Estimating a single risk capital metric that summarizes all risks and explains the risk profile of a financial conglomerate is a critical process of risk management. Particularly in this context, banks find drawing a conclusive roadmap an arduous exercise as there is no standard regulatory guidance, the way it exists for calculation of regulatory capital and it is a difficult exercise to benchmark with peers due to disparities in existing or planned risk profile. For instance while an overwhelming majority of banks cite integrated stochastic scenarios as primary method for risk aggregation within each risk type, this approach has limited appeal when integrating risk across an enterprise. Another source of complexity arises due to mapping of risk with different products/business units originating risk. There are products and business units in which the risks are primarily concentrated in a single risk category. And sometimes there are situations where multiple risk categories are present at product and business unit levels. The challenge is to combine risk across all of the products and business units in an organization. Therefore, decision making in the area of risk aggregation presents the CRO (Chief Risk Officer) with other host of issues and challenges.
This paper is aimed at identifying most of such issues, explain their cause and suggest options that banks can make use for implementing an effective risk aggregation strategy.
For purpose of this discourse, at a high level, we classify the issues into two main categories viz. Qualitative and Quantitative. Issues on the Qualitative (non-modeling) side are linked to understanding of risk characteristics and the resultant risk policy related decision making. Issues that are addressed in this paper are - risk measures and coherence, different notions of capital, the level and direction of risk aggregation - across risk types or business units or all-in-one matrix, calculation of diversification benefits at each level right up to the group level, allocation of diversification benefits to business units, choice of measure for risk aggregation and capital allocation, and harmonization of time and confidence interval in aggregating risk at a group level.
Quantitative issues (modeling related) are about exercising the choice of an appropriate risk aggregation technique available from a standard toolkit that resolves issues like dependence modeling and provides solution using any of the standard approach viz. variance-covariance approach, multifactor approach, copula simulation, etc.