I think it hardly needs mentioning again, but I guess I will: the legislative process in North Dakota probably makes it even more important that we have some confidence in our revenue forecasts. Our legislators are meeting for three months to determine budgets for the next two years. There is always the possibility of a special session if need arises, but you want that to be the truly exceptional case. Now I am not suggesting that anyone will ever get the numbers spot on, 100% accurate, but we can get closer.
In earlier posts I listed many things that are currently wrong with our forecast process. One that I likely neglected, or at least did not emphasize adequately, is a statement of risks to the forecast. Now to be fair, the statement of risks likely does us little good if we do not actually have a meaningful statement of the numbers within the forecast, but I already made that point several times.
One thing I teach my students is to understand the risks in their forecasts. Risk in the forecast is not a flaw, it is a recognition of reality. Sometimes events happen beyond our ability to forecast. The forecaster needs to recognize this and communicate it to their client. Would I suggest that you must include outcomes of Presidential elections in your forecast? No, but if you think there are likely differences in regulatory approach you should be prepared to issue changing guidance after the election results are in. Once again, this just seems to be good practice.
There can also be an asymmetry to the risks faced. There really is no reason to believe that our forecast process errors will be distributed normally, but let’s just say they are. The risks from OPEC changes right now are likely all to the plus side for North Dakota, it is really only a question of the magnitude of any effect on market prices. So unexpectedly effective, or over-the-top, moves on the part of OPEC may not be something you include, but you can list it as a possible risk and the effects would likely mean your forecast understates oil activity and possibly revenues. Is that so difficult?
The desired output of the client is not necessarily two-tailed. In the context of revenue forecasts it may be that they want a 95% confidence interval where the 5% is all in the lower tail. The interpretation then is that we are 95% confident the revenue forecast will be higher than a given amount.
Much of this is cart before horse. I am telling you what is truly excellent and complete when we are dealing with haphazard, partial, and just plain nonsense. However, we are at the point where I think it becomes our own fault for continuing to give those with a bad track record further bites at the apple.
