Over how long should we be tracking DVR? In my own mind, I was happy with aiming at a horizon of 10 years but I can see that others may prefer 15 years. While the longer period reduces the extremes seen for 10 years, 15 years is a very long time. Or is some intermediate period just as good, or better?

What I have done is to look at the time comparisons for the periods 10 years through 15 years, finding the average values and  standard deviations. For both  asset categories, these figures are charted (see below). It is clear that, for equities, increasing the period reduces both the mean of the DVR divergence and its variance. For gilts, there is not much difference.

While I recognise that 15 years can certainly give more information, the problems are having to wait so long and having fewer  measurement intervals over which to  assess the results. It should also be borne in mind that there is less independence over longer periods (the years are repeatedly used).

In fact, 15 years has been a fairly typical forward planning period for SSAP24 purposes (although that accounting standard is being progressively replaced by FRS17). My revised view is 15 is best.

Statistics_Equities

Statistics_Fixed