Six Principles for Forecasting

1
Six Principles for Forecasting

• Diebold, Chapter 2

2
1. Loss Function

• Symmetric vs. Asymmetric Loss Function
• Loss FunctionL(e) where ey-y
• Quadratic Loss Function L(e)e2
• Absolute Loss Function L(e)e

3
Examples

• Electricity industry, short term forecasters are
  called upon to forecast the demand for
  electricity on an hourly basis one day ahead
  taking account of the number and type of
  customers, weather, etc.
• Asymmetry?

4
Examples

• A forecaster for a manufacturer of airplanes may
  be called upon to forecast future demand for a
  model of airplane that is not built yet.
• Asymmetry?
• Forecasts of copies of syllabi needed for class

5
2. Forecast Object

• Event outcome will Fed Chairman be reappointed?
• Event timing Will the economy peak this quarter
  or next?
• Time Series projecting future value of a time
  series what will the price of gasoline be next
  month?

6
3. Forecast Statement

• Point Forecast the price of gasoline will be
  3.19/gallon
• Interval Forecast the price of gasoline will be
  between 2.50 and 3.50/gallon with 90
  probability
• Density Forecast whole distribution of range of
  values
• Probability Forecast this event will happen
  with 65 probability

7
4. Forecast Horizon

• H-step-ahead forecast i.e. 4-step-ahead forecast
  only 1 point 4 periods after the end of the
  time series
• H-step-ahead extrapolation forecast -
  4-step-ahead extrapolation forecast all 4
  periods after the end of the time series

8
5. The Information Set

• Univariate only information about the series
  itself
• Multivariate also have information about
  variables potentially related to the series

9
6. Methods and Complexity

• Parsimony principles other things being equal,
  simple models are preferred to complex models
• KISS principle Keep it Sophisticatedly Simple

10
The Forecasting Process Bails and Peppers
Chapter 1
11
The Forecasting Process

• 1. Determine the purpose and objective of the
  forecast
• 2. Select the relevant theory
• 3. Collect the data
• 4. Analyze the data
• 5. Estimate the initial model

12
The Forecasting Process

• 6. Evaluate the model and make revisions
• 7. Present initial forecasts to management
• 8. Make the final revision
• 9. Distribute the forecast
• 10. Establish monitoring procedures