Dynamic Reinsurance Analysis

Dynamic Financial Analysis ("DFA") is a process that evaluates the impact of the totality of risk on a company's financial condition. DFA looks at the impact of both macro-economic risks (e.g., changes in interest rates, inflation, foreign exchange rates, the price of oil, etc.) and insurance specific risks (e.g., catastrophes, trend, the underwriting cycle, etc.), as well as the correlation among all of the risks.

Typically, DFA involves modeling the impact of the following on the financial condition of the insurer over a fixed timeframe (typically five years):

• underwriting risk (volatility in losses, rate levels, exposures, mix of business, etc.),


• liquidity risk (market risk of assets),


• credit risk (default risk of assets),


• operational risk, and


• correlation among these risks.

Dynamic Reinsurance Analysis ("DRA"), on the other hand, is a process that concentrates on the effectiveness of the risk management of large losses relative to the company's financial position. As such, DRA should be viewed as a subset of DFA, one that focuses on underwriting risk (driven primarily by large losses) and liquidity risk.

While Holborn's DRA model can include credit and operational risks, in most analyses these risks are treated as deterministic, i.e., various risk management options are evaluated in the context of known interest rates, equity returns, inflation, etc. This approach is taken for a number of reasons:

• underwriting and liquidity risk dominate the overall risk profile of an insurer,


• most insurers invest in high quality assets so default risk is not usually material,


• predicting the future behavior of most macro-economic variables (driving credit risk) is extremely difficult, and


• even if we could predict when and the magnitude of discontinuities in the market (the kinds that produce significant credit risk), the reaction of the insurance and reinsurance market, in terms of capacity and price, is difficult to predict.

Therefore, we prefer to run deterministic scenarios and evaluate the effectiveness of the risk management programs in light of a specific set of economic variables.

Based on our experience, we believe that DRA is an effective tool to:

• evaluate alternative risk management programs, particularly reinsurance:


• efficiency of overall structure,


• appropriateness (e.g., per occurrence and annual aggregate retentions),


• sufficiency (e.g., aggregate limits),


• evaluate volatility of gross and net income,


• evaluate risk in the portfolio,


• evaluate impact of different business strategies, and


• build a case for rating review:


• impact of volatility on rating,


• show rating agencies that company has a handle on the risk in its book.

Holborn's DRA model simulates large losses (catastrophic and non-catastrophic losses), applies each of the risk management options and then calculates the impact on the income, cash flow and balance sheet of the company.

Estimates of catastrophe losses can be developed from any of the commercially available cat models (e.g., RMS, EQE, AIR), proprietary models (e.g., Holborn's inland wind model) or subjectively selected. Large property per-risk and casualty losses are estimated from company or Holborn analyses. "Normal" losses are the residual after large losses are deducted from total losses and are modeled through loss and LAE ratios.

With its DRA program, Holborn can analyze the impact of our client's current reinsurance program against any number of alternatives, including the "naked" alternative, i.e., no reinsurance at all. The program can also accommodate any other form of risk mitigation, including cat bonds, surplus notes or contingent capital.

There are a number of metrics clients could use to determine which reinsurance alternative is "best." Holborn's DRA model includes the following metrics:

• Probability of ruin/insolvency (or some higher threshold set by the company),


• Probability of BCAR dropping below the threshold tied to a targeted Best's rating,


• Probability of surplus dropping below threshold (or targeted growth),


• Reward versus risk (measured by looking at an efficient frontier or Sharpe ratios).

In terms of the efficient frontier, the program assumes that reward is measured by net income. Risk, on the other hand, can be measured by the volatility in net income or in some other key management measure such as surplus at some point in time. Risk could also be measured by the likelihood of a "bad" event, where "bad" is defined by management. Examples of "bad" events include:

• Insolvency,


• Loss of Best's rating (or other rating),


• Loss of surplus of more than $X, or


• Excessive leverage ratios (e.g., NWP/PHS or Reserves/PHS).