For several reasons, many forecasts are based on using two or more techniques. Most typically, you will find yourself using some form of a quantitative and judgmental technique. This is because while quantitative techniques tend to be quite rigorous and defensible, judgmental techniques tend to do a better job of allowing for sudden changes in the forecast environment (e.g. Gulf War). Moreover, many researchers like to understand the explicit causal factors that go into a forecast and judgment techniques do a good job of providing that information.
The critical issue is how best to combine techniques. While you may be tempted to do this in a casual, informal manner (and indeed have experience providing reasonably good forecasts), doing so tends to make the forecast hard to justify to others and may undermine the credibility of the whole forecasting process. To enhance the process, researchers tend to encourage the use of a more formal structure. Minimally, you should use an:
• AUDIT TRAIL : This simply means providing a well-documented (on paper) reasoning behind the way you combined forecast methods. For example, what is the reasoning behind the judgmental adjustment to a quantitative forecast you've made, or the particular weights you used in combining separate forecasts (e.g. 50% quantitative model and 50% judgment).
• More formally, you should provide an explicit and systematic representation of the rationale behind the adjustment or combination mechanism used. For example, a judgmental adjustment to a quantitative forecast might begin by first decomposing the rationale into various components (e.g. what are the separate effects of the economy, industry, and/or firm-specific factors on the judgmental adjustment). You might then think about the likelihood of the different possible outcomes and assess their subsequent effect on the judgmental adjustment. In any case, a simple "what-if" analysis of the effect of different assumptions would go far in making your forecasts more defensible.
MarketingProfs uses single