Forecasting & Estimating
rating: 0+x

Definitions

  • Forecasting deals with the future
  • Estimating deals with the present and the past
  • Each have the basic purpose of providing the best possible quantification of some unknown value
  • Both contrast with cost analysis techniques
  • F&E support cost analysis, but cost analysis is used less frequently and with more focus to deal with specific questions or issues

Accuracy of P&E

  • Technical defense
    • Use most accurate techniques available for items of major consequence
    • Build error protection countermeasures into predictions, e.g. introduce systematic bias toward low side in revenue projections and high side in expenditure projections
    • Build contingent defenses against consequences of prediction errors e.g. maintaining asset reserves, imposing controls over expenditures
  • Political defense
    • Political concerns often motivate optimism which run counter to technical defenses. Proponents of a program prefer high revenue estimates and low expenditure estimates, while opponents prefer the opposite
    • Two political defenses are being unbiased and open to new ideas and approaches
    • Unbiased means striving for reasonable levels of technical accuracy in predictions
    • Open means being completely candid that predictions inevitably are in error and about all details of how predictions are made

F&E Techniques

  • Opinion – rely on one or more person’s opinion about future of events. One specific opinion technique is the Delphi Technique, asking experts what their prediction or estimate is and then have them come together to create a consensus
  • Sampling – observing a portion of a group of phenomena and extrapolating from that portion to get an impression of the group as a whole.
    • Most accurate is for past or present.
    • If sampling data from a later period are not consistent with estimate based forecast, then indication that initial forecast was incorrect.
  • Time series – prediction based on the past temporal patterns of the thing being predicted, assumption being that future is like the past.
    • Trendline forecasting is common example.
    • Danger is that it discounts change in initial assumption of rate and direction of change.
  • Associational techniques – more complex than proceeding. They predict a value for one or more variables base on one or more variables.
    • Previous techniques dealt with one variable.
    • Rely on one or more assumptions regarding the relationship of the variables used.
    • Includes correlation, simple and multiple regression, econometric models

Monitoring Estimates and Predictions

  • Relate subsequent information (what was the future, now the present) to test the accuracy of the estimate or prediction
  • Depends on the availability of data and information
  • Can be evaluated mathematically or judgmentally
Add a New Comment