February 11, 2012
Automobile Appraisal Concentration
by Michael Rowe
Hundreds of millions of dollars in claim leakage are the product of an antiquated approach to automobile appraisal triage. This article sets a path to a far more effective approach based on qualitative analysis and advocates for coopetition, especially among small to medium size carriers.
The Property & Casualty Insurance Industry claim automobile appraisal process is highly inefficient, resulting in hundreds of millions of dollars in annual leakage. This leakage is driven by both insufficient quantification of performance variation among appraisal resources and shortcomings in the assignment triage process. This article provides a path from today’s genre-centered, mean-based, macro methodology to tomorrow’s individual resource centered, and micro methodology. A comparison of the financial results produced by the old and new assignment methodologies with the same population of assignment data would empirically demonstrate the superiority of the new methodology and more than
cost justify the transition costs.
Appraisal Genre Versus Individual Resource Driven
The term “genre” is used to describe the traditional high-level appraisal resource categories of Direct Repair Program, Desk Review, Staff, and Independent Appraiser. Within the industry the typical order of priority for each genre is:
- Direct Repair Program
- Desk Estimate (if within parameters)
- Staff Appraiser
- Independent Appraiser
The implicit assumption is that the mean performance within each genre is highest for Direct Repair Programs and lowest for Independent Appraisers, making independent appraisers the choice of last resort. Even if these assumptions were accurate, basic statistics suggest that more than one third of the assignments from such a system will be sub-optimal.
The graph below represents a statistical normal distribution which exists for most populations of similar phenomena. As an example, the height of all human beings would form a similar pattern as would the batting averages of all major league players for the last 100 years. For most populations, the majority of individuals fall around the mean (average) range and then tail off gradually in both directions.
Assuming a statistically significant sample size of Direct Repair Programs, desk estimators, staff, or independent appraisers, and reliable performance measures, each would produce a similar bell curve distribution. Companies that do a better job of managing their appraisal resources should have more compact distributions, but they would still form a bell curve, albeit taller and narrower.
The main point is that, while on average the current hierarchy of appraisal resource utilization might be accurate, using available information it is possible to do much better than that. As a specific example, a highly performing independent appraisal firm could be a better option than a low performing Direct Repair Program.
Taking this to its logical conclusion, the old paradigm about an ordered genre based assignment priority should be eliminated in favor of a best performing resource approach. Over a large population of assignments a best performing resource approach produces a much better financial outcome than does the genre based approach.
Beating a Dead Horse
The industry currently operates on a genre based approach to valuing appraisal resources, meaning that it lumps together all
staff, all Direct Repair Programs, all Independent Appraisers, etc. as if each resource within each genre performed at a consistent level. Let us be clear that we are speaking about the variables of service, severity and expense and combining them into a single weighted composite performance score. When data was scarce, a genre based priority was the “Best Practice” approach, meaning that it represented the most optimized appraisal distribution possible given that state of technology. But with the advent of the information age, trying to force more efficiency out of the old way is not as effective as evolving a tool that leverages the new possibilities.
Building a Better Mousetrap
As a baseline, the assumption is that carriers already have the data they need to construct composite performance ratings for all of their appraisal resources. The standard measures around service, severity and expense typically form the basis for composite performance ratings, but when it comes to questions like how much additional expense or severity am I willing to incur to avoid service degradation, each company has its own tolerance ranges and the construction of individual resource composite performance ratings should reflect those preferences. Individual company preferences are essentially their “secret sauce” that allows them to tune the system to facilitate their competitive strategy. In an extreme example, a non-standard carrier may think differently about the weight they would attach to service versus severity than a standard carrier. Such distinctions are already reflected in the carrier appraisal protocols that vary from company to company. Though the secret sauce may vary, the overall process structure for composite performance rating could be static.
It makes good sense to think about this from the perspective of software requirements because the end game is to design a program that looks at the assignment criteria and matches it to the best available appraisal resource, regardless of genre. This simply means that we want to express the result of qualitative analysis in machine readable code.
A fairly straightforward approach would be to utilize a 1, 2 or 3 rating for each of the three critical performance components of service, severity and expense and then combine the three into a composite rating between 3 and 9. An appraisal resource receiving the best score of 3 for each of the three performance components would earn a composite performance rating of 9, while a resource receiving the lowest rating of 1 for each would be rated a 3. By capturing the assignment criteria in a system, which happens routinely today, the program could identify potential appraisal resources and choose the highest rated resource.
This leaves the question of how to generate the ratings? The best answer would be to allow the program to analyze recent historical data for service and quality and add a static expense table. As part of their “secret sauce” each carrier could decide what result levels would trigger their ratings. As an example: cycle time of 7 days or more earns a 1 rating, from over 5 days to under 7, a 2 rating and 5 days or less a 3 rating. The approach to severity would be similar, but with an added level of complexity because the various quality measures would first need to be composited. For expenses, a static rating table could be incorporated where, as an example, a Direct Repair Program would likely get a 3, Desk Review a 2 and Staff or Independent Appraiser a 1.
Clearly a carrier could opt to expand the rating scale and once a carrier established its parameters, they could be programmed allowing the resulting system to run the same algorithms for each new assignment. The result would be that the best possible resource would be matched to each assignment. If a particular resource began to falter its assignments would automatically be curtailed.
The system contemplated would produce valuable measures of its own success and the key performance indicator would be a trend-line of the average composite resource rating per assignment. For the scale example suggested above, the highest possible average composite resource rating would be 9 and the lowest 3. If, in the initial stages of the system deployment, the baseline result was 6 and as the system ran it increased to 7, 7.5, 8, etc., that would be empirical evidence of improvement that could be financially quantified. The most important job of appraisal managers would be to identify and add as many new quality appraisal resources as possible. The system would provide detailed enough data to indicate exactly what genre of such resources were needed where. The Holy Grail is drawing causal correlations to improvements in customer service surveys, reductions in loss costs and reductions in expenses.
Desk Estimating As The Corner Stone?
Used properly, desk estimating can provide a significant advantage by quickly and cost effectively resolving routine damages making more capacity for appraisal resources better suited to more complex damages. Though dollar amounts tend to drive desk adjustment decisions, in a perfect world the nature of the damages would also be considered. The proliferation of smart phones has opened the door to pre-assignment vehicle damage photos, a potential boon to the triage process, especially related to desk estimating. The very term “desk estimate” is heading for obsolescence as software programs increasingly interact to complete more assignments without human intervention squeezing cycle times into seconds, ensuring high levels of accuracy, and further driving down expense. As long as desk estimates (or eventually system estimates) are properly targeted during assignment they will produce consistently high average composite performance scores. Finally, allowing customers to take their settlement checks and shop for the best repair option brings the benefits of the free market to bear, a better alternative to directing them to average or lower performing Direct Repair Program shops.
The system described becomes exponentially profitable as it increases in scale. The one pitfall compared to the legacy genre approach is that within each genre there existed a sufficient population of data to provide statistical significance. Drilling down to an individual appraisal resource level will reveal that about 20% of appraisal resources lack sufficient data samples for valid ratings, typically such resources are in low concentration rural areas. Given the programmed approach described above, desk estimating would be the likely system selection whenever the damages are of an appropriate nature and that seems logical. Outside of that, the program could default to the legacy genre approach. But there is an even better answer to the scale issue.
According to Wikipedia:
Coopetition occurs when companies work together for parts of their business where they do not believe they have competitive advantage and where they believe they can share common costs. For instance, the arrangement between PSA Peugeot Citroën and Toyota to share
components for a new city car — simultaneously sold as the Peugeot 107, the Toyota Aygo, and the Citroën C1 — qualifies as coopetition. In this case, companies save money on shared costs while remaining fiercely competitive in other areas.
There is no question that customer service and loss cost management impact competitiveness, but price and brand largely drive the insurance buying public’s decisions. Given the fact that scale can increase the financial and customer service benefits of an appraisal program, at the very least, small to medium size carriers should take note. By entering into a “coopetition” style approach to automobile appraisal that invokes the methodologies defined in this article, such carriers could greatly improve their competitive positions vis-à-vis the larger industry carriers, which should be their primary focus.
A collective appraisal program would allow the partner carriers to judge the performance of each appraisal resource based on the collective results produced for all partner companies. As an example, if 50 companies participated in a collective program and a particular appraisal resource had been handling an average of one assignment per month for each carrier, performance for that resource could be judged on the basis of the collective population of 50 appraisals and that resource would understand that all of the assignments would be at risk going forward if performance measures were not maintained.
More importantly, where sufficient collective claim concentrations emerged drive-in claim facilities and Direct Repair Programs leveraged to their full extent driving down costs and speeding up service. Similarly, higher assignment volumes to desk estimating and independent appraisal firms create more loyalty and better rate structures.
Most importantly, if a common appraisal management and quality control structure were employed, the savings from the huge redundancies across competing companies would be immense, and by selecting the best of the best, field quality control oversight becomes both exponentially better and more costeffective. This same line of thinking holds for auto glass, rental vehicles, etc.
In the claim business it is not untypical for claim leaders from each company to believe that their particular practices excel beyond those of all of their competitors, but going back to the normal distribution from statistics, it’s just not reality. Some companies in the population are at the high-end but even for them, the structural limitations presented by low concentration imposes hard physical limitations on what they can accomplish, and even if they were to ostensibly lose some of the potency of their employed acumen, with sufficient scale, it would be compensated for many times over.
The optimum approach combines significant scale, performance based assignments, and Best Practice APD oversight, which can only be attained through coopetition.
If you’re not one of the top 15 automobile writers in the US, you should really start to concentrate.