Advanced Measurement Approach to Operational Risk

Operational Risk: Definition and Approaches

  • The risk of direct or indirect loss resulting from inadequate or failed internal processes, people and systems or from external events
    • strategic and reputational risk is not included
  • Alternative approaches
    • Basic Indicator Approach (BIA)
    • Standardized Approach (SA)
    • Advanced Measurement Approach (AMA)
  • All approaches target for calibration of capital requirement for (next) 12 months time horizon, which is compatible with the treatment of credit risk

Incentives to Move to AMA

  • Capital requirement for operational risk is a new element of regulation of the financial sector; hence more capital is required for this purpose from all regulated institutions
  • AMA is viewed by many national regulators as the long-term goal for most institutions
  • FSA (and some other national regulators) have presented the idea of negative incentives, to be possibly introduced under Pillar 2
  • Institutions to use simple approaches for capital savings (or investment savings) can be made subject to additional capital requirements sufficient to justify adopting the AMA

Advanced Measurement Approach

  • Continues to be a very fluid structure, even after Basel II final document
  • Basic idea is to apply the business line – loss type matrix as developed for the Standardized Approach
  • Minimum capital requirements are to be determined for each cell of the matrix using Advanced Methodologies
  • Total capital requirement for operational risk is then obtained as the direct sum of the capital requirements for the individual cells (no offsetting through correlations allowed)
  • 12 months time horizon is to be used as reference period

FSA interpretation on AMA (CP 189)

  • An important interpretation on the AMA was introduced in the year 2003, presented e.g. by FSA in CP 189
    • consequential losses of market risk and credit risk type from operational risk events have to be included in operational risk capital requirement calculations
  • This implies e.g. that delays in corrective measures – such as detecting and executing a failed hedge – will potentially generate operational losses through market movements or through deterioration of credit quality of assets
  • Consequently, portfolio market-credit value simulations may be included in AMA calculations

Business line – loss type matrix

  • 48 cell matrix combining 8 lines of business (Level 1) and 6 types of losses
  • Exposure indicators are suggested for the qualitative entries of cells
    • these include volume of trades, volume of transactions, value of assets, value of transactions etc.
  • Business lines can be further mapped into Level 2
  • With insurance operations included this mapping increases the number of Level 1 lines of business to 9 and the number of Level 2 lines of business to 24
  • Corresponding number of cells of matrix are 54 and 144 for the Level 1 and Level 2 mappings

Advanced Methodologies

  • Internal Measurement Approach (IMA)
    • Key parameters are EI (exposure indicator), PE (probability of loss event) and LGE (loss given event)
    • EL (expected loss) = EI*PE*LGE
    • Authorities will determine the g (gamma) function which transforms EL into capital requirement for each cell
  • Loss Distribution Approach (LDA)
    • Bank estimates the above three probability distribution functions for each cell
    • Based on these distributions bank then computes the probability distribution function of the cumulative operational loss
    • Capital charge is based on the simple sum of the VaRs of cells
    • Correlations may also be allowed if verified by bank

Implementing IMA in CDFT Platform

  • Implementing straight IMA in CDFT R/V Platform involves relatively simple steps:
    • for each cell of matrix, obtain values of exposure indicator variables
    • for each cell of matrix, obtain point estimates of probabilities of events
    • for each cell of matrix, obtain point estimates of losses given events
    • each estimates are to be forecasts for the next 12 months period
    • for each cell of matrix, compute expected losses and values of g functions
    • the Minimum Capital Requirement to cover operational risk is then obtained as the straight sum of the values of g functions for individual cells

Extending IMA in CDFT Platform

  • Straight IMA calculates just the MCR for operational risk in the standard case
  • IMA can be extended in two ways in CDFT R/V Platform:
    • First, let loss given events for each cell of matrix be random and simulated variables – capital requirements can then be computed at different confidence levels, which may help the bank in planning for its potential capitalization pressures
    • Secondly, let loss given event random variables be correlated with each other in general terms, i.e. so that correlation coefficients may deviate from +1 (the implied case in computing direct sum of cells)
    • These calculations are likely to turn out useful for economic capital and pricing purposes and allow computing the Risk Profile Index for MCR adjustment for bank-specific operational risk profile

Implementing LDA in CDFT Platform

  • Steps to be taken in implementing LDA in CDFT R/V Platform:
    • for each cell of matrix obtain distribution functions for the frequency of operational risk events for 12 months time horizon
    • for each cell of matrix obtain distribution functions for the single operational risk event impact for 12 months time horizon
    • simulate the distributions to obtain the cumulative operational loss distribution function
    • compute the regulator-specified VaR number to obtain directly an estimate for the capital charge for each cell
    • sum up the VaR numbers of cells to obtain the MCR for the whole bank

Advantages of LDA

  • LDA yields a better risk sensitivity than IMA
  • LDA measures unexpected loss directly and not through an RPI adjustment
  • No need for gamma function
  • The bank can determine the business line – loss type matrix itself and can use but is not limited to the regulator-presented sample matrices

Extending LDA in CDFT Platform

  • LDA can be extended in CDFT R/V Platform in different ways:
    • general correlation structure can be introduced between (any) random variables – this improves accuracy of calculations for economic capital/pricing purposes
    • operational risk control measures can be modeled as mathematical control rules – this helps in quantifying the benefits of controls
Related Products
Related Services
Related Solutions