An investigation of age-period-cohort models for individual-level data
When dealing with temporal effects, there is a well-known identification problem that arises as a result of the linear relationship between age, period, and cohort. Kuang, Nielsen, and Nielsen (Kuang et al., 2008) approach this problem in the context of aggregate data by examining estimable function...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
2016
|
| Subjects: |
| Summary: | When dealing with temporal effects, there is a well-known identification problem that arises as a result of the linear relationship between age, period, and cohort. Kuang, Nielsen, and Nielsen (Kuang et al., 2008) approach this problem in the context of aggregate data by examining estimable functions of the parameters of the standard model of temporal effects. These functions are both uniquely identified from data and invariant to transformations of the parameters of the original model. This thesis extends their approach to study the standard model applied to individual-level data. A new test of fit for models of inter-temporal variation in individual-level data is developed, along with an algorithm for the test's implementation. These extensions are demonstrated in an application to BMI data from the Health Survey for England from 1995-2013, which yields suggestive evidence that age and period effects are of primary importance in this data. |
|---|