Speaker: Weiping Chen, senior vice president of Wells Fargo
4 November 2015, Beijing
1.Importance of Quantitative Models in Risk Management
(1)Since the economic crisis, risk management has been increasingly important.
-The economic crisisstarted in 2007, peaked in 2009 and started recovery in 2010.
It is Centered in credit crisis for the housing market, and also included loose origination criteria together with fancy new products.
-Risk management should be strengthened. Self-initiated risk managementfinancial industry, regulator, tightened regulation and CCAR (Comprehensive Capital Analysis & Review) are very important.
(2)Reliance of risk management on quantitative models has increased.
-To help with informed risk management
-To help satisfy the regulatory requirements (CCAR, Basel, etc.)
-Scarcity of modeling resource
2.Important Attributes of a Good Modeler
(1)Integrity
-The most important attribute of a good modeler
-Definition of integrity
-No data/model manipulation/tailoring
-No hide of mistake
(2)Thinking Skills
-An outstanding modeler must be a good thinker.
-It is critical for model design, approach selectionand judgment and model decision.
(3)Business Understanding
-We should understand business needand business meaning of data.
-We should ensure model’s consistency with business processand model’s business sense.
(4)Communication Skills
-We should know clearly about the nature of consulting service.
-We should understand the background of clients.
-We should know about how to communicate in non-statistical language
-It is critical to ensure that a model does meet the need of risk management and it will be used in the right way.
(5)Technical Skills
-A good modeler is very necessary.
-There are also some dangers of over-do, which includes business-model disconnection and unnecessary model complexity.
-Business judgment may be needed to overwrite the data-based technical results.
3.Critical Components of Model Development Process
(1)Data exploration is quite important.
-Data exploration mainly includes data cleaning,distribution and preliminary analysis, data understanding, data validity check and data reconciliation.
-In case of poor data, data exploration becomes the key.
-Data is the oil to the engine, however, huge amount of information also brings about challenge like noise, Computational capacity and analytical approach.
(2)We should make sure that the level of model complexity is appropriate.
-Use a simple model unless you can justify the added complexity
- Benefits of using simple models: Easy to communicate; easy to implement; less chance for mistakes; easy to justify in validation; time saving.
(3)Documentation is rather important as well.
-Documentation is the most important tool to communicate your work.
-A good documentation is supposed to provide reasoning and judgment in model decision making, have good balance between details and high level summary, and keep consistency of results or numbers across chapters; etc.
For further details
All developments regarding IMI can be followed at http://imi.org.cn/.
For further information, contact via imi@ruc.edu.cn or on +86-10-62516755.
