Workers Comp

ContentServer.asp-5.pdf

Home >Business & Finance homework help >Operations Management homework help >Workers Comp

Premium-Rating Model for Workers’ Compensation Insurance in Construction

K. Imriyas1; S. P. Low2; A. L. Teo3; and S. L. Chan4

Abstract: Providing adequate workers’ compensation insurance �WCI� is mandatory by legislation in most countries for employers to hire workers. It is enforced to safeguard the interests of occupational injury victims while eliminating their employers’ financial burden of compensation. WCI’s significance in construction is immeasurable because the construction industry appears to be recording higher injury and fatality rates globally. On the other hand, insurance companies must be meticulous in deciding premiums for WCIs for construction projects. Traditionally, insurers have been adopting the experience modification rating technique for premium-rating of WCI. However, this technique has been proven ineffective for construction applications due to the unique characteristics of construction projects. Hence, a new premium-rating model for construction was developed and tested in the Singapore general insurance industry. The proposed model streamlines structured analyses of project hazards, contractors’ safety management systems, market conditions, and insurers’ internal factors to decide optimal premiums. Moreover, it infuses the partnering notion in WCI for construction. The implementation of the proposed model in the insurance industry would facilitate accident control in the construction industry and thereby loss minimization for insurers.

DOI: 10.1061/�ASCE�0733-9364�2008�134:8�601�

CE Database subject headings: Construction site accidents; Safety; Insurance; Models.

Introduction

The significance of workers’ compensation insurance in construc- tion is overwhelming because the construction industry is per- ceived as one of the more dangerous industries. In the United States, it was reported that the construction industry accounted for only 5% of the United States’ workforce but claimed a dispropor- tionate 20% of all occupational fatalities and 9% of all disabling injuries �National Safety Council 1997�. In Great Britain, con- struction accounted for 31% of all work-related deaths in 2002– 2003 �Haslam et al. 2005�. In Singapore, the construction industry accounted for 29% of the total number of industrial workers, but accounted for 40% of worksite accidents �Chua and Goh 2004�. Hence, the workers’ compensation insurance �WCI� has been rec- ognized as an imperative for construction by legislation to safe- guard the interests of injured workers as well as to ease contractors’ financial burden of compensation. This, on the other hand, implies that the construction WCI is a critical part in the

whole portfolio of any insurer’s business, which entails rigorous attention to risk assessment and premium rating.

Insurance companies have to quote appropriate premiums for construction projects, which should be high enough to cover all the future costs that will arise from the insured risks, yet low enough to meet the market competition �Booth et al. 1999�. In order to decide on an optimal premium for a given project, an extensive study on accident risks is crucial for insurers. The ex- perience modification rating �EMR� is a well-established tech- nique to assess accident risks in construction projects wherein the premium charged to a contractor is based on the contractor’s claims experience �Clough et al. 2005; Hoonakker et al. 2005�. The general notion is that a contractor has control over the loss ratio and is entitled to credit for a good loss-prevention record or, on the other hand, should pay a higher rate if the loss record is poor. A detailed account of EMR can be found in Everett and Thompson �1995�. Nevertheless, many researchers have criticized that EMR is ineffective for construction applications as summa- rized in Table 1. Hence, there is an intense need for developing a new methodology for WCI premium rating of construction projects. The objectives of this study are as follows: 1. To develop an effective model for computing WCI premiums

for construction projects; 2. To develop the conceptual framework of a knowledge based

system �KBS� to automate the model above; and 3. To prototype and verify the KBS.

However, this paper explores only the proposed new premium- rating model due to word limitation. The study is demonstrated in the context of building construction because: �1� time constraint impedes the development of a model to cater to any type of con- struction project; therefore, it is crucial to address the most sig- nificant sector; and �2� building construction is the most significant sector in Singapore where the demand for buildings is around 70% of the total construction demand �BCA 2006�.

1Lecturer, Faculty of the Built Environment, Univ. of New South Wales, Sydney NSW 2052, Australia �corresponding author�. E-mail: [email protected]

2Professor, Dept. of Building, 4 Architecture Dr., National Univ. of Singapore, Singapore 117566.

3Assistant Professor, Dept. of Building, 4 Architecture Dr., National Univ. of Singapore, Singapore 117566.

4Formerly Assistant Professor, Dept. of Building, 4 Architecture Dr., National Univ. of Singapore, Singapore 117566.

Note. Discussion open until January 1, 2009. Separate discussions must be submitted for individual papers. To extend the closing date by one month, a written request must be filed with the ASCE Managing Editor. The manuscript for this paper was submitted for review and pos- sible publication on May 17, 2006; approved on February 4, 2008. This paper is part of the Journal of Construction Engineering and Manage- ment, Vol. 134, No. 8, August 1, 2008. ©ASCE, ISSN 0733-9364/2008/ 8-601–617/$25.00.

JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT © ASCE / AUGUST 2008 / 601

Premium-Rating Variables for Construction WCI

In insurance, unlike other industries, the cost of production is unknown when the contract is sold, and it will be unknown until the policy expires. Therefore, the pricing for insurance must be based on predictions of losses, expenses, and incomes in the fu- ture. Under WCI, the commitment of an insurer is extremely broad; there are no exclusions, and there is no maximum limit on the insurer’s liability �Vaughan and Vaughan 1996�. Setting an appropriate cost for some future contingency of unpredictable timing, frequency, and size requires the estimates of future claims, investment income, administrative expenses, profit, and tax. In addition, the price can profoundly influence the volume of the business attracted �Booth et al. 1999�. Hence, WCI premium rating is a critical management function for any insurer, which should decide on premiums that are high enough to cover all the future costs, yet low enough to meet the market competition. The premium rating is regarded as a two-stage process �Booth et al. 1999�: 1. The costing exercise—this is a scientific method that calcu-

lates the cost of future claims from the insured risks and all associated expenses; and

2. The pricing exercise—this is a commercial adjustment to technical costs that considers broader corporate and market factors.

Predicting Future Claims in Construction WCI

The scientific approach to determine the cost of future claims of a WCI policy entails an assessment of project and contractor- related variables. Under the project-related variables, the follow- ing were identified by various authors: 1. Worrall and Buttler �1988� noted that the wage roll, project

duration, and the expected workers’ compensation liability per wage roll unit need to be accounted for when predicting claims;

2. The Canadian Wood Council �2005� reported that base rates are influenced by the: �1� location of the project; �2� type of construction; and �3� general industry experience on loss his- tory for projects of similar type and location;

3. Lott �2005� added that an assessment of the workers’ risk management program will lead to an effective prediction of the costs to be incurred. This assessment should scrutinize the following aspects:

a. Management commitment and employees participation in workers’ risk control;

b. Workplace hazards, population demographics, and previ- ous incident rates;

c. Implementation of a pertinent risk/safety program; and d. Training of supervisors and workers. As for the contractor-related variables, Groth �1996� com-

mented that WCI premium rating must consider a contractor’s past claims history. Coble et al. �1998� noticed that a contractor’s size has a bearing on premium rates. The Canadian Wood Council �2005� suggested that insurers must consider the following variables: 1. Contractors’ knowledge, experience, and safety

consciousness; 2. Compliance with loss control and underwriting recommen-

dations by the insurer; 3. Acceptance of deductible, that is, a retention of part of the

claims cost by the contractor; 4. Potential future business; and 5. Placement of multiple policies with the same insurer.

Determining Markup for Construction WCI

The adjustment of the technical costs of risks for commercial interests necessitates the consideration of insurers’ corporate fac- tors and market factors. The variables, pertinent to insurers, were recognized by many authors in the literature, viz.: 1. Vaughan and Vaughan �1996� identified three variables that

influence the markup: • The expenses of acquiring and administering the business

�overhead costs�; • The profit required by the insurer; and • The return from the investment of premiums by the insurer.

2. Young �1997� quoted that premium rates are adjusted due to the profit/loss experience of the insurer in the past, and the outstanding claims to the insurer from all projects. The com- petition and the volume of business in the market also influ- ence the markup;

4. Booth et al. �1999� reported that corporate objectives of in- surers will have a significant bearing on markups; and

5. Phifer �1996� observed that the reinsurance premium, which insurance companies pay for risk sharing with reinsurers, do influence the pricing.

Research Method

From the above literature review, 17 variables pertinent to WCI premium rating of construction projects were identified and clas- sified into four categories as described below: 1. Project factor—comprised of wage roll, project duration,

Table 1. Shortcomings of EMR

Author�s� Criticism

1. Everett and Thompson �1995� • The EMR is a complex approach. • The EMR cannot fairly compare the safety records of different contractors. • New contractors are forced to pay more premiums since they are not experience rated. • The premium is more biased toward the ownership of the company.

2. Hinze et al. �1995� • The EMR value is decreased as the project size increased. • Highly paying contractors will have lower EMR values.

3. Coble and Sims �1996� The EMR can be vulnerable to fraud by contractors to obtain low premiums in three ways: • Manipulating the payroll of workers. • Misrepresenting work classification. • Manipulating company ownership.

4. Hoonakker et al. �2005� • The EMR is a lagging indicator. • The EMR is based on worker classifications and not on jobs, which impedes the interpretation of the results.

602 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT © ASCE / AUGUST 2008

project hazard level, and effectiveness of the safety manage- ment system on site;

2. Contractor factor—encompasses contractor’s claims history, placement of multiple policies, expectation of potential busi- ness, cooperation by the contractor, and contractor’s size;

3. Insurer factor—covers corporate objectives, investment in- come from the written premium, overhead costs of insurance, amount of outstanding claims, profit/loss experience in WCI business, and reinsurance cost; and

4. Market factor—includes competition and volume of business in the market.

Subsequently, a questionnaire was designed with the objective of determining the most significant variables. The questions in the questionnaire assessed the significance of the 17 variables for deriving the optimal premium rate on a ten-point Likert scale, whereby 1�“low importance” and 10�“high importance”. The field survey was conducted during November and December 2005 in the Singapore insurance industry, which encompasses a popu- lation of 23 general insurers. An interview questionnaire survey was conducted to collect data because having a questionnaire ad- ministered face to face by an interviewer attains higher response rates than a mail survey. All 23 companies were covered in the survey. At every interview session, the questionnaire was pre- sented to the interviewee for his/her response. Additional expla- nations were rendered when needed. Moreover, interviewees were invited to elaborate on the rationale for their answers.

Profile of Respondents

As shown in Fig. 1, the designations of the interviewees were top management �48%� and middle management �52%�. The top

management interviewees were comprised of managing directors, general managers, senior managers, and managers. The middle management interviewees were assistant general managers, deputy managers, and assistant managers. The mean working ex- perience of the interviewees was 21 years. The minimum and maximum working experiences were 8 and 35 years, respectively. Also, more than 75% of the interviewees had experience of above 10 years. From the interviewees’ profile, it is understood that all of them are well qualified and well experienced in the subject matter. Hence, the data collection is perceived to be reliable.

Data Analysis and Discussion

A statistical analysis was accomplished to identify the most im- portant variables according to Pareto’s 80/20 rule for developing an effective premium-rating model. Descriptive statistics were computed for each category of variables as shown in Table 2. The 17 variables were reorganized into three groups, namely: impor- tant variables, less important variables, and unimportant vari- ables, based on their mean importance ratings, which are further underpinned by their respective medians and standard deviations. Unimportant variables have mean importance ratings �4.00, im- portant variables possess mean importance ratings �7.00, and the rest are less important variables.

Important Variables for Premium Rating

Eight important variables for premium rating were identified out of the 17 variables. A brief discussion on each variable is pro- vided below, based on the qualitative explanations rendered by the interviewees.

Wage Roll In the Singapore insurance industry, WCI premiums are tradition- ally computed by applying a rate on wage rolls of construction projects. There has been a collective agreement among the insur- ers that the preferable WCI premium rate for construction projects is 1% of the wage roll. This rate, however, is merely a bench- mark. Individual insurers set competitive rates around the bench- mark heuristically considering other important variables. The wage roll is also considered as a representation of the project duration since both are positively correlated.

On another perspective, when the wage roll is high, the pre- mium amount is high though the premium rate is low. But, claims and administrative costs will be proportionately less, making it possible for insurers to earn more money by insuring bigger projects. Thus, large projects drive keen competition.

Project Hazard Level The project hazard is directly correlated with the frequency and severity of accidents and thereby the amount of compensation to be paid by the insurer. Therefore, the higher the project hazard, the higher the premium rate. The project hazard has to be assessed in line with the project’s scope. The assessment should peruse if the project involves demolition works, explosive works, excava- tion works, works at heights, works involving lifting, and cranes and/or confined space works. The hazard in each type of work has to be assessed via a rigorous analysis of relevant attributes. For example, the hazard in works at heights can be deduced by the total height of the building. Moreover, the characteristics of the project location/vicinity should also be assessed. Among the at-

Fig. 1. Profile of respondents

JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT © ASCE / AUGUST 2008 / 603

tributes to be scrutinized for project locality are: soil condition, site congestion level, and/or the presence of any chemical/ manufacturing factory or combustible source.

Effectiveness of the Safety Management System on-Site The presence of an effective safety management system on site is significant in reducing claims; thus, the better the safety manage- ment system, the lower the premium rate. All the insurers recog- nized that inspecting the safety management system of the project is essential in inferring potential claims. However, they encoun- tered two constraints, viz.: 1. Lack of guidelines to assess the effectiveness of the safety

management system; and 2. Although a contractor can produce a well-documented safety

plan for a project, its implementation on site is still uncertain as it is inherent to the attitude of the site management team.

Contractor’s Claims History Insurers examine a contractor’s claims records for the past 5 years to deduce the contractor’s safety performance as an alternative to assessing the safety management system. From the survey, it would appear that records with low severity and high frequency infuse higher premiums than records with high severity and low frequency. However, it may be ineffective because the lack of accidents does not imply an absence of accident risks. Moreover, undesirable loss ratios can be a result of both frequency and se- verity of claims.

Competition Triumphing competition is crucial in any business. A well thought out strategy is necessary to assess the competition level by con- sidering the number of competitors, regional economic condition, prestige in the project, and the project size. This can subsequently be used to adjust the markup that the higher the competition the lower the markup.

Corporate Objectives of the Insurer

The markup in a premium coheres with the insurer’s corporate objective�s�. The corporate objective of a Singapore’s insurer at any given time can be one or more of the following: 1. Ensuring the survival of the company; 2. To write a given premium volume; 3. Achieving an overall operating ratio across the whole

portfolio; 4. Achieving an adequate level of profitability or return on eq-

uity; and 5. Achieving a target premium growth rate. For objectives 1 – 3, an insurer may intentionally quote a low premium while for objectives 4 and 5, the same insurer may wish to quote an average premium.

Overhead Costs of Insurance

Overhead costs are key components of a markup; the higher the overhead costs, the higher the markup. Overhead costs for the WCI include brokerage fees and administrative costs for under- writing and claims management. The brokerage fee is a fixed percentage in Singapore’s market; 10% of the premium. Admin- istrative costs vary from company to company, which are esti- mated to be in the range of 15–35% of the underwritten premium. In total, overhead costs amount to 25–45% of the premium in Singapore.

Investment Income from Underwritten Premiums

The underwritten premium can be invested by the insurer on shares, subsidiary companies, real estate, etc. The returns on these investments influence the premium significantly. If the investment returns are high, the insurer can quote attractive premiums as the insurer’s reserve is stable.

Table 2. Descriptive Statistics of Industry Survey Results

Variables

Descriptive statistics

Mean Standard deviation Median Mode Kurtosis Skewness XMin. XMax

Project factor

• Wage roll 8.33 1.32 9 9 0.41 −0.81 5 10

• Project duration 5.10 3.10 6 3 −1.30 −0.12 0 10

• Project hazard level 9.33 0.71 9 10 −0.76 −0.63 8 10

• Effectiveness of safety management system on-site 7.67 2.17 8 8 3.67 −1.74 1 10

Contractor factor

• Contractor’s claims history 8.33 1.21 9 9 −1.18 −0.19 6 10

• Placement of multiple policies 5.00 2.56 5 8 −0.66 −0.66 0 8

• Expectation of potential business 3.76 2.60 4 0 −1.09 0.00 0 8

• Cooperation by the contractor 4.90 3.15 5 2 −1.34 0.01 0 10

• Contractor’s size 3.38 2.92 3 0 −1.26 0.37 0 9

Insurer factor

• Corporate objectives 8.05 1.13 8 7 −0.80 0.32 6 10

• Investment income from underwritten premium 7.24 1.23 7 8 −0.34 0.33 5 10

• Overhead costs of insurance 7.10 1.34 7 8 −1.22 −0.06 5 9

• Amount of outstanding claims 6.62 2.15 7 7 0.14 −0.92 2 9

• Profit/loss experience in WCI business 6.76 2.51 7 10 −1.10 −0.34 2 10

• Reinsurance cost 6.71 1.91 7 9 0.05 −0.67 2 9

Market factor

• Competition 7.38 1.68 7 7 −1.23 −0.14 5 10

• Volume of business in the market 4.52 2.52 5 6 −0.72 0.09 0 9

604 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT © ASCE / AUGUST 2008

Less Important Variables for Premium Rating

Seven variables seem less important for WCI premium rating of construction projects.

Profit/Loss Experience in WCI Business When an insurer experiences a remarkable profit/loss over a pe- riod of time, it is acceptable to adjust the premium rate for new undertakings. If there is a profit growth, the insurer may wish to reduce the premium so as to win more projects, as there is a buffer to assume the risk and to develop the business. In contrast, if the experience is bad, premiums for new projects could be increased with consideration of market trends because losing business is preferred to suffering from loss. However, such ad- justments in the WCI’s context are not encouraged by most of the Singapore insurers because profitable years will have to be bal- anced with less profitable years.

Reinsurance Cost Reinsurance is the principal mechanism that insurance companies use to transfer part or all of the risks assumed through their own underwriting activities. The reasons for reinsuring are as follows: 1. For balance sheet protection—reinsurance is purchased to

protect the solvency of the insurance company; and 2. To increase the capacity—the risk exposure that an insurer

can reasonably accept is restricted by the size of its capital base. Reinsurance allows taking larger risks, which makes the company more attractive to insurance brokers and their clients.

Risk of an insurer can be ceded to a reinsurer via facultative reinsurance and treaty reinsurance. Under the facultative arrange- ment, each risk is underwritten by the reinsurer on its own merits with a separate reinsurance contract. Under the treaty system, reinsurance is underwritten for a class of insurance, annually. The treaty reinsurance is further categorized into proportionate treaty and nonproportionate treaty. The subsets of the proportionate treaty are quota share reinsurance and surplus reinsurance while the subsets of the nonproportionate reinsurance are excess-of-loss �XOL� reinsurance and stop-loss reinsurance. The WCI in Sin- gapore adopts the XOL treaty reinsurance.

A reinsurer decides the XOL treaty premium for an insurer based on the estimated premium income, loss ratio, risk profile of the business covered, and the excess point of the treaty. Insurers can decide upon attractive premiums for their policies, depending on their treaty premiums. However, reinsurance cost is usually less than 10% of the premium. Hence, it is not a major concern for premium rating. Most of the companies incorporate the rein- surance cost into the overhead costs.

Amount of Outstanding Claims to the Insurer When a particular class of insurance, like the WCI, experiences too many claims, the insurer prefers to increase the premium for new projects so as to recover losses. Nevertheless, market com- petition restrains such adjustments. On the other hand, decreasing the premium rate to attract more business for cash flow purposes is also unfeasible because the WCI is the riskiest insurance class in Singapore.

Project Duration When the project duration is longer, the risk exposure of an in- surer is longer and the earned premium, along with the net profit for the year, is reduced. Project duration is therefore a moderate

consideration for some insurers in Singapore. However, the other insurers perceive that the duration is built into the wage roll; thus, it can be negligible for premium rating.

Placement of Multiple Policies by the Contractor When contractors take multiple policies for projects, discounts may be considered as the risks can be spread among policies. Contractor’s all risk �CAR� insurance and the WCI are the main construction insurances. The WCI is ranked as the worse policy by insurers, which earns small premiums but large claims. Mean- while, the CAR insurance has been contrary in this aspect. Mostly, the loss in the WCI is covered by the total premium from both policies. Most of the insurers prefer selling insurance pack- ages rather than individual policies. Hence, the placement of mul- tiple policies is more of a prerequisite than a discounting factor in Singapore.

Cooperation by the Contractor Singapore’s construction industry is heavily reliant on foreign workers from neighboring developing nations such as India, China, Bangladesh, Thailand, and the Philippines. The objective of these workers is to earn money as fast as possible to return to their home countries for a comfortable life. There were incidents in Singapore whereby: �1� some of the injured workers chose to claim by common law to reap greater compensation; and �2� some workers injured themselves or acted recklessly in order to claim compensation �Kyodo 1991�. It is therefore important for insurers to scrutinize the root cause�s� of accident�s� and any fraudulent action�s� of workers to diminish their risks. Contractors’ support is paramount in such situations, which could be a consideration in premium rating. Nonetheless, it is a weak factor because there could still be claims although the contractor is supportive.

Volume of Business in the Market It is acceptable to increase premium rates when the construction industry undergoes an economic boom. Nonetheless, more insur- ers enter the market when the demand is high. This makes the Singapore insurance market constantly competitive.

Unimportant Variables for Premium Rating

Two insignificant variables for premium rating are identified below.

Expectation of Potential Business from the Contractor The expectation of potential business from contractors are irrel- evant for premium rating because construction projects are se- cured through competitive bidding, thus the prediction of potential projects for contractors is difficult.

Contractor’s Size Large contractors with good safety records demonstrate their ca- pacity to use sophisticated and less hazardous construction meth- ods. It is therefore beneficial for insurers to favor large contractors. However, all main contractors employ subcontrac- tors. Thus, subcontractors’ experience, operation methods, and records become key factors. As per insurers’ experience, reported accidents mostly involved subcontractors’ workers whereby the main contractor provided the WCI.

Summary of Survey

The results indicate that the following eight variables are impor- tant for WCI premium rating: �1� wage roll; �2� project hazard

JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT © ASCE / AUGUST 2008 / 605

level; �3� effectiveness of the safety management system on site; �4� contractor’s claims history; �5� overhead costs of insurance; �6� competition; �7� corporate objectives; and �8� investment in- come from underwritten premiums. Hence, these eight variables are encompassed for developing a succinct and optimum premium-rating model.

Proposed Model for WCI Premium Rating of Building Construction Projects

A new model for WCI premium rating of building construction projects was developed, as illustrated in Fig. 2, based on the find- ings of the literature review and the industry survey.

As per the new model, it is proposed that:

1. The net optimal WCI premium for a construction project would have three components as shown in formula �1�

WCI net premium = Risk fee − CCIContractor − SMDClient �1�

where Risk fee�gross price of the risk covered by the in- surer and it has to be paid by the contractor at the underwrit- ing stage and subsequently reimbursed by the client via interim payments; CCIContractor�claim control incentive for the contractor, which is to be furnished at the policy expira- tion stage for controlling the actual claims below the pre- dicted amount; and SMDClient�safety monitoring discount for the client, which is to be furnished at the policy expira-

Fig. 2. Proposed model for WCI premium rating

606 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT © ASCE / AUGUST 2008

tion stage for monitoring the contractor’s safety management on site, which minimized claims in the project.

2. The computation of the net premium would be performed in two stages, viz.:

a. The computation of the risk fee would be carried out at the underwriting stage �construction bidding stage�; and

b. The computation of the CCIContractor, SMDClient, and the net premium would be carried out at the policy expira- tion stage.

Computing Risk Fee

As described above, the risk fee is the gross price for the accident risks that an insurer will have to assume in the project. This is estimated at the bidding stage of a project, according to formula �2�

Risk fee = �1 + K� � C �2�

where C�expected claims in the project; and K�markup for the policy.

The prediction of the expected claims �C� in a project would adopt formula �3�

C = Wage roll � Mean claims unit cost � Risk rate �3�

where Wage roll�total wages of the workers in the project; Mean claims unit cost�average break-even point of workers’ compen- sation claims for building construction projects for the insurer; and Risk rate�degree of risk that would deviate the actual claims

unit cost in the project from the mean claims unit cost. The risk rate for a given project is deduced by a fuzzy algo-

rithm in the fuzzy KBS developed to automate the proposed model. The KBS is not discussed here due to word limitation and it is a potential topic for another paper. However, a glimpse of the algorithm is described in the Appendix and the relationship be- tween variables is illustrated plainly in formula �4� below. Addi- tionally, the rationale for utilizing these three variables in the formula is described below

Risk rate = f a �Wage roll, PHI, PSI� �4�

Wage Roll As per insurers’ experience, the claims rate is not linearly corre- lated with the project size. Thus, an appropriate rate that may not be linearly distributed should be matched with the project size, which is represented by the wage roll size.

Project Hazard Index The hazard level in a project is scrutinized by assessing the project scope and the vicinity/location. Eleven hazardous trades in building projects and their respective attributes for assessing each trade’s hazard were identified in the literature. Based on that, a framework for estimating the PHI for building projects was de- veloped as shown in Table 3. Nevertheless, not every hazardous trade may be applicable to a given project. Relevant trades need to be chosen and hazard rated. Hence, the PHI is derived by the following normalized formula:

PHI = 1

m �DMHscore + EXHscore + SLHscore + FLHscore + RFHscore + ERHscore + CRHscore

+ MTHscore + CsiteHscore + WCHscore + CspaceHscore � �5�

where 0 � m � 11; DMHscore�degree of hazard contributed by demolition works; EXHscore�degree of hazard contributed by ex- cavation works; SLHscore�degree of hazard contributed by scaf- folding and ladder use; FLHscore�degree of hazard contributed by false works; RFHscore�degree of hazard contributed by roof works; ERHscore�degree of hazard contributed by erection works; CRHscore�degree of hazard contributed by crane use; MTHscore�degree of hazard contributed by machinery and tools use; CsiteHscore�degree of hazard contributed by works on con- taminated sites; WCHscore�degree of hazard contributed by weld- ing and cutting works; and CspaceHscore�degree of hazard contributed by works in confined spaces.

The computation of individual scores will exploit the follow- ing algorithm:

IF demolition hazard = true THEN

DMHscore = 1

3 �

5 � a=1

Demolition hazard attribute scorea

ELSE DMHscore = 0

ENDIF �6�

The coefficients of 1/3 and 1/5 are included because the hazard score for demolition works is computed by equally assessing three obligatory attributes on a 1–5 scale, and then the score is

normalized to 1.00. A similar approach is pursued to compute other hazard trade scores too.

Project Safety Index (PSI) The PSI portrays the effectiveness of the site safety management system �0 � PSI � 1.00�. If the safety management system is suf- ficiently robust, the PSI will yield a score of 1.00. Otherwise, it will yield a score of between 0 and 1.00, depending on the degree of safety preparedness. The PSI estimation requires an exhaustive safety audit. Eight safety factors and their subfactors of a safety auditing model for construction projects were identified in the literature. Based on that, a framework for PSI estimation was developed as depicted in Table 4. The PSI is derived based on the following normalized formula:

PSI = 1

8 �PSOscore + RAMscore + SWPscore + STCscore + SIscore

+ SMTscore + SMscore + EMscore� �7�

where PSOscore�adequacy score for project safety organization; RAMscore�adequacy score for risk assessment and management system; SWPscore�adequacy score for safe work practices; STCscore�adequacy score for safety training and competency of people involved; SIscore�adequacy score for safety inspection sys- tem; SMTscore�adequacy score for safe use and maintenance of

JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT © ASCE / AUGUST 2008 / 607

Table 3. Framework for Estimating PHI

Estimating project hazard

1. Demolition works

Rate the level of hazard posed by the following parameters in demolition works in this project.

Low………………………….High • Volume/size of demolition 1 2 3 4 5

• Type of structure 1 2 3 4 5

• Method of demolition 1 2 3 4 5

Trade score

2. Excavation works

Rate the level of hazard posed by the following parameters in excavation works in this project.

• Excavation configuration �depth, width, and length� 1 2 3 4 5

• Geological condition �soil type, water table, etc.� 1 2 3 4 5

• Underground utilities �electrical, water, and sewer lines� 1 2 3 4 5

• Nearby vehicular traffic �vibration and surcharge� 1 2 3 4 5

• Nearby structures 1 2 3 4 5

Trade score

3. Scaffolding and ladder usage

Rate the level of hazard posed by the following parameters in scaffolding and ladder usage in this project.

• Volume of scaffolding and ladder usage 1 2 3 4 5

• Height of the scaffold/ladder to be used 1 2 3 4 5

• Adequacy of design �type of material, member size, bracing, guardrails, platform size, toe board� 1 2 3 4 5

Trade score

4. Falseworks �temporary structures�

Rate the level of hazard posed by the following parameters in falseworks in this project.

•Volume of falsework involved in the project 1 2 3 4 5

• Adequacy of design �material, member size, bracing, guardrails, platform size, toe board� 1 2 3 4 5

Trade score

5. Roof works

Rate the level of hazard posed by the following parameters in roof works in this project.

• Volume of roofing involved 1 2 3 4 5

• Height of the roof 1 2 3 4 5

• Roofing material property such as slippery, brittleness, asbestos etc. 1 2 3 4 5

• Inclination of the roof 1 2 3 4 5

Trade score

6. Erection of steel/precast concrete structures

Rate the level of hazard posed by the following parameters in erection of steel/precast concrete structures in this project.

• Volume of erection work 1 2 3 4 5

• Height of erection work 1 2 3 4 5

• Erection method �partial/full erection at height, labor involvement level� 1 2 3 4 5

Trade score

7. Crane use

Rate the level of hazard posed by the following parameters in lifting and crane use in this project.

• Volume of lifting involved 1 2 3 4 5

• Nature of materials lifted 1 2 3 4 5

•Operating platform 1 2 3 4 5

• Nature of site vicinity �nearby structures, overhead cables, etc.� 1 2 3 4 5

Trade score

8. Construction tools and machinery use

Rate the level of hazard posed by the following parameters in plant and tools use in this project.

• Volume of plant and machinery used 1 2 3 4 5

• Operating platform of plant and machinery �i.e., slope etc.� 1 2 3 4 5

• Site layout 1 2 3 4 5

• Volume of tools used 1 2 3 4 5

• Type of tools used 1 2 3 4 5

Trade score

9. Works on contaminated sites

Rate the level of hazard posed by the following parameters in working on contaminated site in this project.

608 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT © ASCE / AUGUST 2008

machinery and tools regime; SMscore�adequacy score for subcon- tractors’ safety systems; and EMscore�adequacy score for emer- gency management system.

For each safety factor the score is the sum of the attribute scores divided by the number of obligatory attributes, and divided again by the range of the scoring system. For example, PSOscore is computed by equally assessing three obligatory attributes of the factor on a 1–5 scale. Therefore, the PSOscore is calculated as per Eq. �8�. A similar method is adopted to calculate the scores for the other safety factors too

PSOscore = 1

3 �

5 � a=1

PSO attribute scorea �8�

However when there are safety factors with nonobligatory at- tributes, a further filtering is carried out. For example, the com- putation of the adequacy score for safe work practices �SWPscore� needs to analyze three subfactors, viz: • Work procedures; • Permit-to-work �PTW� system; and • Personal protective equipment �PPE� use. The subfactor score for work procedures is computed by equally assessing 13 attributes on a 1–5 scale. However, it is not neces- sary that all the 13 attributes be applicable to a given project. In such cases, irrelevant attributes are marked as not applicable �NA�. Thus, the following normalized equation is adopted to compute the subfactor score:

WPsub-factor score = 1

p �

5 � c=1

WP attribute scorec �9�

Here, p represents the number of applicable attributes to the project for this subfactor. If all 13 attributes are applicable, the weightage of each attribute will be 1/13. Suppose only ten at- tributes are applicable for a particular project, then the weightage will be 1/10 to normalize the cumulative attribute scores. This will therefore maintain a consistency in the safety rating and com- parison of different projects.

Akin to the above subfactor, the equations to compute the subfactor scores for the other two subfactors are normalized as follows:

PTWsub-factor score = 1

q �

5 � d=1

PTW attribute scored �10�

PPEsub-factor score = 1

r �

5 � e=1

PPE attribute scoree �11�

where q and r represent the number of applicable attributes to the project for the subfactors of PTW and PPE, respectively.

The three subfactors are given equal weightage towards com- puting the SWPscore. Hence, the SWPscore equation is written as

SWPscore = 1

3 �WPsub-factor score + PTWsub-factor score

+ PPEsub-factor score �12�

Computing Markup

The derivation of the markup �K� is performed by assessing five variables: �1� overhead costs of the policy; �2� investment income from underwritten premium; �3� corporate objective�s�; �4� com- petition; and �5� contractor’s claims history. The formula for K is written as

K = f b�Overhead costs, Investment income,

Corporate objectives, Competition, Claims history�

�13�

Akin to the risk rate formula, this formula was also translated into a fuzzy algorithm, which is discussed briefly in the Appendix.

Computing CCIContractor and SMDClient

Many insurers appear to have some sort of discount for contrac- tors for keeping their workers’ compensation claims to a mini- mum. However, the method used to compute the discount amount is not well defined. Huang and Hinze �2006� empirically proved that clients’ involvement in safety management during the course of construction significantly improves safety performances on sites. It is therefore equally important to recognize and encourage

Table 3. �Continued.�

Estimating project hazard

• Type of contaminants on the site 1 2 3 4 5

• Quantity of contaminants present 1 2 3 4 5

• Duration of work on contaminated site 1 2 3 4 5

Trade score

10. Welding and hot works

Rate the level of hazard posed by the following parameters in welding and hot works in this project.

• The volume of welding and hot works 1 2 3 4 5

• Location of welding �confined space, underground, on ladders etc.� 1 2 3 4 5

Trade score

11. Works in confined spaces

Rate the level of hazard posed by the following parameters in confined space works in this project.

• The volume of confined space works 1 2 3 4 5

• Confined space configuration 1 2 3 4 5

• Type of activity to be involved �e.g., welding, waterproofing, etc.� 1 2 3 4 5

• Current usage of the confined space �if any� 1 2 3 4 5

Trade score

Total project score �PHI�

JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT © ASCE / AUGUST 2008 / 609

Table 4. Framework for Estimating PSI

Estimating project safety index

�a� Project safety organization

Please rate the adequacy of the duties and responsibilities of the following personnel/team in the project safety organization:

Low…………….High

1. Workplace safety and health coordinator 1 2 3 4 5

2. Workplace safety and health auditor 1 2 3 4 5

3. Workplace safety and health committee 1 2 3 4 5

Section score

�b� Risk assessments and management system

Please rate the adequacy of the following aspects of the risk assessment and management system in the project.

Low…………….High

1. Risk assessment team and responsibilities 1 2 3 4 5

2. Risk assessment procedures 1 2 3 4 5

3. Reporting procedures to workers of identified risks 1 2 3 4 5

4. Control measures for risks identified 1 2 3 4 5

Section score

�c� Safe work practices

�C.1� Work procedures:

Please rate the effectiveness of the work methods and procedures for the following trades. Low……………High

1. Concrete works 1 2 3 4 5 NA

2. Structural steel and precast assembly 1 2 3 4 5 NA

3. Erection and dismantling of scaffolds and falseworks 1 2 3 4 5 NA

4. Works at heights 1 2 3 4 5 NA

5. Demolition works 1 2 3 4 5 NA

6. Excavation works 1 2 3 4 5 NA

7. Piling operations 1 2 3 4 5 NA

8. Welding and cutting works 1 2 3 4 5 NA

9. Works in confined spaces 1 2 3 4 5 NA

10. Works in toxic/contaminated environments 1 2 3 4 5 NA

11. Use of construction plant such as excavators and trucks 1 2 3 4 5 NA

12. Use of cranes 1 2 3 4 5 NA

13. Electrical installation and use 1 2 3 4 5 NA

Subsection score

(C.2) Permit-to-work (PTW) system:

Please rate the effectiveness of the PTW systems for the following trades: Low…………..High

1. Working at heights 1 2 3 4 5 NA

2. Excavation works 1 2 3 4 5 NA

3. Working in confined spaces 1 2 3 4 5 NA

4. Welding and cutting works 1 2 3 4 5 NA

5. Demolition works 1 2 3 4 5 NA

6. Working in toxic/contaminated environments 1 2 3 4 5 NA

Subsection score

(C.3) Personal protective equipment (PPE) use

Please rate the adequacy of the PPE use for the following trades: Low…………..High

1. Concrete works 1 2 3 4 5 NA

2. Structural steel and precast assembly 1 2 3 4 5 NA

3. Erection and dismantling of scaffolds and falseworks 1 2 3 4 5 NA

4. Works at heights 1 2 3 4 5 NA

5. Demolition works 1 2 3 4 5 NA

6. Excavation works 1 2 3 4 5 NA

610 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT © ASCE / AUGUST 2008

Table 4. �Continued.�

Estimating project safety index

7. Piling operations 1 2 3 4 5 NA

8. Welding and cutting works 1 2 3 4 5 NA

9. Works in confined spaces 1 2 3 4 5 NA

10. Works in toxic/contaminated environments 1 2 3 4 5 NA

11. Use of construction plant such as excavators and trucks 1 2 3 4 5 NA

12. Use of cranes 1 2 3 4 5 NA

13. Electrical installation and use 1 2 3 4 5 NA

Subsection score

Section score

�D� Safety training and competency of people involved

�D.1� Safety training to management team

Please rate the adequacy of the safety training to the following personnel in the project: Low……………High

1. Demolition supervisor�s� 1 2 3 4 5 NA

2. Excavation supervisor�s� 1 2 3 4 5 NA

3. Piling supervisor�s� 1 2 3 4 5 NA

4. Lifting supervisor�s� 1 2 3 4 5 NA

5. Scaffold and/or suspended scaffold supervisor�s� 1 2 3 4 5 NA

6. Falseworks supervisor�s� 1 2 3 4 5 NA

7. Welding & cutting supervisor�s� 1 2 3 4 5 NA

8. Confined space work supervisor�s� 1 2 3 4 5 NA

9. Toxic/contaminated environment work supervisor�s� 1 2 3 4 5 NA

10. Project management team members 1 2 3 4 5 NA

Subsection score

(D.2) Certification & safety training of operators

Please rate the adequacy of the certification & safety training of the following operators in the project: Low……………High

1. Crane erector�s� 1 2 3 4 5 NA

2. Crane operator�s� 1 2 3 4 5 NA

3. Riggers�s� 1 2 3 4 5 NA

4. Signal men 1 2 3 4 5 NA

5. Scaffold erector�s� and/or suspended scaffold rigger�s� 1 2 3 4 5 NA

6. Erectors of hoists and lifts 1 2 3 4 5 NA

7. Operators of hoists and lifts 1 2 3 4 5 NA

8. Operators of plant like excavators, bull dozer, etc. 1 2 3 4 5 NA

9. Construction vehicle drivers 1 2 3 4 5 NA

Subsection score

(D.3) In-house safety training to workers

Please rate the adequacy of the following modules of the in-house safety training to workers in the project: Low…………..High

1. Site rules and regulations, and proper use of PPE 1 2 3 4 5

2. Emergency response for various possible incidents 1 2 3 4 5

3. First aid procedures 1 2 3 4 5

4. Safe handling of tools and equipment 1 2 3 4 5

Subsection score

Section score

�E� Safety inspection system

(E.1) Inspection of worksite

Please rate the adequacy of the inspection system for the following items in the project: Low…………..High

1. Excavation by a competent person on a daily basis and after hazardous events �e.g., inclement weather� 1 2 3 4 5 NA

2. Scaffolding by a scaffold supervisor on a weekly basis and after inclement weather 1 2 3 4 5 NA

JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT © ASCE / AUGUST 2008 / 611

Table 4. �Continued.�

Estimating project safety index

3. Falseworks by a PE or other competent person before, during, and after casting and after inclement weather 1 2 3 4 5 NA

4. Demolition by a competent person on a daily basis and after inclement weather 1 2 3 4 5 NA

5. Material loading platform by a competent person on a regular basis and after inclement weather 1 2 3 4 5 NA

6. Temporary structures such as site office, canteen, site hoardings, and concrete batching plant on a regular basis

1 2 3 4 5 NA

7. Specialized structures or operations like use of customized shoring systems by a competent person 1 2 3 4 5 NA

8. General site by a safety personnel or the site manager 1 2 3 4 5 NA

Subsection score

(E.2) Housekeeping

Please rate the adequacy of the housekeeping for the following locations/items in the project: Low……………High

1. Construction worksite 1 2 3 4 5 NA

2. Workers’ quarters 1 2 3 4 5 NA

3. Toilets and washing facilities 1 2 3 4 5 NA

4. Canteen or eating places 1 2 3 4 5 NA

5. Site offices 1 2 3 4 5 NA

6. Storages for materials, tools & wastes 1 2 3 4 5 NA

Subsection score

Section score

�F� Machinery and tools use and maintenance regime

(F.1) Testing & certification of machinery

Please rate the adequacy of the testing & certification of the following machinery in the project Low……………High

1. Lifting gears �12 monthly� 1 2 3 4 5 NA

2. Lifting appliances �12 monthly� 1 2 3 4 5 NA

3. Lifting machines �12 monthly� 1 2 3 4 5 NA

4. Hoists and lifts �6 monthly� 1 2 3 4 5 NA

5. Air receivers �24 monthly� 1 2 3 4 5 NA

6. Explosive power tools �36 monthly� 1 2 3 4 5 NA

Subsection score

(F.2) Inspection of machinery & tools

Please rate the adequacy of the inspection system for the following machinery in the project: Low…………..High

1. Cranes by crane operators on a daily basis 1 2 3 4 5 NA

2. Electrical distribution board by a competent person on a daily basis 1 2 3 4 5 NA

3. Electrical equipment and tools by a competent person on a regular basis �weekly/more frequent� 1 2 3 4 5 NA

4. Construction vehicles like trucks, forklift, bull dozer, etc. by drivers or a designated person on a daily basis 1 2 3 4 5 NA

5. Temporary electrical installation by a licensed electrical worker 1 2 3 4 5 NA

6. Specialized equipment by a competent person 1 2 3 4 5 NA

Subsection score

(F.3) Maintenance of machinery

Please rate the adequacy of the maintenance regime for the following machinery in the project: Low……………High

1. Tower crane�s� 1 2 3 4 5 NA

2. Mobile crane�s� 1 2 3 4 5 NA

3. Gondola�s� 1 2 3 4 5 NA

4. Piling machine�s� 1 2 3 4 5 NA

5. Passenger hoist�s� 1 2 3 4 5 NA

6. Mobile working platform�s� 1 2 3 4 5 NA

7. Construction vehicles like truck, forklift, bulldozer, etc. 1 2 3 4 5 NA

Subsection score

Section score

�G� Subcontractors’ safety systems

clients’ involvements through incentives as a risk control strategy. Hence, CCIContractor and SMDClient are claims experiential incen- tives for the contractor and the client, respectively. The CCIContractor and the SMDClient are preferred to be equal amounts and the following algorithm produces the respective amounts:

If “TAC � C” Then

CCIContractor = SMDClient = �0.5 � � � �C − TAC� Else

CCIContractor = SMDClient = 0

Endif �14�

where TAC�total of actual claims incurred in the project; and ��discounting factor; the � value is at the discretion of the in- surer and it may equate to 2 / 3, 1 / 2, or as the case may be. If an

insurer chooses 2/3 for �, for example, this implies that the in- surer is willing to share the savings equally among the insurer, client, and contractor.

The insurer, the contractor, or the client will not know the amounts for the CCIContractor and SMDClient at the policy under- writing stage. However, the contractor and the client must be informed about the discount/incentive scheme so that they will be motivated to execute the implemented safety management system effectively on site.

Validation of Proposed Model

An empirical test was conducted to ascertain the accuracy and effectiveness of the proposed model. WCI data on four recently completed projects were collected from an insurer. The collected

Table 4. �Continued.�

Estimating project safety index

Please rate the adequacy of the following items of subcontractors in the project: Low…………..High

1. Safe work procedures 1 2 3 4 5

2. Safe use of plant, machinery, and tools 1 2 3 4 5

3. Safety inspection system 1 2 3 4 5

4. Trained operatives and supervisors 1 2 3 4 5

5. Adherence to safety requirements during construction 1 2 3 4 5

Section score

�H� Emergency management system

(H.1) Emergency response plan

Please rate the adequacy of the emergency response plan for the following emergency scenarios in the project: Low…………..High

1. Fire and explosion 1 2 3 4 5

2. Failure and collapse of structures/temporary supports 1 2 3 4 5

3. Failure and collapse of heavy machinery and equipment 1 2 3 4 5

4. Leakage of hazardous substances 1 2 3 4 5

5. Adverse weather and flooding 1 2 3 4 5

Subsection score

(H.2) Emergency response team

Please rate the adequacy, competency, and set responsibilities of the following emergency response team members for various emergency scenarios in the project:

Low………….High

1. Emergency coordinator�s� 1 2 3 4 5

2. Site safety personnel 1 2 3 4 5

3. Designated rescuer�s� 1 2 3 4 5

4. First aider�s� 1 2 3 4 5

5. Specialist operators�s� 1 2 3 4 5

Subsection score

(H.3) Emergency equipment

Please rate the adequacy of the emergency response equipment and facilities for the following emergency scenarios in the project:

Low…………High

1. Fire and explosion 1 2 3 4 5

2. Failure and collapse of structures/temporary supports 1 2 3 4 5

3. Failure and collapse of heavy machinery and equipment 1 2 3 4 5

4. Leakage of hazardous substances 1 2 3 4 5

5. Adverse weather and flooding 1 2 3 4 5

Subsection score

Section score

data were summarized to check the net gain/loss earned by the insurer from the premium. Then, premium computation for all the projects was performed by the underwriter of the company using the proposed model �the fuzzy KBS was used as it is the comput- erized version of the proposed model�. A comparison of the pre- miums is illustrated in Table 5. It was found that the proposed model produced premiums that can cover the risk optimally with- out leading to loss while being concerned with the competition.

Comparison of Proposed Model with EMR Technique

The computation methods of the expected claims in a building project in the proposed model and the EMR approach are formu- lated in Eqs. �15� and �16�, respectively

C = Wage roll � Mean claims unit cost � Risk rate �15�

Standard premium = Payroll unit � Manual rate � EMR

�16�

The major distinctions between these techniques are described below. 1. The payroll unit in the EMR approach is substituted by the

wage roll in the proposed model. The payroll unit accounts for the number of person hours of risk exposure for a par- ticular type of work. It is calculated as

Payroll unit = Employer’s straight time direct labor cost

$ 100

The computation of the payroll unit in a building project needs to consider the amount of pay for each work classifi- cation. There are a number of work classifications in a project, thus making the task tedious. However, the proposed model only needs a single piece of data, i.e., wage roll, and therefore simplifies the prediction;

2. The mean claims unit cost in the proposed model replaces the manual rate in the EMR approach. The manual rate is calculated by using the following formula:

Manual rate = �Benefits paid for a work type + overhead cost for administering�

Straight time payroll

The manual rate gives a rough indication of the risk associated with each work classification. Some work classifications are more likely to result in injuries and fatalities than others. Thus, the manual rate varies among work classifications. The final value for the manual rate for each work classification is based on claims that have been filed for that work classification. The computation of the manual rate involves the collection and processing of claims data for each work classification in a particular state by an external body, which in turn induces insurance companies to rely on external data. On the contrary, the mean claims unit cost in the proposed model reveals the overall risk level associated with the construction WCI business for a particular insurer and therefore simplifies the calculation process. It also eliminates the need for disclosing an insurer’s claims experience to other entities and relying on any external data for premium rating. In addition, the use of a company-specific mean claims unit cost distinguishes one

insurance company from its rivals, therefore giving a competitive advantage; 3 The EMR is replaced by the risk rate in the proposed model.

The EMR accounts for the loss experience of a contractor and is used to modify the manual rate. It helps predict the future losses based on past experience. A contractor with poor safety records will have a higher EMR and pay more for the WCI. However, the EMR is a lagging indicator. The ex- perience period used in the EMR calculation process is the 3-year period ending 1 year prior to the date that the modi- fication becomes effective. The most recent year is not used; although there may have been an improvement in the safety records, it is not accounted for the next 2 years. In contrast, the risk rate in the proposed model accounts for the project hazard level and project safety level at the time of insurance application to derive a real-time adjustment factor;

Table 5. Model Validation

Project

Wage roll Premium charged

by the insurer

�$�

Actual loss ratio

Premium suggested by the proposed model

Possible loss ratio

Possible loss reduction by the proposed model

�$� �$� Actual claims

incurred �$� �%�

A 8,000,000 100,000 70,000 0.70 108,192 0.65 5

B 200,000 3,000 9,000 3.00 6,076 1.48 152

C 800,000 8,000 14,000 1.75 14,733 0.95 80

D 740,000 35,000 8,000 0.23 13,442 0.60 −37

4. The EMR approach is silent with respect to the determination of the markup, whereas the proposed model consists of a module to determine the markup for WCI policies; and

5 Unlike the EMR approach, the proposed model advocates a well-structured risk control strategy via an established incen- tive scheme.

Conclusion

This study proposes a new premium-rating model for construction projects. As per the proposed model, the WCI premium for a construction project has three components: �1� risk fee—paid by the contractor at the underwriting stage and subsequently reim- bursed by the project client for the risk inherent in the project; �2� CCIcontractor—to be returned to the contractor by the insurer at policy expiration if the claims were below the predicted amount; and �3� SMDclient—to be returned to the project client by the insurer at policy expiration if the client had monitored the con- tractor’s safety level so as to minimize accidents. The risk fee is computed by: �1� predicting the potential claims in the project by assessing the wage roll, project hazard level, and the project safety level; and �2� marking up the predicted claims value to account for insurer’s overhead costs, corporate objectives, invest- ment income, and competition. The amounts for CCIcontractor and SMDclient are founded on the total amount of actual claims in the project. If the total actual claims are lesser than the predicted claims at the underwriting stage, the saving will be shared by the insurer with the contractor and the client. The proposed new model also addresses most of the shortcomings of the traditional EMR technique.

Accidents seem to be inevitable on construction sites. Various agencies and institutions are trying to control accidents through different approaches. This study suggests accident control via workers’ compensation insurance premiums for contractors: a cost-driven methodology to minimize accidents. As per the pro- posed WCI premium-rating model, the premium amount for a WCI policy is significantly influenced by the effectiveness of the contractor’s safety management system, which needs to be as- sessed in real time. If a contractor’s real-time safety management

system is robust, the contractor will get cost effective insurance coverage, which will enable the contractor to compete better in tenders. Those contractors who do not have robust safety man- agement systems in place will be penalized by higher premiums. This approach will automatically set risky contractors aside as well as motivate contractors to invest in safety in their organiza- tions for ensuring business continuity. The proposed new model also advocates the partnering phenomenon in WCI for construc- tion, whereby a postproject discount system is introduced to en- courage the involvement of both contractors and clients in reducing claims or improving safety during the course of con- struction. Hence, implementation of the proposed model in the insurance industry would facilitate accident control in the con- struction industry and thereby loss minimization for insurers.

Appendix. Fuzzy Algorithms for Premium Formulas

The fuzzy KBS architecture that automates the proposed premium-rating model is depicted in Fig. 3, which consists of five components: �1� graphical user interface �GUI�; �2� system data- base; �3� intermediate processing unit �IPU�; �4� fuzzy knowledge base; and �5� fuzzy inference engine. The GUI receives input from users and returns outputs after processing the inputs. The system database is partitioned into two components: historical claims repository, and WCI details repository. The historical claims repository stores the data such as wage roll and total paid claims in previous projects. The stored data are utilized to com- pute the real-time mean claims unit cost. The mean claims unit cost dictates the average cost per wage roll that was paid in the past for insuring building construction projects. This can be uti- lized to predict potential claims from new projects, subject to adjustments for projects’ particularity. The WCI details repository stores the policy details that are relevant to building projects until the respective WCI policies become obsolete. The IPU contains three subcomponents, namely a PHI computer—for computing the PHI for the project, based on the project hazard attribute val- ues as fed in by the user, a PSI computer—for computing the PSI for the project, based on the safety attribute values that are input by the user; and a Premium calculator—for calculating the risk

Fig. 3. Fuzzy KBS architecture

Fig. 4. Fuzzy membership functions

fee, CCIcontractor and SMDclient. The fuzzy inference engine, with the support of the fuzzy knowledge base, makes deductions to derive the risk rate that adjusts the mean claims unit cost and the appropriate markup �K� that adjusts the expected claims �C�. The knowledge base consists of fuzzy membership functions for the variables �see Fig. 4� and fuzzy rules that were derived based on the union rule configuration �URC� method. A glimpse of the fuzzy URC rules for the KBS is given below.

Rules: ��IF PHI is Very small THEN risk rate is Very low� OR

�IF PHI is Small THEN risk rate is Low�

……………….. OR

��IF competition is Very strong THEN mark-up is Negative big� OR

�IF competition is Strong THEN mark-up is Negative moderate�

………….

References

Booth, P., Chadburn, R., Cooper, D., Haberman, S., and James, D. �1999�. Modern actuarial theory and practice, Chapman & Hall/CRC, New York.

Building and Construction Authority �BCA�. �2006�. “Working together.” Annual Rep. 2005, �http://www.bca.gov.sg/AboutUs/annual_report_ 05.html� �Aug. 11, 2006�.

Canadian Wood Council. �2005�. “Course of construction insurance ba- sics.” �http://www.cwc.ca/pdfs/Quick_factsInsurance_and_ Construction_Series_No1.pdf� �Sept. 29, 2005�.

Chua, D. K. H., and Goh, Y. M. �2004�. “Incident causation model for improving feedback of safety knowledge.” J. Constr. Eng. Manage., 130�4�, 542–551.

Clough, R. H., Sears, G. A., and Sears, S. K. �2005�. Construction con- tracting, 7th Ed., Wiley, New York.

Coble, R. J., McDermott, M. J., and Carpenter, S. �1998�. “The liability of workers’ compensation exemptions in Florida.” Cost Eng., 40�5�, 34–41.

Coble, R. J., and Sims, B. L. �1996�. “Workers’ compensation fraud in construction.” Trans. Am. Assn. Cost. Eng., 117�3�, C&C.6.1–6.

Everett, J. G., and Thompson, W. S. �1995�. “Experience modification rating for workers’ compensation insurance.” J. Constr. Eng. Man- age., 121�1�, 66–79.

Groth, C. �1996�. “Workers’ compensation.” Construction insurance, bonding, and management, W. J. Palmer, J. M. Maloney, and J. L. Heffron, eds., McGraw-Hill, New York, 125–143.

Haslam, R. A., Hide, S. A., Gibb, A. G. F., Gyi, D. E., Pavitt, T., Atkin- son, S., and Duff, A. R. �2005�. “Contributing factors in construction accidents.” Appl. Econom., 36�5�, 401–415.

Hinze, J., Bren, D. C., and Piepho, N. �1995�. “Experience modification rating as measure of safety performance.” J. Constr. Eng. Manage., 121�4�, 455–458.

Hoonakker, P., Loushine, T., Carayon, P., Kallman, J., Kapp, A., and Smith, M. J. �2005�. “The effect of safety initiatives on safety perfor- mance: A longitudinal study.” Appl. Econom., 36�4�, 461–469.

Huang, X., and Hinze, J. �2006�. “Owner’s role in construction safety: Guidance model.” J. Constr. Eng. Manage., 132�2�, 174–181.

Kyodo. �1991�. “Foreign workers cut own fingers for money, insurers say.” Singapore suspects foreign workers, �http://www.hamline.edu/ apakabar/ basisdata/1991/02/14/0007.html� �Apr. 17, 2006�.

Lott, R. J. �2005�. “Controlling workers’ compensation cost: A risk man- agement program.” �http://www.healthconsultantsusa.com/Risk_ Mgmt_Program_Controlling_Worker_Comp_Costs. pdf� �Jun. 27, 2005�.

National Safety Council. �1997�. Accident facts, Itasca, Ill. Phifer, R. �1996�. Reinsurance fundamentals, Wiley, New York. Vaughan, E. J., and Vaughan, T. M. �1996�. Fundamentals of risk and

insurance, 7th Ed., Wiley, New York. Worrall, J. D., and Buttler, R. J. �1988�. “Experience rating matters.”

Workers’ compensation insurance pricing, D. Appel and P. S. Borba, eds., Kluwer Academic, Boston, 81–94.

Young, V. R. �1997�. “Adjusting indicated insurance rates: Fuzzy rules that consider both experience and auxiliary data.” Casualty Actuarial Society, 84�161�, 734–765.