Identifying ADL type definitions
We classified policies based on terms present in the standard and alternative definitions of total disability in the insurance information document offered by the fund. These were recorded in “standard definition” and “alternative definitions” fields in our data set for each policy offered in each year. The classification was done by first reading through the policies grouped by insurer to identify the text used by that insurer to describe ADL type definitions. We identified two main variants of ADL type definitions in the sample - “Activities of daily living” (ADL) and “Activities of daily work” (ADW). We then applied a function that looked for the text associated with ADL type definitions from a particular insurer and appropriately classified each policy as ADL or ADW, or both.
For example, a definition field in a particular policy containing words that described an ADL type definition would cause that policy to be coded as containing an ADL definition of TPD.
Identifying eligibility criteria
We classified policies based on terms present in the standard and alternative definitions of total disability in the insurance information document offered by the fund. These were recorded in “standard definition” and “alternative definitions” fields in our data set for each policy offered in each year. In order to identify terms used to determine the definition applied to a person making a claim, we split the sample by insurer and went through each of the insurers policies to identify patterns in how a particular eligibility criterion was worded. Once we had identified key phrases that could identify the application of a particular criteria, we used a function to assign a value to a particular policy if it contained a particular criterion.
For example, we identified that the term “working at least 15 hours” was associated with an hours based restriction on the TPD definition applied for a particular insurer and fund. We could then use a function to assign all policies offered by that insurer in the sample a value “15” to a variable representing hours restrictions in order to be able to quantify the number of policies with such a restriction.