The geographic breadth of disasters that struck in 2011 was instructive to insurers and modeling firms, said S. Ming Lee, president and CEO of Boston-based AIR Worldwide Corp.
“It's becoming clear that there are large losses that happen outside of peak zones,” Mr. Lee said during the Monte Carlo, Monaco, meeting.
The complex chain of events that unfolded in the wake of flooding in Thailand floods and the earthquake and tsunami that struck Japan spurred his firm to develop models that will help reinsurers quantify risk surrounding supply chains and business interruption, he said.
William M. Keogh, president of Oakland, Calif.-based Eqecat Inc., noted that many of the most severe losses borne by reinsurers in the past several years have resulted from unmodeled risk.
“The Thai floods were not on a lot of people's radar,” Mr. Keogh said. “Some people presume that if there isn't a model, there isn't a risk. Nothing could be further from the truth.”
Hemant Shah, president and CEO of Newark, Calif.-based Risk Management Solutions Inc., said one lasting result from the past several years is that reinsurers now expect more transparency from modeling firms.
In addition to providing clients with more disclosure and descriptions of the data-based assumptions made in the models, RMS is providing more visibility into its research and development roadmaps.
“Clients can now take control of more of the assumptions of themselves and account for them in their risk management practices,” Mr. Shah said.
“The demand for greater transparency is happening throughout the insurance food chain,” Mr. Lee added.
Jayanta Guin, a senior vice president at AIR Worldwide, noted that under the forthcoming Solvency II requirement in the European Union, risk managers at insurers and reinsurers will need to “own” the risks in their operations.
“It's not enough anymore to just say that you are using a model and treat it as a black box,” Mr. Guin said. “So we are encouraging our clients to question the model, and dig in to the details and understand the uncertainty in the models.”
One result of the trend toward ownership of risk by insurance and reinsurance companies is that they are running catastrophe and other risk models with increased frequency, Mr. Lee said.
“It used to be the models were run annually or quarterly,” he said. “Now, they are run monthly or weekly as the role of analytics takes on much more importance inside companies.”
Robert Muir-Wood, chief research officer at RMS, said this speeding up of the risk modeling processes may have profound implications on the operational aspects of reinsurance, much as high-frequency trading has altered the stock markets. Currently, risk managers may base decisions on information that is many months old, whereas real-time risk assessment may someday enable insurers and reinsurers to manage their risks much more nimbly and granularly.
“The buildup to a renewals period currently happens in slow motion, partially because people can't get a handle at looking at all their risks in real time,” Mr. Muir-Wood said. “Simply speeding things up is really going to be transformative for the reinsurance market.”
Mr. Shah said that his firm is positioning itself to furnish clients not just with models, but also with an overarching risk management technology framework capable of handling big data.
“The era of insurers and reinsurers building ad hoc tools to run ad hoc processes to run models is quickly coming to an end,” Mr. Shah said. “Insurers and reinsurers are looking to systemically embed analytics into their workflows. This requires a redo of system architectures and an ability to operate platforms at scale.”
The elasticity and scalability inherent in cloud computing is well-suited for the task, Mr. Shah added.
Mr. Lee added that AIR offers a hosted version of its models, as well as versions that can run in clients' internal clouds as well as third-party clouds.
Mr. Keogh said considerations of how insurers and reinsurers would “operationalize” the models guided development of Eqecat's new core platform, risk quantification and engineering. In addition to updating 77 of the company's 180 models to coincide with the platform, the company's also focused on data integration issues.
“Usability and integration are now the focus,” he said. “It's almost assumed at this point that you have a good (modeling) methodology.”