Data Integrity and Quality_ and Operational Risk

Document Sample
Data Integrity and Quality_ and Operational Risk Powered By Docstoc
					Data Integrity and Quality, and
       Operational Risk
      Scribe: H. V. Jagadish
    Moderator: Anthony Tomasic
•   Quality Measurement
•   Information Recording
•   Integrity Maintenance
•   Risk Representation
•   Other Information Representation

• Enabled Big Science
       Quality Measurement
• Much debate within the financial industry
  on how we define, measure data quality.
• Seven metrics
  – Completeness, conformity, precision,
    accuracy, timeliness, association, and
• Three dimensions
  – function, methods of measurement, and
    definition of quality.
           Data Recording
• Much will now be recorded, and public.
• Need procedure to make corrections.
• Must represent and record metadata.
• Provenance recording by extending audit
  trails already used by banks.
• Time stamp granularity is concern for fast
  moving phenomena.
• Sniff data on wire for fast alarms.
       Integrity Maintenance
• Required at multiple layers, particularly at
  application level.
• Discrepancies across data sources can
  catch errors (and fraud).
• Patterns across multiple sources may be
  helpful in detecting fraud.
• Need to build logic for what to do when an
  error is detected.
         Risk Representation
• Single number (VAR/Risk Premium)
  cannot be rolled up or manipulated.
• Full state space distribution is too costly.
• How to show risk to decision makers in a
  manner that helps them act.
• How can you drill down from aggregate
  data based on risk factors.
 Other Information Representation
• Formula as attribute value.
  – Treat better than as text field
• Accounting system used is also “data”.
  – Enable queries on accounting rule.
  – Can you measure risk exposure of accounting
     Big Science Questions

Given the opening of the OFR, and
 collection of large data that it promises,
 what new research questions can we ask?
         Big Questions (1 of 2)
• Validate VAR over long time periods.
• Can you model Knightian surprise? Can you predict
  which instruments are more at risk?
• Modeling of Tail dynamics
• Pricing when liquidity breaks down and you cannot
  create a replicating portfolio
• Characterize empirically when weak efficiency of market
• Can trade in the derivative of the security influence the
  value of the underlying asset?
• Tradeoff between capital level required within the bank
  and the rate of growth within the economy.
         Big Questions (2 of 2)
• Structure of financial institution network. Shape of the
  graph. Incentive structures of the individuals.
• Visualization of financial stability for policy makers.
  Which are reasonable indiciators/measures to track.
  How do you summarize effectively to communicate – to
  policy makers, to marketplace.
• Herd behavior or asset bubbles. How do you detect?
  How do you manage?
• Macro-economic models that include securitization and
  financial markets, particularly things that happen outside
  the banking system.
• Develop a new risk-based forward-looking accounting

Shared By: