Exhibit 99.3

Natural Catastrophe Risk Management

We insure and reinsure exposures throughout the world against various natural catastrophe perils.  We manage our exposure to these perils using a combination of methods, including underwriting judgment, CATM (our proprietary risk management system), third party vendor models and third party protection such as purchases of outwards reinsurance and derivative instruments.

Our multi-tiered risk management approach focuses on tracking exposed contract limits, estimating the potential impact of a single natural catastrophe event and simulating our yearly net operating result to reflect an aggregation of modeled underwriting, investment and other risks.  Our Board of Directors regularly reviews the outputs from this process, and we routinely seek to refine and improve our risk management process.

The following discussion should be read in conjunction with the text contained herein entitled  “Our stated catastrophe and enterprise-wide risk management exposures are based on estimates and judgments which are subject to significant uncertainties.”

Exposure Management

We monitor our net reinsurance treaty contract limits that we believe are exposed to a single natural catastrophe  occurrence within certain broadly defined major catastrophe zones.  We provide these limits as a measure of our relative potential loss exposure across major zones in the event a natural catastrophe occurs.

Net Reinsurance Treaty Limits by Zone (1)

(1) For purposes of this presentation, “Mid-Atlantic” includes Georgia, South Carolina, North Carolina, Virginia, Maryland, Delaware and New Jersey; “Northeast” includes New York, Connecticut, Rhode Island, Massachusetts, New Hampshire and Maine; “Gulf” includes Texas, Louisiana, Mississippi and Alabama; “New Madrid” includes Missouri, Tennessee, Arkansas, Illinois, Kentucky, Indiana, Ohio and Michigan; “Northwest” includes Washington and Oregon; “Western European” includes France, Belgium, Netherlands, Luxembourg, Germany, Switzerland and Austria; and “Scandinavia” includes Denmark, Norway and Sweden.

The treaty limits presented are shown net of any outward reinsurance or other third party protection we purchase but have not been reduced by any expected reinstatement premiums. The treaty limits include all business coded as property catastrophe reinsurance (including retrocessional business), property pro rata reinsurance, workers compensation catastrophe reinsurance and event-linked derivative securities, but do not include individual risk business and other reinsurance classes.

For U.S. earthquake, the regional limits shown are for earthquake ground motion damage only, i.e., excluding limits for contracts that do not specifically cover earthquake damage but may provide coverage for fire following an earthquake event. Contracts which provide coverage for multiple regions are included in the totals for each potentially exposed zone, therefore the limits for a single multi-zone policy may be included within several different zone limits.

These treaty limits are a snapshot of our exposure as of January 1, 2012.  As of January 1, 2012, New Madrid earthquake represents our largest concentration of net reinsurance treaty limits among the selected zones. The relative comparison between zones and the absolute level of exposure may change materially at any time due to changes in the composition of our portfolio and changes in our outward reinsurance program.

Single Event Losses

For certain defined natural catastrophe region and peril combinations, we assess the probability and likely magnitude of losses using a combination of industry third-party vendor models, CATM and underwriting judgment. We attempt to model the projected net impact from a single event, taking into account contributions from property catastrophe reinsurance (including retrocessional business), property pro rata reinsurance, workers compensation catastrophe reinsurance, event-linked derivative securities and individual risk business, offset by the net benefit of any reinsurance or derivative protections we purchase and the benefit of reinstatement premiums.

There is no single standard methodology or set of assumptions utilized industry-wide in estimating property catastrophe losses.  As a result, it may be difficult to accurately compare estimates of risk exposure among different insurance and reinsurance companies, due to, among other things, differences in modeling, modeling assumptions, portfolio composition and concentrations, and selected event scenarios.

The table below details the projected net impact from single event losses as of January 1, 2012 for selected zones at selected return period levels using AIR Worldwide Corporation’s CLASIC/2 model version 13.0, one of several industry-recognized third-party vendor models. It is important to note that each catastrophe model contains its own assumptions as to the frequency and severity of loss events, and results may vary significantly from model to model.

As we utilize a combination of third-party models, CATM and underwriting judgement to project the net impact from single event losses, our internal projections may be higher or lower than those presented in the table below.

Net Impact From Single Event Losses by Return Period (in years) (1)

(1) A “100-year” return period can also be referred to as the 1.0% occurrence exceedance probability (“OEP”) meaning there is a 1.0% chance in any given year that this level will be exceeded.  A “250-year” return period can also be referred to as the 0.4% OEP meaning there is a 0.4% chance in any given year that this level will be exceeded.

As of January 1, 2012, our three largest modeled exposures to a single event loss at a 250-year return period were U.S. Hurricane, U.S. Earthquake and European Windstorm.

Our net impact from single event losses may vary considerably within a particular territory depending on the specific characteristics of the event. This is particularly true for the direct insurance and facultative reinsurance portfolio we underwrite. For example, our net impact from a large European windstorm may differ materially depending on whether the majority of loss comes from the U.K. & Ireland or from Continental Europe.

Given the limited availability of reliable historical data, there is a great deal of uncertainty with regard to the accuracy of any catastrophe model, especially when contemplating longer return periods.

Our single event loss estimates represent snapshots as of January 1, 2012.  The composition of our in-force portfolio may change materially at any time due to the acceptance of new policies, the expiration of existing policies, and changes in our outwards reinsurance and derivative protections.

Annual Operating Result

In addition to monitoring treaty contract limits and single event accumulation potential, we attempt to simulate our annual operating result to reflect an aggregation of modeled underwriting, investment and other risks.  This approach estimates a net operating result over simulated twelve month periods, including contributions from certain variables such as aggregate premiums, losses, expenses and investment results.

We view this approach as a supplement to our single event stress test as it allows for multiple losses from both natural catastrophe and other circumstances and attempts to take into account certain risks which are unrelated to our underwriting activities.  Through our modeling, we endeavor to take into account many risks that we face as an enterprise.  However, by the very nature of the insurance and reinsurance business, and due to limitations associated with the use of models in general, our simulated result does not cover every potential risk.

Our Stated Catastrophe and Enterprise-Wide Risk Management Exposures Are Based on Estimates and Judgments Which Are Subject to Significant Uncertainties.

Our approach to risk management, and our estimates of net impact from single event losses such as those provided herein, rely on subjective variables which entail significant uncertainties. For example, in our treaty reinsurance business, the effectiveness of our reinsurance contract zonal limits in managing risk depends largely on the degree to which an actual event is confined to the zone in question and our ability to determine the actual location of the risks insured.  Moreover, in the treaties we write, the definition of a single occurrence may differ from policy to policy and the legal interpretation of a policy’s various terms and conditions following a catastrophic event may be different than we envisioned at its inception.  For these and other reasons, there can be no assurance that our actual net aggregate reinsurance treaty limits by zone, or our net impact from single event loss by return period, will not exceed the Natural Catastrophe Risk Management disclosures provided herein.

In addition, our Natural Catastrophe Risk Management disclosures provided herein involve a substantial number of subjective variables, factors and uncertainties. Small changes in assumptions, which are heavily reliant upon our judgment, can have a significant impact on the modeled outputs.  Further, these disclosures do not take into account numerous real, but non-quantifiable, inputs and risks such as the implications of a loss of our financial strength ratings on our business.  Although we believe that these probabilistic measures provide a meaningful indicator of the relative riskiness of certain events and changes to our business over time, these measures do not predict our actual exposure to, nor guarantee our successful management of, future losses that could have a material adverse effect on our financial condition and results of operations.