Lattice vs. Decision Analysis

Slides:



Advertisements
Similar presentations
Drake DRAKE UNIVERSITY Fin 288 Valuing Options Using Binomial Trees.
Advertisements

Engineering Systems Analysis for Design Richard de Neufville © Massachusetts Institute of Technology Flaw of Averages Slide 1 of 29 Richard de Neufville.
Introducing Information into RM to Model Market Behavior INFORMS 6th RM and Pricing Conference, Columbia University, NY Darius Walczak June 5, 2006.
Introduction to ModelingMonte Carlo Simulation Expensive Not always practical Time consuming Impossible for all situations Can be complex Cons Pros Experience.
Engineering Systems Analysis for Design Richard de Neufville  Massachusetts Institute of Technology Use of Simulation Slide 1 of 19 Use of Simulation.
QA in Finance/ Ch 3 Probability in Finance Probability.
1 Derivatives & Risk Management: Part II Models, valuation and risk management.
Engineering Systems Analysis for Design Richard de Neufville © Massachusetts Institute of Technology Dynamic Programming Slide 1 of 35 Dynamic Programming.
Options for Supply Chain Management Massachusetts Institute of Technology Richard de Neufville  Dec. 5, 2002Slide 1 of 23 Richard de Neufville Professor.
PHARMACOECONOMIC EVALUATIONS & METHODS MARKOV MODELING IN DECISION ANALYSIS FROM THE PHARMACOECONOMICS ON THE INTERNET ®SERIES ©Paul C Langley 2004 Maimon.
Links HOME Options Pricing Model-based pricing of options is a relatively new phenomenon. Until the early 1970's option premiums were determined by offer.
Engineering Systems Analysis for Design Massachusetts Institute of Technology Richard de Neufville  Real Options SDMSlide 1 of 31 Richard de Neufville.
Engineering Systems Analysis for Design Richard de Neufville  Massachusetts Institute of Technology Use of Simulation Slide 1 of 17 Use of Simulation.
Engineering Systems Analysis for Design Massachusetts Institute of Technology Richard de Neufville © Decision Analysis Basics Slide 1 of 21 Calculations.
Heating System for a Group of Condominiums Rory Clune Dept. of Civil & Environmental Engineering, Massachusetts Institute of Technology ESD.71 Application.
Integrated Method for Designing Valuable Flexibility In Systems Case Example: Oil Development Project Abisoye Babajide Richard de Neufville Michel-Alexandre.
© 2013 Pearson Education, Inc., publishing as Prentice Hall. All rights reserved.10-1 American Options The value of the option if it is left “alive” (i.e.,
Chapter 10 Binomial Option Pricing: I. Copyright © 2006 Pearson Addison-Wesley. All rights reserved Introduction to Binomial Option Pricing Binomial.
Copyright © Cengage Learning. All rights reserved. 3 Discrete Random Variables and Probability Distributions.
Engineering Systems Analysis for Design Massachusetts Institute of Technology Richard de Neufville © Arbitrage Enforced Valuation Slide 1 of 39 Arbitrage.
Chapter 6: Random Variables
Introducing Information into RM to Model Market Behavior INFORMS 6th RM and Pricing Conference, Columbia University, NY Darius Walczak June 5, 2006.
Chapter 6: Random Variables
Table 1. Advantages and Disadvantages of Traditional DM/ML Methods
Capital Asset Pricing Model (CAPM)
COSC160: Data Structures Linked Lists
Chapter 6: Random Variables
Managing Short-Term Assets
UNIT-4 BLACKBOX AND WHITEBOX TESTING
Lecture 6 – Binomial trees
Professor Arne Thesen, University of Wisconsin-Madison
Lattice Valuation of Flexibility
Professor of Engineering Systems Civil and Environmental Engineering
Knowledge Organiser Effective Financial Management
Hidden Markov Models Part 2: Algorithms
Options Concepts Richard de Neufville
Samir Mikati, MIT Engineering Systems Division
Java Programming Arrays
Chapter 6: Random Variables
Arbitrage Enforced Valuation + Black-Scholes
Marginal Analysis and Economies of Scale
Lattice Model of System Evolution
Software Construction Lecture 2
Arbitrage Enforced Valuation
Garage case: Simulation Example
Dr. Richard de Neufville
Marginal Analysis Purposes:
Time Format for a New Label
© 2007 John Wiley & Sons Australia, Ltd
Recurrences (Method 4) Alexandra Stefan.
Professor of Engineering Systems Civil and Environmental Engineering
Lattice Valuation of Flexibility
Evaluating and Choosing Preferred Projects
International Economics: Theory and Policy, Sixth Edition
PARADIGM CHANGE IN SYSTEMS ENGINEERING
Marginal Analysis Purposes:
PARADIGM CHANGE IN SYSTEMS ENGINEERING
Marginal Analysis Purposes:
Chapter 6: Random Variables
Monte Carlo Simulation
Hybrid Approach to Option Valuation
Chapter 6: Random Variables
Arbitrage Enforced Valuation Introduction
Chapter 6: Random Variables
Arbitrage Enforced Valuation
Dan Roth Department of Computer Science
Abisoye Babajide Richard de Neufville Michel-Alexandre Cardin
Lattice vs. Decision Analysis
INVESTING IN HUMAN CAPITAL
UNIT-4 BLACKBOX AND WHITEBOX TESTING
Presentation transcript:

Lattice vs. Decision Analysis Richard de Neufville Professor of Engineering Systems and of Civil and Environmental Engineering MIT

Outline Structure Calculations Which more suitable to what situations? Similarity – tree structure Difference – location of decisions Regularity, or not Possibility of Negative Outcomes Calculations Procedure Size limitations Which more suitable to what situations?

Similarity – Tree Structure Both Lattice Model and Decision Analysis have Tree Structure What are differences? Shape: Lattice: 2 to 2 to 2…. DA: unlimited Order: Lattice: states only DA: decision – outcome – decision -- outcome

Dealing with Decisions How does Lattice Model include decisions? “one at a time” is routine (e.g.: ‘close mine’) Anything else requires special treatment Two decisions simultaneously? Can do, but… Sequence of decisions (e.g.: close, then open)? GREAT DIFFICULTY Path dependence the problem

Path Independence: Implicit Assumption of Lattice Analysis Pay Attention – Important point often missed! Model Implicitly assumes “Path Independence” Since all paths to a state have same result Then value at any state is independent of path In practice, this means nothing fundamental happens to the system (no new plant built, no R&D , etc)

When is “Path Independence” OK? Generally for Financial Options (stocks, Foreign exchange, commodities) Why? Random process, no memory…. Often not for Engineering Systems. Why? If demand first rises, system managers may expand system, and have extra capacity when demand drops. If demand drops then rises, they won’t have extra capacity and their situation will differ Process – and result -- then depends on path!

Example for “path independence” Suppose you expand the mine in year 3… What is situation in later stages? In year 4? In year 5? More combinations in 3 dimensions 2 different physical states of system for same outside state (of price)

Regularity of Binomial Model Lattice Model assumes diffusion process is “stationary” probability of next states stays the same throughout periods considered Decision Analysis not limited this way Probabilities can differ after decisions in a stage … and for each stage Ex: P(Environmental penalty) could depend on Decisions made by industry now Changes in government in later stages

Binomial Lattice: Several periods Process continues identically throughout period being considered Period 2 Period 3 Period 0 Period 1 uuuS uuS uS uudS udS S uddS dS ddS dddS

Possibility of Negative Outcomes Lattice does not allow states to shift from positive to negative values: Sign of extreme values (dn S, un S ) same as S This is realistic only for factors which cannot be negative – such as price Lattice does permit negative outcomes, when value model transforms positive state (Cu price) to negative amount (mine profits) Decision Analysis has no limitation

Summary of Structural Comparison Decision Analysis clearly is the more flexible approach

Which Approach Most Suitable? What do you think? Depends on Circumstances What if mixed circumstances?

Differences – Set up for Analysis Lattice Model uses a repetitive process (a recurrence formula) that is similar from stage to stage, and between states. Simple modular process, easy to program Decision Analysis can be different at each stage and step DA Programs available (Crystal Ball, Treeage, etc) Detailed set up required (input of vectors of outcomes, probabilities) that can differ

Differences – Calculation Time Lattice Model Problem Size proportional to N, number of stages Can easily consider 100s of stages However, looking at only 1 decision (to use flexibility or not). Thus results understandable even for large lattice Decision Analysis Problem Size proportion to power of N Looking at more than a few stages becomes complicated – and unintelligible to user 3 stages is the most I’ve seen used effectively

Summary of Analytic Comparison Decision Analysis clearly is the more difficult approach

Lattice model better for single decisions over many stages Summary Lattice Model and Decision Analysis are similar ways of investigating flexibility Each has its own role: Lattice model better for single decisions over many stages Decision Analysis better for complex, irregular processes over couple of stages