MBADF 617: Optimization and Financial Engineering

1
MBAD/F 617 Optimization and Financial Engineering

• Instructor Linda Leon
• Fall 2009
• http//myweb.lmu.edu/lleon/mbad617/

2

• Financial engineering is the use of financial
  instruments to restructure an existing solution
  for a financial problem into one having more
  desirable properties.

• Optimization models can help a manager
  maximize/minimize objectives or just quickly
  produce feasible solutions for highly constrained
  problems

3
Financial Engineering Examples

• Grantham, May, Van Otterloo Co., an investment
  management firm with 26 billion assets,
  developed a mixed integer programming model to
  design portfolios that achieve investment
  objectives while minimizing the number of stocks
  and transactions required.

• GE Capital, a 70 billion subsidiary of GE
  financial services business, developed an
  optimization model to allocate and schedule the
  rental and debt payments of a leveraged lease
  which allowed analysts to target profitability as
  well as optimize NPV of rental payments.

4
Another Example

• GE Asset Management, another subsidiary of GE,
  invests portfolios of assets derived from cash
  flows primarily in corporate and government bonds
  in accordance with risk and regulatory
  constraints. They developed a sequential linear
  programming algorithm to optimize the return of
  more than 30 portfolios valued at over 30
  billion while matching the asset portfolio
  characteristics with the corresponding
  liabilities. (Asset Liability Management)

5
(No Transcript)
6
(No Transcript)
7
Financial Modeling

• Many financial models which use advanced modeling
  and analytical techniques are spreadsheet based
• There is a market demand for more sophisticated
  models and analysis by financial end-users
• Most end-users prefer to develop their own models
  (cost,flexibility)

8
A model is valuable if you make better decisions
when you use it than when you dont!
Symbolic World
Analysis
Model
Results
Interpretation
Abstraction
Management Situation
Decisions
Intuition
Real World
9
Decision Support Models

• Force you to be explicit about your objectives

• Force you to identify the types of decisions that
  influence those objectives

• Force you to think carefully about variables to
  include and their definitions in terms that are
  quantifiable

• Force you to consider what data are pertinent for
  quantification

• Force you to recognize constraints on values that
  variables may assume

• Allow communication of your ideas and
  understanding to facilitate teamwork

10
Decision Models

• Exogenous Variables (inputs)
• Decisions which are controllable
• Parameters which are uncontrollable
• Endogenous Variables (outputs)
• Performance variables, or objective functions,
  that measure the degree of goal attainment
• Consequence variables that display other
  consequences so results can be better interpreted

11
Deterministic vs- Probabilistic Models

• In deterministic models, all of the relevant data
  (parameter values) are assumed to be known with
  certainty.
• In probabilistic (stochastic) models, some
  parameter input is not known with certainty, thus
  causing uncertainty in the other variables.

12
Two General Approaches to Financial Modeling

• Simulation
• Process of imitating the firm so that the
  possible consequences of alternative decisions
  and strategies can be analyzed prior to
  implementation (MBAD/F 619)
• Optimization
• Identifies which decision alternative leads to a
  desired objective given a specified set of fixed
  assumptions (MBAD/F 617)

13
Advantages of End-User Modeling

• End-users get closer to the raw data and the
  assumptions being made
• End-users can customize the models to generate
  information that fits their needs
• End-users can see results easily and immediately,
  which enhances strategy generation and encourages
  risk analysis

14
Disadvantages of End-User Modeling

• Incorrect information is generated by
  inappropriate or inaccurate models (20 to 40
  contain significant errors)
• End-users are overconfident about the quality of
  their own spreadsheets
• Poorly designed models can discourage strategy
  generation and risk analysis
• End-users may not always employ the most
  productive methods for generating insights or may
  misinterpret the generated information

15
Recent spreadsheet research shows

• End users typically do not plan their
  spreadsheets
• End users rarely spend time debugging their
  models
• End users almost never let another person review
  their spreadsheets
• Many end users do not consistently use tools that
  can make modeling productive and insightful

16
Spreadsheets the Sarbanes-Oxley Act

• Section 404 increases the focus on controls
  related to the development and maintenance of
  spreadsheets.
• Two important observations
• the error rate increases with the complexity of
  the model
• the impact of inaccuracy is more serious for
  strategic and financial models than for
  operational models