Derivatives data that is hidden behind expensive firewalls means the derivatives market is not transparent. This means the prices are not fair. The gap from fairness can reach extremes and then there is a crisis.

]]>

==

==

]]>

==

==

]]>

=

==

SLV v RSDMRP 20 year yield start Dec 2000

]]>

www.soa.org/files/pd/2011-orlando-valact-57.pdf

The graphs are hard to read, so you may want to buy the e-book version, which has clear copies of the graphs.

After seeing the graphs showing that SLV does poorly at low rates or transitions from medium to low or low to medium, the actuaries at the session gave Faye Albert and I low ratings for our talk. They said we had little knowledge and should not be invited back to speak again in our speaker evaluations. They said our talk had no important content including these graphs showing SLV fails. SLV is the required model by the NAIC.

Since that talk in September 2011, nothing has been done to fix these problems in the SLV model. I have downloaded the latest version of the spreadsheet and ran it. These problems are model problems not programming problems.

Also, when the interest rate portion of the Academy of Actuaries spreadsheet and stock parts were combined, errors were introduced. I built a replicating system and found some programming errors leading to output errors. No one else reported them. Those were fixed in the spreadsheet. Steve Strommen is the person who coded in the fixes to the Academy of Actuaries spreadsheet. Note these are distinct from the model problems that still persist in SLV.

So out of the entire US life insurance industry that is required to use the scenarios from the Academy of Actuaries spreadsheet, no one built a replicating system to test the spreadsheet. Thus no one found the errors introduced at the time of combining the interest rate and equity parts. Fortunately, I did, and the errors were fixed soon after the combination. Whether companies bothered to download the fixed version and use it is hard to say. Sometimes people use old versions because they get used to them.

How to get the message out?

]]>

==

==

Comparison of SLV to RS-DMRP for December 2000 Start Date US Treasury 5th 95th percentile

One notes the SLV model as calibrated does not deal well with the zero interest rate bound monetary policy, i.e. keeping the short term rate close to zero.

]]>

]]>

RSDMRP v DMRP 5th 95th Percentile Pair Dec 2000 Start 1 year yield US Treasury

The graph shows the 5th 95th percentile pairs of two models. These start from Dec 2000. Also shown is the historical line, H, in green.

DMRP stands for Double Mean Reverting Process. It is often called the 2 factor Black Karasinski model, even though they never published such a model.

RS-DMRP is a regime switching super model of the DMRP.

The graph shows that the 5th percentile of DMRP does not go low enough from 2000 to contain the historical line in the two low interest rate episodes. It also can’t deal with further low rate policy from Ben Bernanke at the Federal Reserve.

The Beaglehole Tenney Double Decay Model of 1991 is an example of a model in the Double Mean Reverting Framework. It is easier to solve because it is a 2 factor multivariate model. It is sometimes called Two Factor Vasicek even though Vasicek never wrote on such a model.

]]>