Planning Analysis:
Forecasting


You should also visit the Calculating Growth Rates page.

Estimates and Forecasts

Types of Forecasts

A. Extrapolation

Extrapolative Forecasting - a method of prediction which assumes that the patterns that existed in the past will continue on into the future, and that those patterns are regular and can be measured. In other words, the past is a good indicator of the future.

Applications: good for developing baseline data. Not as good for assessing the future impacts of a policy change, unless historic data for a similar policy change is available.

Attributes & Limits - simple, cheap and often as or more accurate that complex theoritical models.

Process - plot data and observe patterns.

Key: having a good base of data and understanding the pattern within it--easier said than done

Techniques: best fit, ratios, etc

B. Theoretical Models

Theoretical Forecasting - modeling, or using a construct of how some subsystem of the world functions to predict how things will happen in the future. There are empirical models, which reduce the problem to calculations, and nonempirical models, which do not quantify the problem under scrutiny.

Largely involves modeling. Models specify the linkages between relevant variables and are nothing more than informed speculation about the behavior of some system.

Applications: many rational decisions are made on the basis of models e.g., the law of gravity. (discounting, certain pop models, etc)

Attributes & Limits - more complex than extrapolation.

Process - based on longer-term relationships and research.

Techniques: econometric models, cohort survival, Reilly’s gravity model, discounting

C. Intuitive Prediction

Inductive Forecasting - a set of methods in which the future state is predicted (by persons who have some knowledge that makes them likely to do this accurately), and then data and assumptions necessary to achieve this outcome are deducted.

The most common forecasting methods. Rely primarily on subjective judgement. (seat-of-the-pants planning).

Attributes & Limits - uses retroductive logic eg a future state is described and the retroductive logic is used to find data and assumptions consistent with the forecasted end state.

Process - 1. choose knowledgable persons to interview. 2. anonymity; types - Delphi, cross impact analysis, feasibility assessment, etc.

Guidelines for Selecting a forecasting method

  1. relative accuracy
  2. type of population data available
  3. quality of data available
  4. scale of the analysis
  5. length of projection period
  6. budget & time

Comparison of Population Projection and Estimation Models

 

Relative Complexity

Type of Data that Can Be Used

Estimating or Projecting Technique

Type of Model

Simple

Moderate

Complex

Historical Counts

Vital Components

Other Indicies

Estimation

Projection

Noncomponent:                
Trend Extrapolation

ü

ü

 

ü

 

ü

ü

ü

Comparative Forecast

ü

   

ü

     

ü

Ratio Trend

ü

   

ü

   

ü

ü

Density Ceiling  

ü

ü

ü

     

ü

Ration Correlation  

ü

ü

ü

 

ü

ü

 
Housing Unit  

ü

ü

ü

 

ü

ü

ü

Market Force  

ü

ü

ü

ü

ü

 

ü

Component:                
Residual

ü

     

ü

 

ü

 
Vital Rates  

ü

   

ü

 

ü

ü

Cohort-Survival  

ü

ü

 

ü

 

ü

ü

Cohort-Component  

ü

ü

 

ü

ü

ü

ü

Composite  

ü

ü

ü

ü

ü

ü

ü

Source: Local Population and Employment Projection Techniques. Greenburg, et al. Center for Urban Policy Research, 1979.

 

Comparison of Population Projection and Estimation Models 

 

Appropriate Estimation and Projection Period

Appropriate Estimation and Projection Scale

Appropriate for Studies Involving Numerous Entities

Type of Model Short Middle Long Nation State County Local  
Noncomponent:                
Trend Extrapolation

ü

       

ü

ü

Yes
Comparative Forecast

ü

       

ü

ü

Maybe
Ratio Trend

ü

ü

ü

 

ü

ü

ü

Yes
Density Ceiling  

ü

ü

   

ü

ü

Yes
Ratio Correlation

ü

       

ü

ü

No
Housing Unit`

ü

ü

     

ü

ü

No
Market Force

ü

ü

ü

ü

ü

ü

ü

Yes
Component:                
Residual

ü

   

ü

ü

ü

ü

Yes
Vital Rates

ü

     

ü

ü

ü

Maybe
Cohort-Survival

ü

ü

ü

ü

ü

ü

ü

Yes
Cohort-Component

ü

ü

ü

ü

ü

ü

ü

Depends on Migration Model
Composite

ü

ü

ü

 

ü

ü

ü

Yes

Source: Local Population and Employment Projection Techniques. Greenburg, et al. Center for Urban Policy Research, 1979.


[Home | Syllabus | Schedule | Assignments | ProjectLinks]

This page maintained by Bob Parker, ©2002
October 21, 2003