Forecasting Problems

1
Forecasting Problems

2
Volatility and variability

• Point Estimates will be wrong
• An interval estimate can be more accurate
• confidence interval
• trade-off precision vs. accuracy

3
Bias

• Psychological bias
• optimism vs. pessimism
• Sample bias
• not random, not representative
• Intentional bias
• some forecasts are purposely misleading

4
Mathematical modeling

• Processes
• linear
• exponential
• logistic
• cyclical
• Wrong model Poor forecasts

5
Turning points

• Change from current trend can be very hard to
  predict

6
Forecasting is based on past experience

• Unexpected change is not modeled in forecast
• Consider all assumptions
• explicit and implicit
• Forecasts will be wrong for unexpected shocks

7
Feedback effects

• Behavior changes in response to conditions
• Forecast is based on past behavior
• When change occurs, forecast is inaccurate

8
A feedback example

• Forecast oil will be depleted
• Oil prices increase
• Producers increase exploration
• Consumers conserve and use substitutes
• Result oil reserves are not depleted

9
Forecasts

• Short term forecasts tend to be more reliable
• Mathematical forecasts should only be one input
  into planning
• critical uncertainties
• scenarios