Driver Negligence vs. Odometer Miles

Download Report

Transcript Driver Negligence vs. Odometer Miles

Why Low Credit Scores
Predict More Auto Liability
Claims: Two Theories
Patrick Butler*
American Risk & Insurance Association
August 7, 2007
*National Organization for Women
[email protected]
1
#774.7804
Policy dilemma: get “A” w/o getting “B”
A. Mandatory liability insurance
– Public demands it
– Insurers oppose it because of “B”
B. Price regulation
– To keep mandatory insurance affordable
– Recent example is initiatives to ban or regulate
credit score (CS) pricing
– In response auto insurers commissioned a
large study completed in 2003
2
Figure 1. Liability Claims vs. Credit Scores
(from Miller & Smith 2003)
+90%
-25%
 Claims per 100 car years: +90% to -25% of average
 1.90 / 0.75 = 2.5 times
3
Theory 1
“More driver negligence”
• Basis for theory
– Each liability claim requires a negligent
act by insured car’s driver
• Logic
– Cars of low CS drivers average more
liability claims
– Therefore these drivers are more
negligent
4
Theory 1 (continued)
Explanations
• Miller & Smith 2003 (insurance industry
explanation)
“[Credit-based] scores seem to provide an objective means
of measuring personal responsibility and its effect on
insurance losses.”
• Brockett et al. 2005 (1st academic study)
– Presented at WRIEC, Salt Lake City
– Published 2007
5
Theory 1 (continued)
biological explanation
Brockett & Golden 2007 (J. Risk & Insurance)
• Title (underline added):
“Biological and Psychobehavioral Correlates of Credit
Scores and Automobile Insurance Losses: Toward an
Explication of Why Credit Scoring Works”
• Conclude that the research examined by their study
“suggests that the discussed individualized biological and
psychobehavioral correlates provide a connection between
credit scores and automobile insurance losses.“
6
Model for Theory 1 is the biological explanation for
correlation with driver age
Involvement Rate =
Accidents per 1,000,000 miles
(from Williams 1999)
Driver Age
 Similar age effects confirmed per worker hour of exposure
7
Theory 1
Unaddressed Matters
• Claims & accidents are referenced to driver- and caryears, but individual annual-miles exposures vary widely
• A conflict with risk aversion theory
– Greater financial constraint predicts more risk
aversion, i.e., less negligence
• Insurers report that lower CS also predicts more
uninsured motorist (UM) claims.
– But a UM claim requires non negligence by the
insured car’s driver
8
CS Theory 2
“More miles per insured car year”
• UM & Liability Claims must correlate (+):
– The more miles a group of cars averages, the more
claims per 100 car years the group produces
• more negligence claims, and also
• more non-negligence (UM) claims
• This means that the cars of financiallyconstrained drivers must be averaging
more miles. Why so?
• But first some basics:
9
Basics for Theory 2
• Given – Accidents are a cost of car
operating
• Given – Premiums are a cost of car
owning
• But is the range in annual miles enough to
explain the 2.5 times CS variable range in
claims per 100 car years?
10
Fig. 2. Yes. Household Cars Distributed by
Odometer Miles Shows Range
20%
18%
Mean = 11,801 miles
Median = 9,448 miles
N = 32,153 cars
N = 32,153 cars
16%
WOMEN DRIVERS
AVG = 10,143 MILES
14%
MEN DRIVERS
AVG = 16,553 MILES
12%
10%
Cars
8%
6%
4%
2%
Annual Miles (000)
>30-32
>28-30
>26-28
>24-26
>22-24
>20-22
>18-20
>16-18
>14-16
>12-14
>8-10
>6-8
>10-12
1.0% = 0
miles
>4-6
>2-4
>0-2
0%
4.4% > 32K miles
• Subgroups by car age, and by driver sex and age (despite
difference in averages) span entire annual-miles range
• So cars within insurance classes span low- to high-miles 11
Figure 3 (Recasting of Figure 2)
 The CS range in claims: 2.5 times = 15,000 mi / 6,000 mi
12
Figure 4. Why miles must be individually measured
1
14%
> 15,000 mi.
0.9
0.8
10+ years old
8,758 mi. avg*
0.7
from 1995 NPTS
Quantiles
0.6
0.5
0-2 years old
16,092 mi. avg*
0.4
0.3
0.2
0.1
* Owners' est'd
annual miles
13%
< 6,000 mi.
0
0
5
10
15
20
Annualized Odometer Miles -
25
30
Thousands
 Avg miles explain why new cars average more liability claims than old cars
 But distribution shows many new cars are driven fewer miles than old cars
13
Logic of Theory 2 illustrated
(Hard but sure way to economize)
14
Theory 2 fits other predictors
1.
2.
3.
4.
Zip code income (-)
Education & occupation (-)
No prior insurance (+)
Installment plan (+)
X. Any marker of tight budgets predicts more
liability claims per 100 car years
Therefore, the highest per-car premiums are
charged to those who can least afford them 15
Theory 2 recommends
• Informed by Theory 2, the strong public demand for
enforcing mandatory liability insurance could be
accompanied by
– A strong demand that automobile insurers provide the
audited odometer-mile exposure unit as an option
– At cents-per-odometer-mile class prices this option
would constitute a free-market remedy for the upward
cost-price spiral that the traditional car-year exposure
unit sets off for groups of economizing drivers
– With this option drivers could car pool or take the bus to
save on insurance while keeping their own cars insured
and available for use
16
Table 2. What’s at stake – the challenge for both theories
e. g., Albany, NY: lowest and highest quoted for one car class profile
(same driver age, record, sex, and marital status).
Insurance Company
Premium
ALLSTATE PROP & CAS INS CO
258
GEICO
318
PROGRESSIVE NORTHEASTERN
326
STATE FARM MUT AUTO INS CO
375
GEICO IND CO
492
METROPOLITAN GRP P&C INS CO
641
AUTOONE INS CO
854
ALLSTATE IND CO
1,136
Range: $1,136 / $258 = 4.4 times
17
Conclusion – Research question
Which policy response to the dilemma
“free-market vs. affordability vs. mandatory insurance”
• Theory 1, some issues
– Identify the negligent driver groups on an
accidents-per-1,000,000-mile basis
– Find incentives to reduce negligence per-mile
• Theory 2, some issues (discuss in Q & A)
– Federal surveys show average miles per car year
decreases moderately as household income decreases
– Some high premiums for adult-driver-class cars may
reflect not only higher miles per car, but also the higher
per 1,000,000 mile accident involvement rates of
“undisclosed” young and old drivers sharing the cars
18