•

Non-life Run-off - The claim frequency information for our general casualty and workers’ compensation categories includes direct and assumed open and closed claims by accident year at the claimant level. Reported claims that are closed without a payment are included within our cumulative number of reported claims because we typically incur claim adjustment expenses on them prior to their closure. The claim count numbers exclude counts related to claims within policy deductibles where the insured is responsible for the payment of losses within the deductible layer. Individual claim counts related to certain assumed reinsurance contracts such as excess-of-loss and quota share treaties are not available to us, and the losses arising from these treaties have been treated as single claims for the purposes of determining claim counts. Therefore, each treaty year within the reinsurance contract is deemed a single claim because the detailed underlying individual claim information is generally not reported by our reinsureds.

---

•

StarStone and Atrium - The claim frequency information is determined at the claimant level. Our claims system assigns a unique claim identifier to each reported claim we receive. Each unique claim identifier is deemed to be a single claim, irrespective of whether the claim remains open or has been closed with or without payment. For certain insurance facilities and business produced or managed by managing general agents, coverholders and third party administrators where the underlying claims data is reported to us in an aggregated format, the information necessary to provide cumulative claims frequency is not available. In such cases, we typically record a “block” claim in our system. This also applies to a small amount of assumed reinsurance business that we write where, similarly, the underlying claims data is reported to us in an aggregated format. In such instances, each assumed reinsurance contract is deemed a single claim.

•	Claim counts are presented only on a reported and not on an ultimate basis;

•	Reported claim counts have not been adjusted for ceded reinsurance, which may distort any measures of frequency or severity;

•	For classes of business that have a mix of primary and excess layer exposures, such as our general casualty and workers’ compensation classes of business within our Non-life Run-Off segment, the reported claim counts may fluctuate from period to period between exposure layers, thereby distorting any measure of frequency and severity; and

•	The use of our reported claim frequency information to project ultimate loss payouts by disaggregated disclosure category may not be as meaningful as claim count information related to individual products at a lower level.

---

1.

Retroactive reinsurance cession - Page 185 of our 2016 Form 10-K describes a reinsurance transaction which was recorded in accordance with retroactive reinsurance accounting. StarStone ceded net incurred losses and LAE of $75.7 million under this transaction during the year ended December 31, 2016, comprising $66.1 million of unpaid liability for losses and LAE and paid losses of $9.6 million, which were subject to retroactive reinsurance accounting. The closing ceded unpaid liability for losses and LAE of $66.1 million was disclosed separately as ceded business within the loss reserves roll forward table on page 166 of our Form 10-K. It was therefore not reflected within the current period net incurred losses and LAE total of $415.8 million disclosed within the same table. However, the total retroactive ceded net incurred losses and allocated LAE are reflected in the loss development tables on page 167 of our Form 10-K.

2.

IBNR allocation methodology - In the loss reserves roll forward table on page 166 of our Form 10-K, we utilized an earned premium ratio approach to allocate IBNR between current and prior periods. In the loss development table on page 167, we used the case reserves to IBNR ratio approach, which is a generally accepted methodology, to consistently allocate IBNR by accident year. We acquired StarStone on April 1, 2014 and historical pre-acquisition claims data by accident year, on a gross and ceded basis, was not available. Due to the lack of historical pre-acquisition claims data, we used the case reserves to IBNR ratio approach to consistently allocate IBNR by accident year across the entire StarStone loss development table. The impact of using the two different approaches to allocate IBNR resulted in a difference of $44.4 million in the presentation of the current accident year.

3.

Effect of foreign exchange rate movement - StarStone’s major currencies are the US dollar ("US$"), which is also our functional currency, the Euro (€), the Canadian dollar ("CAD") and the British pound ("GBP"). Foreign currency transactions included in net incurred losses in our statement of earnings and in the loss reserves roll-forward table are recorded at the prevailing average exchange rate at the end of each reporting period. In the loss development tables, however, we have revalued all historical claims data using exchange rates as at December 31, 2016 in order to mitigate the impact of foreign exchange on the claims development disclosed within the loss triangles. This difference was largely attributable to the volatility in the foreign exchange markets following the June 2016 Brexit vote in the United Kingdom, and primarily impacted our GBP exposures.

4.

Unallocated loss and loss adjustment expenses ("ULAE") - The ULAE provision represents our estimate of future costs to administer reported claims. Movements in this provision are included within the loss reserves roll forward table on page 166 in our Form 10-K. The ULAE provision is not recorded on an accident year basis, and accordingly, the total amount of the provision is excluded from the StarStone loss development table on page 167 and is disclosed separately as a reconciling item on page 168 of our Form 10-K. The ULAE provision of $16.8 million as at December 31, 2016 was comprised of $7.2 million and $9.6 million related to the current and prior periods, respectively.

---

a.	The guidance provided by the Financial Accounting Standards Board in ASC 944-40-55-9C indicates that a reportable segment can be considered an appropriate category for use to either aggregate or disaggregate the required disclosures;

b.

The StarStone segment represents 18% of our total liability for losses and LAE as at December 31, 2016, of which the largest individual class of business was General Casualty, representing 5% of our total exposures. The remaining classes of business were not deemed individually significant for the disclosures required by ASC 944-40-50-4B through 50-4G to be separately provided in our 2016 Form 10-K. In our view, an aggregate loss development table for the StarStone segment presents more meaningful information to a user of our financial statements, rather than a single category, such as General Casualty, which represents less than a third of the StarStone segment’s liability for losses and LAE; and

c.

We performed a review of the duration of the components of the liability for losses and LAE within the StarStone segment, by each of the categories shown on page 91 of our Form 10-K. This review indicated that the average duration across the categories was relatively consistent. We determined that the individual categories did not exhibit significantly different characteristics in terms of the timing or the uncertainty of the related cash flows to warrant further disaggregation below the segment level.