Forecasting Product Liability Claims

by Stallard, Eric; Manton, Kenneth G.; Cohen, Joel E.; Weinstein, Jack B.

• ISBN13: 9780387949871
• ISBN10: 0387949879
• Format: Hardcover
• Copyright: 2004-10-30
• Publisher: Springer Verlag

Summary

This volume presents a rigorous account of statistical forecasting efforts that led to the successful resolution of the Johns-Manville asbestos litigation. This case, taking 12 years to reach settlement, is expected to generate nearly 500,000 claims at a total nominal value of over $34 billion. The forecasting task, to project the number, timing, and nature of claims for asbestos-related injuries from a set of exposed persons of unknown size, is a general problem: the models in this volume can be adapted to forecast industry-wide asbestos liability. More generally, because the models are not overly dependent on the U.S. legal system and the role of asbestos as a dangerous/defective product, this volume will be of interest in other product liability cases, as well as similar forecasting situations for a range of insurable or compensable events. The volume stresses the iterative nature of model building and the uncertainty generated by lack of complete knowledge of the injury process. This uncertainty is balanced against the Court's need for a definitive settlement, and the volume addresses how these opposing principles can be reconciled.The volume is written for a broad audience of actuaries, biostatisticians, demographers, economists, epidemiologists, environmental health scientists, financial analysts, industrial-risk analysts, occumpational health analysts, product liability analysts, and statisticians. The modest prerequisites include basic concepts of statistics, calculus, and matrix algebra. Care is taken that readers without specialized knowledge in these areas can understand the rationale for specific applications of advanced methods. As a consequence, this volume will be an indispensable reference for all whose work involves these topics.Eric Stallard, A.S.A., M.A.A.A., is Research Professor and Associate Director of the Center for Demographic Studies at Duke University. He is a Member of the American Academy of Actuaries and an Associate of the Society of Actuaries. He serves on the American Academy of Actuaries Committees on Long Term Care and Social Insurance. He also serves on the society of Actuaries' Long Term Care Experience Committee. His research interests include modelling and forecasting for medical demography and health actuarial practice. He was the 1996 winner of the National Institute on Aging's James A. Shannon Director's Award.Kenneth G. Manton, Ph.D., is Research Professor, Research Director, and Director of the Center for Demographic Studies at Duke University and Medical Research Professor at Duke University Medical Center's Department of Community and Family Medicine. Dr. Manton is also a Senior Fellow of the Duke University Medical Center's Center for the Study of Aging and Human Development. His research interests include mathematical models of human aging, mortality, and chronic disease. He was the 1990 recipient of the Mindel C. Sheps Award in Mathematical Demography presented by the Population Association of America; and in 1991 he received the Allied-Signal Inc. Achievement Award in Aging administred by the Johns Hopkins Center on Aging.Joel E. Cohen, Ph.D., Dr. P.H., is Professor of Populations, and Head of the Laboratory of Populations, Rockefeller University. He also is Professor of Populations at Columbia University. His research interests include the demography, ecology, epidemiology, and social organization of human and non-human populations, and related mathematical concepts. In 1981, he was elected Fellow of the MacArthur and Guggenheim Foundations. He was the 1992 recipient of the Mindel C. Sheps Award in Mathematical Demography presented by the Population Association of America; and in 1994, he received the Distinguished Statistical Ecologist Award at the Sixth International Congress of Ecology.

Author Biography

 Kenneth G. Manton, Ph.D. is Research Professor, Research Director, and Director of the Center for Demographic Studies at Duke University, and Medical Research Professor at Duke University Medical Center's Department of Community and Family Medicine. Dr. Manton is also a Senior Fellow of the Duke University Medical Center's Center for the Study of Aging and Human Development. His research interests include mathematical models of human aging, mortality, and chronic disease. He was the 1990 recipient of the Mindel C. Sheps Award in Mathematical Demography presented by the Population Association of America; and in 1991 he received the Allied-Signal Inc. Achievement Award in Aging administered by the Johns Hopkins Center on Aging.Joel E. Cohen, Ph.D., Dr. P.H., is Professor of Population, and Head of the Laboratory of Populations, Rockefeller University. He also is Professor of Populations at Columbia University. His research interests include the demography, ecology, epidemiology, and social organization of human and non-human populations, and related mathematical concepts. In 1981, he was elected Fellow of the MacArthur and Guggeneheim Foundations. He was the 1992 recipient of the Mindel C. Sheps Award in Mathematical Demography presented by the Population Association of America; and in 1994, he received the Distinguished Statistical Ecologist Award at the Sixth International Congress of Ecology. 

Table of Contents

Overview	1	(16)
Introduction	1	(1)
Asbestos and Health	1	(3)
History of Asbestos	4	(1)
Epidemiological Discovery	5	(1)
Johns-Manville Corporation	6	(1)
Manville Trust	6	(1)
Manville Trust Litigation	7	(2)
Project History	9	(2)
Results	11	(3)
Organization of Monograph	14	(3)
Epidemiology of Asbestos-Related Diseases	17	(44)
Introduction	17	(1)
Design Issues in Studying Occupational Exposure	18	(6)
Measures of Risk	19	(3)
Design Issues	22	(2)
Studies of Health Risks of Occupational Exposures	24	(20)
Health Risks of a Cohort of Insulation Workers Occupationally Exposed to Asbestos	25	(10)
A Case-Control Study of Asbestos Risks in the United States and Canada	35	(2)
Short-Term Amosite Exposure Among Factory Workers in New Jersey	37	(1)
Effects of Chrysotile Exposure Among Miners and Millers in Quebec	38	(2)
Mesothelioma Risks Among World War II Shipyard Workers	40	(2)
Effects of Asbestos Exposure Among a Cohort of Retired Factory Workers	42	(2)
Increases in Disease Risk Associated with Exposure to Asbestos	44	(8)
Effects of Fiber Type on Disease Risks	52	(5)
Simian Virus 40 and Mesothelioma	57	(4)
Forecasts Based on Direct Estimates of Exposure	61	(28)
Introduction	61	(1)
Selikoff's Study: General Description	61	(1)
Data	61	(1)
Model and Methods	62	(1)
Selikoff's Six Tasks	62	(17)
Task 1: Identify the Industries and Occupations Where Asbestos Exposure Took Place	63	(4)
Task 2: Estimate the Number, Timing, and Duration of Employment of Exposed Workers	67	(4)
Task 3: Estimate Risk Differentials Among Occupations and Industries	71	(3)
Task 4: Estimate Dose-Response Models for Cancer Risks	74	(2)
Task 5: Project Future Asbestos-Related Cancer Mortality	76	(1)
Task 6: Estimate and Project Deaths Due to Asbestosis	76	(3)
Sensitivity of Selikoff's Projections	79	(2)
Alternative Projections of Health Implications	81	(8)
Forecasts Based on Indirect Estimates of Exposure	89	(40)
Introduction	89	(1)
Background	89	(4)
Walker's Study: General Description	93	(1)
Data	93	(1)
Model and Methods	94	(1)
Walker's Five Tasks	94	(31)
Task 1: Determine the Effective Number of Past Asbestos Workers	95	(17)
Task 2: Project Mesothelioma Incidence	112	(3)
Task 3: Project Lung Cancer Incidence	115	(4)
Task 4: Estimate Current and Future Asbestosis Prevalence	119	(5)
Task 5: Estimate the Amount of Asbestos-Related Disease Likely to Occur in Women	124	(1)
Asbestos-Related Disease Projections by Other Authors	125	(2)
Conclusions	127	(2)
Uncertainty in Forecasts Based on Indirect Estimates	129	(26)
Introduction	129	(1)
Qualitative Sources of Uncertainty in Walker's Projections	129	(5)
Uncertainties in Either Direction	130	(2)
Why Walker's Projections May Be Too Low	132	(1)
Why Walker's Projections May Be Too High	133	(1)
Sensitivity Analysis of Walker's Projections	134	(9)
Results for Single Parameters	138	(1)
Results for All Variables Jointly	139	(3)
Summary of Uncertainty Results	142	(1)
Further Sensitivity Analysis of Walker's Mesothelioma Projections	143	(9)
Projection Methodology	145	(2)
Alternative Scenarios	147	(2)
Results	149	(3)
Conclusions	152	(3)
Updated Forecasts Based on Indirect Estimates of Exposure	155	(62)
Introduction	155	(1)
Factors Considered	155	(5)
Assumptions	160	(5)
First-Stage Calibration: Overview	165	(4)
Data Preparation	169	(22)
Step 1: Nonmesothelioma Mortality Rates	169	(3)
Step 2: National Estimates of Mesothelioma Incidence Counts	172	(2)
Step 3: Distribution of Age and Date at Start of Asbestos Exposure for Mesothelioma Incidence Among Manville Trust Claimants	174	(15)
Step 4: Normalization of Exposure	189	(1)
Step 5: Intensity of Exposure	190	(1)
Model Estimation	191	(9)
Step 6: Stratification of National Estimates of Mesothelioma Incidence Counts, by Level of Asbestos Exposure	191	(1)
Step 7: Estimation of the IWE Population Exposed to Asbestos Prior to 1975 by Level of Asbestos Exposure	192	(6)
Step 8: Adjustments to Exposure During 1955-1974, by Level of Asbestos Exposure	198	(1)
Step 9: Adjustments to Reflect Improvements in the Workplace During 1960-1974, by Level of Asbestos Exposure	198	(1)
Step 10: Renormalization by Level of Asbestos Exposure	199	(1)
Model Projection	200	(8)
Step 11: Forward Projection of the At-Risk IWE Population by Level of Asbestos Exposure	202	(1)
Step 12: Forward Projection of Mesothelioma Incidence by Level of Asbestos Exposure	202	(6)
Nonparametric Hazard Modeling of Claim Filing Rates: CHR Model	208	(9)
Step 1: Distribution of 1990-1994 Claims by Attained Age, TSFE, and Disease/Injury	208	(1)
Step 2: Estimation of Claim Hazard Rates by Attained Age, TSFE, and Disease/Injury	209	(4)
Step 3: Claim Projections	213	(4)
Uncertainty in Updated Forecasts	217	(34)
Introduction	217	(5)
Analysis S1: Constant Age-Specific Claim Runoff	222	(1)
Analysis S2: Ratio Estimation of Nine Asbestos-Related Diseases -- PTS Model	223	(1)
Analysis S3: Parametric Claim Hazard Rate Model	224	(5)
Analysis S4: Mesothelioma Incidence Function	229	(6)
Sensitivity to the b Parameter	232	(1)
Sensitivity to the k Parameter	233	(2)
Analysis S5: Adjustments to the IWE Exposed Population	235	(1)
Analysis S6: National Mesothelioma Incidence Counts	236	(1)
Analysis S7: Nonmesothelioma Mortality Rates	237	(2)
Analysis S8: Excess Mortality Among Insulation Workers	239	(1)
Analysis S9: Decline in Claim Filing Rates	240	(1)
Overall Sensitivity: Analyses S1-S9	241	(6)
Analysis S10: Manville Trust Calibrations	247	(2)
Conclusions	249	(2)
Forecasts Based on a Hybrid Model	251	(60)
Introduction	251	(1)
Model Overview	252	(3)
First Stage	252	(2)
Second Stage	254	(1)
Data Preparation	255	(18)
Step 1: Nonmesothelioma Mortality Rates	255	(1)
Step 2: Occupation Groups with Significant Asbestos Exposure	256	(1)
Step 3: Distribution of Mesothelioma Claim Counts 1990-1994 by Attained Age at the Time of Claim and TSFE	257	(13)
Step 4: Distribution of Mesothelioma Claim Counts by Age at Start of Exposure and Date of First Exposure	270	(3)
Step 5: Normalization of Exposure	273	(1)
Model Estimation	273	(15)
Step 6: Estimation of the OSHA Model for Mesothelioma	273	(11)
Step 7: Estimation of the Population Exposed to Asbestos Prior to 1975	284	(4)
Model Projection	288	(2)
Step 8: First-Stage Calibration	288	(1)
Step 9: Forward Projection of Mesothelioma Mortality	288	(2)
Second Stage: CHR Forecasting Model	290	(18)
Step 1: Distribution of Disease-Specific Claim Counts for 1990-1994 by Attained Age and TSFE	290	(1)
Step 2: Second-Stage Calibration	290	(4)
Step 3: At-Risk Population Projections	294	(4)
Step 4: Claim Projections	298	(10)
Conclusions	308	(3)
Uncertainty in Forecasts Based on a Hybrid Model	311	(34)
Introduction	311	(2)
Impact of Claim Filing Rules	313	(1)
Baseline Model: SDIS Criterion	314	(1)
Analysis S1: Validated Disease	315	(5)
Analysis S2: Multiple Diseases	320	(5)
Analysis S3: CHR Smoothing	325	(2)
Analysis S4: Exposure Smoothing	327	(1)
Analysis S5: Weibull k Parameter	328	(2)
Analysis S6: Relative Risks of Mesothelioma	330	(2)
Analysis S7: Duration of Exposure	332	(2)
Overall Sensitivity: Analyses S1-S7	334	(2)
Conclusions	336	(9)
Conclusions and Implications	345	(32)
Introduction	345	(2)
Data	347	(3)
Comparisons of Original and Updated Data	350	(4)
Comparisons of Actual and Projected Numbers of Claims	354	(5)
Health and Vital Statistics Data, 1990-1999	359	(15)
Conclusions	374	(3)
References	377	(12)
Index	389