitate judicial review of punitive damages because a ratio would be

negotiated against the backdrop of Supreme Court caselaw on what is

permissible.

178

Scholars have suggested that factfinders might be asked

to announce a multiplier rather than a specific damages amount,

179

and this approach to bifurcation simply extends this into the settle-

ment arena.

B. Protecting Consumers: Damages vs. Market Exclusion

Reverse payment settlements under the Hatch-Waxman Act

180

provide another useful illustration of how settling parties may make

trade-offs to the detriment of third parties, in this case private rather

than governmental third parties. A goal of the statute was to en-

courage generic-drug companies to challenge the validity of patents

protecting brand-name drugs.

181

Patent litigation is expensive, and a

potential generic-drug manufacturer has little incentive to enter a

market if winning would simply mean that the market immediately

becomes open to all generic-drug manufacturers, driving the price of

such drugs down to marginal cost. Congress thus provided that the

178

See cases cited supra note 177. R

179

See, e.g., Sarah G. Cronan & J. Brittany Cross, Predictability in Punitive Damages: Con-

sidering the Use of Punitive Damage Multipliers, 79 D

. C

OUNS

. J. 454, 454 (2012) (“Rather

than awarding a single punitive damage award in each case, under the multiplier approach, a jury

sets a mathematical relationship between punitive and compensatory damages by establishing a

dollar-for-dollar ratio . . . .”).

180

Drug Price Competition and Patent Term Restoration Act of 1984, Pub. L. No. 98-417,

98 Stat. 1585 (codified as amended in scattered sections of the U.S. Code).

181

See Colleen Kelly, The Balance Between Innovation and Competition: The Hatch-Wax-

man Act, the 2003 Amendments, and Beyond, 66 F

OOD

& D

RUG

L.J. 417, 424–25 (2011).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 49 18-JUN-18 16:24

424 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

first generic company to challenge a patent successfully would receive

180 days of exclusivity.

182

Such exclusivity would keep drug prices high

and thus harm consumers, but the hope was that consumers would

benefit ex ante from increased challenges to invalid patents. This logic

parallels the more general argument for patent protection itself, that

patents cause static loss in the short term but provide dynamic bene-

fits.

183

Hatch-Waxman would tolerate the harm produced by 180 days

of exclusivity to provide incentives to challenge the apparently greater

harm of unwarranted patent exclusivity.

In passing the Act, however, Congress failed to anticipate that

litigating parties would be able to construct settlements that would

benefit the generic challenger and the brand-name incumbent at the

expense of the consumer. The mechanism for doing so is a “[r]everse

payment patent settlement[],” in which the brand-name drug manu-

facturer pays the generic manufacturer cash,

184

instead of the more

typical payment of cash by the accused infringer in a patent suit to the

patent holder. The cash payments are typically accompanied by the

generic manufacturer’s agreement to delay its entry into the market.

185

Numerous commentators have argued that such reverse patent settle-

ments have at least the potential to be anticompetitive, harming con-

sumers rather than achieving the statutory goal of striking down

invalid patents.

186

In the absence of cash payments or other benefits transferred

from the brand-name manufacturer to the generic manufacturer, the

negotiation would center around the date at which market entry

would occur.

187

The generic-drug manufacturer would like to enter the

market as soon as possible, and the patent holder would like to delay

entry as long as possible. The weaker the patent, the less delay the

patent holder would win. The stakes, however, are highly asymmetric.

The patent holder faces losing the benefit of exclusivity over poten-

tially the entire remaining life of the patent, while the generic manu-

facturer’s possible gains are limited to 180 days. Both parties can thus

benefit by a cash payment to the generic manufacturer in lieu of ex-

182

21 U.S.C. § 355(j)(5)(B)(iv) (2012).

183

See, e.g., Michael Abramowicz, The Uneasy Case for Patent Races over Auctions, 60

TAN

. L. R

. 803, 809–10 (2007).

184

See, e.g., Elhauge & Krueger, supra note 19, at 284. R

185

See id.

186

See, e.g., id.; Hemphill, supra note 9, at 639 n.40; Herbert Hovenkamp et al., Anticompe- R

titive Settlement of Intellectual Property Disputes, 87 M

INN

. L. R

. 1719, 1720 (2003).

187

See Hemphill, supra note 9, at 635. R

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 50 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 425

clusivity.

188

For example, if the parties anticipate a judgment that

would result in a loss of $10 billion for the brand-name manufacturer

and a gain of $100 million for the generic-drug manufacturer, then a

cash payment of $1 billion to the generic manufacturer would make

both parties much better off.

The only constraint on such settlements has been the prospect of

antitrust liability. A naked cash payment with no other concession

from the brand-name manufacturer might be too obviously anticom-

petitive. So, the cash payment is generally bundled with permission

for the generic-drug manufacturer to enter at some point. This point,

however, may be considerably later than the parties would have

agreed to had there been no possibility of a cash payment. To further

reduce the antitrust risk, Scott Hemphill has shown, litigating parties

have sought to disguise the benefit flowing from the brand-name to

the generic manufacturer.

189

For example, the generic manufacturer

might license some other intellectual property to the brand-name

manufacturer at an inflated price. “The additional term provides an

opportunity,” Hemphill explains, “to overstate the value contributed

by the generic firm and claim that the cash is consideration for the

contributed value, rather than for delayed entry.”

190

In FTC v. Actavis, Inc.,

191

the Supreme Court found that such set-

tlements may be anticompetitive. The facts of Actavis represent a typi-

cal reverse payments settlement. The generic firms promised to delay

entry, and the patent holder agreed to pay them an amount expected

to be over $200 million.

192

The Court found that such a settlement can

be anticompetitive if the patentee’s motive is to avoid competition.

193

Some reverse payments, however, are permissible, because the pay-

ment is not really in exchange for a delay.

194

The reverse payment, for

example, might be for “other services.”

195

This highlights a central

challenge in interpreting reverse payments settlements: because a set-

tlement agreement can encompass a number of different issues, it is

188

See id. at 634–35.

189

Id. at 633.

190

Id. at 663.

191

133 S. Ct. 2223, 2227, 2237 (2013) (applying rule-of-reason analysis to such settlements).

192

The exact amount was tied to the future profits accruing from the patented product,

providing the generic firms some incentive to ensure that the patent holder succeeds in the mar-

ketplace. Id. at 2229.

193

Id. at 2236.

194

Id. at 2237.

195

Id.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 51 18-JUN-18 16:24

426 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

difficult for a court to determine what is given in exchange for a re-

verse payment.

Further complicating the courts’ challenge in reverse payments

cases is uncertainty about the patent’s strength. One might imagine

courts unpacking a reverse payments settlement in part by assessing

the validity of the patent. The Actavis Court indicated that ordinarily

this should not be necessary.

196

Aaron Edlin and several coauthors de-

fend Actavis on the ground that the relevant question is not whether a

patent in fact would be upheld, but the ex ante probability that it

would be upheld.

197

They suggest that “[a] large and unexplained pay-

ment is a strong signal that the patent holder had substantial doubts

that it would win the underlying patent litigation.”

198

Still, Edlin et al.

acknowledge that there may be situations in which the intent behind a

payment is difficult to assess.

199

Joshua Fischman, meanwhile, highlights a complication in assess-

ing reverse payments: The courts wish to use such settlements as re-

flecting a prediction in patent litigation, but such settlements will also

be anticipating how the courts in antitrust cases will evaluate the set-

tlements.

200

The Supreme Court, Fischman notes, seems to assume

that parties “are evidently sophisticated enough to generate a reliable

prediction about the outcome of the patent litigation, yet they are

completely na¨ıve about the potential for antitrust liability.”

201

When

the courts assess settlements made in the shadow of law, there are

feedback effects that complicate the analysis.

202

Fischman outlines

how the courts might apply game theory to generate valid inferences

based on the terms of settlements, but he acknowledges that this

would face serious challenges.

203

What neither Edlin et al. nor Fisch-

man highlight is that all of this complexity arises because disparate

issues are settled together.

The applicability of a bifurcation regime to Hatch-Waxman litiga-

tion would be straightforward. Any portion of a settlement defining

196

Id. at 2236.

197

Aaron Edlin et al., The Actavis Inference: Theory and Practice, 67 R

UTGERS

U. L. R

585, 617–19 (2015).

198

Id. at 618.

199

See id. at 599–60 (discussing situations in which a settlement may also cover unrelated

patent litigation).

200

See Joshua B. Fischman, The Circular Logic of Actavis, 66 A

. U. L. R

. 91, 133

(2016).

201

Id. at 132.

202

See id. at 131–32.

203

Id. at 134–36.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 52 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 427

the degree of delayed entry would be bifurcated from any other terms.

This would make it impossible for the parties to trade entry delay for

cash and would make it impossible for trade-offs to be disguised by

side deals. A generic drug manufacturer would be hesitant to agree to

a short delay period in the hope of receiving cash because the brand-

name manufacturer could renege on the cash component of a deal,

and the brand-name manufacturer would not want to pay cash if the

generic-drug manufacturer could still insist on immediate entry into

the market.

Of course, it is not hard to figure out that making side deals in a

Hatch-Waxman settlement agreement voidable would produce a solu-

tion to the reverse payments problem,

204

though a number of other

more complex approaches have been suggested.

205

But the example

highlights that the possibility of bifurcating settlements ordinarily re-

ceives no consideration because it is not the general approach to set-

tlement negotiations. Legislators ought to consider the possibility of

bifurcating settlements in any context in which a primary purpose of

litigation is to protect the rights of third parties. The law should focus

on ensuring that the party whose interests are to be aligned with those

of the public negotiates at arm’s length on the remedy that affects the

public. Even if there are other issues that must be addressed or that

legitimately may be addressed in a settlement agreement, those issues

can be addressed separately, and a rule preventing multiple remedies

from being combined into a single agreement may be helpful.

Such a rule could be applicable in intellectual property litigation

not involving reverse settlements. Daniel Crane has pointed out that

the trade-offs the settling parties make in reverse payments cases are

in fact present, though less obviously, in many other cases.

206

All pre-

expiration patent settlements involve damages as well as the question

of whether the defendant will exit the market,

207

and so even when the

defendant pays money to the plaintiff rather than the other way

around, the settlement might be different from probabilistic expecta-

tions at trial. The parties have an incentive to make the monetary pay-

204

Hemphill recognizes that barring contemporaneous agreements on delayed entry and

side deals would address the reverse payments problem: “[T]he FTC could set a rule stating that

any conferral of value by a brand-name firm, if made contemporaneously with a generic firm’s

agreement to delay entry, will be considered to exchange payment for delay.” Hemphill, supra

note 9, at 686. R

205

See Hovenkamp et al., supra note 186, at 1756–57 (summarizing proposals). R

206

See Daniel A. Crane, Ease over Accuracy in Assessing Patent Settlements, 88 M

INN

. L.

. 698 (2004).

207

Id. at 700.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 53 18-JUN-18 16:24

428 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

ment less, in exchange for a faster exit from the market,

208

so that they

can mutually benefit from higher prices. For example, parties could

settle a claim in favor of one party and a counterclaim in favor of the

other in order to split markets. Crane points out that “[s]imilar issues

could arise in litigation over covenants not to compete, territorially

limited franchises, trademarks, tortious interference with contracts, or

various kinds of unfair competition or misrepresentation claims.”

209

central goal of each of these areas of law is to protect consumers, even

though the cases are litigated by rights holders and their competitors.

A severable settlements regime could serve as a simple but effec-

tive tool for preventing settling parties from making trade-offs in

favor of market exclusion in any or all of these areas. But would it

pass cost-benefit analysis? The benefit is difficult to quantify because

it would vary from one context to another. The principal cost is that

cases might be more difficult to settle, even if bifurcating settlements

in a context not involving third parties would not increase trial

rates.

210

When settling parties can make trade-offs at the expense of

consumers, then consumers are effectively subsidizing settlement. Par-

ties who go to trial not only face the expense of doing so but also lose

their opportunity to make a trade-off at the consumers’ expense. Once

a case is tried and a final judgment is entered, a payment by one party

to the other to exit a market in which it is legally entitled to be would

arouse considerable antitrust scrutiny. The same payment made in set-

tlement will generally not receive scrutiny. Bifurcating settlements

would eliminate this asymmetry, though the asymmetry could be justi-

fied by the social goal of promoting settlement.

The same logic applies in the case of reverse payment settle-

ments, though the argument for bifurcating settlements may be

greater there if the consumer costs are especially high. But a parallel

to the cost-benefit analysis above can be seen in one argument for

efficiency of reverse payment settlements: allowing a patent holder to

resolve litigation relatively inexpensively increases the returns on

holding a patent and thus the incentive to innovate.

211

Thomas Cotter

208

Id. at 700–01.

209

Id. at 702.

210

See supra Part I (assessing the effect of bifurcation on trial rates).

211

See, e.g., Thomas F. Cotter, Refining the “Presumptive Illegality” Approach to Settle-

ments of Patent Disputes Involving Reverse Payments: A Commentary on Hovenkamp, Janis &

Lemley, 87 M

INN

. L. R

. 1789, 1809 (2003) (“Restricting the parties from settling on terms that

involve reverse payments would decrease the value of at least some valid and infringed pharma-

ceutical patents . . . .”). A complementary argument is that allowing a generic-drug manufacturer

to avoid high litigation costs can increase the incentives to file an Abbreviated New Drug Appli-

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 54 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 429

points out that patentees might pay to settle patent litigation that has

little chance of succeeding,

212

thus reducing the litigation costs associ-

ated with patent ownership. Herbert Hovenkamp, Mark Janis, and

Mark Lemley suggest that only payments in excess of the expected

cost of litigation should be presumptively illegal.

213

The severable-settlements analysis highlights the defense of re-

verse payment settlements: it may be worthwhile to sacrifice con-

sumer welfare in the short term to encourage settlement of litigation.

There are many contexts in which society would not allow this, how-

ever. Apple and Samsung could not settle their smartphone lawsuits in

part by agreeing to raise prices. If reverse payment settlements are

justifiable, the justification must be that the damage to consumers is

relatively small, that litigation costs are particularly pernicious when

they serve as a tax on innovation, that those costs are especially high

for the type of suits at issue, or that the reverse payments are espe-

cially likely to be helpful in promoting settlement. Indeed, those who

defend reverse patent settlements are generally careful to point out

that the efficiency issue is highly case specific: Henry Butler and Jef-

frey Jarosch argue that the courts should apply a rule-of-reason anti-

trust analysis rather than a per se ban or some other presumption.

214

Such an analysis would focus on factors such as the generic manufac-

turer’s entrance date specified in the settlement

215

and the relative size

of the reverse payment.

216

But even if reverse payments should be analyzed on a case-by-

case basis, bifurcation might be beneficial. It will still be difficult for

courts to unpack any particular settlement to determine how strong

the parties believed the patent to be and whether the reverse payment

was necessary to help settle the case. Negotiations could be structured

to make it easier for the court to perform this analysis. For example,

the parties might negotiate along three dimensions: (1) a patentee

cation and thus challenge the patent in the first place. See, e.g., Asahi Glass Co. v. Pentech

Pharm., Inc., 289 F. Supp. 2d 986, 994 (N.D. Ill. 2003) (“A ban on reverse-payment settlements

would reduce the incentive to challenge patents by reducing the challenger’s settlement

options . . . .”).

212

Cotter, supra note 211, at 1807. R

213

Hovenkamp et al., supra note 186, at 1759–61; see also Elhauge & Krueger, supra note R

19, at 304. R

214

See Henry N. Butler & Jeffrey Paul Jarosch, Policy Reversal on Reverse Payments: Why

Courts Should Not Follow the New DOJ Position on Reverse-Payment Settlements of Pharmaceu-

tical Patent Litigation, 96 I

OWA

L. R

. 57, 62 (2010). This position prevailed at the Supreme

Court. See FTC v. Actavis, Inc., 133 S. Ct. 2223, 2237 (2013).

215

Butler & Jarosch, supra note 214, at 116–17. R

216

Id. at 117–18.

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 55 18-JUN-18 16:24

430 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

case-strength dimension; (2) a table translating this strength dimen-

sion into the amount of time that the generic challenger would need to

wait to enter the market if reverse payments were prohibited; and

(3) a table translating this time delay into (a) reverse payments from

the patentee to the generic manufacturer and (b) increases in the

amount of time that the generic challenger must wait to enter. The

strength dimension would be negotiated after dimensions two and

three, and resolution of that dimension would conclude the

settlement.

This differs from other severable-settlement approaches in that

two of the dimensions are functions of the third, rather than

standalone components of the settlement. This approach makes settle-

ment more complicated but greatly facilitates a court’s job in assessing

a settlement by isolating the legally questionable but potentially justi-

fiable trade-off between payments and entry time. The parties will

have appropriate negotiation incentives with respect to the first and

second dimensions, so the court need not focus on them. The parties

might have an incentive to provide large reverse payments along di-

mension (3)(a) in exchange for substantial delays in entry under

(3)(b), but this approach will make the trade-off apparent so that the

court can consider it. At the same time, the court would be able to

consider whether the total amount of reverse payments was justified

in light of the expected litigation costs as of particular dates.

This approach is sufficiently alien to the United States’ litigation

system that we do not anticipate Congress mandating this approach

anytime soon. Nor do we take a position on whether the benefits

would outweigh the costs. But the exercise illustrates that settlements

can be disaggregated into components—some of them conditionals

with other components as their antecedents—that are analytically

distinct.

C. Protecting Class Members: Attorneys’ Fees vs. Damages

There is one area in which the danger of the negotiating parties’

reaching a settlement to the detriment of third parties is so significant

that the possibility of bifurcating settlements has received considera-

ble attention: class action litigation.

217

The courts’ analysis of the bene-

217

See, e.g., William D. Henderson, Clear Sailing Agreements: A Special Form of Collusion

in Class Action Settlements, 77 T

. L. R

. 813, 814–17 (2003) (arguing that courts should adopt

a per se rule rejecting clear sailing agreements in class action litigation, encouraging plaintiffs’

attorneys to focus solely on the benefit to the class and then petitioning the court for attorneys’

fees under the common fund doctrine after the settlement process).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 56 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 431

fits and costs, however, has been incomplete, and commentators have

not considered alternative approaches to structuring severable settle-

ments. A consideration of the caselaw is useful both as an indication

of what the likely objections to bifurcation are and as a means of dem-

onstrating how an appreciation of the theoretical case for bifurcation

can address concerns. The analysis also extends easily to other con-

texts, such as shareholder derivative actions.

218

Under federal law, a class action settlement must be approved by

the court.

219

One reason for this requirement is that the courts wish to

ensure that settlements truly represent the interests of the class mem-

bers.

220

The concern is that the class lawyers might treat themselves as

the real parties in interest, and they may sacrifice class members’ in-

terests to their own advantage.

221

In settling a class action, class law-

yers may be especially concerned about their own fees, and they may

be willing to accept a lower award for the class members in exchange

for higher fees.

222

The defendant’s sole concern is the total amount

that it must pay,

223

and so it will generally be willing to trade higher

fees for an even greater reduction in damages payments to the class.

The result is that, absent bifurcation of settlements, class action settle-

ments will tend at least somewhat to have lower damages and greater

fees than would exist if the class lawyers were negotiating solely with

the interests of the class in mind. The requirement of judicial approval

limits the ability of the negotiators to take advantage of the class, but

courts still sometimes approve settlements that include high attor-

neys’-fee awards and no or minimal value for class members.

224

218

In Delaware, judicial approval of derivative suit settlements is required for much the

same reason that judicial approval of class actions is required. “This rule is designed to prevent

‘private settlements of representative litigation,’ whereby the defendant ‘buys off’ the derivative

plaintiff.” Edward Tsai, Success by Another Name: Recognizing a Limited Exception Under Del-

aware Law to the Indemnification of Derivative Action Settlements, 64 N.Y.U. A

. S

URV

. A

L. 879, 912 (2009) (footnote omitted). Some commentators, however, believe that “Delaware

courts are likely to approve settlements and corresponding awards of attorneys’ fees in exchange

for release of even meritless claims.” Id. at 915.

219

. R. C

. P. 23(e).

220

See In re Gen. Motors Corp. Pick-Up Truck Fuel Tank Prods. Liab. Litig., 55 F.3d 768,

785 (3d Cir. 1995).

221

Id. at 788.

222

See, e.g., Susan P. Koniak & George M. Cohen, Under Cloak of Settlement, 82 V

. L.

. 1051, 1056 (1996) (“We agree with those who argue that lawyer abuse in class actions is

rampant and that the current system, far from keeping this abuse in check, is set up to shield

lawyers from the consequences of their misdeeds.”).

223

Henderson, supra note 217, at 820 (“[A] defendant who has settled a class action lawsuit R

is ultimately indifferent to how a single lump-sum payment is apportioned between the plaintiff’s

attorney and the class.”).

224

For a recent example, see Lane v. Facebook, Inc., 696 F.3d 811, 817 (9th Cir. 2012),

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 57 18-JUN-18 16:24

432 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

In response to this concern, at least one court has held that attor-

neys’-fee negotiations must be bifurcated from negotiation of other

issues. In Prandini v. National Tea Co.,

225

the court rejected a settle-

ment in which bifurcation had not occurred.

226

“Only after court ap-

proval of the damage settlement,” the court held, “should discussion

and negotiation of appropriate compensation for the attorneys be-

gin.”

227

Other courts, however, have not followed Prandini.

228

A Cali-

fornia state court, for example, encouraged but refused to mandate a

bifurcated settlement procedure.

229

Similarly, the Ninth Circuit re-

quires enhanced scrutiny in cases in which attorneys’ fees and dam-

ages are negotiated simultaneously,

230

and the Texas Supreme Court

has emphasized the importance of providing notice to class members

about how fees are to be calculated, coupled with a judicial assess-

ment of the reasonableness of fees.

231

Even in courts nominally follow-

ing Prandini, parties sometimes evade the rule by agreeing to a “clear

sailing agreement,” under which the defendant promises not to con-

test whatever fees the plaintiff lawyers seek up to some level.

232

The U.S. Supreme Court has not explicitly assessed the merits of

bifurcating settlements for class actions, but it has twice indicated

skepticism about requiring severance in a related context: fee determi-

nations under the Civil Rights Attorney’s Fees Awards Act.

233

In a

1982 case, the Court recognized the possibility that negotiators might

make trade-offs to the detriment of the plaintiff,

234

but refused to re-

quire rejection of fee awards made as part of a broader settlement. In

a footnote, the Court worried that such a requirement would prevent

approving a class action settlement in which damages were awarded to a new charity organiza-

tion whose mission related to the plaintiffs’ complaints on a cy pres theory. Judge Kleinfeld

dissented, stating, “This settlement perverts the class action into a device for depriving victims of

remedies for wrongs, while enriching both the wrongdoers and the lawyers purporting to re-

present the class.” Id. at 826 (Kleinfeld, J., dissenting); see also Adam Liptak, When Lawyers Cut

Their Class-Action Clients Out of the Deal, N.Y. T

IMES

, Aug. 13, 2013, at A12.

225

557 F.2d 1015 (3d Cir. 1977).

226

Id. at 1017.

227

Id. at 1021.

228

For a review of different courts’ approaches, see David Brainerd Parrish, Comment, The

Dilemma: Simultaneous Negotiation of Attorneys’ Fees and Settlement in Class Actions, 36 H

OUS

L. R

. 531 (1999).

229

See Ramirez v. Sturdevant, 26 Cal. Rptr. 2d 554, 564 (Ct. App. 1994).

230

See, e.g., Mendoza v. United States, 623 F.2d 1338, 1353 (9th Cir. 1980).

231

Gen. Motors Corp. v. Bloyed, 916 S.W.2d 949, 957–59 (Tex. 1996).

232

Henderson, supra note 217, at 820–21 (discussing such agreements). R

233

42 U.S.C. § 1988(b) (2012).

234

White v. N.H. Dep’t of Emp’t Sec., 455 U.S. 445, 453 n.15 (1982) (summarizing the

argument that there exists “an inherent conflict of interest between the attorney and client”).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 58 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 433

a defendant from being sure of “his total liability from both damages

and fees.”

235

The Court similarly worried about uncertainty four years

later, concluding, “It is therefore not implausible to anticipate that

parties to a significant number of civil rights cases will refuse to settle

if liability for attorney’s fees remains open, thereby forcing more cases

to trial, unnecessarily burdening the judicial system, and disserving

civil rights litigants.”

236

Earlier, we showed analytically that bifurcating settlements can

increase the likelihood of trial, though for different reasons and only

in some circumstances.

237

The Court’s concern that defendants will

settle cases only if they can settle the whole case at once is misplaced

or at least exaggerated. The Court imagines defendants who would be

so concerned about achieving certainty that they would refuse to set-

tle what would ordinarily be the major issues—liability and dam-

ages—because there would remain some residual possibility that the

parties subsequently might be unable to resolve a more minor issue. It

might well make sense for defendants to refuse to settle on the merits

unless they can also resolve attorneys’ fees in a regime, like the pre-

sent one, in which they are permitted to resolve all issues together.

But it would be odd for a defendant concerned about litigation risk to

refuse to take the one step that would greatly reduce such risk and

open the possibility of eliminating it on the ground that this possibility

was not a certainty.

238

There is a reason that settlements on the merits will be less likely

when the issue of attorneys’ fees is deferred: The negotiators’ ability

to make trade-offs to the detriment of the virtually represented plain-

tiffs increases the gains to trade that settlement creates. The class law-

yer and defendant know that if they can reach a settlement, they can

in effect take money from a third party. Yet the Court’s general con-

cern about conflicts suggests that the Court would not endorse con-

flicts on the ground that they facilitate settlement. Placing this

argument aside, settlements on the merits seem more likely because

there are fewer issues that must be resolved. To be sure, there may be

fewer settlements on the attorneys’-fee question itself, but this is a

lesser concern than would be a reduction in settlements on the merits.

235

Id. at 454 n.15. The Court concluded, “Although such situations may raise difficult ethi-

cal issues for a plaintiff’s attorney, we are reluctant to hold that no resolution is ever available to

ethical counsel.” Id. The phrase “reluctant to hold” suggests that the Court’s view was tentative.

236

Evans v. Jeff D., 475 U.S. 717, 736–37 (1986) (footnote omitted).

237

See supra Part I.

238

Defendants have sometimes insisted on global settlements in contexts in which they

faced bankruptcy. See, e.g., Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 597 (1997).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 59 18-JUN-18 16:24

434 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

Adjudicating attorneys’ fees need not require a full exploration of all

issues that would be considered at trial on the merits, and also will

likely be cheaper because the total stakes are lower. Thus, concerns

about bifurcation should be less in this context than in some others,

such as the compensatory-punitive damages context,

239

where the is-

sues that must be tried if either part of a settlement fails are mostly

the same.

We suspect that most courts’ reluctance to insist on bifurcating

settlements in this context, despite the clear acknowledgment that

class lawyers may be sacrificing plaintiffs’ interests for their own,

240

may reflect a discomfort with restriction of the settlement process.

This may in part be attributable to the general view that settlement is

beneficial because it saves resources, but also may simply be a reflec-

tion of what has long been understood as the legal order: parties are

usually free to enter into whatever settlements they like, and where

judicial supervision is necessary to protect nonparties, judges should

simply examine whether settlements are fair, rather than insist on a

particular procedure designed to ensure fairness. Indeed, there is a

legitimate question whether courts, which traditionally have simply

examined settlements for fairness, should decree that settlements that

were not bifurcated are invalid. Perhaps a bifurcation regime should

be created only by the legislature.

Yet it is also possible for parties voluntarily to agree to a bifur-

cated settlement. Certainly, where parties have bifurcated settlement,

that should be given great weight in a court’s assessment of whether a

settlement is fair. If a plaintiff and a defendant have independently

agreed to damages and attorneys’ fees instead of combining them into

a single settlement, that procedural protection should, absent special

circumstances such as repeat-player interaction among the attorneys,

provide enough evidence of the settlement’s fairness. If courts began

to give credit for voluntary bifurcation, they might also begin to feel

more comfortable deducting credit in cases in which there is no bifur-

cation. Indeed, courts already apply enhanced scrutiny in such

cases,

241

and so perhaps all that is needed is for the courts to place

even greater weight on this structural consideration than they already

do. Our hope is that our analysis of bifurcation in general can help

normalize bifurcation so that it will seem less strange to judges. Judges

239

See supra Section III.A.

240

See Koniak & Cohen, supra note 222, at 1056. R

241

See Parrish, supra note 228, at 542–48. R

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 60 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 435

might then feel more comfortable giving greater scrutiny to settle-

ments reached without bifurcation.

At the same time, a recognition in general of the possibility of

bifurcating settlement can lead to more creativity in applying it to ad-

dress concerns. If, for example, the Court were correct that the bifur-

cated-settlements regime it was considering would frustrate

defendants eager for complete resolution, severable settlements might

be preferable. With severable settlements, parties would negotiate

both issues at once. There would be some danger that the other side

would renege, perhaps pretending to be satisfied with one part of a

settlement because it was very happy with the other, but this seems

unlikely given a tentative agreement following a negotiation in which

each side has an incentive to make the best argument it can as to each

issue. Neither side would have an incentive to propose a trade-off

across issues because both parties would recognize that such a settle-

ment would not be stable. So, the parties would work to ensure that

they had arrived at both damages and attorneys’-fees settlements that

were independently mutually agreeable. Conditional on arriving at an

agreement, severable settlements will be more likely to produce a

complete agreement than the first half of a seriatim bifurcated settle-

ment where the second half has not yet been negotiated.

Another possible structure would be seriatim bifurcated settle-

ments with the order of issues reversed. It might be undesirable to

produce a final determination of attorneys’ fees before damages be-

cause it is difficult to assess the value that attorneys have provided

before the case is finished and because once their fees are established,

attorneys might have an incentive to shirk their responsibilities. It

should, however, be possible to produce a schedule of fees dependent

on factors such as the total recovery of the plaintiff class, the total

number of hours worked, and the stage of litigation reached. A court

would need to consider this somewhat critically because of the danger

that the plaintiffs’ attorneys and defendants would agree to a rela-

tively flat fee schedule that would give plaintiffs little incentive to seek

out high damages. But, at least if the fees are to be paid in addition to

damages rather than out of damages, the court would need to focus

only on the difference in fees for different levels of work and success,

rather than the absolute level, because it could be confident that the

plaintiffs’ attorneys would want damages as high as possible and the

defendant would want damages as low as possible. Even if the fees are

to be deducted from damages, the judicial inquiry may be far easier

than a fairness assessment of an ex post fee award. In that situation,

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 61 18-JUN-18 16:24

436 THE GEORGE WASHINGTON LAW REVIEW [Vol. 86:376

the court may have little ability to determine whether the attorneys

had sufficient incentive to fight hard for their clients because it may

have difficulty establishing whether the attorneys thought that they

would make more with greater success.

242

Our purpose is not to argue that severable settlements would be

superior to bifurcated settlements in this context or that resolving at-

torneys’ fees first is necessarily desirable in a regime of bifurcated set-

tlement. Nor do we contend that bifurcating is the best or sole

solution. William Henderson, for example, argues that a guardian ad

litem be appointed to represent the class in the negotiation of attor-

neys’ fees after the settlement on the merits.

243

The guardian ad litem

would then receive a small percentage of the settlement as well.

244

This is another strategy for addressing settlement trade-offs that are

adverse to third parties, and one could also imagine using it in other

contexts, for example by requiring a government lawyer to approve

allocation of damages as compensatory and punitive, or giving repre-

sentatives of consumer groups an explicit seat at the bargaining table

in antitrust negotiations. Meanwhile, there are other strategies that

could be used as alternatives or complements to bifurcation in the at-

torneys’ fees context, such as deeper judicial review or stronger en-

forcement of ethical rules.

245

Our claim is simply that bifurcated

settlements should be on the regulatory menu. It should not be viewed

reflexively as a mechanism that threatens litigant autonomy and the

prospect of settlement.

ONCLUSION

Bifurcating settlements reduces the chance that litigants will

make trade-offs across different components of a settlement. This may

be beneficial when these trade-offs come at the expense of third par-

ties’ interests—for example, the interests of children in being placed

with a parent best able to provide care and still able to provide finan-

cial support, the government in taxing punitive damages, consumers in

low prices, or class members in the allocation of settlement proceeds

between them and their attorneys. By treating these seemingly dispa-

242

Henderson, supra note 217, at 820–21. R

243

Henderson argues that “the specter of a court-appointed advocate challenging the attor-

neys’ fee petition will most likely have a significant impact on bargaining dynamics and trade-

offs proposed by the defendant and class counsel.” Id. at 837.

244

Id. at 831.

245

See, e.g., Arthur B. LaFrance, Public Interest Litigation, Attorneys’ Fees, and Attorneys’

Ethics, 16 E

NVTL

. L. 335, 338–39 (1986) (discussing the ethical dilemmas that defense counsel

face when settlement negotiations are not bifurcated).

\\jciprod01\productn\G\GWN\86-2\GWN202.txt unknown Seq: 62 18-JUN-18 16:24

2018] BIFURCATING SETTLEMENTS 437

rate contexts together, we have aimed to provide a unified theoretical

framework for assessing bifurcation. This framework helps establish

not only when bifurcation may be useful, but also the trade-offs that

exist in different approaches to structuring bifurcation. Our broader

aim is to counter the sense that many lawyers have that there is some-

thing inherently wrong with bifurcation. While bifurcation has not yet

found a foothold in the legal system, there are sound reasons for in-

sisting on it when trade-offs at the expense of third parties are a dan-

ger, and there is little reason to believe that bifurcation will have a

significant adverse impact on the prospects for settlement. Still, as our

analysis of child custody demonstrates, the benefits and costs of bifur-

cation are highly context specific, and further experiments and study

in particular areas may help determine whether bifurcation is war-

ranted and, if so, how it might be structured.