In the theme of this release we analyse option adjusted spreads (OAS). The analysis is motivated by the fact that the weighted OAS published daily by Danish FSA (Finanstilsynet) has lately decreased significantly and currently is very close to zero basis points. The spreads used by Danish FSA are calculated by Danmarks Nationalbank in the DMBS model which enables us to investigate the decrease thoroughly and assess if it is reasonable.
The first part of the analysis shows that Nykredit Index OAS develops similarly to the weighted OAS from the Danish FSA. Since the two key figures are calculated based on the same sample of bonds we conclude that Nykredit’s mortgage model reflects a similar development in spreads as the DMBS model. We find that the OAS decrease is even more pronounced for bonds with higher cou-pons such as 6% 2041 and 7% 2041. The explanation behind this is that the high market prices ob-served for these bonds are difficult to explain in the model. Currently it generates quite high pre-payment rates due to very high prepayment gains and this is not incorporated into the market price, hence we calculate large negative values of OAS.
In general the DMBS model currently tends to overestimate prepayment rates and we find that a plausible explanation for this is too little burnout effect in the model for pool factor levels above 10%. Moreover, the observations are weighted with outstanding amount in the estimation which reduces the effect of bond series with low pool factors on the estimated parameters since they typi-cally have a low outstanding amount as well.
We introduce a new model, DMBSnew (M6), which uses a different specification of the burnout function and estimate this with the observations equally weighted. The prepayment estimates from this model are generally more in line with the actual prepayments and hence the model is better at explaining the high market prices. Considering a two year period the DMBSnew model generates higher and more stable option adjusted spreads and option adjusted durations than the current DMBS model.
Finally, we look into the effect on OAS when using an alternative swap yield curve estimated to OIS rates in the short end. This is compared to the results when using the current swap yield curve which is estimated to CIBOR rates in the short end. Using the alternative swap yield curve implies higher option adjusted key figures (OAPV, OAS and OAD) for short RTL bonds and fixed rate bonds with high coupons and practically unchanged key figures for other types of bonds. Prepay-ment rates are only changed marginally for all bond types when using the alternative swap yield curve.