A while back I was asking about SPIA valuations and Money's Worth Ratios.
I've now made available a calculator that can be used for computing actuarially fair SPIA prices:
http://ift.tt/1Gbqt0Z
It works by combining a mortality table with an interest rate table. For mortality tables I have the 2012 IAM with G2, Social Security AS 120, and an AS 120 table with modified qx values to match a given life expectancy. For interest rates yield curves I have U.S. Treasury TIPS, U.S. Treasury nominals, and U.S. High Quality Markets corporate bonds.
The Python source code is available under the GNU Affero license for anyone interested in inspecting the calculator for any flaws.
I am finding the SPIAs quotes I have seen are quite competitive when measured against these yield curves.
One issue that I don't feel I have fully resolved is applying the 2005-08 Actual/Expected contract duration data. Without any adjustment the 2012 IAM mortality of long time and recent annuitant is seen as the same, when in reality this is not the case. Applying the full contract duration a/e table might seems the best option, but the table is noisy and incomplete. The all ages summary is less noisy and is complete, but information is lost in using it.
I'm not an actuary, so I hope to learn from any feedback you might be able to provide.
I've now made available a calculator that can be used for computing actuarially fair SPIA prices:
http://ift.tt/1Gbqt0Z
It works by combining a mortality table with an interest rate table. For mortality tables I have the 2012 IAM with G2, Social Security AS 120, and an AS 120 table with modified qx values to match a given life expectancy. For interest rates yield curves I have U.S. Treasury TIPS, U.S. Treasury nominals, and U.S. High Quality Markets corporate bonds.
The Python source code is available under the GNU Affero license for anyone interested in inspecting the calculator for any flaws.
I am finding the SPIAs quotes I have seen are quite competitive when measured against these yield curves.
One issue that I don't feel I have fully resolved is applying the 2005-08 Actual/Expected contract duration data. Without any adjustment the 2012 IAM mortality of long time and recent annuitant is seen as the same, when in reality this is not the case. Applying the full contract duration a/e table might seems the best option, but the table is noisy and incomplete. The all ages summary is less noisy and is complete, but information is lost in using it.
I'm not an actuary, so I hope to learn from any feedback you might be able to provide.
SPIA pricing calculator and observations
