A
Actuary_11
Member
In the topic of Hazard rate in Measures of Tail weight in Chapter 16 : Extreme Value Theory
Core Reading says that if the force of mortality increases as a person’s age increases, relatively few people will live to old age (corresponding to a light tail). If, on the other hand, the force of mortality decreases as the person’s age increases, there is the potential to live to a very old age (corresponding to a heavier tail).
So this means that when the hazard rate is decreasing function of x , it has the heavier tail .
But in section 4.3 , page no 29 :-
If alpha < 1, the hazard rate is a decreasing function of x , which tends to lambda from above for large x . This indicates a lighter tail than the exponential distribution with parameter lambda above any fixed threshold x , but the same tail thickness in the limit.
If alpha > 1, the hazard rate is an increasing function of x , which tends to lambda from below for large x . This indicates a heavier tail than the exponential distribution with parameter lambda above any fixed threshold x , but the same tail thickness in the limit.
From the above explanation it means that when we are having decreasing hazard rate , it has the lighter tail , which contradicts the explanation given above .
What is the explanation for the tail weights from the hazard rate function of that distribution ..? Like it will have a lighter tail or heavy tail when the hazard rate is the decreasing function of x and vice versa.
Core Reading says that if the force of mortality increases as a person’s age increases, relatively few people will live to old age (corresponding to a light tail). If, on the other hand, the force of mortality decreases as the person’s age increases, there is the potential to live to a very old age (corresponding to a heavier tail).
So this means that when the hazard rate is decreasing function of x , it has the heavier tail .
But in section 4.3 , page no 29 :-
If alpha < 1, the hazard rate is a decreasing function of x , which tends to lambda from above for large x . This indicates a lighter tail than the exponential distribution with parameter lambda above any fixed threshold x , but the same tail thickness in the limit.
If alpha > 1, the hazard rate is an increasing function of x , which tends to lambda from below for large x . This indicates a heavier tail than the exponential distribution with parameter lambda above any fixed threshold x , but the same tail thickness in the limit.
From the above explanation it means that when we are having decreasing hazard rate , it has the lighter tail , which contradicts the explanation given above .
What is the explanation for the tail weights from the hazard rate function of that distribution ..? Like it will have a lighter tail or heavy tail when the hazard rate is the decreasing function of x and vice versa.