Administrative Director
Division of Workers' Compensation

September 1997

California Division of Workers' Compensation
P.O. Box 420603
San Francisco, CA 94142

Table of Contents


An Overview of the Carve-Out Program

The Carve-Out Participants

Number of Employees Covered

The Data

Analysis of the Claims Data

Data Concerning ADR 14

Data Concerning Vocational Rehabilitation and Light Duty Programs


This report is from the Administrative Director of the Division of Workers' Compensation to the Legislature, as required by Labor Code Section 3201.5(i).

The report covers the second full calendar year -- 1996 -- of the program established by Labor Code Section 3201.5, commonly known as "the carve-out". The program allows unions and employers engaged in construction to create alternatives to the traditional, state-supervised workers' compensation system. Most important, it allows collective bargaining agreements to establish exclusive lists of medical providers and examiners and to establish alternative dispute resolution systems that bypass the state workers' compensation referees.

This report is based on aggregated data from 277 California employers. Together, these employers reported 11,645,276 person-hours of labor and $272,383,804 in wages to employees covered by Section 3201.5. They reported 866 claims filed in 1996. Aggregate incurred losses for these claims were $6,102,437.

The Commission on Health and Safety and Workers' Compensation has initiated an extensive study of the carve-out program and expects to issue a report near the beginning of 1998.

An Overview of the Carve-out Program

Labor Code Section 3201.5 establishes a program in which unions and employers engaged in construction may bargain collectively to create alternatives to the traditional workers' compensation system:

The statute gives the parties leeway to establish the kind of alternative system they want -- but there are two important limits: (1) the agreement may not diminish injured workers' entitlement to compensation payments, and (2) the final step of the "alternative dispute resolution system" [ADR] must be appeal to the 7-member Workers' Compensation Appeals Board [WCAB].

The carve-out program was first established in 1993 by SB 983, authored by Senator Leroy Greene. It was modeled on a similar experiment in Massachusetts, where Bechtel and the Pioneer Valley Building and Construction Trades Council had a collective bargaining agreement governing a single construction project.

When SB 983 first went into effect, several promoters who were clearly outside the intended scope of the program rushed to get the Administrative Director's approval. The Administrative Director received "collective bargaining agreements" from "employers" with no employees and "labor organizations" with no members. One notable "agreement" appeared to be a scheme whereby employers were persuaded to pay money into a "health benefits trust" controlled by the promoter, instead of purchasing a workers' compensation insurance policy as required by law. This placed employers at risk for civil and criminal penalties. The common thread of these early "agreements" was that promoters would approach legitimate employers -- that is, real California companies engaged in real construction -- and persuade them to transfer their employees to the promoter. The promoter would become the nominal employer of the employees, while control over their actual work would remain with the original employer. Conveniently, the promoter would have already signed a collective bargaining agreement with a labor organization that "represented" all the promoter's employees. And, just as conveniently, the agreement would have a carve-out section that was favorable to the employer. The Administrative Director did not recognize any of these "agreements."

In 1994, Senator Greene introduced urgency legislation -- SB 853 -- to amend Section 3201.5, tightening the qualifications of the parties. The term "employer engaged in construction" was more narrowly defined. The employees' bargaining representative was required to be a "union." Parties were now required to submit evidence of their eligibility and receive a "letter of eligibility" from the Administrative Director.

To be eligible, an employer must prove that it is engaged in construction and pays a workers' compensation insurance premium of at least $250,000 per year, or belongs to a multi-employer "safety group" which pays a premium of at least $2,000,000 a year. Self-insured employers are also eligible. To be eligible, a labor organization must prove that it is a bona fide union, free of employer domination or support, whose officers have been elected by secret ballot. The union must submit a copy of the latest LM-2 or LM-3 financial disclosure form that it filed with the United States Department of Labor. These forms give information about the unions' sources of income, expenditures, officers, election dates, and other pertinent matters.

Although the parties must establish that they are "eligible," they do not need the administrative director's approval of the collective bargaining agreement itself.

A copy of the current text of Labor Code Section 3201.5 may be found in Appendix 1.

Regulations implementing section 3201.5 are at 8 California Code of Regulations, Section 10200 - 10204. They may be found in Appendix 2.

The Carve-Out Participants

Through the end of 1996, DWC received eleven applications for letters of eligibility. Nine were granted. One was rejected. One application was never completed. During 1996, one carve-out program involving the Cherne Contracting Corporation and Steamfitters Local 250 came to an end when their construction projects were completed.

The eight currently approved programs are:

Number of Employees Covered

The seven active and reporting carve-out programs reported a total of 11,645,267 person-hours -- equivalent to 5822 full-time employees, figuring 2000 person-hours for one employee-year.

The Data

Labor Code Sections 3201.5(i) and (j) require the administrative director to collect a variety of data from carve-out employers -- payroll, personhours worked, number of claims filed, average cost per claim, etc. -- and to make a report to the Legislature based on the aggregated data. To implement these sections, DWC promulgated 8 Cal.Code Regs. Section 10203, requiring employers to report information to the administrative director on a standard form. Appendix 3 is a sample reporting form.

As noted above, there are seven active carve-out programs that reported activity in 1996. DWC has received data from all 277 employers participating in those programs. As shown in Table 1, they report a total of 11,645,267 person-hours in 1996, and a total payroll of $270,613,970. They report 866 claims filed in 1996 and incurred costs of $6,102,437.

Table 1

Summary of Claims Data1

Number of Employers


Total Personhours


Total Payroll


Total Claims


Total Incurred Costs


Table 2

Claim Frequency and Average Costs Among Seven California Carve-out Programs (1996)

(Each program is identified here by
letter, rather than
name, to protect
Number of Claims per 100 Employees2 Number of Claims per $1 Million Payroll Average Incurred Cost per Claim Number of Indemnity Claims3 per 100 Employees Number of Indemnity Claims per $1 Million Payroll Average Incurred Cost per Indemnity Claim











733 [sic]














































Table 3 shows the components of paid and incurred losses4 for the 866 claims filed in 1996. There were 570 medical-only claims and 296 indemnity claims: 5

Table 3

Cost Components for Claims Filed in Calendar Year 19966

Expense Paid Amount Incurred Amount
(570 claims)
$195,980 $241,318
(296 claims)
Temporary Disability $1,092,346 $1,556,944
Permanent Disability $130,549 $1,007,456
Life Pension 0 0
Death Benefits $285,440 $291,329
Vocational Rehab. $45,590 $523,953
Medical $1,976,491 $2,450,071
Medical-legal $8,569 $31,366
Total Indemnity $3,538,985.00 $5,861,119
GRAND TOTAL $3,734,965 $6,102,437

Table 3 does not include the insurers' operating expenses. Anecdotally, one insurer told us last year that its operating expenses were approximately 2/3 lower in the carve-out program than in its usual workers' compensation business. The insurer attributed this, in part, to the almost complete absence of defense attorney costs.

Table 3 also does not include the costs of operating the ADR process itself -- the ombudsperson's fees and costs, mediator's fees, and so on. Under current programs, the ADR process is administered by, or under the supervision of, a joint labor-management committee or trust. The process is funded by the employers. Employers pay a fee to the trust equal to 1&1/2 or 2% of the insurance premium, or several cents per personhour worked.

It should also be noted that the 866 claims were all filed in 1996. They are all relatively young claims. As claims become older, incurred costs tend to rise, though it is impossible to estimate how much they will rise. One of the largest groups reporting -- the NECA safety group, which is a subset of the NECA/IBEW program -- informs us that incurred losses for claims initiated in 1995 climbed substantially in 1996 -- from $953,981 to $1,487,600.

Table 4

Comparison of Incurred to Expected Losses Among Seven California Carve-out Programs (1996)









Incurred Losses1

Incurred Losses per $100 of Payroll

Expected Losses8

Ratio of Incurred to Expected Losses



































Analysis of the Claims Data

Were losses in the carve-out program lower or higher than losses experienced by other employers in the construction industry?

To answer this question, we compiled Table 4, which compares incurred and "expected" losses of the carve-out employers.

On the 866 claims, there were incurred losses of $6,102,437. Table 4, Column 4, shows incurred losses per $100 in payroll for each of the seven active carve-out programs. In aggregate, the six programs incurred $2.24 for every $100 in payroll.

How does the figure of $2.24 compare to incurred losses of all employers engaged in construction? Table 4 provides "benchmarks" for each carve-out program, based on WCIRB "pure premium rates".

The Workers' Compensation Insurance Rating Bureau publishes "pure premium rates" for every job classification in the state.9 These rates reflect the losses predicted for every $100 in payroll, plus an amount for "loss adjustment expenses".10 Loss adjustment expenses constituted 20.1% of the pure premium rate in 1996. For instance, the pure premium rate for ironworkers on buildings over two stories (Classification #5040) is 14.40. A hypothetical employer who hires ironworkers to erect a steel frame would expect to incur $14.40 in losses for every $100 in payroll, of which $11.52 would be incurred for benefits and $2.88 for loss adjustment expenses. Pure premium rates vary according to the hazards of the job. For instance, the rate for lath installers (#5443) is 8.08; the rate for elevator workers (#5160) is 3.09.

In Table 4, Column 5, we have calculated a benchmark for each of the seven active carve-out programs. They are calculated by weighting the payroll in each job classification, multiplying by the pure premium for each classification, and subtracting loss adjustment expenses. For example, if a particular carve-out program had 60% of its payroll in Classification X, with a pure premium rate of 3.00, and 40% of its payroll in Classification Y, with a pure premium rate of 10.00, the weighted average pure premium for that program would be (.60)(3.00) + (.40)(10.00) = 5.80. Of the 5.80, 20.1% is for loss adjustment expenses (1.16) and 79.9% is for incurred benefit costs (4.63). The benchmark for that hypothetical carve-out program is $4.63. This is roughly what one might predict as incurred losses for that carve-out program per $100 payroll. It is important to note, however, that this does not take into account other factors that might affect individual employers, such as their own safety practices or claims history. Indeed, an employer's loss experience may be consistently much better or worse than the loss experience of all employers in the classification.

Table 4, Column 6, shows the ratio between the incurred losses for each carve-out program and the benchmark for each carve-out program. All seven active carve-out programs incurred losses well below the benchmark. In aggregate, incurred losses were 52% of what the pure premium rates would predict.

In other words, it appears that losses in the carve-out program were lower than would be expected of other construction employers. However, it should be noted that Table 4 is based on claims filed in 1996 -- that is, on data that is quite young. As claims age, incurred costs tend to rise. As noted above, one of the largest groups reporting -- the NECA safety group, which is a subset of the NECA/IBEW program -- informed us that incurred losses for claims initiated in 1995 climbed substantially in 1996 -- from $953,981 to $1,487,600. The ratio of incurred to expected losses for that group went from .34 after 12 months of reporting to .53 after 24 months of reporting ending in December 1996.

Were these apparently low losses due to a random fluctuation in the data?

This report is based on only 11,645,267 person-hours. That is equivalent to only 5823 full-time employees -- a pool that may be too small to provide reliable statistical data. As the number of carve-out employers increases in the future, the statistical pool will expand and the figures presented will be more reliable.

Despite the limitations of the data, we are inclined to think that the low figures are not mere happenstance. Note that losses were low in all seven programs. However, if the loss figures were randomly distributed, one would expect to see some programs showing higher than expected losses and some showing less.

Furthermore, the figures this year are roughly equivalent to the figures in our 1995 report.

Table 5

Cost Comparison between 1995 and 1996



No. of claims






Incurred Cost



Incurred Cost / $100 Payroll



Expected Losses / $100 Payroll



Ratio of Incurred to Expected Losses



The similarity of these figures suggests that neither 1995 nor 1996 was a statistical anomaly.

Were low losses due to carve-out employers' history of low losses? In other words, are carve-out participants self-selected for safety?

We do not have the data necessary to answer this key question.

Were low losses due to fewer claims being filed by carve-out employees?

Carve-out employees filed 866 claims in 1996, equivalent to 15.77 claims for every 100 employees. It is difficult to find an exact comparison for this figure among non-carve-out employees. The California Division of Labor Statistics and Research reports that in 1994 -- the last calendar year for which it has statistics -- there were 12.0 non-fatal occupational injuries and illnesses per 100 employees in the construction industry as a whole. These figures come from injuries and illnesses reported by employers on federal OSHA forms. They are not precisely equivalent to workers' compensation claims, since some injuries might be reported on the OSHA form but not give rise to a claim. Conversely, an employee might file a claim for an injury not reported to OSHA. Nonetheless, there seems to be a rough equivalence between the number of injuries and illnesses that might be expected per 100 employees (12.0) and the number of claims being filed (15.77).

It appears that the carve-out program does not achieve its savings through a decrease in claims.

Data Concerning ADR

Both labor and management indicate that one of their chief aims in participating in the carve-out program is to replace the state's cumbersome and lengthy dispute resolution system with one that is simpler and faster. Of the seven active and reporting carve-out programs in California, six have alternative dispute resolution [ADR] systems.

Of the seven active and reporting carve-out programs, six use ADR. (One carve-out program consists only of an exclusive list of medical providers.) The six ADR programs in California are similar, but have some differences. All six start with an "ombudsperson" -- a neutral person who is available to all parties. The ombudsperson is supposed to resolve disputes at an early stage, or even before they arise. If the ombudsperson is unsuccessful in resolving the dispute, either party may move the matter to the next step, which is typically formal mediation by an outside neutral. Some programs use a joint labor-management committee at this point. If mediation is unsuccessful, the parties turn to an outside, neutral arbitrator -- typically a retired workers' compensation referee. By statute, the decision of the arbitrator may be appealed to the 7-member Workers' Compensation Appeals Board (Labor Code Section 3201.5(a)(1)).

For the six active carve-out programs with ADR, there were 837 claims filed in 1996.11 In our 1995 Carve-Out Report, we attempted to distinguish between claims that were "resolved" in 1995 and those that remained "unresolved" at the end of the year. Unfortunately, there was considerable confusion among the parties about the meaning of "resolved" and "unresolved."12 The confusion continued this year. Accordingly, we have decided not to make a public report based on such unreliable figures.

Despite the confusion, however, several facts are clear:

In two years of carve-out activity, there have been a total of 1356 claims that were subject to ADR (519 filed in 1995 and 837 filed in 1996.) Of these 1356 claims, there have been only 5 mediations and no arbitrations. A very large majority of these claims were somehow resolved without any formal dispute.

Activity Before the WCAB and the Courts

According to the reports filed by carve-out participants, ten employees subject to an ADR system filed applications for adjudication with the WCAB in 1996.

We are not aware of any litigation in 1996 concerning a carve-out claim or program.

Has ADR Reduced the Amount of Formal Dispute Over Claims?

From the very small number of formal mediations and arbitrations and from the lack of litigation in the WCAB and the courts, it appears that the carve-out program has, at least in its first two years, been successful in reducing formal dispute over claims.

It is notable that medical-legal expenses are reported at a very low level -- $31,000 out of more than $6,000,000 total incurred costs -- a mere .5%. This figure is consistent with a low level of formal dispute.

Data Concerning Vocational Rehabilitation and Light Duty Programs

In aggregate, employers reported 4 employees participating in vocational rehabilitation and 65 employees participating in light-duty programs. No statistically valid conclusions can yet be reached from these figures.


The Appendices are not available in this online edition of the report.


1Based on seven active carve-out programs reporting 1996 activity.

2200,000 personhours = 100 full-time employees

3In California, an indemnity claim involves lost time of 3 days or more.

4Incurred loss = amount paid out plus any amount reserved to pay over the life of the claim.

5In California an indemnity claim means a claim involving lost time of at least three days.

6Based on annual reports submitted by all seven active and reporting carve-out programs.

7To protect the confidentiality of the programs, individual payroll and incurred losses for each program are not shown.

8The figures in this column are based on the WCIRB pure premium rates published in the California Workers' Compensation Uniform Statistical Reporting Plan -- 1996. Expected losses are calculated for each program by weighting the payroll in each job classification and multiplying by the pure premium. For example, if a particular carve-out program had 60% of its payroll in Classification X, with a WCIRB pure premium of 3.00, and 40% of its payroll in Classification Y, with a pure premium of 10.00, the weighted average for that program would be (.60)(3.00) + (.40)(10.00) = 5.80. WCIRB pure premium rates include "loss adjustment expenses" of 20.1%. Deducting this 20.1% from the weighted average leaves a figure representing expected losses. The figure shows the expected incurred losses per $100 in payroll. In our hypothetical example, this would be (.799)(5.80) = 4.63. For every $100 in payroll in our example, one would expect $4.63 in incurred losses.

9California Workers' Compensation Uniform Statistical Reporting Plan -- 1996.

10Loss adjustment expenses = the insurer's average costs for handling a claim.

11The raw figure is actually 843; but there were a few omissions and deficiencies in the reports we received from the participants. We have what seems to be complete ADR data on 837 of the claims filed in 1996.

12The term "resolved" was intended to cover all claims where there was a final determination by an arbitrator, all claims which had been settled by compromise & release or stipulation, and all claims in which the insurer literally closed its file because, in its view, no dispute was likely. By contrast, "unresolved" meant merely that the claim was still open on December 31. "Unresolved" did not necessarily mean that a real dispute existed, but only that a dispute remained possible in the future. For instance, a claim that involved an injury that was not yet permanent and stationary on December 31 would be "unresolved", even if no actual dispute had arisen. However, the terms "resolved" and "unresolved" led to a great deal of confusion among the carve-out participants. The parties used the terms inconsistently in their reports.

13On July 17, 1997, in the case of Ramon Becerra v. Eastside Reservoir Project (No. AHM 51304), the WCAB ruled that the proper procedure is for workers' compensation referees to dismiss applications, rather than stay them, if the claim is subject to ADR.