28
Biological and Psychobehavioral Correlates of Risk Taking, Credit Scores, and Automobile Insurance Losses: Toward an Explication of Why Credit Scoring Works Patrick Brockett ([email protected] ), and Linda Golden ( [email protected] ) Presentation to the Casualty Actuarial Society Predictive Modeling Conference on October 11, 2007, Las Vegas, Nevada

Presentation to the Casualty Actuarial Society Predictive Modeling

Embed Size (px)

DESCRIPTION

Biological and Psychobehavioral Correlates of Risk Taking, Credit Scores, and Automobile Insurance Losses: Toward an Explication of Why Credit Scoring Works Patrick Brockett ( [email protected] ), and Linda Golden ( [email protected] ). - PowerPoint PPT Presentation

Citation preview

Page 1: Presentation to the Casualty Actuarial Society Predictive Modeling

Biological and Psychobehavioral Correlates of Risk Taking, Credit Scores, and Automobile

Insurance Losses: Toward an Explication of Why Credit Scoring Works

Patrick Brockett([email protected]),

andLinda Golden

([email protected])

Presentation to the Casualty Actuarial Society Predictive Modeling Conference on October 11, 2007, Las Vegas, Nevada

Page 2: Presentation to the Casualty Actuarial Society Predictive Modeling

Reference for details:

Brockett, Patrick L. and Linda L. Golden “Biological and Psychobehavioral Correlates of Risk Taking, Credit Scores, and Automobile Insurance Losses: Toward an Explication of Why Credit Scoring Works,” Journal of Risk and Insurance, Vol 74(1), March 2007. 23-63.

Available electronically from JSTOR, Blackwell Publishing, accessible from www.ARIA.org, or by emailing the authors

75 Copies available at the meeting

Page 3: Presentation to the Casualty Actuarial Society Predictive Modeling

The most important development in the past two decades in personal lines of insurance may well be the use of an

individual’s credit history as a classification and rating variable to

predict losses.

Page 4: Presentation to the Casualty Actuarial Society Predictive Modeling

Empirical Relationship Demonstrated

The statistical evidence between insured losses and credit score has been repeatedly demonstrated.

Very strong correlation between a bad credit score and increased insurance losses.

Research Examples. . . . . .

Page 5: Presentation to the Casualty Actuarial Society Predictive Modeling

Chart 6Average Incurred Losses Within Each Decile for Policies Grouped by Credit Score Decile

$668

$918

$846$791

$707 $703 $681$631

$584 $568 $558

-

100

200

300

400

500

600

700

800

900

1,000

No creditscore

available

1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th

------ Credit score decile ------

/\/\/\/\

Incurred loss (dollars)

1st credit score decile = lowest credit score

10th credit score decile = highest credit score

Excerpted from University of Texas study conducted for Texas legislature, 2003

Page 6: Presentation to the Casualty Actuarial Society Predictive Modeling

Chart 5Average Relative Loss Ratios By Credit Scores

for Standard Market Data Set

1.07

1.53

1.28

1.061.00 0.99

0.88 0.840.78

0.72 0.76

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

No creditscore

available

1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th

Avg. relative loss ratio

------ Credit score decile ------

1st credit score decile = lowest credit score

10th credit score decile = highest credit score

/\/\/\/\

Excerpted from University of Texas study conducted for Texas legislature, 2003

Page 7: Presentation to the Casualty Actuarial Society Predictive Modeling

Tillman and Hobbs (1949): drivers with bad credit history have repeated crashes at a rate six times higher than those with good credit history.

Washington state study (1968): within the group with a history of no automobile accidents, 64% had good credit and 35% had bad credit-- among group with two or more automobile accidents, 35% had bad credit, -- almost twelve times the percentage (3%) who had good credit

Other correlates: divorce, legal problems, job turnover, lower education

Page 8: Presentation to the Casualty Actuarial Society Predictive Modeling

“…a man drives as he lives.”Tillman and Hobbs, 1949

Research Results Summarized

Page 9: Presentation to the Casualty Actuarial Society Predictive Modeling

The purpose of this research is to present a “missing link” explaining why credit scores are associated with insurance losses.

The outcome of the debate over the use of credit scoring has implications for the social acceptability of Actuarial Standard #12, and has implications for other variables useful for underwriting.

Page 10: Presentation to the Casualty Actuarial Society Predictive Modeling

Heuristic Model

Insured Loss = f(X1,X2) Credit Score = g(Y1,X2)

Where: X1 denotes a vector of automobile specific characteristics,

X2 denotes a vector of person specific psychological (and possibly biological) characteristics, and

Y1 denotes a vector of credit specific attributes

Proposition: The correlation between Insured Losses and Credit Score is high and positive because of the common vector factor X2 (which is in turn correlated with both X1 and Y1 ).

Page 11: Presentation to the Casualty Actuarial Society Predictive Modeling

Risk TakingBehavior(Driving)

Risk TakingBehavior

(Financial)

CreditScore

InsuredAuto

Losses

Bio-chemicalPsycho-

behavioral

Profile

Simplified Model of Conjunctive Influences between Insured Losses and Credit

Page 12: Presentation to the Casualty Actuarial Society Predictive Modeling

The Core Idea Connector between risk taking

behavior in automobile insurance losses and credit scores and financial risk taking is the psychological dimension.

Most easily identified psychological characteristic is the personality type known as “sensation seeking” or “novelty seeking.” It is related to responsibility and risk taking.

Page 13: Presentation to the Casualty Actuarial Society Predictive Modeling

Psychobehavioral Profile of Sensation Seeking/Novelty Seeking

Risky Behaviors

Risky Driving High Risk Occupations

Drinking/ Driving

High Risk Sports

Drinking/ Drug Use

Reduced Personal Responsibility

Overestimation of SkillsIncreased Perceived Benefits

Reduced Perceived RiskReduced Deliberation

1These terms are often used interchangeably in the literature. The “sensation seeking” term comes from Zuckerman (1979) and “novelty seeking” is attributable to Cloninger (1987).

Page 14: Presentation to the Casualty Actuarial Society Predictive Modeling

“If serotonin is the brakes, dopamine is the accelerator in the drive to risky

behavior.”

Zuckerman and Kuhlman, 2000Zuckerman and Kuhlman, 2000

A Biological Component

Page 15: Presentation to the Casualty Actuarial Society Predictive Modeling

Biochemical and Psychobehavioral Profile of Sensation Seeking/Novelty Seeking

Stress

Antisocial

Behavior

Depression

Exploration

Arousal

Amplifies reaction to

stimuli

Impulsivity

SENSATION SEEKING

Low Levels of Cortisol

Low Levels of Serotonin

High Levels of

Testosterone

High Levels of Dopamine

High Levels of Norepinephrin

e

Low Levels of MAO-A

Corticosterone

Low Intellect

Low Education

Low Occupational Status

High SES

Employment

Marriage

Biochemicals

Risk Taking Responses

Socio-cultural Outcomes

Mediating Factors

LEGEND

Low Levels of MAO-B

1These terms are often used interchangeably in the literature. The “sensation seeking” terms comes from Zuckerman (1979) and “novelty seeking” is attributable to Cloninger (1987).

Page 16: Presentation to the Casualty Actuarial Society Predictive Modeling

Influences on sensation seeking and novelty seeking have implications for

automobile insurance losses.

Page 17: Presentation to the Casualty Actuarial Society Predictive Modeling

Comprehensive Overview of Biochemical and Psychobehavioral Influences Related to Paid Automobile Insurances Losses

Testosterone

Corticosterone

Monoamine Oxidase

Cortisol

Serotonin

Norepinephrine

Dopamine

Risk Appraisal Judgments

Risk Perceptions Judgments

Sensation Seeking/Novelt

y Seeking

Impulsive Driving

Decisions

Inattentive to Details or

Environment

Irresponsibility Regarding

Driving Behavior

Other High Risk- Taking

Behaviors

Distractibility/ Lack of Focus

Aggressive/ Antisocial Behavior

3rd Party at Fault Accident

At Fault Accident

Accident Caused by Act

of God

Potential Biochemical Influeners

Driver Psycho-behavioral Profile

Risky Driving Behavior

Accident Characteristics

Actual Loss to Insured

Vehicle Characteristics

Insured’s LossMitigation Activities

Reported Loss to Insurer

Insured’s Possible

Claim Size Build-Up

Insured’s Reporting Decision

Prior Policy Limits &

Policy Coverage Decisions by Insured

Actual Paid

Insurance Losses

Prior Deductible Choice by

Insured

Driver Psychological and Economic Profile Influences

Post-Accident Decisions and Influences on Loss Amount

Loss Incurred by Insurer

Biochemical Psycho-behavioral System Feedback

Age, Gender, Marital Status, Education, SES, Rural/Urban/Inner City Dweller

Driver Characteristics & Demographics

Page 18: Presentation to the Casualty Actuarial Society Predictive Modeling

Financial decision making is also

related to psychobehavioral and

biochemical variables.

Page 19: Presentation to the Casualty Actuarial Society Predictive Modeling

Brown and Harlow (1990) examined blood samples and determined that financial risk taking is related to blood chemistry.

Other research has shown sensation seeking/novelty seeking is related to financial decision making……….

Page 20: Presentation to the Casualty Actuarial Society Predictive Modeling

Reduced risk perception and risk appraisal play an important role in the individual’s propensity for sensation seeking which, in turn, is an integral part of the individual’s financial decision making.

Risk tolerance is evident in both the filing of insurance claims

and excessive credit card use (impulse buying which may be linked to MAO and dopamine or financial stress linked to serotonin, cortisol, dopamine, and norepinephrine).

Debt and poor money management create and are the result of

financial stress which may be linked to serotonin, cortisol, dopamine, and norepinephrine.

Each of these decisions directly impacts the individual’s credit score which is often used as a variable in predicting losses in automobile insurance coverage.

Miraplex and chemically induced risk taking

Page 21: Presentation to the Casualty Actuarial Society Predictive Modeling

…and financial decision making determines, in part, a person’s credit score…

Page 22: Presentation to the Casualty Actuarial Society Predictive Modeling

Comprehensive Overview of Biochemical and Psychobehavioral Influences Related to Credit Score

Monoamine Oxidase

Cortisol

Testosterone

Serotonin

Norepinephrine

Dopamine

Corticosterone

Risk Appraisal Judgments

Risk Perception Judgments

Sensation / Seeking Novelty Seeking

Divorce

Medical Exigency

Impulsive Financial/Purc

hase Decisions

Inattentive to Details or

Environment

Distractable/Unable to

Focus

Irresponsible Regarding Financial or

Credit Obligations

Total Credit Card Debt to Credit Line

Ratio

Defaults on Debts or Derogatory Public

Records

Length of Credit Record

Missed Payment History

Late Payment History

Number of Credit Lines Open

Credit Inquiries in Past 30 Days

Credit Score

Unemployment

Potential Biochemical Influencers

Psycho-behavioral Profile Risky Financial/ Credit Behavior

Economic Exigencies

Credit History Record

Page 23: Presentation to the Casualty Actuarial Society Predictive Modeling

Notice that: The same risk taking correlates show up across realms from driving to financial decision-making.

Why?

Page 24: Presentation to the Casualty Actuarial Society Predictive Modeling

Possible Theoretical Explanations

Risk Homeostasis Theory: all behaviors hold some level of risk and the challenge of driving is to maximize the overall benefits of the behavior. The driver learns to adjust behaviors when a discrepancy is observed

between the observed level of risk and the target level of risk.

(Burns and Wilde 1995; Wilde 2002)

Target Risk Theory: an adaptation of risk homeostasis that necessitates the adjustment of driving behavior so that

perceived risk is in line with target risk.(Wilde 2002)

Page 25: Presentation to the Casualty Actuarial Society Predictive Modeling

The biochemical mechanisms coupled with Wilde’s Homeostasis Theory suggests an intrinsic biological

mechanism at play in the relationship between risk taking and behavior

of all types.

Page 26: Presentation to the Casualty Actuarial Society Predictive Modeling

Irrespective of the viability oftheoretical explanations,

we can graphicallysummarize the

biochemical and behavioral commonalities between credit

scores and insuredloss generation. . . .

Page 27: Presentation to the Casualty Actuarial Society Predictive Modeling

Biological and Psychobehavioral Correlates of Risk Taking, Credit Scores, and Automobile Insurance Losses

Stress

Risky Driving Behavior Risky Financial/Credit Behavior

Antisocial Behavior

Depression

Exploration

Arousal

Amplifies reaction to

stimuli

Impulsive driving decisions

Inattention to details or the environment (road conditions, road

signs, traffic conditions)

Impulsive financial/purchase decisions

Inattention to details or the environment (interest rates, penalty

fees, payment due dates)

Credit Score

Credit History

Insurance Losses

Distractibility/ lack of focus Distractibility/lack of focus (no financial planning, no savings)

Irresponsibility regarding driving behavior (drinking, speeding, light/sign running, unsafe lane

changes

Irresponsibility regarding financial or credit obligations (extravagance,

overextended on credit cards)

Risk Appraisal Judgments

Risk Perception Judgments

Impulsivity

SENSATION SEEKING

Low Levels of Cortisol

Low Levels of Serotonin

High Levels of Testosterone

High Levels of Dopamine

High Levels of Norepinephrine

Low Levels of MAO-A

Corticosterone

Putting All The Relationships Together, We Have . . .

Page 28: Presentation to the Casualty Actuarial Society Predictive Modeling

Thank you very much for your attention.

Questions?

Comments?