How long has the investigation been running?
A sub-committee of the CMI was set up in 1970 to investigate sickness rates under permanent health insurance policies. The first data collected and analysed was for the years 1972 and 1973 with the results being published in CMI Report 2.
Since then, the CMI has collected income protection data for each year for both group and individual business. However, data for years prior to 1975 was found to be unreliable (see CMI Report 7).
How do you collect/analyse data?
We collect census data, ie, in force data at 1st January each year and claims for which at least one payment has been made during the investigation year, from insurance companies, termed contributing offices. The data is in a per policy format. The full format is specified in the coding guide. More information about data submissions and for access to the latest coding guide is available on the CMI data page. The analysis methodology is described in detail in CMI Report 12 and CMI Report 15.
Who should I send data to or ask about the income protection investigation?
For all investigations the CMI use group email addresses in order to manage communications. Any general enquiries about the CMI should be directed to email@example.com. For correspondence about the income protection investigation, or to submit data, use firstname.lastname@example.org. Should you want speak to someone directly, please call the CMI on 020 7776 3820.
How are duplicate policies in the data allowed for?
Individuals may have more than one IP policy leading to multiple policy records that will all have the same claim experience. Inclusion of these duplicates increases the variance of the number of claims (see CMI Report 7). Duplicate claim records are identified as far as possible by matching records by the following data fields:
- Record year
- Age definition
- Deferred period
- Birth month and year
- Date of sickness (day, month and year)
These identified duplicates are removed in analyses of claim termination experience. It is not possible to identify duplicates in the in force data due to the lack of fields capable of differentiating between different policies. Therefore, in analyses of claims inceptions, duplicates are not removed in order to maintain consistency between the claims and in force datasets. The variance is adjusted as described in CMI Report 7.
What are the differences between aggregate, standard and standard* experiences?
The total collected data is referred to as the aggregate data. The data should not include medically substandard lives. Historically such lives have been requested to be excluded from data submissions. However, more recent versions of the coding guide, from version 3.0 onwards, request that these are included and identified as medically substandard.
The main analysis from the 1975-78 quadrennium was carried out on a subset of the aggregate data known as the standard data. This consists of UK policies with no special benefit types (e.g. lump sums), no identifiable underwriting exclusions and no occupational rating. The occupational rating field within the data has been used from the start of the investigation and has two values, rated or not rated. Records where the occupational rating is unknown are excluded from the data (see CMI Report 7).
To make use of the occupational class information, collected with effect from 1991, a new subset of the Aggregate data was defined and named the standard* data. This uses the same criteria as the Standard data but ignores the contents of the occupational rating field. It therefore represents a larger subset than the Standard data (the Standard data is itself a subset of the standard* data), and consists of UK policies with no special benefit types and no identifiable underwriting exclusions (see CMI Report 18).
What is the comparison basis used in the reports?
The actual claims experience is compared against that expected using the graduated rates for the standard male lives experience on individual IP policies for 1975-78. These graduations were presented in CMI Report 12 and are referred to as SM1975-78.
Although the methodology used in these comparisons are largely as set out in CMI Report 12, there have been a few revisions made which are set out in CMI Report 15. The most important of these changes concerns the treatment of non-reported claims in the analysis of claim inception experience.
If you have any questions about the CMI please email