Important note regarding the exposure data used in recent CMI work on mortality projections

Issued November 2007

In August 2007 the CMI became aware that the dataset for Permanent Assurances for Male lives used in recent research into mortality projections included initial exposures but was used as if these were central exposures. As a result, the results published in Working Papers 20, 25 and 27 and the projections in the draft library that used this dataset were inaccurate.

The issue was widely publicised, together with an initial indication of its impact on annuity values. During the re-work the CMI found an error in the projection software in the calculation of deviance residuals used in boot-strapping simulations for the Lee-Carter models which also affects some of the values in Working Paper 25. Registered users of the CMI's illustrative software were also advised of this issue.

This dataset has also been made available to the Profession, together with the illustrative mortality projections software, without specifying what definition had been used for the exposures. As this software fits mu values, users could reasonably expect the dataset to include Central Exposures but it actually contains Initial Exposures. Therefore, the dataset needed to be adjusted before being used with the software. Central Exposures can be estimated for each age and year from the Initial Exposures using the following formulae:

Central Exposure = Initial Exposure - 0.5 x Deaths.

More details on these two issues and corrections to the parts of Working Papers 20, 25 and 27 affected by these issues is contained in a document entitled "Errata to CMI Working Papers 20, 25 and 27 on Mortality Projections". To access this document, please click here.

In addition:

• The versions of Working Papers 20 and 25 on the website have been replaced with revised versions.
• Working Paper 27 has not been revised so that the document on which the CMI sought the views of the Profession remains as it was for consultation; although its content is effectively replaced by Working Paper 30 and the User Guide being issued alongside version 1.0 of the library.
• A new release of the software is being sent to all registered users to correct the calculation of the deviance residuals.

Back to the CMI Home page