Journal of Population Sciences
Table 1 Comparison of the underlying assumptions and properties of the CoDe 2.1 model and the LC model
Properties | Model | |
|---|---|---|
LC | CoDe 2.1 | |
Mathematical form | Models death rates and is linear | Models death probabilities and is non-linear |
Total number of parameters | Two parameters for each age, one parameter for each year, minus two parameters for identification | Six parameters to describe the age pattern plus four parameters for each year |
Modeling of the age pattern | More flexible to capture age-specific effects due to many age-dependent parameters | Fixed mathematical form to describe the age pattern |
Extrapolation to high ages | Additional assumptions are required, such as additional parameters and a two-step approach | Is part of the model. Death probabilities to higher ages are asymptotically equal to parameter g |
Estimating modal age | Limited, only integer values | Possible |
Modeling of the mortality trend | One parameter explains the mortality trends across all ages | For each age group, there is a parameter to explain the mortality trends |
Ability to distinguish between compression and delay | Cannot decompose | Decomposes the two effects and projects them |
Speed of mortality decline | Death rates decrease exponentially for all ages | Non-exponential decay for all age groups |
Incorporation of exogenous variables and expert opinion | Very limited | Possible |
Ability to use as diagnostic tool | Not possible | Possible |