In the theme of this release, we have looked at some possible extensions to the prepayment function. First we examine whether prepayments seem to be seasonality dependent. It is our prior believe that more people move in the summer than in the winter and thereby we expect to observe greater prepayments in the summer months also. We therefore try to include two different dummy variables in the prepayment function; One for July terms and one for October terms. We get a positive coefficient for the July variable but a negative one for the October variable. However, on group level, there does not seem to be any noteworthy effect from including term date variables and so prepayments does not seem to be seasonality dependent. This is further confirmed by including an April dummy in the prepayment function. This variable has positive coefficients and moreover the results seem to be more significant in this case than it was in the two prior cases.
In the second part of the theme we examine whether there seem to be a connection between the age of a bond and prepayment. In the literature this concept is known as seasoning. The hypothesis is that fewer people want to prepay their loans immediately after they have obtained them. The prepayment for a given bond is thus expected to begin at zero and then increase until the bond reach a certain age. After this prepayments are assumed to be constant. We include two variables in the prepayment function. One is an indicator of the rate at which the prepayment increases during the first years of the bond and the second implies when the seasoning effect are no more relevant. Judging from the results, it seems as though prepayment estimates can be improved by correcting for seasoning.