Chapter 3 Flashcards
Techniques for dealing with uncertainty
Set minimum payback period
increase discount rate to submit project to higher hurdle rate
make prudent estimates of outcomes to assess worst situation
assess worst and bestt case
use sensitivity analyssi
Sensitivity formula
NPV Project / NPV of CF subject to uncertainty
Strength of sensitivity analysis
Information presented in form which facilitates subjective judgement
identifies areas of success critical to project
Critical issues strengths
Identifies areas which are critical to the success of project, if undertaken these can be carefully monitored
Sensitivity issues - independence
Assumes changes to variables can be made independently
Sensitivity analysis - probability
Only identifies how far a variable needs to change, not the likelihood that it will change
Sensitivity analysis - no answer
Not optimising technique, only gives information and does not point to decision
Linear regression - predict the
Dependent variable
Linear regression - factors that are thought to impact on dependent variable are called
independent variable
Advantages of linear regression
Simple to use and easy to explain
Can be used to predict impact beyond current estimates
Limitations of linear regression
There is not always a linear relationship
basic models only consider one variable at a time
May identify spurious relationships, correlation is not causation
less meaningful if the data is inaccurate
Advantages of decision trees
Simple ones easy to explain and logical
Can be used to consider different outcomes that can occur based on changes in a number of variables
consider multiple decisions
Limitations of decision trees
Usually restricted to small number of outcomes
large decision trees can be hard to interpret
Advantages of simulation
Gives more info about possible outcomes and probabilities
useful for problems which cannot be solved analytically
Limitations of simulation
Not a technique for making decision, only gives info
can be time consuming without a computer
can be expensive to design and run for complex project
Monte Carlo techniques require assumptions about probability distributions and reltaionships between variables
Advantages of prescriptive analysis
Capability to identify optimum investment decisions whilst considering impact of multiple decisions and variables
Limitations of prescriptive analysis
Creating reliable prescriptive models is complex and requires specialist skills
reliability depends on reliability of data used and ability to predict future outcomes from past data
Reasons data analysis is not always accurate
Inherent bias in the data (intentional/unintentional)
Data may have been intentionally manipulated
Data may have been analysed accurately, but presentation of the data or conclusions may be flawed
Data bias definition
not representative of the population
Selection bias definition
Data is not selected randomly and not representative of the population as not every item has equal chance of being selected
Self selection bias definition
Occurs when individuals elect themselves to be part of a sample (online questionnaire)
Observer bias definition
Occurs when observing and recording, researcher allows assumptions to influence observations
Omitted variable bias definition
Key variables are not included within the data to be analysed
Cognitive bias definition
Relates to human perception and includes bias depending on how data is presented
Confirmation bias definition
When people see data that confirms their beliefs and ignore data that disagrees
Survivorship bias definition
Sample only contains items that survived some previous event
Mean definition
Average- sum of all values divided by number
Standard deviation
Show amount of variability in a data set, showing on average how far each result lies from the mean
risk means variablility of outcomes and can be measured using SD
Coefficient of variation
Ratio of SD to mean
SD/Mean * 100
Normal distribution
Frequency distribution arises frequently in ‘real life’ any distribution that is symmetrical around the mean
% of data within 1SD above/below mean
68%
% of data within 2 SD above/below mean
95.4%
% of data within 3 SD above/below mean
99.7%
Left/negatively skewed distribution has the majority of values concentrated on
RHS of distribution
Right/positive skewed distribution has majority of values concentrated on
LHS of distribution
Left skewed data median mode mean order (left to right)
Mean Median Mode
Right skewed data median mode mean order (Left to right)
Mode, median, mean
Advantages of expected values
Information reduced to single number for each option
Idea of average is understood
Limitations of expected values
Probabilities may be difficult to estimate (could base on past experience or market research)
expected value may not correspond to any expected outcome
unless the same decision has to be made many times, expected value will not be achieved, is therefore not valid way of making decision in one off situation
gives no indication of spread