APPENDIX I: SAMPLE PORTFOLIO INPUTS
APPENDIX 2: SAMPLE PORTFOLIO OUTPUTS
APPENDIX 3: PORTFOLIO ANALYSIS AND REPORTING ISSUES
Risk data is continuously getting richer and new models, running at the bank-wide scale, produce new measures of risk. Bringing these measures to life necessitates appropriate tools for decision makers. Decision makers extend from traders, sales and account officers dealing with clients, to risk managers, plus top management who cannot rely only on aggregated measures.
Risk data extends from observable inputs, such as market prices, to all various credit risk metrics, disaggregated at sub-portfolio and facility levels. Risk data warehouses are critical for organizing the data-gathering process and building up risk data, such as inputs required by the Basel 2 Accord. Moreover, VaR measures require all data for producing market risk and credit risk reports. Bringing the information to life is, in addition to being a bank-wide management challenge, an information technology challenge.
Information technology (IT) plays a key role in banks. Due to the scale of operations of systems of banks, creating the required risk data warehouse with the inputs and outputs of models, and providing links to front ends and reporting modules for decision makers are major challenges. It requires modern tools capable of on-line queries and analyses embedded in front ends and reporting. "Online analysis and processing" (OLAP) systems are presumably tools capable to forward relevant information to end-users whenever they need it. We review some of the reporting and management issues that they need to address.
Traceability of Aggregated Measures and Risk Management
Aggregated measures have a limited usage if it is not possible to trace back the inputs. VaR is not a substitute to all others inputs even though it provides a synthesis of multiple measures. The drawback is that it embeds synthetically underlying the sources of risks. Synthetic measures are convenient but not intuitive anymore and generate a "black-box" effect. "Why is risk high for a particular segment?" is not a simple question for modeled loss volatility or capital. How is the end-user going to disentangle such interdependencies?
Moreover, without access to underlying sources of risks, it is not feasible to control VaR, and VaR remains a passive, although useful, report. VaR type measures of risk require a two-way implementation: from underlying parameters to VaR, and from VaR to the different sources of risk that they represent.
