Manager’s Role in Forecasting

Which department should forecast?

1) User Department

2) Management services, or similar

3) Data processing unit, management information group, or similar

Third possibility general worst, #2 can be ok, #1 is best, but can fail due to lack of know-how. #1 can be helped by:

a) Being aware of the techniques available

b) Incorporating forecasts into management systems

c) Knowing what is likely to go wrong

Guidelines for an Organizations Forecasting System

1) Analyze the decision making systems to be served by the forecasts	List all decisions directly/ indirectly affected by forecasts; users involvement beneficial. Analysis can reveal flaws in decision systems or organizational structure that must be fixed first.
2) Define what forecasts are needed	Forecasts should directly relate to required decisions analyzed in Step 1: match user needs for accuracy, urgency, time period, etc.
3) Develop a conceptual model of the forecasting model	What is ideal data? Idea mode? Specify all influences / factors that can possibly affect available being forecast.
4) Ascertain the data available, and not available	What data can actually be obtained. Indicates how forecast can fall short of ideal.
5) Develop the method by which forecasts are to be made	Develop actual mode. Based on earlier analysis of forecasts needed, accuracy required, and data/resources available. Reduce to 2 or 3 best choices.
6) Test the Method Accuracy (using Past Data) MSE/MAD:	Tests accuracy of smoothing methods only. Hold Out Method has 2 benefits: 1) Test is independent, data being forecast (period held out) not used in establishing forecast model 2) Allows 2 types of forecasting measures to be directly compared in ability to forecast “held out data” 3) Problem: Held out data may not be typical 4)
7) Incorporate judgments into forecasts	Two Reasons: 1) some judgment are valuable, 2) User involvement/ recognition helps develop commitment to & confidence in forecasts. Two Components to incorporating judgments: 1) Obtain consensus (e.g. Delphi Group) 2) Use consensus to adjust forecast, but hold judgment-makers accountable; reduces political maneuvering, creates track record of good/bad judgment-makers.
8) Implement the forecast system	Must have consensus on what problems system has and their solutions. User involvement is precondition for implementation. Implementation based on consensual answers to 4 questions: 1) What are the users’ problems (all potential users) 2) Do all participants agree on identified problems? 3) What possible solutions to problems exist? 4) Can a consensus be reached? If not will participants agree to a trial run?
9) Monitor Performance of the forecast system	Regular evaluation is critical.. manager must ensure forecasts are being used as intended and uncover forecasting inadequacies. Emphasis on regular monitoring of accuracy, usefulness, and forecast utilization Forecasts enable mangers to take action. These actins may change outcomes making forecasts inaccurate. Not a problem, rather this shows forecasting system is worked efficiently and successfully. Essentials: record comprehensive date (qualitative & quantitative) and be willing to face facts and act on them.

Forecasting Errors

· Errors caused by companies not properly organizing the ways in which forecasts are used probably far more important than technical errors. Forecasting, as a system, must be integrated into a company’s management process.

· If forecasting technique wrongly applied, good monitoring will allow quick managerial adjustments to be made (i.e. situation can be salvaged)

· A manager can make substantial contribution to forward planning through a systematic approach to forecasting (9steps) and an awareness of hidden traps (e.g. confusing associative factors with casual factors)

Key Message from the module

Managers have a clear role in ‘managing’ forecasts. And role as practitioners of forecasting. Can use their own data to make forecasts for their own decisions without having to work through management services or data processing units.

This development has several advantages. Link technique and decision is made more easily, person has overall understanding and control; time is saved; re-forecasting are quickly obtained. Pitfalls may be no common database, no common set of assumption, apparent differences to do with data/assumption differences between profitability’s of the project statistical techniques mot as straightforward. Techniques sued by someone with no knowledge. Can or cannot be applied is dangerous. Time series method applied to a random data series is an example. Computer will always give an answer whether it is legitimate or not is another matter.

Management aspect of forecasting gives more prominence as well but disproportionate amount of time spend studying and discussing give a wrong impression of importance relative to management issues.

Forecasting on average is better than the alternative, which is often a guess, frequently not even an educated one.

The volatility in data series puts a premium on good forecasting. Systematic approach to forecasting through the nine guidelines and an awareness of the hidden traps will serve the manger well.