Capital Gains

The 2005 hurricanes not only led to the reconstruction of people’s lives and property, but they also led to the reconstruction of capital models used to test reinsurers’ financial capacity. Ratings agencies are not just relying on capital models in isolation; they are also looking at the context in which these models work. That’s why almost all of the ratings agencies, in addition to changing their capital models, are in some fashion moving towards incorporating catastrophic risk and ERM – a holistic approach to analyzing, quantifying and forecasting financial and corporate viability and vulnerability – into the ratings process. The prospect should serve as a heads up to all industry players, who’ve already had to make adjustments to keep pace with methodology changes.

CAPITAL MODELS: STATIC V. STOCHASTIC

Over the past year, ratings agencies have been evolving their capital model methodologies individually. Standard & Poor’s (S&P’s) is widely cited as a “static” capital model, which is basically described by some as a “narrow, factors-based approach.” According to reinsurance intermediary Guy Carpenter, the S&P’s model looks at factors such as asset risk, reinsurance recoverables, pricing risk and reserve volatility in coming up with capital adequacy projections. While its factor-based analysis remains the same, S&P’s has announced that it will be revising the factors as a result of events last year. Changes to its model will be finalized in 2006 and versions of the new model will be run in 2007.

In contrast to S&P’s static model, Fitch is leaning to a dynamic or ‘stochastic’ approach. With much fanfare, Fitch recently announced its launch of “the world’s first global stochastic model,” called Prism. The new model, inspired by Solvency II, uses a variety of variables to assess capital adequacy – including insurers’ in-house capital models and the regions in which they are based. The model stresses probabilities and assesses the likelihood of outcomes in a way static models do not.

According to the Casualty Actuarial Society, a dynamic, stochastic model is used to describe “the evolution of claims, premiums, reserves and expenses for [a particular] line of business over time.” It focuses attention on such things as the estimation of claims severity, frequencies, the variation of risk propensity due to industry trends and cycles. It then gauges the effect of these dynamics and trends on the three major kinds of insurance arrangements — quota share reinsurence, surplus reinsurance and stop-loss reinsurance.

Fitch says its model will “define available capital, calibrate risk thresholds and assess enterprise risk regionally by the third quarter of this year with full implementation of the model through a Beta version by early 2007.” At this time, Fitch will introduce a ‘cure period’ for companies to amend their reserves in line with the requirements of the new model. Fitch’s model also factors in regulatory requirements by different regions.

Fitch noted Munich Re and Hanover Re have also turned to the use of “sophisticated in-house models” such as stochastic models. “A number of multinationals now have in-house stochastic models which can feed directly into our new model,” Fitch said in a public statement.

Which is better for reinsurers? Static or stochastic models? There is a grey area in this debate, since there are hybrids. Even when a rating agency uses a static, factor-based model, for example, some of the capital charges (such as catastrophe risk) are based on inputs determined stochastically. Similarly, some of the stochastic models use static factors for some of the less important risks.

“There are pros and cons for each type of model,” according to Michele Fleckenstein, the managing director of Guy Carpenter. “A static model is easier to understand and explain than a more complicated stochastic model.

“With static models, stress tests and sensitivity analysis are used to understand the impact of changing certain assumptions and to gain comfort with the level of required capital.

“Conversely, stochastic models generate the statistical probability of achieving a certain capital level and consider the impact of correlations.

“There are many assumptions that enter stochastic models, making it difficult for rating agencies to tailor these models to individual insurers and for insurers to anticipate the impact of managements’ actions on required capital.”

When all is said and done, reinsurance intermediary Guy Carpenter noted in a September 2006 report, Rating Agency Update: Stepping Up to New Criteria, Fitch is predicting insurers will see a 40-65% increase in capital specifically needed to support catastrophe risks.

In the meantime, Moody’s is also revising its catastrophic risk requirements and A.M. Best continues to adjust its Best Capital Adequacy Ratio (BCAR), which was overhauled in the Canadian market last year.

According to Guy Carpenter, “the rating agency methodology changes, coupled with model changes, have had a significant impact of the amount of capital and/or reinsurance protection needed to achieve a given rating. Given this pressure on capital, some companies have reduced exposure, some bought more reinsurance and others accessed non-traditional capital sources such as catastrophic bonds or sidecars.”

Fleckenstein says that although ratings firms aren’t requiring companies to implement ERM or their own capital and catastrophic models as criteria, they are urging such measures as best practices. Still, there remains an air of uncertainty as to what types of models work – and what don’t.

Fleckenstein points to the “compounding effect” that the new capital models, combined with the updated CAT models, have on bottom line numbers. “Loss estimates could be up 10%, or well over 100%,” she says. “The rating agencies say: ‘We’re going to take that bigger number and give you a heavier capital charge.’ That’s the compounding effect.”

Guy Carpenter’s underwriter clients are particularly cautious about the ratings agencies’ changes pertaining to loss estimates. “We have a concern, our clients have a concern. . .that loss estimates changed so much and then the rating agencies are also increasing the associated capital charge,” Fleckenstein says.

But she adds, “No one really know the exact right way to do all this. No one knows if the storm frequency will continue to increase the way that the models are potentially assuming. So the way the ratings agencies were looking at it before could be right or the way they are looking at it now could be right. Time will tell and that’s the concern of the industry.”

David Wilmot, senior vice president and chief agent for Toa Re Canada, says the industry should have been expecting major changes by the ratings agencies to ensure the sustained viability of the industry.

“Of course the bar is being raised,” he says. “The ratings agencies are more concerned with catastrophe exposure and the catastrophic models that are being used. They weight the catastrophe preparedness more heavily than in the past. That’s not surprising…

“There’s always going to be an upward pressure on the ratings agencies to be sure that they’ve captured everything. We are a much more sophisticated industry than we were 10 years ago.”

In fact, ratings agencies are increasingly assessing capital adequacy in the context of ERM.

ERM incorporates a means of assessing a company’s culture, its corporate governance structure, decision-making processes and overall philosophy towards risk in order to deliver context to basic capital models and traditional quantitative analysis. It is primarily achieved through dialogue, troubleshooting, communications and innovation.

All of this might sound touchy-feely, but ratings agencies are incorporating ERM into their capital assessments, and individual companies are beginning to practice or implement customized versions in-house.

According to Prakash Shimpi, ERM practice leader at
Towers Perrin: “ERM is not just about numbers. It’s about building a risk awareness culture within the organization.”

Towers Perrin, a global consultant that includes reinsurance intermediary services, notes that the ratings agencies are tuning in to ERM. Both S&P’s and A.M. Best have recently introduced analysis of insurer’s ERM capabilities when assessing their ratings. Moody’s Investor Services analyzes a firm’s capabilities in risk management in four categories, while Fitch Ratings announced in June it will be looking for companies to make use of information generated by its capital model within an ERM framework.

So what’s the impetus behind the surging recognition of ERM?

In its summer UPDATE bulletin, co-authored by Shimpi, Towers Perrin says: “ERM can be particularly valuable for rating agencies because it considers more than just the quantitative building blocks that contribute to healthy insurance company management. It takes into account qualitative issues such as governance, systems and process, risk analysis and, significantly, the culture of the company.”

“In short,” the bulletin continues, “ERM facilitates a rating agency’s fundamental understanding of the nature of risk an insurer is taking on, the price it is charging to do so and its ability to measure and monitor it.”

ERM AND RATINGS

S&P’s first introduced a more formal measure of ERM in October 2005. At that time, the ratings agency said it would be assessing insurers’ ERM capabilities when assigning its ratings, including a company’s risk management culture, risk controls, strategic risk management, emerging risk management, and its risk and capital models. To date, S&P’s has evaluated the ERM systems in place in at least 125 companies.

“There is no national regulatory agency or industry-wide trade group that has established ERM criteria,” the company said in its announcement. So, “S&P’s is filling the void.”

A.M. Best announced in February that ERM would become a part of its analysis of financial capital adequacy. At the time it noted that although modelling tools such as dynamic financial analysis (DFA) have “been around the insurance industry for more than 10 years…until recently, they’ve not generated broad interest.”

The perceived problems with DFA were multiple, A.M. Best said. “Several stumbling blocks have been the amount of data required to effectively run the model; a fragmented or silo approach to risk management; difficulty in addressing the correlation of risk; and questions over what to do with the model output”, which tempered widespread interest in ERM, the company noted in its announcement.

“However, more recently, interest in the topic of risk management and the use of more sophisticated capital modeling tools have been on the rise. Companies increasingly are asking the questions: What is our aggregate risk profile, what are our most significant individual risks, what are we doing to manage and/or mitigate these risks, and are we being compensated for the risks we’re taking?”

Moody’s Investor Service, meanwhile, includes a top-level business risk assessment to review a company’s risk management capabilities under the following four categories:

* Risk governance

* Risk management

* Risk analysis and quantification

* Risk infrastructure and intelligence.

The agency is also emphasizing a focus on a company’s integrated and strategic view of risks, the timeliness of risk information and policies, and a look at the practical ways risk measurement tools and procedures are implemented.

Fitch includes ERM as a theme component in Prism, its simulation-based economic capital model. For example, Prism delivers an assessment of capital adequacy, along with a basis to evaluate an insurer’s ERM.

For now anyway, none of the ratings agencies intend to weight ERM criteria heavily – if at all – in delivering their final marks. But, the industry could be moving in that direction.

That’s actually good news for the industry, suggests Shimpi. “It is refreshing because it shows they (ratings agencies) do not rely on the numbers as the be all and end all. ERM is not just about numbers. It’s about building a risk aware culture within the organization.”

AN ERM CASE STUDY

The Allstate Corporation was among the first companies to climb onto the ERM bandwagon. Initially launched in 2000, Allstate’s ERM framework set out to develop a system that would combine the benefits of the numbers produced by a stochastic economic capital model with its internal, operational decision-making processes. The premise of combining the two systems was to assess the consequences that one had on the other. At the time it was a novel undertaking.

Larry Moews, Allstate’s chief risk officer, describes the ERM exercise and implementation as a fundamental “change in culture.”

“We launched enterprise risk management really not because the rating agencies were looking at or because it was the latest fad or thing to do,” Moews says. “We did it because it’s good business practice.”

The process of introducing ERM at the multi-faceted, multinational Allstate is described as complex and exhaustive. Moews credits senior management’s innovation and enthusiasm for helping Allstate achieve its new course.

“We have a great team,” he says. “When we started this, there wasn’t any playbook that spelled out: ‘Here’s Step 1, here’s Step 2 and Step 3.'”

The process at Allstate began with the finance department’s development of the stochastic economic capital (EC) model to evaluate the amount of capital the company needed to cover line-of-business and enterprise losses at specific risk levels.

But the results and benefits of the stochastic model were not well understood outside of the finance department – until the ERM framework was established.

That took some work. With the help of Tillinghast consultants, the company identified and prioritized a comprehensive list of all risks across the organization. Allstate’s management then determined its overall risk tolerance, and the likelihood and relative impact of each risk in relation to the overall tolerance.

Risk limits were set. Then an Enterprise Risk Council (ERC) was struck – including the president, CFO, COO, chief investment officer and other division heads from throughout the company’s key operations – to determine the best ways for managing the risks and reviewing any evolving circumstances regularly.

“It really gave us some framework to be able to ask very hard questions,” Moews recalls. “That just leads to great decisions, because we use all these models we have… [Today,] we feel we understand our business better and it gives us better insight, which leads to better decisions, which leads to better earnings and growth and a better stock price.”