01.08.2014 Views

Estimation, Evaluation, and Selection of Actuarial Models

Estimation, Evaluation, and Selection of Actuarial Models

Estimation, Evaluation, and Selection of Actuarial Models

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

12 CHAPTER 2. MODEL ESTIMATION<br />

Exercise 2 Construct the ogive <strong>and</strong> histogram for the following data set.<br />

Payment range Number <strong>of</strong> payments<br />

0—25 6<br />

25—50 24<br />

50—75 30<br />

75—100 31<br />

100—150 57<br />

150—250 80<br />

250—500 85<br />

500—1000 54<br />

1000—2000 15<br />

2000—4000 10<br />

over 4000 0<br />

2.2.4 Empirical distributions for censored <strong>and</strong> truncated data<br />

It is not unusual for data to be incomplete due to censoring or truncation. The formal definitions<br />

are as follows.<br />

Definition 2.15 An observation is truncated from below (also called left truncated) atd if,<br />

when it is below d it is not recorded, but when it is above d it is recorded at its observed value.<br />

An observation is truncated from above (also called right truncated) atu if, when it is above<br />

u it is not recorded, but when it is below u it is recorded at its observed value.<br />

An observation is censored from below (also called left censored) atd if, when it is below d, it<br />

is recorded as being equal to d, but when it is above d, it is recorded at its observed value.<br />

An observation is censored from above (also called right censored) atu if, when it is above u,<br />

it is recorded as being equal to u, butwhenitisbelowu, itisrecordedatitsobservedvalue.<br />

The most common occurrences are left truncation <strong>and</strong> right censoring. Left truncation occurs<br />

when an ordinary deductible <strong>of</strong> d is applied. When a policyholder has a loss below d, heorshe<br />

knows no benefits will be paid <strong>and</strong> so does not inform the insurer. When the loss is above d, the<br />

amount <strong>of</strong> the loss will be reported. A policy limit is an example <strong>of</strong> right censoring. When the<br />

amount <strong>of</strong> the loss exceeds u, benefits beyond that value are not paid <strong>and</strong> so the exact value is not<br />

recorded. However, it is known that a loss <strong>of</strong> at least u has occurred.<br />

When constructing a mortality table, it is impractical to follow people from birth to death. It<br />

is more common to follow a group <strong>of</strong> people <strong>of</strong> varying ages for a few years. When a person joins<br />

a study, he or she is alive at that time. This person’s age at death must be at least as great as the<br />

age at entry to the study, <strong>and</strong> thus has been left truncated. If the person is alive when the study<br />

ends, right censoring has occurred. The person’s age at death is not known, but it is known that it<br />

is at least as large as the age when the study ended. Right censoring also affects those who leave<br />

the study prior to its end due to surrender.<br />

Because left truncation <strong>and</strong> right censoring are the most common occurrences in actuarial work,<br />

they are the only cases that will be covered in this Note. To save words, “truncated” will always<br />

mean truncated from below <strong>and</strong> “censored” will always mean censored from above.<br />

¤

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!