Revisiting the Debate over Attorneys' Contingent Fees: A ... - CiteSeerX

Revisiting the Debate over Attorneys’ Contingent Fees: A Behavioral Analysis Eyal Zamir and Ilana Ritov

ABSTRACT

Building on Kahneman and Tversky’s prospect theory, this paper presents a series of experiments designed to reveal people’s preferences regarding attorneys’ fees. Contrary to common economic wisdom, it demonstrates that loss aversion (rather than risk aversion or incentivizing the lawyer to win the case) plays a major role in clients’ preferences for contingent-fee arrangements. Facing a choice between a mixed gamble and a pure positive one, plaintiffs prefer a contingent fee (framed as a pure positive gamble), even if it yields an expected fee that is 2 or 3 times higher than a noncontingent one (framed as a mixed gamble). At the same time, defendants, who face a choice between two pure negative gambles, are typically risk seeking and prefer fixed fees. Our findings indicate that information problems and lack of alternative fee arrangements probably do not loom large in clients’ choice of fee arrangement. We discuss the policy implications of our findings.

1. INTRODUCTION

Contingent-fee (CF) arrangements are the standard method of financing civil litigation in several types of suits, including personal injuries, colEYAL ZAMIR is Augusto Levi Professor of Commercial Law, Hebrew University of Jerusalem. ILANA RITOV is Professor at the School of Education and the Center for the Study of Rationality, Hebrew University of Jerusalem. We would like to thank Maya Bar-Hillel, Jonathan Baron, Lester Brickman, Yuval Feldman, Chris Guthrie, Ehud Guttel, Assaf Hamdani, Michael Karayanni, Roy Kreitner, Herbert Kritzer, Daphna Lewinsohn-Zamir, David Luban, Barak Medina, Ittai Paldor, Ariel Porat, Eric Posner, Jeffrey Rachlinski, Doron Teichman, Avishalom Tor, David Weisburd, and Kathryn Zeiler for invaluable comments on earlier drafts. Special thanks are due to Asaf Posner for long discussions of the practicalities of the market for legal services in Israel and to Shirley Amsterdam, Roee Ben-Ari, Yoel Ben Or, Netta Corren, Efrat Hurvitz, Marina Motsenok, Maya Shaton, and Gil Tessler for excellent research assistance. Finally, we are grateful for the financial support of the

[Journal of Legal Studies, vol. 39 (January 2010)] 䉷 2010 by The University of Chicago. All rights reserved. 0047-2530/2010/3901-0007$10.00

245

246 / T H E J O U R N A L O F L E G A L S T U D I E S / V O L U M E 3 9 ( 1 ) / J A N U A R Y 2 0 1 0

lection of commercial and retail accounts, and class actions. Under such arrangements, the attorney’s fee is contingent on the success of the claim, which is calculated as a certain percentage of the amount recovered and paid on recovery. A clear advantage of a CF, in comparison with hourly based fee and fixed fee (FF) arrangements, is that it enables plaintiffs of limited financial means to secure otherwise unaffordable legal services (Moliterno 2009, pp. 1300–1306). It increases access to the courts, thereby empowering the underprivileged and strengthening the legal incentives for people to internalize the costs of their injurious behavior (Rickman 1994, pp. 39, 45–47; Schneyer 1998, p. 376). Nevertheless, the desirability, efficiency, and fairness of CFs are hotly debated. Advocates of CFs point out that this arrangement provides clients with credit (since the fee is paid only on recovery) and with a sort of insurance (since the fee is not paid if the claim fails). The CF induces attorneys not to take cases whose expected value is too small, thus saving their clients and the court system the costs involved in pursuing such claims. It also incentivizes the attorney to win the case or attain a favorable settlement and at the same time to avoid investing too much time in handling it (a likely negative effect of hourly fees). Given these additional benefits to the client, it is argued that the effective hourly rate of CFs (the expected fee divided by the number of hours the lawyer works on the case) is very reasonable compared with prevailing hourly rates (Kritzer 1990, pp. 137–43; 2004, pp. 180–218). Critics of current CF practices discard this rosy picture. They argue that the insurance element of CF arrangements is illusory since there is typically no real risk involved in cases that attorneys take on a CF basis (Brickman 1989). Similarly, the credit element should not add much to the fee because the attorney enjoys an excellent means of collection: she deducts her fee from the award before forwarding it to her client. Current CF rates, so it is charged, reflect various market failures, including plaintiffs’ inability to assess the value and prospects of their case and the scope of work involved in handling it, lawyers’ uniform pricing practices, the absence of price advertisements, clients’ prohibitive search costs, and prohibitions against the purchase of tort claims and the brokerage of lawyers’ services (Brickman 2003c). Consequently, CF rates are often disproportionately high. Aware that a large share of the award goes to the attorney, courts and juries increase the compensation they grant Milton and Miriam Handler Foundation and the U.S.–Israel Bi-national Science Foundation.

CONTINGENT FEES /

247

plaintiffs. The debate is thus closely connected to broader issues, such as the alleged litigation explosion (Brickman 1994; Miceli and Segerson 1991), the calls for a tort reform, lawyers’ ethics (Brickman 1996a, 1996b, 2003b), and even (indirectly) defensive medicine and its implications for the American health care system (Dwyer 2006). The CF debate revolves around factual, positive, and normative issues. Hundreds of studies have offered conflicting accounts of the market for legal services, competing explanations for the market’s behavior, and radically different prescriptions for its regulation. Numerous legal scholars, economists, and other social scientists have contributed to this vast body of theoretical and empirical literature. Curiously, no attempt has been made to analyze the issue from the perspective of cognitive psychology.1 This article begins to fill this gap.2 Since the outcomes of the lawyer’s work are uncertain, we suggest analyzing plaintiffs’ choice between CFs and noncontingent fees (that is, hourly fees or global FFs) as a choice between two gambles. According to rational choice theory, the choice between these two gambles depends on each gamble’s expected utility and the subject’s risk aversion. According to Kahneman and Tversky’s (1979) prospect theory, people assign value to gains and losses rather than to final assets or final positions. The value function is defined over changes from a reference point and is normally concave for gains (implying risk aversion), commonly convex for losses (risk seeking), and generally steeper for losses than for gains (loss aversion). People’s choices therefore crucially depend on the way they frame the choice and particularly on their reference point, which determines what changes are perceived as gains or losses. We argue that noncontingent fees are conventionally conceived as exposing the plaintiff to the risk of losing her case and still having to pay the attorney’s fee. It is therefore a mixed gamble in which the plaintiff may either win some gain or bear a loss. In contrast, while a CF arrangement may bring about a smaller gain in the case of success, it eliminates the risk of loss because no fee is paid to the lawyer if the case is lost. It is therefore perceived 1. An exception is Thomas (1998). Thomas argues that in high-profile cases, the attorney’s desire to avoid appearing foolish and the related phenomenon of escalating commitment may mitigate the attorney’s disincentive to invest due to the contingent-fee (CF) arrangement (on this disincentive, see Section 2). Thomas’s analysis focuses on a relatively narrow aspect of CFs, and his insight is relevant to a small minority of cases. 2. The present article is part of a larger research project in which we also study people’s judgments of the fairness of CF arrangements and their effect on prevailing CF rates.


as a pure positive gamble in which the plaintiff may either gain or break even. Rational choice theory predicts that the choice between CFs and noncontingent fees will significantly depend on the chooser’s risk aversion. Assuming diminishing marginal utility of money and standard assumptions about the shape of the utility curve, it predicts that the wealthier a person is, the less she would opt for a CF if an FF would give her a higher expected net gain (the expected award minus the attorney’s fee). Under prospect theory, as long as a person can afford to pay on either a CF or an FF basis, no such correlation between people’s affluence and their preference for CFs is necessarily predicted. Moreover, under rational choice theory, the premium that plaintiffs will be willing to pay to insure against the risk of bearing the legal costs even if their claim fails is not very high, especially if they are well-off. In contrast, prospect theory predicts that if the choice between a CF and an FF is indeed framed as a choice between a pure positive gamble and a mixed one, people will prefer a CF even if the expected fee under a CF arrangement is 2 or 3 times higher than under a noncontingent-fee arrangement. This strong preference for CFs is independent of both people’s absolute wealth and the ratio between their wealth and the cost of legal services. To test these predictions, we designed experiments in which subjects were asked to choose between a CF and an FF under various hypothetical circumstances. These experiments provide clear support for the predictions of prospect theory, thus somewhat diminishing the significance of other explanations for prevailing CF rates. As mentioned above, some commentators argue that current CF rates reflect clients’ information problems or result from the unavailability of alternative fee arrangements. Since the questions in our experiment provided the subjects with all of this information and allowed them to choose between a CF and an FF, it does not seem that their strong preference for CFs stems primarily from information asymmetry or lack of options. Adding real monetary consequences to the experiment did not change the results. Moreover, we ran the same experiment with tort lawyers who were asked to make the same choice assuming that they had suffered a loss and needed an attorney to represent them. Their choices were similar to those of laypersons. The pattern of choice was not affected by the presumably significant differences in affluence between Israeli students and practicing tort lawyers. Framework manipulations did influence the pattern of choice: when respondents were induced to perceive the choice not as one between a mixed gamble and a pure positive one but (unconven-

CONTINGENT FEES /

249

tionally) as between two pure negative gambles, the preference for FFs rose significantly. Finally, we tested the hypothesis that the preference for CFs rests on the incentive effects of CFs. To that end, we ran an experiment in which the type of fee paid by the plaintiff (either CF or FF) was decoupled from the type of remuneration paid to the attorney who handled the case, using an intermediary entity. Such decoupling did not significantly alter plaintiffs’ preferences, which thus indicates that incentives do not play a prime role in this choice. We conclude, therefore, that prospect theory provides a major descriptive explanation for plaintiffs’ preferences. Another intriguing aspect of the market for legal services is that, unlike plaintiffs, defendants only infrequently hire attorneys on a CF basis. Various economic explanations have been offered for this phenomenon, most of which focus on the different characteristics of the typical plaintiff and defendant. On the basis of experimental results, we propose a different explanation, grounded, once again, in prospect theory. We argue that from the defendant’s perspective, the decision is framed as a choice between two pure negative gambles, thus inducing risk seeking. Risk-seeking defendants prefer to assume the risk of having to pay both damages and the lawyer’s FF rather than to pay, at a minimum, a higher CF if the suit is fully dismissed. In addition to their theoretical interest, our findings are of great practical importance, as proposals to regulate CFs are repeatedly considered by legislators throughout the world. For instance, if it is true that plaintiffs’ strong preference for CFs does not stem primarily from information problems, then proposals to impose stricter precontractual disclosure duties on attorneys are somewhat misdirected. In addition, the fact that the strong preference for CFs also characterizes people who face no cash constraints is a relevant factor in considering the distributive effects of CFs and their regulation. Our findings may also have positive and normative implications for other spheres in which people face a choice between a pure positive gamble and a mixed one. People encounter such choices, for example, in financial markets.3 If people have strong preferences for pure positive gambles, then to the extent that similar choices may be framed differently, financial institutions can use 3. For general discussions of financial markets through the lens of prospect theory, see Benartzi and Thaler (1995) and Barberis, Huang, and Santos (2001). Odean (1998) uses prospect theory to analyze investors’ propensity to hold losing investments too long and sell winning investments too soon.


this phenomenon to induce investors to opt for poorer investments by framing them as pure positive gambles. Section 2 of this article reviews the ongoing debate on whether CF rates are excessively high and whether the market for legal services is competitive. Section 3 provides a brief theoretical and comparative background to our experiments. Section 4 describes the findings and normative implications of five experiments we designed to better understand why plaintiffs prefer CFs and why CF rates may be high. Section 5 turns to the defendants’ side. It critically surveys existing attempts to explain why predominantly plaintiffs use CFs, describes an experiment designed to study this issue, and analyzes the implications of the experiment. A conclusion follows.

2 . T H E D E B AT E

In the United States, the conventional flat CF rate is 33 percent of the recovery. When the rate depends on the stage to which the case arrives, it is usually 25 percent if the case does not go to trial, rising to as high as 50 percent if the case reaches the appellate court (Kritzer 2002, pp. 757–61; 2004, pp. 42–43; Brickman 2003a, p. 657). While there is a relatively broad consensus about this typical range of fees, there is disagreement regarding the ratio between ordinary hourly rates charged by lawyers and the effective hourly rates resulting from CF arrangements. Since CFs provide the client with a kind of insurance and credit, it stands to reason that in a competitive market, the effective hourly rate of a CF would be somewhat higher than the rate charged under noncontingent arrangements, such as hourly fee or FF arrangements. However, while some studies maintain that the effective hourly rates of lawyers resulting from conventional CFs are only slightly (and justifiably) higher than standard noncontingent hourly rates (Kritzer 1990, 2004), others claim that CF arrangements often result in effective rates that are several times higher than ordinary hourly rates (Brickman 2003a; Koniak 1996, p. 352). In this context, Brickman (1989, p. 92; 2003a, pp. 79–80) argues that the risk of failure borne by CF lawyers is rather low because they carefully select the cases they take. Kritzer (2004, pp. 17–19) replies that in addition to the risk of complete nonrecovery, CFs must reflect the more common risks of low recovery and spending an unexpectedly large amount of time on the case. Comparable controversies surround the element of credit, as it is argued that thanks to the simple and

CONTINGENT FEES /

251

effective means of collection—deducting the fee from the award before forwarding it to the plaintiff—it should not add much to the fee (Brickman 1989, pp. 118–9). We shall not try to resolve this basically empirical debate (although our findings may shed some light on it). Rather, following most of the literature, we shall assume that the effective hourly rate under CF arrangements is indeed significantly (or at least not insignificantly) higher than under noncontingent-fee arrangements and that this difference cannot be entirely attributed to the insurance and credit elements of CFs (Brickman 2003c; Gross 2006, p. 138; Koniak 1996, p. 352; Metzloff 1991, pp. 100–101). Even Kritzer’s empirical study (2002, pp. 761–68), while estimating that the median effective hourly fee resulting from CFs in early 1990s Wisconsin ($132) was only marginally higher than ordinary hourly fees ($124), found that the mean effective hourly fee ($242) was almost twice as large as the ordinary hourly fee. On the basis of these data and data on federal cases, Kritzer (2004, p. 218) points out that the large gap between the mean and median effective rates comes from the top 10 percent of cases that produce the largest profits for CF lawyers. He estimates that, excluding those 10 percent, an attorney may expect a fee premium of 25–30 percent compared with what hourly rates generate. The premium is much higher once we take into account all cases, including the most profitable ones. Two additional explanations (beside the insurance and credit elements of CF arrangments) have been offered for this difference. One explanation rests on standard market failure analysis. It asserts that, although there are many lawyers willing to represent tort and other claimants, the market is not competitive because of clients’ acute information problems (regarding the expected reward, the risk involved in the suit, the quality of legal services provided, the time required to handle the case, and so on) and various means devised by lawyers to inhibit competition in this sphere (Brickman 2003c, pp. 76–105; Gross 2006, pp. 136–40; Painter 1995, pp. 653–68; Drummonds 1993, pp. 891–96). An alternative explanation does not rest on any market failure but rather on the dual function of any fee arrangement: rewarding the attorney for her time and efforts and creating an optimal incentive for the use of her time and skill. If a lawyer is paid for her time regardless of the outcomes of her services, she may file a legal suit even when it is not in the best interest of the client, spend too much time on the case, and reject attractive settlement offers (Clermont and Currivan 1978, pp. 534–43; Helland and Tabarrok 2003). To create perfect incentives for


the lawyer in all of these regards, lawyers should presumably be allowed to purchase the client’s cause of action for a fixed sum (Rickman 1994, pp. 36–38), or more precisely, the fraction of the cost of litigation borne by the lawyer should equal the fraction of the award or settlement that she obtains (Polinsky and Rubinfeld 2003). If the lawyer receives the entire recovery, she has no reason to spend too much or too little effort in handling the case. However, such arrangements are deemed unethical and illegal (Bond 2002; Moliterno 2003) and in any case would raise serious practical problems of their own (Clermont and Currivan 1978, p. 596). Standard CF arrangements do not perfectly align the attorney’s interests with those of the client, as they induce the attorney to accept settlement offers that are not necessarily in the client’s best interest (Schwartz and Mitchell 1970, pp. 1133–36; Miller 1987; Thomason 1991), yet the higher the lawyer’s share of the recovery (through CFs), the more their interests converge. Clients who wish to maximize their expected net recovery should therefore agree to a CF that is higher than the one reflecting the competitive zero-profit rate, thus creating economic rents for attorneys (Santore and Viard 2001; Hay 1996). While both of these explanations sound reasonable, and the latter was recently supported by a laboratory experiment (McKee, Santore, and Shelton 2007), they are also subject to critique.4 It is not, however, our goal to evaluate these explanations in detail. Rather, we strive to enrich the discussion by offering an additional explanation for why clients agree to pay high CFs. Before turning to describe our experiments, the next section provides a brief theoretical background on the pertinent behavioral theory and a short account of CFs in Israel, where our experiments were conducted.

3 . T H E O R E T I C A L A N D C O M PA R AT IV E B A CK G R O U N D

3.1. Rational Choice Theory versus Prospect Theory

Standard economic analysis, including economic analysis of law, is based on the rational choice theory of human behavior (Korobkin and Ulen 4. For instance, these explanations tend to downplay lawyers’ nonselfish motivations, the importance of reputation, and clients’ ability to know whether the attorney spends appropriate time and effort on their case (Kritzer 1990, pp. 108–11). Furthermore, an empirical study of the amount of time lawyers spend on cases taken on different fee bases reveals a much more nuanced picture than what abstract economic models predict (Kritzer 1990, pp. 111–34). On the role of reputation, see also Kritzer (2004, pp. 219–52).

CONTINGENT FEES /

253

2000). According to this theory, from among the available options, people choose the one that maximizes their expected utility. People’s preferences and choices are assumed to be rational in the sense that they comply with several basic axioms, such as completeness, transitivity, and invariance. Standard rational choice theory also makes important assumptions regarding people’s attitudes to risk. According to the law of diminishing marginal utility, the utility one derives from any additional unit of a good or service, including money, is smaller than the utility one derives from the previous unit. While risk aversion depends on personal inclinations, it is assumed that as a general rule the concavity of people’s utility curve is decreasing and thus the wealthier the individual, the smaller his risk aversion.5 Kahneman and Tversky (1979) proposed a competing, descriptive theory of people’s preferences and choices under risk, the prospect theory (see also Tversky and Kahneman 1992). This theory comprises several elements, all of which violate the axioms of rational choice theory. First and foremost, prospect theory posits that people ordinarily perceive outcomes as gains and losses rather than as final states of wealth or welfare. Gains and losses are defined relative to some reference point. The value function is normally concave for gains (implying risk aversion), commonly convex for losses (reflecting risk seeking), and generally steeper for losses than for gains (indicating loss aversion).6 People’s choices therefore crucially depend on the way they frame any choice. In particular, their reference points determine what changes are perceived to be gains or losses. Ordinarily—but not invariably—people take the status quo as their reference point (Tversky and Kahneman 1991).7 In the years following the publication of the prospect theory, research was conducted on loss aversion that yielded seminal studies of its role 5. On risk aversion, see generally Shavell (1987, pp. 186–205). 6. The subjective value function proposed by prospect theory has the following form: v(x) p xa if x x 0; ⫺l(⫺x)b if x ! 0. The modeling of value in prospect theory allows for the nonlinearity of the value function around the reference point. The parameter l is referred to as the coefficient of loss aversion. A l greater than one implies that the same absolute difference in value will be weighted more heavily when it is in the loss domain than when it is in the gains domain. Kahneman and Tversky estimated a and b to be .88 and l to be 2.25. 7. Ko¨szegi and Rabin (2006) propose a model taking people’s expectations as the pertinent reference point. Prospect theory also posits that people’s risk aversion in the domain of gains and risk seeking in the domain of losses are reversed for low-probability gains and losses (Tversky and Kahneman 1992, p. 306). Guthrie (2000) analyzes a legal application of this element of prospect theory. Another component of prospect theory concerns the subjective weighing of probabilities.


in diverse economic phenomena, including the endowment effect (Kahneman, Knetsch, and Thaler 1990), the status quo bias (Samuelson and Zeckhauser 1988), the equity premium puzzle (Benartzi and Thaler 1995), the tendency to hold on to a losing investment (Staw 1981), and the difficulty of reaching an agreement in negotiations involving losses (Bazerman et al. 2000).8 Although prospect theory speaks directly to preferences among options that vary in gains and losses, little research has addressed the choice between mixed gambles, which involves both gains and losses. Even less research has examined the preference between mixed gambles and pure positive ones. A notable exception is Payne, Laughhunn, and Crum (1980, 1981), who investigated preferences among gamble pairs with equal expected value. At least in such cases, one can infer from their study a preference for pure positive gambles over mixed gambles. A more recent study yielded similar findings, showing that when choosing between multiple-outcome gambles, people prefer gambles with fewer negative outcomes (Birnaum 2006). 3.2. Contingent Fees: A Comparative Comment

Since the experiments were conducted in Israel, it seems useful to briefly describe the relevant Israeli legal and market environment.9 Contingentfee arrangements are as prevalent in Israel as they are in the United States and are regularly used in the same areas of law. Contingent-fee rates are regulated only in specific spheres, primarily in the sphere of death and personal injuries caused by road accidents, where a system of strict liability and compulsory insurance was established in the mid1970s (Kling 2001, pp. 243–58). Contingent-fee rates in Israel are almost as uniform as they are in the United States. The standard flat rate for tort actions (excluding road accidents) is 20–25 percent of the recovery. Sliding scales are sometimes used whereby the CF rate if no suit is filed is 15 percent of the recovery; if a settlement is reached after filing a suit 8. Prospect theory explains not only naive preferences and choices but also judgments and decisions made by sophisticated, professional actors (Coval and Shumway 2005; Olsen 1997; Shapira and Venezia 2001; Roszkowski and Snelbecker 1990), although some studies argue that experienced decision makers can overcome the endowment effect and other manifestations of prospect theory (List 2004; Feng and Seasholes 2005). In addition to the strong experimental evidence, prospect theory has been supported by various empirical studies of different spheres of human activity (Camerer 2000). 9. Contingent fees were traditionally prohibited and are still considered illegal or unethical in most legal systems. The prohibition is, however, gradually eroding in many countries (Kritzer 2004, pp. 256–59).

CONTINGENT FEES /

255

but before the case is tried, 17.5 percent; and otherwise, 20 percent (Kling 2001, p. 249).10 Another comparative remark to be made is that, contrary to the American legal system, in Israel there is no jury system. While this fact decreases the likelihood of megajudgments in personal injury cases and thus reduces the intensity of the debate surrounding CF arrangements, it does not alter the basic elements of the CF debate. As in the United States, legal liability in general and tort liability in particular have significantly and steadily been expanded over the past few decades (Gilead 2003). Against this backdrop, we shall now describe the experiments we conducted to study people’s preferences regarding payment arrangements for legal services.

4 . P L A I N T I F F S ’ C H O I CE

This section describes five experiments conducted in 2007 and 2008 with a view to better understanding when and why plaintiffs prefer CFs and why CF rates may be high. We presented participants with hypothetical cases and asked them to indicate which payment arrangement, CF or FF, they would prefer. 4.1. Experiment 1: Laypersons and Lawyers

The first experiment was designed to examine whether preference for CFs will prevail when the expected fee under a CF arrangement is sub10. According to our survey, which yielded 58 responses from tort lawyers, the common CF rate (excluding road accidents) is 20 percent. We conjecture that two major reasons account for the lower CF rates in Israel compared with the United States. First, unlike their American peers, Israeli lawyers are ethically prohibited from financing the plaintiff’s outof-pocket expenses, such as court fees and expert-witness fees. Thus, they do not charge their clients for this benefit. Second, in Israel—following the English rule and contrary to the American rule—the losing party ordinarily reimburses the winning party for the attorney’s fee and other expenses. This difference implies, first, that the effective CF in Israel is often higher than it initially seems to be. Under a common arrangement, the CF is calculated on the entire sum awarded by the court, including the lawyer’s fee. For instance, if the CF is 25 percent and the court awards the plaintiff, in addition to damages of $100,000, a reimbursement of lawyer’s fee of $20,000, the CF is often calculated as 25 percent of $120,000—that is, $30,000—which is effectively 30 percent of $100,000. Another implication of the Israeli rule is that even a CF arrangement does not guarantee an Israeli plaintiff that she will break even in case of failure. A losing plaintiff is likely to bear at least some of the defendant’s costs. At any rate, in our experiments, the subjects were explicitly asked to assume that the attorney would bear the entire cost of handling the


stantially higher than the FF. In this experiment, the expected value of the CF was set to be over 3 times as much as the FF. We also tested the effect of experience on this preference. Two samples of subjects were used: a sample of naive subjects, or laypersons, consisting of university applicants, undergraduate students, and master’s students at the Hebrew University of Jerusalem, who do not study law, and a sample of tort lawyers. Finally, we examined whether the choice between CF and FF depends on the type of damage for which recovery is sought: property damage or physical injury. 4.1.1. Participants. The sample of laypersons was composed of 77 students and university applicants (47 women, 30 men) who filled out a questionnaire in exchange for 10 new Israeli shekels (NIS).11 Their ages ranged between 18 and 43, with a mean of 24. The expert sample consisted of 66 tort lawyers (29 women, 37 men) who filled out the questionnaire voluntarily. They received and returned it by mail or filled it out during a medical course offered to lawyers as part of their continuing professional education. Ages ranged from 25 to 74, with a mean of 44; experience as a lawyer ranged from half a year to 43 years, with a mean of 15. The lawyer respondents estimated that an average of 78 percent of their entire professional activity is dedicated to civil litigation, including cases that do not go to court (the maximum being 100 percent, and the median being 82.5 percent). In about 70 percent of their civil litigation activity, these lawyers represent plaintiffs rather than defendants. The lawyers indicated that in 90 percent of the cases in which they represent plaintiffs, they charge on a CF basis (the maximum being 100 percent, and the median being 95 percent). While in personal injury claims this percentage rose to 96 percent, in property damage claims it dropped to 78 percent. Only 30 percent of the lawyers had ever represented a defendant on a CF basis. 4.1.2. Experimental Design. We employed six decision problems for which respondents were asked to indicate whether they preferred the

case. This assumption both simplified the questions and made them compatible with the American legal environment. 11. During the period in which our experiments were conducted, 1 new Israeli shekel (NIS) roughly equaled $.25.

CONTINGENT FEES /

257

Table 1. Details of Decision Problems in Experiment 1

Problem 1 2 3 4 5 6

Award

Probability

% Contingent Fee

13,000 10,000 50,000 55,000 100,000 90,000

.50 .90 .95 .45 .90 .85

30 30 50 60 25 30

Claim’s Expected Value

Expected Contingent Fee

Fixed Fee

6,500 9,000 47,500 24,750 90,000 76,500

1,950 2,700 23,750 14,850 22,500 22,950

600 800 7,000 4,500 7,000 7,000

Note. Monetary values are in new Israeli shekels.

CF or the FF arrangement.12 In this and in all of the following experiments, respondents were also given the option to avoid taking any legal action. The numerical details for the six problems are presented in Table 1. These varied in the expected award, the probability of winning, and the proportion of the lawyer’s CF out of the total award. Some of the problems were less realistic than others, yet we wished to examine a relatively wide range of combinations of the three variables. For all problems, the expected contingent fee (ECF; that is, the award times the probability of winning times the percentage of CF) was set to be more than 3 times higher than the FF. More precisely, the ratio between FF and ECF ranged from 1 : 3.21 to 1 : 3.39, averaging 1 : 3.3. Each decision problem was presented to half of the subjects as a case of property damage and to the other half as a case of physical injury. Within each questionnaire, half of the problems were presented as damage-to-property cases and the other half as physical injury cases. The depiction of the problem as a property damage or physical injury case as well as the order of the problems and the order of the CF and FF options were counterbalanced across subjects. The tort lawyers were specifically requested to answer how they would act as a client seeking the services of a lawyer, assuming that they would not handle the case by themselves. No time constraints on filling out the questionnaires were imposed. 12. Hourly fees are more prevalent than fixed fees (FFs). We nevertheless asked respondents to choose between CFs and FFs because making the choice between a CF and an hourly fee would have made the task more complicated, as the amount of hours spent on a case is typically uncertain ex ante. Fixed fees may be inferior to hourly fees in terms of the lawyer’s incentives to invest in the case. However, to the extent that clients are risk averse, an FF is more attractive than an hourly fee. Thus, one can confidently assume that in reality, the preference for CFs over hourly rates would be at least as large as the preference for CFs over FFs.


A typical question read as follows:13 Imagine that you were injured in an accident and broke your leg. A jurist friend estimates that you may claim damages from the injurer, and that the sum that may be reasonably expected is 90,000 NIS. However, getting the damages is not guaranteed, as the injurer may deny his responsibility, in which case a legal process would be necessary. Your friend estimates that there is an 85 percent chance of getting the damages.14 You may choose one of the following three options: A. Ask a lawyer to represent you in claiming the damages for 30 percent of the sum received following his handling of the case (if the damages are obtained, the lawyer will get 27,000 NIS, and if no damages are obtained, the lawyer will get nothing). Assume that the lawyer bears all of the costs involved in handling the case. B. Ask a lawyer to represent you in claiming the damages for a fixed sum of 7,000 NIS (to be paid even if you’ll get no damages at all). Assume that the lawyer bears all of the costs involved in handling the case. C. Avoid handling the matter. What option would you prefer? (circle the appropriate answer): A, B, or C

4.1.3. Results. The percentages of CF and FF choices for each case are presented in Tables A1 and A2 for laypersons and lawyers, respectively. We computed for each subject the percentage of CF choices out of all choices (excluding the choice to avoid legal measures) separately for the personal injury cases and the property damage cases. Laypersons preferred the CF arrangement in 39 percent of the property damage cases and in 46 percent of the personal injury ones. The corresponding percentages for tort lawyers were 42 and 47 percent, respectively (Figure 1). A repeated-measure analysis of variance (ANOVA) of the two percentages by sample (laypersons or lawyers) yielded no significant effect of sample or interaction of the sample with the type of case. Across both samples, the type of case was marginally statistically significant (F [1, 141] p 3.443, p p .066).15 In sum, although the expected fee under the CF arrangement was over 13. Different descriptions of the property damage and physical injury were used to match the varying magnitudes of expected damages. 14. To keep the questions simple, we did not refer to the possibility of obtaining different amounts of damages in different probabilities. This gap leaves some ambiguity in the description, but since this ambiguity is common to all cases, it should not affect the results. 15. The marginal significance of the effect of the type of case (personal injury or property damage) does not appear to stem from lack of attention. The six cases were significantly different from each other (F [5, 650] p 5.503, p ! .001 ), in a repeated-measure analysis of variance [ANOVA] with the six cases as within-subject dependent measures.

CONTINGENT FEES /

259

Figure 1. Percentage of contingent-fee choices for property damage and personal injury

3 times higher than that under the FF arrangement, subjects preferred the CF over the FF arrangement in 43 percent of the choices, across the two samples and two types of cases. Preference for CFs did not depend on the expertise of the responders but was slightly and marginally higher for personal injury cases than for property damage cases. 4.2. Experiment 2: Laypersons with Real Monetary Consequences

The second experiment aimed at examining whether the results obtained in experiment 1 may have been affected by the hypothetical nature of the task. Arguably, in a situation in which a choice has clear monetary consequences for the decision maker, she is more likely to attend to differences in expected value of the available options. In that case, the fact that the expected fee under the CF arrangement would be over 3 times higher than the FF may tilt the balance in favor of the latter option. In order to test this hypothesis, we introduced real monetary consequences, proportional to the ones stated in the case description. 4.2.1. Participants. Twenty-seven students (19 women, eight men) participated in this study. Their ages ranged between 21 and 28, with a mean of 25.


We employed one decision problem referring to bodily injury and using the numerical details of case 6 of experiment 1, which is presented in Table 1. The question was formulated as in experiment 1, except that following the introduction of the three alternatives and before the instruction to mark the chosen alternative, an additional paragraph was inserted. This paragraph indicated that one of four participants would be randomly selected to receive or pay a sum of money according to the outcome of her choice. To imitate the outcomes of the case, those responders would take part in a lottery in which the chance of winning would be 85 percent (they would pull a card from a pack of 20 cards, 17 of which were marked as wins). The sum won would be 1/200 of the awarded damages, namely, NIS 450. A responder who opted for the CF option and drew a winning card would receive this sum minus 30 percent (NIS 135), that is, NIS 315. A responder who opted for the FF option would pay the experimenter 1/200 of the fee indicated in this option, namely, NIS 35. This sum would not be returned. After the payment, the responder would draw a card, and if it was a winning card, she would receive NIS 450 (1/200 of the damages). Otherwise, she would receive nothing. The responders were instructed to approach the experimenter after marking their choice to participate in a lottery that would determine whether they would participate in the real-money lottery. In half of the questionnaires, the CF option was presented first, and in the other half the order was reversed. 4.2.2. Experimental Design.

The order of presentation of the options did not significantly affect choice. Across the two orders, 67 percent of the responders selected the CF arrangement and 33 percent selected the FF. We compared the choices of the participants in this experiment with those of the participants in experiment 1 who responded specifically to the same problem (that is, the personal injury version of this case). As can be seen in Tables A1 and A2, 57 percent of the lay responders and 65 percent of the tort lawyers chose the CF option in this case. Clearly, introducing a real monetary implication did not reduce the preference for CFs.16 4.2.3. Results.

4.3. Experiment 3: Determinants of Preferences

The purpose of the third experiment was threefold. First, we intended to replicate the results of experiment 1 with a different set of values, setting the expected value of the CF to be 2.5 times greater than the FF. 16. Comparison of choice proportions in experiment 1 with those in experiment 2 did not yield significant results: x2(2) p .61, n p 98, p p .73.

CONTINGENT FEES /

261


Problem 1 2 3 4 5 6 7 8

Award

Probability

% Contingent Fee

50,000 50,000 50,000 50,000 250,000 250,000 250,000 250,000

.65 .65 .90 .90 .65 .65 .90 .90

20 55 20 55 20 55 20 55



Fixed Fee

32,500 32,500 45,000 45,000 162,500 162,500 225,00 225,000

6,500 17,875 9,000 24,750 32,500 89,375 45,000 123,750

2,600 7,150 3,600 9,900 13,000 35,750 18,000 49,500


Second, while the overall pattern of preferences in experiment 1 was clear, the significant differences among the six different decision problems were not easily explainable. We thus sought to explore the separate roles of award, probability of success, and rate of CF as determinants of preference for CFs. Finally, in addition to the information provided in experiment 1, subjects were also informed of the number of work hours the case was expected to require.17 Forty-three students (15 men, 27 women, and one who did not indicate gender) participated in the study in exchange for NIS 5. Their ages ranged from 19 to 53, with a mean of 25.

4.3.1. Participants.

4.3.2. Experimental Design. We employed eight decision problems. Each subject was presented with eight scenarios, which varied regarding the award (NIS 50,000 or 250,000), the probability of getting the award (65 or 90 percent), and the CF rate (20 or 55 percent). These three factors were manipulated orthogonally within subjects. In addition, two different presentation orders of the eight problems and two orders of the CF and FF options were counterbalanced across subjects. All cases involved physical injury caused during a medical treatment in a hospital. In all scenarios, the attorney was expected to work 50 hours on the case. Subjects were asked to decide which fee arrangement, CF or FF, they would prefer as a plaintiff. Table 2 presents the details.

Table 3 presents the percentage by which participants preferred a CF over an FF in each case. A repeated-measure ANOVA 4.3.3. Results.

17. This information was added because the experiment also aimed to examine the correlation between subjects’ choices and the perceptions of the fairness of different CF arrangements, which was studied in a separate experiment. See also Zamir and Ritov (2009).


Table 3. Percentage of Contingent-Fee Choices, out of All Nonavoidance Choices (N p 43)

$250,000 Award

$50,000 Award Contingent Fee 20% 55%

Probability of Success p 65%




64.3 68.3

23.8 34.1

92.7 80.0

52.4 53.7

predicting choice in each of the eight cases from claim value, probability of success, and lawyer’s share yielded significant effects for probability of success and claim value (F [1, 42] p 42.743, p ! .001, and F[1, 42] p 17.99, p ! .001, respectively).18 The effect of the lawyer’s share did not approach a significant level (p p .3 ), nor did any of the interactions. Overall, respondents preferred CFs in 4.49 out of eight cases. Thus, in 58 percent of the choices (excluding avoidance), respondents preferred the CF option over the FF even when the ECF was 2.5 times higher than the FF. 4.4. Experiment 4: Exposing Framing Effects

The fourth experiment sought to examine whether framing the choice between different fee arrangements as either a choice between a pure positive and a mixed gamble or between two pure negative gambles would impact people’s preferences. To this end, we presented respondents with two scenarios that differ from one another solely in this respect. 4.4.1. Participants. One hundred twenty-five students (61 women, 64 men) responded to the questionnaire in exchange for NIS 10. Their ages ranged between 18 and 41, with a mean of 25.

We employed four decision problems for which respondents indicated whether they preferred the CF or the FF arrangement (or to avoid taking any legal action). The numerical details for the four problems are presented in Table 4. All the cases involved property damage but varied regarding the expected award, the probability of winning, and the CF rate. We set the CF such that the ECF would be twice as high as the FF. The order of the problems and the order of the CF and FF options were counterbalanced across subjects. 4.4.2. Experimental Design.

18. In each cell of Table 3, no more than three respondents chose the avoidance option.

CONTINGENT FEES /

263


Problem Award Probability 1 2 3 4

40,000 50,000 80,000 100,000

.85 .8 .75 .7

% Contingent Fee



Fixed Fee

20 30 20 30

34,000 40,000 60,000 70,000

6,800 12,000 12,000 21,000

3,400 6,000 6,000 10,500


To induce different framings, two formulations of each problem were used. Participants were randomly assigned to either the gain or the loss framing. A typical decision problem in the loss framing formulation read as follows: Imagine that you have purchased an apartment, and after a while discovered several defects. You hired a contractor who is about to repair the defects for 40,000 NIS. You consider suing the vendor for this sum. If the court will find the vendor liable for the defects, you will be awarded this sum of damages, and you will be able to use it to pay the contractor’s remuneration. The suit’s chances of success are 85 percent. You may choose one of the following three options: A. Ask a lawyer to represent you in claiming the damages for 20 percent of the sum you will save as a result of the fact that the vendor will pay the contractor’s remuneration (if the claim is fully successful and the damages will cover the cost of repair, you will pay the lawyer 8,000 NIS, and if the claim fails, you will pay the contractor the full cost of the repairs, 40,000 NIS, and the lawyer will get nothing). Assume that the lawyer bears all the costs involved in handling the case. B. Ask a lawyer to represent you in claiming the damages for a fixed sum of 3,400 NIS (to be paid even if you will receive no damages, and you will have to pay the contractor’s remuneration of 40,000 NIS out of your own pocket). Assume that the lawyer bears all the costs involved in handling the case. C. Pay the contractor the full cost of the repairs, and avoid all legal action.

The gain-framing question described a similar scenario. However, instead of referring to a contract made with a contractor, it merely stated that the claimed damages were based on the cost of repairs and that if the court finds the vendor liable for the defects, the respondent would be free to use the awarded money for any purpose.


The percentages of CF and FF choices for each case are presented in Table A3. Across all orders, participants in the gain condition preferred the CF over the FF in 57 percent of the four cases (out of nonavoidance choices), while participants in the loss condition preferred the CF over the FF in only 40 percent of the four cases.19 4.4.3. Results.

4.5. Experiment 5: Neutralizing Attorneys’ Incentives

The purpose of the fifth experiment was to examine the client’s preference for CFs over FFs when the client’s payment is decoupled from the lawyer’s fee arrangement. In these questionnaires, the clients chose among the same two options for their agreement with a firm, but the firm hired a lawyer to handle the case on either a CF or an FF basis, irrespective of how the client paid. Thus, preference for a CF or an FF agreement with the firm could not be motivated by the client’s consideration of the lawyer’s incentive. 4.5.1. Participants. Forty-four students (23 men, 21 women) participated in the study in exchange for NIS 5. Their ages ranged from 20 to 41, with a mean of 24. 4.5.2. Experimental Design. We used the same set of cases used in experiment 3 (presented in Table 2). However, the legal fees were paid to a firm handling such cases. The firm hired a lawyer to work on the case, and the lawyer was paid either an FF or a CF. A typical decision problem read as follows: Imagine that your left hand was harmed during a medical treatment in a hospital and you are considering claiming damages from the hospital. The estimation is that, if the court will find the hospital liable, you will get 50,000 NIS in damages, that the chances of success are 65 percent, and that the lawyer who will handle the case will have to spend about 50 working hours. You may choose one of the following three options: A. Ask a firm that handles such claims to hire a lawyer who will handle the damages claim for 20 percent of the sum received following his handling of the case (if the damages are obtained, you will pay the firm 10,000 NIS, and if no damages are obtained, you will pay the firm nothing). Assume that the firm is bearing all the costs involved in handling the case 19. For the main effect of framing, F(1, 116) p 6.627 , p p .01, in an ANOVA of percentage of CF choices by framing condition and the two order factors. The order factors did not yield any significant effects or interactions.

CONTINGENT FEES /

265

and that the firm is paying the lawyer some fixed sum for his work, irrespective of the outcomes of his work. B. Ask a firm that handles such claims to hire a lawyer who will handle the damages claim and pay it a fixed sum of 2,600 NIS (that you will pay the firm even if you will get no damages). Assume that the firm is bearing all the costs involved in handling the case and that the firm is paying the lawyer some fixed sum for his work, irrespective of the outcomes of his work. C. Avoid handling the matter.

The method of lawyer’s remuneration by the firm was varied between subjects. Thus, in half of the questionnaires, instead of the words “the firm is paying the lawyer some fixed sum for his work, irrespective of the outcomes of his work” (in both options A and B), the phrase was “the firm is paying the lawyer a certain percentage of the sum obtained following his work.” As in experiment 3, the other three factors of interest, namely, the claim sum, the probability of success, and the CF, were manipulated orthogonally within subjects. In addition, the order of the two CF and FF options was counterbalanced across subjects. As in experiment 3, all cases involved physical injury. Across all cases, the CF was preferred 50 percent of the time. Although preference for the CF seems somewhat greater when the lawyer was paid contingent on the outcome than when the lawyer’s pay was not contingent (55 versus 46 percent), this difference does not reach a significant level (t [42] p 1.209, p p .23). A comparison between the preference for CFs in this experiment—in which the lawyer’s incentives were normatively irrelevant—and the preference for CFs in experiment 3—where incentivizing the lawyer could have played a role in the subject’s choices—does not yield a significant difference between the two.20 Table 5 presents the percentage of choices in which participants preferred a CF over an FF. We again used a repeated-measure ANOVA of the choice in each of the eight cases by claim value, probability of success, and lawyer’s share. The analysis essentially replicated the results of ex-

4.5.3. Results.

20. Even if one separately compares the results of each of the two conditions of experiment 5 with the results of experiment 3, the effect is not statistically significant. When comparing the condition in which the lawyer’s fee is contingent with the results of experiment 3, there is clearly no significant difference (F [1, 104] p .459 , p p .5 ). When comparing the condition in which the lawyer is paid a FF with the results of experiment 3, the difference is closer to reaching statistical significance but is still not significant (F [1, 104] p 3.227, p p .075).

Lawyer is paid on a contingent-fee basis: Contingent fee p 20% Contingent fee p 55% Lawyer is paid on a fixed-fee basis: Contingent fee p 20% Contingent fee p 55% 21.1 26.3 20.0 25.5

47.8 52.2


68.4 76.5


$50,000 Award

Table 5. Percentage of Contingent-Fee Choices, out of All Nonavoidance Choices (N p 44)

79.2 70.8

94.7 88.9


28.0 45.8

21.1 50.0


$250,000 Award

CONTINGENT FEES /

267

periment 3, showing a significant effect of the probability of success and of the claim value (F [1, 36] p 49.98, p ! .001, and F [1, 36] p 15.162, p ! .001, respectively)21 but no significant effect of the CF percentage (p p .2) and no significant interactions. 4.6. Analysis and Normative Implications

These results shed light on several major issues in the theoretical and policy debates concerning CFs. Specifically, we argue that these results 1. establish that prospect theory provides a convincing explanation for plaintiffs’ preference for CFs; 2. indicate that while incentivizing the lawyer is an important factor in choosing between different fee arrangements, it is not the only, and probably not even the primary, factor; loss aversion, as predicted by prospect theory, is plausibly more important; 3. seem to refute the claim that CF rates are excessive primarily because of asymmetric information; information problems probably do not loom large on the plaintiffs’ preference for CFs, and thus imposing disclosure duties is unlikely to have a very significant effect on the market. In experiment 1, in which the ratio between FFs and ECFs, calculated by multiplying the CF rate by the probability of recovery and by the sum of recovery, was approximately 1 : 3.3, 43 percent of the subjects nevertheless preferred the CF.22 In experiment 3, in which the ratio between the FF and the ECF was 1 : 2.5, 58 percent of the subjects preferred the CF.23 In experiment 4 (with regard to the subjects who made the choice under the conventional framing) and in experiment 6, described in Section 5.2 (with regard to the subjects who made the choice as plaintiffs), in which the ratio be4.6.1. Loss Aversion versus Risk Aversion.

21. In each cell of Table 5, no more than two respondents chose the avoidance option. 22. These results were probably not affected by the hypothetical nature of the decision problems, as the addition of real monetary consequences in experiment 2 did not change the respondents’ choices. If anything, the addition of monetary incentives strengthens the preference for CFs. 23. In another experiment, subjects were provided with a table of six different FFs, ranging from one-quarter to one-half of the ECF (25 percent, 30 percent, . . . 50 percent), and for each FF they were asked to choose between the FF and a CF. Out of the 35 subjects who seemed to have understood the questions, 13 invariably opted for a CF, thus indicating that they prefer a CF even when the expected CF (ECF) is more than 4 times higher than the FF. Eight subjects invariably opted for an FF, thus indicating that they prefer FFs even if the ECF is less than twice as high as the FF. For the remaining 14 subjects, the average indifference point, across choices, was a ratio of 38 percent between the FF and the ECF, which is a ratio of 1 : 2.6.


tween the FF and the ECF was 1 : 2, 57 percent of the subjects preferred the CF. These results provide important clues to understanding plaintiffs’ behavior in the market for legal services. First, the results reinforce the descriptive, explanatory force of prospect theory compared with expected utility theory. In our experiments, most subjects preferred CFs over FFs even when the ECFs were 2 or 2.5 times higher than the FF and 43 percent of the subjects opted for a CF even when the ECF was around 3.3 times higher than the FF, which seems implausible under conventional rational choice theory yet is perfectly compatible with prospect theory. As explained earlier, a unique feature of the choice between a CF and an FF is that it resembles a choice between a pure positive gamble and a mixed gamble.24 Our findings of a strong preference for a CF even when the ECF is much larger than the FF thus extend the findings of Payne, Laughhunn, and Crum (1980, 1981), demonstrating the preference for pure-gain gambles over mixed gambles when the two are equal with respect to expected value.25 Our analysis assumes that plaintiffs frame the choice between CF and FF as a choice between a pure positive and a mixed gamble. Arguably, a plaintiff who suffers injury or loss may view her preinjury or preloss position as the pertinent reference point and thus view even positive recovery as a loss, as long as the recovery is smaller than her losses (Korobkin and Guthrie 1994, pp. 120–24; 1997, pp. 95–99). There are, however, good reasons to believe that this is not ordinarily the case (Rachlinski 1996, pp. 128–30; Hogarth 1987, p. 105; Babcock et al. 1995; Guthrie 2000, p. 182; 2003; van Koppen 1990). Studies have indicated that people internalize losses and gains and adapt their reference points rather quickly (Kahneman, Knetsch, and Thaler 1991; Tversky and Kahneman 1991, pp. 1041–42; Frederick and Loewenstein 1999). Thus, the time lapse between the point of injury or loss and the point at which a plaintiff faces the choice of fee arrangement is likely to make her view the current, postinjury or postloss position as the status quo and thus view her reward as a gain (Rachlinski 1996, p. 129 n.65). It is possible that the fact that in our experiments the subjects were 24. A similar observation was made by Rachlinski (1996, p. 129 n.64) regarding the effect of different fee arrangements on the framing of decisions whether to accept a settlement offer. 25. We also conducted a series of experiments in which subjects were asked to choose between pure positive and mixed gambles straightforwardly presented as gambles rather than as choices pertaining to legal services. The results of these experiments were even more dramatic than those described in this article. See note 38.

CONTINGENT FEES /

269

informed about their loss or injury at the same time they were asked to choose the fee arrangement—rather than at a later point in time—explains why the preference for a CF in our experiments was weaker than in the real world. Perhaps some of the subjects framed both options as belonging to the domain of losses and thus acted as risk seekers.26 This is not to say that under no real-world circumstances will plaintiffs frame the choice of fee arrangement as belonging to the domain of losses. As the results of experiment 4 nicely demonstrate, there may be peculiar cases in which at least some plaintiffs would perceive the choice of fee arrangement as involving losses. We submit, however, that these cases are rare and that plaintiffs conventionally perceive the choice of a fee arrangement as a choice between a mixed gamble and a pure positive gamble. In fact, it took us considerable effort to envisage the scenario used in experiment 4 in which we managed to induce respondents to perceive the choice as involving losses. The significant gap between the respondents’ choice pattern in the two conditions of experiment 4 clearly indicates the effect of framing on the choice of fee arrangements.27 Furthermore, death, personal injury, and comparable claims often include a substantial component of damages for nonmonetary injuries. In such cases, damages are likely to put the plaintiff in a better monetary position even compared with her preaccident positions. The same is certainly true with regard to punitive damages, which are not intended to compensate the plaintiff for any loss she has suffered. Framing the choice of fee arrangements in terms of losses is therefore particularly implausible in cases involving a significant component of nonmonetary damages. This conjecture may explain the (marginally) stronger preference for CFs in the cases pertaining to personal injury (46 percent of the laypersons and 47 percent of the lawyers), compared with the property damage cases (39 and 42 percent, respectively) seen in experiment 1. This conjecture is also supported by the reports of the tort lawyers who participated in experiment 1, according to which CFs are used in 96 percent of the cases in which they represented plaintiffs with personal injuries but only 78 percent of the cases in which they represented plaintiffs with property damages. Our results are compatible with prospect theory and incompatible 26. Additional explanations for this gap are discussed below. 27. Note that there was no difference in wealth between the two conditions in experiment 4; thus, the dissimilar choice patterns cannot be attributed to a wealth effect.


with the standard notion of risk aversion in two additional important respects. First, according to the standard economic notion of risk aversion, people who are more affluent are expected to be less risk averse with regard to any given financial risk. If this is true, we should expect a significant difference between the choices made by students and those made by practicing attorneys, who can safely be assumed to be much richer on average.28 Experiment 1 shows that lawyers’ preferences are almost identical to those of students and university applicants. Note that our claim regarding the loss aversion (rather than risk aversion) of presumably well-to-do respondents does not imply that financial constraints do not affect preferences for CFs. Certainly, such constraints are a primary advantage of CF arrangements for people who could not otherwise afford legal representation. We do argue that the manifested preferences of plaintiffs who can afford to hire an attorney on alternative bases and opt for a CF are best explained by loss aversion. In the real world, in areas such as personal injury cases, practically all plaintiffs—rich and poor—hire their attorneys on a CF basis. Second, according to the standard notion of risk aversion, subjects’ choices should have depended on the magnitude of the sum involved. For any given probability of risk, the larger the sum, the more people are presumed to be risk averse. In fact, however, the preference for a CF was manifest even when the sums were very small. For example, in case 1 of experiment 1, 50 percent of the lawyers in the personal injury scenario and 37 percent of the lawyers in the property damage scenario preferred a CF over an FF, despite the fact that the ECF was 3.25 times higher than the FF and the FF was only NIS 600 (approximately $150). This extreme loss aversion is perfectly compatible with prospect theory but can hardly be explained by rational choice theory, at least under standard assumptions concerning the shape of people’s utility curve (Rabin 2000). Our results nicely coincide with the empirical findings reported by 28. The annual tuition at Israeli public universities (such as the Hebrew University of Jerusalem) is around $2,500. While people from privileged socioeconomic backgrounds are overrepresented in the student body of leading universities, students are generally not wealthy. Most of them work for a living but almost invariably work at temporary, parttime, and relatively low paying jobs. Many of them take out loans to finance their studies or are supported by their parents. In contrast, the economic position of practicing lawyers with a mean professional experience of 15 years is considerably above the mean income of the general population. Various surveys indicate that the net average income of practicing lawyers, depending on their experience, is 3–10 times higher than working students’ average salaries.

CONTINGENT FEES /

271

Kritzer (1998, pp. 305–6): “Interestingly, most of the lawyers who told me that they had recommended to a client that a matter be handled on an hourly basis found that clients almost always opted for the contingency fee when advised of even the slightest possibility of a downside risk.” Our findings indicate that people are willing to pay much higher fees under CF agreements than under FF agreements. This result reinforces the estimations (discussed in Section 2) that the average effective hourly rates under a CF arrangement are considerably higher than under alternative arrangements. 4.6.2. Loss Aversion versus Efficient Incentives. Numerous studies of different fee arrangements have analyzed the incentives they create for the lawyer, especially given that most clients lack effective means of monitoring the lawyer’s performance. The incentive effect is considered not only a key factor in choosing among different fee arrangements but also a central explanation for why CF rates are set above the competitive zero-profit rate. A low CF rate may decrease the attorney’s incentive to invest time and skill in the case, thus decreasing—rather than increasing—the plaintiff’s expected net recovery (the award minus the lawyer’s fee). Can the incentive effects of CF arrangements explain the choices made by the participants in our experiments? In all five experiments, the chances of receiving the award were presented in the opening paragraph of each question, which thus implies that the prospects of obtaining the award were independent of the choice of fee arrangement presented later in the question. However, this formulation did not entirely rule out the possibility that at least some respondents assumed, consciously or unconsciously, that a CF would be preferable because it would increase the chances of getting the award or the magnitude of recovery. To neutralize the possible incentive effects of different fee arrangements, experiment 5 clearly and explicitly decoupled the fee arrangement made between the plaintiff and the firm from the fee arrangement made between the firm and the lawyer who handles the case. While this decoupling somewhat decreased the preference for CFs, from 58 percent in experiment 3 to 50 percent in experiment 5, this decrease was not significant.29 Even if the decrease were significant, this difference would have meant only that incentives do play some role in the choice between 29. This nonsignificant effect of the lawyer’s fee arrangement is in line with research in other domains showing relative insensitivity to the role of incentives (Baron and Ritov 1993; Baron 1994).


different fee arrangements—an observation that is obviously true. The striking result is that even when incentives were neutralized, 50 percent of the respondents opted for a CF when the fee they were expected to pay (ECF) was 2.5 times higher than the FF. This result is in tension with the assumption implicit in the economic analyses, namely, that incentives are (and should be) the primary concern of the parties in shaping the fee arrangement (Choi 2003; Emons 2000, 2006; Emons and Garoupa 2006; Polinsky and Rubinfeld 2003; Santore and Viard 2001). It may be contended that even in experiment 5 the incentive effects of the CF were not absolutely neutralized. Some subjects might have surmised that even if the lawyer is paid a FF, paying the firm on a CF basis would induce it to use other means, such as monitoring the lawyer more closely, to increase the probability of winning the case. At the same time, a firm that is paid on a FF basis may hire less experienced lawyers or pay CF rates that are too low to create sufficient incentives for success. While it is impossible to rule out such conjectures altogether, given the design of the experiment, we believe that its results do cast serious doubt on the centrality of incentives in the present context. Experiment 2, in which the real monetary results of the respondents’ choice were determined by drawing a card from a pack (rather than by anybody’s efforts or experience), provides further support for this conclusion.30 It is difficult to quantify the relative importance of incentives and loss aversion in people’s choice of a CF and in setting the CF rate. Inasmuch as they accurately reflect people’s behavior in the real world, our experiments demonstrate that loss aversion is certainly a major factor, and probably the primary one. A competing explanation for the prevalence of the CF and its high rate is that the market for legal services is not competitive because of clients’ acute information problems as to the expected reward, the risk involved in the suit, the quality of legal services provided, the time required to handle the claim, and so forth (Brickman 2003c). While our findings do not refute this claim, they diminish its significance for two cumulative reasons. First, in all of our experiments, subjects were provided with clear information concerning the reward they might expect and the probability of success. In experiment 3, subjects were also provided with information about the time 4.6.3. Asymmetric Information?

30. See also note 38, where we describe the results of experiments involving abstract gambles.

CONTINGENT FEES /

273

required to handle the case. In fact, since the participants faced a choice between a CF and an FF, and not between a CF and an hourly rate, information about the amount of work necessary to handle the case was irrelevant. Similarly, since in our experiments there was no indication that the choice between the different options would affect the quality of legal services, this information was not relevant either (or so one can reasonably assume). Thus, while it is true that clients often lack relevant information, our experiments indicate that the availability of all relevant information is unlikely to have a major effect on their preferences. A majority of respondents preferred a CF over an FF even when the data presented to them indicated that the ECF was 2 or 2.5 times higher than the FF. Even more important, experiment 1 established that the preferences of experienced tort lawyers who regularly charge their clients on a CF basis—and who are therefore very knowledgeable about the advantages and disadvantages of CF arrangements—are almost identical to those of laypersons. It is therefore quite clear that lack of information does not play a significant role in plaintiffs’ choices of fee arrangements.31 This is not to say that plaintiffs are sufficiently informed when they hire a lawyer on a CF (or any other) basis or that additional information would not be of use to them. For example, our results support the intuition that lawyers have an incentive to portray the prospects of success in any case brought to them as lower than what they truly estimate them to be, to strengthen the client’s tendency to opt for a CF. As Table 3 shows, for any claimed sum and for any CF rate, respondents’ preference for a CF over an FF rose dramatically when the probability of success dropped from 90 to 65 percent. Furthermore, the percentage of respondents who opted for a CF in our experiments is appreciably lower than the percentage of plaintiffs, in both Israel and the United States, who resort to CF arrangements in such spheres as personal injury. The fact that in such cases almost all lawyers charge their clients on a CF basis must therefore reflect addi31. Our findings regarding the preferences of experienced tort lawyers may—or may not—bear on the question whether sophisticated repeat players overcome the “biases” predicted by prospect theory. To begin, this controversy is largely irrelevant in the present context, as most of the plaintiffs in the areas in which CF arrangements are used are in fact unsophisticated one-shot players. As for the choices made by the tort lawyers, while it may be argued that they contradict the contention that experienced professionals act in accordance with the predictions of rational choice theory, it may also be argued that these people are experts in representing plaintiffs, not in being plaintiffs.


tional factors, including the latter’s information problem and financial constraints. Obviously, one party’s preferences are not the only factor determining the terms of a bilateral transaction. Lawyers’ unwillingness to offer alternative payment arrangements surely plays a significant role as well. An additional factor that may play a role is people’s documented excessive discount rate: people tend to underestimate future costs and benefits compared with present ones (Soman 1998; Shelley 1994). Since under CF arrangements payment is ordinarily deferred to the time of recovery, clients may fail to appreciate its true magnitude. In addition, whereas in our experiments, to avoid excessive complexity, subjects were offered a choice between a CF and an FF, in the actual market for legal services the major alternative to a CF is an hourly fee. Hourly fees are considerably less attractive than FFs because of the uncertainty concerning the scope of work the case will require. Finally, the gap between an ECF and other fee arrangements may in reality be smaller than in our experiments. In any case, on the basis of our findings, and especially considering the similarity between laypersons’ and professionals’ preferences, we doubt that imposing disclosure duties on lawyers would make a substantial difference in market behavior. The same holds true for requiring lawyers to offer their prospective clients a choice between different fee arrangements. Despite the explicit, and even highlighted, availability of such alternatives in our experiments, respondents manifested a strong preference for CF arrangements. Arguably, people have a right to obtain all relevant information and to choose among different options before making a contract, even if this information and these options are unlikely to affect their decisions (Bymers 1972; Whitford 1973). We have no quarrel with this view, but one must remember that the imposition of such duties increases the transaction costs, which are then divided between the parties. Another argument might be that if additional information and more options are unlikely to affect plaintiffs’ behavior, the fact that plaintiffs’ preferences deviate from the predictions of standard rational choice theory may justify a paternalistic regulation, for example, by capping CF rates. While we accept that legal paternalism is sometimes desirable and even efficient (Zamir 1998), we doubt that there is enough reliable information regarding the CF market to warrant such regulation on paternalistic grounds. Even if people may gain more by displaying less loss aversion, the complexity of factors affecting the choice of fee arrangement (including clients’ financial constraints, their loss aversion, and the

CONTINGENT FEES /

275

incentive effects of different arrangements), as well as the incompleteness of the empirical data, militate against paternalistic regulation in this sphere. Moreover, to the extent that one is interested in promoting people’s utility rather than people’s wealth, and given the documented prevalence of loss aversion, one should not regulate CF arrangements with a view to inducing people to opt for noncontingent-fee arrangements. If anything, CFs should be encouraged. Finally, from a deontological perspective, limiting people’s freedom for their own sake seems much more justifiable when it is done to prevent loss and suffering than when it is done to ensure greater monetary gain (Zamir and Medina 2010, p. 345).

5. DEFENDANTS’ CHOICE

5.1. The Puzzle and Proposed Solutions

Thus far, our discussion has revolved around plaintiffs’ recourse to CF arrangements and the characteristics of this market. In principle, CFs may be used by defendants just as they are used by plaintiffs. Once a suit is filed, the defendant and her attorney could agree that the attorney’s fee will be a certain percentage of that part of the claim that will be dismissed. For instance, a defendant who is sued for $100,000 may agree to pay her attorney one-third of the difference between this sum and the sum she will have to pay the plaintiff. In this example, if the plaintiff is awarded, say, $70,000 in damages, the defendant will pay her attorney a fee of $10,000 (one-third of $30,000) and no fee if the plaintiff is awarded $100,000. However, while CF arrangements are prevalent among plaintiffs’ attorneys, they are quite rare among defendants’ lawyers (Litan and Salop 1994).32 This paradoxical behavior is puzzling because many of the explanations for the use of CFs apply equally to both plaintiffs and defendants (Dana and Spier 1993). A CF removes the risk that the defendant will have to pay both the full amount of the claim and her attorney’s full fixed or hourly fee. Hence, riskaverse defendants should presumably prefer a CF just as risk-averse plaintiffs do. Similarly, just as a CF mitigates the incentive problem of plaintiffs’ attorneys, it can mitigate the same problem for defendants’ 32. According to the information provided by the tort lawyers who participated in experiment 1, while they charge on a CF basis in 90 percent of the cases in which they represent plaintiffs, only 30 percent of them, in an average professional career of 15 years, had ever represented a defendant on this basis.


lawyers. While an attorney paid by the hour may spend too much time defending a lawsuit when the prospects of successful defense are slim, an attorney paid on a CF basis would refrain from such a waste of time and advise her client to settle the case or even pay the entire claim (that is, refuse to take the case). Several explanations have been offered for this puzzle. First, a CF on the defendant’s side is more difficult to calculate and administer. Presumably, the sum of money recovered by a plaintiff reasonably reflects the merit of her case and may thus serve as a basis for calculating the CF. In contrast, the sum claimed by a plaintiff may reflect strategic considerations and is therefore an unsafe basis for determining the “savings” to the defendant due to her attorney’s skill and efforts (Epstein 2004; Danzon 1983; Dana and Spier 1993; Molot 2009, pp. 378–79). When a claim has a substantial component of general damages, such damages may not even be specified by the plaintiff. Also, since plaintiffs’ attorneys may deduct their fees from the damages paid by the defendant, and thus have an effective means of collecting their fees, they can give their clients credit through a CF at a relatively low cost.33 In contrast, defendants’ attorneys enjoy no such means and may thus be more reluctant to extend credit through CFs. Other explanations for the relative rarity of CF use by defendants focus on the characteristic differences between plaintiffs and defendants, especially in personal injury cases (Dana and Spier 1993; Danzon 1983). Defendants are often large firms or insurance companies. Unlike poor plaintiffs, they need neither the credit CF attorneys afford their clients nor the risk-sharing element of CFs. Furthermore, as sophisticated repeat players, defendants are better positioned to assess the merits of cases brought against them. Hence, while CFs can efficiently cope with plaintiffs’ inability to evaluate the merit and value of their claims (because a CF guarantees that the attorney will take a case only if it is worthwhile to pursue), such a means is arguably unnecessary on the defendants’ side. Finally, large firms can more effectively monitor the activities of their attorneys and are therefore less vulnerable to the conflicts of interest 33. Arguably, this means of collection also makes the payment of the attorney’s fee less painful. Relatedly, Lowenstein and O’Donoghue (2006, p. 199) point out that withholding income tax is less salient, and hence less painful, than requiring people to write a check to the government. This phenomenon, to the extent that it exists, may also be explained according to prospect theory, because it is easier for the plaintiff not to receive the attorney’s share in the first place than to get the entire award and then part with a large share of it.

CONTINGENT FEES /

277

inherent in fixed or hourly fees. Lawyers who wish to maintain profitable relationships with large defendant firms have strong incentives to act in the best interests of their clients. Some of these explanations, however, rest on questionable generalizations, and none of them account for the entire picture. A conspicuous difficulty of the explanations resting on the typical differences between plaintiffs and defendants is that even large firms use CFs when hiring attorneys as plaintiffs in tort actions and for collecting commercial and retail accounts.34 At the same time, even nonwealthy, unsophisticated, and risk-averse defendants do not commonly use CFs (Danzon 1983). As for the asymmetric information and monitoring costs, even a large commercial defendant is unlikely to possess the professional expertise of its attorneys. The asymmetric-information and agency problems explaining the prevalence of CFs on the plaintiffs’ side are therefore applicable to defendants as well.35 In the same vein, the alleged difficulties of collecting CFs by successful defendant attorneys may be overcome quite easily by prepayment arrangements subject to the possibility of repayment by the attorney (Danzon 1983). Finally, while it is true that the sum claimed by a plaintiff may be inflated, the defendant and her attorney may agree that the benchmark for calculating the CF will be a lower sum. A parallel argument is made regarding CFs for plaintiffs, namely, that often the appropriate benchmark for calculating the attorney’s CF should be higher than zero, as it is clear that even absent any effort by the lawyer, the defendant will offer—or even has already offered—some positive payment to settle the case.36 It is not our goal to undermine or discard the above explanations for the plaintiffs/defendants puzzle. Rather, we propose a more general explanation that complements the existing ones. Our explanation draws on prospect theory and on an experiment we conducted to investigate the issue. 34. Kakalik and Pace (1986, pp. 95–96) estimate that 88 percent of organization tort plaintiffs in state courts and 85 percent of organization tort plaintiffs in federal courts pay their lawyers on a contingency basis. See also Aronson (1980, pp. 75–76), Kritzer (1990, pp. 58–59, 189 n.6), Kritzer (1998, p. 306), Schneyer (1998, pp. 378–81), and Morgan and Stoner (2004), which note the increase in CF patent suits. Even the government sometimes hires lawyers on a CF basis. See Erichson (2000, pp. 1–18, 35–40). 35. Moreover, as Bebchuk and Guzman (1996, pp. 61–62) argue, under the assumption that defendants are better able than plaintiffs to monitor and control the lawyer’s decisions, a CF arrangement could provide them with a strategic advantage in the negotiation for settlement. 36. For a description of a reform proposal based on this insight, see Horowitz (1995).


5.2. Experiment 6: Plaintiffs and Defendants

This experiment was designed to test whether the preference for CFs over FFs depends on the role of the decision maker as a plaintiff or defendant. By keeping all aspects of the case and the fee arrangement options constant—varying only the role of the decision maker—we can determine to what extent her preferences are affected by the role played. As defendants are more likely than plaintiffs to frame both the CF and the FF outcomes as pure losses, we expected the CF arrangement to lose its unique appeal when the options were considered from a defendant’s perspective. Hence, we predicted a preference reversal, whereby a CF would be preferred over an FF when participants responded as plaintiffs and an FF would be more attractive when participants responded as defendants. Seventy-seven students (35 women, 42 men) responded to the questionnaire in exchange for NIS 10. Their ages ranged between 19 and 30, with a mean of 24.

5.2.1. Participants.

5.2.2. Experimental Design. We employed six decision problems for which respondents were asked to indicate whether they preferred the CF or the FF arrangement (or avoiding any action). The numerical details for the six problems are presented in Table 6. Again, the problems varied in the expected award, the probability of winning, and the CF rate. The type of damage (property or personal) was not indicated. The CF was set so that the ECF would be approximately twice as high as the FF. The three questions were presented twice in each questionnaire: once as a scenario involving a plaintiff and once as a similar scenario involving a defendant. The order of plaintiff or defendant framing of the problem, the order of the problems, and the order of the CF and FF options were counterbalanced across subjects. The plaintiff scenarios were similar to those used in experiments 1 and 3. A typical defendant scenario read as follows: Imagine that a suit was filed against you for damages of 100,000 NIS for harm caused to the plaintiff, that he claims you are liable for. A jurist friend estimates that if you defend against the suit, there is a 90 percent chance that the suit will be dismissed, and you will not have to pay the plaintiff anything. You may choose one of the following three options: A. Ask a lawyer to represent you in the legal proceedings for a fee of 25 percent of the sum you will save as a result of the dismissal of the

CONTINGENT FEES /

279


Problem

Damage

Probability


1 2 3

100,000 150,000 60,000

.90 .80 .95

90,000 120,000 57,000

% Contingent Fee


Fixed Fee

25 30 35

22,500 36,000 19,950

11,000 18,000 10,000


claim (if the suit will be dismissed and you will not have to pay the plaintiff anything, you will pay your lawyer 25,000 NIS, and if you will be obliged to pay the damages, you will not pay your lawyer any fee). Assume that the lawyer bears all the costs involved in the legal proceedings. B. Ask a lawyer to represent you for a fixed sum of 11,000 NIS (that you will have to pay him even if you will be obliged to pay the damages of 100,000 NIS). Assume that the lawyer bears all the costs involved in the legal proceedings. C. Pay the damages of 100,000 NIS. 5.2.3. Results. The percentages of CF and FF choices for each case are presented in Table 7. We computed for each case the percentage of CF choices out of all choices made in the defendant role, excluding the choice to avoid legal measures, and the corresponding percentage for choices made in the plaintiff role. Across all orders, participants preferred CFs over FFs in 57 percent of the three cases in which they acted as plaintiffs (out of nonavoidance choices) and in only 29 percent of the three cases in which they acted as defendants.37

5.3. Analysis and Implications

Our experiment demonstrates that even in the absence of any of the typical differences between plaintiffs and defendants—their affluence, risk aversion, sophistication, being one-shot or repeat players, means of collecting attorneys’ fees, or the ease of calculating the fee—people display very different preferences regarding the choice between CFs and FFs when they are positioned in the role of plaintiffs or defendants. While a clear majority of plaintiffs (57 percent) preferred CFs over FFs when the ratio between the FF and the ECF was approximately 1 : 2, when acting as defendants, more than two-thirds (71 percent) preferred 37. In a repeated-measure ANOVA with percentage of CF choices in the two roles as a within-subject dependant factor and the two order factors as between-subject independent predictors, F(1, 73) p 32.13, p ! .001.


Table 7. Experiment 6: Percentage of Contingent-Fee and Fixed-Fee Choices (N p 77)

Plaintiff: Contingent fee Fixed fee Defendant: Contingent fee Fixed fee

(1)

(2)

(3)

67.5 32.5

75.3 23.4

27.3 71.4

29.9 68.8

36.4 61.0

18.2 79.2

Note. Values do not add up to 100 because of the third option, to avoid legal measures.

FFs over CFs under similar conditions. The fact that the same people were asked to make the choice as both plaintiffs and defendants excludes any explanation having to do with differences between subjects. Since the respondents were inexperienced laypersons, it is extremely unlikely that they were aware of the prevailing fee arrangements, either for plaintiffs or for defendants. Conformity with common usage is therefore not a plausible explanation either. The results of our experiment are perfectly compatible with prospect theory and in particular with the characterization of people as risk averse in the domain of gains (that is, when they act as plaintiffs) and as risk seeking in the domain of losses (that is, when they act as defendants). For the defendant, an FF implies that she will have to pay a total sum ranging from the FF only (if the claim is fully dismissed) to the claimed sum plus the FF (if the plaintiff gets her entire claim). The CF, in contrast, narrows this range and thus reduces the defendant’s risk. Under a CF arrangement, while the minimal loss may be higher than under an FF arrangement, the maximal loss is capped by the claimed sum because if the claim is fully successful, the defendant’s attorney gets no fee. Thus, rational choice theory, which views the choice between CFs and other fee arrangements as basically similar for plaintiffs and defendants, crucially misses the different ways plaintiffs and defendants frame this choice. Possibly the fact that all respondents were presented with both the plaintiff and the defendant scenarios, with the same numerical details in both framings, highlighted the similarity between the situations and thus reduced the framing effect.38 While we do not rule out the possibility 38. In another experiment, the subjects were presented with choices between monetary gambles. Three out of four framings resembled the plaintiff’s position and one the defendants’ position, and each subject was presented with only one type of framing. Whereas

CONTINGENT FEES /

281

that a defendant who has profited from violating the plaintiff’s rights may view even a judgment requiring her to pay a certain amount of money as decreasing her gains rather than involving a loss, for the reasons discussed in Section 4.6 regarding plaintiffs’ framing, we submit that these cases are quite rare. In addition, it should be noted that very often the losses and injuries suffered by the plaintiff are not accompanied by any correlative gains for the defendant. Admittedly, the gap between the expressed preferences for CF under the two conditions in experiment 6 is smaller than the difference between the actual behavior of plaintiffs and defendants in the real world. In addition to the fact that our respondents were faced with the two types of framing in the same questionnaire, this difference plausibly indicates that the factors discussed in Section 5.1 also impact preferences for CFs.

6. CONCLUSION

People’s preferences, judgments, and choices play a central role in the market for legal services, as they do in any sphere of human behavior. Past scholarship has analyzed this market from legal, comparative, ethical, economic, political-economic, and sociological perspectives yet curiously disregarded its behavioral aspects. As this article demonstrates, cognitive psychology in general—and prospect theory in particular—can greatly contribute to a better understanding of the particular characteristics of this market and to a resolution of some of the major descriptive and normative debates surrounding it. Inter alia, our findings and analysis shed new light on the role of loss aversion in people’s choice of CF arrangements, the extent of people’s preferences for such arrangements, the (rather limited) weight clients attribute to incentivizing the lawyer in their choice of fee arrangements, and the (again limited) role of information problems and lack of alternative fee options in the market. We do not argue that the behavioral perspective is necessarily more important than other points of view. Hence, our analysis incorporates numerous insights from other brands of scholarship. We are also aware of the limits of experimental studies of the kind presented in this article. We do, however, believe that behavioral factors play a key role in the CF market and that the experimental results provide strong evidence in the mixed versus pure positive gambles, 73–88 percent of the subjects opted for the pure positive gamble, only 17 percent of the subjects opted for the corresponding option in the pure-loss framing.


about these factors. Our results also illuminate other contexts in which people face comparable choices. The picture emerging from our research is sharper than previous ones but not at all simpler. Contrary to some depictions of the CF market, we conclude that lawyers are neither saints nor devils. The positive and normative picture is rather complex and nuanced; so should be the legal treatment of CF arrangements. APPENDIX

Table A1. Experiment 1: Laypersons’ Responses to Six Decision Problems

Personal injury: CF FF Property damage: CF FF Overall: CF FF

(1)

(2)

(3)

(4)

(5)

(6)

52.5 42.5

40.5 56.8

20.0 77.5

59.5 29.7

37.5 62.5

56.8 40.5

40.5 48.7

27.5 70.0

29.7 70.3

50.0 47.5

35.1 64.9

42.5 55.0

46.8 45.5

33.8 63.6

24.7 74.0

54.6 39

36.4 63.6

49.4 48.1

Note. Values are percentages of subjects who prefer a contingent fee (CF) or a fixed fee (FF), across subjects in each cell. The values do not add up to 100 because of the third option, to avoid legal measures. N p 77.

Table A2. Experiment 1: Lawyers’ Responses to Six Decision Problems

Personal injury: CF FF Property damage: CF FF Overall: CF FF

(1)

(2)

(3)

(4)

(5)

(6)

46.7 46.7

44.4 47.2

20.7 72.4

40.0 51.4

43.3 56.7

64.7 35.3

33.3 57.6

23.3 73.3

31.4 60.0

26.7 70.0

69.4 30.6

37.9 62.1

39.7 52.4

34.8 59.1

26.6 65.6

33.8 60.0

57.6 42.4

52.4 47.6

Note. Values are percentages of subjects who prefer a contingent fee (CF) or a fixed fee (FF), across subjects in each cell. The values do not add up to 100 because of the third option, to avoid legal measures. N p 66.

CONTINGENT FEES /

283

Table A3. Experiment 4: Framing Effects

Gain: CF FF Loss: CF FF

(1)

(2)

(3)

(4)

54.8 43.5

45.2 51.6

59.7 38.7

62.9 33.9

39.7 57.1

23.8 73.0

44.4 47.6

42.9 50.8

Note. Values are percentages of contingent fee (CF) and fixed fee (FF). The values do not add up to 100 because of the third option, to avoid legal measures. N p 125. R E F E R E N CE S

Aronson, Robert H. 1980. Attorney-Client Fee Arrangements: Regulation and Review. Washington, D.C.: Federal Judicial Center. Babcock, Linda, Henry S. Farber, Cynthia Fobian, and Eldar Shafir. 1995. Forming Beliefs about Adjudicated Outcomes: Perceptions of Risk and Reservation Values. International Review of Law and Economics 15:289–303. Barberis, Nicholas, Ming Huang, and Tano Santos. 2001. Prospect Theory and Asset Prices. Quarterly Journal of Economics 116:1–53. Baron, Jonathan. 1994. Nonconsequentialist Decisions. Behavioral and Brain Sciences 17:1–42. Baron, Jonathan, and Ilana Ritov. 1993. Intuitions about Penalties and Compensation in the Context of Tort Law. Journal of Risk and Uncertainty 7: 17–33. Bazerman, Max H., Jared R. Curhan, Don Moore, and Kathleen L. Valley. 2000. Negotiation. Annual Review of Psychology 51:279–314. Bebchuk, Lucian Arye, and Andrew T. Guzman. 1996. How Would You Like to Pay for That? The Strategic Effects of Fee Arrangements on Settlement Terms. Harvard Negotiation Law Review 1:53–63. Benartzi, Shlomo, and Richard H. Thaler. 1995. Myopic Loss Aversion and the Equity Premium Puzzle. Quarterly Journal of Economics 110:73–92. Birnaum, Michael H. 2006. Evidence against Prospect Theories in Gambles with Positive, Negative, and Mixed Consequences. Journal of Economic Psychology 27:737–61. Bond, Paul. 2002. Making Champerty Work: An Invitation to State Action. University of Pennsylvania Law Review 150:1297–1341. Brickman, Lester. 1989. Contingent Fees without Contingencies: Hamlet without the Prince of Denmark? UCLA Law Review 37:29–137. ———. 1994. On the Relevance of the Admissibility of Scientific Evidence: Tort System Outcomes Are Principally Determined by Lawyers’ Rates of Return. Cardozo Law Review 15:1755–97.


———. 1996a. ABA Regulation of Contingency Fees: Money Talks, Ethics Walks. Fordham Law Review 65:247–335. ———. 1996b. Contingency Fee Abuses, Ethical Mandates, and the Disciplinary System: The Case against Case-by-Case Enforcement. Washington and Lee Law Review 53:1339–79. ———. 2003a. Effective Hourly Rates of Contingency-Fee Lawyers: Competing Data and Non-competitive Fees. Washington University Law Quarterly 81: 653–736. ———. 2003b. The Continuing Assault on the Citadel of Fiduciary Protection: Ethics 2000’s Revision of Model Rule 1.5. University of Illinois Law Review, pp. 1182–1216. ———. 2003c. The Market for Contingent Fee–Financed Tort Litigation: Is It Price Competitive? Cardozo Law Review 25:68–128. Bymers, Gwen. 1972. Seller-Buyer Communication: Point of View of a Family Economist. Journal of Home Economics 64:59–163. Camerer, Colin F. 2000. Prospect Theory in the Wild: Evidence from the Field. Pp. 288–300 in Choices, Values, and Frames, edited by Daniel Kahneman and Amos Tversky. Cambridge: Cambridge University Press. Choi, Albert. 2003. Allocating Settlement Authority under a Contingent-Fee Arrangement. Journal of Legal Studies 32:585–610. Clermont, Kevin M., and John D. Currivan. 1978. Improving on the Contingent Fee. Cornell Law Review 63:529–639. Coval, Joshua D., and Tyler Shumway. 2005. Do Behavioral Biases Affect Prices? Journal of Finance 60:1–34. Dana, James D. Jr., and Kathryn E. Spier. 1993. Expertise and Contingent Fees: The Role of Asymmetric Information in Attorney Compensation. Journal of Law, Economics, and Organization 9:349–67. Danzon, Patricia Munch. 1983. Contingent Fees for Personal Injury Litigation. Bell Journal of Economics 14:213–24. Drummonds, Henry H. 1993. The Law and Ethics of Percentage Contingent Fees in Oregon. Oregon Law Review 72:859–900. Dwyer, Casey L. 2006. An Empirical Examination of the Equal Protection Challenge to Contingency Fee Restrictions in Medical Malpractice Reform Statutes. Duke Law Journal 56:611–41. Emons, Winand. 2000. Expertise, Contingent Fees, and Insufficient Attorney Effort. International Review of Law and Economics 20:21–33. ———. 2006. Playing It Safe with Low Conditional Fees versus Being Insured by High Contingent Fees. American Law and Economics Review 8:20–32. Emons, Winand, and Nuno Garoupa. 2006. US-Style Contingent and UK-Style Conditional Fees: Agency Problems and the Supply of Legal Services. Managerial and Decision Economics 27:379–85. Epstein, Richard. 2004. Cases and Materials on Torts. New York: Aspen. Erichson, Howard M. 2000. Coattail Class Actions: Reflections on Microsoft,

CONTINGENT FEES /

285

Tobacco, and the Mixing of Public and Private Lawyering in Mass Litigation. U.C. Davis Law Review 34:1–48. Feng, Lei, and Mark S. Seasholes. 2005. Do Investors Sophistication and Trading Experience Eliminate Behavioral Biases in Financial Markets? Review of Finance 9:305–51. Frederick, Shane, and George Loewenstein. 1999. Hedonic Adaptation. Pp. 302–29 in Well-Being: The Foundations of Hedonic Psychology, edited by Daniel Kahneman, Ed Diener, and Norbert Schwarz. New York: Russell Sage Foundation. Gilead, Israel. 2003. Israel. In International Encyclopedia of Laws: Tort Law, edited by Sophie Stijns. The Hague: Kluwer Law International. Gross, Leonard. 2006. Are Differences among the Attorney Conflict of Interest Rules Consistent with Principles of Behavioral Economics? Georgetown Journal of Legal Ethics 19:111–54. Guthrie, Chris. 2000. Framing Frivolous Litigation: A Psychological Theory. University of Chicago Law Review 67:163–216. ———. 2003. Prospect Theory, Risk Preference, and the Law. Northwestern University Law Review 97:1115–63. Hay, Bruce L. 1996. Contingent Fees and Agency Costs. Journal of Legal Studies 25:503–33. Helland Eric, and Alexander Tabarrok. 2003. Contingency Fees, Settlement Delay and Low-Quality Litigation: Empirical Evidence from Two Datasets. Journal of Law, Economics, and Organization 19:517–42. Hogarth, Robin. 1987. Judgment and Choice. 2d ed. Chichester: John Wiley & Sons. Horowitz, Michael. 1995. Making Ethics Real, Making Ethics Work: A Proposal for Contingency Fee Reform. Emory Law Journal 44:173–211. Kahneman, Daniel, Jack L. Knetsch, and Richard H. Thaler. 1990. Experimental Tests of the Endowment Effect and the Coase Theorem. Journal of Political Economy 98:1325–48. ———. 1991. The Endowment Effect, Loss Aversion, and the Status Quo Bias. Journal of Economic Perspectives 5:193–206. Kahneman, Daniel, and Amos Tversky. 1979. Prospect Theory: An Analysis of Decision under Risk. Econometrica 47:263–91. Kakalik, James S., and Nicholas M. Pace. 1986. Costs and Compensation Paid in Tort Litigation. Santa Monica, Calif.: Institute for Civil Justice, RAND. Kling, Gabriel. 2001. Ethics for Lawyers (in Hebrew). Tel Aviv: Israel Bar. Koniak, Susan P. 1996. Principled Opinion: Response to Brickman. Fordham Law Review 65:337–54. Korobkin, Russell, and Chris Guthrie. 1994. Psychological Barriers to Litigation Settlement: An Experimental Approach. Michigan Law Review 93:107–92. ———. 1997. Psychology, Economics, and Settlement: A New Look on the Role of the Lawyer. Texas Law Review 76:77–141.


Korobkin, Russell B., and Thomas S. Ulen. 2000. Law and Behavioral Science: Removing the Rationality Assumption from Law and Economics. California Law Review 88:1051–1144. Ko¨szegi, Botond, and Matthew Rabin. 2006. A Model of Reference-Dependent Preferences. Quarterly Journal of Economics 121:1133–65. Kritzer, Herbert M. 1990. The Justice Broker: Lawyers and Ordinary Litigation. New York: Oxford University Press. ———. 1998. The Wages of Risk: The Returns of Contingency Fee Legal Practice. DePaul Law Review 47:267–319. ———. 2002. Seven Dogged Myths Concerning Contingency Fees. Washington University Law Quarterly 80:739–94. ———. 2004. Risks, Reputations, and Rewards: Contingency Fee Legal Practice in the United States. Stanford, Calif.: Stanford University Press. List, John A. 2004. Neoclassical Theory vs. Prospect Theory: Evidence from the Marketplace. Econometrica 72:615–25. Litan, Robert E., and Steven C. Salop. 1994. Reforming the Lawyer-Client Relationship through Alternative Billing Methods. Judicature 77:191–97. Lowenstein, George, and Ted O’Donoghue. 2006. “We Can Do This the Easy Way or the Hard Way”: Negative Emotions, Self-Regulation, and the Law. University of Chicago Law Review 73:183–206. McKee, Michael, Rudy Santore, and Joel Shelton. 2007. Contingency Fees, Moral Hazard, and Attorney Rents: A Laboratory Experiment. Journal of Legal Studies 36:253–73. Metzloff, Thomas B. 1991. Resolving Malpractice Disputes: Imaging the Jury’s Shadow. Law and Contemporary Problems 54:43–129. Miceli, Thomas J., and Kathleen Segerson. 1991. Contingent Fees for Lawyers: The Impact on Litigation and Accident Prevention. Journal of Legal Studies 20:381–99. Miller, Geoffrey P. 1987. Some Agency Problems in Settlement. Journal of Legal Studies 16:189–215. Moliterno, James E. 2003. Broad Prohibition, Thin Rationale: The “Acquisition of an Interest and Financial Assistance in Litigation” Rules. Georgetown Journal of Legal Ethics 16:223–57. ———. 2009. A Golden Age of Civic Involvement: The Client Centered Disadvantage for Lawyers Acting as Public Officials. William and Mary Law Review 50:1261–1308. Molot, Jonathan T. 2009. A Market in Litigation Risk. University of Chicago Law Review 76:367–439. Morgan, Paul, and Bruce Stoner. 2004. Reexamination vs. Litigation—Making Intelligent Decisions in Challenging Patent Validity. Journal of the Patent and Trademark Office Society 86:441–63. Odean, Terrance. 1998. Are Investors Reluctant to Realize Their Losses? Journal of Finance 53:1775–95.

CONTINGENT FEES /

287

Olsen, Robert A. 1997. Prospect Theory as an Explanation of Risky Choice by Professional Investors: Some Evidence. Review of Financial Economics 6: 225–32. Painter, Richard W. 1995. Litigating on a Contingency: A Monopoly of Champions or a Market for Champerty? Chicago-Kent Law Review 71:625–97. Payne, John W., Dan J. Laughhunn, and Roy Crum. 1980. Translation of Gambles and Aspiration Levels Effects in Risky Choice Behavior. Management Science 26:1039–60. ———. 1981. Further Tests of Aspiration Levels Effects in Risky Choice Behavior. Management Science 27:953–58. Polinsky, A. Mitchell, and Daniel L. Rubinfeld. 2003. Aligning the Interests of Lawyers and Clients. American Law and Economic Review 5:165–88. Rabin, Matthew. 2000. Risk-Aversion and Expected-Utility Theory: A Calibration Theory. Econometrica 68:1281–92. Rachlinski, Jeffrey J. 1996. Gains, Losses, and the Psychology of Litigation. Southern California Law Review 70:113–85. Rickman, Neil. 1994. The Economics of Contingency Fees in Personal Injury Litigation. Oxford Review of Economic Policy 10:34–50. Roszkowski, Michael J., and Glen E. Snelbecker. 1990. Effects of “Framing” on Measures of Risk Tolerance: Financial Planners Are Not Immune. Journal of Behavioral Economics 19:237–46. Samuelson, William, and Richard Zeckhauser. 1988. Status Quo Bias in Decision Making. Journal of Risk and Uncertainty 1:7–59. Santore, Rudy, and Alan D. Viard. 2001. Legal Fee Restrictions, Moral Hazard, and Attorney Rents. Journal of Law and Economics 44:549–72. Schneyer, Ted. 1998. Legal-Process Constraints on the Regulation of Lawyers’ Contingent Fee Contracts. DePaul Law Review 47:371–412. Schwartz, Murray L., and Daniel J. B. Mitchell. 1970. Economic Analysis of the Contingent Fee in Personal-Injury Litigation. Stanford Law Review 22: 1125–62. Shapira, Zur, and Itzhak Venezia. 2001. Patterns of Behavior of Professionally Managed and Independent Investors. Journal Banking and Finance 25: 1573–87. Shavell, Steven. 1987. Economic Analysis of Accident Law. Cambridge, Mass.: Harvard University Press. Shelley, Marjorie K. 1994. Gain/Loss Asymmetry in Risky Intertemporal Choice. Organizational Behavior and Human Decision Processes 59:124–59. Soman, Dilip. 1998. The Illusion of Delayed Incentives: Evaluating Future EffortMoney Transaction. Journal of Marketing Research 35:427–37. Staw, Barry M. 1981. The Escalation of Commitment to a Course of Action. Academy of Management Review 6:577–87. Thomas, Robert E. 1998. Psychological Impact of Scrutiny on Contingent Fee Attorney Effort. West Virginia Law Review 101:327–84.


Thomason, Terry. 1991. Are Attorneys Paid What They’re Worth? Contingent Fees and the Process of Settlement. Journal Legal Studies 20:187–223. Tversky, Amos, and Daniel Kahneman. 1991. Loss Aversion in Riskless Choice: A Reference-Dependent Model. Quarterly Journal of Economics 106: 1039–61. ———. 1992. Advances in Prospect Theory: Cumulative Representation of Uncertainty. Journal of Risk and Uncertainty 5:297–323. van Koppen, Peter J. 1990. Risk Taking in Civil Law Negotiations. Law and Human Behavior 14:151–65. Whitford, William C. 1973. The Functions of Disclosure Regulation in Consumer Transactions. Wisconsin Law Review, pp. 400–70. Zamir, Eyal. 1998. The Efficiency of Paternalism. Virginia Law Review 84: 229–86. Zamir, Eyal, and Barak Medina. 2010. Law, Economics, and Morality. New York: Oxford University Press. Zamir, Eyal, and Ilana Ritov. 2009. Notions of Fairness and Contingent Fees. Unpublished manuscript. December 21. Hebrew University of Jerusalem, Faculty of Law and School of Education. http://ssrn.com/abstractp1526644.