Literature Review: Final Offer Arbitration

Table of Contents

Table of Contents................................................................................................................... 1

Table of Figures..................................................................................................................... 1

I. INTRODUCTION............................................................................................................. 2

Labor Code 4065............................................................................................................. 2

II. HISTORY OF §4065 AND ITS INTERPRETATION........................................................ 2

III. REVIEW OF THE LITERATURE................................................................................... 3

A. Increasing the cost of arbitration to increase the likelihood of settlement......................... 4

B. Convergence versus divergence in offers...................................................................... 5

C. Convergence and preferred settlements........................................................................ 5

D. Divergence in outcomes.............................................................................................. 6

E. Risk aversion and lower offers..................................................................................... 6

F. Offers as information................................................................................................... 7


A. Caveats to generalizing from the literature to workers' compensation.............................. 7

B. Final offer arbitration in workers' compensation............................................................. 8

C. Narrowing the uncertainty about decisions and increasing probability of settlement........ 10


A. Data Description....................................................................................................... 10

B. Differences by office of region................................................................................... 11

C. Does baseball arbitration cause opposing offers to converge?....................................... 12

D. Does Baseball Arbitration cause awards for similar disabilities to diverge?.................... 13

V. DISCUSSION................................................................................................................. 15

Recommendations.......................................................................................................... 16

Bibliography......................................................................................................................... 18


Table of Figures

Figure 1: Hypothetical distribution of awards for the same disability………………………..9

Figure 2: Offers under baseball arbitration in sample………………………………………12

Figure 3: Mean rating: Baseball arbitration and contested claims at DEU………………...13

Figure 4: Mean and variance of awards: Baseball arbitration and conventional…………...13

Figure 5: Distribution of final ratings: Baseball arbitration v. all other………………….…14

Figure 6: Value of PD rating valued at maximum weekly earnings……………………….…15


Final offer arbitration was introduced into the workers' compensation decision process as a result of the 1993 reforms. Labor code 4065 is often referred to as "baseball arbitration" because it is the same type of arbitration that is used for resolve baseball contract salary disputes.  Under final offer arbitration, the workers' compensation judge (WCJ) is restricted, in determining a permanent disability rating, to selecting the final offer of one of the parties.  This contrasts with the process pursued in the majority of permanent disability decisions where the WCJ has the discretion to award permanent disability on the basis of the 'range of evidence'.

California Labor Code §4065. 

(a) In cases where either the employer or the employee have obtained evaluations of the employee's permanent impairment and limitations from a qualified medical evaluator under Section 4061 and either party contests the comprehensive medical evaluation of the other party, the workers' compensation judge or the appeals board shall be limited to choosing between either party's proposed permanent disability rating.

(b) The employee's permanent disability benefit awarded under paragraph (a) shall be adjusted based on the disability rating selected by the appeals board.  If the appeals board chooses the permanent disability rating recommended by the employer, then the employee's permanent disability benefit award shall be reduced by the cost of the employee's comprehensive medical-legal evaluation.  If the judge chooses the permanent disability rating recommended by the employee, the permanent disability benefit award shall be increased by the cost of the employer's comprehensive medical-legal evaluation.


            In this report, we will begin by reviewing the history of Labor Code §4065 and its implementation and interpretation.  Then we will review the current state of understanding within the academic literature concerning the impact of final offer arbitration on various characteristics of the decision process and outcomes, both settlements and awards.  We then apply and contrast this research, conducted in other venues, to the characteristics of workers' compensation.  In the following section we review the findings based on a sample of 400 decisions handed down by workers' compensation judges during a two-week period in August of 1999.  Finally, we conclude with a discussion and recommendations. 



            Section 4065 was part of a package of reforms adopted by the 1993 Legislature which also included provisions changing the appellation of worker's compensation judges to referee, restricting their gifts, honoraria, and travel, and requiring them to participate in continuing education courses including ethics and conflict of interest.  The general tenor of the legislation suggests that at least some of the legislators lacked confidence in the ability of the WCJs to make just awards.  Consistent with this supposition is information received by the Commission from a representative of the Division of Workers" Compensation that the purpose was to preclude decisions based on "the range of the evidence."  In Minniear v. Mt. San Antonio Community College Dist. (1996) 24 CWCR 261, 61 CCC 1055, the Appeals Board said that the purpose of §§4060-4065 was to expedite resolution of medical issues by restricting the number of medical-legal evaluations. 

            To implement §4065, the Appeals Board adopted new WCAB Rules 10632 and 10633.  The former provides:

Where the provisions of Labor Code Section 4065 apply, the workers' compensation judge shall receive into evidence both qualified medical evaluators' evaluations and the evaluation(s) of the treating physician(s), together with the "proposed ratings" submitted by the parties.

Rule 10633 specifies the content of proposed ratings and requires the WCJ to give the reasons for selecting one rating over the other.  Pursuant to Rule 10633, a WCJ is permitted to make a determination as to how he or she would rate the case if unfettered by §4065.  Sheppard v. WCAB (1997) 62 CCC 993 (writ denied). 

            The language of the WCAB Rules and at least subdivision (a) of §4065 is relatively straightforward and very little interpretative case law has developed.  An issue that did arise and was decided is that a proposed rating can be based on a comprehensive medical evaluation by the primary treating physician. Gallo Glass Co. v. WCAB (Reeves) (1998) 63 CCC 450 (writ denied).

            In a recent case, Fay v. Golden Eagle Ins. Co., WCK 27669, Feb. 24, 1999, initiated under §4060, to which neither baseball arbitration nor the treating physician presumption applies, the Appeals Board reaffirmed the range of the evidence rule.  This rule, which was given the blessing of the California Supreme Court in Liberty Mut. Ins. Co. v. IAC  (Serafin)(1948) 33 C2d 89, 13 CCC 267, and has been followed ever since, permits the trier of fact to resolve conflicts in the medical evidence by finding any degree of disability between the extremes established by the medical opinion.



Final offer arbitration first suggested by Stevens (1966) has been used since the 1970s almost exclusively in two venues, 1) to settle public sector labor contract issues where the right to strike is not available to labor, and 2) within Major League Baseball (MLB) to resolve salary disputes for players who are not yet eligible for free agency and can only bargain with their current club.  Within these two venues, final offer arbitration (FOA) also know as baseball arbitration has been studied extensively.  The studies involve both theoretical modeling of the various parties' decision processes in the face of uncertainty, and empirical studies evaluating actual offer and award data.  The findings of these studies offer insight into how baseball arbitration can be expected to affect the process within workers compensation and the expected outcomes for each party.

            In this discussion we will consider the equivalence of three decision processes:   1) conventional arbitration where the arbitrator is constrained only in that the decision cannot be outside the two offers proposed by the parties, 2) FOA where the arbitrator is constrained to choose one offer or the other, and 3) negotiated settlements where the parties reach a decision prior to and often under the 'threat' of arbitration.  For purposes of this discussion, we will follow the approach of Ashenfelter and Bloom (1984) and assume it is reasonable to equate the decision by a Workers' Compensation Judge (WCJ) at hearing with conventional arbitration.[1]

            There are several areas across which researchers attempt to evaluate decision processes.  First, how do they affect the likelihood of settlement?  Second, how do they affect the quality of the award in the absence of settlement, that is, generally, how does the award compare to the potential range of negotiated settlements?  Third, are parties affected differently by the decision process, potentially disadvantaging one party against the other?  The following section explores the current research on each of these issues.  In the subsequent section, parallels and distinctions are drawn between this research and the implementation of FOA in workers' compensation.

A. Increasing the cost of arbitration to increase the likelihood of settlement

            It is generally accepted that negotiated settlements are superior to arbitrated settlements.  First the parties reach a mutually acceptable agreement.  Second, the cost of arbitration is avoided.  Consequently, one criteria for comparing decision processes is the their relative impact on the probability that the parties will settle prior to arbitration. The final decision process affects the probability of settlement by narrowing (or widening) the differences between the parties' positions, and increasing (or decreasing) the costs, both indirect and direct, of failure to settle.

            The final decision process, whether it be the traditional tort process, or an arbitration alternative motivates the parties to settle by defining the likely range of outcomes.  (See Mnookin and Kornhauser, 1979).  Under arbitration an independent third party imposes a binding settlement on the two negotiating parties.  Because there is a range of outcomes and the actual outcome under arbitration is not known with certainty, ex ante, the uncertainty about the decision imposes costs on the parties.  This cost can be avoided if the parties come to an agreement prior to the arbitration deadline.  Hence, one objective of interest arbitration is to provide an incentive to the parties to negotiate their own agreement, rather than risk an unfavorable outcome at arbitration.

Stevens (1966) has argued that FOA "generates just the kind of uncertainty…that is calculated…to compel them [the parties] to seek security in agreement."   Farber (1980) emphasizes that "while direct cost may be important, the primary channel through which FOA imposes cost on the parties is fundamentally different.  It is uncertainty concerning the arbitrator's award combined with the risk aversion of the parties which is hypothesized to make /FOA a costly alternative."

Early proponents of FOA argued that given risk averse parties, offers would tend to converge towards the arbitrators expected award.  This convergence in turn would reveal more information about the parties' preferences for a negotiated settlement and consequently lead to more frequent settlement.  

Chelius and Dworkin (1980) argue that FOA raises the costs of failing to reach a negotiated settlement, thus increasing the likelihood the parties will settle. The potential outcomes under FOA impose higher costs because the final offers of the parties are outside the zone where both parties would prefer to settle, called the contract area.  On the other hand, based on theoretical modeling of parties actions, Farber (1980) argues (as discussed in the next section) that FOA reduces the likelihood of settlement because final offers will diverge from those under conventional arbitration. Consequently, the offers may contain less information about parties' actual preferences about a negotiated settlement and reduce the likelihood that the parties will agree.

The question is essentially an empirical one because the two effects operate in different directions. Based on empirical evidence, most authors argue for FOA leading to more frequent settlement.  Farber and Bazerman (1989) find empirical evidence supports the contention that FOA results a higher proportion of negotiated settlements, and cite both laboratory studies (Neale and Bazerman, 1983; Grigsby and Bigoness, 1982, and Notz and Stark, 1978) and field studies (Kochan and Baderschneider, 1978). Coleman, Jennings, and McLughlin (1992) also find empirical evidence in several settings supporting more frequent settlement under FOA.[2]

B. Convergence versus divergence in offers

Early proponents of final offer arbitration argued that FOA would lead to convergence in the offers of the two parties (Donn 1977; Feuille, 1975; Olson, 1979; Staudohar, 1979; Farber and Bazerman, 1989).   The theory originating with Stevens (1966) was that conventional arbitration had a "chilling" effect on negotiations and offers because the parties were motivated to make extreme offers when facing an arbitrator who was thought to "split the difference."  Subsequent research tended to support the contention that conventional arbitration causes parties to make more extreme offers under the assumption that the offers affect arbitrators' decisions (Overton and Wortman, 1974, Rhemus 1976; and Wheeler, 1974)

Coleman, Jennings, and McLaughlin (1992) however, find "there is significant disagreement between labor relations theorists and those in the decision sciences regarding final-offer arbitration' convergence properties. Decision scientists, upon examining the mathematical properties of the process, have widely determined that the claim of convergence is not justified (Brams and Merril, 1983; Whitman, 1986). Rather they argue that offers under FOA will diverge from those under conventional arbitration. Coleman, et. al. go on to argue that convergence, which they expect based on empirical evidence, can be made consistent with models that dictate divergence if the parties have other motivations involving continued relations between parties (employment relationship, player motivation) that are outside the single dimension (e.g., salary) that is being arbitrated.

C. Convergence and preferred settlements

However, convergence of the offers under FOA compared to conventional arbitration is not a sufficient condition for 'better' decisions by the arbitrator given that the arbitrator can choose only one or the other.  'Better' here meaning that the arbitrator's decision results in an award that would have been equivalent or preferred, ex ante, by both parties over their final positions and the related uncertainty about the final outcome.  Farber (1979) constructs a model of the parties' decision processes in which parties' decisions depend on uncertainty about the arbitrator's preferences, expectations about the other party's preferences, and the relative risk-aversion of each party.  Variations of this model have been used by numerous other researchers. The model results in a Nash Equilibrium, that is, a pair of offers where neither party can improve their expected outcome at arbitration by raising or lowering their offer.  At equilibrium, the offers of each party still fall outside of the area where both parties would be willing to settle, referred to as the contract space, where the combined utility of both parties would be superior to that of either offer.  Curry (1993) finds the same model applies equally well to FOA and conventional arbitration, that is, under both types of arbitration, the offers at arbitration will be outside of the contract space. 

D. Divergence in outcomes

In the absence of convergence between parties' offers under FAO relative to conventional arbitration, the outcomes faced by parties will be distributed in a manner that has important policy implications.  Since the arbitrator under FOA is restricted to a choice between the offers, in the absence of convergence in the offers the outcomes will not only diverge from the preferred contract space, but the awards will show greater variance than under conventional arbitration.  That is parties in similar circumstances will be subject to substantially different awards based on small differences in either party's final offer and/or the preferences of the arbitrator.  Burgess and Marbarger (1993) find just such divergence in the outcomes when analyzing Major League Baseball arbitration relative to negotiated settlements.  When management wins offers were 9% lower than negotiated settlements for comparable players.  When players win, the decision was 14% higher than comparable players.  However, this comparison is between FOA and negotiated settlements, and not FOA and conventional arbitration.

In the absence of greater convergence in the offers under FOA than conventional arbitration, FOA has to result in increased variance in the outcomes.  The arbitrator has to select one offer or the other. It remains an empirical question whether in a given venue, such as workers' compensation, conventional arbitration offers can diverge sufficiently from both the contract space and offers under FOA that they drive arbitrators' decisions outside the contract zone and result in greater variance in awards for similar circumstances than FOA.  To find equal or greater variance for conventional arbitration awards, researchers need to understand two characteristics, the divergence of the offers under conventional arbitration relative to FOA and the distribution of arbitrators decisions in conventional arbitration in the face of divergent offers and the ability to compromise between them.

E. Risk aversion and lower offers

Related to the issue of variance in outcomes under FOA is the impact of this variance on specific parties. If risk aversion affects offers, and if offers affect decisions, and if risk aversion is unevenly distributed between parties, FOA may reduce outcomes on average for one side.

Ashenfelter and Bloom (1983) and Farber (1980) argue that the more risk averse party will on average make a lower, more concessionary offer.  Somers (1977), Olson (1978) and Bloom (1979) find that the more risk averse party in public sector negotiations wins more often.  Consequently, research suggests that the more risk averse party will win more often under FOA, but will 'win' less in each instance.  Neither group of studies examines the combined effect of the lower offer and the increased probability of winning.

Curry (1993) argues, based on a theoretical model, that the more risk averse party receives, on average, less compensation.  That is, the combined effect of the lower offer and increased probability of winning works on average to shift a portion of the 'pie' from the more risk averse to the less risk averse party.

F. Offers as information

As discussed earlier, offers represent information about each party's preferences.  This information is meant to signal (or misinform) the opposing side about a party's preference and lead, hopefully, to a favorable settlement. However, in the absence of a settlement, final offers become information upon which an arbitrator may base a decision.

Bazerman and Farber (1985) have tested the question of whether arbitrators preferences are affected by the offers of the parties, independent of the facts of the case.    They find that independent of the facts, offers do affect arbitrators' decisions of the correct decision.  This issue is problematic in baseball arbitration and public sector labor disputes where the arbitrator is expected to form his preferences based, in the most part, on factors that are known quantities, separate from the offers of the parties. (See Dworkin and Chellius for a discussion of the factors under FOA in MLB.).

            In conventional arbitration, the mechanism that drives parties to propose more extreme offers is precisely this expectation that offers affect arbitrators' decisions, even when the informational content of the offers should be non-existent.  However, when analyzing arbitrator behavior, Bazerman and Farber (1986) find that arbitrators treat the informational content of the offers as an inverse function of the distance between offers.  That is, as offers become more extreme, at least relative to each other, arbitrators act as thought the offers contain less information about the correct settlement. Consequently, one final value to FOA is that, if it is found that offers converge under FOA compared to conventional arbitration, the offers would also have more information content under final offer arbitration than under conventional arbitration.  Thus, the more important offers are as information in a particular venue, the more important it is that the offers converge.


A. Caveats to generalizing from the literature to workers' compensation

It should be noted that conclusions about FOA in these two well-studied venues (public sector labor disputes and Major League Baseball) should be drawn cautiously when evaluating baseball arbitration in workers' compensation.  First, public sector contracts involve incremental changes to prior contracts, negotiated by experienced and informed parties.  These often involve repeat arbitration over a period of years.  Both labor negotiations and MLB involve considerations by both parties of a continuing employment relationship between parties that may enter into consideration of the quality of final offers.  The informational content of the actual offers at arbitration is often of secondary importance to or even independent of an arbitrator's decision process. 

Within the workers' compensation process before the WCAB, one party may be poorly informed and inexperienced at the arbitration process. As will be shown in the study section, in the majority of cases, the differences between final offers represent very large potential differences in outcomes, differences that can not be mitigated by future awards as is possible with repeat bargaining.  Also, the employment relationship between the parties has more often than not ended prior to arbitration, which in any case is often between the worker and a third party (insurer, TPA).  Consequently there is little to mitigate either party's effort to maximize the award at the other party's expense.

Finally, and most critical to generalizing from the models considered above, within workers compensation, most of the information for the arbitrator's final decision is contained within the offers in that the each offer is connected to a competing medical-legal report.

In addition, Ashenfelter's (1987) theoretical framework requires that arbitrators are exchangeable. Consequently, Ashenfelter suggests that,  "Since both management and the union have veto power in selecting arbitrators, arbitrators have an incentive to render decisions that echo those of past arbitrators." This theoretical condition cannot be imposed in modeling workers' compensation where the arbitrators income is not based on workload, but rather selection of the arbitrator is to a first approximation, based on random assignment. 

B. Final offer arbitration in workers' compensation

In workers' compensation, the insurer is a repeat player, and consequently spreads the risk of any decision over many cases.  The injured worker on the other hand faces a one-time event. In this sense, workers' compensation differs substantially from public sector labor negotiations, which involve incremental changes and repeat negotiations, and MLB where the parties both face substantial impacts of infrequent decisions. Injured workers face decisions that can mean the difference between an award of a few hundred dollars and one of tens of thousands of dollars.  (In the following section we will detail the differences in dollars represented by the final offers in a sample of baseball arbitrations drawn from the WCAB.) Consequently, the worker is likely, on average, to be the more risk averse party. 

The evidence in the literature strongly supports the contention that FOA increases the probability of settlement.  (This could be confirmed by examining the pattern of settlements at different WCAB offices with different customs in applying baseball arbitration.)  As noted above, settlements are generally preferred to arbitration decisions because they avoid the uncertainty imposed by the arbitrator.  However, if as Mnookin in and Kornhauser (1979) propose parties "bargain in the shadow of the law", then negotiated settlements in the face of FOA are likely to reflect, on average, the characteristics of the arbitrated settlements.  If so, then an arbitration process that favors the less risk averse party will lead to negotiated settlements that are likewise more favorable to the less risk averse party.  Consequently, more frequent settlements are an important but not a sufficient quality for identifying a preferred system. The quality of the settlements may outweigh the advantages of higher settlement rates.

            The following diagram suggest a possible distribution of awards under baseball arbitration and conventional arbitration (decisions by a WCJ).  The figure demonstrates why risk plays such an important role under final offer arbitration. This diagram is only meant to be expository and is not based on underlying data.  However, it is suggestive of the distribution of outcomes for similar disabilities that might result under baseball arbitration in workers' compensation.  In this example, both distributions have the same average award (35.5%).  However, in this example, under conventional arbitration, 95% of the injured workers receive a settlement within 4.5% of their 'true' disability (the mean).  Under baseball arbitration, 2/3rds of the injured workers in this hypothetical example receive a settlement that is more than 4.5% higher or lower than their 'true' disability. On average, justice would look the same (35.5% rating) but at the individual level, a substantial number of workers would be under or over compensated.

Figure 1: Hypothetical distribution of awards for the same disability

Figure 1: Hypothetical distribution of awards for the same disability

Risk aversion on the part of injured workers poses an additional issue.  If risk aversion results in lower average awards to the more risk averse party, as suggested by the models of behavior discussed in the literature section, then workers on average will settle for or be awarded lower permanent disability ratings in the face of final offer arbitration.  The spreading of the distribution of awards under baseball arbitration, both higher and lower than the distribution under conventional arbitration will tend to mitigate, in dollar terms, the effect of risk aversion on the average award.  This is because as the PD% increases, the dollar value of the award increases more rapidly.  This may have the effect of increasing the total dollars awarded under baseball arbitration. But, it will also exacerbate the difference in settlements faced by similarly disabled workers.

C. Narrowing the uncertainty about decisions and increasing probability of settlement

Farber (1980) argues that the contract zone depends in part on the uncertainty about the preferences of the arbitrator.  The contract zone can be shrunk, divergence in awards reduced, and the likelihood of settlement increased as the uncertainty about the arbitrator preference is reduced.

The uncertainty with respect to arbitrators' awards is magnified in workers' compensation by a total lack of published information about the distribution of these awards.  Expanding on the findings of this study would create information about the distribution of WCJs' awards.  This would likely improve the frequency and quality of negotiated settlements and increase the convergence of the offers proposed by the parties when negotiation fails.

            In addition, empirical evidence contradicting anecdotal evidence of arbitrator's tendency to 'split the difference' might also improve the quality of offers in conventional arbitration.  The analysis of arbitrators (WCJs) decisions conducted within this study, but not reported here, found a significant number of decisions where the arbitrator choose the offer of one party rather than compromising between the offers based on the "range of evidence."  This is consistent with experimental studies of arbitrator behavior in other venues which find a substantial tendency of arbitrators to award an amount exactly equal to one party's offer.  Anecdotal evidence in workers' compensation is that the WCJ splits the difference, while empirical evidence may lead to different conclusions. Knowledge of this empirical evidence may lead to better decision making by the parties.

In analyzing the outcomes within workers' compensation, researchers have an opportunity to compare outcomes under both forms of arbitration. The results section of this report, based on a small sample of FOA arbitrations, will present data that suggests that outcomes diverge under FOA compared to conventional arbitration.  This is problematic where the system is meant to deliver "substantial justice' to each injured worker.



A. Data Description

The data set for the study was drawn from all decisions (Findings and Awards, Findings and Orders) handed down by WCJs during a two-week period (8/9/99-8/20/99).  Eliminated at the start were decisions approving or suspending Compromise and Release Agreements (C&Rs) and Stipulated Settlements (Stips) and decisions on issues such as taking a case off calendar, lien trials, and similar decisions that did not involve resolution of the case-in-chief. This left a sample of 329 decisions.

In addition, 43% of the remaining decisions did not involve issues related to permanent disability. For example a large number of cases involved expedited hearings over medical treatment.  Another 17% of cases involved an issue of AOE/COE that was decided against the applicant (making the PD issue moot).  This left 134 decisions in the sample where PD was an issue.


Finding: Of all decisions involving Permanent Disability awards (including findings of disability = 0%), 12% were resolved through baseball arbitration.


This sample could be drawn more narrowly. It is appropriate to exclude the claims were PD was resolved without competing reports, for example, based on an AME report, or only the treating physicians report, or where the date of injury was prior to the introduction of baseball arbitration. 


Finding: Of all decisions involving permanent disability awards where baseball arbitration could have been used, that is dates of injury after 1/1/94, and involving competing medical legal reports, 20% were resolved using baseball arbitration. 


In another 9% of the decisions, baseball arbitration may have been used, but the decision did not positively state that baseball arbitration was involved.  Baseball arbitration was assumed only if the WCJ 1) indicated "Baseball Arbitration", 2) indicated Labor Code Section "4065", or 3) that the decision in favor of one doctor's report caused the WCJ to shift the cost of the medical-legal report between the applicant and the defense.  In 9% of the cases the judge indicated a decision in favor of one doctor's report over another but did not indicate any of the above three characteristics.

Consequently, for this sample of claims, baseball arbitration was used at least 20% and possibly as high as 30% of the time when it was the appropriate method for resolving the issue of the permanent disability rating.  This number is substantially higher than earlier estimates based expectations of parties and interviews with WCJs.

These data suggest that the true frequency of the use of baseball arbitration as perceived by system participants is being obscured by the extraordinary number of other issues that WCJs are called upon to decide in the normal course of business, and the number of instances where permanent disability decisions cannot be determined by baseball arbitration because of limitations imposed by statute.

These data confirm that while baseball arbitration is being used in a substantial minority of cases, it is being avoided in the majority of cases.   


B. Differences by office of region. 

Based on the small sample and limited time frame it is impossible to reach conclusions regarding any regional differences in the use of baseball arbitration.  However, no office or region was clearly outside reasonable expectations based on overall frequency. The rates of baseball arbitration were almost identical between Northern and Southern California (18.5% and 20.7%).


C. Does baseball arbitration cause opposing offers to converge?

Baseball arbitration is meant to bring the parties closer together in their final offers, making the decision of the WCJ one that is between less extreme offers. The anecdotal wisdom is that in the absence of final offer arbitration, the parties act as though the WCJ will 'split the difference' between two competing PD estimates.  Consequently, when the WCJ will decide on the 'range of evidence' the parties may adopt more extreme positions, at least up to the point that the position will be discounted by the WCJ.[3] 

            Baseball arbitration is meant to reduce this tendency to choose extreme evaluators or to present extreme offers.  However, this conclusion is not supported by research in other venues, as described in the previous literature review section. 

Below we present a very preliminary look at the distribution of offers across the set of Findings and Awards that involved baseball arbitration. The data on actual decisions is very limited.  The sample includes the 11 baseball arbitrations where the offers of both parties were given along with the final award.  These are listed in the following table.

Figure 2: Offers under baseball arbitration in sample


Defense Rating

Applicant Rating

'Winning' party



































These data at first glance to not suggest that the offers are very close for those claims that are resolved through baseball arbitration.  We cannot conclude from these data that these differences are any larger than for offers under conventional resolution by a WCJ, because there are no data currently available offers made in cases that go to hearings and are decided without baseball arbitration.  This would be the best comparison for ratings offered in baseball arbitration. 

However, data is available from the Disability Evaluation Unit database covering all claims where both applicant and defense evaluations were submitted for rating.  In the following table (Figure 3) we present data on the average differences in competing ratings in the DEU data and for the list of offers at baseball arbitration (from Figure 2). The table also gives the average rating for each group. 

Figure 3: Mean rating under baseball arbitration and contested claims at DEU




Mean Rating


Mean difference between applicant and defense rating

Claims rated by DEU for both applicant and defense evaluations





Baseball Arbitration offers







The difference in ratings, on average, is higher in baseball arbitration, but the mean rating is also higher, likely reflecting the tendency of more severe cases to go to trial.  Higher ratings (more serious claims) offer more opportunity for larger differences in the opposing ratings.  That said, there is not an indication in this small dataset that the final offer arbitration process results in a narrowing of the differences in ratings offered by the opposing parties. 

D. Does Baseball Arbitration cause awards for similar disabilities to diverge?

            As indicated in the background research, in the absence of convergence in the final offers under final offer arbitration relative to conventional arbitration, the outcomes must diverge.  In the small sample available, we do not find convergence in offers, suggesting the variance in final awards for similar injuries will increase under baseball arbitration.

            Figure 4 compares the average award under baseball arbitration with the average award for all other cases that involved permanent disability awards. Consistent with expectations from the literature there was no difference in the mean award under baseball arbitration.  However, the variance (as reflected in the Standard Error) was higher for awards under baseball arbitration.

Figure 4:

Mean and variance of awards under baseball arbitration and conventional arbitration


Mean Award

Std. Error

Baseball Arbitration



Not Baseball Arbitration



This variance is further demonstrated in the chart below, Figure 5. As displayed, the awards in this small sample result in a disturbing pattern. The distribution of ratings when the case does not involve baseball arbitration follows an expected curve that shows frequent smaller ratings and declines steadily as the rating level increases. The pattern in baseball arbitration could be flatter or even bimodal, that is the ratings may be more variant so that smaller ratings and larger ratings are more frequent and intermediate ratings less frequent.

The distribution displayed in Figure 5 involves all awards in the sample. This is unlike Figure 1, which is a hypothetical example of the distribution of awards under baseball arbitration where all claims have the same underlying disability.  However, if baseball arbitration results in the kind of variance in awards for similar disabilities hypothesized in Figure 1, the resulting distribution for all ratings would tend to look like the distribution in Figure 5. 

If the same result were found in a large sample of claims, this would be a troubling finding. First, the average rating may be the same, but the match of any individual worker's rating to that worker's disability would be poor.  Second, accepted economic theory postulates that if the match is poorer, the workers on average are worse off, because under compensation reduces a worker's utility more than overcompensation increases a worker's utility. Third, economic theory as reviewed earlier suggests that the more risk averse party will have poorer outcomes on average under this type of arbitration. Finally, if the average award is the same, but the distribution of awards is more dispersed for a given disability, employers will face higher costs.

Figure 5: Distribution of final ratings: Baseball arbitration v. all other

Figure 5: Distribution of final ratings: Baseball arbitration v. all other

Finally, the large differences in the offers in nearly all of the cases involving baseball arbitration mean that each side faces the potential for very different awards possibly for very small variations in either party's final offer or the preferences of the WCJ.  The following table, Figure 6, shows the differences in dollars to the injured worker that would result from acceptance of one offer over the other.



Figure 6: Value of PD rating valued at maximum weekly earnings

Defense Rating %

Defense award $s

Applicant Rating %

Applicant award $s

Difference       $s













$106,375 + Life pension

$99,340 +    Life pension





























$118,795 + Life Pension

$65,372  +   Life pension













            CHSWC was informed that WCJ's are having problems with application of §4065 and that many of them are trying to avoid it.  This is confirmed by reported cases.  It is often impossible to tell from the record whether §4065 is applicable, i.e., whether the dispute arose under §4061.  See Vergara v. State Comp. Ins. Fund (1996) 24 CWCR 204.  Some WCJs ask the parties about prelitigation proceedings when framing the issues and record the answers in the minutes, but many apparently do not (see Bud of Calif. v. WCAB (Diaz) 62 CCC 375).  Other WCJs may use hyper technical reasons for finding that a party has waived the right to "baseball arbitration."  See Paula Ins. Co. v. WCAB (Diaz) 62 CCC 375. 

            Parties are equally adept at avoiding "baseball arbitration."  For example, when the treating physician finds the disability permanent and stationary and describes factors of disability, the objecting party will object to the determination that the disability is permanent and stationary under §4062 rather than to the extent of disability under §4061.  QME's examining under §4062 are required to "address all contested medical issues" including PD, and when the PD issue is heard by a WCJ, the party argues that §4065 is inapplicable because the evaluation was not "under Section 4061."

            More important is the reason for the reluctance to use "baseball arbitration," i.e., the result is often unfair.  Experienced triers of fact in the workers' compensation field know that more often than not an applicant's true disability lies somewhere between the description of PD obtained by the applicant and that procured by the defendant.  Under §4065, however, a WCJ is forced to award too much or too little.  For example, in Sheppard, supra, the WCJ determined the actual rating to be 13-3/4 percent, but he was compelled to find it to be the 7-1/2 percent proposed by the employer because it was closer than the 25 percent proposed by the applicant.

            WCJs' frustration with being deprived of the right to determine PD on the basis of the range of the evidence was eloquently described by WCJ Jules Greenberg in an article entitled Directions in the Law (Calif. Workers' Comp. Enquirer, Vol. 15, No. 9, p. 15) in which he suggested that represented workers may be being over compensated as a result of the combination of the treating physicians' presumption of accuracy and "baseball arbitration."  He wrote that the "usual scenario" is for the worker's attorney to seize medical control from an "industrial medical clinic" under §4600 after 30 days and arrange for treatment by a physician whose final evaluations "contain the usual exaggerated puffery in the work restrictions to advance the value of the report 'just enough' to get [a] higher rating."  The employer objects to the rating and secures a QME evaluation with "a washout or lower rating."  Due to the combined effect of §§4062.9 and 4065, the trier of fact is forced to select an inordinately high evaluation when the truth lies somewhere between the two evaluations.

            Under current procedures, a defendant may submit only one proposed rating.  If it has a reasonable, but not indisputable, claim that the disability should be apportioned it faces the dilemma of whether or not to propose an apportioned rating.  Failure to guess correctly can result in an inaccurate rating.  Thus, in San Diego Gas & Elec. v. WCAB (Morgan)(1997) 62 CCC 384, the defendant proposed an apportioned PDR of 8 percent.  Although the WCJ said that he would have considered defendant's unapportioned rating of 20 percent, he was compelled to accept the 33-1/2 percent rating proposed by the applicant.

            Where several parts of the body are to be rated, the parties face a problem similar to that where apportionment is claimed.  Including a part of the body in issue will throw the rating off if the WCJ finds that part of the body was not injured.  Omitting it will result in an inadequate rating if the WCJ finds that part of the body to have been injured.

            Finally, attorneys, particularly applicants' attorneys, are faced with the problem of selecting a rating that the WCJ is likely to accept without selling their clients short.  Applicants are likely to be upset when their attorneys propose ratings that they consider too low.  This discontent can develop into something more serious if the WCJ indicates in his opinion that he would have found greater disability if it had been proposed.


            The literature review, the preliminary data analysis, and legal and anecdotal evidence all seem to agree that there are problems with the implementation of final offer arbitration in workers' compensation. But changes to this system should be considered carefully.  The data analysis here is performed on a small sample of final offers and awards under baseball arbitration.  In addition, this preliminary study did not draw information on the offers made under conventional WCAB hearing rules.

            Consequently, the Commission should expand the data on final offer arbitration by drawing a highly targeted sample of baseball arbitrations and conventional arbitration awards.  For both, information on offers and awards should be obtained. The preliminary  sampling conducted here has given the Commission the necessary understanding to draw a narrowly defined sample that will include the correct claims, but require a minimum of resources.

            If the larger sample confirms the problems suggested by the preliminary analysis, there would appear to be enough non anecdotal evidence to conclude that baseball arbitration is neither working satisfactorily nor producing fair ratings.  Because it precludes determinations based on the range of the evidence, it would appear that nothing short of repeal would remedy the situation.  The range of the evidence rule does not, however, guarantee fair ratings unless it is employed with common sense and impartiality.  Repeal is, therefore, likely to be dependent on restoration of the Legislature's confidence in the adjudicatory process.  How to do this could be the subject of an entire study.

            If the Commission decides to recommend repeal of the treating physician presumption, it might consider including repeal of §4065 with that recommendation.  Although, our statistics on lack of savings from the presumption will not support repeal of §4065, but inadequate treating physician reports and lack of savings are not the only difficulties with the presumption.  As indicated by the cases and WCJ Greenberg's article, many of the practical problems with the presumption also apply to, and are aggravated by, baseball arbitration.

            Baseball arbitration question as a part of an overall review of the adjudicatory process in workers' compensation, the objective of which is to devise a means of providing adequate, consistent, and predictable permanent disability awards.  Baseball arbitration has the same defect as was raised in objection to the fast-track system proposed by RAND, i.e. an injured worker is entitled to be compensated for his or her individual disability and not for the average disability or by the result of a guessing game.

            On the other hand, a larger sample of awards could find that baseball arbitration is producing positive results, more frequent settlements and offers that are more reasonable.  If this is the case, the current research suggest two likely recommendations. First, the research suggests that offers will converge if the uncertainty about the WCJs probable award is reduced.  This can only occur if the Division or the Commission make available data about offers and decisions.  These data would greatly improve parties' understanding of the process and reduce their uncertainty about probable outcomes. 

Data on offers and awards would also accomplish the same purpose under conventional arbitration.  That is, improving the understanding of parties about probable outcomes given sets of offers under any decision process will encourage settlement and improve the quality of offers. Consequently, regardless of the type of decision process ultimately selected, information on offers and awards could be collected and published by the Division or Commission.

Finally, if final offer arbitration is retained in workers' compensation, rules should be standardized so that parties' expectations about facing FOA would be consistent case to case.  Baseball arbitration is being used in a minority of cases and avoided in a substantial portion of cases.  Application of baseball arbitration should be made more consistent by applying the current statute as written.  Nothing in the current statute restricts baseball arbitration in the majority of claims with competing medical reports.



1.      Ashenfelter, O., and D. E. Bloom, 1984. Models of arbitrator behavior: Theory and Evidence. American Economic Review v. 74 (March, 1984):111.

2.      Ashenfelter, Orley; Currie, Janet; Farber, Henry S.; Spiegel, Matthew. An experimental comparison of dispute rates in alternative arbitration systems. Econometrica v60, n6 (Nov, 1992):1407.

3.      Ashenfleter, O. Arbitration and the negotiation process.  American Economic Review, v.77 (1987):342

4.      Babcock, Linda C.; Olson, Craig A. The causes of impasses in labor disputes.     Industrial Relations v31, n2 (Spring, 1992):348-360.

5.      Bloom, D. E. Is arbitration really compatible with bargaining? Industrial Relations, v.20 (1981):233.

6.      Bloom, D. E. The effect of final offer arbitration on the salaries of municipal police officers in New Jersey. Working Paper #129. Industrial Relations Section. Princeton University (December, 1979)

7.      Brams, S.J., and S. Merril. Equilibrium strategies for final-offer arbitration: there is no median convergence. Management Science v.29(August, 1983).

8.      Bretz, Robert D., Jr.; Thomas, Steven L. Perceived equity, motivation, and final-offer arbitration in major league baseball. Journal of Applied Psychology v77, n3 (June, 1992):280.

9.      Burgess, Paul L.; Marburger, Daniel R.  Do negotiated and arbitrated salaries differ under final-offer arbitration?  Industrial and Labor Relations Review v46, n3 (April, 1993):548-559.

10.  Coleman, B. Jay; Jennings, Kenneth M.; Mclaughlin, Frank S. Convergence or divergence in final-offer arbitration in professional baseball. Industrial Relations v32, n2 (Spring, 1993):238-247.

11.  Curry, Amy Farmer; Pecornio, Paul.  The use of final offer arbitration as a screening device.  Journal of Conflict Resolution v37, n4 (Dec, 1993):655.

12.  Donn, C. D. Games final arbitrators must play. Industrial Relations v. 16  (October, 1977):306

13.  Farber, H. S. and M. H. Bazerman.  Divergent expectations as a cause of disagreement in bargaining: evidence from a comparison of arbitration schemes. Quarterly Journal of Economics v.104(February, 1989)

14.  Feuille, P. Final-offer arbitration and negotiating incentives. Arbitration Journal v.33(September, 1977)203.

15.  Feuille, P. Final-offer arbitration and the chilling effect. Industrial relations v.14(October, 1975)302.

16.  Grigsby, David, and William Bigoness. The effects of third party intervention on pre-intervention bargaining behavior. Journal of Applied Psychology, v.67(1982):549.

17.  Kochan, T.A., and Jean Baderschneider. Determinants of reliance on impasse procedures: Police and firefighters in New York State. Industrial and Labor Relations Review, v.31(1978):431.

18.  Marburger, Daniel R.; Scoggins, John F.  Risk and final offer arbitration usage rates: evidence from major league baseball.  Journal of Labor Research v17, n4 (Fall, 1996):735.

19.  Mnookin, Robert and Tony Kornhauser. Bargaining in the shadow of the law. Yale Law Journal, v.88 (1979):970

20.  Neale, Margaret A., and Max H. Bazerman, The role of perspective taking ability in negotiating under different forms of arbitration. Industrial and Labor Relations Review, v.34(1983):378.

21.  Notz, W.W. and F. A. Starke. Final offer versus conventional arbitration as means of conflict management. Administrative Science Quarterly, v.23(1978):189.

22.  Olson, C. A. Final-offer arbitration in Wisconsin after five years. Proceedings of the Thirty-First Annual Meeting of the Industrial Relations Research Association. 111-119.

23.  Olson, Craig A.; Jarley, Paul. Arbitrator decisions in Wisconsin teacher wage disputes.      Industrial and Labor Relations Review v44, n3 (April, 1991):536.

24.  Olson, G. A. Does final-offer arbitration allow bargaining that conventional arbitration chills? Monthly Labor Review v.102(May, 1979)38.

25.  Overton, C.E., and M. S. Wortman. Compulsory Arbitration: A strike alternative for Police? Arbitration Journal v.28(March, 1974):40

26.  Rehmus, C. M. 1976.  Public Employees: A survey of some critical problems on the frontier of collective bargaining. Labor Law Journal v.27(September, 1976):88.

27.  Somers, P. C.  An evaluation of final-offer arbitration in Massachusetts. Journal of Collective Negotiations v.6 (1977):38.

28.  Staudohar, P. D. Results of final-offer arbitration. California Management Review, v.18 (Fall, 1975).

29.  Stevens, Carl. Is compulsory arbitration compatible with bargaining? Industrial Relations, v.5 (February, 1966).

30.  Wheeler, H. N. Is compromise the rule in fire fighter arbitration? Arbitration Journal v.29(September, 1974):176.

31.  Whitman, D.  Final-offer arbitration Management Science v. 32(December, 1986):1551.


[1] The major differences between the hearing process at the WCAB and conventional arbitration are 1) arbitrators are commonly not held to strict rules followed by the courts and 2) usually, arbitrators decisions are not subject to appeal or higher review.  Neither of these substantially affects the discussion here.

[2] This issue can be answered within WCAB decisions where both FOA and conventional arbitration operate side-by-side.  The likelihood that parties settle between hearing and decision could measure the likelihood that each type of arbitration results in a settlement.

[3] Actually, if the parties assume the judge will compromise between the two positions, each parties strategy will be to make its offer more extreme to the point where additional movement to the extreme would reduce the credibility of the position by enough to reduce that parties gain (positive for the applicant, negative for the defense).  In essence the parties trade-off three factors, their rating proposal, their guess as to the opposing parties rating proposal, the credibility of their proposal if they move one more unit to the in either direction. This process and the final offer arbitration process each involve complex game theory issues and an economic theory called Nash Equilibrium