All statistical estimates make use of some models and assumptions. At times of change, these approaches are less likely to reflect what is actually taking place, as we generally use methods which do not react to every change. So it is a paradox that at times of recession the statistical estimates are both most important and least certain.

One solution is to have a different approach during recession. But then we have to make a decision about when a recession starts in order to trigger the change; and this process itself may trigger a discontinuity in the measurements just when we are trying to estimate change. So ideally we want processes that work well both in normal times and recessions.

More research on the best approaches to measurement during times of recession are needed. The papers [below] give some pointers and ideas, but it would be great to have a discussion and share ideas on this problem. If you have any papers or reports which would be suitable, please send an email to repository@enbes.org to contribute.