Business Forecasting Topic 11 Flashcards
forecasts involve judgment in what ways
- choice of method
- how much past data
- which variables to include
direct (widely used) - wholly based on judgment or judgmental adjustments to statistical
benefits of management judgment to forecasts:
- account special events (strike, gov initiative)
- compensate for deficiencies/absence of historical data
- allow managers to have control over variables
- acceptable forecasts - managers = sense of ownership
(complex stats method = lack transparency and credibility)
problems of judgment related to humans
- limited info processing capacity
- simple mental strategies = cope with complexities
Heuristics
are simple strategies that humans, animals, organizations, and even machines use to quickly form judgments,
3 types of heuristics
- availability
- anchoring and adjustment
- representatives
availability
- likelihood of future events -> how easily similar past events are recalled
- over influenced by recent/vivid/events in press
fear e.g. violent crime more than obesity, terrorism more than smoking or food poisoning
anchoring and adjustment
- making estimate, start with initial value and adjust from this
- estimates tend to be close to anchor even if implausible value
- underestimate upward trends (severe for exponential trends) - stay close to. most recent values
Representativeness
extrapolating time series = expect future pattern to be representative of past pattern
- try reproduce perceived past pattern in forecasts - much of this past pattern is noise
narrative fallacy
expert in investing explanations for these random movements
- many movements in sales = random = people brilliant at inventing explanations
5 problems with judgment
- hindsight bias -> wrongly think could have easily predicted - knew it all along
- overconfidence -> better forecasts than really are despite messages from computer
- political biases - dishonesty to please boss, Sandbagging - deliberately low
- optimism bias - calculated
- group biases - conformity
6 ways of making the best use of judgment
- Delphi method
- decomposing
- feedback
- adjusting if appropriate
- record reasons
- combine judgment with statistical
Delphi method
- avoid danger of face to face with groups
- individual and anonymous forecasts & reasons
- results tallied and summarised
- re-polling
- repeated until consensus emerges - median estimate = used as forecast
adv and dis of Delphi
adv: large/geographic groups, no dominance, change minds
dis: little chance of persuasion, choosing panel members (experts = over analyse/not open minded)
decomposition
make judgmental tasks easier through breaking down to smaller easier tasks
adjustments -> price reduction, promotion, new customer
only if this is practical
feedback on performance
MAPE or MPE
on bias - more useful than feedback on accuracy