The Statistics of Forecasting - Part III

The final part of the three part series deals with forecasting bias. How are forecasts derived? Well as we discovered in the second part of the series it is impossible to guess as to what new technologies or natural disasters will happen. Therefore in order to provide an educated guess we need a point of reference. That point of reference is where we are today. So the first problem is that all "guessers" are at different starting points so their points of reference and hence their predictions will be biased towards their personal situations.

Consider a "guesser" living in London versus a "guesser" living in India or China. Certainly very different circumstances and hence very different points of reference from which to begin to make a forecast. The Londoner would look at the dismal economic growth of Europe and the housing market in and around the United Kingdom plus the massive deficit that his government is running. The Chinese "guesser" thinks to herself that while the rest of the world is having its problems things are good in China and we have a massive surplus. Housing prices may be high but we all believe in our government to support us and with this support we will become the next super power. Certainly very different outlooks and this will bias the forecast considerably.

The next issue is that the forecast must be based on something and humans base their forecasts on the past. Our past should point to our future right? Not in the least and this is the main reason why forecasts are notoriously incorrect. While the past can repeat itself, there is no reason to believe that the timing or the magnitude of the future move will be anything like the past. Furthermore, while history is thought to repeat itself it is impossible to have a true comparison as the circumstances at the time of the "similar" event were more than likely very different to the conditions faced today. Different leaders, different geographical regions and different ideologies just to name a few.

Consider the housing crisis of the 1980s compared to today's housing crisis. While we could (and many analysts do) use the 1980's crisis as a barometer to gauge how the recovery process will unfold, to use it in a vacuum would result in a severely flawed forecast. While the 1980's crisis caused some problems and dragged down some savings and loan institutions it did not wreak havoc with the global economy and the United State was not $14 trillion in a hole. These are very different animals and should therefore not be compared too closely.

Now please do not think that the past cannot be used as a point of reference. I certainly believe that a study of the past is critical to prepare for the future, but always be prepared to remain flexible as the current situation will not completely replicate the past.