Content uploaded by Keren Weinshall Margel
Author content
All content in this area was uploaded by Keren Weinshall Margel on Dec 14, 2015
Content may be subject to copyright.
1
Cost–Benefit Analysis of Class Actions:
An Israeli Perspective
by
Alon Klement and Keren Weinshall-Margel∗
We propose an analytical framework for evaluating the effectiveness of class ac-
tions, in which their costs are estimated against the benefits they produce in at-
taining three objectives: law enforcement and deterrence; access to courts; and
compensation. We outline parameters for measuring the social costs and benefits
relevant to these objectives and evaluate them for Israeli class actions, based on
original data including all class actions filed between 2006 and 2012 (n=2,056).
Findings indicate that class actions did not substantially facilitate access to courts
and compensation, and that they had limited success in realizing law enforcement
and deterrence. (JEL: K40, K41)
1 Introduction
The class action is a unique procedure. It addresses a problem where violation of
a legal duty results in dispersed harm to numerous individuals. When each individ-
ual’s loss does not justify pursuing its recovery in court, producers of mass harm
might not be sufficiently deterred from violating their legal duties. The class action
addresses this problem by aggregating the small individual claims into one lawsuit.
It creates a procedural mechanism that renders the aggregate harm “marketable,”
thus incentivizing representing plaintiffs and attorneys to identify suitable causes of
action and to pursue them in court.
Absent a preexisting organizational framework that addresses the problems of
collective action (like the one pertaining to corporations or labor unions), the class
action mechanism must not only incentivize litigation, but also stipulate procedures
for controlling and monitoring it. The self-appointed agents – the class representative
and her attorney – must be prevented from exploiting the class for their own benefit.
∗Alon Klement: Faculty of Law, Tel-Aviv University, Tel Aviv; Keren Weinshall-
Margel (corresponding author): Faculty of Law, Hebrew University, Jerusalem, Is-
rael. Professor Klement was supported by the Israel Science Foundation (ISF), Re-
search Grant 709/12. Dr. Weinshall-Margel was supported by the I-CORE program
of the Planning and Budgeting Committee and the Israel Science Foundation (grant
no. 1821/12). Part of the data analyzed was collected in collaboration with the Is-
raeli Courts Research Division. We thank Christoph Engel and Michael Heise for their
thoughtful comments.
Journal of Institutional and Theoretical Economics – ISSN 0932-4569
DOI: 10.1628/093245616X14472269022780 – ©2016 Mohr Siebeck
Alon Klement and Keren Weinshall-Margel2 JITE 172
At the same time, defendants must also be protected against frivolous litigation.
Indeed, the problems of agency, on the one hand, and of frivolous litigation, on the
other, are often considered to be the main impediments of the class action procedure,
undermining its ability to realize its objectives (Hensler et al., 2000).
Agency problems and frivolous litigation affect both the objectives of the class
action and its costs, but are very difficult to measure. Our study measures, instead, the
end products of class action litigation – its benefits and its costs. The paper identifies
and explains the relevant parameters for empirically assessing these benefits and
costs, using observational data and analysis. However, we also note the inherent
difficulties of such an empirical evaluation and, consequently, the limitations of an
overall observational–empirical assessment of the effectiveness of class actions.
We propose an analytical framework for examining the effectiveness of class
actions. For this purpose, we define their three main objectives: (1) enforcing the
law and deterring future violations; (2) exercising the right of access to the court,
especially for disadvantaged groups or individuals; (3) providing compensation for
injured parties. Against these objectives, we analyze the private and public costs of
class action litigation. Optimal class actions must fully realize their goals, while at
the same time minimize their costs.
Using the analytical model, we examine the application of the Israeli 2006 Class
Action Law (“the CAL” or “the law”). Starting in the late 1980s, the Israeli legislature
has incorporated class action procedures into substantive laws in various sectors such
as consumer protection, banking, and securities. In 2006, these arrangements were
replaced with a uniform arrangement stipulated in the CAL. The study examines
empirical findings of the implementation of the CAL since it entered into effect in
April 2006 and through the end of the 2012 court term (August 31, 2012).
The research method is primarily based on a quantitative analysis of all motions to
certify a class action filed during the period under study (hereinafter – “class action
cases”) (2,056 cases). We created a unique database that includes over 200 variables
measuring different aspects of all class action cases, including characteristics of
plaintiffs, defendants, their lawyers, the represented groups, the causes of action,
and all court proceedings and decisions in each case.Trends and data from analyzing
class action cases are compared with findings from a reference group that includes
a random sample of civil cases heard during the same period in Israeli district courts
(n=278 civil cases). The comparison provides a point of reference to the findings
and is intended to facilitate a deeper understanding of the costs and benefits of class
actions, as opposed to ordinary civil litigation.
The paper proceeds as follows: Following the introduction, section 2 analyzes
the costs and benefits of the class action in light of its objectives, as well as the
inherent risks in the mechanism. We present a model for evaluating the effectiveness
of class actions and discuss the empirical parameters relevant for assessment, as
well as their limitations. Section 3 outlines the historical and legal background for
the 2006 CAL in Israel. Section 4 presents our research method and sources of
information. Section 5 discusses the empirical findings: in the first subsection, we
analyze key trends and characteristics in motions to certify class actions; in the
Cost–Benefit Analysis of Class Actions(2016) 3
second subsection, we focus on case outcomes and evaluate the extent to which they
realize the objectives of the class action; the third subsection includes an in-depth
analysis of the use of monitoring tools stipulated by the CAL for contending with
the unique challenges of class actions – the principal–agent dilemma and the danger
of frivolous lawsuits; the fourth subsection analyzes the costs of class actions in
Israel. Section 6 concludes.
2 The Effectiveness of Class Actions: An Analytical Framework
2.1 The Social Objectives of Class Action Litigation
We divide the objectives of the class action mechanism into three main categories:
exercising the right of access to court; providing compensation for injured parties;
and enforcing the law and deterring against its violation. We discuss each of these
objectives in turn.
2.1.1 Exercising the Right of Access to Court
By its very nature, a class action does not allow all represented plaintiffs to fully
exercise their right of due process in court. A represented plaintiff is not a party
to the class action proceeding, and she may intervene in it only with the court’s
explicit permission. Moreover, the outcome of the class action can be binding upon
the represented plaintiff as res judicata, thus denying her future individual access
to courts. However, the class action does facilitate class members’ access to court,
especially in cases where the value of their individual claims, net of their costs,
is negative. Therefore, a class action substitutes class members’ right of collective
access to the court for their individual respective rights.
Class actions should therefore be distinguished according to the value of class
members’ individual rights. When violated rights have low importance for each
plaintiff, pursuing them in court has no inherent value as well. This is the case, for
example, in consumer lawsuits pertaining to deception or overcharging, where the
only damage caused is monetary damage of low value. On the other hand, other
class actions aggregate claims whose value for each plaintiff is likely to be high
– for example, discrimination and disabled-accessibility lawsuits. Such individual
lawsuits might, nevertheless, be economically unfeasible, and therefore they would
not be filed absent an aggregating mechanism. Others may be individually filed, but,
given the economies of scale enjoyed by defendants, the litigation would be tilted
in their favor.
When the class action substitutes viable individual claims, its value in facilitating
collective access to courts must be weighed against its cost in denying individual
access. The importance of a class action in realizing the right of access to courts is
thus highest when the claim has significant value for the individual, but she cannot
exercise her litigation right – either because she lacks the necessary information and
Alon Klement and Keren Weinshall-Margel4 JITE 172
resources, or because she fears the defendant’s retaliation if she sues him (e.g., in
employment litigation).
2.1.2 Compensating Injured Parties
Monetary compensation is a mere transfer of money from the defendant to the
plaintiffs. The latter’s gain is identical to the former’s loss. Therefore, a significant
social value of compensation inheres in its reallocation of risk, from risk-averse
plaintiffs to risk-neutral (or, at least, less risk-averse) defendants (Shavell, 2004).
Examination of the insurance market indicates that people do not insure against
small losses. For example, most people do not insure themselves against the risk of
suffering a minor injury while on a trip or the risk of losing a very small amount of
money. In other words, they refuse to pay any premium for such insurance. Similarly,
people do not insure themselves against nonmonetary damages (such as pain and
suffering or an infringement of their autonomy). Therefore, the value attributed to
future compensation for such damages cannot be high.
Hence, compensation is socially valuable mainly for individual monetary losses
that are sufficiently high, yet individuals are not insured against them. If individual
lawsuits are not filedin such cases, then class actions may be necessary to serve this
social goal. Still, the social value of class actions is obviously not the ex post measure
of losses, but the ex ante (discounted by the probability of loss) risk premium that
individuals would be willing to pay for being compensated.
2.1.3 Enforcing the Law and Deterring against its Violation
When the individual value for each plaintiff is high, a class action may serve as
an effective mechanism for law enforcement and deterrence. However, the primary
importance of enforcement and deterrence is in cases where the value of the indi-
vidual remedy is low. In these cases, enforcement of the law and deterrence against
its violation are the principal, and often sole, objectives of the class action. The
small damages sustained by each member of the group of plaintiffs, which are not
significant for any of them individually, amount to substantial social loss when
added together. Consequently, there is social value in enforcing the law, averting
ongoing violations and deterring future ones. Since these lawsuits would not be filed
by individual plaintiffs, the defendant would have lower incentives to comply with
the law absent the class action mechanism.
It is important to distinguish between the objectives of enforcement and the
objective of deterrence: Enforcement of the law and prevention of its violation are
usually accomplished through injunctive remedies. In this way, the court rectifies an
existing situation in which the law isviolated, starting from the time the injunction is
issued. Deterrence is attained through monetary remedies. A defendant who expects
to be sued for the losses he causes will internalize these losses and take efficient steps,
in advance, to avoid such violation. In order to achieve deterrence, compensation
does not necessarily need to be paid to plaintiffs (Eisenberg and Engel, 2014). It is
Cost–Benefit Analysis of Class Actions(2016) 5
only necessary that defendants pay it, and that it equal the actual losses they caused.
This is important, as we explain below, in cases in which it is difficult to identify
and compensate injured plaintiffs.
2.2 The Social Costs of Class Action Litigation
A class action makes it possible to aggregate numerous claims in a single proceeding.
When each of these claims would have been individually litigated, combining them
into one aggregate litigation saves resources for the plaintiffs, the defendant, and
the court. Hence, in such cases, the saving of litigation costs in itself may justify
using the class action procedure. However, in all cases in which individual lawsuits
would not have been filed, introducing a class action entails additional costs. This
is true for most class actions.
Class actions impose additional costs not only because they render litigation
viable, but also because they require special procedural mechanisms to overcome the
principal–agent and frivolous-litigation problems. The three mechanisms designed
to overcome these problems are certification of a lawsuit as a class action, approval
of a settlement or a voluntary dismissal, and compensation of the class representative
and the class attorney. Courts actively supervise and monitor class actions through
these mechanisms. Yet, these procedures areresource-intensive and time-consuming
– for the parties and for the court.
2.3 An Analytical Framework for Evaluating the Effectiveness of Class Actions
What is the net social surplus from class actions? We present an analytical framework
that can be empirically estimated, subject to limitations we explore below. The
challenge is to create a single metric for evaluating both the costs and the benefits
of the class action.
Absent class actions, defendants’ spending on satisfying their legal duties is
denoted C0
P; given these costs, defendants decide on a level of activity (private,
business, or public), whose social benefit is denoted U0; and despite these means
of precaution, individuals suffer losses as a result of intentional or unintentional
violations, negligent or not, whose total cost is denoted D0. Assuming that in this
situation individual claims would not be filed, the overall social welfare, denoted W0,
would be equal to the benefit from the activity minus the costs of precaution and the
losses caused:
W0=U0−C0
P−D0.(1)
When class actions are available, we denote their overall litigation costs by C1
L.
These costs consist of plaintiff, defendant, and court costs. Defendants’ spending
on satisfying their legal duties is C1
P, and the benefit from their activity is U1.
Consequently, the social losses, given these costs of care and levels of activity,
are D1. Three additional social benefits are the value of duties enforced through
injunctive remedies, denoted E1, the value of collective access to court, denoted A1,
Alon Klement and Keren Weinshall-Margel6 JITE 172
and the insurance value of compensation, denoted I1. Therefore, the overall social
welfare when class actions can be filed is
W1=U1−C1
P−D1+A1+I1+E1−C1
L.(2)
Class actions are socially beneficial if they raise overall net social welfare, that
is, if
W1>W0.
Substituting from (1) and (2), denoting by ΔXthe difference in the variable X
between its value with class actions and its value in their absence, and rearranging,
we get
ΔU−ΔCP−ΔD>C1
L−E1−A1−I1.
The left-hand side equals the deterrence value of class actions. The right-hand side
consists of all ex post costs and benefits of class actions.
While the analytic model is straightforward, trying to empirically estimate it using
observational data is much less obvious. In particular, the deterrence value can only
be estimated by observing the change in defendants’ behavior in expectation of
being subject to class actions. Such observations are difficult to obtain. Indirect
observations, such as stock market data or opinion survey data, may be collected to
estimate the deterrence effect. However, this study focuses on case characteristics
and outcomes and therefore may only qualitatively estimate to what extent these
outcomes could affect defendant behavior.
The right-hand side is more conducive to empirical estimation using observations
of case files and their outcomes. On the cost side, we observe different parameters
that indicate the amount of time spent in court (such as number of hearings, length
of proceedings, case weights, etc.). Absent direct data on litigation spending by the
parties, we can only estimate it based on these observations.
On the benefit side, we observe various characteristics of the cases that indicate
their access to court value. In particular, we can identify cases in which individual
harm or causes of action is substantial, so that taking such cases to court may
serve class members’ rights of access to court. Enforcement value can be estimated
by observing case outcomes, and in particular the injunctive remedies awarded.
Finally, compensation value, although measured based on its ex ante insurance
premium value, which cannot be directly observed, can nevertheless be estimated
by observing case characteristics and outcomes, focusing on monetary remedies.
We elaborate on our estimation methodology for each parameter in the next section.
2.4 Previous Empirical–Observational Studies of Class Actions
Most empirical research on class actions has been conducted in the U.S., where the
class action mechanism has been in effect in its current form since 1966. Two projects
conducted a comprehensive observational analysis to evaluate the implementation
of class actions in general: Hensler et al. (2000) aggregated descriptive statistics
on U.S. class actions from various sources and conducted qualitative case studies
Cost–Benefit Analysis of Class Actions(2016) 7
of ten class actions, and Willging, Hooper, and Niemic (1996) analyzed all cases
terminated within a two-year period in four federal district courts.1Few empirical
studies have focused on class action settlements and explored their outcomes (e.g.,
Fitzpatrick, 2010); on attorney fees in class action settlements (e.g., Eisenberg and
Miller, 2004b, 2010); and on the role of opt-outs and objectors (e.g., Eisenberg and
Miller, 2004a).
Other studies have attempted to evaluate the effects of specific reforms such as
the Securities Litigation Reform Act of 1995 (e.g., Choi, 2007) and the Class Action
Fairness Act of 2005 (e.g., Lee III and Willging, 2008), or significant Supreme Court
decisions like Amchem and Ortiz, which imposed substantial limitations on class
action settlements (e.g., Niemic and Willging, 2002).
Finally, significant empirical research has focused on class actions and other
aggregative mechanisms in specific substantive areas, such as securities litigation
(e.g., Choi, 2007) and mass torts (e.g., Carroll et al., 2005).
Outside the U.S., no jurisdiction except Canada and Australia has had sufficient
numbers of class actions to conduct a meaningful quantitative study. For an empirical
study of class actions in Australia, see Morabito (2009, 2010); and in Canada, see
Pritchard and Sarra (2010) (focusing on securities class actions). As mentioned
above, this is the first study presenting and analyzing data on Israeli class action
filings and outcomes.
3 Class Actions in Israel – A Legal Background
The Israeli parliament (the Knesset) enacted the CAL on January 1, 2006. The new
law superseded all previous statutory arrangements, which allowed class actions
for specific causes of action, and it set a comprehensive procedure for filing and
litigating class actions. The legislation responded to the Supreme Court’s call for
comprehensive and uniform regulation of class actions in LCA 3126/00 State of
Israel v. E.S.T. Management and Manpower, Ltd. 57 P.D. 220 (2003).
The new law was modeled after the American FRCP Rule 23.2This model is
based on several fundamental principles. First, in order to pursue a class action,
a plaintiff must motion the court to certify it as such, and certification is conditioned
on various statutory requirements. Second, all plaintiffs who are members of the
class defined in the certification decision are bound by its outcome, unless they
actively opt out of the class action. Third, the possible remedies in a class action
include monetary compensation, even if such compensation is not uniform among
all class members. Fourth, the attorney’s fee is contingent upon winning the case or
a settlement, and it depends, inter alia, on the value of the remedy awarded.
1One should note, however, that Willging, Hooper, and Niemic (1996) pertain only
to federal courts. The scope of state class actions remains uncertain, since state courts
do not separately docket these cases.
2This choice was first made in enacting the specific statutory arrangements preced-
ing the CAL, and applied in the CAL as well.
Alon Klement and Keren Weinshall-Margel8 JITE 172
Although the CAL adhered to the basic features of the U.S. class action, it deviates
from the American model in various respects. Most significantly, the Israeli legisla-
ture, concerned about the potential adverse ramifications of class actions, has opted
for a substance-specific framework by designating the possible causes of action
that may be brought as a class action.3It thus departed from the trans-substantive
American rule 23, which allows a class action to be filed on any ground. As of today,
the CAL specifies 13 causes of action for submitting class actions. These include
consumer claims (including insurance and banking), corporate lawsuits pertaining
to securities and antitrust, claims related to environmental hazards, employment-
related claims, lawsuits aimed at providing equality for the disabled or for minority
groups, and restitution claims for unlawful fees collected by state authorities.
Furthermore, the CAL has also incorporated various arrangements that deviate
from the U.S. model. Thus, the CAL allows for representation by nonprofit organi-
zations and a few (specified) state regulators; it provides the court with an option
to certify an opt-in class action, in which class members should explicitly express
their willingness to join the lawsuit in order to be bound by its outcome; it explicitly
allows for cy-près remedies; it allows for government funding of class actions; and
it provides specific defenses for the state as a defendant.
Finally, the CAL also departs from the American model in some of its specific
certification requirements. Most importantly, it requires the court to examine the
merits as a condition for class certification, whereas the American procedure does
not favor that (Eisen v. Carlisle & Jacquelin, 417 U.S. 156 (1974)).
4 Research Method and Data
We coded all motions to certify a class action (“cases”) filed in Israel during the
period starting in April 2006 (when the CAL entered into effect) and ending in
August 2012 (the end of the 2012 court term). Our data set includes 2,056 such
cases. An analysis of all class actions in a “census study” method allows for in-
depth statistical analysis, subject to minimal reliability tests that are required for
smaller samples.
For each case, we coded over 200 variables, including information regarding the
class representatives and their attorneys, the defendants and their attorneys, causes
of action and other characteristics of the motion to certify a class action, remedies
requested, and more. By August 2013, 1,397 of the 2,056 cases were resolved
(“resolved claims”). For these cases the database also includes information about
case outcomes and other characteristics of the proceedings.
The cases were drawn from computerized court records and were coded by law
students who read all pertinent documents: pleadings, motions, hearing protocols,
closing arguments,and court decisions. A second tier of encoders randomly sampled
approximately 10% of these cases for accuracy and internal reliability. Coding
3See details in the explanatory remarks for the class action legislation, in Govern-
mental Law Proposals 211, pp. 251–257 (Hebrew).
Cost–Benefit Analysis of Class Actions(2016) 9
was found to be consistent in over 95% of the variables coded. Since the Israeli
court records were computerized only in 2006, there is no reliable information on
class actions filed before the CAL. It is clear, however, that class actions were not
frequently used in Israel before the CAL.
When possible, we compared class action data with a reference group of individual
lawsuits filed during the same period in the district courts. The comparison aimed
to provide a comparative perspective and context to our findings. The comparison
group included a representative sample of civil lawsuits filed in the district courts
as a court of first instance (n=278).4(The overwhelming majority of class action
cases are filed in the district courts.)
5 Findings
To analyze the costs and benefits of class actions, it is necessary to distinguish
among different types of claims, remedies, and represented plaintiff classes. This
section examines the characteristics as featured in motions to certify a class action.
5.1 Trends in Filing Requests for Certification of Class Actions
5.1.1 Number of Motions to Certify a Class Action
Over 2,000 motions to certify a class action were filed from the enactment of the
2006 law until the end of August 2012. During this period the number of class
actions increased dramatically, as shown in Figure 1.
The number of individual lawsuits filed during this period remained constant.5
This holds true, in particular, for lawsuits that could have been substituted by
class actions, such as consumer small claims against businesses or employment
claims.6Thus, there is no indication that enactment of the CAL had any substitution
effect on individual litigation by aggregating numerous individual claims in a single
procedure. Apparently, class actions mostly aggregated claims that would not have
been individually filed otherwise.
Two implications follow: First, the CAL has increased the litigation caseload
as well as the overall costs of litigation by plaintiffs, defendants, and courts. In
section 5.4 we estimate these additional costs. Second, the CAL had no adverse
impact on individual access to courts. The law has facilitated collective access to
the courts in cases that would not have been filed otherwise.
4These cases involved claims of over 2.5 million NIS, or pertained to real estate
ownership. They are included in a broad database of information on civil cases in var-
ious courts. See Weinshall-Margel and Taraboulos (2014).
5See the Israeli judiciary reports: http://elyon1.court.gov.il/heb/haba/dochot/doc/
hofesh_meida2013.pdf (last accessed August 7, 2015).
6This is based on samples taken from cases in the labor courts and small claims
courts, as well as administrative cases between 2006 and 2012, together with data on
all cases filed in these courts from 2000 to 2012. See note 11.
Alon Klement and Keren Weinshall-Margel10 JITE 172
Figure 1
Number of Class Actions Filed, 2006–2012
5.1.2 Type and Categories of Motions to Certify a Class Action
As noted, a class action may be certified only if its cause of action is listed in
the CAL’s second appendix. There are 13 causes listed, and Table 1 presents the
frequency of motions to certify a class action, according to these causes of action.
Our findings indicate that about 78% of the motions to certify a class action were
consumer claims: About 68% were filed under Article 1 of the appendix, which
refers to all claims customers may have against businesses (“dealers” as this term is
defined in section 1 of the Consumer Protection Law). These include, among others,
consumer protection, contracts, and even tort actions against businesses. About
6% were filed under Article 2 of the appendix, which refers to insurance-related
claims. And about 4% were filed under Article 3 of the appendix, which refers to
customer claims against banking institutions. An additional 14% of the motions
to certify were filed against state authorities, under Article 11 of the appendix,
which refers to claims against government or local authorities for the restitution of
unlawfully collected payments. Only a small minority of the motions were based
on corporate or securities law, antitrust law, environmental law, employment law, or
antidiscrimination laws.
To facilitate our cost–benefit analysis, we divided the causes of action into five
categories. Our categorization is similar to conventional international categorization
in the field:7The first category, consumer claims, includes all claims of customers
against businesses, insurers, and banks (Articles 1–3), as well as a small number of
motions filed against advertisers for sending “spam” messages in violation of sec-
7See categorizations in various countries on the website http://globalclassactions.
stanford.edu/about (last accessed August 7, 2015), as well as different ways of catego-
rization in the U.S. in Hensler et al. (2000, pp. 52–58).
Cost–Benefit Analysis of Class Actions(2016) 11
Table 1
Class Actions Filed by Cause of Action
Cause of action specified in the application for certification of the class Frequency
action claim (in accordance with the categories defined in the CAL)
Article 1 – A claim against a “dealer,” as defined under the Consumer Pro-
tection Law, in matters between a dealer and a customer
67.7%
Article 2 – An insurance-related claim against an insurer, insurance agent,
or provident fund management company
5.5%
Article 3 – A claim against a banking corporation pertaining to client–bank
relationships
3.8%
Article 4 – A claim pursuant to the Antitrust Law 0.7%
Article 5 – A claim arising from the ownership, possession, sale, or pur-
chase of a security, as defined in the Corporation Law and in the Securities
Law
1.7%
Article 6 – A claim regarding an environmental nuisance 1.1%
Article 7 – A claim of discrimination under the Prohibition of Discrimina-
tion in Products, Services and Entry into Places of Entertainment and Pub-
lic Places Law, 2000
0.1%
Article 8 – A labor discrimination claim, under the Equal Opportunities at
Work Law or under the Equal Pay for Male and Female Employees Law
0.1%
Article 9 – A claim under the Equal Rights for Disabled Law 1%
Article 10 – A labor-related claim. 2.9%
Article 11 – A claim against governmental or local authorities for the resti-
tution of amounts collected unlawfully
14.0%
Article 12 – A claim against an “advertiser,” as defined in the Communica-
tions (Telecommunications and Broadcasts) Law.
1.4%
Article 13 – A claim against a company operating a central pension clear-
ing system, regarding transmission of data or finances via that system
0
Total 100%
2004
tion 30A of the Communications Law (Article 12). The second category, corporate
claims, includes all claims based on either corporate, securities, or antitrust laws
(Articles 4–5). The third category includes claims against a governmental author-
ity, where, due to specific defenses allowed by the CAL to the state, restitution is
often unavailable or limited, and therefore the remedy often awarded is an injunc-
tion against further unlawful collection of payments and overcharges (Article 11).
A fourth category, employment and civil rights claims, aggregates causes of action
related to employment (Article 10) and discrimination, either in a labor context or
otherwise (Articles 7–9). The two categories were combined due to the low number
Alon Klement and Keren Weinshall-Margel12 JITE 172
of lawsuits in them, and because the right of access to the court is particularly
dominant in both, given the type of claims and the vulnerability of the plaintiffs.
The fifth category includes environmental claims filed against a defendant who is
responsible for an environmental hazard (Article 6). This category is similar to the
mass torts category in the U.S.
Figure 2 compares the distribution of class actions filed in Israel with the markedly
different distribution in the U.S.8About 78% of class actions in Israel are consumer
claims. This high percentage remained stable throughout the years examined. In
the U.S., consumer claims consist of only 25% of the total number of class actions
filed, and the other two significant categories are corporate (securities and antitrust)
claims (19%) and employment and civil rights claims (29%). Corporate claims in
Israel comprise only about 2% of all class actions, and claims related to civil rights
and employment together account for only 4% of the class actions. While in the
U.S. mass torts consist of 9% of class actions, environmental claims in Israel are
particularly infrequent, comprising only about 1% of the total. The only category
with similar distribution in both countries is claims against the government – 14%
Figure 2
Class Actions by Categories, Israel versus U.S.
8Distribution data from the U.S. is from Deborah R. Hensler et al. (2000, p. 53).
The distribution includes an analysis of all 1,020 judicial decisions in class actions
reported in the years 1995–1996 in federal and state courts. We did not find a more
up-to-date report of the distribution of all of the class actions. However, it seems that
changes stemming, inter alia, from the enactment of the Class Action Fairness Act, 28
U.S.C. §§ 1332(d), 1453, 1711–1715, in 2005 may have led to an increase in the num-
ber of consumer claims in the U.S. (see Lee III and Willging, 2008).
Cost–Benefit Analysis of Class Actions(2016) 13
of the class actions in Israel and 11% of the U.S. total. The different distribution of
claims indicates that class actions are not necessarily serving the same goals in the
two legal regimes, even though the procedure is similar in both.
5.1.3 Characteristics of Motions to Certify a Class Action
In order to best identify the class action goals in practice, this section summarizes
various characteristics of motions to certify a class action, by case category.
Table 2 presents descriptive data on the identity of defendants, the estimated
number of class members, and the remedies sought, by class action category. The
table does not include information on the type of class representative. This is because
nearly all of the class representatives – 99.3% – were plaintiffs who had a personal
cause of action, and were part of the class they sought to represent.9The CAL
allows nonprofit organizations to represent a class when the court is convinced that
filing the class action by a class member would prove difficult (section 4(a)(3) of the
CAL). Yet, only 13 such class actions were filed. Furthermore, although the CAL
specifically empowers three public regulators to submit motions to certify a class
action on grounds of discrimination and environmental hazards,10 none of them had
used this power as of August 2012.
The first column in Table 2 describes the type of class action defendants. As
expected, the vast majority of consumer and corporate claims are filed against
companies (94% and 99% respectively), and all governmental claims are submitted
against government authorities. Government authorities are also sued in 38% of
employment and civil rights cases and in 21% of environmental claims.
The second and third columns in Table 2 denote the type of remedy sought.
Nearly all motions to certify (96.2%) seek monetary compensation. In 56.3%,
an injunction or declaratory remedy is sought, usually in addition to monetary
compensation. A breakdown by class action category indicates, as expected, that
monetary compensation is particularly common in consumer lawsuits (97%), and
that it is slightly less common in employment and civil rights claims (84%) and
in environmental class actions (82%). Injunctive or declaratory reliefs are most
frequent in consumer claims, and in employment and civil rights claims (57% in
each category).
The fourth column displays statistics about the estimated number of class mem-
bers, according to the certification motion. The representative attorney often makes
these estimations without any concrete information regarding the exact number.
Nonetheless, the data indicates significant and expected variation among the differ-
ent class action categories. The represented class is smallest in cases of employment
and civil rights (a median of 6,070 members and an average of 34,448) and largest in
9The analysis includes data on the first three class representatives in each request.
Over 80% of the requests were filed by a single representative, and 5% included more
than three.
10 The Commission for Equal Rights of Persons with Disabilities, the Israel Nature
and Parks Authority, and the Equal Employment Opportunities Commission.
Alon Klement and Keren Weinshall-Margel14 JITE 172
Table 2
Characteristics of Certification Requests, by Category of Class Action
Type Defendant type Monetary Injunctive Estimated no. Individual gross Total compen-
compen- or of members sums (NIS) in sation requested
sation declaratory in the repre- representing for the class –
remedy sented group plaintiff’s claim gross NIS amount
Consumer Company 94%
97% 57%
MED 100,620 193 30,000,000
Claims Governmental 4% AVE 546,955 4,386 359,865,918
Nonprofit 1% SD 1,225,025 35,034 3,682,724,972
organization n930 1,397 1,426
Corporate Company 99%
86% 30%
MED 1,000,000 4,670 54,109,871
Claims Governmental 1% AVE 1,221,135 102,453 595,440,522
SD 1,225,025 338,413 1,722,646,540
n11 32 38
Employment Company 61%
84% 57%
MED 6,070 3,017 23,271,972
and Governmental 38% AVE 34,448 11,675 294,256,345
Civil Rights Nonprofit 1% SD 72,983 25,820 6,009,111,034
Claims organization n52 55 79
Claims
94% 43%
MED 10,300 726 6,000,000
against a Governmental 100% AVE 98,560 6,170 31,108,206
Governmental SD 420,098 19,485 91,301,936
Authority n99 242 225
Environmental Company 74%
82% 35%
MED 75,000 2,000 33,996,613
Claims Governmental 21% AVE 210,446 5,174 273,891,403
Nonprofit 5% SD 492,069 9,385 636,628,691
organization n16 18 17
Total Company 80% 96.2% 56.3% MED 100,000 250 25,450,000
Governmental 19% AVE 482,252 6,715 435,275,100
Nonprofit 1% SD 1,144,349 57,060 3,566,396,932
organization 1,978 1,130 n1,108 1,753 1,794
Cost–Benefit Analysis of Class Actions(2016) 15
environmental and consumer lawsuits (in consumer lawsuits the median is 100,620
and the average is 546,955).
In the fifth column, we examine the class representative’s claim as an indicator for
class members’ average claim value. Unlike the number of class members, which is
based on mere speculation, the value of the representative’s claim must be supported
by real evidence. The bottom row in the table shows that the median representative’s
claim value is NIS 250. We note that in over 70% of the motions this value is less
than NIS 1,000, and that in about 22% it ranges between NIS 1,000 and NIS 10,000.
Comparison of the class representative’s claim value in the various categories
shows that it is lowest in consumer lawsuits, followed by class actions against
governmental authorities, environmental claims, civil rights, and work-related cases;
it is highest in corporate class actions, in which the variance is also considerable.
Although the individual claims are low, the total compensation requested for the
class is very high, as shown in the last column in Table 2. Themedian class compen-
sation requested in all cases is NIS 25,450,000, and the average is NIS 435,275,100.
This average is36 times higher than the average compensation demanded in a refer-
ence group of civil cases in the district courts. Note however the very high standard
deviation for the total amount of compensation requested (SD =3,566,396,932).11
There are no significant differences among claim categories, except for in-class
actions against government authorities, which are characterized by a relatively low
estimated class compensation – with a median of NIS 6 million and an average of
about NIS 31 million. This is in line with the Israeli law, which limits the period
for which plaintiffs can receive compensation in this category to two years prior
to the date of filing the certification motion. The high class compensation in other
categories is not explained by the estimated number of class members, the class
representative’s claim, or the interaction between them. We developed various re-
gression equations in an effort to understand the total compensation sum, but these
did not yield any significant results or explanations. It should be observed, however,
that whereas in standard civil cases plaintiffs must pay a court fee that equals 2.5%
of their claim, class actions are exempt from such fees. Therefore, class represen-
tatives bear no direct cost for overestimating the total class claim. Class actions’
outcomes, which we discuss below, reinforce our conclusion that motions to certify
highly overstate total class compensation.
5.1.4 Discussion
Our analysis of the motions to certify a class action indicates the following: most
of these motions are filed for consumer-related causes of action (78%); the primary
remedy in these cases is compensation; the value of individual lawsuits is particularly
low; the estimated number of group members represented in the class action is
high, and the total compensation sum demanded is also high. Such lawsuits may
serve primarily one objective – enforcement of the law and deterrence against its
11 The range is NIS 1–10 million in 38.4% of the requests.
Alon Klement and Keren Weinshall-Margel16 JITE 172
violation. Their utility in exercising the right of access to the courts and in providing
compensation to injured parties is minimal.
On the other hand, there is a very low number of certification motions in lawsuits in
which the objectives of access to courts and compensation areimportant. Only 4% of
the motions are filed for employment and civil rights causes of action. These lawsuits
are characterized by a low number of plaintiffs relative to consumer claims, as well
as relatively high individual value and a relatively high frequency of requests for
declaratory remedies and injunctions, which are likely to have significant economic
repercussions.
About 14% of the class actions are claims against governmental authorities for
unlawful collection of payments. The CAL allows the authority to cease the allegedly
unlawful collection and thus prevent certification of the class action. In addition, even
when the government authority does not take advantage of this option, restitution
to the class is limited to 24 months preceding the filing of the certification motion.
Therefore, the primary objective these class actions serve is law enforcement and
prevention of future infringement.12
Finally, the percentage of motions to certify a class action in the categories of
corporate and environmental lawsuits was very low. Therefore, the objectives of the
CAL were hardly attained in these categories during the period under study.
5.2 Outcomes of Class Action Proceedings
5.2.1 Decisions in Motions to Certify a Class Action
To litigate a case as a class action, the court must certify it as such. Only after the
certification motion is accepted can the case proceed to the next stage of litigation.
Tocertify the case as a class action, the court must find that the following conditions
are satisfied: There are common issues of fact or law pertaining to the class; there
is a reasonable likelihood that these issues will be decided in favor of the class;
litigating the case as a class action will be efficient and fair; the class will be
adequately represented and its representation will be in good faith (section 8(a) of
the CAL). In addition, the court must find that the cause of action is listed in the
CAL’s second appendix
Thus, class certification constitutes a significant sorting mechanism aimed at
guaranteeing that the law realizes its objectives, and that it minimizes the prob-
lems of agency and frivolous litigation. Its implementation affects not only judicial
resolution of the cases filed, but also their voluntary or consensual termination in
dismissal or settlement, before or after class certification (Klement, 2011, p. 32). At
the same time, certification imposes significant costs on the court and the parties.
12 This category may realize the citizens’ right of access vis-à-vis the authority,
even if the individual sums of the lawsuit are low and even if the authority is ulti-
mately released from its obligation to repay the sums it collected in a class action. The
authority is still subject to individual lawsuits that are not precluded by the CAL.
Cost–Benefit Analysis of Class Actions(2016) 17
Our findings indicate that in the overwhelming majority of cases, the court did
not render a decision on the motion to certify a class action. Courts decided the
motion to certify in 171 cases, which was less than 10% of all 1,397 resolved cases.
Of these 171 cases, the class action was certified in only 49 cases (about 29%).
Since the average time from filing to rejection of a motion to certify is shorter than
the average time from filing to certification,13 we also examined the outcomes in
cases opened before 2010 (a total of 102 cases). We found that a similar fraction
– about one-third of the motions to certify (34%) – were accepted, whereas the
remaining two-thirds were rejected. We found no significant difference in the rates
of certification among the various class action categories.
A study of the U.S. federal courts found that a judicial decision on the motion
to certify was made in over one-third of the cases, and two-thirds of the motions to
certify were approved (Willging and Wheatman, 2006). Thus, Israeli courts make
class certification decisions in a much lower percentage of the cases, and their
certification rate is much lower, compared with U.S. federal courts.
The most common grounds for rejection pertain to the plaintiff and his attorney
– 41% of the requests were rejected after the court determined that the class repre-
sentative lacked personal grounds, and 20% were rejected after the court found that
the plaintiff did not adequately represent the group. Yet, in all of these cases the
court found at least one additional cause for not certifying the class action.14 In 34%
of the rejected requests, the additional reason for rejection was that the likelihood
of the collective questions being decided in favor of the group was not sufficiently
high.
5.2.2 Closing Decisions
Class action proceedings may be resolved in a number of ways, listed in Table 3.
Overall, 1,397 cases were resolved between 2006 and 2012.15 No significant dif-
ferences were found in the distribution of closing decisions among the various
categories of class actions, with the exception of claims against a government au-
thority, as we explain below.
Over 57% of the cases were concluded by the class representative’s voluntary
dismissal of her lawsuit prior to certification as a class action. Voluntary dismissals
have no binding (res judicata) effect overclass members, but they must nevertheless
13 On average, 539 days from filing to rejecting decisions (n=121 cases,
SD =334 days) versus 626 days from filing to a decision to certify (n=49,
SD =357 days).
14 Consequently, the court did not exercise its authority to certify the request con-
ditional upon replacing the class representative, according to section 8(c) of the CAL.
15 In order to verify that the data does not include distortions stemming from the
fact that the duration of the proceeding is different is each of the forms of closure, we
checked the distribution of the forms of closure only among 639 cases that opened by
2010. This comparative examination showed a rather similar distribution of cases, the
salient difference being the higher rate of settlement prior to certification of the lawsuit
as a class action (22.4%) and a lower rate of dismissal (49.8%).
Alon Klement and Keren Weinshall-Margel18 JITE 172
Table 3
Closing decision Frequency
Voluntary dismissal before the suit’s certification as a class action 57.3%
A settlement before the suit’s certification as a class action 14.7%
Involuntary dismissal following a motion for dismissal (before certification) 3.6%
The court accepted the governmental authority’s cease and desist notice 6.2%
Rejection of the motion to certify the suit as a class action 8%
Dismissal for plaintiff’s lack of action 4.3%
Voluntary dismissal after the suit’s certification as a class action 0.1%
Settlement after the suit’s certification as a class action 0.3%
A judgment after the suit’s certification as a class action 0.8%
Other (consolidation of proceedings, technical) 4.8%
Total 100%
1,379
be approved by the court. If the court finds it proper, it might reject the motion
for voluntary dismissal, and replace the class representative or attorney. Yet, our
findings indicate that courts approved the voluntary dismissal in more than 99% of
these cases.
Motions for voluntary dismissal were coded according to the reasons provided
by the plaintiff. In about 20% of the motions (158 motions), the plaintiff stated
that the case became moot because the defendant satisfied his legal obligations
after the case was filed.16 In these cases, the class action mechanism seems to have
proved effective in realizing the objective of law enforcement, using lower resources
compared to full-blown litigation.
On the other hand, about 60% of the motions for voluntary dismissal (455 mot ions)
were submitted after the plaintiff recognized that the likelihood of a favorable
decision for the class was low. Thus, about a third of all class action cases that were
concluded could be considered frivolous lawsuits.17 Clearly, these cases served no
social objective. The same could be said about the 3.6% of the cases ending with
involuntary dismissal following a motion for dismissal (before certification) and
the 4.3% of the cases that were dismissed due to the class representative’s lack of
action.
16 The remaining 20% are due to various other reasons.
17 This conclusion should be qualified, as the plaintiff might realize the low prob-
ability of success after discovery. However, discovery procedures are rare before class
certification, and therefore cannot explain most voluntary dismissals.
Cost–Benefit Analysis of Class Actions(2016) 19
14.7% of the cases were settled before the suit’s certification as a class action.
These settlements are binding upon all class members as res judicata, deny future
access to courts, and thus require the court’s approval. In practice, of the 222 motions
to approve a settlement, the court approved without change 73% of the settlements
(in 163 cases), 20% were approved after making amendments (in 43 cases), and 7%
were completely rejected (in 16 cases). Most approved settlements included relief
for the represented class, thus realizing to different degrees compensation objectives
(see elaborate data below).
Only 0.3% of the cases – 4 cases – ended with a settlement after the suit was
certified as a class action, and 0.8% of the cases were adjudicated on the merits after
the suit’s certification as a class action. The latter refers to only 10 cases, of which
plaintiffs prevailed in nine and defendants prevailed in only one.
In 32% of the claims against governmental authorities (87 cases, which constitute
6.2% of all resolved cases), the court accepted the authority’s cease and desist notice,
thus resolving the case before certification. These cases promote law enforcement,
yet they do not compensate class members and are limited in their ability to realize
deterrence. (However, the court may award compensation for the class representative
and attorney fees – see below.)
5.2.3 Remedies Awarded
Remedies were awarded in 250 class actions resolved through settlements before
and after certification (176 and 4 cases respectively), cease and desist notices (36
cases), voluntary dismissals (25), or judgments (9). In an additional 176 cases, the
defendant satisfied his legal obligations after the case was filed, presumably as
a result of the filing. Considering the latter, in 426 cases – 30.8% of all resolved
cases – the class or public benefited from the class action. This percentage was
significantly lower than in ordinary civil lawsuits, where plaintiffs received remedy
in 78% of the cases (including those ending in settlements).
Table 4 features descriptive data on the various types of remedies granted. Note
that in 50 of the 250 cases, more than one type of remedy was awarded.
These findings demonstrate the limited effectiveness of the Israeli class action in
realizing its objectives:
Compensating Injured Plaintiffs. In over 77% of the cases in which some mone-
tary remedy was awarded, there was only a partial overlap between class members
who were allegedly harmed, and those receiving the remedy. This was mainly true
for discounts, coupons, and cy-près distributions.
The CAL allows the court to award discounts, coupons, and cy-près donations
as a class remedy. It nevertheless sets strict requirements for using them, includ-
ing a requirement that identifying class members and compensating them would
prove impractical (section 20(c) of the CAL). Nonetheless, our findings indicate
that these remedies were frequently used, undermining the goal of compensa-
tion.
Alon Klement and Keren Weinshall-Margel20 JITE 172
Table 4
Type of Awarded Remedies
Monetary Discounts or Cy-près Injunctions Declaratory
compensation coupons donations remedy
nof claims in
which the remedy
was awarded
49 72 96 99 10
AVE remedy
value in NIS
3,786,889 14,661,525 304,900 – –
SD 8,177,351 74,218,786 522,927
MED 1,500,000 500,000 90,000
In 96 cases the remedy was a cy-près donation, transferred to a public cause rather
than to members of the class. Furthermore, 85% of the donations were channeled
to nonprofit institutions devoted to assisting the infirm, the poor, or children, even
though there was no proximity between the interests of the recipients and those
being pursued by class members.18
In 72 cases the remedy awarded was discounts or coupons, which are mostly
received by the defendants’ future customers, independently of their member-
ship in the represented class. It should be noted that the use of discounts and
coupons appears to be more widespread in Israel than in the U.S. (For example, they
were awarded in less than 10% of class settlements in federal courts (Fitzpatrick,
2010).)
Enforcing the Law and Deterring against its Violation. Law enforcement is re-
alized through injunctive and declaratory relief, awarded in a total of 109 cases.
Additionally, voluntary dismissals after the defendants satisfied their legal duties
were registered in 158 cases. Most of these cases were consumer class actions.
Finally, 87 of the class actions filed against governmental authorities were dis-
missed when the defendants agreed to cease their allegedly unlawful fee collection.
Therefore, overall 25.6% of all concluded class actions were resolved with some
law-enforcement remedy.
The findings about deterrence are less satisfactory. Unlike compensation, deter-
rence does not require any overlap between harmed class members and those who
receive the class action remedy. However, it is presumably affected by the monetary
value of this remedy. Table 4 displays the median and average total amount awarded
in class actions, by type of remedy.The average compensation when direct monetary
remedy was awarded (49 cases) is about NIS 3.8 million, and the median is NIS
18 Compare the requirement in the American Law Institute’s Principles of the Law
of Aggregate Litigation, section 3.07.
Cost–Benefit Analysis of Class Actions(2016) 21
1.5 million. These are relatively low sums, only four times higher than the respective
remedies awarded in civil proceedings in the comparison group, and much lower
than the sums cited in requests for class certification (on average NIS 435 million;
see Table 2).
Furthermore, initial findings indicate that the actual costs for defendants of dis-
counts and coupons are much lower than their reported values. The relatively high
value of the discount and coupon arrangements displayed in Table 4 assumes their
full realization. However, the defendant was ordered to report the actual distribution
of coupons and discounts in only 36% of the 72 cases involving these remedies,
and the few reports that were actually submitted feature a realization rate of less
than 25%. Indeed, the low realization ratio of discounts and coupons may be at-
tributed to lack of incentive mechanisms. For example, in about half of the cases
in which this remedy was awarded, its realization was not automatic but required
additional action by customers. Moreover, the attorney’s fee was contingent upon
the actual realization of the remedy in only 27% of the cases where monetary relief
was awarded.
In regard to cy-près donations, defendants often enjoy reputational benefits when
making these donations. Furthermore, in many instances, the donations would have
been made anyway, regardless of the class action. Thus, it is questionable to what
extent such remedies indeed create a deterrent effect.
Table 5 shows the results of a linear regression model examining the factors in-
fluencing the monetary value of the awarded remedy. The dependent variable is the
recovered remedyin NIS, transformed to logs, in all cases in which monetary reme-
dies were awarded. We used logs to satisfy the regression’s normality assumption,
as remedy amounts were positively skewed The model is significant at an adjusted
R2of 0.34. As expected, the class compensation sought in the motion to certify was
positively correlated with the awarded remedy, but the individual remedy shows
no such correlation. The highest level of compensation is awarded in class actions
against a governmental authority. This finding is surprising, given the CAL’s two-
year limit for total compensation in these class actions, as compared to the usual
seven-year limitation period.
5.2.4 Discussion
Examination of the outcomes of class actions demonstrates their mixed results in
realizing their goals. About one-third of the cases ended in voluntary dismissals, in
which plaintiffs recognized that their likelihood of prevailing was low. Only 3.5%
of the cases ended with direct monetary compensation to class members, and the
average total compensation per class action was small. An additional 12% awarded
discounts, coupons, or cy-près donations, whose realized cost to defendants and
value for plaintiffs is doubtful. Thus, class actions seem to have had limited impact
on compensation and deterrence.
At the same time, 25.6% of class actions provided some form of regulatory
enforcement, either through future-looking injunctive and declaratory remedies,
Alon Klement and Keren Weinshall-Margel22 JITE 172
Table 5
Linear Regression for Remedy Amount
Length of proceeding 0.001∗∗
(0.000)
Number of court hearings −0.04
(0.087)
Voluntary dismissal −1.721∗
(0.837)
Judgment or settlement after certification −0.420
(0.761)
Monetary remedy 1.218∗∗
(0.417)
Donation −1.165∗∗∗
(0.354)
Consumer 2.163∗
(1.09)
Work and discrimination 1.467
(1.39)
Against a governmental authoritya2.79∗∗
(1.348)
Remedy sought for the represented group (log) 0.252∗∗
(0.108)
Individual remedy claimed (log) 0.95
(0.68)
Constant 5.458∗∗
(2.36)
Model sig. 0.0001
Observations 112
Adj. R20.3
Notes: Dependent variable is remedy amount in NIS transformed to log.
∗Significance at 10%; ∗∗ significance at 5%; ∗∗∗ significance at 1%; stan-
dard errors in parentheses. aThe reference category for the case type
dummy variables is “environmental claims.” There were not enough ob-
servations to include corporate claims in the regression model.
through agreement by private (mostly corporate) defendants to satisfy their legal
duties, or by consent of government authorities to cease their allegedly unlawful
collection of fees. This seems to be the most significant social objective satisfied by
Israeli class actions.
Cost–Benefit Analysis of Class Actions(2016) 23
5.3 Attorney Fees, Class Representative Compensation, and Cost Shifting
To realize the objectives of deterrence and compensation, courts must use incentive
schemes that minimize the principal–agent problem and discourage frivolous suits.
This requires attorney fees and representative compensation to increase with their
input of resources, their output of class remedy, and the social value (not necessarily
measured as its economic value) of the class action. In addition, costs may be shifted
to the class representative in cases where lawsuits are evidently frivolous. Yet, our
findings indicate that Israeli courts fail to implement these principles.
5.3.1 Attorney Fees and Class Representative Compensation
Attorneys and class representatives received fees and compensation in 513 cases.
Class remedies were awarded in only 250 cases, and an additional 176 cases ended
with voluntary dismissals after the defendants satisfied their legal duties. Thus,
the class representatives were awarded fees in 87 cases in which the class gained
no benefit. This indicates either that these representatives misused the class action
threat for their own benefit at the expense of class members, or that those lawsuits
had no value for class members, yet defendants were willing to compensate class
representatives to dispose of them.
On average attorney fees were NIS 102,551 (median =40,000, SD =382,442),
and compensation for class representatives was NIS 32,367 (median =10,000,
SD =209,364). As a percentage of the remedy (when monetary remedy was award-
ed), the class representative’s share averaged nearly 17% and the attorney’s fee
was about 28%. The fee percentage is similar to the percentage reported for
class settlement in U.S. federal courts (Eisenberg and Miller, 2004b; Fitzpatrick,
2010).
Table 6 displays the distribution of amounts and percentages of attorney fees
and class representatives’ compensation, according to case resolution. Fees and
compensation were paid in 234 cases that terminated in voluntary dismissal prior to
certification of the class action. One-third of these cases were resolved without any
benefit for the represented group or the public. The table also shows that the average
amount and percentage of compensation and fees were highest in cases resolved
in settlement prior to class action certification. These findings do not accord with
the input principle for awarding compensation and fees. Moreover, they encourage
early settlements and dismissals, instead of discouraging them to overcome the
inherent agency problem. They are therefore inconsistent with the output principle
as well.
In order to test the prevalence of this pattern, we created four regression models for
examining the amount and percentage of fees and compensation. Table 7 presents
these models. Dependent variables in models 1 and 3 are fee or compensation
amounts transformed to logs; the fee or compensation percentages in models 2
and 4 are transformed to square roots. The independent variables include indicators
that correspond with input and output factors.
Alon Klement and Keren Weinshall-Margel24 JITE 172
Table 6
Compensation for Class Representative and Attorney’s Fees, by Type of Remedy
Case outcome Class represen- Class represen- Attorney’s fee Attorney’s fee
tative’s com- tative’s com- (NIS) (% of remedy)
pensation pensation
(NIS) (% of remedy)
Voluntary 13,605 17.1% 35,260 15%
dismissal (n=234 (n=17 (n=234 (n=3
SD =25,589 SD =82% SD =33,998 SD =12.2%
M=5,000) M =15.4%) M =25,000) M =18.7%)
Settlement before 58,930 16.9% 189,761 28.1%
certification (n=195 (n=129 (n=193 (n=107
SD =341,188 SD =28.5% SD =612,654 SD =28.2%
M=20,000) M =6%) M =70,000) M =20%)
Settlement after 21,666 15.9% 51,667 35.3%
certification (n=3(n=2(n=3(n=2
SD =33,998 SD =12.8% SD =28,431 SD =20.8%
M=25,000) M =15.9%) M =60,000) M =35.3%)
Judgment 34,250 8.1% 152,500 22.8%
(n=8(
n=5(n=6(n=6
SD =30,311 SD =5.5% SD =180,354 SD =27.3%
M=25,000) M =7.7%) M =92,500) M =19.8%)
Cease and 23,436 – 85,225 –
desist notice (n=73 (n=74
SD =30,920 SD =33,998
M=12,000) M =25,000)
Direct input factors are positively correlated with the fee and compensation
amounts. Thus, the amounts of attorney fees and representative compensation are
positively correlated with the number of court hearings and the number of days from
filing to closure of the case.
As for output factors, our regressions feature positive correlation between remedy
value and fee and compensation amounts, and a negative correlation between remedy
value and fee and compensation percentages. These findings are similar to those
found in federal class action settlements in the U.S. (Fitzpatrick, 2010; Helland
and Klick, 2007; Eisenberg and Miller, 2004b). In other words, the amounts of
attorney fees and class representative’s compensation rise as the remedy to the class
increases, whereas their percentages decline.
However, findings pertaining to the correlation between representative fees and
class action resolution reinforce our conclusion from the descriptive analysis in
Table 6. Even after allowing for the value of remedy and duration of proceedings,
the attorney’s fee and compensation in settlements concluded prior to the stage of
Cost–Benefit Analysis of Class Actions(2016) 25
Table 7
Regression Models of Fee and Compensation Amounts and Percentages
Model 1: Model 2: Model 3: Model 4:
Fee amount Fee % Compensation Compensation
amount %
Gross remedy 0.02∗∗ −0.162∗∗∗ 0.024∗∗ −0.109∗∗∗
in NIS (log) (0.009) (0.017) (0.01) (0.009)
Length of 0.001∗∗∗ 0.000∗0.001∗∗∗ 0.0008∗
proceeding (0.000) (0.000) (0.000) (0.000)
Number of 0.087∗∗ −0.006 0.101∗∗ −0.004
court hearings (0.032) (0.017) (0.034) (0.000)
Voluntary −0.462∗∗ 0.964∗∗∗ −0.257∗0.676∗∗∗
dismissal (0.151) (0.188) (0.158) (0.106)
Judgment or −0.901∗∗ −1.94 −0.732∗∗ −0.08
settlement after (0.303) (1.48) (0.382) (0.079)
certification
Cease and 0.461 – −0.091 –
desist noticea(0.288) (0.382)
Constant 10.126∗∗∗ 2.5∗∗ 8.84∗∗∗ 1.66∗∗∗
(0.322) (0.230) (0.336) (1.32)
Model sig. 0.0001 0.0001 0.0001 0.0001
Observations 512 112 497 112
Adj. R20.321 0.578 0.328 0.662
Notes: Dependent variables are fee or compensation amount transformed to logs, and fee
or compensation percentage transformed to square roots. Dummy variables for case cate-
gories are included in the model but not reported. A joint test of their significance and each
dummy variable fails to reject the null hypothesis. ∗Significance at 10%; ∗∗ significance at
5%; ∗∗∗ significance at 1%; standard errors in parentheses. aThe reference category for case
resolution variables is precertification settlement.
class certification are higher than in class actions that terminate after certification
– whether through settlement or judgment. Moreover, the percentage of attorney’s
fees and compensation in models 2 and 4 is higher in cases that were resolved in
dismissal than for settlements in the precertification stage. Regression models show,
therefore, that the earlier the class action is resolved, the higher the percentage of
compensation and fees. This might encourage inefficient settlements and frivolous
litigation.
5.3.2 Costs Shifted to Losing Class Representatives
Israeli courts enjoy broad discretion whether to order losing plaintiffs to reimburse
prevailing defendants for their expenses (including attorney fees) and if so, to
Alon Klement and Keren Weinshall-Margel26 JITE 172
determine the amount of such reimbursement. Our findings indicate that the class
representative was ordered to pay the defendant’s expenses in only 161 cases.
This represents about 16.6% of the cases in which no remedy was awarded. In
comparison, court expenses were awarded to the defendant in a similar proportion
(20%) of the civil cases in the district courts when the plaintiff did not win any
remedy. The amount of costs awarded was also similar in class actions and in the
comparison group of civil cases – a median of NIS 15,000 in both; the average
amount in class actions was about NIS 40,000 (SD =197,222), compared to about
NIS 27,000 in other civil cases (SD =33,212). We found no correlation between
the costs awarded and defendants’ actual legal costs, as reflected in the indicators
regarding filings of responses to class certification motions.
Although it is impossible to know what percentage of the class actions that
conclude without remedy are indeed frivolous lawsuits, it seems reasonable to
assume that many of the 455 class actions that ended in voluntary dismissal after
the plaintiff declared that the likelihood that the case will be decided in favor of the
class is low may be regarded as frivolous. The court ordered fees to be shifted to the
plaintiff in only 33 such cases.
5.4 Court Resources Invested in Class Actions
We do not have direct information about the costs of class action proceedings. One
indicator of these costs, however, is judicial time spent on these proceedings. We
combined our findings regarding the number of class action proceedings with other
studies that pertain to the judicial workload in Israeli class actions. A recent study
examining judicial workloads in various types of cases found that class actions are
one of the five most time-consuming case types (of more than 100 case types in
the Israeli judiciary). For example, a district court judge invests an average of about
21 hours in a civil case and about 34 hours in a class action (Weinshall-Margel,
Galon, and Taraboulos, 2015). Figure 3 shows the average time a judge invests in
class actions according to case resolution and different stages in processing the case.
The figure indicates that judges invest a lot of time in cases ending after the
suit’s certification as a class action – 106 and 99 hours on average when the case
ends with judgment or settlement, respectively. Yet, judges also invest a substantial
amount of time in cases that end without serving many of the class objectives.
For example, it takes a judge 12 hours to process cases ending with involuntary
dismissals, and 7 hours for the majority of cases ending with voluntary dismissals
before certification. In fact, in voluntary dismissals due to the plaintiff’s recognition
of the small chance of winning, the judicial time invested was even higher, with 1.3
lengthy hearings on average compared to 0.9 such hearings in other cases ending
with voluntary dismissals.
Our data indicates that class actions are conducted largely via written motions.
An average of 12 motions are submitted in class actions (median =10, SD =5),
compared to 10 motions in civil cases. Judges also report a significant difference in
the amount of time invested in each motion: nearly two hours of work for class action
Cost–Benefit Analysis of Class Actions(2016) 27
Figure 3
Average Judicial Time for Processing Class Actions, by Case Resolution
motions, compared to less than an hour in civil cases.19 It seems that lawyers too
invest considerable resources in litigating class actions. For example, as an indicator
for the time invested by lawyers, we found that the average number of pages in
motions to certify a class action is 17.9 (median =15, SD =10.6), compared to 7.9
pages submitted to courts on average in a district civil claim (median =7, SD =5.6).
We do not have other indicators for the time invested by the sides in class actions,
but we assume it correlates with the high judicial inputs.20
Our findings indicate that class actions impose a disproportionately heavy work-
load on the justice system. In 2012, class actions comprised less than 0.5% of
the number of civil and administrative cases in the justice system, but imposed
a workload that was six times that size – 3% of the total workload.
6 Conclusion
Six years after its enactment, the CAL has generated a significant amount of class
action litigation. However, our findings indicate that its success in realizing its goals
19 See the method of measuring judicial invested times in Weinshall-Margel, Galon,
and Taraboulos (2015, pp. 20–26).
20 When calculating the costs of class actions it is also necessary to take into ac-
count the secondary litigation they generate – and appeal cases in particular. Most of
these appeals are heard by the Israeli Supreme Court. Data provided by the courts in-
dicates that from August 2010 through August 2013, 26 appeals were filed on interim
decisions in class actions and 37 appeals were submitted on final class action judg-
ments. Appeals of final judgments are especially costly, as they are decided by a panel
of three Supreme Court justices. However, we do not have information on the results
of these proceedings or the precise input invested in them.
Alon Klement and Keren Weinshall-Margel28 JITE 172
is limited. The law’s main contribution lies in facilitating some regulatory enforce-
ment as well as a potentially low level of deterrence, mostly in consumer law and in
restitution lawsuits against government authorities. Yet, during these first six years,
the CAL has failed to produce a significant amount of litigation in the important
categories of corporate (including antitrust), employment and discrimination, and
environmental lawsuits.
The costs of class actions during these six years was not insubstantial, as their
litigation consumed resources by plaintiffs, defendants, and courts. Moreover, about
one-third of the cases were recognized by plaintiffs as having no merit, and were
thus voluntarily dismissed. These cases produced a net social loss.
In order to improve the benefit–cost ratio of Israeli class actions, courts should take
steps to encourage filings in categories that may provide significant social value – in
deterrence compensation and access to courts. At the same time, courts should better
screen and discourage frivolous lawsuits. These policy goals may be obtained using
the procedural mechanisms of class certification, settlement approval, attorney fees,
and plaintiff compensation, as well as cost shifting. Our findings indicate that court
practices in implementing these mechanisms do not always conform to principles
that would guarantee their effectiveness in generating greater net social value from
the CAL.
References
Carroll, S. J., D. Hensler, J. Gross, E. M. Sloss, M. Schonlau, A. Abrahamse, and J. S.
Ashwood (2005), Asbestos Litigation, RAND Corporation, Santa Monica (CA).
Choi, S. J. (2007), “Do the Merits Matter Less after the Private Securities Litigation Reform
Act?” The Journal of Law, Economics, & Organization, 23(3), 598–626.
Eisenberg, T., and C. Engel (2014), “Assuring Civil Damages Adequately Deter: A Public
Good Experiment,” Journal of Empirical Legal Studies, 11(2) 301–349.
— and G. Miller (2004a), “The Role of Opt-Outs and Objectors in Class Action Litigation:
Theoretical and Empirical Issues,” Vanderbilt Law Review, 57(5), 1529–1567.
— and G. P. Miller (2004b), “Attorney Fees in Class Action Settlements: An Empirical Study,”
Journal of Empirical Legal Studies, 1(1) 27–78.
— and — (2010), “Attorney Fees and Expenses in Class Action Settlements: 1993–2008,”
Journal of Empirical Legal Studies, 7(2), 248–281.
Fitzpatrick, B. T. (2010), “An Empirical Study of Class Action Settlements and their Fee
Awa rds,” Journal of Empirical Legal Studies, 7(4), 811–846.
Helland, E., and J. Klick (2007), “The Effect of Judicial Expedience on Attorney Fees in
Class Actions,” The Journal of Legal Studies, 36(1), 171–187.
Hensler, D. R., N. M. Pace,B. Dombey-Moore, B. Giddens, J. Gross, and E. K. Moller (2000),
“Class Action Dilemmas: Pursuing Public Goals for Private Gain,” RAND, Santa Monica
(CA).
Klement, A. (2011), “Class Action Settlements” (in Hebrew), Mishpatim, 41, 5–95.
Lee III, E. G., and T. E. Willging (2008), “The Impact of the Class Action Fairness Act
on the Federal Courts: An Empirical Analysis of Filings and Removals,” University of
Pennsylvania Law Review, 156, 1723–1764.
Morabito, V. (2009), “An Empirical Study of Australia’s Class Action Regimes, First
Report: Class Action Facts and Figures,” http://globalclassactions.stanford.edu/sites/
default/files/documents/Australia_Empirical_Morabito_2009_Dec.pdf, last accessed Sep-
tember 23, 2015.
Cost–Benefit Analysis of Class Actions(2016) 29
— (2010), “An Empirical Study of Australia’s Class Action Regimes, Second Re-
port: Litigation Funders, Competing Class Actions, Opt Out Rates, Victorian Class
Actions and Class Representatives,” http://globalclassactions.stanford.edu/sites/default/
files/documents/Vince%20Morabito%202nd%20Report.pdf, last accessed September 23,
2015.
Niemic, B., and T. Willging (2002), “Effects of Amchem/Ortiz on the Filing of Federal Class
Actions,” Report to the Advisory Committee on Civil Rules, The Federal Judicial Center,
Washington (DC).
Pritchard, A. C., and J. P. Sarra (2010), “Securities Class Actions Move North: A Doctrinal
and Empirical Analysis of Securities Class Actions in Canada,” Alberta Law Review, 47(4),
881–926.
Shavell, S. (2004), Foundation of Economic Analysis of Law, Harvard University Press,
Cambridge (MA).
Weinshall-Margel, K., I. Galon, and I. Taraboulos (2015), “Measuring Judicial Workload” (in
Hebrew), Mishpatim, 44, 769–834.
— and I. Taraboulos (2014), “Israeli Civil Proceedings Dataset,” Israeli Courts Re-
search Division, http://elyon1.court.gov.il/heb/Research%20Division/doc/Israeli_Civil_
Proceedings_Dataset.zip, last accessed August 7, 2015.
Willging, T. E., L. L. Hooper, and R. J. Niemic (1996), “Empirical Study of Class Actions
in Four Federal District Courts,” Final Report to the Advisory Committee on Civil Rules,
The Federal Judicial Center, Washington (DC).
— and S. R. Wheatman (2006), “Attorney Choice of Forum in Class Action Litigation: What
Difference Does it Make,” Notre Dame Law Review, 81(2), 591–654.
Alon Klement
Tel-Aviv University
39040 Tel Aviv
Israel
klement@tauex.tau.ac.il
Keren Weinshall-Margel
The Hebrew University of Jerusalem
91905 Jerusalem
Israel
keren.weinshall@mail.huji.ac.il
