UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN

Actuarial Science Program

DEPARTMENT OF MATHEMATICS

 

Math 478 / 568

Actuarial Modeling

 

 

 

 

Rick Gorvett

Spring, 2012

 

374 Altgeld Hall

 

 

Phone: 244-1739

213 Gregory Hall

9:30-10:50 am TuTh

 

Office Hours:

3-4 pm Tues, 3-4 pm Wed,

or by appointment

Class Website:

 

E-mail: gorvett@illinois.edu

http://www.math.uiuc.edu/~gorvett/m478s12/home.html

 

 

 

Required Texts

 

Loss Models: From Data to Decisions, by Klugman, Panjer, and Willmot, 3rd edition

 

Chapter 8, Credibility, Foundations of Casualty Actuarial Science, Fourth Edition, 2001, Casualty Actuarial Society (FCAS) http://www.soa.org/files/pdf/C-21-01.pdf

 

Topics in Credibility Theory http://www.soa.org/files/pdf/c-24-05.pdf

 

 

Course Overview

 

In the course of her/his work, an actuary frequently must develop mathematical models to describe the insurance loss process. Such modeling is performed under a variety of conditions: with large or small amounts of data, for different types of insurance coverages, etc. This course is intended to introduce the student to the intuitive underpinnings, the mathematical foundations, and the real-world applications of actuarial modeling theory.

 

This course is essentially divided into thirds:

 

Topic

Anticipated Timing

Loss modeling

First third of course

Model selection and parameter estimation

Second third of course

Credibility theory and simulation

Final third of course

 

In the Loss modeling section, we will study how to mathematically model the severity (size), the frequency (number), and the aggregate value (total dollar amount) of claims sustained by the policyholders of an insurance company. Such mathematical models are critical for developing appropriate prices for insurance policies, and for determining the level of loss reserve liabilities to be entered on the financial statements of an insurer. This section will also address risk measures. The Estimation section will involve the determination and estimation of actuarial models and their parameters based on data. The Credibility section will examine how to determine the relative weights to be given to various sources of data in calculating premium rates and loss reserve levels. All of these three general topics are essential to the effective performance of the actuarial function, and are found on professional actuarial Exam 4/C.

 

 

Accessibility Statement

 

To ensure that disability-related concerns are properly addressed throughout the semester, students with disabilities who require reasonable accommodations to participate in this class are asked to contact me within the first two weeks of class.

 

 

Honors Projects

 

For those involved in a Campus or College honors program, such as James Scholar, you are welcomed to perform an honors project in association with this class. Such a project would involve an opportunity for undergraduate research relevant to actuarial science. Please let Prof. Gorvett know of your interest in the first few weeks of the class.

 

 

Course Schedule

 

Some key dates for the class are noted below:

 

January 17 (Tuesday)

First day of class

February 16 (Thursday)

Exam # 1

March 15 (Thursday)

Exam # 2

May 1 (Tuesday)

Last day of class

May 9 (Wednesday, 1:30 pm)

Exam # 3 (Final Exam)

 

Practicing actuaries may be invited to give guest presentations to the class on several occasions during the semester.

 

 

Grading

 

Course grades will be determined based on the following weights:

 

Homework and other assignments

25 %

Project

10 %

Exams 1 and 2

40 % (20% each)

Final Exam

25 %

 

Exams may not be made up or taken at different times, except in extremis. In general, the grade weight for a missed exam (provided there is a valid excuse) will accrue equally to any remaining exams. Homework assignments will be approximately weekly, and are due at the beginning of the class on the due date. Late homework assignments will not be accepted. A missed homework assignment may not be made up. However, in calculating your course score and grade, your lowest homework assignment score will be dropped. Class attendance is expected; serious attendance problems may result in a grade lower than that indicated by the weighting system above.

 

Please note that the final exam schedule is prescribed by the university; in general, instructors are not permitted to change the final exam timing of their courses.

 

Graduate students taking this class (Math 568) for four hours of credit will be required to complete significant additional work. Performance on this additional work will impact the overall grade of a graduate student.

 

This class employs a zero-tolerance approach to cheating. In the event that a student cheats (cheating includes, but is not limited to, giving or receiving aid on an exam, or participating in the submission of an in-class assignment by one student on behalf of another student) that student will be penalized in accordance with the 2011-12 UIUC Student Code (particularly, Article I, Section 403, Penalties for Infractions of Academic Integrity).

 

 

Other Important Information

 

The topic list for CAS/SOA Exam 4/C can be found at

http://www.beanactuary.org/exams/preliminary/exams/syllabi/2012-05-exam-c.pdf

 

 

Class Summaries

 

Date

General Topic

Specific Class Topics

Assignments

Textbook Reading

Other Items / Links

Jan. 17

Introduction and motivation.

Review of syllabus.

Motivation via modeling examples

 

Ch. 1

Exam 4/C Syllabus

Day 1 Slides

Jan. 19

Random variables

Conditional probabilities and Bayes Theorem.

Survival functions.

Hazard rates.

HW#1

Ch. 2

Textbook Exercises Ch 2

Jan. 24

Statistical quantities

Moments.

Truncation and censoring.

Deductibles and limits.

Per loss and per payment.

 

Ch. 3.1 to 3.2

Textbook Exercises Ch 3

Jan. 26

Losses and distributions

Limited expected values.

Mean excess loss.

Per loss and per payment random variables.

HW#2

Ch. 8.1 to 8.3

HW#1 Solutions

Sample Exam C Problems 1

Jan. 31

Losses and distributions (cont.)

Allocating losses to parties.

Using limited expected values.

Loss elimination ratio.

 

Ch. 8.3 to 8.5

Sample Exam C Problems 2

Textbook Exercises Ch 8

Feb. 2

Losses and distributions (cont.)

Parametric and scale distributions.

New distributions from old.

Mixing distributions.

Frequency distributions.

HW#3

Ch. 4

Ch. 5.1 to 5.2

Ch. 6.1 to 6.5

HW#2 Solutions

Sample Exam C Problems 3

Textbook Exercises Ch 4

Textbook Exercises Ch 5

Feb. 7

Aggregate loss models

Combining frequency and severity.

Convolutions.

Moments of aggregate distributions.

 

Ch. 9.1 to 9.5

Sample Exam C Problems 4

Textbook Exercises Ch 9

Feb. 9

Aggregate loss models (cont.)

Stop-loss insurance.

(a,b,0) and (a,b,1) distributions.

 

 

HW#3 Solutions

Sample Exam C Problems 5

Feb. 14

Review for Exam # 1

 

 

 

Exam 1 from Last Year

Feb. 16

Exam # 1

114 David Kinley Hall.

Regular class time.

3x5-inch notecard allowed.

 

 

 

Feb. 21

Empirical models

Handed back and discussed exam.

Severity inflation and leveraged effect.

Types of risks in modeling.

 

 

Exam 1 Results

Exam 1 Solutions

Feb. 23

Empirical models (cont.)

Quality of estimators.

Point and interval estimation.

Hypothesis testing.

HW#4

Ch. 12

Sample Exam C Problems 6

Feb. 28

Empirical models (cont.)

Complete vs incomplete data.

Individual vs group data.

Complete data analytics.

 

Ch. 13

Sample Exam C Problems 7

Mar. 1

Empirical models (cont.)

Modified data analytics.

Kaplan Meier.

Nelson Aalen.

Kernel models.

HW#5

Ch. 14

HW#4 Solutions

Sample Exam C Problems 8

Mar. 6

Parameter estimation

Method of moments.

Percentile matching.

Maximum likelihood.

 

Ch. 15

Textbook Exercises Chs 12 to 15

Sample Exam C Problems 9

Sample Exam C Problems 10

Mar. 8

Parameter estimation (cont.)

MM and PM problems.

MLE with modified data.

 

Ch. 15

HW#5 Solutions

Sample Exam C Problems 11

Sample Exam C Problems 12

Mar. 13

Parameter estimation (cont.)

Asymptotic variance.

Information.

 

 

Exam 2 from a Prior Year

(Answers to be provided in class on March 13.)

Sample Exam C Problems 13

Mar. 15

Exam # 2

141 Wohlers Hall.

Regular class time.

3x5-inch notecard allowed.

 

 

 

Mar. 27

Model Selection

Handed back and discussed Exam # 2.

Graphical model evaluation techniques.

 

Ch. 16

Exam 2 Results

Exam 2 Solutions

Mar. 29

Model Selection (cont.)

Hypothesis tests: KS, AD, Chi square, likelihood ratio.

HW#6

Ch. 16

Sample Exam C Problems 14

Apr. 3

Model Selection (cont.)

Finish up hyp tests.

Ranking models.

VaR and TVaR.

 

Ch. 16, Ch. 3

Sample Exam C Problems 15

Apr. 5

Credibility

Motivation.

Limited fluctuation or classical credibility.

 

Ch. 20

HW#6 Solutions

Apr. 10

Credibility (cont.)

Classical (cont.).

Class project description and data have been distributed to teams.

Ch. 20

Sample Exam C Problems 16

Apr. 12

Credibility (cont.)

Buhlmann credibility

 

Ch. 20

 

Apr. 17

Credibility (cont.)

Buhlmann credibility (cont.)

 

Ch. 20

Sample Exam C Problems 17

Apr. 19

Credibility (cont.)

Bayesian credibility

HW#7

Ch. 20

 

Apr. 24

Credibility (cont.)

Bayesian credibility (cont.)

 

Ch. 20

Sample Exam C Problems 18

Textbook Exercises Ch 20a

Textbook Exercises Ch 20b

Apr. 26

Credibility (cont.)

Buhlmann vs Bayesian credibility.

Buhlmann-Straub credibility.

 

Ch. 20

HW#7 Solutions

Textbook Exercises Ch 20c

May 1

Credibility (cont.)

Empirical Bayes estimation.

 

Ch. 20

Exam 3 from a Prior Year

(Answers to be provided in class on May 1.)

Sample Exam C Problems 19

Sample Exam C Problems 20

May 9

Final Exam

213 Gregory Hall.

1:30 pm start time.

3x5-inch notecard allowed.