Using Pricing to Motivate Workplace Safety

1
Using Pricing to Motivate Workplace Safety

• International Forum on Disability Management
• October 2006
• Presenter Gregory Krohm
• IAIABC
• Research in collaboration with
• Mike Barth,
• Georgia Southern University
• Bob Klein,
• Georgia State University

2
Overview

• Experience rating responds to philosophical and
  practical values
• The effectiveness of experience rating has been
  in dispute
• New statistical evidence shows that it works to
  reduce claims!
• But is the reduction in claims due to
• Claims management?
• Making work safer?

3
Does it All Boil Down to Economics?

• Economic theory says
• Firms would be expected to invest in
  safety up to point where marginal cost of safety
  investment marginal benefits of safety
• But are employers
• able to self assess their safety practices and
• do they rationally react to economic
  incentives????

4
Rate Making Generally

• Workers compensation has one of the most complex
  and data intensive systems of rate making in any
  insurance system
• Particularly in Australia, USA and countries that
  use private insurance systems
• Public and private insurers track losses and
  premiums for each covered employer
• Ultimate incurred losses are very difficult to
  estimate for rate making because it takes so long
  for the ultimate costs of some claims to be known
  
• Rate making systems are very data/computer/reporti
  ng intensive
• Hence, they are not suitable for developing
  countries or countries starting radical new forms
  of insurance

5
Theory of Experience Rating

• Class rates are imperfect measures of loss
  expectations of an individual employer
• An employers own loss experience should be used
  to modify class rates
• The weight given to an employers own experience
  should be driven by statistical credibility
• Only 5-25 of all employers are good candidates
  for experience rating

6
Methods and Basis for the Formula

• Worldwide, experience formulas are VERY diverse
  in structure
• The actuarial foundation is quite sophisticated
  in the USA
• A uniform model (based on National Council on
  Compensation Insurance) is used in most states
• A primary goal is to levelize loss ratios after
  the application of experience rating
• Not necessarily to induce safety

7
Is the Actuarial Goal Met?

• In the USA, the modification factors do a
  surprisingly good job of leveling out LR costs to
  insurers across employers in the same industry
• Before the application of the mod all employers
  in the same class pay the same rate
• After the mod, premiums produce a loss ratio
  similar for all employers

8
Do Employers Understand Experience Rating?

• Agents take pains to sell their expertise in
  helping clients manage their X-mod
• The most popular risk management software has
  built in reports to help manage the mod factor
• Industry oriented articles try to educate
  employers about the importance of mod factors
• The mod has become a common litmus test for the
  quality/reliability of a potential subcontractor
• Mods above 1 can disqualify firms from working
  for a general contractor

9
Popular Risk Management Software Allows Employers
to Track and Manage Mods
10
How Do Savvy Employers Manage Their Mods?

• Positive Methods
• Identify sources of claims and engage in loss
  control
• Minimize lost time to WC by voluntarily paying
  wage for time away from work
• Using in-house medical treatment to minimize
• outside, reportable treatment, and
• Worsening of medical problems (infection,
  swelling, etc)

• Artificial/Unhelpful Methods
• Summary denial of claim made by employee
• Non-reporting of recognized WC claims to
  insurer/jurisdiction
• Appealing claims to jurisdiction (legal contests)
• Inducing employee to file with non-WC payer
• Safety incentives that incent employees to not
  file claims

11
Good Example of Two Methods of Managing Claims
Source Camping Magazine, May 1994, Ed Schirick
12
Research on Effectiveness of Experience Rates

• Empirical research findings consistent with
  experience rating link to safety/claims but
  relationship indirectly measured.
• Distribution of Firm Characteristics (e.g., size)
  claims experience trends.
• Ruser (1985), Worrall Butler (1988)
• Regulation and Claims Experience
• Danzon Harrington (2001), Barkume Ruser (2001)
• Event Studies
• Employer opinion surveys

13
Previous Statistical Research

• The outstanding defect of previous statistical
  research is that it did not measure experience
  rating with the actual experience of individual
  employers
• They used coarse approximations
• Losses for very dissimilar employers
• Capturing experience rating solely through
  employer size
• Nor did they consider the dynamics of how
  employers react to changes in their mod factors

14
New Findings

• We obtained individual experience data from the
  rating bureaus in MN, PA, WI and a similar set
  for CA
• We estimated the individual employer reactions to
  changes in their mod factors
• Estimates were made with an improved statistical
  technique
• The model accounted for size of employer and
  inherent riskiness of each industry class/group

15
Econometric ResultsGood Theoretically and
Statistically
The only parameter less .99 confidence interval
16
Statistical Interpretation

• As an employer grows in employment size, claims
  grow more slowly than employment
• Very large employers have fewer claims than
  others
• The mod factor itself is positively related to
  average claim rate
• Mod goes up, then average claims go up
• The change in mod is negatively related to
  average claim rate
• Mod goes from 1.5 to 2, then claims go down
• NOTE these are AVERAGE responses, not true for
  every employer
• Think of them as the odds of something happening,
  all other things being equal

17
Common Sense Explanation

• The change in premium gets the employers
  attention more than a bill with the same premium
  level
• Looking at a changed bill, the employer will
  discover any large change in the mod factor and
  begin the process of asking why?
• The discovery process will sort out fluke changes
  (salesmans car was rear ended) from systematic
  changes in safety practices (hiring a lot of new
  workers)
• Employer will either
• game the claims reporting
• ask how claims can be made lower

18
Summary

• Class rates dont affect safety, only individual
  experience rating matters
• Econometric analysis strongly indicates that a
  behavioral response from changes in mod factors
• The change may be to game the system or to
  improve safety
• Larger employers seemed to understand X-Mods
• They have the knowledge and resources to respond
• Small employers have much less incentive and
  ability to respond to X-Mods

19
Thank You

• If you have questions, please contact me at
• gkrohm_at_iaiabc.org
• This presentation can be downloaded from the
  IAIABC website
• www.iaiabc.org then go to Library