For more than two decades, actuaries, econometric modelers, and legal teams have grappled with the challenge of forecasting asbestos liabilities, which stubbornly refuse to taper off. A commonly accepted actuarial method for estimating asbestos claim costs has been to apply a decay pattern based on epidemiological mesothelioma incidence curves, namely, the Nicholson Curve, Stallard Curve, and Price Curve. Adaptations of these epidemiological decay curves have been used as proxies for the gradual decrease in asbestos claim filings (referred to as claim decay) when estimating future asbestos claim liabilities. As many actuaries and modelers have observed, these curves have underestimated the persistence of asbestos claim tail runoff, particularly with regard to liabilities for individual defendants facing legacy asbestos product liability allegations. This article explores what has caused the curves to underestimate asbestos claim decay and will discuss specific considerations for estimating future asbestos liability claim costs.

Background on the Nicholson, Stallard, and Price curves for forecasting asbestos liabilities

The Nicholson Curve (published 1982) traces back to work by William Nicholson and colleagues.¹ They sought to quantify the number of U.S. workers exposed to asbestos across key industries from the 1940s through 1970s and predict the resulting asbestos-related disease incidence, especially mesothelioma. The study utilized disease incidence data from the Manville Trust (composed of asbestos injuries arising from Johns Manville historical operations) and U.S. Bureau of Labor Statistics employment data. The estimated decay curve suggested a peak in asbestos-attributable cancer cases in the early 2000s, followed by a sharp decline ending in 2030.

The Stallard Curve (originally published 2001; updated in 2004), developed by Eric Stallard and collaborators, also utilized the Manville Trust asbestos claims data to refine Nicholson’s forecasting methodology by translating the disease incidence projection into estimated claim counts by incorporating exposure-specific cohort details, such as latency and worker/exposed cohort age.² This framework was designed with the intent to better capture the varying nature of asbestos exposure by industry. The Stallard Curve projected a later peak of asbestos-related personal injury claim filings between 2010 and 2015, with an ultimate end of filings by approximately 2050.

The Price Curve (originally published 2004; updated in 2009), created by Dr. Bertram Price, was differentiated from the previous curves by utilizing mesothelioma incidence data from the National Cancer Institute’s Surveillance, Epidemiology, and End Results Program (SEER).^3,4 Using the SEER data, Dr. Price observed that the female population displayed a relatively constant incidence of mesothelioma even though the female population was not exposed to the same degree of prolonged occupational asbestos exposure. This finding, along with epidemiological literature investigating the causes of mesothelioma, warranted a separation of mesothelioma cased into those attributable to occupational asbestos exposure and others classified as “background cases” (i.e., cases presumed to arise from low-level, nonoccupational, or unknown source of exposure). The Price Curve produced an estimate of asbestos-linked mesothelioma incidence as a proportion of total U.S. mesothelioma incidence. Later commentary (including a 2022 update) reiterates that mesothelioma has multiple risk factors and a persistent background rate. Dr. Price’s analysis concluded that, “around 2040 virtually all mesothelioma cases in the US will be background cases.”⁵

Decay curve application to asbestos liabilities. The mesothelioma decay curve methodologies arguably provided valuable countrywide population-level projections, but the curves projected an earlier peak and shorter duration compared to actual asbestos claim filings. The design of these curve studies, built upon disease incidence across the broad population, does not account for discrete economic behaviors influencing the timing and quantity of claim filings as discussed in the following section. These behaviors introduce greater potential for model bias.

Challenges for curve methodologies

Litigation drivers – Plaintiff firms

Asbestos lawsuit filing and settlement patterns are materially influenced by law firm capacity, advertising effort, and the U.S. legal system’s litigation structures. While nationwide mesothelioma diagnoses have begun to exhibit downward trend in recent years, the number of asbestos claim filings has persisted at a level above what modelers anticipated.⁶

Plaintiff firms have devoted considerable time and resources to developing their asbestos litigation capacity and advertising strategies, both of which have been aided by the rise of third-party litigation funding. The plaintiff firms have developed expertise in cultivating caseloads and producing settlements. This is an example of the commonly observed economic phenomenon in which mature markets in decline tend to dry up slowly due to the accumulated resources, expertise, and infrastructure devoted to the revenue source at hand. Firms operating in mature markets have incentive to continue operations and maximize the value of the market development costs. Often, established firms possess efficiencies that allow them to operate profitably at thinner margins. As a result, litigation arising from asbestos product liability will likely continue as long as it remains profitable for law firms to pursue.

Litigation drivers – Jurisdiction

In addition to plaintiff firm expertise, favorable legal jurisdictions also contribute to the mismatch between actual versus projected asbestos claim rates. Those familiar with asbestos claim filing venues are aware of the concentration in Madison and St. Clair counties, both located in southeastern Illinois, which for years have held the top rank for U.S. asbestos lawsuit filings. Other high-volume jurisdictions include New York City, Philadelphia, and Cook County, Illinois. The jurisdictions that are viewed as plaintiff-friendly court venues also exhibit a concentration of plaintiff firms.

Characteristics associated with favorable jurisdictions include permissive forum-non-conveniens (e.g., no local connection required), centralized or high-speed dockets, streamlined procedural rules, and advantageous jury-pool attributes. These venue features contribute to a materially nonhomogeneous geographic distribution of filing activity, which complicates the application of incidence-curve methodologies since these curves do not explicitly account for venue favorability or jurisdiction-specific caseload concentrations. The evolving jurisdictional mix has, in turn, influenced settlement values and the timing of claim resolution, introducing additional complexity into modeling efforts.

Attribution challenge. Often, lung diseases (including certain cancer types) are inherently difficult to attribute to singular cause. In these cases, claimants often seek compensation by filing a product liability claim, asserting that asbestos contributed materially to their condition. Moreover, there is a relatively narrow group of defendants known for their role in the manufacture and retail of asbestos-containing products. The existence of broadly accepted causal epidemiological evidence coupled with a well-defined defendant group has resulted in a litigation strategy that can be reliably reproduced. To further complicate matters, asbestos-related diseases have long latency periods, meaning that information used to attribute the source of the exposure becomes difficult to produce and verify. This often makes it difficult for defendants to dispute the allegations and for courts to determine each defendant’s proportional share of liability.

Calibration source data. A key limitation of traditional epidemiological curve methodologies lies in the nature of their calibration data. The Nicholson, Stallard, and Price curves were all parametrized using employment data as an exposure base, that is, individuals potentially exposed to asbestos while employed. The projected asbestos claims, therefore, are from a blend of many defendants (some declining, some still highly active). This calibration exposure base will not exactly match the decay pattern of claims from a single corporate entity.

Nationwide incidence data ignores defendant-specific dynamics that shape persistence and claim cost. A few examples of this are the nature and time period of defendant’s exposure, claimant demographics such as age, and jurisdictional concentrations. A defendant repeatedly receiving claims from individuals allegedly exposed while manufacturing asbestos in the 1960s and 1970s will likely see cases subside sooner than a defendant with a material share of claimants alleging exposure during maintenance of asbestos-containing machinery in more-recent decades.

Incidence data is also a combination of occupational, secondary, and product exposures, making for less homogeneity between modeling assignments. For example, a corporation that manufactured asbestos insulation will experience a far different claim trajectory than one whose products or operations created a peripheral (incidental/limited) exposure to asbestos.

Moreover, in recent years, talcum powder exposure has emerged as a source of a significant number of claim filings with similar disease allegations and a growing share of claimants alleging secondary exposure (i.e., indirectly exposed family members). These developing factors were not observable at the time of epidemiological studies and were not influential in the resulting decay curves with the exception of the Price study, which attempted to reflect background disease incidence (albeit not talc specific).

Correlation of assumption. Economic incentives and behavioral factors tend to shift together, often in ways that are difficult to quantify in a reserving context. A model that assumes independence or stable historical correlations can break down and can become vulnerable to model error, that is, the gap between model estimates based on statistical assumptions and actual financial results. Actual market dynamics and intertwined behavioral elements, particularly within the legal landscape, can amplify volatility. Such shifts alter not only singular assumptions but render historical relationships less predictive of the future. It may be beneficial to incorporate stress or sensitivity tests to gain a better representation of uncertainty.

Concluding thoughts on forecasting asbestos liabilities

For these reasons, effective asbestos claim modeling requires consideration of defendant-specific exposure characteristics and an analysis of historical claim experience. Epidemiological curves can provide an initial framework for modeling, but the accuracy of “off-the-shelf” application is hindered by the complexity of real-world phenomena. For practitioners estimating future asbestos claim cost, this necessitates that decay curve methodologies be supplemented with market- and company-specific knowledge, awareness of risk factors, and recognition of blind spots in analysis methodologies.

¹ Nicholson, W.J., Perkel, G., & Selikoff, I.J. (1982). Occupational exposure to asbestos: population at risk and projected mortality--1980-2030. American Journal of Industrial Medicine, 3(3), 259–311. Retrieved December 17, 2025, from https://onlinelibrary.wiley.com/doi/10.1002/ajim.4700030305.

² Stallard, E. (2001). Product liability forecasting for asbestos-related personal injury claims: A multidisciplinary approach. Annals of the New York Academy of Sciences, 954, 223–244. https://doi.org/10.1111/j.1749-6632.2001.tb02754.x.

³ Price, B. (1997). Analysis of current trends in United States mesothelioma incidence. American Journal of Epidemiology, 145(3), 211–218. Retrieved December 17, 2025, from https://academic.oup.com/aje/article-abstract/145/3/211/182582?redirectedFrom=fulltext.

⁴ National Cancer Institute’s Surveillance, Epidemiology, and End Results Program. (April 2003). SEER 1973–2000 public use data (November 2002 submission) [Data set]. National Institutes of Health. Available from https://seer.cancer.gov/data/.

⁵ Price, B. (2022). Projection of future numbers of mesothelioma cases in the US and the increasing prevalence of background cases: An update based on SEER data for 1975 through 2018. Critical Reviews in Toxicology, 52(4), 317–324. https://doi.org/10.1080/10408444.2022.2082919.

⁶ U.S. Cancer Statistics Working Group. (June 2025). U.S. cancer statistics data visualizations tool. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, & National Cancer Institute. Retrieved December 15, 2025, from https://www.cdc.gov/cancer/dataviz.