After the delivery of Enterprise Wide Stress Testing (EWST) on August 12, 2015, CCFA Risk Study group held the fifth study session on December 7, 2015. The topic is on Operational Risk. It was an honour to invite two experienced practitioners in the field of operational risk to lead the round table discussion. They are: Baifang Xing, Manager of Risk Methodologies at RBC, and Shunbai Li, Manager of Internal Rating Management at Scotiabank. Both speakers have extensive experience in the operational risk area. Yicent Chen, Model Risk Specialist from BMO was the facilitator. Other practitioners from leading Canadian financial institutions also joined the discussion.

The session covered Advanced Measurement Approach (AMA) methodology specifications, model challenges and limitations from model development perspective. A high-level overview of the operational risk capital model methodology was provided by the speakers. At first the speakers explained the definition of operational risk from Basel Guideline and in practical terms. Then they moved onto explaining the main factors that are associated with operational risk and operational risk capital calculation. In the end, they shared some of the on-going challenges industry faces from regulators and banks perspective.

Operational risk is defined as the risk of loss resulting from inadequate or failed internal process, people and systems or from external events. In practice, it contains all risk types that can’t be classified under market risk, credit risk, or any of the major risk categories. Basel uses two dimensions to define the characteristics of operational risk: event types and business lines. For operational risk capital modelling, a two dimensional table is created by setting event types as column and business lines as row. Each cell of this table contains a unit of measure which is used for loss severity and loss frequency estimation. There are two fundamental assumptions for the Advanced Measurement Approach: (1) it assumes loss events happen with the frequency they happened before, and (2) the magnitude of the loss, while random, is adequately captured by the past. This approach is being developed on the top of basic indicator approach and the standardised approach which are simpler and more straightforward. Modeller uses four different data inputs: internal data, external data, scenario data and Business Environment and Internal Control Factors (BEICF). Even though the model has been studied for more than ten years globally, significant challenges still exist as followings: internal and external data scarcity, capital volatility from low frequency and high severity loss event, truncated data due to reporting bias, inflation adjustment, foreign exchange adjustment, and common root loss event.

Due to the nature of the operational risk definition and the data used, it is a challenge to compare the operational risk Advanced Measurement Approach among different banks. Each bank used different approach and dataset to solve the problem. 

就企业整体压力测试（EWST）研讨后，CCFA于2015年12月7日迎来了第五次风险管理研讨会，课题是操作风险。我们荣幸地邀请到了在这一领域的两位资深从业者领导为我们主导这次的圆桌讨论，他们是加拿大皇家银行（RBC）的风险方法经理Baifang Xing，以及丰业银行（Scotia bank）的内部评级管理经理Shunbai Li。两位演讲者在操作风险领域都有丰富的经验。担任本次研讨会主持的是满地可银行（BMO）的模型风险专家Yicent Chen。参加讨论的还有来自加拿大各大金融机构的专业人士。

这次的讨论包括高级计量法（AMA）方法规范, 以及从模型发展的角度来看，模型的挑战性和局限性。演讲者对操作风险资本模型方法进行了概述。首先，演讲者们讲解了巴塞尔协议中对于操作风险的定义。之后，他们又对操作风险的主要因素和计算进行了讲解。最后，他们以监管机构和银行的角度，分享了一些产业正持续面临的挑战。

操作风险定义为，由于不适当的或失败的内部流程，人为和系统或外部事件造成损失而带来的风险。在实践中，它包含了不能被归类于市场风险，信用风险，或任何重大风险的其它所有风险类型。巴塞尔使用二维来定义操作风险的特征：事件类型和业务种类。操作风险资本模型，是一个将事件类型设置为列，业务种类设置为行的二维表格，该表格的每一个单元格包含一个用于损失额度和损失频率估计的度量单位。关于高级计量法（AMA）有两个基本假设：①假定损失事件及它的发生频率曾经发生过；②损失大小，过去偶尔会捕捉到。这种方法是建立在基本指标法和标准法上的，也是更简单明确的模范方法。分析员使用了四种不同的输入数据：内部数据，外部数据，场景数据以及商业环境与内部控制因素。即使该模型在全球已经被研究了超过十年，但仍然存在重大的挑战，如下：内部和外部数据不足，低频率和高度损失事件引起的资本波动，由于报告偏差引起的数据损坏，通货膨胀的调整，外汇的调整, 以及常见的根本损失事件。

根据操作风险的定义和使用的数据的性质，比较不同银行的操作风险高级计量法（AMA）也是一个挑战。各银行采用不同的方式和数据集来解决这个问题。